Preclinical Repurposing of Sitagliptin as a Drug Candidate for Colorectal Cancer by Targeting CD24/CTNNB1/SOX4-Centered Signaling Hub

Despite significant advances in treatment modalities, colorectal cancer (CRC) remains a poorly understood and highly lethal malignancy worldwide. Cancer stem cells (CSCs) and the tumor microenvironment (TME) have been shown to play critical roles in initiating and promoting CRC progression, metastasis, and treatment resistance. Therefore, a better understanding of the underlying mechanisms contributing to the generation and maintenance of CSCs is crucial to developing CSC-specific therapeutics and improving the current standard of care for CRC patients. To this end, we used a bioinformatics approach to identify increased CD24/SOX4 expression in CRC samples associated with poor prognosis. We also discovered a novel population of tumor-infiltrating CD24+ cancer-associated fibroblasts (CAFs), suggesting that the CD24/SOX4-centered signaling hub could be a potential therapeutic target. Pathway networking analysis revealed a connection between the CD24/SOX4-centered signaling, β-catenin, and DPP4. Emerging evidence indicates that DPP4 plays a role in CRC initiation and progression, implicating its involvement in generating CSCs. Based on these bioinformatics data, we investigated whether sitagliptin, a DPP4 inhibitor and diabetic drug, could be repurposed to inhibit colon CSCs. Using a molecular docking approach, we demonstrated that sitagliptin targeted CD24/SOX4-centered signaling molecules with high affinity. In vitro experimental data showed that sitagliptin treatment suppressed CRC tumorigenic properties and worked in synergy with 5FU and this study thus provided preclinical evidence to support the alternative use of sitagliptin for treating CRC.


Introduction
Colorectal cancer (CRC) is the third most prevalent cancer, leading to high mortality in women and men worldwide [1,2].Despite the advances in surgical procedures, systemic treatments, overall survival, and disease-free survival in patients with advanced metastatic CRC remain poor [3,4].Distant metastasis and the development of treatment resistance have been attributed to a subpopulation of CRC cells, commonly referred to as cancer stem cells (CSCs) [5,6].CSCs are defined as possessing unlimited cell division and self-renewal ability and are characterized by expressing CD24CD24, CD44, CD133, OCT4, aldehyde dehydrogenase 1 (ALDH1), and CXC-chemokine receptor 4 (CXCR4) [7].The presence and emergence of CSCs are associated with disease progression, treatment failure, and relapse.Studies have indicated that targeting and eliminating CSCs is a rational and potential venue for achieving the complete remission of CRC [8].However, the CRC tumor core and tumor microenvironment (TME) are highly heterogeneous and complex, presenting a daunting task for CSC-specific therapeutic development.Thus, a better understanding of CSC biology and its interactions with the TME is essential.
Vigorous efforts have been invested in identifying CSC markers, and recently, dipeptidylpeptidase 4 (DPP4, also known as CD26) has been reported as a marker of stemness (CSCs) in several cancers and has been associated with distant metastasis [9][10][11][12].DPP4 is a multifunctional cell surface protein reported to play a pivotal role in various cancer types, including CRC, through its interaction with the tumor microenvironment and is widely expressed in the T-lymphocytes [13][14][15].When upregulated in CRC, DPP4 was shown to promote cancer development, progression, and metastasis, making it a potential and novel therapeutic target for CRC [16][17][18].Notably, DPP4 inhibitors (DPPi) have been developed and safely used for treating type II diabetes.This discovery prompted researchers to investigate DPP4i's potential usage for treating cancer patients.
Multiple studies have shown that approximately 70% of CRC patients exhibit high expression levels of CD24, which has also been demonstrated to possess CSC properties [19,20].Others have shown that the genetic alterations on β-catenin prompt the βcatenin/T cell transcription factor (TCF) signaling in CRC [21,22].β-catenin/TCF4 pathway at the transcriptional level is regulated by SOX4 [23].SOX4 has recently been identified in various cancer types and implicated in cancer initiation.However, little is known about its involvement in CSCs.These observations suggest molecular connections between CD24/SOX4-centered signaling and DPP4/β-catenin, making them potential therapeutic targets.Gene association and enrichment analyses showed that CD24/SOX4-centered signaling hub was associated with cancer cell metabolism, immune responses, ECM regulation, and Wnt/β-catenin pathways.Based on these preliminary data, we aim to provide preclinical support for sitagliptin (a DPP4 inhibitor) as an inhibitor of CSC in CRC.

Identification of Differentially Expressed Genes (DEGs) in CRC
Differentially expressed genes (DEGs) from four (4) GEO microarray datasets of CRC patients (accession numbers GSE21510, GSE32323, GSE110223, and GSE110224) were selected based on the obtained p-values (<0.05), and cut-off values.The expression data based on the number of patients and number of expressed genes, including upregulated and downregulated genes, are represented in volcano plots (Figure 1A-D); moreover, the 59 overlapping upregulated DEGs can be obtained from the constructed Venn diagram (Figure 1E).Furthermore, all the overexpressed genes obtained from overlapping genes are presented on a heatmap diagram (Figure 1F).

CD24/CTNNB1/SOX4 Are Overexpressed in Colorectal Cancer (CRC)
The 59 overlapping genes were obtained from microarray datasets, including CD2 CTNNB1, and SOX4 oncogenes.To validate this, we explored GEPIA2 online bioinform ics tool with default settings.As anticipated, the obtained results showed that the mRN levels of CD24/CTNNB1/SOX4 oncogenes were upregulated in CRC tissues as compar with normal tissues (Figure 2A-C).Moreover, we used the HPA as an independent to to validate the expression of CD24/CTNNB1/SOX4 in CRC tissue, and the results reveal the high expression levels of these oncogenes in CRC tissues as compared with norm tissues, with medium staining intensities displayed on each tissue, and p < 0.05 being co sidered statistically significant (Figure 2D-I).For further analysis, we explored the c oportal tool to predict the co-expression of CD24/CTNNB1/SOX4 in CRC.Based on t analysis, CD24 with SOX4, CTNNB1 with CD24, and SOX4 with CD24 were found to co-expressing in CRC, with positive Spearman and Pearson values and a p-value < 0. considered significant (Figure 2J-L), with p < 0.05 considered statistically significant.LGALS2, HMGCS2, ZG16, CTNNB1, GUCA2, GCG, CLCA1, CD24, CD177, AQP8, and ABCG2).

CD24/CTNNB1/SOX4 Are Overexpressed in Colorectal Cancer (CRC)
The 59 overlapping genes were obtained from microarray datasets, including CD24, CTNNB1, and SOX4 oncogenes.To validate this, we explored GEPIA2 online bioinformatics tool with default settings.As anticipated, the obtained results showed that the mRNA levels of CD24/CTNNB1/SOX4 oncogenes were upregulated in CRC tissues as compared with normal tissues (Figure 2A-C).Moreover, we used the HPA as an independent tool to validate the expression of CD24/CTNNB1/SOX4 in CRC tissue, and the results revealed the high expression levels of these oncogenes in CRC tissues as compared with normal tissues, with medium staining intensities displayed on each tissue, and p < 0.05 being considered statistically significant (Figure 2D-I).For further analysis, we explored the cBioportal tool to predict the co-expression of CD24/CTNNB1/SOX4 in CRC.Based on the analysis, CD24 with SOX4, CTNNB1 with CD24, and SOX4 with CD24 were found to be co-expressing in CRC, with positive Spearman and Pearson values and a p-value < 0.05 considered significant (Figure 2J-L), with p < 0.05 considered statistically significant.in normal colon tissues compared with CRC tissues.Overexpression of CD24/CTNNB1/SOX4 is associated with poor prognosis in CRC.Co-expression of (J) CTNNB1 with CD24 (K) CTNNB1 with SOX4, (L) SOX4 with CD24 in CRC.The staining quality was <75%, and * p < 0.05 was considered statistically significant.

Increased CD24/CTNNB1/SOX4 Expression Is Associated with Poor Prognosis in Colon Cancer Patients
We first searched and analyzed a dataset containing 12 CRC patients' samples (66,050 cells using a 10× genomics platform) [24] (Figure 3A).We first explored an established cancer stemness marker, CD24 (Figure 3B) and an EMT marker for SOX4 expression (Figure 3C).CD24 expression was predominantly in the malignant and epithelial cell clusters.CD24/ CTNNB1/SOX4 expression was detected in multiple cell clusters, including malignant, endothelial, epithelial, fibroblast, myofibroblast, and DC cell clusters (Figure 3D).We then queried the TCGA datasets using GEPIA2 software.We found that increased CD24/CTNNB1/ SOX4 expression was associated with shorter overall survival (Figure 3E,F We first searched and analyzed a dataset containing 12 CRC patients' samples (66,050 cells using a 10× genomics platform) [24] (Figure 3A).We first explored an established cancer stemness marker, CD24 (Figure 3B) and an EMT marker for SOX4 expression (Figure 3C).CD24 expression was predominantly in the malignant and epithelial cell clusters.CD24/CTNNB1/SOX4 expression was detected in multiple cell clusters, including malignant, endothelial, epithelial, fibroblast, myofibroblast, and DC cell clusters (Figure 3D).We then queried the TCGA datasets using GEPIA2 software.We found that increased CD24/CTNNB1/SOX4 expression was associated with shorter overall survival (Figure 3E,F).
signaling networks and participated in cellular communication, immune/metabolism ulation, and ECM remodeling.

Heterogeneity of CRC Cancer-Associated Fibroblasts (CAFs) and Exploration of CD24+ CAFs in CRC
A seminal study by Qi et al. indicated that heterogenous CAF populations infilt the CRC tumor microenvironment [25].One subtype of CRC-infiltrating CAFs was CD (the first column denoted by the red asterisk, Figure 5A).High expression level COL1A1 and COL3A characterize these CD24+ CAFs.Further analyses classified diffe CAF-associated gene sets by transcriptional factor (TF) regulon.CD24+ CAFs were fo which were most significantly associated with TF regulon: POU5F1 (OCT6), PKNOX2, BPA, and ZNF71 [25] (Figure 5B).This TF regulon participates in the Wnt-mediated morigenesis, and multiple intracellular intrinsic and innate immune systems [26,27].A the TIMER2.0program analysis results showed that increased SOX4 expression was p itively associated with the increased expression of multiple oncogenes, including EG FN1, KRAS, and mTOR (Figure 5C).CTNNB1/SOX4 expression in CRC tumors was nificantly associated with tumor-infiltrating CAFs (Figure 5D).

Heterogeneity of CRC Cancer-Associated Fibroblasts (CAFs) and Exploration of CD24+ CAFs in CRC
A seminal study by Qi et al. indicated that heterogenous CAF populations infiltrate the CRC tumor microenvironment [25].One subtype of CRC-infiltrating CAFs was CD24+ (the first column denoted by the red asterisk, Figure 5A).High expression levels of COL1A1 and COL3A characterize these CD24+ CAFs.Further analyses classified different CAFassociated gene sets by transcriptional factor (TF) regulon.CD24+ CAFs were found which were most significantly associated with TF regulon: POU5F1 (OCT6), PKNOX2, CEBPA, and ZNF71 [25] (Figure 5B).This TF regulon participates in the Wnt-mediated tumorigenesis, and multiple intracellular intrinsic and innate immune systems [26,27].Also, the TIMER2.0program analysis results showed that increased SOX4 expression was positively associated with the increased expression of multiple oncogenes, including EGFR, FN1, KRAS, and mTOR (Figure 5C).CTNNB1/SOX4 expression in CRC tumors was significantly associated with tumor-infiltrating CAFs (Figure 5D).The results were generated using TIMER2.0software.

In Silico Molecular Docking Analysis of Sitagliptin against CD24 with Sitagliptin
After establishing the DDP4′s connection with a CD24/SOX4-centered signaling hub, we decided to explore whether the DPP4 inhibitor, sitagliptin, could target colon CSCs.We applied molecular docking simulations using the Autodock vina tool, which showed the high binding abilities of sitagliptin when docked with DPP4 and CD24.The Gibbs free energy obtained from the sitagliptin-DPP4/CD24 (PDB:2ONC) complex exhibited a binding affinity of (Δ = −7.4 and Δ = −9.6 kcal/mol), respectively (Figure 6A,B).Our careful analysis of the interactions between sitagliptin and DPP4 and CD24 revealed a higher number of van der Waals forces, salt bridge, pi-pi stacked, pi-pi shaped, pi-alkyl, and most importantly, two conventional hydrogen bonds and their minimal distance constraints, as shown in the accompanying table Interestingly, sitagliptin is a DPP4-specific inhibitor, and our docking analysis predicted that sitagliptin was bound to CD24 with an even higher affinity than DPP4.The results were generated using TIMER2.0software.

In Silico Molecular Docking Analysis of Sitagliptin against CD24 with Sitagliptin
After establishing the DDP4 ′ s connection with a CD24/SOX4-centered signaling hub, we decided to explore whether the DPP4 inhibitor, sitagliptin, could target colon CSCs.We applied molecular docking simulations using the Autodock vina tool, which showed the high binding abilities of sitagliptin when docked with DPP4 and CD24.The Gibbs free energy obtained from the sitagliptin-DPP4/CD24 (PDB:2ONC) complex exhibited a binding affinity of (∆ = −7.4 and ∆ = −9.6 kcal/mol), respectively (Figure 6A,B).Our careful analysis of the interactions between sitagliptin and DPP4 and CD24 revealed a higher number of van der Waals forces, salt bridge, pi-pi stacked, pi-pi shaped, pi-alkyl, and most importantly, two conventional hydrogen bonds and their minimal distance constraints, as shown in the accompanying table Interestingly, sitagliptin is a DPP4-specific inhibitor, and our docking analysis predicted that sitagliptin was bound to CD24 with an even higher affinity than DPP4.

Molecular Targeting of β-Catenin and SOX4 by Sitagliptin
Next, we performed a molecular docking analysis for key molecules in the CD24/SOX4-centered signaling hub, SOX4, and β-catenin by association.We demonstrated that sitagliptin formed stable interactions with β-catenin (or CTNNB1) (PDB: IJPW)

Molecular Targeting of β-Catenin and SOX4 by Sitagliptin
Next, we performed a molecular docking analysis for key molecules in the CD24/SOX4centered signaling hub, SOX4, and β-catenin by association.We demonstrated that sitagliptin formed stable interactions with β-catenin (or CTNNB1) (PDB: IJPW) and SOX4 (PDB:3U2B).Specifically, β-catenin and SOX4 formed a stable complex with sitagliptin exhibiting Gibbs free energies of ∆ = −7.4 and ∆ = −9.1 kcal/mol, respectively (Figure 7A,B).We used the discovery studio (BIOVIA) to analyze our results further.We identified interactions between sitagliptin, which displayed van der Waals interactions, halogens, pi-cation, alkyl, three conventional hydrogen bonds, and the minimum binding distances, as shown in the accompanying table below. .We used the discovery studio (BIOVIA) to analyze our results further.We identified interactions between sitagliptin, which displayed van der Waals interactions, halogens, pi-cation, alkyl, three conventional hydrogen bonds, and the minimum binding distances, as shown in the accompanying table below.(2.81 Å) and TRY631 (2.64 Å), for SOX4 in complex with sitagliptin.The accompanying table shows several interactions, including amino acids, van der Waals forces, carbon-hydrogen bonds, halogens, and pi-pi stacked.

Sitagliptin Decreases the Viability of CRC Cells
The efficacy of sitagliptin on the viability of CRC cells was evaluated to confirm the predicted outcomes.The results indicated that sitagliptin reduced the viability of DLD-1 and HCT116 cells with IC50 values of 140 µM and 270 µM, respectively (Figure 8A).Additionally, sitagliptin was found to have inhibitory effects on migration, colony and sphere formation of DLD-1 and HCT116 cells (Figure 8B-E).

Sitagliptin Decreases the Viability of CRC Cells
The efficacy of sitagliptin on the viability of CRC cells was evaluated to confirm the predicted outcomes.The results indicated that sitagliptin reduced the viability of DLD-1 and HCT116 cells with IC50 values of 140 µM and 270 µM, respectively (Figure 8A).Additionally, sitagliptin was found to have inhibitory effects on migration, colony and sphere formation of DLD-1 and HCT116 cells (Figure 8B-E).

Sitagliptin Synergistically Enhanced the Anti-Tumorigenic Activities of 5FU in CRC Cells
The synergetic combination index (CI index) in the range of 0.06-6.29 and 0.65-6.75 was obtained for DLD-1 and HCT116 (Figure 9A,B), thus indicating that sitagliptin and 5FU act synergistically to suppress the viability of CRC.Furthermore, colony formation assay demonstrated that, the combination of sitagliptin and 5FU significantly decreased the colony formation of DLD-1, as compared with 5FU single treatment (Figure 9C).Moreover, the combination treatment also suppressed the sphere forming ability of DLD-1 and HCT116, as compared with single treatment (Figure 9D).Accordingly, western blot analysis following combination of sitagliptin and 5FU, exhibited more suppressive effects on DPP4, CD24, β-catenin, and SOX4 centered signaling hub, and GAPDH was used as internal control, as compared with a single treatment (Figure 9E).

Sitagliptin Synergistically Enhanced the Anti-Tumorigenic Activities of 5FU in CRC Cells
The synergetic combination index (CI index) in the range of 0.06-6.29 and 0.65-6.75 was obtained for DLD-1 and HCT116 (Figure 9A,B), thus indicating that sitagliptin and 5FU act synergistically to suppress the viability of CRC.Furthermore, colony formation assay demonstrated that, the combination of sitagliptin and 5FU significantly decreased the colony formation of DLD-1, as compared with 5FU single treatment (Figure 9C).Moreover, the combination treatment also suppressed the sphere forming ability of DLD-1 and HCT116, as compared with single treatment (Figure 9D).Accordingly, western blot analysis following combination of sitagliptin and 5FU, exhibited more suppressive effects on DPP4, CD24, β-catenin, and SOX4 centered signaling hub, and GAPDH was used as internal control, as compared with a single treatment (Figure 9E).

Discussion
Despite the significant advances in treatment modalities, colorectal cancer (CRC) remains one of the leading malignancies, with high numbers of reported cases and causes of death globally [28].The major causes of CRC-related deaths are associated with cancer recurrence and metastasis.Most studies have shown that approximately 60% of CRC patients are expected to develop metastasis [29,30].Although the current approaches for the treatment of CRC have improved significantly over the last decade, the overall survival rate for most patients is still under 5 years.Several studies have identified subpopulations of CRC that are more resistant to therapeutics, such as chemotherapy and radiotherapy [5,6,8].Chemoresistance remains an obstacle in CRC patients and is a major factor in Fluorouracil (5FU) therapeutic resistance in CRC patients [31].5FU-based therapy comes with severe adverse effects, and many patients eventually develop resistance [32,33].Thus, there is an urgent need to identify novel and reliable biomarkers associated with cancer stemness and drug resistance for drug development in CRC.Dipeptidyl peptidase 4 (DPP4) oncogene is frequently expressed in cancer tissue and has been reported to promote cancer development, progression, and metastasis in CRC, making it a potential and novel therapeutic target [16][17][18].Sitagliptin, an anti-diabetic drug with cytotoxic activities, was approved by the FDA in 2006 as an inhibitor of DPP4 [15,34,35].Recently, studies have shown that sitagliptin inhibits mCRC in vivo [9,14].In the present study, we evaluated the anticancer activities of sitagliptin as a target for CD24/CTNNB1/SOX4, which

Discussion
Despite the significant advances in treatment modalities, colorectal cancer (CRC) remains one of the leading malignancies, with high numbers of reported cases and causes of death globally [28].The major causes of CRC-related deaths are associated with cancer recurrence and metastasis.Most studies have shown that approximately 60% of CRC patients are expected to develop metastasis [29,30].Although the current approaches for the treatment of CRC have improved significantly over the last decade, the overall survival rate for most patients is still under 5 years.Several studies have identified subpopulations of CRC that are more resistant to therapeutics, such as chemotherapy and radiotherapy [5,6,8].Chemoresistance remains an obstacle in CRC patients and is a major factor in Fluorouracil (5FU) therapeutic resistance in CRC patients [31].5FU-based therapy comes with severe adverse effects, and many patients eventually develop resistance [32,33].Thus, there is an urgent need to identify novel and reliable biomarkers associated with cancer stemness and drug resistance for drug development in CRC.Dipeptidyl peptidase 4 (DPP4) oncogene is frequently expressed in cancer tissue and has been reported to promote cancer development, progression, and metastasis in CRC, making it a potential and novel therapeutic target [16][17][18].Sitagliptin, an anti-diabetic drug with cytotoxic activities, was approved by the FDA in 2006 as an inhibitor of DPP4 [15,34,35].Recently, studies have shown that sitagliptin inhibits mCRC in vivo [9,14].In the present study, we evaluated the anticancer activities of sitagliptin as a target for CD24/CTNNB1/SOX4, which were previously demonstrated to promote CRC progression, stemness, and metastasis [36][37][38].Studies have shown that DPP4 modulates the T-cell mechanism and promotes tumor initiation [39][40][41].Moreover, others have demonstrated the significant role of DPP4 in CRC development, progression, and association with therapeutic resistance [15,[42][43][44].
DPP4 has also been demonstrated to interact with the tumor microenvironment [13][14][15], suggesting its essential role in CRC.Herein, we explored computational simulations to identify potential target genes upregulated in CRC.We performed transcriptomics data analysis from GEO, which revealed 59 upregulated oncogenes in CRC tissues, compared with normal tissue; interestingly, among the top genes expressed were CD24/CTNNB1/SOX4.To validate the expression of this oncogenic signature in CRC, we utilized the GEPIA2 bioinformatics tool, which confirmed the overexpression of these oncogenes as expected.To validate further, explored the human protein atlas to identify the expression of these oncogenes at a tissue level and found high expression levels of these genes in CRC tissues as compared with tissues with medium-staining intensities displayed on each tissue.To further analyze, we found that high expressions of CD24, CTNNB1, and SOX4 oncogenic signatures were associated with a shorter overall survival rate in CRC, as shown on the Kaplan-Meier plots (Figure 3D-F).
To analyze the possible protein-protein interactions (PPI) among DPP4, CTNNB1, CD24, and SOX4 oncogenic signatures, we applied the STRING analytic tool to try to connect DPP4 with CD24/CTNNB1/SOX4 signaling.We identified high-confidence interaction between DPP4 and CD24/CTNNB1/SOX4; these interactions were associated with gene neighborhood, gene fusion, gene co-occurrence, and co-expression among the genes, as well as enriched gene ontology involving biological processes (BP), and KEGG pathways, which, interestingly, were associated with immune response, hence suggesting the potential inhibitory effects of the DPP4 inhibitor on CD24/CTNNB1/SOX4 oncogenic signaling.
These findings prompt us to predict the protein-ligand interaction further, using molecular docking to evaluate the potential binding affinities of sitagliptin with DPP4/CTNNB1/ CD24/SOX4 oncogenic signatures.The results from the in silico molecular docking analysis revealed unique binding energy between sitagliptin with DPP4, CTNNB1, CD24, and SOX4 with Gibbs free energy values of (−7.4 kcal/mol, −7.4 kcal/mol, −9.6 kcal/mol, and −9.1 kcal/mol); these observations strongly suggested that sitagliptin could form a stable complex with DPP4/CTNNB1/CD24/SOX4 oncogenic signature.We further investigated sitagliptin's therapeutic potential using in vitro models and found that sitagliptin and 5FU acted synergistically to suppress the viability of CRC cells and decreased migration ability, colony, and tumorsphere formation compared with 5FU single treatment.Collectively, these results suggest that a combination of sitagliptin and 5FU could serve as a novel strategy to improve the efficacy of 5FU in colorectal cancer.

Microarray Data Acquisition and Identification of Differentially Expressed Genes in CRC
The transcriptomic data were downloaded from the Gene Expression Omnibus (GEO) database (https://www.ncbi.nlm.nih.gov/geo/accessed on 4 April 2023) [45], accession numbers GSE21510 [46], GSE32323 [47], GSE110223 [48], and GSE110224 [48]; microarray data processing were processed and analyzed using GEO2R, an R-based web application that identifies differentially expressed genes (DEGs) [45].The determination of the correct fold change cut-off value in the transcriptomic data analysis remains a great challenge [49]; herein, we considered the Benjamin-Hochberg accuracy methods for p-value adjustment of the false discovery rate (FDR), as previously reported [50].We considered p value less than 0.05 and fold change greater than 1.5 as the criteria for DEG selection to the upregulated genes in tumor samples compared with normal adjacent samples.Bioinformatics and evolutionary genomic diagram was used for constructing Venn diagrams to visualize overlapping (http://bioinformatics.psb.ugent.be/webtools/Venn/accessed on 4 April 2023) [51].

Identification of Differentially Expressed Genes (DEGs) in CRC
The messenger-RNA (mRNA) levels of differentially expressed genes (DEGs) of in tumorous versus normal tissues from various cancers from The Cancer Genome Atlas (TCGA) database were analyzed using GEPIA2 (http://gepia2.cancer-pku.cn/accessed on 4 April 2023) bioinformatics software with default settings.To validate the expression levels of identified DEGs in CRC [52], to validate these results, we explored the Human Protein Atlas (HPA) database for immunohistochemistry (IHC) (https://www.proteinatlas.org/ accessed on 4 April 2023) to compare expression levels between tumor samples and normal samples.

Interpretation of Gene Co-Expression in CD24/CTNNB1/SOX4 Genes Network
Interpretation of gene expression network was analyzed using the Network Analyst 3.0 tool (https://www.networkanalyst.ca/accessed on 28 May 2023), a comprehensive analytical visual platform which integrates PPI networks and gene co-occurrence networks and interprets gene expression networks.Herein, we used Enrichment Map, a sub-tool of network analyst, with a Bipartite view, to determine the enrichment of co-expressed genes in KEGG pathways [57][58][59].

Distribution and Expression of CD24/CTNNB1/SOX4 in Colon Tissue
A clear understanding of single-cell RNA expression in a specific cell type in colon tissue is crucial for the investigation of specific and potential cancer biomarkers that may promote tumorigenesis [60].We used the HPA database (https://www.proteinatlas.org/accessed on 28 May 2023) to retrieve the cell-type atlas showing the single-cell RNA sequencing (scRNA-seq) data of DPP4, CTNNB1, CD24, and SOX4 expression in colon tissue [61].The expression of each gene in the specified cell types was explored using the interactive UMAP plots and bar chart [62].

Cell Culture and Reagents
The DLD-1 and HCT116 human colon cancer cell lines were acquired from the American Type Culture Collection (ATCC) based in Manassas, VA, United States of America.In brief, each cell line was cultured and subsequently passaged at 90% cell confluence using Dulbecco's Modified Eagle Medium (DMEM) sourced from Invitrogen, Life Technologies in Carlsbad, CA, USA.The cells were then stored under standard incubator conditions, maintaining a temperature of 37 • C with 5% humidified CO 2 .

Cell Viability Assay
We assessed cell viability using sulforhodamine B (SRB reagent from Sigma-Aldrich, Taipei, Taiwan, following established protocols) [63].HCT116 and DLD-1 cells at confluency of more 90% were plated at 7000 cells/well in 96-well plates for 24 h.After a 48 h treatment with sitagliptin, we added 10% TCA, refrigerated at 4 • C for an hour, washed twice with distilled water, and stained with 0.4% SRB for 30 min at room temperature.After two washes with 1% acetic acid, plates air-dried overnight.The protein-bound stain was dissolved in a 20 mM Tris-buffer solution on an orbital shaker for 15 min, and absorbance was measured at 560 nm wavelength using a microplate reader (Molecular Devices, Sunnyvale, CA, USA).

Cell Migration Assay
CRC cells (HCT116 and DLD-1) were seeded at a density of 10 5 cells in 100 µL of media per well in two-well plates with a silicon insert.After 24 h of incubation, the media were removed, and the insert was gently taken out.Fresh media (1.5 mL) containing sitagliptin (11.25 µM) were added.Wound images were captured immediately (0 h) and after 24 h using an Olympus CKX53 Cell Culture Microscope.Wound closure was quantified with NIH ImageJ software (version 1.8.0) (https://imagej.nih.gov/ij/accessed on 2 July 2023).

Colony-Formation Assays
To evaluate how sitagliptin treatment influences colony formation in colon cancer cell lines, we conducted a colony formation assay based on Franken et al.'s protocol [64].Specifically, 500 cells were seeded in Corning ® 6-well plates (Sigma-Aldrich) and treated with sitagliptin at concentrations equivalent to the 40% inhibitory concentration (IC40) values for HCT116 and DLD-1.The cells were allowed to grow for a minimum of one week.Colony quantification was performed using a Cell3iMager neo-scanner, and inhibitory effects on colonies were expressed as percentages relative to control colonies.

Immunoblot Analysis
The DLD-1 and HCT116 cell line groups, both treated with sitagliptin along with the control group, underwent trypsinization for harvesting.Total protein lysates from treated and untreated cells were collected using RIPA buffer.Subsequently, 20 µg of total lysates were separated via SDS-PAGE using the Mini-Protean III system (Bio-Rad, New Taipei City, Taiwan) and transferred onto polyvinylidene difluoride membranes with the Trans-Blot Turbo Transfer System (Bio-Rad) [65].Membranes were then incubated overnight in the refrigerator at −4 • C with primary antibodies, followed by a 1 h incubation with secondary antibodies the next day, list of antibodies are shown on Table 1.Enhanced chemiluminescence (ECL) detection kits (ECL kits; Amersham Life Science, California, CA, USA) were used to detect proteins of interest, and BioSpectrum ® Imaging System (Upland, CA, USA) was employed for image capture and analysis.[66].Specifically, 2000 colon cancer cells were seeded per well in six-well ultra-low-attachment plates (Corning, Corning, NY, USA) using serum-free media.After 7 days, tumor spheres, characterized as dense, non-adherent spheroid-like masses with a diameter over 50 µm, were counted using an inverted phase-contrast microscope.

Data Analysis
Pearson's correlations were used to assess correlations between DPP4/CTNNB1/CD24/ SOX4 expressions in multiple cancer types.The statistical significance of DEGs was evaluated with * p < 0.05 was accepted as being statistically significant.The Kaplan-Meier curve was employed to present patient survival in different cancer cohorts; Western blot quantification was evaluated using Image J (Version 1.53t).

Conclusions
In summary, we identified overexpression of CD24, CTNNB1, and SOX4 signatures in CRC, revealing their association in promoting tumorigenic properties, cancer stemness, and resistance to therapy.In addition, the in silico molecular docking analysis revealed high binding energies between sitagliptin DPP4, CTNNB1, CD24, and SOX4, predicting its potential inhibitory functions against these targets.Subsequently, we validated the above findings in the CRC cell line model, where the combination of sitagliptin and 5FU suppressed the migration, colony, and tumorsphere formation of DLD-1 and HCT116 cells, accompanied by the reduced DPP4/CTNNB1/CD24/SOX4 expression.

Figure 2 .
Figure 2. CD24/CTNNB1/SOX4 oncogenes are overexpressed in CRC tissues as compared with normal tissues.(A-C), high expression levels of CTNNB1, CD24, and SOX4 in CRC.HPA staining intensity was analyzed based on medium staining (D-I) intensities of CD24, CTNNB1, and SOX4 in normal colon tissues compared with CRC tissues.Overexpression of CD24/CTNNB1/SOX4 is associated with poor prognosis in CRC.Co-expression of (J) CTNNB1 with CD24 (K) CTNNB1 with

Figure 2 .
Figure 2. CD24/CTNNB1/SOX4 oncogenes are overexpressed in CRC tissues as compared with normal tissues.(A-C), high expression levels of CTNNB1, CD24, and SOX4 in CRC.HPA staining intensity was analyzed based on medium staining (D-I) intensities of CD24, CTNNB1, and SOX4 in normal colon tissues compared with CRC tissues.Overexpression of CD24/CTNNB1/SOX4 is associated with poor prognosis in CRC.Co-expression of (J) CTNNB1 with CD24 (K) CTNNB1 with SOX4, (L) SOX4 with CD24 in CRC.The staining quality was <75%, and * p < 0.05 was considered statistically significant.

Figure 5 .
Figure 5. CD24/CTNNB1/SOX4-centered signaling in CRC tumor cells and cancer-associated fibroblasts predicts poor survival in CRC patients.(A) Violin plots demonstrate the heterogeneity of CAFs.A red asterisk marks the CD24+ CAFs in the first column.(B) A heatmap indicates the top transcriptional factor (TF) SCENIC regulons.The top-ranking TFs associated with CD24+ CAF are listed in red.Data in (A,B) were adapted from Qi et al. [25].(C) A table demonstrates SOX4-coexpressing oncogenes.The number in each box indicates Spearman's correlation coefficient.(D,E) Expression level versus CAF-infiltration level plots for CTNNB1/SOX4 in TCGA CRC databases.The results were generated using TIMER2.0software.

Figure 5 .
Figure 5. CD24/CTNNB1/SOX4-centered signaling in CRC tumor cells and cancer-associated fibroblasts predicts poor survival in CRC patients.(A) Violin plots demonstrate the heterogeneity of CAFs.A red asterisk marks the CD24+ CAFs in the first column.(B) A heatmap indicates the top transcriptional factor (TF) SCENIC regulons.The top-ranking TFs associated with CD24+ CAF are listed in red.Data in (A,B) were adapted from Qi et al. [25].(C) A table demonstrates SOX4-co-expressing oncogenes.The number in each box indicates Spearman's correlation coefficient.(D,E) Expression level versus CAF-infiltration level plots for CTNNB1/SOX4 in TCGA CRC databases.The results were generated using TIMER2.0software.

Figure 8 .
Figure 8. Sitagliptin displayed inhibitory effects on the viability of CRC Cells (A).The 50% inhibitory concentration (IC50) values for HCT116 and DLD-1 cell lines.(B,C) Sitagliptin suppressed the migration ability (D) colonies and (E) sphere formation of DLD-1 and HCT116 cells.Images of the colonies and spheroids are shown in the left panel, with quantification of the results presented in the right panel.Statistical significance is denoted by ** (p < 0.05) and *** (p < 0.01).

Figure 8 .
Figure 8. Sitagliptin displayed inhibitory effects on the viability of CRC Cells (A).The 50% inhibitory concentration (IC50) values for HCT116 and DLD-1 cell lines.(B,C) Sitagliptin suppressed the migration ability (D) colonies and (E) sphere formation of DLD-1 and HCT116 cells.Images of the colonies and spheroids are shown in the left panel, with quantification of the results presented in the right panel.Statistical significance is denoted by ** (p < 0.05) and *** (p < 0.01).

Figure 9 .
Figure 9. Sitagliptin treatment reduced tumorigenic properties of CRC cells and enhanced 5FU efficacy.(A) Sita enhanced 5FU efficacy in both DLD-1 and HCT116 cell lines.IC50 values are shown.(B) Combination of sitagliptin and 5FU, improved efficacy of 5FU in CRC CI index of sitagliptin and 5FU combination treatment in DLD-1 and HCT116 cells.The combination index (CI) <1 defines synergism; CI of 1 indicates additive effect while >1 indicates antagonism.The DLD-1 and HCT116 cell lines were exposed to a combination treatment of 5FU and sitagliptin for 7 days.(C) Images of colonies and (D) spheroids were taken, followed by counting.Western blotting analysis showed significant suppressive effects of 5FU with sitagliptin on DPP4, CD24, β-catenin, and SOX4 expression compared with 5FU alone.GAPDH was used as an internal control.Statistical significance is denoted by *** (p < 0.01) (E).

Figure 9 .
Figure 9. Sitagliptin treatment reduced tumorigenic properties of CRC cells and enhanced 5FU efficacy.(A) Sita enhanced 5FU efficacy in both DLD-1 and HCT116 cell lines.IC50 values are shown.(B) Combination of sitagliptin and 5FU, improved efficacy of 5FU in CRC CI index of sitagliptin and 5FU combination treatment in DLD-1 and HCT116 cells.The combination index (CI) <1 defines synergism; CI of 1 indicates additive effect while >1 indicates antagonism.The DLD-1 and HCT116 cell lines were exposed to a combination treatment of 5FU and sitagliptin for 7 days.(C) Images of colonies and (D) spheroids were taken, followed by counting.Western blotting analysis showed significant suppressive effects of 5FU with sitagliptin on DPP4, CD24, β-catenin, and SOX4 expression compared with 5FU alone.GAPDH was used as an internal control.(E) Statistical significance is denoted by *** (p < 0.01).

Author
Contributions: N.M. wrote the manuscript; J.-W.S. and P.-L.W., helped with data collection and analyses; A.T.H.W. and Y.-J.H. designed and oversaw the study.All authors have read and agreed to the published version of the manuscript.Funding: Jing-Wen Shih and Yan-Jiun Huang are supported by the TMU-TMUH research grant (111TMU-TMUH-09).Alexander TH Wu is supported by the following research grants from Taipei Medical University (TMU111-AE2-I14-3) and the National Science and Technology Council (112-2314-B-038-019-). ).