Author’s Accepted Manuscript FMS-like tyrosine kinase 3 (FLT3) inhibitors: molecular docking and experimental studies Baratali Mashkani, Mohammad Hossein Tanipour, Mohammad Saadatmandzadeh, Leonie K. Ashman, Renate Griffith www.elsevier.com/locate/ejphar

PII: DOI: Reference:

S0014-2999(16)30072-3 http://dx.doi.org/10.1016/j.ejphar.2016.02.048 EJP70482

To appear in: European Journal of Pharmacology Received date: 15 July 2015 Revised date: 14 February 2016 Accepted date: 15 February 2016 Cite this article as: Baratali Mashkani, Mohammad Hossein Tanipour, Mohammad Saadatmandzadeh, Leonie K. Ashman and Renate Griffith, FMSlike tyrosine kinase 3 (FLT3) inhibitors: molecular docking and experimental s t u d i e s , European Journal of Pharmacology, http://dx.doi.org/10.1016/j.ejphar.2016.02.048 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting galley proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

FMS-like tyrosine kinase 3 (FLT3) inhibitors: molecular docking and experimental studies Baratali Mashkania,b*, Mohammad Hossein Tanipoura,c, Mohammad Saadatmandzadehd, Leonie K. Ashmanb, Renate Griffithe a

Department of Medical Biochemistry, School of Medicine, Mashhad University of Medical Sciences, Mashhad,

Iran. Email: [email protected] b

School of Biomedical Sciences and Pharmacy, University of Newcastle, Callaghan, NSW 2308, Australia. Emails:

[email protected]. [email protected] c

Student

Research

Committee,

Mashhad

University

of

Medical

Sciences,

Mashhad,

Iran.

Email:

[email protected] d: Department of Chemistry, Faculty of Sciences, Ferdowsi University, Mashhad, Iran e

School of Medical Sciences/Pharmacology, UNSW Australia, Sydney, NSW 2052, Australia. Email:

[email protected] *

Corresponding author: Baratali Mashkani. Department of Medical Biochemistry, School of

Medicine, Mashhad University of Medical Sciences, Mashhad, Iran. Email: [email protected]. Tel: +98 (51) 3800 2375; Fax: +98 (51) 3882 8574 Abstract Activating mutations in FMS-like tyrosine kinase 3 (FLT3) occur in 25% of acute lymphoid and 30% of acute myeloid leukemia cases. Therefore, FLT3 is a potential therapeutic target for small molecule kinase inhibitors. In this study, protein-ligand interactions between FLT3 and kinase inhibitors (CEP701, PKC412, sunitinib, imatinib and dasatinib) were obtained through homology modelling and molecular docking. A cellular system for experimental testing of the inhibitors was also established by expressing wildtype and internal tandem duplication mutant FLT3

Page 1 of 33

(FLT3-WT and FLT3-ITD) in FDC-P1 cells. Imatinib and dasatinib could not be docked into any of the FLT3 models, consistent with their lack of activity in the experimental assays. CEP701, PKC412 and sunitinib interacted with the ATP-binding pocket of FLT3, forming Hbonds with Cys694 and Glu692. Based on the EC50 values in the cell proliferation assay, CEP701 was the most potent inhibitor; sunitinib and PKC412 were ranked second and third, respectively. Sunitinib was the most selective inhibitor, followed by PKC421 and CEP701. The potency of sunitinib and to a lesser extent CEP701 in inhibition of FLT3 autophosphorylation was lower than the cell proliferation inhibition, indicating that inhibition of FLT3 downstream proteins may contribute to the cellular effects. It was shown in this study that the docking procedure was able to differentiate FLT3 inhibitors from ineffective compounds. Additionally, interaction with the phosphate binding region in the ATP-binding pocket increased potency at the cost of selectivity. These findings can be applied in designing highly effective and selective inhibitors for FLT3 and other related kinases.

Keywords: FLT3; Docking; GOLD; Cell Proliferation inhibition; Autophosphorylation Inhibition; Small Molecule Kinase Inhibitors; Chemical compounds studied in this article Imatinib (PubChem CID: 5291); Dasatinib (PubChem CID: 3062316); Sunitinib (PubChem CID: 5329102);

Page 2 of 33

PKC412 (PubChem CID: 9829523); CEP701 (PubChem CID: 126565) 1. Introduction FMS-like tyrosine kinase-3 (FLT3) is normally expressed by early hematopoietic stem/progenitor cells and it disappears during differentiation (Gotze et al., 1998). Overexpression of FLT3 mRNA was reported in acute myeloid leukaemia (AML), acute lymphoid leukaemia (ALL) and the blast crisis of chronic myeloid leukaemia (CML) (Carow et al., 1996). Internal tandem duplications (ITD) in the juxtamembrane (JM) domain are the most frequent mutations causing constitutive activation of the FLT3 kinase domain (Meshinchi and Appelbaum, 2009). Mutant forms of FLT3 have been reported in 25% of childhood ALL (Armstrong et al., 2004) and 30% of AML patients (Griffin, 2001; Levis and Small, 2003), in myelodysplastic syndrome (MDS) and chronic myelomonocytic leukemia (CMML)(Daver et al., 2013). Additionally, they have a confirmed role in transformation of MDS to AML (Shih et al., 2004), in AML relapse and poor response to stem cell transplantation therapy (Tiesmeier et al., 2004), as well as in autoimmune diseases (Andersson et al., 2012; Singh et al., 2012; Whartenby et al., 2005).Therefore, FLT3 is a potential therapeutic target for small molecule kinase inhibitors (SMIs) in leukemia (Ustun et al., 2009) and autoimmune diseases (Whartenby et al., 2005; Whartenby et al., 2008). In recent years, many different small molecules have been reported to have inhibitory effects on FLT3 kinase. Among them, quizartinib (AC220) (Kampa-Schittenhelm et al., 2013), AKN028 (Eriksson et al., 2012), BPR1J-097 (Lin et al., 2012) and ENMD-2076 (Fletcher et al., 2011) are under investigation in different stages of preclinical and clinical studies. Lestaurtinib

Page 3 of 33

(CEP701) (Smith et al., 2004), midostaurin (PKC412) (Weisberg et al., 2002) and sunitinib (Sutent) (Levis and Small, 2004) have also been suggested as possible drugs for FLT3 inhibition, but they are also active against a wide range of other kinases (Fabbro et al., 2000; Levis and Small, 2004). Structural studies of the interactions of the FLT3 kinase with the inhibitors could help in understanding important factors in potency and selectivity. However, there is only one published FLT3 kinase domain crystal structure (1RJB) in an autoinhibited conformation (Griffith et al., 2004). Alternatively, homology modelling and ensemble docking is of particular importance in kinases with their well-known conformational flexibility (Cowan-Jacob et al., 2009). We have previously demonstrated the validity of this approach for docking into the FMS kinase domain (Mashkani et al., 2010). FDC-P1 (ATCC No. CRL-12103) is a factor-dependent murine early myeloid cell line requiring murine IL-3 (Dexter et al., 1980) or GM-CSF (Foster et al., 2004) for growth. However, expression of tyrosine kinases including FLT3 can abrogate IL-3 dependency in FDC-P1 cells which has been related to the induction of c-Myc (Cleveland et al., 1988; Rossner et al., 1994). In this study, potency and selectivity of CEP701, PKC412 and sunitinib were investigated for inhibition of cell proliferation and autophosphorylation in the FDC-P1 cells expressing human FLT3. By relating the docking to the biological assay results, this study reveals the characteristics of the druggable binding pocket of FLT3 and identifies the residues that can be targeted for development of new and selective inhibitors.

2. Materials and methods 2.1 Homology modelling Page 4 of 33

Homology models of FLT3 were built and validated as described before (Mashkani et al., 2010), based on crystal structures of an inhibited conformation of KIT with imatinib (1T46) (Mol et al., 2004), 2GQG, an active conformation of the Abl kinase with dasatinib (Tokarski et al., 2006), 3G0E, an inhibited conformation of KIT with sunitinib (Gajiwala et al., 2009), 1PKG, an active conformation of KIT with ADP (Mol et al., 2003), and 2I1M, an inhibited conformation of FMS in complex with an arylamide molecule (Schubert et al., 2007), all obtained from the RCSB Protein Data Bank (PDB) as templates. SWISS-MODEL and the Swiss-Pdb Viewer were used for preparation and evaluation of the models (Guex and Peitsch, 1997). The name of the target sequence (FLT3) followed by “model” and the name of the template structure (1T46) was used to designate each homology model (i.e. FLT3model1T46). The models were used for docking of CEP701, dasatinib, imatinib, PKC412 and sunitinib (Fig. 1) without any further minimization or modification. 2.2 Docking procedure GOLD Suite 5.2.2 software (the Cambridge Crystallographic Data Centre (CCDC), Cambridge, UK) was used for molecular docking of the inhibitors into 1RJB and five different homology models of FLT3. The centre of ligands in the superimposed crystal structures of FMS (21IM) (Schubert et al., 2007), KIT (1T46) (Mol et al., 2004) and KIT (1PKG) (Mol et al., 2003) was used as the centre for docking. The wizard in GOLD was used for preparation of both receptors and ligands for ensemble docking as instructed in the manual. Briefly, water molecules were removed and hydrogens were added to the loaded FLT3 model structures (receptors); then the inhibitors (ligands) were also added to the setting. The binding pocket radius was set to 13Å to accommodate all ligands in the superimposed structures. The ChemScore function with ChemScore kinase parameters was used for scoring. Docking results were analysed for docking Page 5 of 33

score and the interactions between docked poses of ligand and the receptor. A pdb file of the highest scored pose was extracted from docking results for further analysis in the Discovery Studio Visualizer 4 (DSV4) (Accelrys, CA, USA). In an attempt to improve the docking results, 10 residues from the binding pocket in 2Å distance from the best docked poses were set as flexible and the docking procedure was repeated. 2.3 FLT3 expression The FDC-P1 cells routinely maintained in DMEM/10% FCS supplemented with murine GM-CSF as previously described (Frost et al., 2002). Where indicated, FDC-P1 cells that had been transduced with FLT3 WT and ITD cDNA (FD-FLT3 cells) were grown in FLT3L instead of GM-CSF. The DNA constructs for FLT3 wild type and FLT3-ITD mutant were supplied by Dr. Hitoshi Kiyoi, Nagoya University School of Medicine, Nagoya, Japan. FLT3 fragments from those constructs were subcloned into MSCV-IRES-GFP vector to achieve high FLT3 expression and to allow use of GFP for cell sorting. DNA was introduced into FDC-P1 cells by retrovirusmediated gene transfer as previously described (Roberts et al., 2007). FDC-P1 cells expressing FLT3-WT (FD-FLT3-WT) were selected using FLT3L as growth factor and FD-FLT3-ITD cells in growth factor-free medium. FLT3L used in this experiment was expressed in Pichia pastoris using a similar procedure as for CSF-1 (Mashkani et al., 2013). FD-FLT3 cells were sorted (FACS Aria, BD Biosciences, CA) for expression of GFP to obtain a homogenous population. FLT3 expression on sorted cells was confirmed by flow cytometry using FLk2/FLT3 (sc-19635) antibody (Santa Cruz Biotech., Santa Cruz, CA) and sheep anti-mouse IgG conjugated with phycoerythrin (Chemicon, Billerica, MA) as secondary antibody. 2.4 Cell proliferation assay

Page 6 of 33

Serial dilutions of imatinib and PKC412 (Novartis, Basel, Switzerland), dasatinib and sunitinib (Chemietek, Indianapolis, IN) and CEP701 (LC Laboratories, Woburn, MA) were prepared from 10 mM stock solutions of each drug in dimethyl sulfoxide (DMSO, Ajax Finechem, NSW, Australia) in 100 μl DMEM (Dulbecco’s Modified Eagle’s Medium) containing 10% FCS (Fetal Calf Serum) and either 2x FLT3L or GM-CSF in 96 well plates (Nunc, Roskilde, Denmark). FD-FLT3 cells were washed with growth factor-free (GF-Free) medium and resuspended in GF-Free DMEM. 100 μl of the cell suspension containing 5x104 cells were added to each well. Plates were incubated at 37 ˚C with 5 % CO2 for 48 h. Then 20 µl of resazurin reagent [300 μM resazurin, 78 μM methylene blue, 1 mM potassium hexacyanoferrate III and 1 mM potassium hexacyanoferrate II (all from Sigma, St. Louis, MO)] were added into each well and the plates were incubated for 4 h. Fluorescence intensity of the product resorufin, proportional to the number of viable cells per well, was measured by a FLUOstar OPTIMA (BMG LABTECH, Offenburg, Germany) with excitation at 530 nm and emission at 590 nm. Results were analyzed using GraphPad Prism 4.3 (GraphPad Software, San Diego, CA). The EC50 value was determined by fitting a non-linear curve to all data points of fluorescence graphed against the logarithm of drug concentration. The ratio of EC50 values in the presence of GM-CSF and FLT3L (EC50 value of GM-CSF/FLT3L) indicated selectivity for wild type and constitutively activated mutants of FLT3 relative to other (unspecified) targets downstream of GM-CSF receptor.

2.5 Inhibition of autophosphorylation FD-FLT3 cells were grown in the presence of murine GM-CSF to a density of 106 cells/ml, washed, and incubated in FCS- and GF-Free medium for 3h at 37˚C. Then 107 cells were added Page 7 of 33

to 1 ml of FCS- and GF-free medium containing different concentrations of drugs and incubated for 30 min at 37˚C. Subsequently the cells were cooled on ice before being stimulated with FLT3L for 5 min. The cells were washed twice with cold PBS ( Phosphate Buffer Saline) and lysed with ice-cold 1% NP40 in TSE (50mM Tris-HCl, 150mM NaCl, 1mM EDTA pH 8.0) with complete protease inhibitor cocktail (Roche, Basel, Switzerland), 5mM sodium fluoride, 5mM tetra sodium pyrophosphate, 5mM sodium vanadate, and 1mM phenylmethylsulfonyl fluoride (Sigma). The lysates were centrifuged to remove cell debris and protein concentration determined using a MicroBCA kit (Pierce, Rockford, IL). Preliminary experiments indicated the presence of FLT3 degradation products in the whole cell lysate of FD-FLT3 ITD. Therefore antiFLT3 antibody SF-1.340 (Santa Cruz Biotech.) which recognizes FLT3 extracellular domain was used for its immunoprecipitation. This was not necessary in the case of FD-FLT3-WT cells where those interfering degradation products were not observed. Samples (whole cell lysate of FD-FLT3-WT

cells

and

immunoprecipitated

FLT3-ITD)

were

resolved

on

6%

SDS/polyacrylamide gels under reducing conditions along with Precision Plus protein Standards (BioRad, Hercules, CA). Separated samples were transferred to nitrocellulose membrane. The membrane was blocked with 1% BSA (Sigma) for 1 h and probed for phosphotyrosine using a cocktail of 4G10 (Upstate, Temecula, CA) and pY20 (BD Biosciences) antibodies. The FLT3 protein was detected using sc-479 antibody (Santa Cruz Biotech.) which recognizes the intracellular domain. Bound antibody was detected with secondary anti-mouse or anti-rabbit IgG antibody conjugated with HRP. Chemiluminence was measured using a Fuji imaging system (Fujifilm, Tokyo, Japan) and quantitated using Multi Gauge V3 software (Fujifilm).The intensity of phosphorylated tyrosine in FLT3 (pY) for each sample was firstly normalised to total FLT3 to compensate for the variations of protein loaded in each well. Then the values were normalised to

Page 8 of 33

the lowest concentration (approximating zero) for each drug as 100% phosphorylation to calculate the pY% for each sample. These experiments were carried out three times and the results were consistent with each other. 3. Results 3.1 Docking small molecule inhibitors into the FLT3 kinase domain Structures of the small molecule inhibitors used in this study are presented in Fig. 1. The protein structures used for docking included 1RJB (Griffith et al., 2004), and five homology models representing active and inactive structures of FLT3 prepared and evaluated as described in section 2.1. The docking procedure was validated by redocking imatinib, sunitinib and dasatinib into the crystal structures 1T46 (Mol et al., 2004), 3G0E (Gajiwala et al., 2009) and 2GQG (Tokarski et al., 2006), respectively. The docking parameters were changed until the docked poses were highly similar with less than 2Å root mean square deviation (RMSD) from the pose in the original crystal structure. The main changes consisted of increasing population size and max operation times up to 1000 and 1000000, respectively. 1RJB represents an autoinhibited conformation of the FLT3 kinase domain, no ligand is crystallised, and part of the juxtamembrane domain blocks the binding cleft. Due to the closed structure of 1RJB, none of the five inhibitors could fit in either the ATP-binding or substrate pockets. Space filling representation of residues Gln575, Leu576, Gln577, Lys644, Phe691, Cys694, Leu818, DFG motif residues (Asp829, Phe830, Gly831) and Leu832 shows that they were interfering with the docking (Fig. 2). In the next step, all the inhibitors were docked into all the FLT3 homology models. The best results were obtained from docking CEP701 and PKC412 into active conformations of Page 9 of 33

FLT3 built based on the 1PKG structure; sunitinib was docked fairly well into a FLT3 inactive model based on 1T46. However, sunitinib could not be docked into an inactive conformation model based on 3G0E or the active conformation models. Imatinib and dasatinib could only be docked shallowly into inactive conformations of FLT3 based on 1T46 and 2I1M, and had interactions with some of the residues on the protein surface. 3.1.1 CEP701 Among the six different structures, CEP701 was docked best into the ATP binding site (in the hinge region) of a FLT3 model based on the active conformation of the KIT kinase domain (1PKG). Docked energy (Chemscore.DG) for the best pose was -48.6 kcal/mol. The molecule formed hydrogen bonds (H-bonds) with Glu692 and Cys694 in the hinge loop, as well as with Arg815 in the catalytic loop of the C-terminal lobe of the kinase domain. It also had hydrophobic and π-interactions with Val624, Leu616 in the nucleotide binding loop of the N-terminal kinase loop and with Tyr693 of the hinge (Fig. 3). 3.1.2 PKC412 Docking PKC412 into models of the active and inactive conformations of FLT3 showed that this molecule also prefers the active conformation based on 1PKG. Docked energy for the best pose of PKC412 was -54 kcal/mol. It formed the same hydrogen bonds (H-bonds) with Glu692 and Cys694 and the same hydrophobic and π-interactions with Val624, Leu616 and Tyr693 as does CEP701 (Fig. 4). PKC412 overall docking is very similar to CEP701 except for its benzylamide tail, which forms an additional H-bond with Ser618 in the nucleotide binding loop. However, PCK412 could not form any H-bonds with Arg815 as a result of blockade of one of the OH groups with a Page 10 of 33

methyl group as indicated by a circle in Fig. 4, and the benzylamide substitution of the other OH group.

3.1.3 Sunitinib The best results were obtained when sunitinib was docked into the inactive conformation of FLT3 based on1T46 (KIT structure cocrystallised with imatinib) with docked energy of -40 kcal/mol. It formed H-bonds with Glu692, Cys694 and Asp698 in the hinge, but did not form any other interactions with the kinase (Fig. 5A). When FLT3Model1T46 was superimposed on 3G0E (KIT structure cocrystallised with sunitinib), the best docked pose was very similar to the position of sunitinib in the crystal structure (Fig. 5B), with the ‘tail’ (see Fig. 1), which is not resolved in the 3G0E, exposed to the solvent. We expected some differences between the docked pose and the crystal structure, because the gate-keeper residue (Thr670) in KIT is replaced by Phe691 in FLT3. As a result, sunitinib could not be docked properly into FLT3Model3G0E. When this model was superimposed on FLT3Model1T46, the positions of the Phe691, Cys694 and Gly697 sidechains constitute a spatial hindrance for correct docking (Fig. 6). Sidechains for those three residues were also set to flexible; however, this did not significantly improve the docking results. Further analysis showed some perturbation in the protein backbone in the area of Gly697, which may also contribute to the docking failure. The bulkier ligand imatinib in 1T46 has created a larger space, so that the Phe691 sidechain in the FLT3 model based on 1T46 is orientated such that docking of sunitinib is possible.

3.1.4 Comparing the docked poses of CEP701, PKC412 and sunitinib

Page 11 of 33

Superimposing the FLT3 models containing docked CEP701, PKC412 and sunitinib revealed that all these inhibitors have rings occupying the space for adenine and ribose in the ATP binding pocket (Fig. 7). This indicates that H-bond formation with Glu692 and Cys694 in the hinge loop is the common feature for all three molecules. CEP701, but not PKC412 and sunitinib, formed H-bond with Arg815 located in the catalytic loop of the kinase domain which is the segment normally would interact with phosphate groups of the ATP.

3.1.5 Dasatinib and imatinib Due to the spatial hindrance by the bulky side chain of Phe691, imatinib and dasatinib could not be docked correctly into the binding pockets in the FLT3 kinase in the models even in those based on the crystal structures 1T46 and 2GQG, respectively. They could only be docked shallowly and had some interactions with residues on the protein surface. Lack of any interaction with the key residues in the binding pocket was the common feature in docking imatinib and dasatinib into all FLT3 conformations.

3.2 Expression of FLT3 in FDC-P1 cells It was demonstrated by immunofluorescence and flow cytometry analysis, that 94 and 92% of the FD-FLT3-WT and FD-FLT3-ITD cells had high levels of FLT3 protein expression (Fig S1). Despite having strong bands in the FLT3-expressing cells in Western blot analysis of phosphotyrosine and FLT3, there was no detectable protein band around 165 or 130 kDa in the control samples (transfected with empty vector). This indicated that the expressed human FLT3 overwhelms the FDC-P1 native proteins with similar molecular weight. As a result, the whole cells lysate could be used for the analysis of both total and phosphorylated FLT3. However, Page 12 of 33

immunoprecipitation was used for FLT3-ITD protein analysis to reduce the background caused by degradation of this constitutively active mutant.

3.3 Inhibition of FD-FLT3 cell proliferation 3.3.1 Inhibition of FD-FLT3-WT cell proliferation Cell proliferation assays showed that CEP701, PKC412 and sunitinib could inhibit the growth of FD-FLT3-WT cells, but dasatinib and imatinib were not effective (Fig. 8) even at concentrations 10 times higher than the EC50 value for FD-FMS (Mashkani et al., 2010). As indicated in Table 1, sunitinib with an EC50 value of 7.9 nM and EC50 value ratio between GMCSF and FLT3L stimulated cells of >12.5 appeared to be the most potent and selective among the five compounds tested in these experiments. CEP701 was more potent than PKC412 with EC50 values of 12 and 47 nM, respectively. The EC50 value of CEP701 in the presence of GMCSF was almost 3.5 times greater than in the presence of FLT3L showing some selectivity for blockade of FLT3-WT signalling upon stimulation by FLT3L, but its selectivity was lower than for PKC412 with an EC50 value ratio of GM-CSF/FLT3L of about 10.

Table 1. Comparison of EC50 values of the small molecules for inhibition of cell growth of FD-FLT3-WT in the presence of either GM-CSF or FLT3L. Each EC50 value was calculated as the average of three independent experiments ± standard deviation and with four replicates in each experiment. CEP701 (nM)

Dasatinib (nM)

Imatinib (nM)

PKC412 (nM)

Sunitinib (nM)

EC50 value in GM-CSF

42.3±12

>100

>2000

>500

>100

EC50 value in FLT3L

12.2±1.2

>100

>2000

47.1±1.2

7.9±0.9

Page 13 of 33

GM-CSF/FLT3L ratio

3.5

--

--

>10

>12

3.3.2 Inhibition of FD-FLT3-ITD cell proliferation The response of FD-FLT3-ITD cells to different concentrations of the tested compounds are presented in Fig. 9. Imatinib and dasatinib could not inhibit the growth of these cells. The EC50 value of CEP701 for the FLT3-WT and the ITD mutant were the same, but the ITD mutant was slightly less sensitive to PKC412 (Table 2). The EC50 value of sunitinib for FLT3-ITD was 27 nM, three times greater than for FLT3-WT. Based on these findings, CEP701 was the most potent and PKC412 was the most selective among these inhibitors against the cells expressing FLT3-ITD.

Table 2. Comparison of EC50 values of the small molecules for inhibition of cell growth of FD-FLT3-ITD in the presence of either GM-CSF or FLT3L. Details of experiments as for Table 1. CEP701 (nM)

Dasatinib (nM)

Imatinib (nM)

PKC412 (nM)

Sunitinib (nM)

EC50 value in GM-CSF

34.2±6.3

>100

>2000

310±13

>100

EC50 value in FLT3L

14.3±1.6

>100

>2000

66±12

27.3±3.6

GM-CSF/FLT3L ratio

2.4

--

--

4.7

>3.6

3.4 Inhibition of FLT3 phosphorylation 3.4.1 Inhibition of FLT3-WT phosphorylation Protein bands of 160 and 130 kDa represent the fully glycosylated (mature) and less glycosylated (immature) forms of FLT3, respectively. Some phosphorylation of FLT3 was

Page 14 of 33

observed even without stimulation by FLT3L (Fig. 10, Lane C-/-) which may be due to the relatively high level of FLT3 expression on FD-FLT3-WT cells (Armstrong et al., 2003). Dasatinib and imatinib failed to inhibit autophosphorylation of FLT3 despite using a concentration much higher than their EC50 values for wild type KIT and FMS. This supports the results of failed docking and their lack of effect in the cell proliferation assays. CEP701, PKC412 and sunitinib inhibited the autophosphorylation of FLT3-WT upon stimulation by FLT3L (Fig. 10). FLT3 autophosphorylation was reduced to 59% of the control at 50 nM concentration of PKC412, which was quite close to the results of the cell proliferation assay (EC50 value of 47 nM). CEP701 reduced the phosphorylation of FLT3 down to about 40% at the concentration of 50 nM compared with its EC50 value in the cell proliferation assay of 12 nM. Although sunitinib was also a potent inhibitor of FLT3-WT-induced cell proliferation (EC50 value of 8 nM), a concentration of 100 nM was required to reduce the autophosphorylation down to 37 % of the drug control. This suggests that for CEP701 and sunitinib, off-target effects on downstream signalling proteins contributed to the inhibition of cell proliferation.

3.4.2 Inhibition of FLT3-ITD phosphorylation Neither of the control samples in this experiment was treated with the drugs, but the FLT3L– stimulated control (C-/+) had higher levels of phosphorylation compared to the unstimulated control (C-/-). This may suggest that, even though FLT3-ITD mutant has enough constitutive activity to support factor-independent cell proliferation, it is not fully phosphorylated in the absence of FLT3L. Inhibition of phosphorylation of FLT3-ITD protein by CEP701, PKC412 and sunitinib was dependent on their concentration (Fig. 11). CEP701 and PKC412 decreased the tyrosine phosphorylation down to 61 and 75% at the concentrations of 10 and 50 nM, Page 15 of 33

respectively, which was comparable with their EC50 values of 14 and 60 nM in the cell proliferation assay. Sunitinib, on the other hand, failed to appreciably inhibit autophosphorylation of FLT3-ITD kinase at concentrations close to its EC50 value of 27 nM measured through the cell proliferation inhibition assay. This means that sunitinib is not highly selective for FLT3 kinase. 3.4.3 Comparing EC50 values from phosphorylation inhibition of FLT3- WT and -ITD Table 3 shows the EC50 values of the kinase inhibitors (CEP701, PKC412 and sunitinib) for inhibition of FLT3 phosphorylation. It indicates that CEP701, with EC50 values of 39 and 33 nM, does not differentiate between FLT3-WT and –ITD. However, PKC412 and sunitinib were much effective against FLT3-WT than the –ITD mutant. The EC50 values of PKC412 and sunitinib were 35 and 79 nM against FLT3-WT and 109 and 188 nM for FLT3-ITD. Table 3. EC50 values of the kinase inhibitors for inhibition of FLT3 phosphorylation. The EC50 values were calculated from analysing the intensity of the phosphorylated FLT3 bands in different Western blotting experiments. Three replicate experiments were performed in each case and the data are presented as Mean±S.D.

FLT3-WT

FLT3-ITD

CEP701 (nM)

39±15

33±9

PKC412 (nM)

35±2

109±24

Sunitinib (nM)

79±10

188±65

3.5 Comparing docking results with EC50 values

Page 16 of 33

Table 4 shows that the overall outcome of the docking with GOLD was consistent with the cell based assays. Using GOLD, the best docking energies for CEP701, PKC412, and sunitinib were -48, -54, and -40 kcal/mol, respectively. This indicates that GOLD succeeded to identify sunitinib as a less competent inhibitor of FLT3 autophosphorylation than CEP701 and PKC412. Dasatinib and imatinib inhibited neither the cell proliferation nor FLT3 autophosphorylation, and could not be docked properly into the ATP-binding pocket. The inhibition of cell proliferation by sunitinib, and to some extent by CEP701, has been demonstrated to be due to additional factors, rather than FLT3 inhibition alone. Thus, the EC50 values in this assay do not correlate well with docking energies of the compounds.

Table 4. Comparing the EC50 values from inhibition of cell proliferation and phosphorylation with docked energies. NA: Not applicable docked energy where the ligand could not be docked into the FLT3 active or inactive conformations. EC50 for cell based assay

EC50 for phosphorylation inhibition

Docked energy for the best fitted pose

(nM)

assay (nM)

(kcal/mol)

Compound

FLT3

WT

ITD

WT

ITD

Active

Inactive

CEP701

12

14

39

33

-48

NA

PKC412

47

66

35

109

-54

NA

Sunitinib

8

27

79

188

NA

-40

4. Discussion The structure of small molecule inhibitors (SMIs) determines their interaction mode with the target as well as their potency and selectivity. Structural analysis of KIT and Bcr-Abl kinases in complex with imatinib and dasatinib, respectively (Mol et al., 2004; Tokarski et al., 2006), showed that these compounds only cover the space for adenine in the ATP binding site and Page 17 of 33

interact partially (dasatinib) or extensively (imatinib) with the catalytic area towards the Cα helix in front of the kinase cleft. Since the catalytic area is involved in recognition of the substrate peptide, it is less conserved among different kinases. Therefore, interaction with this area improves the selectivity of the inhibitors. However, replacement of the gate-keeper residue, which separates the substrate binding area from ATP-binding pocket, in KIT and Bcr-Abl mutants with hydrophobic amino acids renders them insensitive to imatinib (Tamborini et al., 2006) and dasatinib (Tokarski et al., 2006), respectively. The solution was designing new compounds such as ponatinib (AP24534) with a 2,3-dihydroimidazo [1,2-b] pyridazine which resembles a purine structure connected to the methylbenzene in the catalytic area with an ethyne linker, which is longer and more hydrophobic than NH in imatinib (O'Hare et al., 2009) to target T315I Bcr-Abl. However, the threonine gate-keeper residue in KIT and Abl kinase, is replaced by a phenylalanine (Phe691) which is not only nonpolar but also more bulky leaving less space for designing effective and selective inhibitors against FLT3 kinase. It was shown in this study that having Phe691, as the gate-keeper residue in FLT3, prevented imatinib and dasatinib but not CEP701, PKC412 and sunitinib from being docked into the hinge region of the ATP-binding pocket. Superimposing docked poses (Fig. 7) showed that sunitinib, CEP701 and PKC412 have a ring system resembling the purine structure and function by forming H-bonds with Cys694 and Glu692 in the hinge region of FLT3, which is highly conserved among kinases. These H-bonds are important for anchoring the binding between the SMIs and the kinase domain. This enables these compounds to compete with ATP for binding to the ATP binding pocket of the kinases. Since this area is more conserved among the kinases, the molecules which exclusively bind to the ATP binding pocket are not highly selective.

Page 18 of 33

It was reported that ITD mutations in the juxtamembrane domain of FLT3 not only rendered the FDC-P1 cells factor independent, but also changed the drug response (Mashkani et al., 2014). Sunitinib was more potent against the inactive conformation of FLT3 with its preferential docking into FLT3Model1T46 and lower EC50 values of 69 for FLT3-WT autophosphorylation compared to 500 nM for FLT3-ITD. Therefore, sunitinib can be used as an example of how conformational changes caused by activating mutations may affect the efficiency of kinase inhibitors even if the mutated residues are unrelated to the inhibitor binding pocket. This finding is consistent with studies on KIT where despite having similar modes of binding to both WT and constitutively active D816H mutant KIT, the EC50 value for the active state of the kinase is much higher than for inactive forms (Gajiwala et al., 2009). In inactive kinase conformations, the side chain of the phenylalanine in the DFG motif of the activation loop points towards the ATP binding site, occupies the space for the ATP phosphate groups, and prevents ATP access to its binding site, without interfering with sunitinib binding. Therefore, small molecule inhibitors with relatively low affinity (i.e. sunitinib) can also bind to the kinase and further stabilise the inactive conformation. However, a high affinity SMI is required to compete with the high concentration of intracellular ATP in binding to constitutively active kinases. Furthermore, the low activity of sunitinib in the FLT3 phosphorylation inhibition assay compared to its much higher potency in inhibiting cell proliferation, may suggest that it was active against the downstream FLT3 signaling proteins. Inhibition of at least eight kinases including FMS (Mashkani et al., 2010), KIT, PDGFRα and β, as well as VEGFR1-3 has also been reported by sunitinib (Roskoski, 2007). Since sunitinib has two fused rings, which are too big to fit into the catalytic area, it therefore only binds to the hinge region and occupies a small area in the hinge region of the kinase domain; thus it is not a highly selective kinase inhibitor.

Page 19 of 33

CEP701 was a competent inhibitor of FLT3 in the cell proliferation assay, but not highly selective. Phosphorylation of both FLT3-WT and the ITD mutant was also decreased by 50% at the concentration of 33-39 nM of CEP701 which is almost 3 times higher than its EC50 value in the cell proliferation assay. This indicates that its off-target activity played an important role in the cell proliferation inhibition assay. PKC412 appeared less potent but more selective for FLT3 compared to CEP701.The results of the FLT3-WT phosphorylation inhibition assay for PKC412 were consistent (EC50 value of 35 nM) with the cell proliferation assay (EC50 of 47) for FLT3WT. However, its EC50 values against the ITD mutant in inhibition of cellular proliferation and autophosphorylation assays were 66 and 108 nM, respectively. These findings indicate that CEP701 is equally effective on FLT3-WT and ITD, but PKC412, similar to sunitinib, was less effective on the active mutant. PKC412 and CEP701 have similar structures with several fused rings which are too bulky to fit into the catalytic area, but interact with the ATP binding area which is highly conserved among kinases. CEP701 and PKC412 not only occupy the space for adenine, but also have a ring (tetrahydro-2H-pyran in PKC412 and tetrahydrofuran in CEP701) which occupies the space for the ribose moiety of ATP. Both these compounds also have an aromatic ring which is in the vicinity of a hydrophilic pocket (created by the conserved lysine and the side chain of the aspartate from the DFG motif), which would normally be occupied by phosphate groups of ATP. It may be possible to add some H-bonding substituents to this aromatic ring and increase the potency against kinases and improve the bioavailability by increasing water solubility of the molecules. Since the ATP binding area is highly conserved among kinases, the structural characteristics of CEP701 and PKC412 which interact with almost all the residues important in ATP binding

Page 20 of 33

adversely affect selectivity of these compounds for specific kinases compared to imatinib and dasatinib. CEP701 has two hydroxyl groups (H-bond donors) which allow H-bonds with Arg782 in FMS (Mashkani et al., 2010) and Arg815 FLT3. This arginine in the catalytic loop is highly conserved and it is adjacent to Asn816 (equivalent of Asn797 in KIT which interacts with phosphate groups of ADP in 1PKG). This interaction may have contributed to the potency of CEP701 against FLT3. PKC412 cannot form either of these two H-bonds due to methylation of one of the two OH groups and substitution with a benzylamide tail. The amide oxygen of this tail forms a H-bond with Ser618 in the nucleotide binding loop. This residue is not conserved between kinases. This may explain the better FLT3 selectivity of PKC412 in cell proliferation assays compared to CEP701. The similar EC50 values for PKC412 in cell proliferation and autophosphorylation assays also indicate that PKC412 has some FLT3 selectivity. In other words, interaction with the phosphate binding region in the ATP-binding pocket increased potency of CEP701 at the cost of its selectivity for FLT3 kinase.

5. Conclusions This study showed that docking with GOLD was able to differentiate FLT3 inhibitors from ineffective compounds, and provided proper interpretation of results based on mainly the number of effective interactions, rather than solely docking energies. GOLD could distinguish between more potent inhibitors of FLT3 autophosphorylation, CEP701 and PKC412, than the less potent sunitinib in terms of docking score. The relationships between structures and biological activities derived in this study will be useful in designing more potent and selective drugs for treatment of diseases not only associated with FLT3, but also other kinases.

Page 21 of 33

Acknowledgements Novartis kindly provided the compounds PKC412 and imatinib. The experimental part of the project was funded by the Anthony Rothe Memorial Trust and an NHMRC Principal Research Fellowship (LKA). The in silico analysis of this project was funded by the grant number 921682 from Mashhad University of Medical Sciences. References:

Andersson, S.E., Svensson, M.N., Erlandsson, M.C., Dehlin, M., Andersson, K.M., Bokarewa, M.I., 2012. Activation of Fms-like tyrosine kinase 3 signaling enhances survivin expression in a mouse model of rheumatoid arthritis. PLoS One 7, e47668. Armstrong, S.A., Kung, A.L., Mabon, M.E., Silverman, L.B., Stam, R.W., Den Boer, M.L., Pieters, R., Kersey, J.H., Sallan, S.E., Fletcher, J.A., Golub, T.R., Griffin, J.D., Korsmeyer, S.J., 2003. Inhibition of FLT3 in MLL. Validation of a therapeutic target identified by gene expression based classification. Cancer Cell 3, 173-183. Armstrong, S.A., Mabon, M.E., Silverman, L.B., Li, A., Gribben, J.G., Fox, E.A., Sallan, S.E., Korsmeyer, S.J., 2004. FLT3 mutations in childhood acute lymphoblastic leukemia. Blood 103, 3544-3546. Carow, C.E., Levenstein, M., Kaufmann, S.H., Chen, J., Amin, S., Rockwell, P., Witte, L., Borowitz, M.J., Civin, C.I., Small, D., 1996. Expression of the hematopoietic growth factor receptor FLT3 (STK-1/Flk2) in human leukemias. Blood 87, 1089-1096. Cleveland, J.L., Dean, M., Wang, J.Y., Hedge, A.M., Ihle, J.N., Rapp, U.R., 1988. Abrogation of IL-3 dependence of myeloid FDC-P1 cells by tyrosine kinase oncogenes is associated with induction of c-myc. Curr. Top. Microbiol. Immunol. 141, 300-309. Cowan-Jacob, S.W., Mobitz, H., Fabbro, D., 2009. Structural biology contributions to tyrosine kinase drug discovery. Curr. Opin. Cell Biol. 21, 280-287. Daver, N., Strati, P., Jabbour, E., Kadia, T., Luthra, R., Wang, S., Patel, K., Ravandi, F., Cortes, J., Qin Dong, X., Kantarjian, H., Garcia-Manero, G., 2013. FLT3 mutations in myelodysplastic syndrome and chronic myelomonocytic leukemia. Am. J. Hematol. 88, 56-59. Dexter, T.M., Garland, J., Scott, D., Scolnick, E., Metcalf, D., 1980. Growth of factordependent hemopoietic precursor cell lines. J. Exp. Med. 152, 1036-1047. Eriksson, A., Hermanson, M., Wickstrom, M., Lindhagen, E., Ekholm, C., Jenmalm Jensen, A., Lothgren, A., Lehmann, F., Larsson, R., Parrow, V., Hoglund, M., 2012. The novel tyrosine kinase inhibitor AKN-028 has significant antileukemic activity in cell lines and primary cultures of acute myeloid leukemia. Blood Cancer J 2, e81. Fabbro, D., Ruetz, S., Bodis, S., Pruschy, M., Csermak, K., Man, A., Campochiaro, P., Wood, J., O'Reilly, T., Meyer, T., 2000. PKC412--a protein kinase inhibitor with a broad therapeutic potential. Anticancer Drug Des. 15, 17-28. Page 22 of 33

Fletcher, G.C., Brokx, R.D., Denny, T.A., Hembrough, T.A., Plum, S.M., Fogler, W.E., Sidor, C.F., Bray, M.R., 2011. ENMD-2076 is an orally active kinase inhibitor with antiangiogenic and antiproliferative mechanisms of action. Mol. Cancer Ther. 10, 126137. Foster, R., Griffith, R., Ferrao, P., Ashman, L., 2004. Molecular basis of the constitutive activity and STI571 resistance of Asp816Val mutant KIT receptor tyrosine kinase. J. Mol. Graph. Model. 23, 139-152. Frost, M.J., Ferrao, P.T., Hughes, T.P., Ashman, L.K., 2002. Juxtamembrane mutant V560GKit is more sensitive to Imatinib (STI571) compared with wild-type c-kit whereas the kinase domain mutant D816VKit is resistant. Mol. Cancer Ther. 1, 1115-1124. Gajiwala, K.S., Wu, J.C., Christensen, J., Deshmukh, G.D., Diehl, W., DiNitto, J.P., English, J.M., Greig, M.J., He, Y.A., Jacques, S.L., Lunney, E.A., McTigue, M., Molina, D., Quenzer, T., Wells, P.A., Yu, X., Zhang, Y., Zou, A., Emmett, M.R., Marshall, A.G., Zhang, H.M., Demetri, G.D., 2009. KIT kinase mutants show unique mechanisms of drug resistance to imatinib and sunitinib in gastrointestinal stromal tumor patients. Proc. Natl. Acad. Sci. U. S. A. 106, 1542-1547. Gotze, K.S., Ramirez, M., Tabor, K., Small, D., Matthews, W., Civin, C.I., 1998. Flt3high and Flt3low CD34+ progenitor cells isolated from human bone marrow are functionally distinct. Blood 91, 1947-1958. Griffin, J.D., 2001. Point mutations in the FLT3 gene in AML. Blood 97, 2193A-2193. Griffith, J., Black, J., Faerman, C., Swenson, L., Wynn, M., Lu, F., Lippke, J., Saxena, K., 2004. The structural basis for autoinhibition of FLT3 by the juxtamembrane domain. Mol. Cell 13, 169-178. Guex, N., Peitsch, M.C., 1997. SWISS-MODEL and the Swiss-PdbViewer: an environment for comparative protein modeling. Electrophoresis 18, 2714-2723. Kampa-Schittenhelm, K.M., Heinrich, M.C., Akmut, F., Dohner, H., Dohner, K., Schittenhelm, M.M., 2013. Quizartinib (AC220) is a potent second generation class III tyrosine kinase inhibitor that displays a distinct inhibition profile against mutant-FLT3, PDGFRA and -KIT isoforms. Mol. Cancer 12, 19. Levis, M., Small, D., 2003. FLT3: ITDoes matter in leukemia. Leukemia 17, 1738-1752. Levis, M., Small, D., 2004. Small molecule FLT3 tyrosine kinase inhibitors. Curr. Pharm. Des. 10, 1183-1193. Lin, W.H., Jiaang, W.T., Chen, C.W., Yen, K.J., Hsieh, S.Y., Yen, S.C., Chen, C.P., Chang, K.Y., Chang, C.Y., Chang, T.Y., Huang, Y.L., Yeh, T.K., Chao, Y.S., Chen, C.T., Hsu, J.T., 2012. BPR1J-097, a novel FLT3 kinase inhibitor, exerts potent inhibitory activity against AML. Br. J. Cancer 106, 475-481. Mashkani, B., Griffith, R., Ashman, L., 2014. Differences in growth promotion, drug response and intracellular protein trafficking of FLT3 mutants. Iran J Basic Med Sci 17, 867-873. Mashkani, B., Griffith, R., Ashman, L.K., 2010. Colony stimulating factor-1 receptor as a target for small molecule inhibitors. Bioorg. Med. Chem. 18, 1789-1797.

Page 23 of 33

Mashkani, B., Odell, A.F., Byrnes, E.M., Griffith, R., Ashman, L.K., 2013. Expression of biologically active human colony stimulating factor-1 in Pichia pastoris. Protein Expr. Purif. 88, 93-97. Meshinchi, S., Appelbaum, F.R., 2009. Structural and functional alterations of FLT3 in acute myeloid leukemia. Clin. Cancer Res. 15, 4263-4269. Mol, C.D., Dougan, D.R., Schneider, T.R., Skene, R.J., Kraus, M.L., Scheibe, D.N., Snell, G.P., Zou, H., Sang, B.C., Wilson, K.P., 2004. Structural basis for the autoinhibition and STI-571 inhibition of c-Kit tyrosine kinase. J. Biol. Chem. 279, 31655-31663. Mol, C.D., Lim, K.B., Sridhar, V., Zou, H., Chien, E.Y., Sang, B.C., Nowakowski, J., Kassel, D.B., Cronin, C.N., McRee, D.E., 2003. Structure of a c-kit product complex reveals the basis for kinase transactivation. J. Biol. Chem. 278, 31461-31464. O'Hare, T., Shakespeare, W.C., Zhu, X., Eide, C.A., Rivera, V.M., Wang, F., Adrian, L.T., Zhou, T., Huang, W.S., Xu, Q., Metcalf, C.A., 3rd, Tyner, J.W., Loriaux, M.M., Corbin, A.S., Wardwell, S., Ning, Y., Keats, J.A., Wang, Y., Sundaramoorthi, R., Thomas, M., Zhou, D., Snodgrass, J., Commodore, L., Sawyer, T.K., Dalgarno, D.C., Deininger, M.W., Druker, B.J., Clackson, T., 2009. AP24534, a pan-BCR-ABL inhibitor for chronic myeloid leukemia, potently inhibits the T315I mutant and overcomes mutation-based resistance. Cancer Cell 16, 401-412. Roberts, K.G., Odell, A.F., Byrnes, E.M., Baleato, R.M., Griffith, R., Lyons, A.B., Ashman, L.K., 2007. Resistance to c-KIT kinase inhibitors conferred by V654A mutation. Mol. Cancer Ther. 6, 1159-1166. Roskoski, R., Jr., 2007. Sunitinib: a VEGF and PDGF receptor protein kinase and angiogenesis inhibitor. Biochem. Biophys. Res. Commun. 356, 323-328. Rossner, M.T., McArthur, G.A., Allen, J.D., Metcalf, D., 1994. Fms-like tyrosine kinase 3 catalytic domain can transduce a proliferative signal in FDC-P1 cells that is qualitatively similar to the signal delivered by c-Fms. Cell Growth Differ. 5, 549-555. Schubert, C., Schalk-Hihi, C., Struble, G.T., Ma, H.C., Petrounia, I.P., Brandt, B., Deckman, I.C., Patch, R.J., Player, M.R., Spurlino, J.C., Springer, B.A., 2007. Crystal structure of the tyrosine kinase domain of colony-stimulating factor-1 receptor (cFMS) in complex with two inhibitors. J. Biol. Chem. 282, 4094-4101. Shih, L., Huang, C., Wang, P., Wu, J., Lin, T., Dunn, P., Kuo, M., 2004. Acquisition of FLT3 or N-ras mutations is frequently associated with progression of myelodysplastic syndrome to acute myeloid leukemia. Leukemia 18, 466-475. Singh, P., Hoggatt, J., Hu, P., Speth, J.M., Fukuda, S., Breyer, R.M., Pelus, L.M., 2012. Blockade of prostaglandin E2 signaling through EP1 and EP3 receptors attenuates Flt3Ldependent dendritic cell development from hematopoietic progenitor cells. Blood 119, 1671-1682. Smith, B.D., Levis, M., Beran, M., Giles, F., Kantarjian, H., Berg, K., Murphy, K.M., Dauses, T., Allebach, J., Small, D., 2004. Single-agent CEP-701, a novel FLT3 inhibitor, shows biologic and clinical activity in patients with relapsed or refractory acute myeloid leukemia. Blood 103, 3669-3676. Page 24 of 33

Tamborini, E., Pricl, S., Negri, T., Lagonigro, M.S., Miselli, F., Greco, A., Gronchi, A., Casali, P.G., Ferrone, M., Fermeglia, M., Carbone, A., Pierotti, M.A., Pilotti, S., 2006. Functional analyses and molecular modeling of two c-Kit mutations responsible for imatinib secondary resistance in GIST patients. Oncogene 25, 6140-6146. Tiesmeier, J., Muller-Tidow, C., Westermann, A., Czwalinna, A., Hoffmann, M., Krauter, J., Heil, G., Ganser, A., Serve, H., Verbeek, W., 2004. Evolution of FLT3-ITD and D835 activating point mutations in relapsing acute myeloid leukemia and response to salvage therapy. Leuk. Res. 28, 1069-1074. Tokarski, J.S., Newitt, J.A., Chang, C.Y., Cheng, J.D., Wittekind, M., Kiefer, S.E., Kish, K., Lee, F.Y., Borzillerri, R., Lombardo, L.J., Xie, D., Zhang, Y., Klei, H.E., 2006. The structure of Dasatinib (BMS-354825) bound to activated ABL kinase domain elucidates its inhibitory activity against imatinib-resistant ABL mutants. Cancer Res. 66, 57905797. Ustun, C., DeRemer, D.L., Jillella, A.P., Bhalla, K.N., 2009. Investigational drugs targeting FLT3 for leukemia. Expert Opin Investig Drugs 18, 1445-1456. Weisberg, E., Boulton, C., Kelly, L.M., Manley, P., Fabbro, D., Meyer, T., Gilliland, D.G., Griffin, J.D., 2002. Inhibition of mutant FLT3 receptors in leukemia cells by the small molecule tyrosine kinase inhibitor PKC412. Cancer Cell 1, 433-443. Whartenby, K.A., Calabresi, P.A., McCadden, E., Nguyen, B., Kardian, D., Wang, T., Mosse, C., Pardoll, D.M., Small, D., 2005. Inhibition of FLT3 signaling targets DCs to ameliorate autoimmune disease. Proc. Natl. Acad. Sci. U. S. A. 102, 16741-16746. Whartenby, K.A., Small, D., Calabresi, P.A., 2008. FLT3 inhibitors for the treatment of autoimmune disease. Expert Opin Investig Drugs 17, 1685-1692.

Fig. 1. Structures of small molecule inhibitors used in this study as well as ADP and the arylamide molecule. The ‘tail’ highlighted in sunitinib which is not resolved in the crystal structure 3G0E, is indicated by a circle. Fig. 2. Space filling representation of residues interfering with ligand docking into the 1RJB structure. 1RJB was superimposed on 1PKG, 1T46 and 2I1M and the steric clashes between the ligands in the crystal structures and FLT3 residues were detected using DSV. Carbon, nitrogen, oxygen and hydrogen atoms are colored in grey, blue, red, and white, respectively. Fig. 3. CEP701 docked into FLT3Model1PKG. Docking was performed using GOLD and the exported data analysed using Accelrys DSV. H-bonds, hydrophobic and π-interactions are shown Page 25 of 33

with green and orange lines, respectively. The direction of arrows for H-bonds indicates the donor molecule. Nitrogen, oxygen, carbon and hydrogen atoms are shown in blue, red, black and white colours, respectively. Non-polar hydrogens were removed from the CEP701 molecule for clarity. Fig. 4. PKC412 docked into FLT3Model1PKG.The circle shows the OH group blocked by a methyl group. Details of the docking and data analysis procedure were as described for Fig. 3. Fig. 5. Sunitinib docked into FLT3Model1T46. (A) Interactions of docked sunitinib with FLT3 residues in FLT3Model1T46. (B) FLT3Model1T46 with docked sunitinib superimposed on 3G0E to compare the position of sunitinib in FLT3Model1T46 (yellow) with 3G0E (red). The solvent accessible surface of the atoms on the tail is represented by a blue halo around the atom in panel A. Details of the docking and data analysis procedure were as described for Fig. 3. Fig. 6. Clashes between sunitinib and FLT3Model3G0E residues indicated by red dashed lines. FLT3Model3G0E was superimposed on FLT3Model1T46 containing docked sunitinib and clashes were detected using DSV. The residues from FLT3Model1T46 were turned off for clarity. Fig. 7. Superimposing the best docked poses of CEP701 (green), PKC412 (red) and sunitinib (black). The important residues in ligand binding are illustrated as sticks. The hinge regions from FLT3 models built based on 1PKG and 1T46 are shown as solid ribbons. The docking procedure and analysis of the results were as described for Fig. 3. Fig. 8. Inhibition of FD-FLT3-WT cell proliferation by small molecule kinase inhibitors. Cells were cultured in the presence of the indicated concentrations of drugs with either GM-CSF or FLT3 ligand as growth factor. After 48h the relative number of viable cells per well was Page 26 of 33

determined based on fluorescence of resorufin produced by reduction of resazurin. Data points are shown as mean ± S.D. of four replicate wells. The figure shows a representative of three independent experiments. Fig. 9. Inhibition of FD-FLT3-ITD cell proliferation by small molecule kinase inhibitors. Details of experiment and data analysis are as explained for Fig. 8. Fig. 10. Inhibition of FLT3-WT autophosphorylation by small molecule inhibitors. FD-FLT3WT cells were treated with drugs for 30 min then pulsed with FLT3L for 5 min on ice and lysed. Whole cell lysates were separated by PAGE, and then FLT3 and phosphotyrosine (pY) were detected by Western blot. Samples treated with CEP701 and dasatinib were run on one gel, imatinib and PKC412 on another one and sunitinib in the third gel. The figure was prepared by aligning all three figures based on position of molecular weight markers on all the gels. In this study, only intact forms of FLT3 (160 and 130 kD) were analysed and degraded forms with molecular weight less than 130 kD were not included in the final analysis. The intensity of each phosphorylation band was firstly normalised to total FLT3 of the same sample and then the phosphorylation of the lowest drug concentration (as it is very close to zero compared to the highest concentration of each drug) was used to calculate pY% for each sample. EC50 values were calculated by plotting percentage of autophosphorylation inhibition against log10 of drug concentrations and fitting a linear regression curve using GraphPad Prism software. The fully glycosylated (160 kDa) and immature (130 kDa) forms of FLT3 are indicated with arrows. C-/shows the control sample neither treated with drugs nor stimulated with FLT3L. C+/- shows the cell lysate of FD-FLT3-WT cells without drug treatment, but stimulated with FLT3L. This figure is a representative of three independent repeats of this experiment with similar results.

Page 27 of 33

Fig. 11. Inhibition of FLT3-ITD kinase domain phosphorylation by small molecule inhibitors. FD-FLT3-ITD cells were treated and lysed as described for FD-FLT3-WT cells in Fig. 10, and then the FLT3 protein was immunoprecipitated from the cell lysates prior to electrophoresis. The fully glycosylated (160 kDa) and immature (130 kDa) forms of FLT3 are indicated with arrows. Details of figure preparation and Western blot experiment and data analysis are as explained for Fig. 10.

Page 28 of 33

Page 29 of 33

Page 30 of 33

Figure 6:

Page 31 of 33

Figure 7

Page 32 of 33

Page 33 of 33