Pharmacoinformatics and Preclinical Studies of NSC765690 and NSC765599, Potential STAT3/CDK2/4/6 Inhibitors with Antitumor Activities against NCI60 Human Tumor Cell Lines

Signal transducer and activator of transcription 3 (STAT3) is a transcriptional regulator of a number of biological processes including cell differentiation, proliferation, survival, and angiogenesis, while cyclin-dependent kinases (CDKs) are a critical regulator of cell cycle progression. These proteins appear to play central roles in angiogenesis and cell survival and are widely implicated in tumor progression. In this study, we used the well-characterized US National Cancer Institute 60 (NCI60) human tumor cell lines to screen the in vitro anti-cancer activities of our novel small molecule derivatives (NSC765690 and NSC765599) of salicylanilide. Furthermore, we used the DTP-COMPARE algorithm and in silico drug target prediction to identify the potential molecular targets, and finally, we used molecular docking to assess the interaction between the compounds and prominent potential targets. We found that NSC765690 and NSC765599 exhibited an anti-proliferative effect against the 60 panels of NCI human cancer cell lines, and dose-dependent cytotoxic preference for NSCLC, melanoma, renal, and breast cancer cell lines. Protein–ligand interactions studies revealed that NSC765690 and NSC765599 were favored ligands for STAT3/CDK2/4/6. Moreover, cyclization of the salicylanilide core scaffold of NSC765690 mediated its higher anti-cancer activities and had greater potential to interact with STAT3/CDK2/4/6 than did NSC765599 with an open-ring structure. NSC765690 and NSC765599 met the required safety and criteria of a good drug candidate, and are thus worthy of further in-vitro and in-vivo investigations in tumor-bearing mice to assess their full therapeutic efficacy.


Introduction
Despite advances in biomedical research, cancer remains a public health concern and is currently ranked the second leading cause of global mortality [1,2]. The etiology of cancer is often multifactorial, involving an interplay between genetic and epigenetic factors which amount to dysregulation of molecular networks, proteins, RNA, and DNA in favor of cell growth and proliferation [3,4]. The survival, growth, and metastasis of tumor cells depend on cellular differentiation, proliferation, angiogenic, and apoptotic mechanisms [5], which are controlled by a range of protein kinases and signal transduction pathways [6,7]. Cyclin-dependent kinases (CDKs) are a family of serine/threonine protein kinases with catalytic and regulatory subunits [8]. Out of the nine CDKs so far identified, four (CDK1, CDK2, CDK4, and CDK6) play important regulatory roles in the cell cycle; CDK4/D-cyclin, CDK2/E-cyclin, and CDK6/D-cyclin regulate the G 1 to S phase of the cell cycle transition, while the complex of CDK1 or CDK2 with cyclin A regulates the S to G 2 transition [9]. Under physiological conditions, the catalytic and regulatory (cyclin) subunits of CDKs remain dissociated. However, periodic complexation of a cyclin with its catalytic unit leads to its activation and phosphorylation of a variety of downstream target proteins required for cell cycle progression [10,11]. CDKs regulate cell cycle transitions via phosphorylation and subsequent inactivation of the retinoblastoma (Rb) protein, a tumor suppressor that prevents cell cycle transition [12]. Thus, the inactivation of this Rb protein allows the free flow and progression of cells into the cell cycle, leading to multi-cell cycles, cell proliferation, and eventual development into cancer cells [12].
A number of key oncogenic abnormalities, including amplification of cyclin D; inactivation of CDKN2A (p16); and deletions/mutations upstream of cyclin D, such as activating mutations of phosphatidylinositol 4,5-bisphosphate 3-kinase catalytic subunit alpha (PIK3CA)/the B-raf proto-oncogene, serine/threonine kinase (BRAF), and phosphatase and tensin homolog (PTEN) deletion [13,14], were identified as contributing factors to hyperactive CDKs and consequently deregulated the cell cycle. Therefore, since dysfunctional cell cycle regulation via oncogenic aberrations of CDKs is a hallmark of all human cancers [15], pharmacological targeting of CDKs will undoubtedly affect cancer proliferation and survival [16,17]. Hence, CDK inhibitors have been developed and evaluated; however, disappointingly, while first-generation inhibitors of CDK were non-selective and besieged with toxicity [18,19], second-generation CDK4/6 inhibitors, although showing promising outcomes, are plagued with acquired resistance, which develops in almost all cases [20] due to the activation of other oncogenic pathways, including c-Myc, signal transducer and activator of transcription 3 (STAT3), and phosphatidylinositol 3-kinase (PI3K)/AKT/mammalian target of rapamycin (mTOR) signaling pathways [21,22].
STAT3 is a cytoplasmic transcription factor involved in a number of biological processes including cell differentiation, proliferation, survival, and angiogenesis [23]. Overexpression of STAT3 is associated with poor clinical prognoses of cancer patients [24]. It is therefore not surprising that the STAT3 signaling axis has long been explored in cancer therapy owing to its roles in tumor formation, metastasis, and therapeutic failure [25,26]. Preclinical studies revealed that aberrant STAT3 expression mediates immunosuppression of tumor cells [27,28], while inhibition of STAT3 re-sensitizes therapy-resistance breast cancer cells to palbociclib treatment [29]. Summing up the above clinical and preclinical evidence, it is convincing that identifying and validating novel CDK inhibitors capable of simultaneously targeting STAT3 signaling may open up new windows for long-lasting and multilayered tumor control [30].
The translational value of knowledge of cancer biology into developing effective prognostic/diagnostic and therapeutic strategies for clinical practice remains disappointing [31]. However, increasing knowledge of the molecular basis of tumorigenesis, applications of multi-omics approaches, and molecular simulations based on a structural analysis of receptor-ligand interactions have jointly contributed to the identification of more-reliable predictive markers and the discovery of novel, less-toxic, and target-specific anticancer Biomedicines 2021, 9,92 3 of 22 agents [31][32][33][34]. At present, substantial attention is being focused on small molecules for targeted therapy in cancer treatments [35]. NSC765599 and NSC765690 are small molecule derivatives of salicylanilides (Pub-Chem CID: 60202556; NDMC101), which were previously synthesized and evaluated for biological activities in our Lab [36]. We had previously conducted a series of chemical modifications of the lead molecule, NDMC101, to yield the open ring (NSC765599) [37] and close ring (NSC765690) [38] derivatives. Herein we demonstrated that both NSC765690 and NSC765599 exerted antitumor activities in vitro against panels of NCI60 human tumor cell lines. We further identified and validated CDK2/4/6 and STAT3 as druggable candidates for the compounds, through in silico and molecular simulation of ligand-receptor interaction studies. Hence, our data provide evidence that expressions of CDK2/4/6 and STAT3 can be directly regulated by NSC765690 and NSC765599 with consequent antitumor implications in multiple cancer types.

In Vitro Anticancer Screening against 60 Full NCI Cell Panels of Human Tumor Cell Lines
NSC765690 and NSC765599 ( Figure 1) were submitted to the National Cancer Institute (NCI) for the screening of its panel of NCI60 cancer cell lines. The preliminary single-dose screening of the two compounds was conducted against 60 full NCI cell line panels comprising melanomas, leukemia, central nervous system (CNS) cancers, NSCLC, renal cancer, breast cancer, ovarian cancer, and prostate cancer in agreement with the protocol of the NCI. Following single-dose testing at 10 µM, the two compounds were selected for five-dose screening against the same panels of cancer cell lines. As described previously [39,40], protocols for NCI60 cell five-dose screening involved seeding of about 5000~40,000 cells/well (depending on the doubling time of individual cell lines) in 96-well plates, followed by treatment with NSC765690 or NSC765599 at concentrations of 0.01, 0.1, 1.0, 10, and 100 µM and incubation at 37 • C in 5% humidified CO 2 for 48 h. Cells were fixed with a sulforhodamine B (SRB) solution [41] followed by a series of washing and staining to determine their viability. Growth inhibition was calculated relative to cells without drug treatment and time-zero control. Results of the five-dose assay are represented in terms of a dose-dependent curve, tumor growth inhibition (TGI), LC 50 (concentration needed to kill 50% of cells by cytotoxic activity) for each cell line tested, and GI 50 (concentration needed to inhibit 50% of cancer cell growth) [42]. Results were presented as cell growth relative to the untreated cell control and to the time zero number of cells. Growth inhibitions were indicated by values between 0 and 100, while lethality (cytotoxic effect) was indicated by values less than 0. ines 2021, 9, x FOR PEER REVIEW 3 of 23 reliable predictive markers and the discovery of novel, less-toxic, and target-specific anticancer agents [31][32][33][34]. At present, substantial attention is being focused on small molecules for targeted therapy in cancer treatments [35]. NSC765599 and NSC765690 are small molecule derivatives of salicylanilides (Pub-Chem CID: 60202556; NDMC101), which were previously synthesized and evaluated for biological activities in our Lab [36]. We had previously conducted a series of chemical modifications of the lead molecule, NDMC101, to yield the open ring (NSC765599) [37] and close ring (NSC765690) [38] derivatives. Herein we demonstrated that both NSC765690 and NSC765599 exerted antitumor activities in vitro against panels of NCI60 human tumor cell lines. We further identified and validated CDK2/4/6 and STAT3 as druggable candidates for the compounds, through in silico and molecular simulation of ligand-receptor interaction studies. Hence, our data provide evidence that expressions of CDK2/4/6 and STAT3 can be directly regulated by NSC765690 and NSC765599 with consequent antitumor implications in multiple cancer types.

In Vitro Anticancer Screening against 60 Full NCI Cell Panels of Human Tumor Cell Lines
NSC765690 and NSC765599 ( Figure 1) were submitted to the National Cancer Institute (NCI) for the screening of its panel of NCI60 cancer cell lines. The preliminary singledose screening of the two compounds was conducted against 60 full NCI cell line panels comprising melanomas, leukemia, central nervous system (CNS) cancers, NSCLC, renal cancer, breast cancer, ovarian cancer, and prostate cancer in agreement with the protocol of the NCI. Following single-dose testing at 10 μM, the two compounds were selected for five-dose screening against the same panels of cancer cell lines. As described previously [39,40], protocols for NCI60 cell five-dose screening involved seeding of about 5000~40,000 cells/well (depending on the doubling time of individual cell lines) in 96-well plates, followed by treatment with NSC765690 or NSC765599 at concentrations of 0.01, 0.1, 1.0, 10, and 100 μM and incubation at 37 °C in 5% humidified CO2 for 48 h. Cells were fixed with a sulforhodamine B (SRB) solution [41] followed by a series of washing and staining to determine their viability. Growth inhibition was calculated relative to cells without drug treatment and time-zero control. Results of the five-dose assay are represented in terms of a dose-dependent curve, tumor growth inhibition (TGI), LC50 (concentration needed to kill 50% of cells by cytotoxic activity) for each cell line tested, and GI50 (concentration needed to inhibit 50% of cancer cell growth) [42]. Results were presented as cell growth relative to the untreated cell control and to the time zero number of cells. Growth inhibitions were indicated by values between 0 and 100, while lethality (cytotoxic effect) was indicated by values less than 0.

Identifying the Molecular Targets and Therapeutic Classes of NSC765599 and NSC765690
NSC765599 and NSC765690 were screened for potential drug target using SwissTarget prediction algorithm, which predicts potential drug target based on the principle of similarity [43]. In addition, we also employ the computer-aided Prediction of Biological

Identifying the Molecular Targets and Therapeutic Classes of NSC765599 and NSC765690
NSC765599 and NSC765690 were screened for potential drug target using SwissTarget prediction algorithm, which predicts potential drug target based on the principle of similarity [43]. In addition, we also employ the computer-aided Prediction of Biological Activity Spectra (PASS) web resources to predict the potential drug targets [44]. The activity patterns (fingerprints) of both NSC765599 and NSC765690 were correlated to NCI synthetic compounds, standard agents and molecular targets using the DTP-COMPARE algorithms [45]. The NSC numerical IDs were used as "seed" while GI 50 , TGI and LC 50 were set as the endpoints.

In Silico Molecular Docking Analyses
The three-dimensional (3D) structure of palbociclib (CID: 5330286) and SH-4-54 (CID: 72188643) were retrieved in SDF file format from the PubChem database, while the 3D structures of NSC765690 and NSC765599 were drawn out in sybyl mol2 format using the Avogadro molecular builder and visualization tool vers. 1.XX [46] and were subsequently transformed into the protein data bank (PDB) format using the PyMOL Molecular Graphics System, vers. 1.2r3pre (Schrödinger, LLC). The PDB file of the 3D structure of the receptors and crystal structures of apo CDK2 (PDB; 4EK3), CDK4/cyclin D3 (PDB; 3G33), CDK6/cyclin (PDB; 1JOW), and STAT3 (PDB; 4ZIA), were retrieved from the Protein Data Bank. The PDB file formats of the ligands (NSC765690, NSC765599, and palbociclib) and the receptors (STAT3; CDK2, 4, and 6) were subsequently converted into the Auto Dock Pdbqt format using AutoDock Vina (vers. 0.8, The Scripps Research Institute, La Jolla, CA, USA) [47]. Pre-docking preparation of the receptors followed the removal of water molecules, while hydrogen atoms and Kolmman charges were added accordingly. Molecular docking studies were performed using Autodock VINA software and by following the protocols described in our previous study [3]. The docking results based on hydrogen bonds and electrostatic and hydrophobic interactions of the best pose of the ligand-receptor complexes were expressed as binding energy values (kcal/mol). PyMOL software was used to visualize H-bond interactions, binding affinities, interacting amino acid residues, binding atoms on the ligands and receptors, and 3D graphical representations of ligand-receptor complexes, while 2D graphical illustrations of ligand-binding interactions were further visualized using Discovery studio visualizer vers. 19.1.0.18287 (BIOVIA, San Diego, CA, USA) [48].

Pharmacokinetics, Drug-Likeness, Toxicity and Medicinal Chemical Analyses
The drug-likeness, medicinal chemistry, and pharmacokinetics, including the adsorption, distribution, metabolism, excretion, and toxicity (ADMET) properties of NSC765690 and NSC765599 were analyzed using SwissADME software developed by the Swiss Institute of Bioinformatics [49]. The drug-likeness properties were analyzed in terms of Ghose (Amgen), Egan (Pharmacia), and Veber (GSK), and more importantly, the Lipinski (Pfizer) rule-of-five, as well as cLogP, molecular mass, hydrogen acceptor, hydrogen donor, and molar refractive index [50] for drug-likeness and drug discovery. The Abbot Bioavailability Score was calculated based on the probability of the compound to have at least 0.1 (10%) oral bioavailability in rats or measurable Caco-2 permeability [51], while gastrointestinal absorption and brain penetration properties were analyzed using the Brain Or IntestinaL EstimateD permeation (BOILED-Egg) model [52]. The acute toxicity (LD 50 ) in rats and environmental toxicity were predicted using GUSAR software [53].

Data Analysis
Spearman's rank correlation was used to assess the correlations of NSC765599 and NSC765690 fingerprints with NCI synthetic compounds, standard agents, and molecular targets. The COMPARE correlation threshold was set to ≥45 common cell lines, ≥0.1 correlation coefficient, and ≥0.05 standard deviation. The growth inhibition by NSC765690 and NSC765599 in single-dose assay was obtained by subtracting the positive value on the plot from 100, i.e., a value of 40 would mean 60% growth inhibition.

NSC765690 and NSC765599 Exhibited Anti-Proliferative Effects on NCI60 Human Cancer Cell Lines
Both NSC765690 and NSC765599 exhibited anti-proliferative effect against all the 60 panels of NCI human cancer cell lines ( Figure 2). Furthermore, single-dose treatment with NSC765690 and NSC765599 also demonstrated cytotoxic effects against some cell lines. As indicated by the percentage growth altered by treatment, melanoma (SK-MEL-5, SK-MEL-2, and MALME-3M), renal (A498, UO-31, and CAKI-1), leukemia (HL-60, molt-4, and RPMI-8226), and breast (MDA-MB-468, T-47D, and HS 578T) cancer cell lines were the most responsive to NSC765690 treatment. For NSC765599, melanoma, renal cancer, leukemia, and the breast cancer cell line were also sensitive to cytotoxic effects of NSC765599 at 10 µM. However, panels of prostate, ovarian, and colon cancer cell lines were less responsive to NSC765690 and NSC765599 treatment ( Figure 2). These primary single dose screening results clearly indicated the anti-proliferative activities of NSC765690 and NSC765599 against different kinds of human cancer cell lines, and thus are worthy of further evaluation for dose-dependent activities.

NSC765690 and NSC765599 Exhibited Anti-Proliferative Effects on NCI60 Human Cancer Cell Lines
Both NSC765690 and NSC765599 exhibited anti-proliferative effect against all the 60 panels of NCI human cancer cell lines ( Figure 2). Furthermore, single-dose treatment with NSC765690 and NSC765599 also demonstrated cytotoxic effects against some cell lines. As indicated by the percentage growth altered by treatment, melanoma (SK-MEL-5, SK-MEL-2, and MALME-3M), renal (A498, UO-31, and CAKI-1), leukemia (HL-60, molt-4, and RPMI-8226), and breast (MDA-MB-468, T-47D, and HS 578T) cancer cell lines were the most responsive to NSC765690 treatment. For NSC765599, melanoma, renal cancer, leukemia, and the breast cancer cell line were also sensitive to cytotoxic effects of NSC765599 at 10 μM. However, panels of prostate, ovarian, and colon cancer cell lines were less responsive to NSC765690 and NSC765599 treatment ( Figure 2). These primary single dose screening results clearly indicated the anti-proliferative activities of NSC765690 and NSC765599 against different kinds of human cancer cell lines, and thus are worthy of further evaluation for dose-dependent activities.   and breast cancer cell lines were most sensitive to NSC765690 treatment, while leukemia, CNS, ovarian, and colon cancer cell lines were the least sensitive to NSC765690 ( Table 1). Analysis of LC 50 values also indicated that the melanoma cell lines were the most sensitive to drug treatments, with NSC765690 being the most active (Table 1).

Figure 3.
Dose-cytotoxic response curves of NSC765690 against panels of 60 NCI human cancer cell lines. The growth percentage value of +100 on the Y-axis represents the growth of untreated cells, the 0 value represents no net growth, while −100 represents the complete death of cells.

NSC765599 and NSC765690 Shared Similar NCI Anti-Cancer Fingerprints and Molecular Targets of Cell Cycle Transition Proteins
DTP-COMPARE analysis results showed that the 10 NCI synthetic compounds that are most correlated with NSC765690 and NSC765599 anti-cancer fingerprints are small molecules (MW: 168.19~506.5 g/mol), with p values in the range of 0.44~0.62. In addition, NSC765690 (p-value range of 0.33~0.74) and NSC765599 (p-value range of 0.15~0.36) shared anticancer mechanism with a number of standardized drugs in the NCI database ( Table 2). The molecular target fingerprints generated also showed that both compounds shared a similar positive correlation (p = 0.1~0.3) with the expression of multiple genes, most of which are proteins involved in cell cycle progressions (Table 3). Biomedicines 2021, 9, x FOR PEER REVIEW 7 of 23    The GI 50 = 50% growth inhibition, TGI = total growth inhibition, LC 50 = 50% loss of cells.

CDK2/4/6 and STAT3 Are Potential Druggable Candidates for NSC765690 and NSC765599
Using NSC765690 and NSC765599 as query molecules on the SwissTargetPrediction algorithm, a computer-based drug target prediction tool that identifies the most probable macromolecular targets of a small molecule, on the basis of similarity with a known actives compound in the library [54], we identified a number of targetable proteins, most of which are CDKs and associated proteins. Among these predicted targets, we found that kinases, cytochrome P450, enzymes, and electrochemical transporters were the most occurring targeted classes (Table 4, Supplementary Figure S1). Specifically, STAT3, four members of CDKs (CDK1/2/4/9), and three cyclins were among the top-ranked targeted proteins, while NSC765599 was predicted to target CDK2/4/5/6. Other top-ranked proteins targetable by NSC765599 are shown in Table 4. Coherent with the SwissTarget prediction, the PASS analysis of NSC765690 and NSC765699 also predicted (all pa > pi) inhibitions of CDKs and STAT3 amongst other activities (Table 5).

NSC765599 and NSC765690 Met the Required Criteria of Drug-Likeness and Safety
NSC765690 met the required criteria of a good drug candidate in terms of lipophilicity, polarity, flexibility, solubility, saturation, and molecular weight, while NSC765599 slightly violated the required range of solubility (Log S(ESOL) = 0~6) and lipophilicity (XlogP3 = −0.7~5) having slight outlying values of −5.71 and 5.47 respectively. Both compounds demonstrated good synthetic accessibilities, highly probable GIA absorption, and bioavailability, but poor BBB permeation ( Supplementary Figures S6 and S7, Table 7). Predicted ecotoxicity and acute toxicity for different administration routes of NSC765690 and

Discussion
Despite advances in treatment modalities, cancer survival ratios are still disappointing; thus, developing a novel chemotherapeutic strategy may improve the prognosis of the cancer patient [55]. At present, substantial attention is being focused on small molecules for targeted therapy in cancer treatments [35]. In the present study, we demonstrated the anti-proliferative and dose-dependent cytotoxic effect of NSC765690 and NSC765599 against panels of NCI human cancer cell lines, with NSC765690 demonstrating higher activities than NSC765599. However, there was the least amount of cell lethality in panels of leukemia, CNS, ovarian, and colon cancer cell lines, suggesting that NSC765690 and NSC765599 are not generally toxic to growing cell lines, but display some degree of cytotoxic preference for NSCLC, melanoma, renal, and breast cancer cell line.
The NCI60 cell lines have been well-characterized for genetic and protein expression patterns, and, a computational tool, the DTP-COMPARE algorithm, has been developed to identify the molecular targets and mechanism(s) of action for the unknown compound from known drugs as well as the known molecular targets in the NCI databases [56]. Interestingly, our analysis of correlation patterns of NSC765690 anticancer fingerprint with mechanistically known NCI-Standard Agents suggests that NSC765690 might be targeting DNA replication and cell cycle. NSC765599, on the other hand, shares similar (p = 0.15~0.36) antitumor mechanistic fingerprints with known inhibitors of growth factor receptor and cell cycle transition, and DNA damage inducer. However, the modest correlation with NCI-standard agents suggests that NSC765599 may have a unique mechanism of action not common to standard agent database. The strong antitumor pattern correlation (p = 0.44~0.6) between NSC765690 and NSC765599 fingerprint with NCI synthetic compounds suggests a mechanism of action similar to that of the seed compounds. Unfortunately, the top-ranked correlated compounds antineoplastic-643812 (p = 0.61) and N-(3-chloro-2-methylphenyl)-2-hydroxy-3-nitrobenzamide (p = 0.62) are yet to be mechanistically investigated, while combretastatin A-4 with p-value 0.58 (NSC765690) and p-value 0.53 (NSC765599) have been reported for the mechanism of action, indicating cell cycle arrest and inhibition of tubulin [57], hence providing hypothesis about the possible mechanism of actions of NSC765690 and NSC765599. The high number of cell cycle proteins and cyclin and cyclindependent kinases identified as molecular target fingerprints of both NSC765690 and NSC765599 strongly suggest that cyclin and cyclin-dependent kinase pathways are the core molecular targets of NSC765690 and NSC765599 predicted by COMPARE. However, the correlations (p = 0.1~0.3) were low, indicating the possibility of a new mechanism not captured by COMPARE.
The COMPARE prediction, together with in-silico SwissTarget and PASS prediction collectively identified STAT3/CDK2/4/6 as the most probable target for NSC765690 and NSC765599, and thus were further studied for ligand-receptor interactions. Interestingly, we found that NSC765690 and NSC765599 docked well into the receptor cavities of STAT3, CDK2, CDK4, and CDK6, with CDK2 being the most favored receptor while CDK4 appeared to be the least favored. Comparatively, the higher binding affinities and closer proximity of NSC765690 to the receptors suggested that NSC765690 is a better ligand for CDK2, CDK4, and CDK6 than is NSC765599.
NSC765690 and NSC765599 are both derivatives of NDMC101. A close look at the structure-related activities of NSC765690 and NSC765599 revealed that cyclization of salicylanilide core scaffold (Figure 1) was the structural feature associated with enhanced anticancer activity and better docking profiles of NSC765690 with STAT3/CDK2/4/6 compared to NSC765599 with an open-ring structure (Figures 7 and 8). Interestingly, when the same receptors (CDK2/4/6) were docked with palbociclib, a clinical CDK inhibitor, higher binding energies (lower binding affinities) were recorded for palbociclib compared to NSC765690, thus making the latter the more-favored ligand for CDK2/4/6.
The unique stability of NSC765690 in the binding sites of CDK2, CDK4, and CDK6 could be attributed to the larger number of H-bond and pi interactions including pi-cation, pi-anion, pi-alkyl, pi-pi stack, and pi-pi T-shaped interactions between NSC765690 and the receptors. The large numbers of pi interactions, which mostly involve charge transfer, help in intercalating the NSC765690 in the receptors' binding cavities. The higher affinity of NSC765690 was also associated with the presence of a larger number of Van der Waal forces created on its backbone with respective amino acids Phe82, Leu83, Asp145, Ala144, Asn132, Gln131, Thr14, Thr158, Glu12, Gly13, Gly11, Asp86, Gln85 and His84 of CDK2; Leu64, Arg60, Leu113, Ala63, Leu31, Phe163, Phe160, Leu161, Glu76 and Arg114 of CDK4; and Phe71, Ser296, Leu295, Phe127, and His73 of CDK6, which undoubtedly created a strong cohesive environment, thereby stabilizing the complex formed [58]. These higher numbers of interactions undoubtedly contributed to the higher affinity that NSC765690 has for CDK2, CDK4, and CDK6 than do NSC765599 and palbociclib. Palbociclib has been actively applied in multiple preclinical models and was approved for targeting CDK4/6 as anticancer therapy for breast cancer; however, acquired resistance occurs in almost all cases [20]. Although docking STAT3 with a known inhibitor, SH-4-54, revealed a higher number of conventional H-bond and halogen bond interactions with STAT3, the pi interactions were fewer, and consequently, the overall binding affinity was less negative. Therefore, higher pi interactions also contributed to the higher binding affinity that NSC765690 has for STAT3 than does SH-4-54. Our molecular simulations of ligand-receptor interactions validated the computational prediction of STAT3/CDK2/4/6 as druggable candidates for NSC765690 and NSC765599. The present study, therefore, provides useful molecular docking-based evidence of alternative small molecules that could target CDK4/6 as well as CDK2 and STAT3. They thus offer a wider targeted therapy with less probability of suffering drug resistance The concept of drug-likeness gives useful guidelines for identifying potential drug candidates during the early stage of drug discovery and development [59]. Estimation of rodent acute toxicity (LD 50 ), an adverse effect that follows a single dose exposure to a substance, is an important task in drug design and risk assessment of chemicals [60][61][62]. The LD 50 values predicted for oral, intravenous, intraperitoneal, and subcutaneous administration indicated that both NSC765690 and NSC765599 were non-toxic and will be well-tolerated by the experimental animals. In addition, NSC765690 and NSC765599 met the criteria of a good drug candidate ( Table 7), suggesting that the compounds were druglike molecules and thus possessed the potential to be considered oral drug candidates. Collectively, findings from this study have opened the door for new research direction for NSC765690 and NSC765599 as a novel and potent anti-cancer agent. We found these compounds worthy of further investigation, and both in vitro and in vivo studies in tumorbearing mice are currently ongoing in our lab to assess the full therapeutic efficacies of these compounds.

Conclusions
In conclusion, this study demonstrated the anti-proliferative effect of NSC765690 and NSC765599 against the 60 panels of NCI human cancer cell lines, and dose-dependent cytotoxic preference for NSCLC, melanoma, renal, and breast cancer cell lines. STAT3/CDK2/ 4/6 signatures appear to be potential druggable candidates for the small molecules and interacted with high binding affinity, indicating the potential of NSC765690 and NSC765599 to dysregulate expressions of STAT3/CDK2/4/6 signature and, consequently, compromised cell survival. Further in-vitro and in-vivo confirmation studies are ongoing in our lab.
Supplementary Materials: The following are available online at https://www.mdpi.com/2227-9 059/9/1/92/s1. Figure S1: Pie chart showing the repartition of protein classes of potential druggable candidates for NSC765690 and NSC765599. Figures S2-S5: Docking profiles of NSC765690, NSC765599, palbociclib with CDK2/4/6 and STAT3. Figure S6: BOILED-Egg model of brain or intestinal estimated permeation of NSC765599 and NSC765690. Figure  Data Availability Statement: The datasets generated and/or analyzed in this study are available on reasonable request.