Evaluation of Protein Kinase Inhibitors with PLK4 Cross-Over Potential in a Pre-Clinical Model of Cancer

Polo-like kinase 4 (PLK4) is a cell cycle-regulated protein kinase (PK) recruited at the centrosome in dividing cells. Its overexpression triggers centrosome amplification, which is associated with genetic instability and carcinogenesis. In previous work, we established that PLK4 is overexpressed in pediatric embryonal brain tumors (EBT). We also demonstrated that PLK4 inhibition exerted a cytostatic effect in EBT cells. Here, we examined an array of PK inhibitors (CFI-400945, CFI-400437, centrinone, centrinone-B, R-1530, axitinib, KW-2449, and alisertib) for their potential crossover to PLK4 by comparative structural docking and activity inhibition in multiple established embryonal tumor cell lines (MON, BT-12, BT-16, DAOY, D283). Our analyses demonstrated that: (1) CFI-400437 had the greatest impact overall, but similar to CFI-400945, it is not optimal for brain exposure. Also, their phenotypic anti-cancer impact may, in part, be a consequence of the inhibition of Aurora kinases (AURKs). (2) Centrinone and centrinone B are the most selective PLK4 inhibitors but they are the least likely to penetrate the brain. (3) KW-2449, R-1530 and axitinib are the ones predicted to have moderate-to-good brain penetration. In conclusion, a new selective PLK4 inhibitor with favorable physiochemical properties for optimal brain exposure can be beneficial for the treatment of EBT.

CFI-400437 is an indolinone-derived, ATP-competitive kinase inhibitor [60] with high selectivity for PLK4, but also displays lower levels of inhibitory activity for the other members of the PLK family. CFI-400437 was found to significantly impair the growth of breast cancer cells in vitro, as well as decreasing tumor size in mouse xenograft tumor models [60].
Centrinone and centrinone B are selective, reversible PLK4 inhibitors (PLK4i) developed using the pan-AURK inhibitor VX-680 as a template. Centrinone reversibly depleted centrioles by preventing centriole assembly in HeLa cells resulting in a p53-dependent cell cycle arrest in G1 phase. Centrinone B exhibited comparable outcomes with centrinone [59].
R1530 is a member of a series of pyrazolobenzodiazepine compounds, which were identified as mitotic inhibitors. R1530 is a multi-kinase inhibitor which was shown to target all the five members of the PLK family. It induced polyploidy that led to apoptosis of cancer cells by interfering with the mitotic checkpoint kinase BubR1, which is likely to be a result of its PLK4 inhibition [58].
Axitinib was developed as a tyrosine kinase inhibitor, specifically with high affinity for VEGFR1-3 [67]. Prior to FDA approval, axitinib was found to inhibit PLK4 with an IC 50 value of 4.2 nM [67]. Axitinib was used during the synthesis of CFI-400437, as a model of a potent PLK4i [60]. In 2012, axitinib was approved by the FDA for use in renal cell carcinoma [57,68,69] and has since been in clinical trials to treat thyroid [70] and advanced non-small cell lung [71] cancer as well as melanoma [61,72].
KW-2449 is an analog of CFI-400945 that we identified in an attempt to find a candidate molecule with potentially better brain exposure based on its chemical properties. It has been previously described as a multi-kinase inhibitor of FLT3, ABL, ABL-T315I, and AURKs. Potent inhibitory growth effects on leukemia cells with FLT3 mutations were achieved by inhibition of the FLT3 kinase, with subsequent down-regulation of phosphorylated-FLT3/STAT5, G1 arrest, and apoptosis [63,64].
Alisertib is a pyrimidobenzazepine, ATP competitive inhibitor for the AURK family with the highest selectivity for Aurora kinase A (AURKA) [73]. AURKA has previously been described to be overexpressed in AT/RT [74] and alisertib is currently in a phase II clinical trial for its treatment (NCT02114229).
It is estimated that >95% of approved drugs lack sufficient blood-brain-barrier (BBB) penetration to allow efficacy [75][76][77]. The tissue exposure challenge is even greater for PKI drugs [78]. Briefly, an initial informatics analysis of differences in molecular properties between approved CNS drugs and PKI drugs approved or in public domain clinical trials databases identified a cluster of three molecular properties that distinguished PKIs from approved CNS drugs. CNS drugs clustered around a multiproperty profile of molecular weight (MW ≤ 400), lipophilicity (LogP ≤ 4), and polar surface area (PSA ≤ 80), with extant PKIs often falling outside this multi-property profile, especially PSA. This pioneering observation was updated and extended to undisclosed internal CNS candidates within Pfizer as well approved CNS drugs [79,80]. The investigators proposed a CNS multiproperty profile, based on analysis of their proprietary database, that could be used for late stage medicinal chemistry optimization [79]. Continuing trends in multi-property consideration for CNS drug candidate design and refinement are now embedded in commercially available computational packages such as ACD Labs Percepta as a virtual screen for early alerts related to potential brain penetration.
Herein, we evaluate multiple small molecule PKI with potential PLK4 crossover for their effects in a pre-clinical cellular model of EBT, potential to occupy the PLK4 active site in virtual docking analyses, and their potential liability for BBB penetration based on computed multi-property profile.
Our aim was to evaluate the abovementioned inhibitors ( Figure 1) in a pre-clinical model of embryonal tumors (RT and MB) in regard to their anti-cancer properties and potential for brain exposure. The predicted binding modes of these inhibitors were also examined, with the goal of identifying the structural requirements for potential to binding to the PLK4 active site.

Docking Simulations
Docking simulations using the PLK4 catalytic cavity and each of the various PKIs with PLK4 crossover activity ( Figure 1) demonstrated that all of the tested inhibitors, except axitinib, exhibited similar interactions within the binding cavity, engaging in H-bonding with backbone residues Glu-89 and Cys-91 (Figure 2A-H). Only axitinib did not engage in H-bonds at the hinge region ( Figure 2I) with Glu-89 and Cys-91. Instead, axitinib exhibited two H-bonds with the backbone of Leu-17 and the side chain of Arg-98 at the ribose pocket ( Figure 2I). Interestingly, alisertib ( Figure 2A) and axitinib ( Figure 2B) engaged in an additional cation-π interaction with Lys-40, while the empirical structure of centrinone [59] ( Figure 2C) and the docking pose of centrinone B ( Figure 2D) both engaged in cation-dipole interactions with this residue (−NH 3 + F). These findings may explain the higher docking scores obtained for these inhibitors (Table 1). Compounds CFI-400437 and CFI-400945 displayed extra H-bonding with Gly-18 and Arg-98, respectively, and a similar exploration of the phosphate-binding site ( Figure 2E,F,I,J). Compounds KW-2449 and R1530 only exhibited polar interactions at the hinge region that may explain their lower scores as presented in Table 1. Noteworthy, the docking findings also highlighted the potential of KW-2449 and R1530 to guide further structure-activity relationship (SAR) studies as well as the design of novel compounds with improved exploration of the PLK4 binding cavity.
Overall, centrinone and centrinone B represent those with higher computed risk based on total 237 polar surface area (TPSA) > 190 Å 2 as well as > 600 MW. Clearly, future studies for brain tumor 238 treatment research warrant an exploration of dose-dependent brain exposure in order to provide a

250
Oxygen is depicted in red. Nitrogen is depicted in blue. In panels C and D, sulfur is depicted in

257
Aurora kinase B (AURKB), and Aurora kinase C (AURKC) were also determined for each of the 258 inhibitors included in this study. Results shown in Figure 3 and Table 3    Additional hydrophobic interactions have also been observed through the cavity for all inhibitors especially-but not exclusively-with Leu-17, Val-25, Ala-38, and Leu-142 (Figure 2A-H). The overlapping poses of the compounds ( Figure 2J) indicated a similar exploitation of the binding cavity by the eight compounds examined. Therefore, compounds bearing polar groups like piperazine (CFI-400437 and KW-2449) and morpholine (centrinone and CFI-400945) had a better fit within the hydrophilic regions of the pocket (i.e., phosphate-binding site and ribose pocket, Figure 2I,J), or turned their polar moieties to the solvent-exposed region.

Physical Property Analysis
The logBB is a logarithm index that express the brain/plasma concentration ratio of drugs, and it is indicative of the permeation of drugs through the BBB. The logBB values from ACD Labs Percepta 2016 (Build 2911) identify differences among the inhibitors, including those with a common indazole core ( Table 2).  Overall, centrinone and centrinone B represent those with higher computed risk based on total polar surface area (TPSA) > 190 Å 2 as well as > 600 MW. Clearly, future studies for brain tumor treatment research warrant an exploration of dose-dependent brain exposure in order to provide a firm experimental foundation.

Kinase Assays
Kinase assays were performed to determine the half-maximum inhibitory concentrations (IC 50 ) for PLK4. Because PLK4 has high catalytic similarities with AURKs [60,81,82], the IC 50 for AURKA, Aurora kinase B (AURKB), and Aurora kinase C (AURKC) were also determined for each of the inhibitors included in this study. Results shown in Figure 3 and Table 3 indicate the following: (1) CFI-400437 had the lowest IC 50 value for PLK4 (1.55 nM) followed by centrinone (2.71 nM) and CFI-400945 (4.85 nM); (2) centrinone and centrinone B were the most specific to PLK4; (3) CFI-400437 inhibited both AURKB and AURKC at concentrations <15 nM, while CFI-400945 inhibited AURKB at a higher concentration (70.7 nM) and AURKs A and C at even higher concentrations (188 nM and 106 nM, respectively); (4) high IC 50 for PLK4 was observed for KW-2449 (52.6 nM) which inhibited all three AURKs more efficiently than PLK4; (5) finally and not surprisingly, alisertib, which is known to be a primary inhibitor of AURKA, had the highest IC 50 for PLK4 (62.7 nM) with the best inhibitory activity of all AURKs.

Cell-Based Studies
The evaluation of the impact of each inhibitor over cell phenotype (proliferation, viability, colony formation, senescence, and cell cycle/polyploidy) demonstrated that: (1) Decrease in viability and proliferation was consistent with the results of the kinase assay ( Figure 3 and Table 3) being more significant for CFI-400437 and centrinone (Table 4, Figures 4 and 5), followed by CFI-400945, which has been previously tested in our system [36,37]; (2) although KW-2449 had the second highest IC 50 on the kinase assay, significant impact over viability/proliferation, inhibition of colony formation, induction of senescence and polyploidy were observed, but in higher concentrations (1-2 µM) ( Figure 6); (3) centrinone, which is the most selective inhibitor for PLK4, together with centrinone B, did not induce polyploidy in the cell lines tested ( Figure 5). AURKB proper function during mitosis is crucial for ensuring the prevention of mitotic errors, endoreduplication, and polyploidy [83][84][85]. Therefore, this phenomenon could stem from the absence of AURKB-associated inhibition, which may be necessary for induction of polyploidy [86].

KW-2449
Although kinase assays reveal this ligand to possess a high IC 50 for PLK4 and cell-based assays indicated a phenotype consistent with PLK4 inhibition at high concentrations, the docking simulations indicated that KW-2449 has the potential to guide further structure-activity relationship (SAR) studies and the design of novel compounds with improved exploration of the PLK4 binding cavity. For this reason, and because of the favorable scores for BBB penetration, we further studied KW-2449.

KW-2449 Potential for Cardio Toxicity and Drug-Drug Interaction Risk
Screening of a PKI for human ether-a-go-go-related gene (hERG) potassium channel inhibition has become an early step in testing for potential drug dependent long QT syndrome that is linked to sudden death [87]. Binding of KW-2449 at the hERG channel was not detected under standard assay procedures at the highest concentration tested (IC 50 > 10 µM) (Supplementary Figure S1). These data suggest that KW-2449 is not in the high-risk category for cardiovascular toxicity. Similarly, cytochrome P450 (CYP450) enzymes, which are drivers of first-pass metabolism for orally administered drugs, can contribute to toxicities or therapeutic failures [88] with the two of the most significant in risk being CYP3A4 and CYP2D6, which showed an IC 50 value >10 µM (the highest tested concentration) (Supplementary Figure S1).

KW-2449 Kinase Inhibition Screen
We performed a kinase selectivity screen and established that KW-2449 has low selectivity, inhibiting PLK4 among several other kinases. The screening revealed that 215 out of 486 (44%) kinases had their activity inhibited above 80% by KW-2449 at 10 µM (Supplementary Table S1). The IC 50 for selected kinases was validated by a 10-point titration curve analysis. The IC 50 values of AURKA, AURKB, and AURKC were 45.8 nM, 23.8 nM, and 23.2 nM, respectively while the IC 50 value of PLK4 was even higher: 52.2 nM (Figure 3 and Table 3). In summary, KW-2449 showed low selectivity and efficient PLK4 inhibitory effect inducing a phenotype consistent with PLK4 inhibition in EBT cells. Key features suggest a potential for low cardiotoxicity or drug-drug interaction risk.

399
In order to identify improved inhibitors for use in the treatment of brain cancers, it will be 400 important to design compounds that are not only selective for PLK4, but that also have appropriate 401 lipophilicity (logP) and as well as modest total polar surface areas (TPSA) to allow for brain penetration. Thus, the use of multi-method predictive modules such as central nervous system

Discussion
PKIs have promised to overcome major disadvantages of traditional cancer treatments as they potentially discriminate between non-malignant and rapidly proliferating cancer cells, leading to fewer off-target effects and lower toxicity for the patients. We previously demonstrated overexpression of PLK4 in embryonal tumors including RT, pediatric MB, and CNS-NB, for which only highly toxic and poorly effective treatments are available. Our preliminary findings suggested that targeting PLK4 with small-molecule inhibitors may represent a novel strategy to treat EBT and possibly other tumors of the brain [36][37][38]. We have demonstrated that oral CFI-400945 delayed tumor growth and improved survival in a xenograft model of AT/RT [35]. However, CFI-400945 is at the cusp of the multi-property profile predictive of efficient brain exposure and therefore, this favorable outcome may be a result of the disruption of the BBB by the xenografting procedures, rather than the ability of the molecule to penetrate the brain. This type of BBB disruption can be responsible for over-predicting the efficacy of agents that otherwise would not penetrate the brain. This explanation is one of the reasons why certain agents that were shown to be active in preclinical studies failed in clinical trials [89]. A brain penetrant PLK4i could be used to treat not only EBT but other aggressive CNS tumors with PLK4 overexpression like GBM [34]. Therefore, we conducted the present study in order to understand the potential of available PLK4is and explore the possibility to develop new chemical entities with higher PLK4 selectivity and efficient BBB penetration.

Polyploidy and Aurora Kinases (AURKs)
Polyploid cells have more than two paired homologous chromosomes. Aberration of the homeostatic diploid state, resulting in polyploidy, has been implicated in a number of cancers [90][91][92][93], but it can also increase cancer cells' susceptibility to cytotoxic therapies [94]. Cell fusion, cytokinesis failure, and endoreduplication are three mechanisms that can lead to polyploidization [92].
It has been demonstrated that the well-balanced action of the PLK and AURK families of protein kinases is essential for normal mitotic function. AURKs are members of a conserved family of serine/threonine PKs involved in cell cycle progression [85]. AURKA is involved in organizing cell polarity, regulating the assembly and organization of mitotic spindles, centrosome maturation, and regulation of mitotic checkpoints [81,95]. AURKB is actively involved in mitosis, chromosomal alignment and segregation, formation of the cleavage furrow, and control of checkpoints, and therefore prevents the occurrence of aneuploidy [83,85,86]. In fact, inhibition of AURKB has been found to lead to polyploidy in vitro in both cancer and non-cancer cells [83,84,86,96]. AURKC is the least-studied member of the family. It is known to be highly expressed in testis and involved in the coordination of meiotic spindles during spermatogenesis [81,82]. While the active sites of AURKB and AURKC are identical, the active site of AURKA differs by only three amino acids [82]. Whereas inhibitors specific to AURKB and pan-AURK inhibitors induced significant polyploidy in vitro, the phenotype was not observed when subjected to inhibition by an AURKA selective inhibitor [84].
PLK4 shares 37% residue homology with that of AURKA, making the active site of PLK4 more structurally similar to the members of the AURK family than to PLK1 [60,72,81,82]. This high degree of homology between the ATP-dependent catalytic domains could explain the observed inhibition of AURKs when subjected to PLK4is. Because of these catalytic similarities between PLK4 and AURKs, we evaluated the kinase inhibitory activity of all three AURKs for each PLK4i.
Previously, we have observed induction of polyploidy after treating EBT cells with CFI-400945 [36,37]. We interpreted this outcome as a result of DNA endoreduplication in the absence of centriole replication, which would have been resultant from PLK4 inhibition. However, our results showed that centrinone, the most selective inhibitor for PLK4, did not induce polyploidy of the cancer cells ( Figure 3B), which raises the hypothesis that the major factor responsible for polyploidy is inhibition of AURKB rather than PLK4.
Although the majority of drug developing efforts have focused on selectivity for a single biological target, there has been evidences implying that more selective drugs are less likely to work in the clinic compared with drugs that target a set of molecular pathways implicated in the disease pathophysiology [77]. In this scenario, aiming for a PKI targeting multiple desirable targets may be of greater interest.

Considerations for Future Development of a Brain Penetrant PLK4i and its Therapeutic Use
In order to identify improved inhibitors for use in the treatment of brain cancers, it will be important to design compounds that are not only selective for PLK4, but that also have appropriate lipophilicity (logP) and as well as modest total polar surface areas (TPSA) to allow for brain penetration. Thus, the use of multi-method predictive modules such as central nervous system multi-parameter optimization (CNS MPO) should be considered in the compound optimization process. KW-2449 and R1530 appear to have reasonable scores for brain exposure while centrinone and centrinone B are the most selective in terms of their kinase profiles. These results together with further molecular docking studies may lead to the design of compounds with improved selectivity and brain penetration, thus allowing for improved cancer therapeutics.
Kinase inhibitors can be used in combination with cytotoxic chemotherapy, radiation therapy, or other STMs. In previous studies, we established the synergism of the PLK4i CFI-400945 with cytotoxic chemotherapy drugs [35]. A key challenge in the clinics is to identify the most efficient combination of kinase targets and then develop treatment combinations for specific cancer types. Since the PLK4 inhibition has a well-defined mechanism of action, combination with drugs that act on cooperating pathways are potentially significant. For example, Kawakami et al. [28] found that combined treatments with CFI-400945 and seliciclib, a CDK2 inhibitor, exerted both additive and synergistic effects to reduce lung cancer cell growth. Therefore, combining a PLK4i not only with cytotoxic drugs but also with drugs that affect cooperating pathways may be promising.

Inhibitors Tested
We studied 8 PKI with PLK4 cross-over potential:

Structural Analysis of the Inhibitors
The crystallographic structure of PLK4 was obtained from the Protein Data Bank (PDB) with access code 3COK (https://www.rcsb.org/structure/3cok). This structure has a resolution of 2.25 Å. All the inhibitors' structures were built and energy-minimized with the density functional theory (DFT) method Becke-3-Lee Yang Parr (B3LYP) with the standard 6-31G* basis set available in the Spartan'14 program (Wavefunction, Inc., Irvine, CA, USA). Molecular docking was performed with the GOLD 5.2 program (CCDC). The scoring function used was 'GoldScore', which is the default function of the GOLD 5.2 program. Hydrogen atoms were added to PLK4 according to the data inferred by the GOLD 5.2 program on the ionization and tautomeric states. The docking interaction cavity in the protein was established with a 10 Å radius from the co-crystalized ligand (phosphoaminophosphonic acid-adenylate ester). The number of genetic operations (crossover, migration, mutation) in each docking run that was used in the search procedure was set to 100,000. Docking simulations were performed six times for each inhibitor. The figure of the best-and most frequent-docking pose for each compound was generated by the PyMOL Delano Scientific LLC program (Palo Alto, CA, USA). The experimental structure of centrinone bond to PLK4 was obtained from the PDB with the access code 4YUR. CNS MPO scores were calculated as per the original method described by Pfizer. Calculated logP, log D, TPSA, MW, pKa, and HBD were obtained using ACD Labs Percepta 2016 (Build 2911). The logBB values were obtained using ACD Labs Percepta software v16 (Advanced Chemistry Development, Inc., Toronto, On, Canada).

Kinase Assay
Each inhibitor's potency was determined by generating 10-point IC 50 curves from a 4-fold dilution series in DMSO (1 mM). Curves were generated from the compound concentration and the corresponding percent (%) inhibition calculated for each concentration tested. For PLK4, we used the LanthaScreen Eu Kinase Binding Assay (Thermo Fisher Scientific, Carlsbad, CA, USA) that utilizes an Alexa Fluor conjugated "tracer" and an Eu-labeled anti-tag antibody to measure binding of a compound to the kinase target. For AURKs (AURKA, AURKB and AURKC) we used the Z' LYTE Kinase Assay (Thermo Fisher Scientific, Carlsbad, CA, USA) that determines the differential sensitivity of phosphorylated and non-phosphorylated peptide substrates to proteolytic cleavage using a FRET-based readout.

Cell Culture
Six well-established and extensively characterized cell lines were used in this study: The rhabdoid cells MON provided by Dr. Delattre (Institute Curie, Paris, France) [98], the AT/RT cell lines BT-12 and BT-16, which have been extensively used in preclinical studies [99], and two MB cell lines (DAOY and D2830). The cells were cultured as previously described.

Cell-Based Assays
For each cell line and each inhibitor, we performed proliferation, viability, senescence, colony formation, and cell cycle analysis, as described below.

Proliferation Assay
To evaluate cell proliferation, the TACS MTT Cell Proliferation Assays (Trevigen, Gaithersburg, MD, USA) were used. The RT cell lines (MON, BT-12, and BT-16) were plated at 2 × 10 3 cells on each well of a 96-well plate. The MB cell lines (DAOY and D283) were plated at 5 × 10 4 and 5 × 10 5 cells per well, respectively. The absorbance was measured after 24, 48, 72, and 96 h at concentrations ranging from 0.001 to 10 µM of the inhibitor. All assays were performed in triplicates.

Viability Assay
Cell viability was assessed using the Presto Blue™ Cell Viability reagent (Thermo Fisher Scientific, Carlsbad, CA, USA) following the manufacturer's instructions. Cells were plated in 96-well plates at the same densities described above for the proliferation assay. The fluorescence was measured after 24, 48, 72, and 96 h at concentrations ranging from 0.001 to 10 µM of the inhibitor. All assays were performed in triplicate.

Clonogenic Assay
All cell lines were seeded at 200 cells per well in six-well plates and incubated overnight. After 24 h, the treatment medium was added and changed once a week. Cells were treated with 0.01, 0.05, 0.1, 0.2, 0.5, 1, 5, and 10 µM of PLK4i and 0.1% DMSO as control. After 14 days, the cells were washed twice with 1x PBS, fixed with formalin, and stained with Cresyl violet (ACROS Organics, Pittsburgh, PA, USA). Colonies were counted using the ImageJ software (www.imagej.nih.gov, accessed on April 21, 2017).

Beta-Galactosidase Assay
Senescence was evaluated using the Beta-galactosidase assay (Senescence Cells Histochemical Staining Kit CS0030 (Sigma, St Louis, MO, USA) and clonogenic recovery assay as previously described.

Cell Cycle Analysis
Cell cycle analysis was performed by flow cytometry of cells stained with Propidium Iodide (PI) (Thermo Fisher Scientific, Carlsbad, CA, USA) according to manufacturer's instructions. Cells treated with the top PLK4i candidates and 0.1% DMSO (control) were fixed in 80% ethanol overnight, stained with PI, and then subjected to flow cytometric analysis using a BD Fortessa instrument (BD Biosciences, San Jose, CA, USA). Data was analyzed using Modfit LT from Verity Software House (Topsham, ME, USA). All experiments were performed in triplicate.

Kinase Activity Screening
The inhibitory activity of KW-2449 on various kinases was assessed through 486 biochemical kinase assays using 10 µM of the compound (SelectScreen Kinase Profiling Services-Thermo Fisher Scientific, Carlsbad, CA, USA). These assays utilized various formats appropriate to the kinase, its substrate, and its activity. The IC 50 for selected kinases was further measured by 10-point titration.

Drug Safety and Toxicology
Analysis of KW-2449 against a panel of P450s isoenzymes (1A2, 2B6, 2C8, 2C9, 2C19, 2D6, 2J2, 3A4, and 3A5) and substrates was performed using 10-point titrations, in duplicates to determine IC 50 values (SelectScreen P450 Profiling Service-Thermo Fisher Scientific, Carlsbad, CA, USA). To determine if the compound binds the cardiac hERG channel, the IC 50 value was quantitated using a 10-point titration, in duplicates in the Predictor hERG fluorescence polarization assay (SelectScreen hERG Screening Service-Thermo Fisher Scientific, Carlsbad, CA, USA).

Conclusions
Our analyses demonstrated that: (1) CFI-400437 s inhibition of PLK4 had the greatest impact overall, but similar to CFI-400945, its physiochemical properties are not optimal for brain exposure. Also, their phenotypic anti-cancer impact may, in part, be a consequence of the inhibition of AURKs, especially AURKB. (2) Centrinone and centrinone B are the most selective PLK4is but they are the least likely to penetrate the brain. (3) KW-2449, R1530, and axitinib are the ones predicted to have moderate-to-good BBB penetration. (4) The compound with the most chemically favorable molecular properties for brain exposure, KW-2449, was also the least selective for PLK4.
In conclusion, we showed that PLK4 can be inhibited by a variety of structurally disparate molecules, with varying degrees of in vitro phenotypic effects and that a selective PLK4i with favorable physiochemical properties for optimal brain exposure will be beneficial for the treatment of aggressive brain tumors with PLK4 overexpression. Finally, there is a need to improve the existing inhibitors for better selectivity and brain exposure and/or to develop new chemical entities with these desirable characteristics.