Synthesis of fadraciclib, 2-pentanol, 3-[[6-[[(4,6-dimethyl-3-pyridinyl)methyl]amino]-9-(1-methylethyl)-9Hpurin-2-yl]amino]-, (2R,3S)

To a stirred solution of (4,6-dimethyl-pyridin-3-ylmethyl)-(2-fluoro-9-isopropyl-9H-purin-6-yl)-amine (300 mg 0.84mmol) in n-BuOH / DMSO (5 ml, 4:1) at room temperature under an argon atmosphere, DIEA (1.7 ml, 10 eq, 8.4 mmol) was added followed by (2R,3S)-3-amino-pentan-2-ol (0.5g, 4.8 mmol). The flask was fitted with a condenser and the reaction mixture was placed in a preheated oil bath at 140°C and stirred at this temperature for 72 h. The reaction mixture was allowed to cool to room temperature and the solvent was evaporated in vacuo. The residue was partitioned between ethyl acetate (50 ml) and water (50 ml), the aqueous phase was extracted with more EtOAc (2 x 25 ml) and combined organic phase was washed with brine (50 ml), dried (MgSO₄) and evaporated in vacuo. The residue was purified by flash gradient column chromatography on silica gel eluted with CHCl₃: MeOH (100: 0 →95: 5), to afford 55 mg of pure title compound. For further details on the synthesis of the series, refer to Wilson et al. [32].

NMR spectrum data: δ_H (250 MHz, CDCl₃) 095 (3, H, t J 7.5, CHCH₂CH₃), 1.06 (3 H, d, J 7.5, CHCH₃OH) 1.48 [6 H, d, J 7.5 CH(CH₃)₂], 2.24 (3 H, s, CH₃), 2.4 (3 H, s, CH3), 3.92–3.82 (2 H, m, NHCH₂), 4.67–4.45 (3 H, m, CHEtCHMeOH), 6.16 (1 H, s, br, NH), 6.87 (1 H, s ArH), 7.37 (1 H, ArH), 8.31 (1 H, s, ArH); δ_C (250 MHz, CDCl₃) 160.11 (C), 157.68 (C), 154.57 (C), 149.42 (CH), 146.38 (C), 134.54 (CH), 129.24 (C), 124.84 (CH), 71.52 (CH), 59.65 (CH), 40.33 (CH₂), 24.94 (CH₂), 23.89 (CH₃), 23.52 (2 x CH₃), 1737 (CH₃), 12.57 (CH₃); m/z 398 (M + H)

Determination of kinase inhibitory potency and selectivity in vitro using recombinant enzymes

Kinase assays for CDK2, CDK4, CDK7 and CDK9 were carried out as described previously using recombinant CDK/cyclins generated at Cyclacel Ltd, Dundee, UK [29,34,35].

Profiling of 1 μM fadraciclib against a 256-kinase selectivity panel was performed at Carna Biosciences. Enzymes that inhibited >50% at the tested concentration were selected for IC₅₀ determination. IC₅₀ values were determined from a 10-point concentration curve using ATP concentrations approximating K_m values. The value obtained for the reaction control (complete reaction mixture) was set as 0% inhibition and for the background (no enzyme) as 100% inhibition. The percentage inhibition of each test solution was calculated, which allowed the IC₅₀ value to be determined from concentration versus percentage inhibition curves by fitting to a four parameter logistic curve.

Mining publicly available data for the dependence of cancer cell lines on CDKs inhibited by fadraciclib

We analyzed data available from the Cancer Dependency Map to determine the effects of CRISPR-Cas9 knockdown of CDKs inhibited by fadraciclib on the viability of >700 cancer cell lines, including those used in the present study. Results were shown as a CERES gene effect that scores the Dependency of cell lines on a given gene Score ((https://depmap.org/portal/).

Cell lines and tissue culture conditions

The work presented in this report was performed on a panel of human cancer cell lines, including Colo205, TNBC, AML and several cell lines of non-malignant origin. A list of the cell lines used in the project, along with a brief description and culture conditions, is presented in S1 Table.

Colo205 was derived from a colorectal carcinoma and obtained from ECACC. The breast cell line panel consisted of lines derived from two TNBC tumors and a HER2-amplified breast tumor. In addition, two immortalized cell lines were derived from healthy, non-malignant breast tissue. All breast panel cell lines were obtained from ATCC, with the exception of Cal51 which was purchased from DSMZ.

Seven of the thirteen AML cell lines contain MLL rearrangements (MLLr) or partial tandem duplications (PTD): EOL-1 (MLL-PTD), ML-2 (MLL-AF6), MOLM-13 (MLL-AF9), MV4-11 (MLL-AF4), NOMO-1 (MLL-AF9), OCI-AML2 (MLL-PTD), THP-1 (MLL-AF9) and 6 contain no MLL abnormalities: HEL, HL60, Kasumi-1, KG-1, OCI-AML5 and PL-2. All AML cell lines were purchased from DSMZ with the exception of THP-1 and HL-60, which were sourced from ECACC and ATCC, respectively.

All cell lines were grown in tissue culture medium recommended by the supplier in incubators at 37°C and 5% CO₂. Information on the cell lines including their classification, origin, supplier, doubling time, tissue culture media, MLL gene status and other molecular and genetic characteristics are summarized in S1 Table.

Determination of cell viability using the resazurin reduction assay

Solid tumor cell lines were seeded in 100 μl medium into 96-well plates at an appropriate cell number determined by a cell dilution assay (Cal51–2000 cells/well; Colo205, MCF10A and 184A1–3000 cells/well and MDA-MB-468, HCC1954–4000 cells/well). AML suspension cells were seeded in 96-well plates at approximately 3000 cells per well. Cells were incubated overnight at 37°C, 5% CO₂. The following day, 100 μl fresh medium containing increasing concentrations of appropriate compound was added and the cells were incubated for up to 72 h at 37°C, 5% CO₂. For pulse treatment of solid (adherent) tumor cell lines, medium was aspirated and replaced with fresh drug-free medium. For pulse treatment of suspension lines, plates were centrifuged at 478 xg for 5 min at the end of the incubation period with compound, medium containing compound was carefully aspirated and replaced with fresh compound-free medium. A total of 72 h after the start of treatment, 20 μl of a 10X stock of resazurin sodium salt (Sigma-Aldrich, Dorset, UK) prepared in medium was added to the cells and incubated for up to 3 h. Absorbance was measured on the Wallac Victor plate reader (Cambridge, UK) at 544–595 nm. IC_50/70/90 calculations were determined in XLFit.

Cell viability analysis by flow cytometry

AML cells were seeded at 5 x 10⁵ cells/ml in a 6-well plate and incubated with the test compound or vehicle control (DMSO) for 6, 8 or 10 h. At the end of the incubation period, cells were collected into Falcon tubes, pelleted via centrifugation (5 min at 478 x g), and medium containing compound was aspirated. The cells were resuspended in 4 ml of fresh compound-free medium and returned to the 6-well plates for further incubation. Following compound addition, 100 μl (at 24 h), 75 μl (at 48 h), 50 μl (at 72 h) or 30 μl (at 96 h) of a homogeneous suspension of cells were removed from the culture, made up to 100 μl with medium and mixed with 100 μl Guava Viacount reagent (Millipore, Nottingham, UK) in a 96-well plate. Samples were incubated for 20 min in the dark, mixed and then analyzed by GuavaSoft 2.6 software on the Guava Easycyte 8HT (Millipore). Guava Viacount software was used to gate and calculate the proportion of viable and dead cells in the population.

Cell cycle analysis by flow cytometry

Colo205 cells were seeded at 8 x 10⁵ cells per 10 cm dish in a total of 9 ml medium and incubated overnight at 37°C, 5% CO_2. The following day 1 ml of medium containing 10x test agent or vehicle was added to the cells, which were then incubated for 24 h at 37°C, 5% CO_2. Cells were collected, prepared, propidium iodide stained for DNA content analysis, and analyzed as described previously [29]. Analysis was performed using BD Cell Quest 3.3 software on a BD LSR flow cytometer.

TNBC cell lines were seeded at 1 million cells per 10 cm dish in a total of 9 ml medium and incubated overnight at 37°C, 5% CO_2. The following day 1 ml of 10x stock of fadraciclib or fresh medium were added to the cells for an 8 h pulse before the medium was removed and replaced with fresh compound-free medium. Cells were incubated for a further 16 h at 37°C, 5% CO₂ before samples were collected in 50 ml Falcon tubes (including all cells in suspension in original medium, cells collected from the PBS wash and adherent cells collected from trypsinisation). The pooled collected cells were washed once with PBS, followed by fixation in ice-cold 70% ethanol overnight at -20°C. 5 x 10⁵ cells, as recommended by the cell cycle kit instructions, were washed once in 1 ml PBS, centrifuged at 200 x g for 5 min then resuspended in PI solution from Guava cell cycle reagent kit from Millipore (4500–0220) for 30 min before analysis by GuavaSoft 2.6 software on a Guava Easycyte 8HT.

AML cells were seeded in a 6-well plate at 5 x 10⁵ cells/ml and incubated with test compound for 16 h. Cells were collected and washed once with PBS, followed by fixation in ice-cold 70% ethanol overnight at -20°C. Samples were prepared for propidium iodide staining of DNA to determine the proportion of cells in different cell cycle phases within the population as per the Guava Cell Cycle Reagent manufacturer’s instructions. Briefly, cells were washed once in PBS and then resuspended in 100 μl PBS. An equivalent volume of Guava Cell Cycle Reagent was added to the cells and incubated for 20 min in the dark. The stained cell suspension was then resuspended and analyzed by GuavaSoft 2.6 software on a Guava Easycyte 8HT.

Quantitative Real-Time (RT) PCR analysis

AML cells were seeded at 5 x 10⁵ cells/ml in appropriate medium and treated for up to 3 hours with various concentrations of fadraciclib or vehicle control (DMSO). Cells were collected every hour, washed once in PBS, pelleted by centrifugation and pellets were snap-frozen in liquid nitrogen and stored at -20°C. Total RNA was extracted using the RNAeasy minikit (Qiagen, Manchester, UK) according to the manufacturer’s instructions. Quantitative real-time PCR was carried out on the Lightcycler II (Roche, Basel, Switzerland) using the One-Step RT-PCR SYBRGreen kit as recommended by the manufacturer and using validated gene-specific primers from Qiagen. The cycling conditions were 50°C for 20 min for the RT step, then 95°C for 15 min and 40 cycles of 94°C for 15 s, 55°C for 20 seconds and 72°C for 20 s. Beta-2 microglobulin (B2M) was used as a housekeeping control for normalization of the data. Each sample was tested in duplicate. Fold changes in gene expression were calculated and normalized to B2M expression.

Western blot analysis

Protein samples for Western blot analysis were prepared as follows. AML cells were seeded at 5 x 10⁵ cells/ml and solid tumor cell lines were seeded at approximately 1 million cells per 10cm dish in the appropriate medium and incubated overnight. Cells were treated for up to 24 h with various concentrations of fadraciclib or DMSO as vehicle control. Cells were collected at various time-points as indicated and centrifuged for 5 min at 478 x g. Medium was removed and cells were washed once in PBS. Cells were pelleted by centrifugation and the pellets were snap-frozen in liquid nitrogen and stored at -20°C until analysis. Pellets were lysed in NP40 cell lysis buffer (Life Technologies, Paisley, UK), PMSF, sodium beta-glycerophosphate and protease inhibitor cocktail (1:1000, Sigma-Aldrich) and sonicated. The soluble protein fraction was obtained after centrifugation at 14,000 x g for 10 min. The protein yield was determined using the BCA protein determination kit (Pierce, Thermo Fisher Scientific, Northumberland, UK) as per the manufacturer’s instructions.

Total protein (25μg) was mixed with 4x gel loading buffer and 1x reducing agent and separated in precast NuPAGE 10%, 4–12% Bis-Tris or 3–8% Tris-Acetate polyacrylamide gels using denaturing electrophoretic conditions (Life Technologies). Proteins were transferred to nitrocellulose membranes using an IBlot transfer unit (Life Technologies). Membranes were blocked in 5% nonfat milk in PBS with 0.05% Tween 20 (PBST) for 2 hours and incubated overnight at 4°C with primary antibody diluted in 5% milk in PBST, or 3% BSA for phospho-specific antibodies. Membranes were washed 3 x 10 min in PBST and incubated for 1 h in 5% milk in PBST containing horseradish peroxidase-conjugated secondary antibody. Membranes were washed as before and incubated with enhanced chemiluminescence solution and exposed to X-ray film (Amersham, VWR, Leicestershire, UK). Protein bands were quantified using ImageJ software. Alternatively, Clarity ECL solution was used and images were captured and analyzed on the Chemidoc (both Biorad, UK). The antibodies and their final concentrations used are listed in S2 Table.

In vitro combination studies of fadraciclib with BCL2 inhibitors

Venetoclax (ABT-199), ABT-263 and ABT-737 were purchased from Selleck Chemicals. To assess potential synergistic interactions, the concomitant treatment regimen involved simultaneous treatment of cells for 72 h, alongside suitable controls of cells treated with the individual compounds alone for 72 h. At the end of the treatment period, cell viability was assessed using the Alamar Blue Cytotoxicity Assay (see above). Combination Index (CI) values were calculated using the method of Chou and Talalay [36]. A combination index (CI) value of 0.9–1.1 indicated an additive drug interaction, a CI value of >1.1 was considered antagonistic and CI <0.9 was considered synergistic.

Assessment of in vivo antitumor activity

Synthesis and characterization of fadraciclib

We disclose for the first time (Fig 1A) the chemical structure of fadraciclib, a second generation clinical candidate based on the aminopurine scaffold of the CDK inhibitor seliciclib. Seliciclib is an ATP competitive inhibitor of CDK2, CDK7 and CDK9, with an average anti-proliferative activity (IC₅₀) in tumor cells of around 15 μM [29]. Seliciclib has been studied extensively in preclinical and clinical settings in both oncology and non-malignant, proliferative and autoimmune conditions NCT00999401, ISRCTN36667085, NCT02160730 and NCT02649751) [30]. It has moderate aqueous solubility and is readily absorbed when administered orally. However, seliciclib is rapidly converted to a less potent carboxylic acid analogue through first pass oxidative metabolism [37]. Medicinal chemistry efforts to increase target CDK inhibitory potency and enhance the metabolic stability of seliciclib resulted in the synthesis of 6-pyridylmethylaminopurines with increased aqueous solubility, metabolic stability and significantly improved target inhibition and anti-proliferative activity [32,33].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Enhanced potency and conserved mechanism of action of CCT068127 and fadraciclib (CYC065) over seliciclib.

A. Chemical structures of seliciclib, and the 6-pyridylmethylaminopurine derivatives, CCT068127 and the clinical candidate, fadraciclib (CYC065). B. CDK selectivity profile of seliciclib, CCT068127 and fadraciclib (CYC065). * IC₅₀ > 10 μM. C. Confirmation of mechanism of action of the derivative compounds. Colo205 colon cancer cells were treated with vehicle (V), seliciclib (S) (26.6 μM), fadraciclib (CYC065) (C) (0.62 μM) or alvocidib (flavopiridol) (F) (0.26 μM) at equivalent of 2 x IC₅₀ for up to 24 h. Cells were harvested at the indicated time-points and the expression levels of RNA pol II, RNA pol II CTD pSer2, RB and MCL1 were examined by Western blotting. D. Colo205 cells were treated for 24 h with 2 x IC₅₀ seliciclib, fadraciclib (CYC065), alvocidib (flavopiridol) or vehicle control, then harvested and fixed for propidium iodide staining to examine the cell cycle profile.

https://doi.org/10.1371/journal.pone.0234103.g001

The clinical development candidate selected from this series is fadraciclib. The synthetic procedure and spectral data are described in Materials and Methods. Fig 1A shows the chemical structure of fadraciclib together with seliciclib and the previously described lead compound CCT068127 [33].

To confirm retention of the desired CDK selectivity, and select the most promising compounds from this series, analogues were screened for inhibition of a panel of CDK enzymes encompassing the principal targets of seliciclib, CDK2, CDK7 and CDK9, and counter-screened against CDK4. The CDK selectivity profiles of seliciclib and the purine analogues are similar, but not identical (Fig 1B). Seliciclib inhibits CDK2, CDK7, and CDK9, in that order, whereas fadraciclib and CCT068127 are more selective for CDK2 and CDK9 and inhibit CDK7 to a much lesser extent. All three compounds exhibit low potency against CDK4. IC₅₀ values for fadraciclib against CDK2 and CDK9 are approximately 20-fold lower than those for seliciclib, which is likely to result in improved cellular anti-proliferative activity in vitro.

To further characterize fadraciclib, its potency was investigated in Colo205 human colon cancer cells in comparison to seliciclib, CCT068127 and alvocidib. The IC₅₀ value of 13.3 μM for seliciclib in Colo205 (S3 Table) was in agreement with the reported average anti-proliferative IC₅₀ for this compound [25]. In contrast, the IC₅₀ values for the two most potent analogues, fadraciclib and CCT068127, were 0.31 μM and 0.82 μM, respectively, which is comparable to alvocidib (0.129 μM S4 Table). The increased potency of fadraciclib was confirmed in a human tumor cell line panel (S5 Table), in which it exhibited an approximately 34-fold enhanced anti-proliferative activity compared with seliciclib.

The cellular mechanism of action of fadraciclib was confirmed in Colo205 cells by examining the impact of treatment for up to 24 h by Western blotting (Fig 1C) and flow cytometry (Fig 1D). In agreement with the observed changes in key CDK targets after seliciclib treatment, fadraciclib treatment resulted in a rapid and robust decrease in RNA pol II C-terminal domain (CTD) pSer2 (a target of CDK9), loss of MCL1 within 3 h, and a decrease in RB pThr821 (a potential CDK2 site) within 8 h. Interestingly, the effect of fadraciclib on these targets were sustained up to 24 h. In contrast, there was evidence of recovery in the levels of RNA pol II CTD pSer2 and particularly MCL1, despite the continued presence of seliciclib or alvocidib (previously reported in [38]).

Together with seliciclib and alvocidib for comparison, the cellular consequences of fadraciclib was examined by flow cytometry in Colo205 cells (Fig 1D).The results revealed the predominant cellular effect to be induction of cell death, as measured by an increase in the sub-G1 population. The robust increase in percentage of cells in sub-G1 made it difficult to draw conclusions on any specific cell cycle perturbations induced by the compounds, although there was evidence of a decrease in cells in both G1 and S phases. The absence of a marked cell cycle arrest is in agreement with the published literature for this class of compounds [33] indicating that the predominant mechanism of action (at concentrations equivalent to 2 x IC₅₀) is related to inhibition of CDKs regulating transcription resulting in induction of apoptosis.

In summary, fadraciclib is significantly more potent against CDK2 and CDK9 and has enhanced effects compared to seliciclib on cell growth and induction of cell death. Fadraciclib shows a more sustained effect on MCL1 in comparison with seliciclib or alvocidib, resulting in its selection as the preferred clinical candidate.

Kinome profiling

Identifying compounds with the appropriate CDK inhibitory profile to maximize effects on transcription, cell cycle and cancer cell death, while retaining sufficient selectivity to avoid unwanted collateral toxicities, is challenging. Due to its essential nature, inhibition of CDK1 appears not to be desirable [39].

To more fully characterize the kinase selectivity of fadraciclib, in vitro screening was performed with 1 μM fadraciclib against a panel of 256 kinases (Carna Biosciences). At this concentration, fadraciclib inhibited only 9 kinases >50%, consisting of 7 CDK enzymes, and 2 CDK-like kinases (Fig 2 and S6 Table). IC₅₀ values for fadraciclib against each of these enzymes were determined at ATP concentrations close to the respective K_m values Fadraciclib was shown to be most potent against CDK2/cyclin A (IC₅₀ = 5 nM), followed by CDK5/p25, CDK9/cyclin T1 and CDK3/cyclin E1 (IC₅₀ values of 21, 26 and 29 nM, respectively) (Table 1). Fadraciclib is more than 40-fold less potent against CDK7/cyclin H, CDK4/cyclin D3 and CLK2 and more than 100-fold less potent against CLK1 and CDK1/cyclin B (Table 1). Like seliciclib, fadraciclib is extremely selective for CDK over non-CDK enzymes.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. In vitro CDK selectivity of fadraciclib (CYC065).

Fadraciclib (CYC065) (1 μM) was profiled against a 256 kinase selectivity panel. The value obtained for the reaction control (complete reaction mixture) was set at 0% inhibition, and the value of the background (no enzyme) was set as 100% inhibition. Percentage inhibition of each enzyme by fadraciclib (CYC065) is indicated in the graph. Nine CDK and CDK-like enzymes were inhibited greater than 50%.

https://doi.org/10.1371/journal.pone.0234103.g002

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. CDK inhibitory potency and selectivity of fadraciclib (CYC065).

IC₅₀ values determined from a 9-point concentration curve at around K_m [ATP] (Carna Biosciences). The most potently inhibited kinases (IC₅₀ < 30nM) are highlighted in bold.

https://doi.org/10.1371/journal.pone.0234103.t001

To determine the potential contribution of inhibiting CDK3 and CDK5, as compared to CDK2 and CDK9, we analyzed data publicly available from genome-wide CRISPR-Cas9 screens in >700 cancer cell lines (from the Cancer Dependency Map project). The results show that in almost all cancer cell lines tested, including the cell lines described in S5 Table, genetic loss of CDK3 and CDK5 has a minimal impact on cell viability. This contrasts with dependency of cancer cell lines revealed following the genetic loss of CDK2 and particularly CDK9 (S6 Fig). Thus the predominant pharmacological effect of fadraciclib on cancer cells is likely to be mediated by inhibition of CDK2 and CDK9.

Characterization of fadraciclib mechanism of action in target cancer and non-transformed cell lines

Treatment strategies for TNBC represents an unmet medical need, as it is one of the most highly proliferative and aggressive types of breast cancer with a poor prognosis [40]. CDK2/9 inhibitors appear to be promising treatments for TNBC for several reasons. CDK2/cyclin E activity is elevated in a significant proportion of TNBC [41]. MCL1 is commonly amplified and correlates with a poor prognosis [9]. TNBC tumors express high levels of the MYC oncogene product compared to other breast cancer subtypes [42]. These features may contribute to the aggressive and refractory nature of TNBC tumors. The properties of fadraciclib, including inhibition of CDK2 activity and depletion of transcripts with a short half-life, such as MCL1 and MYC, as a result of inhibition of CDK9 [43] provide a strong therapeutic rationale to test this agent in TNBC models.

Throughout the study of CDK9 targeting transcriptional inhibitors (such as CCT068127, alvocidib, dinaciclib) it is clear that the onset of biochemical events and subsequent induction of apoptosis are rapid [29,34,44]. Within approximately 30 minutes, RNA pol II CTD pSer2 levels are diminished, followed by loss of MCL1 within a few hours, triggering an irreversible commitment to cell death [29,34,44], This suggests that while clinical exposure of several hours may be required to commit cells to apoptotic death, more prolonged target inhibition may not be necessary.

Indiscriminate inhibition of proliferation for prolonged periods is likely to be universally toxic to both malignant and healthy tissues, leading to significant gastrointestinal and bone marrow toxicity, regardless of CDK-specificity [45]. For example, in the Phase 1 trial assessing the safety and efficacy of a 72 h infusion schedule of alvocidib, toxicities were deemed to be unacceptable and the schedule was not efficacious [46,47]. These observations provided the impetus for utilizing clinically relevant durations and exposures of fadraciclib in preclinical studies, as guided by preclinical animal pharmacokinetics and toxicology, to minimize toxicity to healthy tissues.

We therefore examined the efficacy and cellular mechanism of action of fadraciclib in a small panel of breast cancer cell lines, including two human TNBC cell lines (Cal51 and MDA-MB-468) and the HER2-amplified HCC1954 breast cancer cell line. Non-transformed MCF10A and 184A1 cell lines were also included to assess differences in cellular response between cancer and non-transformed cell lines, which may have implications for the potential therapeutic window. Since we had observed complete loss of RNA pol II CTD pSer2, MCL1 and RB pThr821 by 8 h in Colo205 cells, an 8 h pulse treatment was selected and compared with cellular cytotoxicity of a continuous 72 h treatment in the breast panel (Table 2). The IC₅₀ value established from a standard 72 h viability assay was determined to be ≤0.4 μM in all five cell lines. After 8 h pulse treatment with fadraciclib, the cellular IC₅₀ value remained in the submicromolar range in all three breast cancer cell lines. In contrast, non-transformed MCF10A and 184A1 cells were much less sensitive to pulse treatment (IC₅₀ = 3.9 μM), revealing a difference in sensitivity between the cancer and non-transformed models tested using the pulse dosing schedule.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Short pulse treatment with fadraciclib (CYC065) reveals a therapeutic window in vitro.

https://doi.org/10.1371/journal.pone.0234103.t002

Unsynchronized cell lines were treated with up to an 8 h pulse of 0.5 or 1 μM fadraciclib. Levels of key proteins related to CDK9 inhibition were determined (Fig 3A) together with biochemical evidence of cell death (Fig 3B). Reduced levels of RNA pol II CTD pSer2 were observed by 4–8 h fadraciclib treatment in all cell lines tested (Fig 3A). Consequently, there was also a clear reduction in the levels of both MCL1 and MYC protein in all cell lines. Importantly, PARP cleavage, indicating apoptotic induction, was only observed in the three breast cancer cell lines and not in non-transformed MCF10A cell line (Fig 3A). This demonstrates that although upstream effects of the compound are similar in cancer and non-transformed cell lines, fadraciclib only induces apoptosis in cancer cell lines at the concentrations used. This may reflect a higher dependency on MCL1 and MYC, or apoptotic priming, in cancer cell lines.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Fadraciclib (CYC065) is selectively cytotoxic to breast cancer cells.

A. Representative human breast cancer cell lines and non-transformed cells derived from healthy breast tissue were exposed to fadraciclib (CYC065) for up to 8 h, and the expression levels of selected proteins was examined by Western blotting. B. Cells were treated with vehicle, 0.5 μM or 1.0 μM fadraciclib (CYC065) for an 8 h pulse, then cells were harvested at 24 h for cell cycle analysis by propidium iodide staining. Error bars denote SD.

https://doi.org/10.1371/journal.pone.0234103.g003

To further understand the difference in fadraciclib sensitivity between breast cancer and non-transformed cell lines, flow cytometry profiles were analyzed after an 8 h pulse of fadraciclib followed by incubation in drug-free medium for a further 16 h (Fig 3B). Consistent with the accumulation of cleaved PARP by Western blotting exclusively in cancer cell lines, induction of a sub-G1 population, indicative of cell death, was observed only in the cancer cell lines. In contrast, the non-transformed cell line MCF10A displayed an increase in the G1 population (Fig 3B), potentially indicative of cell cycle arrest. The differential sensitivity of cancer cells to apoptosis, together with the use of the pulse schedule, suggests a promising therapeutic window for fadraciclib, with implications for future clinical development.

Sensitivity of AML, a MCL1-dependent disease, to fadraciclib

MCL1 has been shown to be critical for survival of AML cells across multiple genetic subtypes, including the MLLr and FLT3-ITD subtypes, which are associated with a poor prognosis [48,49]. Because of the direct involvement of CDK9/cyclin T in MLL-driven transcription, it is predicted that CDK9 inhibition may have a significant impact in MLL-rearranged disease. We therefore selected a panel of human AML cell lines, ensuring adequate representation of MLLr and FLT3-ITD subtypes, to examine the activity of fadraciclib in AML.

Seven of the cell lines carry MLL gene rearrangements (www.dsmz.de/catalogues/catalogue-human-and-animal-cell-lines.html), including translocation with different partners such as AF4 (MV4-11), AF6 (ML-2) and AF9 (MOLM-13, Nomo-1, THP-1), and partial tandem duplications (EOL-1 and OCI-AML2). In addition, two cell lines (MOLM-13 and PL21) carry FLT3-ITD, one has a FLT3 mutation (MV4-11) and another amplified MYC (HL60).

As a first step, several of the AML cell lines were used to establish the minimum duration of pulse treatment to achieve optimal cellular response. As an example, Kasumi-1 cells were treated with 0.5 μM or 1.0 μM fadraciclib for up to 6 h and effects on levels of MCL1 protein and cleaved PARP were determined every hour (S1A Fig). MCL1 levels were strongly suppressed at 1 h and further depleted as treatment continued. Cleaved PARP was detectable at 2 h and reached a plateau at approximately 5–6 h following compound addition, suggesting that MCL1 depletion preceded the induction of apoptosis rather than being a consequence of it. A Viacount assay was employed to determine the proportion of dead cells and confirmed that a 6 h pulse treatment was sufficient to induce apoptosis in up to 80% of the cells (S1B Fig). Effect on viability of 6 h-, 8 h- and 10 h-pulses was also compared using the resazurin reduction assay with 72 h continuous treatment in AML cell lines. No significant differences in IC_50/70/90 values were observed between pulse and prolonged continuous treatments (data not shown). Based on these results, a 6 h pulse of exposure was selected for further characterization of AML cell line sensitivity to fadraciclib.

Asynchronously growing AML cells were treated with increasing concentrations of fadraciclib for 6 h and then maintained for an additional 66 h in compound-free medium. Cell viability was assessed using the resazurin reduction cytotoxicity assay and the IC_50/70/90 values determined (Fig 4A). The majority of AML cell lines were highly sensitive to fadraciclib treatment. Six hour incubation in low micromolar or sub-micromolar concentrations of fadraciclib resulted in 90% inhibition of cellular proliferation and yielded IC₅₀ values very similar to 72 h continuous treatment (9 out of 13 cell lines; S7 Table). On average, MLLr/MLL-PTD AML cell lines tested were more sensitive to fadraciclib than AML cell lines without MLLr/MLL-PTD (IC_50/70/90 values; p = 0.048, 0.035 and 0.03, respectively; see inset Fig 4A). The only exception in the MLLr/MLL-PTD subset was the THP-1 cell line which was less sensitive than all other MLLr/MLL-PTD cell lines. Interestingly, a similar correlation between MLL status and sensitivity to fadraciclib was observed in a panel of ALL cell lines (S2 Fig).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Reliance of AML on MCL1 confers sensitivity to fadraciclib.

A. Graph depicting IC_50/70//90, values determined by resazurin reduction cytotoxicity assay after a 6 h pulse treatment with fadraciclib (CYC065) followed by 66h of release. The graph indicates the average of at least 3 independent experiments. Inset: Comparison of IC₉₀ values for a 6 h pulse with fadraciclib (CYC065) in cell lines with or without MLLr/MLL-PTD (p = 0.03). B. Robust induction of cell death in AML cell lines, especially those with MLLr/MLL-PTD. Viability was assessed at 48 h following a 6 h pulse with fadraciclib (CYC065) at the indicated concentrations using Viacount assay. Error bars denote SD. C. Impact of fadraciclib (CYC065) on MLL-mediated transcription. Gene expression was measured by quantitative real-time PCR and normalised to B2M, housekeeping gene. THP-1 cells were treated with 0.5 or 1.0 μM fadraciclib (CYC065), and then harvested at 1, 2 or 3 h to examine the effect of fadraciclib (CYC065) treatment on gene expression. The fold decrease in gene expression of the MLL-mediated genes, HOXA9 and MEIS1, is compared with MCL1. Error bars denote SD. D. Examining the mechanism of action of fadraciclib (CYC065) in AML cell lines with MLLr/MLL-PTD. MOLM-13 and THP-1 are AML cell lines were treated with 0.5 μM or 1 μM fadraciclib (CYC065) for up to 24 h. The protein expression of BCL2 family members, HOXA9 and cleaved PARP were determined by Western blotting. E. Deciphering the mechanism of action of fadraciclib (CYC065) in Kasumi-1 (non MLLr/MLL-PTD cell line). Cells were treated with 0.5–2.0 μM fadraciclib (CYC065) for up to 24 h, and the expression levels of key proteins were examined by Western blotting. RNA pol II CTD pSer2 is a direct substrate of CDK9 and RB pThr821 is a proposed CDK2 site. Actin is included as a loading control. Samples for 2 μM fadraciclib (CYC065) at 24h not included in the total RB and RB pThr821 Western blots. F. Examining the expression levels of BCL2 family members and induction of apoptosis in the resistant HEL cell line. HEL cells were exposed to 0.5 μM or 1.0 μM fadraciclib (CYC065) for up to 24 h, and the expression levels of several BCL2 family members and the cleavage of PARP (Ab detecting both total and cleaved PARP Santa Cruz (Insight) sc-7150) were examined by Western blotting. Viability of HEL cells was also measured at 6 h and 24 h.

https://doi.org/10.1371/journal.pone.0234103.g004

Of the AML cell lines without MLLr/MLL-PTD, three were as sensitive to fadraciclib as were the MLLr/MLL-PTD cell lines. However, three lines (HEL, OCI-AML5 and PL21) were significantly less sensitive to fadraciclib than all other AML cell lines, with IC₉₀ values in the range of 10 to 20 μM (see also section: Potential Patient Stratification Markers).

To determine whether fadraciclib caused cell growth arrest or cell death in individual AML cell lines, a Viacount assay was used to quantify the number of viable and dead cells after 6 h pulse treatment with fadraciclib (Fig 4B). The number of dead cells at different concentrations correlates well with the IC_50/70/90 values determined in the cytotoxicity assay, suggesting that the predominant consequence of fadraciclib treatment is cell death rather than growth arrest. The Viacount data also highlight that the sensitive cell lines undergo apoptosis, reaching almost 100% non-viable cells with only 1 μM fadraciclib. In the less sensitive cell lines, apoptosis is triggered to a lesser extent, for example reaching approximately 50% in KG-1 and with no appreciable increase in apoptosis in HEL with up to 2 μM fadraciclib.

The impact of fadraciclib on the cell cycle was also examined by propidium iodide staining and flow cytometry (S3 Fig) which confirmed a robust increase in the sub-G1 fraction in sensitive cell lines (≥ 0.5 μM fadraciclib), indicative of induction of cell death. These data are consistent with the expected outcome of CDK9 inhibition and demonstrate that short exposures to fadraciclib are capable of producing commitment to cell death in target cells. These findings also imply that continuous target inhibition may not be required for anti-cancer activity in vivo.

Characterization of fadraciclib mechanism of action in AML cell lines with or without MLLr/MLL-PTD

MLL binds to promoters of HOX genes, such as HOXA7 and HOXA9, to maintain their expression [49]. These HOX proteins regulate hematopoiesis and are normally expressed only in early hematopoietic progenitors. MLL fusion proteins directly and constitutively up-regulate HOX expression via recruitment of P-TEFb (CDK9/cyclin T), DOT1 and other proteins involved in transcriptional elongation [48]. Persistent expression of HOX genes along with expression of another up-regulated HOX cofactor, MEIS1, appears to be necessary and sufficient to cause and maintain leukemia [49]. CDK9 inhibition may therefore be expected to be highly effective in CDK9/MLL-driven leukemogenesis via inhibition of leukemogenic transcription [50].

To determine the effect of fadraciclib on MLL-driven transcription, the mRNA levels of HOXA9 and MEIS1 were measured in THP-1 cells treated with 0.5 or 1 μM fadraciclib for up to 3 h (Fig 4C). Fadraciclib inhibited HOXA9 and MEIS1 transcription similarly to MCL1.

The impact of fadraciclib on HOXA9 protein levels was then examined in THP-1 and MOLM-13 cells by immunoblotting (Fig 4D). HOXA9 was decreased by fadraciclib treatment, beginning at 6 h. Interestingly, HOXA9 levels failed to recover in MOLM-13, a sensitive MLLr cell line, whereas in THP-1 cells loss of HOXA9 was both modest and transient, and was equivalent to levels in control cells at 24 h.

Significantly, MCL1 followed a very similar pattern to HOXA9 in these two cell lines, with its protein levels remaining depleted in MOLM-13 but being restored to levels similar to or even exceeding levels in the DMSO vehicle control-treated THP-1 cells. These findings are also reflected in the pattern of cleaved PARP. In the sensitive MOLM-13 cell line, there was a robust concentration-dependent increase in cleaved PARP at 3 h, which was maintained throughout the duration of the experiment. In contrast, robust increases in cleaved PARP were observed only in THP-1 cells at 1 μM fadraciclib and the levels appeared to be greatly diminished by 24 h, despite the continued presence of the drug. This apparently transient apoptotic response hints at a possible recovery or compensatory mechanism in less sensitive cells, which can overcome–to some extent–the effects of fadraciclib, and could explain the reduced sensitivity of this cell line to CDK9 inhibition.

Asynchronously growing Kasumi-1 cells (non MLLr/MLL-PTD), which were highly sensitive to fadraciclib, were treated with increasing concentrations of fadraciclib for up to 24 h and analyzed by Western blotting (Fig 4E). As expected, the CDK9 inhibitory activity of the compound resulted in a rapid decrease of RNA pol II CTD pSer2, MCL1 and MYC protein levels, observed after only 3 h of treatment. The latter finding is consistent with a very short half-life for MYC of less than 20 minutes [51]. Downregulation of the pro-survival proteins preceded the accumulation of p85 PARP cleavage product, a marker of apoptosis, which was evident at 3 h and more pronounced 6 h after treatment initiation. A decrease in RB pThr821 level, related to CDK2 inhibitory activity, was detected at 6 h. This coincided with a band shift for both RB pThr821 and total RB protein, indicating dephosphorylation [52]. Depletion and subsequent loss of RB protein was most evident at 8 h and 24 h. Effects of fadraciclib on CDK2 and CDK9 substrates were sustained to the end of the period tested (24 h) as was observed earlier in Colo205 colon cancer cells.

In summary, the core biochemical events occurring in all cancer cell lines examined appear to be similar, involving loss of RNA pol II CTD pSer2, MCL1 and induction of apoptosis. In addition, AML cell lines carrying MLLr/MLL-PTD appear to show loss of HOXA9, which may also contribute to the loss of viability.

Potential patient stratification markers

Hematological malignancies, such as AML, have been demonstrated to be very sensitive to MCL1 downregulation resulting in a rapid apoptotic response [17]. In our panel, the majority of AML cell lines were very sensitive to fadraciclib, undergoing rapid apoptosis. The presence of the MLLr/MLL-PTD appears to enrich for sensitivity, although it is not apparent from the analysis of the biochemical events in cell lines with and without MLLr/MLL-PTD why that is the case. In addition, within the non MLLr/MLL-PTD group, further markers would be beneficial to differentiate responders from non-responders in this subgroup.

As a first step to understanding the inherent resistance of certain cell lines to fadraciclib, biochemical events were examined in the highly insensitive HEL AML cell line following compound treatment (Fig 4F). Interestingly, similar effects of fadraciclib on MCL1 were observed in this cell line as with the sensitive lines, and at similar concentration levels. However, appearance of cleaved PARP was restricted to the 24 h time-point and was most apparent at 1 μM fadraciclib. Although MCL1 was depleted throughout the duration of the experiment, this alone did not appear to be sufficient to trigger apoptosis in the HEL cell line. However, depletion of BCL2L1 and increase in BAX resulting from a more prolonged treatment with fadraciclib coincided with modest induction of cell death (Figs 4F & S3). This observation is consistent with the reported dependence of HEL cells on BCL2L1 rather than MCL1 for survival [17]. MCL1 belongs to the BCL2 family of proteins which regulate apoptosis, primarily at the mitochondria through the intrinsic apoptotic pathway [53]. These data suggest an important interplay between various components of the BCL2 family in determining cellular responses to MCL1 loss. This indicates that the balance of anti-apoptotic (e.g. MCL1, BCL2, BCL2L1) and pro-apoptotic (e.g. BAX and BAK) proteins might determine sensitivity to transcriptional inhibitors affecting this apoptotic pathway.

To investigate this further, untreated AML cell lysates were analyzed by immunodetection for levels of MCL1, BCL2, BCL2L1, BAX and BAK (Fig 5A). Higher sensitivity of MLLr/MLL-PTD cell lines appears to correspond with high levels of the pro-apoptotic protein BAX. Interestingly, THP-1, the least sensitive MLLr/MLL-PTD cell line of this group, had comparatively low levels of BAX pro-apoptotic protein compared with all other MLLr/MLL-PTD cell lines.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. BCL2 family member expression correlates with sensitivity to fadraciclib (CYC065).

A. Steady state levels of BCL2 family members in the human AML cell line panel. Lysates were prepared from actively proliferating cell lines and the expression of various BCL2 family members was examined by Western blotting. B. Steady state levels of BCL2 family members in selected solid tumour cell lines. C. Induction of cell death. Cleavage of PARP was measured by Western blotting (upper) and % dead cells was assessed by Viacount viability assay at 6 h and 24 h.

https://doi.org/10.1371/journal.pone.0234103.g005

In terms of the non MLLr/MLL-PTD cell lines, HEL, the least sensitive of all AML cell lines tested, had substantially high levels of BCL2L1, which can functionally substitute for MCL1 in binding and neutralizing pro-apoptotic BCL2 family members [44]. High BCL2L1 level was also accompanied by a very low level of the pro-apoptotic protein BAK in this cell line. Relatively low levels of BAK were also observed in two other cell lines that were less sensitive to fadraciclib, OCI-AML5 and PL-21.

It has been reported that apoptosis mediated by MCL1 depletion requires BAK [6]. It thus seems plausible that in the absence of BAK, cells may be more resistant to apoptosis in response to MCL1 depletion [6]. The comparison of IC₅₀ values with BAK levels suggests a reciprocal relationship between these two parameters (Fig 5A). These findings suggest that the inherent resistance of HEL and the other two non MLLr/MLL-PTD cell lines might be related to the low level of the pro-apoptotic protein BAK and are consistent with previous findings [6,44]. For instance, Wei et al. [6] established that, out of over 40,000 genomic features measured, the top feature that predicted sensitivity to transcriptional inhibitor compounds was low expression of BCL2L1. They also determined that BAK deficiency protected cells from these compounds.

To expand on the AML findings, human solid tumor cell lines were selected from a larger panel used to characterize dinaciclib-induced apoptosis [44]. The selected cell lines were A2780 (ovarian cancer), PC3 (prostate cancer), MDA-MB-231 (breast cancer) and A549 and H23 NSCLC cell lines. H23 cells have been described in the literature as MCL1-dependent [54]. Fadraciclib sensitivity of these cell lines was determined by comparing IC₅₀ values obtained with a 6 h pulse versus continuous treatment for 72 h (S4 Fig). A2780 and H23 cells were highly sensitive to fadraciclib, yielding submicromolar IC₅₀ values upon exposure to a short pulse. In contrast, the A549, MDA-MB-231 and PC3 cancer cell lines were sensitive to fadraciclib only after more prolonged treatment.

Untreated cell lysates from these solid tumor cell lines were then examined for expression of MCL1, BCL2L1 and BCL2 to highlight any differences in expression pattern between the two sensitive and the three less sensitive lines (Fig 5B). As expected, the two sensitive cell lines showed high levels of MCL1 and negligible levels of BCL2L1. The less sensitive cell lines displayed the reverse pattern, with substantial levels of BCL2L1 and negligible levels of MCL1. BCL2 was variably expressed and showed no correlation with sensitivity to fadraciclib in this panel. Consistent with observations in the literature for other transcriptional inhibitors [44], it appears in this sample of solid tumors that high expression of MCL1, or indeed a high MCL1:BCL2L1 ratio may confer sensitivity to fadraciclib. Analysis of mRNA levels similarly confirmed that high MCL1:BCL2L1 ratio confers sensitivity to fadraciclib (S5 Fig). Treatment of all five solid tumor cell lines with 1 μM fadraciclib confirmed that only A2780 and H23 cells showed any evidence of apoptosis as determined by appearance of cleaved PARP, accompanied by loss of MCL1, and increase in the proportion of dead cells (Fig 5C).

Fadraciclib combines synergistically with BCL2 inhibitors in AML and ALL cell lines

There are multiple examples of human cell lines that are dependent on MCL1 for survival [see introduction and refs 6,8,9,10,11,12]. These cell lines are highly sensitive to transcriptional inhibition by compounds such as fadraciclib and are characterized by high MCL1:BCL2L1 ratios, and/or detectable levels of BAK. In contrast, there are several cell lines whose survival depend on other BCL2 family members, such as BCL2 or BCL2L1, or in fact may be co-dependent on several anti-apoptotic members. In the latter cell lines, simultaneous inhibition of more than one anti-apoptotic family member may provide an effective mechanism to ablate tumor cells (Fig 6A). To test this hypothesis using cancer cell lines identified in the present study to be less sensitive to fadraciclib, we determined the effects of fadraciclib combined with BCL2 inhibitor venetoclax (ABT-199), or BCL2/BCL2L1/BCL2L2 (bcl-w) inhibitors ABT-263 or ABT-737 in THP-1 (Fig 6B). Similar results were observed with two ALL cell lines, Jurkat and SEM (data not shown).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 6. Dual targeting of MCL1 and BCL2 is highly synergistic at inducing cell death.

A. Schematic diagram of BCL2 family members, interaction and inhibitors. Venetoclax (ABT-199) is a selective BCL2 inhibitor, whereas ABT-263 and ABT-737 inhibit BCL2, BCL2L1 and BCL2L2. Bcl-x_L = BCL2L1, Bcl-w = BCL2L2, A1 = BCL2A1 / Bfl-1. B. Fadraciclib (CYC065) combines synergistically with inhibitors of the BCL2 family in THP-1 cells. y axis–combination index (CI); x-axis–fraction affected (Fa). CI values were calculated using the method of Chou and Talalay [36]. CI values of 0.9–1.1 indicate an additive drug interaction, CI values of >1.1 are considered antagonistic and CI values <0.9 are considered synergistic.

https://doi.org/10.1371/journal.pone.0234103.g006

THP-1 cells were treated in a 7 x 7 matrix of concentrations spanning the IC₅₀ value of each compound: fadraciclib (0.43 μM), ABT-199 (0.06 μM), ABT-263 (0.27 μM) and ABT-737 (0.43 μM). Cells were treated concomitantly for 72 h and compared with suitable controls of cells treated with the individual compounds alone. Combination Index (CI) values were calculated using the method of Chou and Talalay [36]. Fadraciclib combined synergistically with venetoclax, ABT-263 and ABT-737 with median CI values of 0.76, 0.74 and 0.57, respectively. Similar results were observed in HEL cells, except that much higher concentrations of venetoclax were required to observe synergy (data not shown). This suggests that high venetoclax concentrations may be necessary to inhibit BCL2L1 as well as BCL2 in this cell line consistent with the observation that HEL cells show significant overexpression of and is dependent on BCL2L1.

Fadraciclib has potent anti-AML activity in vivo

Overall, the cellular pharmacology data showed that fadraciclib effectively downregulates pro-survival and leukemogenic proteins and that short pulse treatment is sufficient to substantially decrease cell viability of the majority of AML cell lines tested via apoptotic induction. To test if these effects will translate into anti-leukemic activity in vivo we used two subcutaneous xenograft models EOL-1 (with MLL-PTD) and HL60 (non MLLr/MLL-PTD AML with MYC amplification).

SCID mice with established subcutaneous EOL-1 tumors were randomized into groups of 6 animals and assigned to vehicle control, fadraciclib at 40 or 55 mg/kg, or cytarabine 100 mg/kg (Fig 7A). Cytarabine (a nucleoside analogue, incorporated into DNA) was used as a positive control as this is one of the main chemotherapeutic agents used in pediatric and adult patients with AML and ALL, including those with MLL aberrations. Fadraciclib administration at 40 and 55mg/kg resulted in 97% and 100% tumor growth inhibition (TGI), respectively, on Day 19 (study completion; p<0.0001 versus vehicle control). In comparison, cytarabine treatment inhibited tumor growth by 38%. Overall, the treatments were well tolerated with only transient body weight loss of 2% and 3% in the fadraciclib 55 mg/kg and cytarabine groups respectively (data not shown).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 7. In vivo efficacy of fadraciclib (CYC065) in AML xenograft models.

A. Potent anti-tumour activity of fadraciclib (CYC065) in the EOL-1 MLL-PTD AML xenograft model. Fadraciclib (CYC065) was administered orally at 40 or 55 mg/kg qd x 5 for 2 weeks. AraC was administered using an optimised schedule for this model, consisting of 100mg/kg i.p. on days 1–5. The median TGI achieved on day 9 was 97% and 95% for 55 and 40 mg/kg fadraciclib (CYC065), respectively, compared with 41% for AraC. B. Anti-tumour activity in HL60 (MYC amplified non-MLLr/MLL-PTD) AML xenograft model. Fadraciclib (CYC065) was administered at 70 mg/kg orally over 5 days for 2 weeks, and delayed the growth of HL60 xenografts (T-C >12.6 days) and TGI on day 11 was 90%, which was significantly different to the vehicle control group (p<0.001). Compound treatment was well tolerated in both models at all doses with a mean body weight loss of 8% at the highest dose.

https://doi.org/10.1371/journal.pone.0234103.g007

HL60 was selected as a representative, sensitive, non MLLr/MLL-PTD AML cell line. Nude mice with subcutaneous HL60 tumors were randomized into groups of 10 animals and assigned to vehicle control or fadraciclib treatment groups (Fig 7B). Fadraciclib administration at 70mg/kg delayed the growth of HL60 leukemia xenografts with a T-C value of more than 12.6 days. Two animals achieved tumor regression and one was determined to be tumor-free. TGI on Day 11 was 90%, significantly different versus vehicle control (p< 0.001). An average of 8% body weight loss was observed in the treatment group (data not shown).

In summary, fadraciclib was well tolerated and showed promising efficacy in two subcutaneous AML xenografts as a single agent in agreement with the essential role for MCL1 in AML survival.