Pro-Oncogenic c-Met/EGFR, Biomarker Signatures of the Tumor Microenvironment are Clinical and Therapy Response Prognosticators in Colorectal Cancer, and Therapeutic Targets of 3-Phenyl-2H-benzo[e][1,3]-Oxazine-2,4(3H)-Dione Derivatives

Genetic and environmental factors play important roles in cancer progression, metastasis, and drug resistance. Herein, we used a multiomics data analysis to evaluate the predictive and prognostic roles of genetic and epigenetic modulation of c-MET (hepatocyte growth factor receptor)/epidermal growth factor receptor (EGFR) in colorectal cancer (CRC). First, we found that overexpressions of c-MET/EGFR were associated with the infiltration of tumor immune cells and cancer-associated fibroblasts, and were of prognostic relevance in CRC cohorts. We also observed that genetic alterations of c-MET/EGFR in CRC co-occurred with other gene alterations and were associated with overexpression of messenger (m)RNA of some cancer hallmark proteins. More specifically, DNA-methylation and somatic copy number alterations of c-MET/EGFR were associated with immune infiltration, dysfunctional T-cell phenotypes, and poor prognoses of the cohorts. Moreover, we describe two novel gefitinib-inspired small molecules derivatives of 3-phenyl-2H-benzo[e] [1,3]-oxazine-2,4(3H)-dione, NSC777205 and NSC777207, which exhibited wide-spectrum antiproliferative activities and selective cytotoxic preference for drug-sensitive and multidrug-resistant melanoma, renal, central nervous system, colon, and non-small cell lung cancer cell lines. We further provided in silico mechanistic evidence implicating c-MET/EGFR/phosphatidylinositol 3-kinase (PI3K)-mammalian target of rapamycin (mTOR) inhibition in anticancer activities of those compounds. Our overall structure-activity relationship study revealed that the addition of an –OCH3 group to salicylic core of NSC777207 was not favorable, as the added moiety led to overall less-favorable drug properties as well as weaker anticancer activities compared to the properties and activities demonstrated by NSC777205 that has no –OCH3 substituent group. Further in vitro and in vivo analyses in tumor-bearing mice are ongoing in our lab to support this claim and to unravel the full therapeutic efficacies of NSC777205 and NSC777207 in CRC.


INTRODUCTION
According to 2020 global cancer statistics, with more than 1.9 million new cases and 935,000 deaths, colorectal cancer (CRC) ranked third in terms of incidence, but second in terms of mortality in 2020 (Sung et al., 2021). CRC is regarded as a marker of socioeconomic development, as incidence rates tend to rise uniformly with an increasing human development index (HDI) in countries undergoing major transitions (Bray, 2014;Fidler et al., 2016). In particular, incidence rates have been steadily increasing in Asia, Europe, and South America (Arnold et al., 2017;Arnold et al., 2020). These increases correspond to adoption of lifestyles that are associated with risk factors, such as being overweight, having decreased physical activity, and becoming more sedentary (Siegel et al., 2020). In addition, increased consumption of cigarettes, alcohol, and red or processed meats, and decreased intake of dairy products, grains, and fibers in those regions have greatly contributed to the observed trends (Clinton et al., 2020).
Both genetic and environmental factors play important roles in the etiology of CRC. To date, several molecular hallmarks have been associated with CRC (Colussi et al., 2013). These hallmarks of CRC are acquired through progressive accumulation of genetic and epigenetic alterations that inactivate tumor-suppressor genes and activate oncogenes (Grady and Carethers, 2008;Kuipers et al., 2015). At the molecular level, activation of growth factor receptors (GFRs) was shown to be involved in the rapid growth of cancer cells (Normanno et al., 2006). The etiology of most cancers can be linked to aberrant intra-and intercellular communication associated with GFR-mediated pathways. Activated GFRs aid blood vessel formation, cell migration, metastasis, and the inhibition of apoptosis. Among these GFRs, the epidermal growth factor receptor (EGFR) and hepatocyte growth factor receptor (HGFR) play central roles in the pathogenesis and progression of different carcinoma types (Normanno et al., 2001).
HGFR, also known as tyrosine-protein kinase (c-Met), is a tyrosine kinase receptor (RTK) that in humans is encoded by the MET gene, while the EGFR belongs to the ErbB family of RTKs, and is a trans-membrane protein. The binding of ligands to the extracellular domain of these receptors induces the formation of receptor homo-or heterodimers and subsequent activation of the intrinsic tyrosine kinase domain (Oprita et al., 2021), which facilitates recruitment of proteins that initiate a signaling cascade, integrating numerous signaling pathways that lead to specific cellular responses that favor angiogenesis, high nutrient supplies, cell migration, tumor growth, and metastasis of CRC (Sattler et al., 2011;An et al., 2018).
Several treatment modalities exist, including immunotherapy, radiotherapy, neoadjuvant and palliative chemotherapies, laparoscopic surgery for primary disease, and more-aggressive resection of metastatic disease, and these provide alternatives for patients with primary and metastatic CRC (Stintzing, 2014;Kuipers et al., 2015). However, these treatment strategies have limited success rates in terms of prognoses and long-term survival. For these reasons, searching for relevant predictive biomarkers that can inform treatment decisions and developing novel therapeutic strategies with high efficacy and minimal side effects are impetuses for the research world. To this end, we used a multiomics data analysis to evaluate the predictive and prognostic roles of genetic and epigenetic modulation of c-MET/EGFR in CRC. Moreover, we describe two novel gefitinibinspired small molecules, NSC777205 and NSC777207, with wide-spectrum antiproliferative activities and selective cytotoxic preferences for melanoma, renal, central nervous system (CNS), colon, and non-small cell lung cancer (NSCLC) cell lines, and provide in silico mechanistic evidence implicating c-MET/EGFR/phosphatidylinositol 3-kinase (PI3K)/mammalian target of rapamycin (mTOR) inhibition in anticancer activities of these compounds.

MATERIALS AND METHODS
In Silico Evaluation of the Drug Likeness, Pharmacokinetics, Acute Toxicity, and Cytotoxic Activities of NSC777205 and NSC777207 Against Cancer Cell Lines We analyzed the drug-likeness, PKs, medicinal chemistry, and toxicity of NSC777205 and NSC777207 using SwissADME software (http://www.swissadme.ch) (Daina et al., 2017), and computer-aided Prediction of Biological Activity Spectra (PASS) web resources (http://way2drug.com/dr) (Poroikov et al., 2019). We used the blood-brain barrier (BBB) Prediction Server (https://www.cbligand.org/BBB/) which operates based on support vector machine (SVM) and LiCABEDS algorithms on four types of fingerprints of 1,593 reported compounds  to analyze the BBBpermeation ability of NSC777205 and NSC777207 based on the permeation threshold of 0.02. In addition, we also used the Brain Or IntestinaL EstimateD permeation method (BOILED-Egg) model (Daina and Zoete, 2016) to further analyze the brainand intestinal-permeation abilities of the compounds based on their lipophilicity and polarity. We used the CLC-Pred (Cell Line Cytotoxicity Predictor) modules of the PASS server (http://www. way2drug.com/PASSonline) (Poroikov et al., 2019) created based on a training set of data on cytotoxicity retrieved from ChEMBLdb (vers. 23) (https://www.ebi.ac.uk/chembldb/), to predict the cytotoxic activities of NSC777205 and NSC777207 against various cancer cell lines.

Differential Expression and Survival Analysis of Tyrosine-Protein Kinase/ Epidermal Growth Factor Receptor in Colorectal Cancer
Expression levels of c-MET/EGFR in CRC tissues from The Cancer Genome Atlas (TCGA) database were compared with expression levels in normal tissues using the TNMplot module of the Kaplan-Meier Plotter (https://www.tnmplot.com/). Furthermore, we used the Human Protein Atlas (HPA) database (www.proteinatlas.org) to assess the immunohistochemical (IHC) profile of c-MET/EGFR in tumor samples from CRC patients. We used the MEXPRESS algorithm (https://mexpress.be/index.html) to depict the mechanisms of c-MET/EGFR dysregulation in colorectal cancer (Koch et al., 2019). To analyze the prognostic relevance of the gene signatures, we used the PREdiction of Clinical Outcomes from Genomic profiles (PRECOG) server (https://precog.stanford.edu/precog_ metaZ.datatable.php) (Gentles et al., 2015) to split CRC patient cohorts into high and low c-MET/EGFR expression groups, by setting the median expression level as the expression threshold. Kaplan-Meier survival plots were used to present the survival ratio of cohorts with hazard ratios (HRs), 95% confidence intervals (CIs), and log-rank test p values.

Analysis of the Effects of Tyrosine-Protein Kinase/Epidermal Growth Factor Receptor Expression, and Genetic and Epigenetic Alterations on Immune Infiltration, Dysfunctional T-Cell Phenotypes, and Prognostic Relevance in Colorectal Cancer
We used the cBioPortal server (http://www.cbioportal.org/) to mine the Colorectal Adenocarcinoma (TCGA, PanCancer Atlas) dataset for c-MET/EGFR genetic alterations including mutations, copy number alterations, gene mutation co-occurrences, and microbiome signatures and evaluated the prognostic relevance of the alterations in 594 CRC patients using survival analyses of the cohorts (Cerami et al., 2012;Gao et al., 2013). Protein-protein interaction (PPI) networks and functional enrichment analyses including Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways and biological processes enriched in genes that were associated with c-MET/EGFR alterations were analyzed using the Search Tool for Retrieval of Interacting Genes (STRING, vers. 10. 5, (https://www.string-db.org) with the adjusted threshold confidence set to 0.400 (Szklarczyk et al., 2019). We used the IMmune Estimation Resource (TIMER2.0) algorithm (http:// timer.cistrome.org/) to analyze correlations of c-MET/EGFR expression with somatic copy number alterations and infiltration of six tumor-infiltrating immune cell subsets (B cells, cluster of differentiation 4 (CD4) T cells, CD8 T cells, macrophages, neutrophils, and dendritic cells) in CRC. We also used the TIMER server to analyze correlations of c-MET/EGFR expressions and cancer-associated fibroblast (CAF) infiltration. In addition, we analyzed the prognostic relevance of these associations by employing a protocol described in a previous study (Lawal et al., 2021a). Briefly, TCGA CRC cohorts on the TIMER server were categorized into four groups of low CAF+low c-MET/EGFR, low CAF+high c-MET/EGFR, high CAF+low c-MET/EGFR, and high CAF+high c-MET/EGFR and used Kaplan-Meier survival plots to analyze the cumulative survival of the cohorts. We also used Tumor Immune Dysfunction and Exclusion (TIDE) (http://tide.dfci.harvard.edu) (Jiang et al., 2018) to analyze the effects of genetic (somatic copy number) and epigenetic (DNA methylation) alterations of c-MET/EGFR on dysfunctional T-cell phenotypes, risk factors, and survival of CRC cohorts.
In Vitro Anticancer Screening of NSC777205 and NSC777207 NSC777205 and NSC777207 were evaluated for in vitro anticancer activities against 60 panels of human tumor cancer cell lines representing leukemia (six cell lines), NSCLC (eight cell lines), colon cancer (seven cell lines), CNS cancers (six cell lines), melanomas (nine cell lines), ovarian cancer (seven cell lines), renal cancer (eight cell lines), prostate cancer (two cell lines), and breast cancer (five cell lines) through the National Cancer Institute (NCI). Ranges of 5,000-40,000 viable cancer cells were seeded into each well of 96-well plates and incubated at 37°C with 5% CO 2 , 95% air, and 100% relative humidity for 24 h. After incubation, cells were treated with either NSC777205 or NSC777207 at a single dose of 10 μM and further incubated for 48 h (Shoemaker, 2006;Holbeck et al., 2010). A sulforhodamine B (SRB) staining protocol (Vichai and Kirtikara, 2006) was used to determine cell viability. After determining satisfactory antiproliferative activities of a single dose, both NSC777205, and NSC777207 were further subjected to multiple-dose screening at five concentrations of 0, 0.1, 1.0, 10, and 100 μM using the same protocol described above. The activity of each of the drug on each cell line was calculated using four measurements parameters as described here. In these equations, Tz is the absorbance at time 0; C is the absorbance of the control after 48 h without treatment; Ti is the absorbance of drug-treated cells after 48 h; GI 50 is the concentration of the drug causing a 50% reduction in cell growth; TGI is the concentration causing complete inhibition of cell growth, and LC 50 is the concentration causing 50% cell death. For a drug whose maximum dose tested (100 μM) did not meet the required effect on a particular cell line, the value for that parameter was expressed as greater than the maximum concentration tested (>100 μM).

DTP-COMPARE Analysis and in Silico Identification of Mechanistic Targets for NSC777205 and NSC777207
We used the private COMPARE module of the NCI-COMPARE program (https://dtp.cancer.gov/databases_tools/compare.htm) to correlate anticancer activity fingerprints of NSC777205 and NSC777207 with NCI synthetic compounds, standard agents, and molecular targets based on established relationships between cell responses to therapy and gene expression profiles of the cell lines (Paull et al., 1989). The NSC numerical IDs were used as the "seed," whereas GI 50 , TGI, and LC 50 values were set as endpoints (Lawal et al., 2021b). Analyses were conducted based on program default settings of a minimum correlation of 0.1, minimum common cell line counts of 40, minimum coefficient of variation of 0.01, and maximum return of 200, while results were generated in rank ordered lists of the most highly correlated NCI compounds and targets. In addition to the COMPARE algorithm, we also identified potential targets of NSC777205 and NSC777207 using three different in silico target identification algorithms, including the PharmMapper Server, a pharmacophore mapping algorithm with statistical modules (http://lilab-ecust.cn/pharmmapper/index.html) (Liu et al., 2010), the SwissTargetPrediction algorithm (http://www. swisstargetprediction.ch/), which operates on the basis of similarity of the queried molecule with known active drugs (Gfeller et al., 2013), and the computer-aided PASS web resources (http://way2drug.com/dr) (Poroikov et al., 2019).

Statistical Analysis
All statistical analyses were conducted according to each server's instructions. We used the Wilcoxon test to compare statistically significant differences in c-MET/EGFR expressions between CRC cohorts and healthy tissues. Purity adjustment and partial Spearman's correlations were used to analyze c-MET/EGFR expression correlations with infiltration of CAFs and various immune cells. Gene alteration co-occurrences on the cbioportal were considered significant only at an adjusted p value of <0.05. The KEGG and gene ontologies (GOs) were visualized using GraphPad prism software. DTP-COMPARE analyses were conducted using Pearson's correlations. All survival analyses are presented using Kaplan-Meier plots. GI was calculated relative to cells without drug treatment and the time-zero control. Statistical significance was denoted as *p < 0.05, **p < 0.01, and ***p < 0.001.

Tyrosine-Protein Kinase/Epidermal Growth Factor Receptor are Deregulatory Expressed and are Associated With Oxaliplatin Resistance and Poor Prognoses of Colorectal Cancer Patients
We conducted a differential expression analysis between the CRC tumorous and adjacent normal tissues in TCGA database. We found that c-MET expression is higher (p 1.24e-07) while EGFR expression is lower (p 4.67e-04) in colorectal cancer cohorts compared to adjacent normal cohorts. ( Figure 1A). Correlation analyses also indicated that c-MET expression was strongly correlated with EGFR expressions in COAD (r 0.497, p 5.95e-30) and READ (r 0.545, p 3.01e-14) ( Figure 1B). Furthermore, we explored the HPA database for IHC data of c-MET/EGFR protein expressions based on tissue microarrays (TMAs) in CRC cohorts ( Figure 1C). We found that all CRC patient samples presented for EGFR (% 66.66 COAD and 33.33% READ) and c-MET (58.33% COAD and 41.66% READ) expression profiling were positive for c-MET (Antibody: CAB005282) and EGFR (Antibody: HPA018530) signals Frontiers in Pharmacology | www.frontiersin.org August 2021 | Volume 12 | Article 691234 ( Figure 1D). Mean ages of patients with high c-Met and EGFR expression profiles were 72.08 and 75.25 years, respectively. Totals of 58.33 and 83.33% of patients had very high intensities (>75%), while 41.66 and 8.33% had medium intensities of c-Met and EGFR expression, respectively ( Figure 1E). Only 8.33% of patients had a low EGFR expression intensity ( Figure 1E). Our analysis of the mechanisms of gene dysregulation indicated that EGFR expression is significantly associated with colon polyps (p 0.028) and primary tumors p 0.04029) but less associated with metastasis or recurrent tumor. Furthermore, both c-MET and EGFR expressions show significant correlation with copy number alterations in colorectal cancer ( Figure 1F). To assess the prognostic relevance of gene signatures, we conducted a survival analysis and found that CRC cohorts with higher expression profiles of c-Met and cohorts with lower expression profiles of EGFR exhibited low survival rates ( Figure 1G) and resistance to oxaliplatin chemotherapy.  (H) Gene expression profile between oxaliplatin sensitive and resistance colorectal cancer cohorts. The strength of correlations between the genes is reflected by the purity-adjusted partial Spearman's rho value and estimated statistical significance, where a value of r 1 means a perfect positive correlation and a value of r −1 means a perfect negative correlation. *p < 0.05; **p < 0.01; ***p < 0.001.
Frontiers in Pharmacology | www.frontiersin.org August 2021 | Volume 12 | Article 691234 5 was negatively correlated with tumor purity in COAD (r -0.0716) and READ (r -0.18717, p 0.026802), while MET exhibited no significant correlations with tumor purity in COAD (r 0.01029, p 0.836045) and READ (r 0.017459, p 0.83777). Collectively, the above results indicate that c-MET/ EGFR expressions were correlated with immune infiltration in COAD and READ ( Figure 2A). Furthermore, we used the abundances of six immune cells and expression levels of c-MET/EGFR to construct univariate and multivariate Cox regression models. The univariate analysis indicated that higher infiltration levels of CD4 + T cell, macrophages, neutrophils, and dendritic cells were correlated with poor survival outcomes in COAD (Supplementary Figure 1), while low infiltration levels of CD8 + T cell, B cells, macrophages, neutrophils, and dendritic cells were correlated with poor survival outcomes in READ. A multivariate Cox regression model revealed that lower infiltration levels of CD4 + T cells (hazard ratio (HR) 0.008, p 0.015) were risk factors for COAD (Table 1).
We analyzed CAF correlations with expression profiles of c-MET/EGFR in CRC samples. We found that EGFR expressions were correlated with CAF infiltration in COAD (partial corr. 0.281, p 2.25e-06) and READ (partial corr. 0.32, p 2.11e-03). Similarly, c-Met expressions were positively correlated with CAF infiltration in COAD (partial corr. 0.139, p 4.91e-03) and READ (partial corr. 0.214, p 1.14e-02) ( Figure 2B). To evaluate the prognostic relevance of these associations, we classified the cohorts into four groups; low CAF+ low c-Met/ EGFR, low CAF+ high c-Met/EGFR, high CAF+ low c-Met/EGFR, and high CAF+ high c-Met/EGFR, and we found that the high CAF+ high c-Met/EGFR cohort had a low overall survival rate ( Figure 2C).   We mined TCGA PanCancer Atlas dataset for c-MET/EGFR genetic alteration data and evaluated the prognostic relevance of alterations in CRC patients. We found that 3.0 and 4.0% of the 594 CRC patients (63.6% of COAD, 26.1% of READ and 10.3% of mucinous adenocarcinoma (MUAD)) in the dataset respectively harbored genetically altered EGFR and c-MET ( Figure 3A). Alterations in EGFR were mainly mutations (2.36%), amplifications (0.51%), and multiple alterations (0.17%), while alterations in c-MET were mutations (3.2%), amplifications (0.17%), and deep deletions (0.34%) ( Figure 3B). By stratifying the mutations, we found 19 mutations consisting of 17 missense and two truncating mutations of EGFR, while c-MET mutations were stratified into 22 missense mutations, four truncations (all nonsense mutations; R1170* and S851*), and one in-frame deletion (I1130del) ( Figure 3C). A comparative survival analysis indicated that CRC patients with genetic alterations of c-MET/EGFR exhibited shorter overall survival (OS), DFS, and progressive-free survival than cohorts without c-MET/EGFR alterations ( Figure 3D).

Genetic Alterations of Tyrosine-Protein Kinase/Epidermal Growth Factor Receptor in CRC Co-Occurred With Other Gene Alterations and are Associated With Overexpression of mRNAs of Some Cancer Hallmark Proteins
We analyzed the frequencies of the co-occurrence of gene alterations with c-MET/EGFR genetic alterations, and found the co-occurrence of genetic alterations in a total of 19,434 genes, enriched in both c-MET/EGFR altered and non-altered cohorts ( Figures 4A, B). Genes with significantly (all p < 10 −4 ) enriched genetic alteration co-occurrences with EGFR alterations included SMARCD2, TRIM7, MCMDC2, HNRNPUL1, BEST3, GRB10, PDE6C, and SNX19, while SLC22A2, DST, AKT3, and TAB2 were the only significantly (all p < 10 −4 ) enriched gene alterations that co-occurred with c-MET alterations (Supplementary Table 3). However, TTN, APC, TP53, SYNE1, DNAH11, BRAF, PIK3CA, COL12A1, DNAH3, and SRCAP had the highest mutation frequencies in both EGFR altered and nonaltered cohorts, while TTN, APC, DST, BRAF, TP53, ZC3H13, DNAAH8, SYNE1, USP34, and DNAH9 were the most frequently mutated genes associated with c-MET altered and non-altered cohorts ( Figure 4C).
We further constructed PPI networks of overexpressed mRNAs in c-MET/EGFR altered cohorts ( Figure 4D) and performed enrichment analyses. We found that overexpressed mRNAs were enriched in several KEGG pathways associated with CRC development, progression, and drug resistance. The top enriched pathways (with a strength of enrichment (SOE) of >1 and p < 0.05) included cell adherent junctions, EGFR tyrosine kinase inhibitor resistance, central carbon metabolism, CRC, pancreatic cancer, hepatocellular carcinoma, gastric cancer, and melanomas ( Figure 4E), while enriched GOs (with a SOE of >1 and p < 0.05) included the c-MET, ERBB2, and EGFR signaling pathways, MAPKK activation, and negative regulation of cellular responses to drugs ( Figure 4F).

DNA-Methylation and Somatic Copy Number Alterations of Tyrosine-Protein Kinase/Epidermal Growth Factor Receptor are Associated With Immune Infiltration, Dysfunctional T Cell Phenotypes, and Poor Prognoses of Colorectal Cancer Patients
We also analyzed associations of different somatic CNAs, such as arm-level gains, high amplifications, deep deletions, arm-level deletions of c-MET/EGFR and immune cell infiltration in CRC. We found that arm-level gains of EGFR were associated with B cell (p < 0.005), CD8 + T cell (p < 0.001), neutrophil (p < 0.005), and dendritic cell infiltration in COAD, and macrophage (p < 0.005) infiltration in READ. Arm-level gains of c-MET were positively correlated with CD8 + T cell, B cell, neutrophil, and dendritic cell infiltration in colon cancer and dendritic cells infiltration in READ, while deep deletions of c-MET were associated with CD8 + T cell and neutrophil infiltration of tumors in READ ( Figure 5A). Furthermore, CNAs of both c-MET and EGFR were associated with dysfunctional T-cell Frontiers in Pharmacology | www.frontiersin.org August 2021 | Volume 12 | Article 691234 8 phenotypes ( Figure 5B) and poor prognoses ( Figure 5C) of CRC patients. c-MET was found to be hypomethylated, while EGFR was hypermethylated in CRC ( Figure 5D). These methylation statuses of c-MET and EGFR in CRC were also consistently associated with the tumor stage ( Figure 5D), dysfunctional T cell phenotypes ( Figure 5E), and shorter OS of CRC patients ( Figure 5F). Collectively, these findings indicated that the DNA-methylation and somatic CNAs particularly the armlevel gains of c-MET/EGFR were associated with immune infiltration, dysfunctional T cell phenotypes, and poor prognoses of CRC patients.
NSC777205 and NSC777207 Displayed Desirable Characteristic of a Good Drug Candidate and Exhibited In Silico Cytotoxic Activities Against Colon, Brain, Lung, and Bone Cancer Cell Lines NSC777205 (3-(4-chloro-2-fluorophenyl)-2H-benzo[e][1,3] oxazine-2,4(3H)-dione) and NSC777207 (3-(4-chloro-2-fluorophenyl)-7-methoxy-2H-benzo[e][1,3]-oxazine-2,4(3H)dione) were obtained at high purity (>95%) via condensation reactions between salicylic acid and the 2,4-disubstituted aniline moieties of gefitinib to produce a salicylanilide which was further cyclized to the 3-phenyl-2H-benzo[e] [1,3]-oxazine-2,4(3H)-dione derivatives, NSC777205 and NSC777207, as summarized in Figure 6A. Our general in silico analysis of drug-like and toxicity characteristics of NSC777205 and NSC777207 suggested that both compounds exhibited suitable physicochemical properties (lipophilicity, polarity, flexibility, solubility, saturation, molecular weight, and other properties) of a good drug candidate. The compounds were both lead-like and satisfied Lipinski's rules for drug likeness candidates. Furthermore, our in silico analysis suggested that both compounds can be easily absorbed in the gastrointestinal tract and are well permeant to the blood-brain barrier (BBB) especially NSC777205 which demonstrated 2-fold higher permeation of the BBB than NSC777207 based on in silico estimation ( Figure 6B; Table 2). Estimates of the acute toxicity in rats revealed high LD 50 values for NSC777205 and NSC777207 from oral, intravenous, intraperitoneal, and subcutaneous administration, while bioaccumulation factor (log10) and other markers of eco-toxicity also identified both compounds as having low toxicities (Table 2). Collectively, these findings suggested that NSC777205 and NSC777207 , which were obtained via intermolecular hybridization and condensation reactions between salicylic acid (R 1 H) or 2-hydroxy-4-methoxybenzoic acid (R 1 OCH 3 ) and 2,4-disubstituted aniline moieties of gefitinib to produce salicylanilides which were further respectively cyclized to produce NSC777205 (when R 1 H) and NSC777207 (when R 1 OCH 3 ). Cyclization occurred in the presence of methyl chloroformate and pyridine, while hybridization occurred in the presence of tetrahydrofuran (THF). (B) Blood-brain barrier (BBB) permeation curves of NSC777205 and NSC777207 as measured by the support vector machine (SVM) and LiCABEDS algorithms of the BBB prediction server. (C) PASS predicted cytotoxic activities of NSC777205 and NSC777207 against various cancer cell lines. We initially screened NSC777205 and NSC777207 for anticancer activities against the full NCI-60 panel of human tumor cell lines. Interestingly, we found that with single-dose (10 μM) treatment, both NSC777205 and NSC777207 demonstrated antiproliferative activities against all of the NCI-60 cell line panels of breast, prostate, renal, ovarian, colon, melanoma, CNS, leukemia, and non-small cell lung cancers, with NSC777205 demonstrating higher activities than NSC777207. In addition, most of the leukemia (cell death (CD) 0.77-10.28%), NSCLC (CD 4.2-50.81%), CNS (CD 013.71%), melanoma (CD 0.57-46.24%) renal (CD 7.01-30.91%), breast (CD 10.48%), prostate, and ovarian cancer cell lines demonstrated some cytotoxic response to NSC777205 and NSC777207 single-dose treatment, while panels of colon cancer cell lines were completely insensitive to the cytotoxic activity of the compounds at 10 μM. Collectively, these studies demonstrated that both NSC777205 and NSC777207 exhibited a wide spectrum of anticancer activities but demonstrated selective and exclusive antiproliferative activities against all panels of colon cancer cell lines (Supplementary Figure 2).
With five-dose screening, NSC777205 and NSC777207 demonstrated dose-dependent cytotoxic effects against the panel of NCI-60 human tumor cell lines ( Figures 7A, B). With the exception of HT29 and COLO 205 (GI 50 11.90-20 µM), both NSC777205 and NSC777207 exhibited high anticancer activities with GI 50 concentrations of <6 µM against all cell line panels ( Figure 8A). Total cell growth inhibitory (TGI) concentrations of the compounds revealed that NSCLC (1.71-3.68 µM), leukemia (7.51-33.80 µM), CNS (7.51-33.80 µM), melanoma (4.196-34.90 µM), and colon (12.40-58.80 µM) cancer cell lines were the most sensitive to NSC777205 treatment, while a lower but similar trend of anticancer activity against the panel of cell lines was also observed for NSC777207 ( Figure 8B).  Figure 8C). Collectively, the anticancer study revealed that NSC777205 and NSC777207 exhibited a wide spectrum of antiproliferative activities and selective cytotoxic preferences for renal cancer, melanoma, CNS cancer, colon cancer, and NSCLC cell lines with NSC777205 demonstrating higher potencies than NSC777207.

Tyrosine-Protein Kinase/Epidermal Growth Factor Receptor/Phosphatidylinositol 3-Kinase/Mammalian Target of Rapamycin Signaling Pathways are Implicated in the Anticancer Activities of NSC777205 and NSC777207
In order to adequately identify potential drug targets of our compounds, we used three different in silico drug target identification algorithms and also the DTP-COMPARE algorithm. Using PharmMapper, c-MET and EGFR were both identified as potential targets for NSC777205 and NSC777207. The pharmMapper algorithm identified c-MET as potential targets for NSC777205 and NSC777207 by pharmacophore mapping of one hydrogen bond and four hydrophobic interactions, with normalized fit scores of 0.5876 and 0.5884, while EGFR was mapped with NSC777205 and NSC777207 by two hydrophobic reactions and two hydrogen interactions with normalized fit scores of 0.6981 and 0.6986 (Supplementary Table 5). In addition, the PI3K regulatory subunit alpha, several MAPKs, and several serine/threonine-protein kinases were identified as potential targets for NSC777205 and NSC777207 (Supplementary Table 5). The PASS prediction server also identified c-MET and EGFR as potential targets for NSC777205 and NSC777207 with pa>0.3 and pi<0.06. In addition, inhibition of MAPK, inositol 1,4,5-trisphosphate 3-kinase, and growth factors among other activities were predicted for NSC777205 and NSC777207 (Supplementary Table  6). However, SwissTargetPredition for NSC777207 returned serine/ threonine-protein kinases, mTOR, several subunits of PI3K, MAPK, and ErbB-2 among others, while AKT, EGFR, mTOR, several subunits of PI3K, and growth factors were predicted for NSC777205 (Supplementary Table 7). Altogether, c-MET, EGFR, PI3K, and mTOR were commonly identified targets of NSC777205 and NSC777207 by the three drug target identification algorithms we employed. Furthermore, results of the DTP's COMPARE algorithm indicated that the anticancer fingerprints of NSC777205 and NSC777207 showed strong correlations (p 0.58-0.86, common cell line count (CCLC) 47-58) with fingerprints of some NCI synthetic compounds (Table 3). These synthetic compounds are small molecules with molecular weights ranging 277.8-559.5 g/mol. The COMPARE analysis of NSC777205 and NSC777207 anticancer fingerprints with NCI investigational drugs revealed strong associations (CCLC 55-58, p 0.3-0.61) with a number of c-MET/EGFR/PI3K/ AKT/mTOR inhibitors, including JNJ38877618, MTX211, SGX-523, altiratinib, TAS-6417, voxtalisib, PI-103 HCL, AMG-458 (c-MET), CC115, and others ( Table 4). In addition, the COMPARE analysis identified MET, EGFR, PIK3R2, RPS6KB1, CDK6, IGHMBP2, and ErbB2 as some of the major targets that were correlated with the anticancer fingerprints of NSC777205 and NSC777207 (Table 4).
Collectively this study provides a hypothesis implicating C-MET/EGFR/PI3K/AKT/mTOR signaling pathways in the anticancer activities of NSC777205 and NSC777207.

DISCUSSION
Despite the availability of various treatment strategies, CRC remains among the top ranked most diagnosed and leading causes of cancer deaths globally (Sung et al., 2021). In order to achieve a reasonable shift in the global burden of CRC, the importance of diagnostic and prognostic biomarkers cannot be overemphasized. In the present study, our differential expression and survival analyses indicated that c-MET/EGFR can be an effective biomarker for CRC progression, immune infiltration, CAF infiltration, and poor prognoses, and thus can serve as attractive targets for exploring the development of chemotherapeutic agents against CRC. We found that c-MET and EGFR expression levels were independent biomarkers for clinical prognosis, and their high expressions were correlated with tumor progression and a poor prognosis of CRC. Abnormal expressions of c-MET and EGFR were reported in various human cancers where they support the proliferation and survival of cancer cells (Nicholson et al., 1990;Normanno et al., 2006;Zhang et al., 2018;Chakraborty et al., 2019;Fujiwara et al., 2020;Lee et al., 2020). c-MET and EGFR regulate the synthesis and secretion of several angiogenic growth factors, including basic fibroblast growth factor (bFGF), vascular endothelial growth factor (VEGF), and interleukin (IL)-8, in tumor cells (De Luca et al., 2008). Taken together, these findings suggest that the c-MET/EGFR system is an important mediator within the tumor microenvironment (TME) that results in enhanced tumor growth.
Furthermore, we explored TCGA clinical database to evaluate the frequency and prognostic consequences of c-MET/EGFR genetic alterations in CRC. We found that 3.0 and 4% of the cohorts that respectively harbored genetically altered EGFR and MET had poor prognoses and shorter survival times compared to cohorts without c-MET/EGFR alterations. Genetic alterations of c-MET and EGFR were previously identified and implicated in the progression of several cancers (Bastien et al., 2015;Piccirillo et al., 2015;Yamaura et al., 2020) and were also implicated in the resistance of most cancers to any generation of EGFR-TKIs including the latest third-generation series (Paez et al., 2004;Pao et al., 2005;Guo et al., 2020a). Of importance, we analyzed concurrent gene alterations with c-MET/EGFR genetic alterations, and found that EGFR alteration cooccurred with genetic alterations of SMARCD2, TRIM7, MCMDC2, HNRNPUL1, BEST3, GRB10, PDE6C, and SNX19, while alterations of SLC22A2, DST, AKT3, and TAB2 co-occurred with c-MET alterations in CRC. Genetic alterations and concurrent alterations of c-MET/EGFR with other genetic alterations have been well reported in lung cancer (Stuart and Sellers, 2009;Tang et al., 2018;Guo et al., 2020b;Gini et al., 2020), but were underreported in other solid cancers especially CRC, and hence our study is significant in this regard. A recent study reported concurrent alterations of p53 (60-65%), RTKs (5-10%), PIK3CA/KRAS (3-23%), Wnt (5-10%), and cell cycle pathways MYC (7-25%) and NKX2-1 (10-15%) in EGFR-altered cohorts (Gini et al., 2020). Another more-recent study also identified concurrent genetic alterations of EGFE with MEK, ALK, KRAS, ROS1, TP53, PIK3CA, and PTEN in NSCLC (Guo et al., 2020b), while Tang et al. (Tang et al., 2018) reported coexistent genetic alterations involving ALK, RET, ROS1, and MET in 15 cases of lung adenocarcinoma (Tang et al., 2018). In addition to concurrent gene alterations in CRC patients, we found that the genetic alteration of EGFR was associated with overexpression of mRNA levels of TRIM7, BARX2, CFAP43, TPBG, HPSE, GNB5, MARK3, FUT8-AS1, SS18 and KCNK1, while genetic alteration of MET was associated with the overexpression of mRNA levels of SMAD4, MBP, IER3IP1, OXCT1, ALKBH5, C18ORF25, WDR76, TYMS, LINC00909, and GPR3 in CRC cohorts. We also found that these overexpressed mRNAs were enriched in several KEGG pathways including cell adherent junctions, central carbon metabolism, and EGFR TKI resistance, indicating a potential functional role in driving resistance to therapy. We explored the PPI network and pathways of these mRNAs from a different perspective, and we concluded that multiple components and multiple pathways led to the development and progression of CRC. However, we found that c-MET, ERBB2, and EGFR signaling pathways, MAPKK activation, and negative regulation of cellular responses to drugs appeared to play important roles in the development of CRC. Our findings are therefore relevant in predicting and classifying evolutionary trends of genetic alterations during therapy and could guide the selection of combination therapies appropriate for patients with resistance to TKIs in CRC.   Van der waal forces  VAL1010  VAL101  GLY719  LS1110  ASN1167  ARG1166  1ASP10  LYS745  PHE1089  MET1211   12  PHE723  GLY1163  ALA1226  PHE997,  MET790  LEU792  ARG841  LYS728  ASN842  ASN996  ASN996  THR854  LEU792  GLY796  PRO794  CYS797  LYS728  GLY1085  GLU1015  ILE1084  THR847  TYR1234  Hydrophobic Interactions Van der waal forces  GLN846  GLN846  ASP861  THR2207  GLU2196  ALA1971  PHE698  TRP201  HIS295  LEU2204  MET2199  ILE1939  CYS869  GLY868  GLY868  LEU1900  GLY2203  GLU2196  PRO789  PRO789  MET842  GLY2203  THR2207  MET2199  TYR867  ILE870  GLN1937  GLN2200  PRO866  PHE898  GLY2203  TRP201  GLN846  THR2207  LEU865  ASN2147  LEU864  π -sulfur, π-electron cloud between the Aromatic rings of ligands and lone pair of electron cloud of sulfur atom in the receptors; π -π stacked, π-electron cloud between the Aromatic rings; π -π T-shaped, T shaped π-electron cloud between the Aromatic rings; π -alkyl, π-electron cloud between the Aromatic ring of ligand and alkyl group of ligand; Ӑ, Angstrom.
The TME is a complex ecological niche that plays important roles in tumor growth and progression. The TME consists of several components including tumor cells and adjacent non-tumor cells, such as infiltrating immune cells, fibroblasts, signaling molecules, and the extracellular matrix (ECM) (Azizi et al., 2018;Pearce et al., 2018;Zou et al., 2020). Tumor-infiltrating immune cells are well associated with angiogenesis, tumorigenesis, and metastasis of tumor cells, which can in turn regulate the quantity and subsets of immune cells (Taddei et al., 2013). Oncogenic signaling molecules are known to be drivers of immune infiltration and immunotherapeutic responses (Lawal et al., 2021a). Consequently, the present study revealed associations of c-MET/EGFR expressions with infiltration of various immune cells in CRC and predicted poor survival of the cohorts. We also found that both genetic and epigenetic alterations of c-MET/EGFR were associated with immune infiltration, dysfunctional T-cell phenotypes, and poor prognoses of CRC patients, hence adding another layer of prognostic relevance of c-MET/EGFR in CRC.
CAFs are apoptotic-resistant fibroblasts that promote tumorigenic and metastatic properties by secreting cytokines and remodeling the ECM (Erez et al., 2010;Giannoni et al., 2010). We found that associations between c-MET/EGFR expression profiles and CAF infiltration resulted in low cumulative survival of CRC cohorts. These findings could be attributed to the fact that oncogenes induce the CAF phenotype (Lisanti et al., 2013), which in turn inhibits T-cell expansion by mediating the hyper-expression of immune checkpoint molecules in the TME, consequently leading to T-cell allergies (Harryvan et al., 2019), immune therapy failure, aggressive tumor growth, and high death rates in those cohorts. Coherent with our results, a previous study reported that CAFs promote EGFR-TKI resistance via c-MET/IGF-1/ANXA2 signaling in NSCLC (Yi et al., 2018).
Another study also reported that CAFs regulate the plasticity of tumor-initiating cells in liver cancer through c-Met/FRA1/HEY1 signaling (Lau et al., 2016), while Lau et al. (2017) reported that CAFs promote the metastasis of ovarian cancer via tumor necrosis factor (TNF)-α/transforming growth factor (TGF)-α/EGFR signaling. Altogether, our study suggested that c-MET/EGFR play important roles in CAF-mediated tumorigenesis and host immune responses to CRC. It is therefore reasonable to conclude that c-MET/EGFR are important biomarker signatures of tumorigenesis, tumor immune invasion, and poor prognoses, and thus can serve as attractive targets for the development of immune/chemotherapeutic strategies against CRC. In addition, our study may be clinically useful in prognosis assessment and follow-up management of immunotherapy of CRC. Immunotherapies are more effective and promising for some tumor patients than others, and our study therefore suggested that c-MET/EGFR could be an important contributor to those less sensitive to CRC immunotherapy compared to other cancer types as reported in a previous study (Banerjea et al., 2005), hence providing a rationale for a combination of c-MET/EGFR antagonists and immunotherapy in CRC. From another perspective, since EGFR aberrations co-occurred with other genetic assaults, developing a potent c-MET/EGFR inhibitor with multi-target potential capable of inhibiting multiple pathways with minimal side effects may open up new windows for long-lasting, multilayer control of CRC. Hence we evaluated NSC777205 and NSC777207, gefitinib-inspired small molecules derivatives of niclosamine, for that purpose.
On the basis of PK studies during the early stages of drug discovery possibly reducing attrition rates (Kola and Landis, 2004), we evaluated NSC777205 and NSC777207 for in silico drug-likeness and ADME profiles, and found that both NSC777205 and NSC777207 possessed good druglikeness properties and obeyed Lipinski's rule-of-five, a physicochemical-based rule set to evaluate drug candidates for suitability as oral remedy (Pollastri, 2010). The number of rotatable bonds is important in the stereoselectivity of drug molecules for optimal binding with target receptor molecules, while the topological polar surface area (TPSA) and molar refractivity are important for a drug's transport and biodistribution (Veber et al., 2002;Prasanna and Doerksen, 2009). Values of the TPSA, molar refractivity, and other PK parameters observed for NSC777205 and NSC777207 were within the permissible range of a good drug molecule. The overall analysis of drug-likeness studies strongly suggested that NSC777205 and NSC777207 possessed good drug-likeness properties for optimal drug transport, bioavailability, intestinal availability, membrane permeability, and eventual interactions with receptor molecules and bioactivity. Interestingly both NSC777205 and NSC777207 also demonstrated in silico cytotoxic activities against colon cancer, brain, lung, and bone cancer cell lines with NSC777205 demonstrating higher activity (Pa 0.31-0.658) than NSC777207 (Pa 0.303-0.565). BBB penetration is an important limiting factor in the use of chemotherapy for treating glioblastomas and other CNSassociated diseases (Moretti et al., 2015). Interestingly, our in silico study suggested that both NSC777205 and NSC777207 had good BBB permeation ability, with NSC777205 demonstrating about 2-fold higher ability than NSC777207. Notably, the synthetic accessibilities of these compounds were lower than 3.5, suggesting that they can relatively easily be synthesized. The overall analysis indicated that NSC777205 demonstrated more-favorable drug likeness, ADMET, and cytotoxic properties than NSC777207, thus suggesting better therapeutic prospects of the former than the latter. Considering these promising PK, ADMET, and biological properties, we synthesized these compound by adapting and optimizing our previously established protocols (Chen et al., 2014) and evaluated their anticancer activities against the NCI-60 human tumor cell lines.
To our delight, the in vitro anticancer study revealed that both NSC777205 and NSC777207 exhibited wide spectra of antiproliferative activities with GI 50 values of <6 µM against NCI-60 cell line panels with selective cytotoxic preferences for melanoma, renal, CNS, colon, and NSCLC cell lines, suggesting that the compounds were not generally toxic to growing cells but were selective for melanoma, renal, CNS, colon, and NSCLC cells. This observation corroborates with the in silico structure-activity-related cytotoxic predictions for NSC777205 and NSC777207 against cancer cell lines, which indicated that both compounds exhibited cytotoxic activities against colon (HCT-116 and RKO) brain (Hs683), lung (A549) and bone cancer (SJSA-1) cell lines with NSC777205 demonstrating higher activity (Pa 0.31-0.658) than NSC777207 (Pa 0.303-0.565). Of importance, among all cell lines, colon cancer cell lines stood out uniquely with all seven cell lines demonstrating strict antiproliferative sensitivity and no cytotoxic response to single-dose (10 μM) treatment of either NSC777205 or NSC777207. This suggests that NSC777205 and NSC777207 would be very useful in growth inhibition and treatment of CRC. Tumors often develop resistance to a wide range of anticancer drugs. Therefore, drug resistance represents an important problem in cancer chemotherapy (Baguley, 2010;Alfarouk et al., 2015). Intriguingly, NSC777205 and NSC777207 demonstrated higher anticancer activities on multidrug resistant NC1-60 human tumor cell lines including the SK-OV-3, OVCAR-3 OVCAR-4, and NCI/ADR-RES cell lines than the corresponding naïve cell lines of the same origin (OVCAR-8 and IGROV1). These cell lines have been well profiled and established to be resistant to multiple clinical drugs (Louie et al., 1985;Lorenzi et al., 2009). It is therefore plausible that since NSC777205 and NSC777207 demonstrated promising activity against these cell lines, they may be useful in treating multidrug-resistant cancers. Furthermore, among the NCI-60 cell lines investigated in this study, the leukemia RPMI-8226 and melanoma SK-MEL-28 cell lines were identified to harbor EGFR mutations which led to their resistance to 12 TKIs, including erlotinib (Ikediobi et al., 2006;Liu et al., 2007). It is therefore noteworthy that both NSC777205 and NSC777207 exhibited high antiproliferative activity against these cell lines, hence suggesting their applicability in treating TKIresistant tumors. In line with the in silico cytotoxic activities and ADMET properties, our in vitro anticancer study also indicated that NSC777205 had higher anticancer activities than NSC777207 against the nine cancer types evaluated. These higher activities of NSC777205 could therefore be attributed to its estimated higher BBB permeation and favorable ADME and physicochemical properties that give it better stability and a greater ability to interact with targets than NSC777207. Hence, we evaluated the possible mechanistic properties of these compounds via an in silico study and molecular docking simulations of ligand-receptor interactions.
Identification of molecular targets is an important aspect of developing novel small molecules for chemotherapeutic purposes, and a number of computational tools that have been developed and widely employed for these purposes have had translational success in preclinical and clinical practice. However, due to differences in embedded principles for target identification by these tools and also the vast diversity of targets, it is recommended to use multiple computational tools and identify common targets predicted for certain drug candidates (frontiers). In compliance with this recommendation, we used four computational tools including the COMPARE algorithm developed by the USA NCI to identify the most probable targets for NSC7772015 and NSC777207. In the COMPARE analysis, the anticancer fingerprints of NSC777205 and NSC777207 corresponded to anticancer fingerprints of known inhibitors of c-MET/EGFR/mTOR/PI3K pathways. Furthermore, the most highly correlated NCI synthetic compound (p 0.78) was 3-(2-fluoro-4-iodophenyl)-7methoxy-2H-benzo[e] (Fidler et al., 2016;Sung et al., 2021) oxazine-2,4(3H)-dione, whose detailed mechanistic targets were reported to be PI3K, AKT, mTOR, MEK, and EGFR (Lawal et al., 2021c). In addition, N-(2,4-difluorophenyl)-2', 4'-difluoro-4-hydroxybiphenyl-3-carboxamide and (4-(6-(2,4-difluorophenyl)-2,4-dioxo-2H-benzo[e] (Fidler et al., 2016;Sung et al., 2021)-oxazin-3(4H)-yl) benzonitrile which also showed significant correlations with anticancer fingerprints of NSC777205 and NSC777207 were mechanistically reported, and signal transducer and activator of transcription 3 (STAT3)/cyclin-dependent kinases (CDKs)/PIK3CB/EGFR were identified among other targets (Lawal et al., 2021b). c-MET phosphorylates and triggers intracellular signaling cascades that lead to activation of key molecules such as the extracellular signalregulated kinase 1/2 (ERK1/2), PI3K/AKT signaling, and STAT3 (Organ and Tsao, 2011;Wood et al., 2021). It is therefore plausible that the COMPARE analysis inferred that both NSC777205 and NSC777207 may also alter the expression capacity of STAT3 in addition to the promising mechanistic target of the c-MET/EGFR/PI3K/mTOR pathways. Collectively, we found that c-MET/EGFR/P13K/ mTOR satisfied the target criteria set by the four tools we used. These tools hence suggested that c-MET/EGFR/P13K/ mTOR are the most probable targets and mechanisms of action of NSC7772015 and NSC7772017.
Molecular docking aids the identification of drug lead molecules by unraveling possible interactions between a drug candidate and protein targets, providing structureactivity relationships and therapeutic mechanisms of drug candidates prior to in vitro or in vivo validation. Noncovalent interactions such as hydrogen bonds, hydrophobic, ionic, and π-interactions, and van der Waals forces occurring between ligands and receptors are important for stabilization of a ligand within the binding cavity of a receptor (Zhao and Huang, 2011). Our docking analysis revealed that NSC777205 and NSC777207 docked well into the binding cavities of c-MET, EGFR, mTOR, and P13K by various non-covalent interactions and with binding affinities ranging -6.8∼-8.9 kcal/mol. These collective interactions would lead to strong stabilization of NSC777205 and NSC777207 in receptor cavities, which could lead to competitive or non-competitive alterations in the normal expressions and functional properties of the proteins, thus compromising the sustenance and survival of the oncogenic pathways that depend on the proteins for their hyperactivity. Hydrophobic interactions play important roles in the formation of ligand-receptor complexes. Although NSC777205 and NSC777207 shared common interacting residues with the receptors, in most cases, NSC777205 demonstrated more-robust and a higher number of hydrogen bonds, hydrophobic contacts, and van der Waals forces with the receptors than did NSC777207. These higher hydrophobic interactions would enhance the stabilization of the energetically favored NSC777205 in an open conformation surrounding the receptors and further increase its biological activity (Patil et al., 2010;Lawal et al., 2021c). In addition, the higher van der Waals forces would create a strong cohesive environment, that further stabilizes the complex (Arthur and Uzairu, 2019). Both NSC777205 and NSC777207 demonstrated similar tendencies for inhibiting c-MET and EGFR as revealed by the most common interacting residues and similar binding affinities to the binding sites of c-MET and EGFR; however, NSC777205 had higher hydrophobic contacts, H-bonds, halogens, and π-interactions with EGFR than did NSC777207. In addition, the higher number of H-bonds and van der Waals forces around the ligand backbone could be responsible for the higher affinity that NSC777205 had for PI3K than did NSC777207.
Both NSC777205 and NSC777207 exhibited high similarity with gefitinib and crizotinib, known clinical EGFR and c-MET inhibitors, respectively, in terms of common interacting residues with EGFR and c-MET and similar binding affinities with EGFR, suggesting that both NSC777205 and NSC777207 could be as promising as gefitinib and crizotinib at inhibiting EGFR and c-MET, serving as potential dual targets and thus superseding the strength of individual clinical drugs in arresting cancer proliferation and aggressiveness for which EGFR and c-MET are implicated. This multi-target potency of NSC777205 and NSC777207 could be attributed to the observed activities of NSC777205 and NSC777207 against NCI cell lines that are known to be multi-drug resistant including those that are resistant to more than 10 panels of TKIs (Liu et al., 2007).
Collectively, our study revealed that structural hybridization strategies yielded novel and potential c-MET/EGFR/PI3K/mTOR multi-target compounds, NSC777205 and NSC777205, with promising prospects for treating melanoma, renal, CNS, colon, and NSCLC cell lines. Although NSC777205 and NSC777207 share common interacting residues with the receptors, the overall structure-activity relationship study revealed that the addition of an -OCH 3 group to the salicylic core of NSC777207 was not favorable, as the added moiety did not contribute significant interactions with the receptorbinding residues and also led to overall less-favorable ADMET, physicochemical, and drug-likeness properties and weaker anticancer activities against the panel of NCI cell lines compared to the properties and activities demonstrated by NSC777205 that has no -OCH 3 substituent group. However, further in vitro and in vivo analyses in tumor-bearing mice are ongoing in our lab to support this claim and to unravel the full therapeutic efficacies of NSC777205 and 777207 in CRC.

CONCLUSION
In conclusion, our results indicated that, c-Met/EGFR are important biomarker signatures of cancer-associated fibroblasts and tumor immune infiltration, and are of clinical relevance in colorectal cancer. Genetic and epigenetic alterations of c-MET/ EGFR were associated with dysfunctional T-cell phenotypes, and poor prognoses of the cohorts. c-MET/EGFR/PI3K/mTOR are therapeutic targets for NSC777205 and NSC777207 with consequent selective cytotoxic preferences for melanoma, renal, CNS, colon, and NSCLC cell lines. However, the removal of an -OCH 3 substituent group from the salicylic core of NSC777205 was responsible for more highly favorable druglikeness properties and anticancer activities, and more-robust interactions with c-MET/EGFR/PI3K/mTOR than those of NSC777207.

DATA AVAILABILITY STATEMENT
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation, to any qualified researcher.

AUTHOR CONTRIBUTIONS
BL wrote the manuscript; helped Y-CW, with data collection and analyses; H-SH synthesized and provided NSC777205 and NSC777207; Y-CW, AW and H-SH designed and oversaw the study. All authors have read and agreed to the published version of the manuscript.