Preclinical validation of a novel compound targeting p70S6 kinase in breast cancer

Generation of FS-115, a potent, selective and orally bioavailable inhibitor of p70S6K1

We recently published that inhibition of p70S6K1 activity strongly impaired local relapse in a mouse model of breast cancer recurrence [9]. In order to develop candidate drugs for clinical trials and translate our results from the bench to the bedside, we have tested a novel inhibitor of p70S6K1. We particularly looked to compounds that were potent, selective, cell penetrant and orally bioavailable. Among the different compounds, FS-115 exhibited particularly high selectivity and potency and was thus selected for further analyses. In particular, it blocked the kinase activity of p70S6K1, in vitro, with an IC50 of 0.035μM, as determined by using a radiometric ATP-competitive kinase assay (Fig. 1A). Importantly, it only marginally inhibited the activity of p70S6K2, a closely-related homologue of p70S6K1 (IC50 of 2.06μM) and AKT2 (IC50 of 23.8μM) (Fig. 1B-C). FS-115 also inhibited some tyrosine kinases of the Src family and PRK2 (protein kinase C-related protein kinase-2) and ROCK-II (Rho-associated protein kinase II), while it was substantially inactive towards other kinases of the PI3K pathway (Fig. 1D and Supplementary Table 1).

FS-115 efficiently suppresses p70S6K activity in breast cancer cells

Our previous studies demonstrated that wound fluids, drained from BC patients for 24 hours post-surgery (hereafter WF), are very rich in cytokines and growth factors and very efficiently stimulate proliferation, motility and invasion and that activation of p70S6K signaling is a common event following exposure of breast cancer cells to WF [10,24].

Thus, we tested the ability of FS-115 to counteract the potent activation of the p70S6K signaling induced in time course experiments of breast cancer cells stimulated with WF. To this aim, we serum starved MDA-MB-231 cells (triple negative subtype), pre-treated for 30 minutes with different doses of FS-115 and then stimulated with WF for 20 minutes, a condition in which p70S6K is highly activated [10]. Western Blot analyses of phosphorylated ribosomal protein S6 (pSer235/236- and pSer240/244-RPS6), the direct downstream substrate of p70S6K1, indicated that, following WF stimulation, FS-115 efficiently reduced p70S6K activity already when used at 10μM, at lower but comparable levels to what was observed using PF-4708671, a commercially available p70S6K inhibitor, used in most of the following experiments as control (Fig. 2A). Since it is known that prolonged treatment with mTOR inhibitors can lead to hyper activation of AKT, due to suppression of negative feedback loop mechanism [4], we checked the effect of FS-115 on this pathway. We noticed that FS-115 elicited a partial activation of AKT, but only when given at the highest dose (50μM). Similar results were harvested also in MDA-MB-453 (HER2-positive subtype) and MCF-7 (luminal subtype) breast cancer cells (Supplementary Figure 1).

Due to the relative in vitro activity displayed by FS-115 on c-Src, we also tested whether this signaling pathway was altered following treatment. We looked at c-Src expression and phosphorylation and, also, at the phosphorylation of STAT3, cofilin and FAK, three proteins frequently activated downstream of Src, in MDA-MB-231 cells stimulated with WF. However, no consistent effect of FS-115 on Src signaling pathway was detectable in MDA-MB-231 cells (Fig. 2B). In MDA-MB-453 c-Src expression and activation was barely detectable following WF stimulation and not affected by FS-115 (Supplementary Figure 1). On the other hand, in MCF-7 cells activation of c-Src was quite efficiently prevented by FS-115 treatment (Supplementary Figure 1), suggesting that the spectrum of activity for this inhibitor may vary depending from the cancer subtypes.

No effect on other members of p70S6K-related pathway (mTOR, 4EBP1) or on unrelated pathway (MAPK) was observed following FS-115 treatment (Fig. 2A).

Altogether, these experiments indicate that, following strong activating stimuli, FS-115 treatment efficiently and specifically suppresses p70S6K activity in breast cancer cells.

FS-115 strongly impacts on the survival and tumorigenic potential of breast cancer cells

Our recent data suggested that an intact p70S6K signaling is important for cell survival, particularly when cells are challenged by “stringent” microenvironmental conditions. From previous studies, we know that MDA-MB-231 cells grow very well also in stringent culture conditions, such as plating at very low density, but robust p70S6K activity is necessary for their survival and for colony formation [10]. Colony formation assay on cells plated at a very low density (1×10³/100mm dish) demonstrated that treatment with FS-115 was highly potent in suppressing the ability of MDA-MB-231 cells to survive and form colonies in such condition, even when added at very low concentrations (Fig. 2C and D). Thus, FS-115 treatment was capable to suppress ability of breast cancer cells to survive to harsh environment, such as when they are challenged to grow as isolated colonies.

Next, we tested anchorage independent growth in soft agar, in the presence of WF as external source of growth stimuli, with or without FS-115 or PF-4708671, at different doses. Impairment of p70S6K activity, by either compounds, significantly decreased the cell ability to survive and grow in anchorage independence (Fig. 2E and F). Both number and size of the colonies were affected, signifying that, in this context, activity of p70S6K was necessary for both survival and proliferation of breast cancer cells (Fig. 2E and F). This result again supported that, under stringent conditions, cells rely on a robust p70S6K signaling for their survival and that FS-115 efficiently counteracts this process.

Pharmacokinetic and pharmacodynamic profile of FS-115

Altogether, the above experiments supported that FS-115 was a promising compound, since it not only significantly and specifically blocked p70S6K activity but also well recapitulated our previous findings on the relevance of p70S6K activity in breast cancer cells. Given the good performance of FS-115 in vitro, we decided to evaluate the potential of this new compound in vivo.

We first tested the pharmacokinetic profile of FS-115 in male CD-1 mice, looking at profiles from both intravenous (i.v.) and oral (p.o.) routes of administration. FS-115 was formulated using a solution comprised of 3% DMSO, 0.1M aqueous HCl (stoichiometric amount) and a 20% w/v aqueous solution of β-cyclodextrin (dose volume 5 ml/kg). Following a single dose of FS-115 at 10 mg/kg p.o. and 2 mg/kg i.v., plasma was sampled at 8 time points (n=3 mice per time point, terminal sampling) over 24 hours, followed by protein precipitation using methanol and analysis by LC-MS/MS. The plasma profiles are shown in Figure 3A. FS-115 was determined to exhibit an oral bioavailability in excess of 95%, an i.v. half-life of 0.83 hr, clearance of 25 ml/min/kg and a volume of distribution of 1.64 L. We next tested the brain-penetrance of FS-115 by a separate discrete oral pharmacokinetic experiment in male CD-1 mice. Following a single dose of FS-115 at 25 mg/kg p.o., plasma and brain was sampled at 3 time points (0.5, 1.1 and 2.6 hr), followed by protein precipitation using acetonitrile and analysis by LC-MS/MS. FS-115 was found to be highly brain-penetrant, with a ratio of (brain : plasma) of (4.0 : 1) at 2.6 hr (Fig. 3B).

For pharmacodynamic profile, female athymic nude mice were implanted with MDA-MB-231 cells into the fat pad of the thoracic mammary gland. When tumors reached 50-100 mm3, mice were randomly assigned to treatment groups of either vehicle QD x 3 PO or FS-115 125 mg/kg BID x 3 PO (6 doses in total). Twelve hours after the final dose animals were sacrificed, tumors resected and homogenized. Protein extracts were then tested in ELISA assay for the expression of total/phospho-S6 (ser 240/244) and total/phospho-AKT (ser 473). FS-115 treated animals exhibited significantly reduced expression of phospho-S6 compared to vehicle-treated animals (Fig. 3C). It is important to note that the effect of FS-115 was long lasting in vivo, the samples for ELISA having been taken 12 hours following the final dose. Furthermore, there was no effect on phospho-AKT, indicating that a pro-survival response had not been initiated by dosing of FS-115 (Fig. 3C).

Treatment with FS-115 reduces tumor take rate and growth

Based on these encouraging data, we moved to the evaluation of the effects of FS-115 in tumor initiation. To this aim, on day -1 animals were randomly assigned to treatment groups and drug treatment commenced. FS-115 was dosed orally daily (QD) at 250mg/kg or twice a day (BID) at 125mg/kg regimen, for 21 or 30 consecutive days (Fig. 4A and B). On day 0, MDA-MB-231 cells (1×10⁶ in Matrigel) were implanted into the fat pad of the thoracic mammary gland. Measurement of tumor volume was commenced 7 days following implantation. Treatment with FS-115 significantly reduced the take-rate of MDA-MB-231 tumors, with only 40% of animals having a palpable tumor at the end of the study. In contrast 100% of vehicle-treated controls had tumors (Fig. 4B and C). When tumors appeared, a significant reduction in tumor volume was apparent in all treated animals, already from day 9 after implantation (Fig. 4D, compare black line with any colored line from treated animals). Animals from Group 1 (black line, solid circles) converge in a same line with Group 2 (black line, open circles) until day 21, since they are equally treated and then culled. The same is true for animals from Groups 3 (red line, solid circles) and Group 5 (red line, open circles). Although not significantly, mean tumor volume of animals treated with 125mg/kg BID was consistently less than that of animals treated with 250mg/kg QD (Fig. 4D, compare blue and red lines). Group 4 animals (green line) received treatment for 21 days, followed by a recovery period of 7 days. It is interesting to note that during this recovery period tumor growth was faster than that observed in animals from Group 5 (red line) who continued to receive compound (Fig. 4D, compare green and red lines). However, mean tumor weight was significantly different between groups receiving vehicle versus each corresponding treatments (Fig. 4E).

Necropsy of vehicle-treated control animals revealed metastatic spread to lung and pancreatic tissues. Consistently, their lung tissue weighed significantly more than that collected from animals treated with FS-115 at 250mg/kg QD or 125mg/kg BID and animals sacrificed after receiving 30 doses of vehicle had lung tissue heavier than that from animals sacrificed 9 days earlier (mean weight 0.194g versus 0.17g) (Fig. 4F). Nodules on the surface of lungs were visible in 6/10 vehicle-treated animals (with a mean of 3 nodules per animal). In contrast, only 4/20 FS-115 treated animals had lung nodules with a mean of 1 nodule per animal (Fig. 4B and F). This pattern was also seen with pancreatic tissue (Fig. 4H). Liver weights were not significantly different (not shown). Thus, treatment with FS-115 reduced tumor initiation and spontaneous metastatic spreading of MDA-MB-231 cells to distant organs.

Peri-surgical treatment with FS-115 suppresses breast cancer local recurrence

We next evaluated the efficacy of FS-115 in pre-clinical setting, using the mouse model of breast cancer recurrence recently set up in our laboratory [9,10]. Briefly, MDA-MB-231 breast cancer cells were bilaterally injected in nude mice mammary fat pads (MFP). Once grown, primary tumor masses were surgically removed and mice were followed-up for 8 weeks (Fig. 5A). Our previous studies demonstrated that in this context MDA-MB-231 breast cancer cells give rise to locally recurrent disease in 50-70% of cases and that impairment of p70S6K activity, by different means, strongly decreases the rate of recurrence formation of these cells [9,10]. We thus tested the activity of FS-115, treating mice with 125 mg/kg twice a day (BID) x 3, in a peri-surgical treatment schedule (administration at day -1, 0 and +1, respect to surgery time) (Fig. 5A). Although primary tumor masses had grown at similar extent (Fig. 5B, see insets), the evaluation of recurrence formation at long term (8 weeks) indicated that FS-115 was highly effective in reducing the rate of recurrence formation (Fig. 5B) (disease-free survival: 42% vs 84% in CTR vs FS-115-treated mice, p=0.02). Thus, this experiment supported that peri-surgical inhibition of p70S6K1 with FS-115 could represent a promising therapeutic agent to treat breast cancer patients at high risk to develop recurrent disease.

Treatment with FS-115 strongly suppresses metastatic spread of breast cancer cells

Finally, FS-115 was tested for efficacy in a mouse model of metastasis. Female athymic nude mice were intracardially inoculated with luciferase-tagged MDA-MB-231 cells (MDA-MB-231-Luc), to allow for evaluation and treatment of brain metastases. FS-115 was dosed orally QD at 250mg/kg or BID at 125mg/kg regimen, for 21 or 70 days, with dosing commencing the day prior to inoculation (Fig. 5C and D). At the mid-point of study (36 days after inoculation), 80% of vehicle-treated animals had visible metastatic deposits whereas none of compound-treated animals had visible metastatic deposits (Fig. 5E and F; Suppl. Fig. S2A). It is interesting to note that mice treated for 21 days (groups 3 and 5) showed no metastatic deposits, suggesting that a shorter duration of dosing may be sufficient to control metastasis. However, to better evaluate metastatic spread and longer term outcomes, all animals were imaged and sacrificed at day 70 (Fig. 6A and B). To this aim, brain, liver and lungs were removed and imaged individually (Fig. 6C and F; Suppl. Fig. S2B and C). Review of flux readings (which show severity of metastatic burden) showed that 70-days treatment regimen was most effective overall. Comparing groups 3 and 5, BID dosing was more effective than QD dosing and, FS-115 125 mg/kg BID x 70 (group 6) was the most effective regimen, with no metastasis detected in liver or brain and only minimal signal in lungs.

Treatment with FS-115 is well tolerated in mice

Tolerability studies were performed in female athymic nude mice used for the experiments described above (Fig 6A), in which FS-115 was dosed orally QD at 250mg/kg or BID at 125mg/kg regimen, for 70 consecutive days. Bodyweight and general clinical signs were annotated daily and used as a measure of drug tolerance. No animals were sacrificed due to weight loss or illness, while 1 death in each of vehicle groups, due to metastatic burden, occurred (Fig. 6B). Treatment with FS-115 QD at 250mg/kg resulted in significantly decreased bodyweight respect to vehicle, however this never dropped below 10% of the initial (Fig. 6G). Dosing with 125mg/kg BID resulted in significantly less bodyweight loss than 250mg/kg QD (Fig. 6G), suggesting that this regimen might be better tolerated. The vehicle treated animals showed a decline in bodyweight post day 49 (probably concurrently with metastatic spread), while this was not observed for FS-115-treated animals (Fig. 6G). It is to consider that the slow weight gain observed in the FS-115-treated cohort is likely related also to the role of p70S6K1 in controlling organ size, as suggested by the phenotypes observed in p70S6K1 KO mice [25].

Study approval

Animal experiments on tumor local recurrence were reviewed and approved by the CRO of Aviano Institutional Organism For Animal Wellbeing (OPBA) and by the Italian Ministry of Health. Experiments were conducted according to that committee's guidelines. Metastasis, tumor initiation, PK and PD protocols used have been approved by the Axis Bioservices Animal Welfare and Ethical Review Committee, and all procedures were carried out under the guidelines of the Animal (Scientific Procedures) Act 1986. PK studies were performed via two separate external providers, Nikem (IT) and Pharmidex (UK), under approval and in accordance with the internal ethics policy in operation at the time of the study.

Wound Fluids (WF) from breast cancer patients were collected at CRO of Aviano. Scientific use of biological material was approved by the Ethics Committee of the CRO of Aviano, Italy. Specific informed consent was obtained from each patients.

Kinase profile of FS-115

FS-115 ability to block p70S6K1 kinase activity was tested using radiometric ATP-competitive kinase assay (Eurofins Millipore KinaseProfiler™ Service Assay). The broader kinase selectivity profile was determined by counter-screening FS-115 against a panel of 268 kinases, excluding p70S6K1, using the Eurofins Millipore KinaseProfiler™ Service. Full details are available at https://www.eurofins.com/pharma-services/pharma-discovery-services/services/in-vitro-pharmacology/kinases.aspx. In brief, the kinase assay method used is as follows: FS-115 was prepared at 50x the final assay concentration in 100% DMSO. In a final reaction volume of 25 μL, human p70S6K1 protein of 1-421 amino acid length with T412E mutation is incubated with 8 mM MOPS pH 7.0, 0.2 mM EDTA, 100 μM KKRNRTLTV, 10 mM Mg acetate and [γ-33P-ATP] (specific activity approx. 500 cpm/pmol, concentration as required). After incubation for 40 minutes at room temperature, the reaction is stopped by the addition of 5 μL of a 3% phosphoric acid solution. 10 μL of the reaction mixture is then spotted onto a P30 filtermat and washed three times for 5 minutes in 75 mM phosphoric acid and once in methanol prior to drying and scintillation counting. The concentration of ATP used was 15 μM.

Cell culture and development of stable cell lines

MDA-MB-231 (basal, ER-, PR-, HER2-), MDA-MB-453 (luminal, ER-, PR-, HER2-), MCF-7 (luminal, ER+, PR+, HER2-) mammary carcinoma cell lines were obtained from ATCC (LGC Standards) and grown in Dulbecco modified Eagle medium (DMEM, Lonza) supplemented with 10% fetal bovine serum (FBS, SIGMA). All cell lines were authenticated by BMR Genomics srl (Padova, Italy), according to Cell ID ^TM System (Promega) protocol and using Genemapper ID Ver 3.2.1, to identify DNA STR profiles. MDA-MB-231-luciferase-expressing cells (MDA-MB-231-luc) derived from MDA-MB-231 cells originally purchased from ATCC. For in vivo experiments, these cells were stably transfected with pGL4.51 (Promega) using Lipofectamine 2000 (Life Technologies), by Axis Bioservices (Northern Ireland, UK).

Preparation of protein lysates and immunoblotting analysis

MDA-MB-231, MDA-MB-453, MCF-7 mammary carcinoma cell lines were serum starved in DMEM containing 0.1% bovine serum albumin (BSA, SIGMA) and then stimulated with 5% WF for 20 minutes. Where indicated, cells were also pre-treated for 30 minutes with PF-4708671 (p70S6K1 inhibitor, 10 or 50 μM, SIGMA) or FS115 (10 or 50 μM).

To extract total proteins cells were scraped on ice using cold RIPA lysis buffer (1% NP40; 50 mM Tris HCl pH 8; 150 mM NaCl; 0.1% SDS; 0.5% Na-Deoxycholate) plus a protease inhibitor cocktail (Complete™, Roche) and supplemented with 1 mM Na₃VO₄ (SIGMA), 10 mM NaF (SIGMA) and 1 mM DTT (SIGMA).

For immunoblotting analysis, proteins were separated in 4-20% SDS-PAGE (Criterion Precast Gel, Biorad) and transferred to nitrocellulose membranes (GE Healthcare). Membranes were blocked with 5% dried milk in TBS-0.1% Tween20 or in Odyssey Blocking Buffer (Licor, Biosciences) and incubated at 4°C overnight with primary antibodies. Then, membranes were incubated 1 hour at RT with IR-conjugated (Alexa Fluor 680, Invitrogen or IRDye 800, Rockland) secondary antibodies for infrared detection (Odyssey Infrared Detection System, Licor).

Primary antibodies directed against: AKT (sc-1618), ERK1 (sc-94), p70S6K1 (sc-8418), Src (sc-19), STAT3 (sc-482) and vinculin (sc-7694) were purchased from Santa Cruz; pT202/204 ERK1/2 (#9101), pS473 AKT (#4060), pT389 p70S6K1 (#9234), S6 (#2217), pS235/236 S6 (#4858), pS240/244 S6 (#5364), pS2448 mTOR (#2971), mTOR (#2972), pT37/46 4EBP1 (#2855), 4EBP1 (#9644), pY416 Src (#2101), pY705 STAT3 (#9131), pS3 Cofilin (#3311) were purchased from Cell Signaling; Cofilin (#612145), FAK (#610088) were purchased from BD Biosciences; pY397 FAK (#44624G) was purchased from ThermoFisher Scientific; Actin (#A5060) was purchased from SIGMA.

Wound fluid collection. Drainage wound fluids (WF) were collected over the 24hrs after surgery from unselected patients undergone breast-conserving surgery, as described previously [24,37]. The assays were then performed using pools of all fluids.

Colony assay

For colony assay, MDA-MB-231 untreated or pre-treated for 24 hours with PF-4708671 (p70S6K1 inhibitor, 1, 2, 5, 10 μM, SIGMA) or FS-115 (1, 2, 5, 10 μM) were trypsinized, counted and seeded at density 1000 cells/100 mm dish and incubated in complete growth medium in the presence of the inhibitors or left untreated. Two weeks later plates were stained with crystal violet and colonies were counted.

Soft Agar assay

To evaluate the anchorage-independent cell growth, MDA-MB-231 cells (1.5×10⁴) were re-suspended in 2 ml top agar medium (DMEM-10% FBS or SFM-3% WF, 0.4% Low Melting Agarose, SIGMA) in the presence of PF-4708671 (p70S6K1 inhibitor, 10 or 20μM, SIGMA) or FS-115 (10 or 20μM), and quickly overlaid on a previously gelified 0.6% bottom agar medium (DMEM-0.1% BSA, 0.6% LowMelting Agarose, SIGMA). The experiments were performed in six-well tissue culture plates, in triplicate. Fresh medium was added to the wells twice a week as a feeder layer. After three weeks, the number of colonies was counted at 4X magnification.

In vivo experiments

Pharmacokinetics (PK) studies were conducted by Nikem (IT) and Pharmidex (UK), in male CD-1 mice, looking at profiles from both intravenous (i.v.) and oral (p.o.) routes of administration. FS-115 was formulated using a solution comprised of 3% DMSO, 0.1M aqueous HCl (stoichiometric amount) and a 20% w/v aqueous solution of β-cyclodextrin (dose volume 5 ml/kg). Following a single dose of FS-115 at 10 mg/kg p.o. and 2 mg/kg i.v., plasma was sampled at 8 time points (n=3 mice per time point, terminal sampling) over 24 hours, followed by protein precipitation using methanol and analysis by LC-MS/MS. Brain-penetrance of FS-115 was determined by a separate discrete oral PK experiment in male CD-1 mice. Following a single dose of FS-115 at 25mg/kg p.o., plasma and brain was sampled at 3 time points (0.5, 1.1 and 2.6 hr), followed by protein precipitation using acetonitrile and analysis by LC-MS/MS.

Pharmacodynamics (PD) was conducted by Axis Bioservices in 6 female athymic nude mice (Hsd:Athymic Nude-Foxn1^nu) implanted with MDA-MB-231 cells into the thoracic mammary fat pad (1×10⁶ in Matrigel). The mice were randomly assigned to treatment groups of either vehicle QD x 3 PO or FS-115 125 mg/kg BID x 3 PO (6 doses in total) when their tumors reached 50-100 mm3. Vehicle used for both treatment groups was 5% DMSO, 95% aqueous hydroxypropyl-β-cyclodextrin (20% w/v). Twelve hours after the final dose animals were sacrificed and the tumor resected. Tumor tissue was homogenized in three volumes Pathscan Sandwich ELISA lysis buffer (Cell Signaling Technologies). Protein concentration was measured using a commercial BCA kit (Sigma) and ELISAs for phospho-S6 240/244, total-S6, phospho-AKT 473 and total-AKT carried out according to manufacturer's instructions (Cell Signaling Technologies).

For local recurrence studies, primary tumors were established by injection of 2×10⁶ MDA-MB-231 control bilaterally in the fat pads of the thoracic mammary glands of female nude mice (Hsd:Athymic Nude-Foxn1nu, aged 6-8 weeks, Harlan). After the appearance of palpable primary tumors, animals were orally treated with FS-115 (125mg/kg) twice a day (BID) in a peri-surgical treatment schedule (day -1, day 0 and day +1 respect to surgery). Mice were sacrificed at the end point of the experiment (8 weeks from surgery) unless they displayed suffering behavior, in adherence with ethical guidelines of CRO of Aviano Institutional Organism For Animal Wellbeing (OPBA). Tumors, recurrences, lymphnodes and other organs were collected and stored for subsequent analyses.

Tumor initiation, tumor growth and metastasis studies were conducted by Axis Bioservices (Northern Ireland, UK) in female athymic nude mice (Hsd:Athymic Nude-Foxn1^nu), aged 6-8 weeks.

For metastasis study, animals were initially assigned to treatment groups. On day 0, 1×10⁵ MDA-MB-231-luc cells were diluted in 100μl HBSS and loaded into a 27-gauge needle for intracardiac injection. The following day, animals were first injected with D-luciferin (i.p. at 150mg/kg) and then imaged using a Xenogen IVIS machine. Evidence of light only in the cardiac area is indicative of a failed implantation and so any such animals were discarded from the study. FS-115 (or vehicle) was dosed orally via QD (at 250mg/kg) or BID (at 125mg/kg) regimen, with dosing commencing the day prior to inoculation (day -1) and ending 70 days after inoculation. Animals were imaged on 2 occasions during the study, on day 38 from commencement of dosing (mid-point) and on day 70 (end of study).

For tumor initiation and growth studies, female nude mice were randomly assigned to treatment groups: FS-115 (or vehicle) was dosed orally via QD (at 250mg/kg) or BID (at 125mg/kg) regimen, with dosing commencing the day prior to inoculation (day -1) and ending 21 or 30 days after cell implantation. MDA-MB-231 cells (1×10⁶ in matrigel) were implanted into the second mammary fat pad, on day 0. Time to appearance and tumor volume were estimated respectively by palpation and caliper measurements.

In all in vivo experiments, dosing volume of FS-115 was 10 ml/kg and formulation was 5% DMSO, 95% aqueous hydroxypropyl-β-cyclodextrin (20% w/v). Dosing formulation was made up fresh daily. Bodyweight and general clinical signs (e.g. starred fur, lack of movement, difficulty breathing) were annotated daily and used as a measure of drug tolerance.

Statistical analysis

The IC50 graphs were plotted using XLFit version 5.3 (ID Business Solutions). Sigmoidal dose-response (variable slope) curves are fitted using non-linear regression analysis. The results are presented as the percentage of kinase activity relative to the DMSO-treated control. Data were examined using the two-tailed Student t test or unpaired two-tailed Mann-Whitney U test. Differences were considered significant at p < 0.05. The computer software PRISM (version 4, GraphPad, Inc.) was used to make graphs and all statistical analyses, unless otherwise specified.