Abstract

Simple Summary

   Glioblastoma multiforme (GBM) remains to be the most frequent malignant
   tumor of the central nervous system (CNS), which accounts for
   approximately 54% of all primary brain gliomas. Current treatment
   modalities for GBM include surgical resection, followed by radiotherapy
   and chemotherapy with temozolomide (TMZ). However, due to its genetic
   heterogeneity, GBM tumors always recur, due to treatment reasistance.
   The aim of this study was to identify molecular gene signatures,
   responsible for cancer initiation, progression, resistances and to
   treatment, metastasis, and also evaluate the potency of our novel
   compounds SJ10 as potential target for CCNB1/CDC42/MAPK7/CD44 oncogenic
   signatures. Accordingly, we used computational simulation and identify
   these signatures as regulators of the cell cycle in GBM, which leads to
   cancer development and metastasis. We also showed the antiproliferative
   and cytotoxic effects of SJ10 compound against a panel of NCI-60 cancer
   cell lines. This suggests the potential of the compounds to inhibit
   CCNB1/CDC42/MAPK7/CD44 in GBM.

Abstract

   Current anticancer treatments are inefficient against glioblastoma
   multiforme (GBM), which remains one of the most aggressive and lethal
   cancers. Evidence has shown the presence of glioblastoma stem cells
   (GSCs), which are chemoradioresistant and associated with high invasive
   capabilities in normal brain tissues. Moreover, accumulating studies
   have indicated that radiotherapy contributes to abnormalities in cell
   cycle checkpoints, including the G[1]/S and S phases, which may
   potentially lead to resistance to radiation. Through computational
   simulations using bioinformatics, we identified several GBM oncogenes
   that are involved in regulating the cell cycle. Cyclin B1 (CCNB1) is
   one of the cell cycle-related genes that was found to be upregulated in
   GBM. Overexpression of CCNB1 was demonstrated to be associated with
   higher grades, proliferation, and metastasis of GBM. Additionally,
   increased expression levels of CCNB1 were reported to regulate
   activation of mitogen-activated protein kinase 7 (MAPK7) in the G[2]/M
   phase, which consequently modulates mitosis; additionally, in clinical
   settings, MAPK7 was demonstrated to promote resistance to temozolomide
   (TMZ) and poor patient survival. Therefore, MAPK7 is a potential novel
   drug target due to its dysregulation and association with TMZ
   resistance in GBM. Herein, we identified MAPK7/extracellular regulated
   kinase 5 (ERK5) genes as being overexpressed in GBM tumors compared to
   normal tissues. Moreover, our analysis revealed increased levels of the
   cell division control protein homolog (CDC42), a protein which is also
   involved in regulating the cell cycle through the G[1] phase in GBM
   tissues. This therefore suggests crosstalk among
   CCNB1/CDC42/MAPK7/cluster of differentiation 44 (CD44) oncogenic
   signatures in GBM through the cell cycle. We further evaluated a newly
   synthesized small molecule, SJ10, as a potential target agent of the
   CCNB1/CDC42/MAPK7/CD44 genes through target prediction tools and found
   that SJ10 was indeed a target compound for the above-mentioned genes;
   in addition, it displayed inhibitory activities against these oncogenes
   as observed from molecular docking analysis.

   Keywords: glioblastoma multiforme (GBM), temozolomide (TMZ),
   chemoradioresistance, genetic heterogeneity, bioinformatics, molecular
   docking, National Cancer Institute (NCI)-60

1. Introduction

   Glioblastoma multiforme (GBM) is one of the most frequent malignant
   tumors of the central nervous system (CNS) [[36]1,[37]2], which
   accounts for approximately 54% of all primary brain gliomas, with a
   yearly incidence of 3.2 per 100,000 adults globally [[38]3,[39]4], and
   is classified as grade IV by the World Health Organization (WHO)
   [[40]5]. It is associated with poor clinical outcomes, with fewer than
   10% of patients reaching a 5-year survival rate after diagnosis
   [[41]6,[42]7]. The current treatment modalities for GBM include
   surgical resection, followed by radiotherapy and chemotherapy with
   temozolomide (TMZ) [[43]8,[44]9,[45]10]. However, due to genetic
   heterogeneity, GBM tumors always recur mainly at the resection site,
   leading to an overall median survival of only 15 months following the
   initial diagnosis [[46]5,[47]11]. Therefore, understanding molecular
   mechanisms and invasive characteristics of GBM is pivotal as an
   essential strategy for developing more-effective therapeutics.
   Resistance to treatment in GBM is also associated with glioblastoma
   stem cells (GSCs), which may potentially assist GBM cancer cells to
   escape irradiation [[48]11,[49]12,[50]13]. Studies showed that GCSs are
   resistant to TMZ chemotherapy, thus promoting radio resistance through
   DNA damage response activation [[51]11].

   One of the stem cell markers that is commonly expressed in various
   cancer types, including GBM is cluster of differentiation 44 (CD44), a
   surface adhesion receptor which promotes cancer progression and
   metastasis [[52]14]; its expression in GBM cells is also crucial for
   GBM invasion and migration [[53]15]. CD44-expressing cells were shown
   to escape exogenous DNA damage from radiation-induced double-stranded
   breaks (DSBs); it ultimately promoted tumor recurrence and resistance
   to radiation [[54]16]. Moreover, accumulating studies indicated that
   radiotherapy contributes to abnormalities in cell cycle checkpoints,
   including the G[1]/S and S phases, which may potentially lead to
   resistance to radiation [[55]17,[56]18]. Cyclin B1 (CCNB1) is one of
   the cell cycle-related genes that was reported to be a potential
   biomarker in GBM [[57]19,[58]20]. Overexpression of CCNB1 was
   demonstrated to be associated with higher grades, proliferation, and
   metastasis of GBM [[59]21]. Additionally, CCNB1 was shown to regulate
   cell mitosis at the G[2]/M phase through interacting with
   cyclin-dependent kinase 1 (CDK1) [[60]22]. Thus, CCNB1/CDK1 may be
   potential diagnostic and prognostic markers of GBM
   [[61]23,[62]24,[63]25].

   Mitogen-activated protein kinase 7 (MAPK7) is a member of MAPKs, which
   regulate signaling transduction cascades [[64]26] and are associated
   with multiple cellular processes, such as cell proliferation and
   survival [[65]27,[66]28]. High expression levels of MAPK7 were
   identified in GMB tumors compared to adjacent normal brain tissues. In
   clinical settings, MAPK7 promotes resistance to TMZ and poor patient
   survival; however, its role in GBM still remains to be further
   investigated [[67]29]. Increased expression levels of CCNB1 were
   reported to regulate activation of MAPK7 in the G[2]/M phase, which
   consequently modulates mitosis [[68]28]. Therefore, MAPK7 is a
   potential novel drug target due to its dysregulation and association
   with TMZ resistance in GBM [[69]30,[70]31,[71]32]. Moreover, cell
   division control protein 42 homolog (CDC42) is a protein which is also
   involved in regulating the cell cycle through the G[1] phase.
   Overexpression of CDC42 in GBM was demonstrated to promote tumor cell
   invasion and migration; additionally, CDC42 was associated with low
   survival rates and drug resistance in GBM patients [[72]33,[73]34].
   This therefore suggests crosstalk among CCNB1/CDC42/MAPK7/CD44
   oncogenic signatures in GBM through the cell cycle.

   Integrated bioinformatics analyses have been extensively applied in the
   early stages of drug discovery and development and have significantly
   accelerated the process, as well as reduced costs. Computational
   simulation approaches and molecular structural analyses of
   ligand-protein interactions have contributed to the identification and
   prediction of novel diagnostic and prognostic biomarkers in cancer
   research [[74]35,[75]36]. In this study, we explored Microarray Data
   Extraction and predicted overexpressed and downregulated genes in GBM
   tumors; moreover, we utilized online prediction tools to further
   identify and validate expressions of our genes of interest, the
   CCNB1/MAPK7/CDC42/CD44 oncogenes and also predicted patients’ clinical
   outcomes in GBM under the same settings. To date, only bevacizumab,
   everolimus and TMZ are the common FDA-approved drugs for brain tumor
   treatment [[76]37]. Fortunately, with our innovative lab techniques, as
   mentioned in previous preliminary studies in drug discovery
   [[77]38,[78]39], we synthesized
   9-chloro-6-(piperazin-1-yl)-11H-indeno[1,2-c]quinolin-11-one (SJ10), a
   quinolone and piperazine derivative, with anticancer activities
   ([79]Figure 1). In addition to our earlier studies, quinoline
   derivatives were demonstrated to possess anticancer activities by
   inducing DNA double-strand breaks and apoptosis [[80]39,[81]40].
   Therefore, in this study, we performed drug target predictions and
   identified CCNB1/MAPK7/CDC42/CD44 oncogenic signatures as potential
   drug candidates of SJ10, and further performed ligand-protein binding
   simulations using in silico molecular docking, which validated
   CCNB1/MAPK7/CDC42/CD44 as druggable candidates of SJ10. The
   antiproliferative and cytotoxic effects of SJ10 were evaluated in vitro
   using the US National Cancer Institute (NCI)-60 central nervous system
   (CNS) cell lines to determine responses of single-dose and
   dose-dependent treatments with SJ10 [[82]41].

Figure 1.

   [83]Figure 1
   [84]Open in a new tab

   Rationale of NSC772862 (SJ10) and some of the representative drugs in
   natural products and marketed drugs.

2. Material and Methods

2.1. Dataset Collection

   Gene expression profiles (GEPs) from GBM patient samples were
   downloaded from the Gene Expression Omnibus (GEO) database
   ([85]http://www.ncbi.nlm.nih.gov/gds/, accessed on 4 August 2021), an
   international public repository for high-throughput microarray data
   [[86]42]. (3) different expression profile datasets obtained, viz.,
   [87]GSE4290, [88]GSE68848 and [89]GSE30563 were further analyzed using
   GEO2R ([90]https://www.ncbi.nlm.nih.gov/geo/geo2r/, accessed on 4
   August 2021), an interactive online platform to identify differentially
   expressed genes (DEGs) [[91]43], which was used to identify DEGs
   between GBM tumor samples and normal samples. The Benjamini–Hochberg
   adjustment was made to p values (adj. p) to control the false discovery
   rate (FDR) and maintain the balance between the possibility of
   false-positives and the detection of significant genes. The fold-change
   (FC) threshold was set to 1.5, and adj. p < 0.05 was considered
   statistically significant. Venn diagrams were constructed using the
   Bioinformatics & Evolutionary Genomics (BEG) online tool
   ([92]http://bioinformatics.psb.ugent.be/webtools/Venn/, accessed on 4
   August 2021).

2.2. Identifying Molecular Targets and Therapeutic Classes of SJ10

   Potential SJ10 target genes were predicted using an open-source web
   tool based on the Prediction of Biological Activity Spectra (PASS)
   [[93]44] ([94]http://www.way2drug.com/passonline/predict.php, accessed
   on 17 August 2021). In addition, we explored the Swiss target
   prediction tool ([95]http://www.swisstargetprediction.ch/, accessed on
   17 August 2021), a web-based algorithm that uses the principle of
   similarity to predict drug targets of bioactive small molecules
   [[96]45,[97]46] as an independent tool to further validate the
   predicted potential target genes of SJ10 ([98]Table 1).

Table 1.

   Prediction of Biological Activity Spectra (PASS) and Swiss target genes
   and classes of the SJ10 compound.
   SwissTarget Prediction PASS Prediction Results
   Target Gene Target Class PA PI Activities
   AKT1 Kinase 0.703 0.054 MAP kinase kinase 4 inhibitor
   MAPK8 Kinase 0.652 0.014 Histidine kinase inhibitor
   TTK Kinase 0.598 0.088 Cyclic AMP phosphodiesterase inhibitor
   PIM2 Kinase 0.513 0.024 MAP3K5 inhibitor
   CDK9 Kinase 0.487 0.037 Antineoplastic (glioblastoma multiforme)
   SLC6A3 ECT 0.529 0.091 Protein kinase inhibitor
   EGFR Kinase 0.422 0.004 Focal adhesion kinase inhibitor
   CDK2 Kinase 0.445 0.040 Cyclin B1 inhibitor
   MAPK7 Kinase 0.435 0.037 Apoptosis agonist
   MAPK9 Kinase 0.415 0.021 Protein kinase B gamma inhibitor
   CCNA2 CDK2 Kinase 0.541 0.152 MAP kinase kinase 7 inhibitor
   CCND1 CDK4 Kinase 0.395 0.020 Transcription factor STAT3 inhibitor
   CDK1 CCNB1 Other cytosolic protein 0.407 0.049 T cell inhibitor
   CDK2 CCNA1 CCNA2 Other cytosolic protein 0.407 0.055 Wee-1 tyrosine
   kinase inhibitor
   MAPK1 Kinase 0.353 0.042 CDC42 inhibitor
   MAPK3 Kinase 0.406 0.102 Check point kinase 2 inhibitor
   [99]Open in a new tab

   Pa > Pi, Pa, probability of being active; Pi, probability of being
   inactive.

2.3. DEG Identification by the Tumor Immune Estimation Resource (TIMER)

   Expression profiles of genes showing differential expression between
   GBM tumor and adjacent non-tumor tissues in The Cancer Genome Atlas
   (TCGA) database were analyzed with TIMER
   ([100]https://cistrome.shinyapps.io/timer/, accessed on 9 August 2021),
   a web-based tool for the analysis of interactions between genes of
   interest and immune cells. The relative gene expression level is
   indicated as transcripts per million (TPM) and the expression value was
   normalized by log transformation. Moreover, we explored the Chinese
   Glioma Genome Atlas (CGGA) ([101]http://www.cgga.org.cn/index.jsp,
   accessed on 16 August 2021) to analyze gene expression correlations
   between the two datasets into positive and negative correlations, with
   positive Pearson correlation coefficients and p < 0.05 considered
   statistically significant.

2.4. Validation of DEGs in GBM

   To validate expression levels of identified DEGs in GBM, we explored
   the Human Protein Atlas (HPA) database for immunohistochemistry (IHC)
   ([102]https://www.proteinatlas.org/, accessed on 13 September 2021) to
   compare expression levels between tumor samples and normal samples. The
   HPA database represents the protein expression in 44 major human
   tissues and some cancer tissues by IHC [[103]47]. Statistical analyses
   were performed using the statistical package for social sciences (SPSS)
   vers. 21.0 (Chicago, IL, USA) and the p value was determined using the
   Mann–Whitney U-test. Moreover, for further analysis, the predicted
   genes were validated by an independent bioinformatics tool, the CGGA
   [[104]48].

2.5. Protein-Protein Interaction (PPI) Network Construction and Functional
Enrichment Analysis

   To assess PPIs, we used the search tool for the Retrieval of
   Interacting Genes/Proteins database (STRING,
   [105]https://string-db.org/, accessed on 21 September 2021), a web tool
   developed to analyze interactions of PPIs, such as physical and
   functional associations [[106]49]. Functional enrichments with the
   clustering network were also retrieved from the STRING analysis, and
   they included gene ontology (GO) involving biological processes (BPs)
   and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways, with p <
   0.05 considered significant. For further analysis, we used Network
   Analyst ([107]https://www.networkanalyst.ca/, accessed on 21 September
   2021), a web-based visual analytics platform for comprehensive gene and
   protein expression profiling [[108]50,[109]51]. In this platform, we
   used the SIGnaling Network Open Resource (SIGNOR 2.0) and selected the
   BP database to analyze enriched co-expressed genes.

2.6. Predictions of Patient Clinical Outcomes with Radiomics Signature
Construction

   To predict prognostic outcomes in GBM, radiomics signatures were
   constructed using the GlioVis database
   ([110]http://gliovis.bioinfo.cnio.es/, accessed on 28 August 2021), an
   online portal used for analysis of brain tumor expression [[111]52].
   The distribution of radscores and maximally selected rank statistics of
   DEGs were used to determine the optimum cutoff values for the CCNB1,
   CDC42, MAPK7, and CD44 oncogenes, in order to evaluate overall survival
   (OS).

2.7. Receiver Operating Characteristic (ROC) Curves and Kaplan-Meier (KM)
Analyses Were Used to Validate the Prognostic Values of the CCNB1, CDC42,
MAPK7, and CD44 Oncogenic Signatures in GBM Samples

   To evaluate and validate the diagnostic and prognostic significance of
   CCNB1, CDC42, MAPK7, and CD44 in GBM patients, we used ROC curve, which
   was retrieved from ([112]https://kmplot.com/analysis/, accessed on 1
   September 2021), and further explored the GlioVis database for the KM
   analysis. The ROC curve was based on true positive (sensitivity) and
   false positive (specificity) rates in GBM patients. We evaluated
   whether the test measurement had a specific condition. We assessed the
   area under the curve (AUC), and an AUC of 0.5 indicated no
   discrimination, while an AUC of 1.0 indicated discrimination of the
   curve that includes all possible decision thresholds from a diagnostic
   test result, which were patients who experienced disease onset and
   individuals who did not.

2.8. Evaluation of Drug Likeness, Pharmacokinetics (PKs), and Medicinal
Chemistry of SJ10

   Identifying novel and potential drug candidates in the early stage of
   drug discovery and development is crucial, as it reduces time and
   costs; herein, we applied the drug-likeness concept based on specific
   criteria [[113]53,[114]54]. We explored the SwissADME algorithm
   developed by the Swiss Institute of Bioinformatics
   ([115]http://www.swissadme.ch/index.php, accessed on 2 September 2021),
   and molecular in silico (molsoft) tools
   ([116]https://molsoft.com/mprop/, accessed on 2 September 2021), to
   evaluate the PKs, drug likeness, medicinal chemistry friendliness,
   adsorption, distribution, metabolism, excretion, and toxicity (ADMET)
   properties of SJ10 [[117]55,[118]56]. We analyzed the drug-likeness
   properties according to the Lipinski (Pfizer) rule-of-five), Ghose
   (Amgen), Veber (GSK), and Egan (Pharmacia), and further showed
   relationships between PK and physicochemical properties [[119]57].
   Moreover, we analyzed the gastrointestinal absorption (GIA) and
   brain-penetration properties using the brain or intestinal estimated
   permeation (BOILED-Egg) model [[120]58]. The Abbot Bioavailability
   Score was determined based on the probability of the compound having at
   least 10% oral bioavailability in rats or measurable Caco-2
   permeability [[121]59].

2.9. In Vitro Anticancer Screening of SJ10 against NC1-60 CNS Cells

   SJ10 was submitted to the National Cancer Institute (NCI)-Development
   Therapeutics Program (DTP) to be screened for potential
   antiproliferative and cytotoxic effects against a panel of NCI-60 CNS
   cell lines, in agreement with the outlined protocol of the NCI
   ([122]https://dtp.cancer.gov/, accessed on 11 September 2021). The
   compound was tested at an initial dose of 10 μM. Results showed that
   SJ10 exhibited antiproliferative activities against CNS cell lines.

2.10. Molecular Docking Analysis

   Receptor-ligand interactions were predicted using a molecular docking
   analysis, a technique used to predict the predominant binding ability
   of a ligand with a protein’s three-dimensional (3D) structure
   [[123]60]. To assess possible interactions of SJ10 with target genes
   predicted and selected from the Swiss-target and PASS prediction tools,
   we performed a docking analysis of SJ10 with the CCNB1, CDC42, MAPK7,
   and CD44 oncogenes. For further analysis, we used the Food and Drug
   Administration (FDA)-approved standard inhibitors of CDC42 and MAPK of
   CASIN and BAY-885, respectively. Accordingly, the 3D structure of SJ10
   was assembled with the Avogadro molecular visualization tool [[124]61],
   the 3D structures of CASIN (CID:2882155) and BAY-885 (CID:134128280)
   were downloaded from PubChem as SDF files, and the files were
   subsequently converted to PDB format using PyMol software
   ([125]https://pymol.org/2/, accessed on 6 October 2021). In addition,
   the crystal structures of CCNB1 (PDB:2B9R), CDC42 (PDB: 2ODB), MAPK7
   (PDB:4H3Q), and CD44 (PDB:1UUH) were downloaded from the Protein Data
   Bank (PDB). For further processing, we converted all PDB files to PDBQT
   file format using autodock software
   ([126]http://autodock.scripps.edu/resources/adt, accessed on 6 October
   2021) and, finally, performed docking. To visualize and interpret the
   docking results, we applied BIOVIA discovery studio software for
   analysis [[127]62].

2.11. Statistical Analysis

   Pearson’s correlations were used to assess correlations of
   CCNB1/CDC42/MAPK7/CD44 expressions in GBM cancer types. The statistical
   significance of DEGs was evaluated using the Wilcoxon test. * p < 0.05
   was accepted as being statistically significant.

3. Results

3.1. Identification of DEGs in GBM

   Gene expression profiles (GEPs) from GBM samples and normal brain
   samples tallied from different studies were extracted from the
   microarray dataset. The analytical results showed that 100, 256, and 30
   GBM samples and normal samples were, respectively, obtained from the
   [128]GSE4290, [129]GSE68846, and [130]GSE30563 datasets. Further
   analysis with a Venn diagram demonstrated 87 overlapping upregulated
   genes from the three datasets ([131]Figure 2A) and 50 overlapping
   downregulated genes from the same database ([132]Figure 2B). Moreover,
   [133]Figure 1 is the heatmap of overexpressed overlapping genes.
   [134]Figure 2D–F shows volcano plots of GBM tumor samples compared to
   normal samples, the volcano plot revealed the statistical significance
   of the difference between tumor and normal samples through −10 log and
   −2 log fold change, respectively. The p-value in the volcano plot was
   used to indicate threshold indicators for adjusted p-values, which was
   further used to show all the genes that are statistically
   differentially-expressed with an adjusted p value threshold of 0.05
   considered significant.

Figure 2.

   [135]Figure 2
   [136]Open in a new tab

   Differentially-expressed genes (DEGs) in glioblastoma multiforme (GBM)
   extracted from the [137]GSE4290, [138]GSE68846, and [139]GSE30563
   microarray datasets. (A) Venn diagram of 87 selected overexpressed
   overlapping DEGs. (B) Venn diagram of 50 selected downregulated
   overlapping DEGs. (C) is the heatmap of overexpressed overlapping
   genes. (D–F) Volcano plots of DEGs from the [140]GSE4290, [141]GSE68848
   and [142]GSE30563 datasets with red and blue dots respectively
   representing upregulated and downregulated genes (at p < 0.05).

3.2. Evaluation of Drug Likeness, PKs, and Medicinal Chemistry of the SJ10
Compou

   We explored the SwissADME algorithm developed by the Swiss Institute of
   Bioinformatics and molecule in silico (molsoft) to evaluate the PKs,
   drug likeness, medicinal chemistry friendliness, adsorption,
   distribution, metabolism, excretion, and toxicity (ADMET) properties of
   SJ10 [[143]55,[144]56]. We analyzed the drug-likeness properties
   according to the Lipinski (Pfizer) rule-of-five)), Ghose (Amgen), Veber
   (GSK), and Egan (Pharmacia), and further showed relationships between
   the PK and physicochemical properties [[145]57]. Moreover, we analyzed
   the GIA and brain-penetration properties, using the brain or intestinal
   estimated permeation (BOILED-Egg) model [[146]58]. The Abbot
   Bioavailability Score was determined based on the probability of the
   compound having at least 10% oral bioavailability in rats or measurable
   Caco-2 permeability [[147]59] ([148]Figure 3).

Figure 3.

   [149]Figure 3
   [150]Open in a new tab

   SJ10 passed the required physicochemical properties, medicinal
   chemistry, pharmacokinetics (PK), and drug-likeness criteria. (A,B)
   Structure of the SJ10 (NSC772862) small molecule, bioavailability radar
   (BA), displaying the six physicochemical properties of absorption
   including lipophilicity (XLOGP3 = 3.90), molecular weight (349.10
   g/mol), polarity (PSA = 37.08 Å^2), solubility (Log S (ESOL) = −4.7),
   flexibility (rotation = 4), saturation (fraction Csp3 = 0.2), and pKa
   of the most basic or acidic group (= 0.5) of the SJ10 compound. In
   addition, the SJ10 compound demonstrated a highly-probable GIA
   absorption, bioavailability score (55%) and good synthetic
   accessibility (2.89). (C) The compound passed the blood–brain barrier
   (BBB) with a score of 4.98, and further displayed a drug-like model
   score of −0.68. A structural characterization of the compounds was done
   with the help of spectroscopic studies including IR, proton NMR, 13C
   NMR, MS, and elemental analysis (D).

3.3. CCNB1/CDC42/MAPK7/CD44 Oncogenic Signatures Are Overexpressed in GBM

   A bioinformatics analysis through TIMER online web tool with default
   settings showed significantly increased messenger (m)RNA levels of
   CCNB1/CDC42/MAPK7/CD44 in pan cancers, including GBM tumor tissues
   compared to normal tissues from TCGA) ([151]Figure 4A–D). Relative gene
   expression levels are indicated as transcripts per million (TPM) and
   the expression value was normalized by log transformation of the
   statistical significance, as evaluated by the Wilcoxon test, with p
   value significant codes: 0 ≤ *** < 0.001 ≤ ** < 0.01 ≤ * < 0.05 ≤ . <
   0.1. We further explored the CGGA tool with default settings to
   investigate correlations among the CCNB1/CDC42/MAPK7/CD44 oncogenes.
   When all four genes were combined for analysis, the predicted results
   showed positive correlations ranging r = 0.43~0.67 of CCNB1 with CDC42,
   CCNB1 with MAPK7, CCNB1 with CD44, and MAPK7 with CD44 in GBM patients
   ([152]Figure 4E–H), with positive Pearson correlation coefficients and
   p < 0.05 considered statistically significant.

Figure 4.

   [153]Figure 4
   [154]Open in a new tab

   CCNB1/CDC42/MAPK7/CD44 oncogenic signatures are overexpressed in
   glioblastoma multiforme (GBM). (A–D) Increased mRNA levels of
   CCNB1/CDC42/MAPK7/CD44 in pan cancers, including GBM tumor tissues,
   compared to normal tissues from The Cancer Genome Atlas (TCGA). The
   relative gene expression level is indicated as transcripts per million
   (TPM), and expression values were normalized by log transformation of
   the statistical significance as evaluated by the Wilcoxon test, with p
   value significant codes: 0 ≤ *** < 0.001 ≤ ** < 0.01 ≤ * < 0.05 ≤. <
   0.1. (E–H) Correlation analysis of CCNB1/CDC42/MAPK7/CD44 oncogenes
   revealed correlations among all four genes when combined for analysis.
   Predicted results showed positive correlations ranging r = 0.43~0.67 of
   CCNB1 with CDC42, CCNB1 with MAPK7, CCNB1 with CD44, and MAPK7 with
   CD44 in GBM samples, with positive Pearson correlation coefficient and
   p < 0.05 considered statistically significant.

3.4. Validation of CCNB1/CDC42/MAPK7/CD44 Oncogenic Signature Expressions in
GBM

   To validate expression levels of the CCNB1/CDC42/MAPK7/CD44 gene
   signatures in GBM, we explored the HPA database for IHC to compare gene
   expression levels between GBM tumor tissues and normal samples. CCNB1
   displayed medium staining, with strong intensity and quantity (25%)
   ([155]Figure 5A), while CDC42 displayed medium staining, with moderate
   intensity and quantity (75%) ([156]Figure 5B), and MAPK7 and CD44
   displayed high staining with strong intensity and quantity (75%)
   ([157]Figure 5C,D) in GBM tissues as compared to normal tissues. For
   further analysis, the GlioVis database showed increased mRNA expression
   levels of CCNB1/CDC42/MAPK7/CD44 oncogenes in GBM tissues compared to
   non-tumor tissues ([158]Figure 5E–H). In addition, we explored the
   CGGA, an independent glioma database, and validated expressions of the
   CCNB1/CDC42/MAPK7/CD44 gene signatures in WHO grade II, III, and IV GBM
   tumors using the Analysis of variance (ANOVA) ([159]Figure 5I–L), with
   p < 0.05 considered statistically significant.

Figure 5.

   [160]Figure 5
   [161]Open in a new tab

   Increased expressions of CCNB1/CDC42/MAPK7/CD44 oncogenic signatures in
   glioblastoma. (A) CCNB1 displayed medium IHC staining, with strong
   intensity and quantity (25%). (B) CDC42 displayed medium IHC staining,
   with moderate intensity and quantity (75%). (C,D) MAPK7 and CD44
   displayed high IHC staining with strong intensity and quantity (75%),
   in GBM tissues as compared to normal tissues. (E–H) Increased mRNA
   expression levels of CCNB1/CDC42/MAPK7/CD44 oncogenes in GBM tissues
   compared to non-tumor tissues from a GlioVis database analysis. (I–L)
   Expressions of CCNB1/CDC42/MAPK7/CD44 gene signatures in WHO grade II,
   III, and IV GBM tumors using the Analysis of variance (ANOVA), with p <
   0.05 considered statistically significant in all datasets. All images
   can be found online.

3.5. Immunofluorescent (IF) Staining of the U251-MG GBM Human Cell Line

   To further validate expressions of the CCNB1/CDC42/MAPK7/CD44 genes in
   GBM, we explored HPA IF staining, using the U251-MG GBM cell line. The
   following antibodies were used for staining: CCNB1 (HPA030741), CDC42
   (CAB004360), MAPK7 (CAB018561), and CD44 (CAB000112). Staining results
   of the U251-MG cell line exhibited the location of genes, with
   antibodies shown in green, nuclei in blue, and microtubules in red.
   CCNB1 was localized in the cytosol, CDC42 was detected in microtubules,
   while the localization of MAPK7 was in the nucleoplasm and CD44 was
   found in plasma membranes ([162]Figure 6).

Figure 6.

   [163]Figure 6
   [164]Open in a new tab

   HPA staining results of the U251-MG cell line exhibited the locations
   of genes. Antibodies are shown in green, nuclei in blue, and
   microtubules in red. (A) CCNB1 was localized in the cytosol, (B) CDC42
   was detected in microtubules, (C) while the localization of MAPK7 was
   in the nucleoplasm, and (D) CD44 was found in plasma membranes. All
   images are available online.

3.6. PPI Network Construction and Functional Enrichment Analysis

   To assess PPIs, we used the STRING database
   ([165]https://string-db.org/, accessed on 21 September 2021), a web
   tool developed to analyze interactions of PPIs, such as physical and
   functional associations. The clustering analysis had nine nodes and 15
   edges, with an average local clustering coefficient of 0.917 and a PPI
   enrichment p value of 0.0293. Moreover, the interaction score
   confidence was set to > 0.4, and considered most significant. Active
   interactions were based on text mining, experiments, databases,
   co-expressions, neighborhood, gene fusion and co-occurrence
   ([166]Figure 7A). Functional enrichments with the clustering network
   were also retrieved from the STRING analysis, and they included gene
   ontology (GO) involving BPs and KEGG pathways, with p < 0.05 considered
   significant ([167]Figure 7B,C). For further analysis, we used Network
   Analyst ([168]https://www.networkanalyst.ca/, accessed on 21 September
   2021), a web-based visual analytics platform for comprehensive gene and
   protein expression profiling [[169]50,[170]51]. In this platform, we
   used the SIGnaling Network Open Resource (SIGNOR 2.0) and selected the
   BP database to analyze enriched co-expressed genes ([171]Figure 7D).
   The signaling network analysis of KEGG pathway enrichment showed
   co-expressions of CCNB1/CDC42/MAPK7/CD44 oncogenes in the same network
   cluster, and results were viewed from the network topology in a force
   atlas layout analyzed from the Igraph R package ([172]Figure 7E).

Figure 7.

   [173]Figure 7
   [174]Open in a new tab

   Protein–protein interaction (PPI) network revealed interactions among
   the CCNB1/CDC42/MAPK7/CD44 oncogenes in glioblastoma multiforme (GBM).
   (A) The clustering network consisted of nine nodes and 15 edges, with
   an average local clustering coefficient of 0.917 and a PPI enrichment p
   value of 0.0293. Moreover, the interaction score confidence was set to
   >0.4, and p < 0.05 was considered statistically significant. Active
   interactions were based on text mining, experiments, databases,
   co-expressions, neighborhood, gene fusion, and co-occurrence. (B) The
   top biological processes (BPs), (C) KEGG pathways, and (D) Signaling
   network analysis from the BP database, showing that co-expressions of
   the CCNB1/CDC42/MAPK7/CD44 oncogenes displayed enrichment in the cell
   cycle, regulation of molecular function, positive regulation of
   cellular processes, regulation of protein metabolic processes, and
   protein phosphorylation among others (red bubble). (E) Signaling
   network analysis of the KEGG pathway enrichment analysis showed
   co-expression of CCNB1/CDC42/MAPK7/CD44 oncogenes in the same network
   cluster.

3.7. Predictions of Patient Clinical Outcomes with Radiomics Signature
Construction

   Prognostic outcomes of GBM patients were predicted by exploring
   radiomics signatures constructed using the GlioVis database, and
   distributions of Radscores and maximally selected rank statistics were
   used to determine optimal cutoff values for the CCNB1, CDC42, MAPK7,
   and CD44 oncogenes. The obtained cutoff scores (Radscores) were 2.83,
   6.62, 3.9, and 3.48, respectively ([175]Figure 8A–H). This analysis
   therefore showed that patients with lower Radscores generally displayed
   better OS; however, since the CCNB1/CDC42/MAPK7/CD44 oncogenes were
   shown to be highly expressed in GBM, herein, they also exhibited high
   Radscores and, consequently, worse prognoses. Therefore, predicted
   expressions of the CCNB1, CDC42, MAPK7 and CD44 oncogenes exhibited
   significant roles in the cell cycle, and thus are potential prognostic
   biomarkers in GBM.

Figure 8.

   [176]Figure 8
   [177]Open in a new tab

   Optimal cutoff score calculations of Radscores of CCNB1, CDC42, MAPK7,
   and CD44 expressions in glioblastoma multiforme (GBM). (A,B) Radscores
   plot of CCNB1 with a cutoff value of 2.83. (C,D) Radscores plot of
   CDC42 with a cutoff value of 6.62. (E,F) Radscore plot of MAPK7 with a
   cutoff value of 3.9. (G,H) Radscores plot of CD44 with a cutoff value
   of 3.48. The low radscores are indicated in blue and high radscores are
   indicated in red. This analysis shows that patients with lower
   Radscores generally displayed better overall survival. Therefore,
   predicted expressions of the CCNB1, CDC42, MAPK7, and CD44 oncogenes
   exhibited significant roles in the cell cycle, and are thus potential
   prognostic biomarkers for GBM.

3.8. High Expressions of CCNB1, CDC42, MAPK7, and CD44 Were Associated with a
Poor Prognosis in GBM

   To evaluate and validate prognostic significant values of CCNB1, CDC42,
   MAPK7, and CD44 in GBM patients, we used an ROC curve and KM analysis.
   The ROC curve was based on true (sensitive) and false (selective)
   positive rates of responses in GBM patients. AUC scores of CCNB1,
   CDC42, MAPK7, and CD44 were 0.534, 0.515, 0.541, and 0.538,
   respectively ([178]Figure 9A–D). The KM analysis and log-rank test
   showed significantly prolonged OS times in the low-risk group compared
   to the high-risk group, with each subtype displaying different cutoff
   values; p < 0.05 was considered statistically significant ([179]Figure
   9E–H). This indicated that the CCNB1, CDC42, MAPK7, and CD44 oncogenic
   signatures possessed potential diagnostic abilities in GBM. To evaluate
   whether the test measurements had specific conditions, we assessed the
   AUC, and an AUC of 0.5 indicated no discrimination, while an AUC of 1.0
   indicated discrimination. The curve that included all possible decision
   thresholds from a diagnostic test result were patients who had
   experienced disease onset and individuals who had no thresholds from a
   diagnostic test result, which were patients who had experienced disease
   onset and individual thresholds.

Figure 9.

   [180]Figure 9
   [181]Open in a new tab

   High expression levels of the CCNB1, CDC42, MAPK7, and CD44 oncogenes
   were associated with a poor prognosis in glioblastoma multiforme (GBM).
   (A–D) Time-dependent ROC analysis according of the true positive
   (sensitivity) and false positive (specificity) rates of survival,
   assessed by the prognostic accuracy based on AUC values. CCNB1 (AUC:
   0.552), CDC42 (AUC: 0.566), MAPK7 (AUC: 0.536), and CD44 (AUC: 0.565).
   An AUC of 0.5 indicates no discrimination, while an AUC of 1.0
   indicates discrimination. (E–H) Kaplan–Meier analysis predicted a
   significant prolonged overall survival time in the low-risk group
   compared to the high-risk group. The analysis was based on the optimal
   cutoff point from the low- and high-risk groups, and p < 0.05 was
   considered significant.

3.9. In Vitro Anticancer Screening of SJ10 against NC1-60 CNS Cell Lines

   SJ10 was submitted to the NCI-DTP for screening for potential
   antiproliferative and cytotoxic effects against a panel of NCI-60 CNS
   cell lines, in agreement with the outlined protocol of the NCI. The
   compound was tested at an initial dose of 10 μM. Results showed that
   SJ10 exhibited antiproliferative activities against several CNS cell
   lines. The compound growth inhibition (GI) percentage showed that
   SNB-19 cells were more sensitive, with GI of 67.25%, followed by SF-539
   at 36.91%, SF-268 at 34.33%, SF-295 at 23.75%, and SNB-75 at 22.34%, as
   shown in [182]Figure 10A. The compound was further evaluated with
   dose-dependent treatment, since it exhibited antiproliferative
   activities at an initial dose of 10 μM. Accordingly, SF-268 displayed
   complete growth inhibition (−100%), followed by U251 at −96%, SNB-75 at
   −84%, SF-539 at −79%, SF-295 at −76%, and SNB-19 at −42%.
   Sulforhodamine B (SRB) dual-pass staining was used to further
   investigate the in vitro 50% growth inhibition (GI[50])/50% inhibitory
   concentration (IC[50]), and results ranged 1.14~2.15 μM in the CNS cell
   lines, with SNB-75 more sensitiv at 1.14 μM, followed by U251 at 1.59
   μM, SF-268 at 1.64 μM, SNB-19 at 1.67 μM, SF-539 at 1.69 μM, and SF-295
   at2.15 μM, showing a smaller response to SJ10 ([183]Figure 10B,C).

Figure 10.

   [184]Figure 10
   [185]Open in a new tab

   In vitro anticancer screening of SJ10 against NC1-60 CNS cell lines (A)
   The initial dose was 10 μM. Results showed that SJ10 exhibited
   antiproliferative activities against various CNS cell lines. The
   compound growth inhibition (GI) percentage showed that SNB-19 was more
   sensitive, with GI of 67.25%, followed by SF-539 at 36.91%, SF-268 at
   34.33%, SF-295 at 23.75%, and SNB-75 at 22.34%. (B) Dose-dependent
   treatment results. SF-268 displayed complete growth inhibition at
   −100%, followed by U251 at −96%, SNB-75 at −84%, SF-539 at −79%, SF-295
   at −76%, and SNB-19 at −42%. (C) SRB dual-pass staining was used to
   further investigate the in vitro GI[50]/IC[50], and results ranged
   1.14~2.15 μM for the CNS cell lines, with SNB-75 cells more sensitive
   at 1.14 μM, followed by U251 at 1.59 μM, SF-268 at 1.64 μM, SNB-19 at
   1.67 μM, SF-539 at 1.69 μM, and SF-295 less responsive at 2.15 μM.

3.10. Molecular Docking Analysis

   The potential inhibitory effects of SJ10 were assessed using a docking
   analysis. Results of receptor-ligand interactions obtained from the
   Autodock tool revealed putative binding affinities of SJ10 with CCNB1
   (−7.9 kcal/mol), CDC42 (−7.8 kcal/mol), MAPK7 (−8.4 kcal/mol), and CD44
   (−7.0 kcal/mol). When compared to standard inhibitors of CASIN (CID:
   2882155) for CDC42 and BAY-885 (CID: 134128280) for MAPK7, they showed
   lower binding energies of −7.4 and −7.3 kcal/mol, respectively. For a
   further analysis, we used Pymol and Discovery Studio to visualize the
   analytical results. The SJ10/CCNB1 complex displayed interactions by
   conventional hydrogen (H) bonds with ALA128 (2.07 Å) and ARG68 (2.73
   Å). The interactions were stabilized by van der Waals interactions
   (ASN130, LEU129, PHE131, GLY132, PHE131, ASN130, GLY134, and PRO136),
   pi-sigma (GLY132), and pi-alkyl (LEU17, LEU17, and ARG135) displayed in
   their binding pockets. The SJ10/CDC42 complex displayed van der Waals
   interactions (THR25, PHE28, SER30, THR17, and TYR40), pi-sigma (ILE21),
   and pi-alkyl (PHE18, LYS27, and PRO29) in their binding pockets, while
   SJ10/MAPK7 exhibited conventional hydrogen bond with SER153 (2.23 Å)
   and was further stabilized by van der Waals interactions (LYS114,
   GLY34, TRP192, and THR193), carbon hydrogen bond (GLU33), pi-sigma
   (THR190), pi-pi stacked (TYR113), and pi-alkyl (PRO152 and LYS151) in
   their binding pockets. Interactions between the SJ10/CD44 complex
   displayed van der Waals interactions (THR102, GLY103, ARG90, LEU70,
   TYR79, SER71, ILE96, and ARG78), carbon hydrogen bond (CYS77), pi-pi
   T-shaped (TYR42) and pi-alkyl (ILE91) in their binding pockets
   ([186]Figure 11). For further analysis, we used the FDA approved
   standard inhibitors of CDC42 and MAPK7, CASIN and BAY-885 respectively.
   The interaction of CDC42 in complex with CASIN exhibited binding energy
   of (−7.4 kcal/mol) and MAPK7 in complex with BAY-885 displayed binding
   energy of (−7.3 kcal/mol), these results exhibited a much lower binding
   affinities as compared to SJ10, this suggesting the potential
   inhibitory effects of SJ10 in GBM expression CCNB1, CDC42, MAPK7, and
   CD44 oncogenic signatures ([187]Figure 12, [188]Table 2).

Figure 11.

   [189]Figure 11
   [190]Open in a new tab

   In silico docking results of SJ10 in complex with the CCNB1, CDC42,
   MAPK7, and CD44 oncogenes in 2D representations. (A) The CCNB-SJ10
   complex exhibited a putative binding energy of −7.9 kcal/mol, and
   displayed interactions by conventional H-bonds (green) with ALA128 and
   ARG68, and short binding distances of 2.07 and 2.73 Å, respectively.
   (B) The CDC42-SJ10 complex showed a binding energy of −7.8 kcal/mol,
   and displayed van der Waals interactions (THR25, PHE28, SER30, THR17,
   and TYR40), pi-sigma (ILE21), and pi-alkyl (PHE18, LYS27, and PRO29) in
   their binding pockets. (C) The MAPK7-SJ10 complex displayed a unique
   binding energy of −8.4 kcal/mol, and further showed conventional
   hydrogen bonds (SER153), with a shorter binding distance of 2.23 Å. (D)
   The CD44-SJ10 complex showed a binding energy of −7.8 kcal/mol, and
   exhibited van der Waals interactions (THR102, GLY103, ARG90, LEU70,
   TYR79, SER71, ILE96, and ARG78), carbon hydrogen bonds (CYS77), pi-pi
   T-shaped (TYR42), and pi-alkyl (ILE91) in their binding pockets.

Figure 12.

   [191]Figure 12
   [192]Open in a new tab

   In silico docking results CDC42 and MAPK7 with standard inhibitors (A)
   interaction of CDC42 in complex with CASIN exhibited binding energy of
   (−7.4 kcal/mol). (B) MAPK7 in complex with BAY-885 displayed binding
   energy of (−7.3 kcal/mol), in 2D representations.

Table 2.

   Analytical summary table showing interactions of SJ10 with
   CCNB1/CDC42/MAPK7/ CD44 oncogenes.
   SJ10-CCNB1 Complex (=−7.9 kcal/mol) SJ10-CDC42 Complex (=−7.6 kcal/mol)
   Type of interactions and number of bonds distance of interacting Amino
   acids Type of interactions and number of bonds distance of interacting
   Amino acids
   Conventional Hydrogen bond (2) ALA128 (2.07 Å) and ARG68 (2.73Å) Van
   der Waals forces THR25, PHE28, SER30, THR17, and TYR40
   Van der Waals forces ASN130, LEU129, PHE131, GLY132, PHE131, ASN130,
   GLY134, and PRO136 Pi-Sigma ILE21
   Pi-Sigma GLY132 Pi-alkyl PHE18, LYS27, PRO29
   Pi-Alkyl LEU17, LEU17, and ARG135
   SJ10-MAPK7 Complex (=−8.4 kcal/mol) SJ10-CD44 Complex (=−7.0 kcal/mol)
   Type of interactions and number of bonds distance of interacting Amino
   acids Type of interactions and number of bonds distance of interacting
   Amino acids
   Conventional Hydrogen bond (1) SER153 (2.23 Å) Van der Waals forces
   THR102, GLY103, ARG90, LEU70, TYR79, SER71, ILE96, and ARG78
   Van der Waals forces THR102, GLY103, ARG90, LEU70, TYR79, SER71, ILE96,
   and ARG78 TRP192, THR193 Carbon hydrogen bond CYS77
   Carbon hydrogen bond CYS77 pi-pi T-shaped TYR42
   Pi-sigma THR190 Pi-Alkyl ILE91
   Pi-Alkyl ILE91
   Pi-Pi stacked TYR113
   Pi-alkyl PRO152 and LYS151
   [193]Open in a new tab

4. Discussion

   Despite improvements in standard therapies, including surgical
   resection, radiation, and chemotherapy with TMZ, patients with GBM
   still exhibit poor clinical outcomes, with a median survival of only
   about 15 months [[194]63], mainly due to GBM’s biological and genetic
   heterogeneity. Therefore, understanding molecular mechanisms and
   invasive characteristics of GBM is pivotal as an essential strategy for
   developing more-effective therapeutics. Integrated bioinformatics
   analyses have been extensively applied in the early stages of drug
   discovery and development and have significantly accelerated the
   process and reduced costs. In the current study, we applied
   computational simulation analyses to predict and identify dysregulated
   gene networks and pathways leading to resistance to radio- and
   chemotherapies (TMZ). Accumulating studies have demonstrated that
   therapeutic resistance in GBM is also associated with GSCs, which may
   potentially assist GBM cancer cells escape irradiation. Others have
   shown that GCSs are resistant to TMZ chemotherapy, thus promoting
   radioresistance through DNA damage-response activation
   [[195]11,[196]12,[197]13].

   SJ10 (NSC7772862) is a small molecule and a derivative of a quinolone
   and piperazine derivative and was recently synthesized in our
   laboratory. Through application of the Swisstarget and PASS prediction
   tools ([198]Table 1), we predicted CCNB1, CDC42, MAPK7, and CD44
   oncogenic signatures as target genes for SJ10. Moreover, we explored
   the SwissAMDE and molsoft algorithms to evaluate the PK, drug-likeness,
   medicinal chemical friendliness, and ADMET properties of SJ10
   [[199]55,[200]56]. The compound successfully passed the required
   physicochemical properties, medicinal chemistry, PK, and drug-likeness
   criteria. Bioavailability radar, displaying the six physicochemical
   properties of absorption-included lipophilicity (XLOGP3 = 3.90),
   molecular weight (349.10 g/mol), polarity (PSA = 37.08 Å^2), solubility
   (Log S (ESOL) = −4.7), flexibility (rotation = 4), saturation (Fraction
   Csp3 = 0.2), and pKa (=0.5) of the SJ10 compound. In addition, the SJ10
   compound demonstrated highly probable GIA absorption, a bioavailability
   score (55%), and good synthetic accessibility (2.89). The compound
   reached the BBB with a score of 4.98, and further displayed a drug-like
   model score of (−0.68) ([201]Figure 2).

   We identified significantly increased mRNA levels of the
   CCNB1/CDC42/MAPK7/CD44 oncogenes in pan cancers, including GBM tumor
   tissues compared to normal tissues from TCGA, using the TIMER
   bioinformatics tool. These results were further validated using the HPA
   and GlioVis database analyses, which showed similar outputs displaying
   overexpression of CCNB1/CDC42/MAPK7/CD44 gene signatures in WHO grade
   II, III, and IV GBM tumors using an ANOVA. For further analysis, we
   used the STRING online web tool and showed that CCNB1/CDC42/MAPK7/CD44
   actively interacted with each other in the same clustering network,
   based on text mining, experiments, databases, co-expressions,
   neighborhood, gene fusion, and co-occurrence and also exhibited
   enrichment of GO involving BPs and (KEGG pathways, with p < 0.05
   considered significant ([202]Figure 6B,C). Furthermore, we predicted
   patients’ clinical outcomes using the Radiomics signature constructed
   from the GlioVis database, to determine optimal cutoff values for the
   CCNB1, CDC42, MAPK7, and CD44 oncogenes. The obtained cutoff scores
   (Radscore) were 2.83, 6.62, 3.9, and 3.48, respectively ([203]Figure
   7). The analysis therefore showed that patients with lower Radscores
   generally displayed better OS; however, since the
   CCNB1/CDC42/MAPK7/CD44 oncogenes were shown to be highly expressed in
   GBM, herein, they also exhibited high Radscores, which consequently led
   to worse prognoses. Therefore, predicted expressions of the CCNB1,
   CDC42, MAPK7, and CD44 oncogenes exhibited significant roles in the
   cell cycle, and thus are potential prognostic biomarkers for GBM. In
   addition, MAPK7 is a potential novel drug target due to its
   dysregulation and association with TMZ resistance in GBM. Herein, we
   showed that targeting MAPK7 in GBM tumors can potentially improve the
   strength of TMZ in suppressing tumor cells [[204]30,[205]31,[206]32].

   The potential anticancer activities of SJ10 were evaluated against NCI
   human CNS cell lines. Accordingly, an initial dose of 10 μM exhibited
   antiproliferative activities against the CNS cell lines, as shown in
   [207]Figure 9A. The compound was further evaluated with dose-dependent
   treatment, since it exhibited antiproliferative activities at an
   initial dose of 10 μM. Accordingly, SJ10 displayed complete growth
   inhibition at −100% against SF-268 cells, followed by U251 at −96%,
   SNB-75 at −84%, SF-539 at −79%, SF-295 at −76%, and SNB-19 at −42%. SRB
   dual-pass staining was used to further investigate in vitro
   GI[50]/IC[50] values, and results ranged 1.14~2.15 μM in the CNS cell
   lines, with SNB-75 more sensitive at 1.14 μM, followed by U251 at 1.59
   μM, SF-268 at 1.64 μM, SNB-19 at 1.67 μM, SF-539 at 1.69 μM, and SF-295
   at 2.15 μM, showing a weaker response to SJ10. Finally, we evaluated
   the potential inhibitory effects of SJ10, which were assessed using a
   docking analysis. The results of receptor-ligand interactions obtained
   from the Autodock tool revealed higher binding energies of SJ10 with
   CCNB1 (−7.9 kcal/mol), CDC42 (−7.8 kcal/mol), MAPK7 (−8.4 kcal/mol),
   and CD44 (−7.0 kcal/mol) compared to the standard inhibitors of CASIN
   (CID: 2882155) for CDC42 and BAY-885 (CID: 134128280) for MAPK7, which
   showed lower respective binding energies of −7.4 and −7.3 kcal/mol
   (Fig.12). Therefore, the above-mentioned results suggest that SJ10
   exhibits drug-like characteristics, with anticancer activities and is a
   potential oral drug candidate. Further in vitro and in vivo studies are
   both currently in progress in our laboratory.

5. Conclusions

   In summary, our obtained results showed that CCNB1/CDC42/MAPK7/CD44
   oncogenic signatures are potential biomarkers of GBM
   therapeutic-resistant tumors, and potential drug targets of our novel
   small molecule, SJ10. We further showed that SJ10 exhibits anticancer
   activities against a panel of NCI human CNS cancer cell lines when
   administered at an initial dose of 10 μM and also in a dose-dependent
   manner. We evaluated receptor-ligand interactions using a docking
   analysis and identified unique and higher binding energies of SJ10 in
   complex with the CCNB1, CDC42, MAPK7, and CD44 oncogenes, compared to
   their interactions with two FDA-approved inhibitors. Further in vitro
   and in vivo studies are both currently in progress in our laboratory.

Acknowledgments