Signaling Pathways in Clear Cell Renal Cell Carcinoma and Candidate Drugs Unveiled through Transcriptomic Network Analysis of Hub Genes

Abstract

Clear cell renal cell carcinoma (ccRCC) is a type of kidney cancer. It advances quickly and often metastasizes, making the prognosis for patients challenging. This study used weighted gene co-expression network analysis (WGCNA) to study gene expression data of different stages of ccRCC obtained in the GEO database. The analysis identified three significant highly preserved gene modules across the datasets: GSE53757, GSE22541, GSE66272, and GSE73731. Functional annotation and pathway enrichment analysis using DAVID revealed inflammatory pathways (e.g., NF-kB, Hippo, and HIF-1 pathways) that may drive ccRCC development and progression. The study also introduced the involvement of viral infections associated with the disease in the metabolic reprogramming of ccRCC. A drug repurposing analysis was also conducted to identify potential drug candidates for ccRCC using the upregulated and downregulated hub genes. The top candidates are ziprasidone (dopamine and serotonin receptor antagonist) and fentiazac (cyclooxygenase inhibitor). Other drug candidates were also obtained, such as phosphodiesterase/DNA methyltransferase/ATM kinase inhibitors, acetylcholine antagonists, and NAD precursors. Overall, the study’s findings suggest that identifying several genes and signaling pathways related to ccRCC may uncover new targets, biomarkers, and even drugs that can be repurposed, which can help develop new and effective treatments for the disease.

1. Introduction

Renal cell carcinoma (RCC) accounts for 3% of all cancers worldwide. The clear cell type is the most common RCC. It is responsible for 80% of all RCCs [1,2]. A clear cell appearance of the tumor cells, caused by the accumulation of lipids and glycogen in the cytoplasm, is the characteristic feature of ccRCC [3]. RCC is categorized according to the TNM classification system. The cancer stages are categorized in terms of tumor size, degree of lymph nodes, and metastasis [4]. In the cases of RCC, both Stage 1 (≤7 cm) and 2 tumors (≥7 cm) are still confined in the kidney. On the other hand, Stage 3 has already spread to the nearby parts of the kidney or lymph nodes, and Stage 4 indicates metastasis to distant sites [5,6,7,8,9]. In terms of survival rates of patients, stage 1 has 90.4%, stage 2 has 83.4%, stage 3 has 66.0%, and the rate is 9.1% for stage 4 [6,7]. The development of ccRCC mutations or loss of the VHL gene disrupts the oxygen regulation response, which accumulates hypoxia-inducible factors such as HIF-1α and HIF-2α. This, in turn, activates certain genes like vascular endothelial growth factor (VEGF), which contribute to the physiology of the tumor and promote cancer [10,11,12,13]. Inflammation may also worsen the ccRCC [14,15,16,17] since significant leukocyte infiltrates, including CD8+ and CD4+ T cells, are linked with higher tumor grade and poor prognosis in ccRCC [14,18].

The early stages of ccRCC are usually managed with surgery or conventional treatments. Still, as the disease progresses to a later stage, the treatment becomes more complex and requires multiple therapies with a combination of tyrosine kinase/immune checkpoint/mTOR inhibitors [19,20]. Along with that, drug resistance and severe side effects are also considered problems in treating ccRCC [3,20,21,22]. Currently, researchers are exploring how combining traditional treatments with new approaches might address these problems. However, it has been found that simply adding existing therapies together only resolves some of the issues [20]. That is why new treatments are needed. Studies have shown also that treatment responses can vary due to many factors (e.g., genetic differences, inflammation). This emphasizes the need for a deeper understanding of ccRCC [3].

To address these challenges, bioinformatics techniques such as weighted gene co-expression network analysis (WGCNA) can be used. WGCNA analyzes the relationship of genes in line with their expression patterns across different diseases or stages/grades of a particular disease. It helps identify important gene groups that can be analyzed to determine key genes that are significant or related to the disease’s development, progression, and even pathways associated with the disease [23]. By identifying hub/key genes, researchers can also use this to identify existing drugs that can be repurposed based on their gene signatures and patterns. These genes may also uncover significant pathways that can reveal therapeutic opportunities. Recent studies employing WGCNA have explored diseases [24,25] including cancer. This suggests that bioinformatics analysis may lead to discovering new biomarkers and important pathways involved in the disease [24,25,26].

This study used four gene expression datasets (GSE53757, GSE22541, GSE66272, and GSE73731) to analyze different stages of ccRCC. Bioinformatics tools such as WGCNA, DAVID, STRING, and Cytoscape were used to identify gene modules. These gene modules are used to determine functional annotations, pathways, and key genes associated with ccRCC. The identified hub genes were also used to facilitate the repurposing of existing drugs through drug repurposing analysis.

2. Materials and Methods

2.1. Data Retrieval and Preliminary Processing

Gene expression data from DNA microarray analyses of tumor samples from patients with ccRCC stages 1 to 4 were obtained from the Gene Expression Omnibus (GEO) of the National Center for Biotechnology Information (NCBI) [https://www.ncbi.nlm.nih.gov/geo/, accessed on 12 June 2024]. Only datasets pre-processed using the GPL570-HG-U133 Plus 2 Affymetrix Human Genome U133 Plus 2.0 Array were selected and used to ensure consistency and avoid/minimize variations from probe design or normalization differences [24]. The study analyzed 113 samples: 24 from stage 1 (S1), 21 from stage 2 (S2), 27 from stage 3 (S3), and 44 from stage 4 (S4). Detailed information regarding the samples is provided in

Table 1. Information of the GEO datasets used in the study.

Accession No.	GSE53757	GSE22541	GSE66272	GSE73731
Condition	ccRCC Stage 1 (S1)	ccRCC Stage 2 (S2)	ccRCC Stage 3 (S3)	ccRCC Stage 4 (S4)
Type	Expression Profiling by Array
Platform	GPL570—HG-U133 Plus 2 Affymetrix Human Genome U133 Plus 2.0 Array
Source	Primary Tumor Samples
No. of Samples	24	21	27	44

.

The datasets were processed in R (v.4.4.0) using the affy package from Bioconductor v3.18 (www.bioconductor.org, accessed on 27 June 2024). This package performed background correction, quantile normalization, and log-2 transformation of the data. The outliers were identified through boxplots and dendrograms. The control probes were removed to reduce variation. Gene expression data were also filtered for genes above the 20th percentile. Probes in all datasets were retained, and samples lacking values after log-2 transformation were removed using the goodSamplesGenesMs function of the WGCNA R package. The probe IDs were converted to gene symbols with the AnnotationDbi function and the hgu133plus2.db [24].

2.2. Weighted Gene Co-Expression Network Analysis (WGCNA)

2.2.1. Determining the Optimal Soft-Thresholding Power for Scale-Free Network

Scale-free topology fit against the power index (1 to 20) was calculated using the pickSoftThreshold function in the WGCNA package in R. The soft-threshold power (β) was selected as the value that met the scale-free topology criterion. This analyzes gene degree distributions using power-law distributions. This assessment also includes checking the network connectivity with linear patterns on a log–log scale. Additionally, a plot of scale-free topology fit against β was used to identify the power at which the fit stabilized or was sufficiently high. To vary the chosen β, an approximate straight-line relationship was generated and examined using soft-connectivity (k) values [27].

2.2.2. Construction of Networks and Identification of Modules

The adjacency matrices for network construction were created using Pearson’s correlation with the “signed” network type. Topological overlap measure (TOM) dissimilarities were computed by raising the adjacency matrices to β. This emphasizes strong correlations and downweighs weak ones. Gene clustering was performed hierarchically using the flashClust function. Dendrograms were generated using the hclust function and the “average” method. Gene modules were identified using cuttreeHybrid using deep split parameters from 0 to 3 [23]. Stable modules across different parameters suggest good clustering, while varying results indicate a need for further validation.

2.3. Module Preservation Analysis

Module preservation analysis measures how well gene group/module connectivity patterns are maintained across different networks and conditions. It determines whether these gene groups remain biologically significant across several diseases or conditions. Module preservation analysis was carried out using the modulePreservation function of the WGCNA package on ccRCC S1 to S4 datasets. The parameters used are “signed” network type, 1000 permutations, and a minimum module size of 100. Eigengene-based connectivity (kME) was computed with the moduleEigengenes function to identify module genes based on their correlation with module eigengenes [23].

2.4. Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) Pathways Analysis

Genes obtained from each preserved module were subjected to functional and pathway enrichment analysis using the Database for Annotation, Visualization, and Integrated Discovery (DAVID) (https://david.ncifcrf.gov/home.jsp, accessed on 28 June 2024). Gene Ontology (GO) analysis was carried out to identify biological processes (BP), molecular functions (MF), and cellular components associated with ccRCC. The KEGG database was also used to determine important pathways. Results with a p-value of less than 0.05 were considered statically significant and were considered for further analysis.

2.5. Protein–Protein Interaction (PPI) Network Analysis

The networks of each module were constructed using a high confidence level score of 0.7 in STRING v 12.0 (https://string-db.org/, accessed on 6 September 2024). Following that, their screened PPI networks were imported into Cytoscape v3.10.1 for further analysis. Topological analysis was conducted to identify hub genes using the CytoHubba plugin of Cytoscape. The study includes calculating their degree centrality, closeness centrality, betweenness centrality, and maximum neighborhood component (MNC). For each metric, the top 20 hub genes were identified and selected. Then, the hub genes common to all four metrics were identified and used in the study. A separate PPI network was also generated to visualize the connection of these hub genes.

2.6. Survival Analysis of Hub Genes

Survival analysis, including overall survival (OS) and disease-free survival (DFS), was conducted using the Gene Expression Profiling Interactive Analysis (GEPIA) (http://gepia.cancer-pku.cn/, accessed on 6 September 2024) webserver for the identified hub genes. The KIRC patients were categorized into high- and low-expression groups based on the median expression levels of the cancer samples. Then, a plot was generated using the Kaplan–Meier method. Hub genes with a p-value less than 5 were considered statistically significant and included in the study.

2.7. Drug Repurposing Analysis

The top hub genes in each module were classified as upregulated or downregulated using GEO2R. These hub genes were used for drug repurposing analysis using the Drug Repurposing Encyclopedia (DRE) (https://www.drugrep.org/drugrepurposinganalysis, accessed on 6 September 2024) webserver. These identify potential drugs based on transcription profiles from MSigDB (Molecular Signatures Database) and CMap (Connectivity Map) [28]. Only the drugs with a false discovery rate (FDR) below 0.05, a negative Tau, and a known mechanism of action were considered in the analysis.

3. Results

3.1. Weighted Gene Co-Expression Network Analysis (WGCNA)

3.1.1. Scale-Free Network Analysis of the ccRCC Datasets

In WGCNA, gene expression matrices are normalized and outliers are removed from the data [21,27]. This results in filtered data (Figure S1) with outliers removed (Figure S2). Following this, 26,434 common genes are used to build the networks. The large number of genes in the analysis suggests robust analysis [23,24,26,29]. Figure 1 presents the scale-free topology fit index as a function of soft threshold (β) values (1–20). From this figure, the point at which the fit stabilized was at β = 10. Thus, β = 10 was selected to analyze all the datasets since this may promote a robust gene network with biologically relevant cluster construction.

Figure 2 illustrates that the log–log plot for the ccRCC S1 dataset had the highest fit, with an R² value of 0.96. This indicates that the S1 dataset is the most appropriate for constructing the WGCNA network and identifying modules due to its relevant biological expression patterns [30]. Along with that, an R² value near 0.9 is also considered ideal. The higher R² value supports the idea that biological networks tend to exhibit scale-free characteristics, making this dataset a strong candidate for representing functionally important networks.

3.1.2. Network Construction and Module Identification

In WGCNA, meta-analysis consists of mapping module eigengenes from different datasets onto a reference dataset. This influences the robustness of the network and clustering resolution based on TOM 23. For this study, the ccRCC S1 dataset was selected as the reference. It also represents the early stage of ccRCC. A soft threshold power of β = 10 was used to construct the adjacency matrix. The S1 dataset was selected as the reference for the analysis, and the identified modules are displayed in Figure 3. This dataset’s expression profiles may provide a basis for understanding ccRCC S1 to S4. The sensitivity plots (Figure S3) also suggested that these module sets were stable even as sensitivity increased, presumably because of either strong gene correlations or fewer significant clusters. WGCNA identified seven gene co-expression modules as follows, each with a distinct color and a total number of genes: blue (2500 genes), brown (2497 genes), red (1242 genes), grey (2500 genes), green (1557 genes), yellow (2320 genes), and turquoise (2500 genes).

3.2. Module Preservation Analysis

Module preservation examines how well-defined modules are conserved across ccRCC datasets. This analysis reflects the reproducibility of co-expression networks and biological relevance. Module preservation in the ccRCC S1 dataset was assessed from the S2–S4 datasets. Modules with a Z-score value greater than 10 were considered highly preserved, and larger modules represented stronger patterns of connectivity [23,24]. Figure 4 illustrates the Z-score values of each module’s genes from different stages. Seven modules were generated with high preservations of blue, brown, and red. Although the turquoise module had 2500 genes, it was not strongly consistent across all datasets and was excluded. The blue module showed the highest score value in all stages. The number of modules passing the threshold was almost equal across datasets, except for S3, which had a higher score value than S2, before somewhat decreasing in S4. This is perhaps because S4 included genes that were not expressed during earlier stages of ccRCC.

3.3. Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) Pathways Analyses

The datasets were analyzed using eigengene-based connectivity (kME) by correlating the expression profile of each gene with module eigengenes. The genes were ranked by their kME values. The top genes were selected for functional analysis using DAVID GO categories (BP, CC, MF) and KEGG pathways. Enriched GO and KEGG terms for each module are provided in

Table A1. Top KEGG pathways analysis results in the highly preserved modules—Blue, Brown, and Red—from the Kyoto Encyclopedia of Genes and Genomes (KEGG) (https://www.genome.jp/kegg (accessed on 28 June 2024)).

KEGG	Term	Count	p-Value
Blue	Pathways in cancer	68	1.27 × 10−5
	Viral carcinogenesis	33	5.01 × 10−5
	Ras signaling pathway	29	0.01091
	TGF-β signaling pathway	18	0.002582
	Renal cell carcinoma	16	1.63 × 10−4
Brown	Pathways in cancer	65	2.69 × 10−6
	PI3K-Akt signaling pathway	46	3.74 × 10−5
	MAPK signaling pathway	37	4.46 × 10−4
	Hippo signaling pathway	23	7.59 × 10−4
	Renal cell carcinoma	11	0.017668
Red	Pathways in cancer	66	2.69 × 10−6
	MAPK signaling pathway	43	1.50 × 10−6
	PI3K-Akt signaling pathway	41	5.81 × 10−4
	Ras signaling pathway	30	7.19 × 10−4
	HIF-1 signaling pathway	17	0.003199

,

Table A2. Top KEGG pathways analysis results in the other modules—Turquoise, Yellow, and Green—from the Kyoto Encyclopedia of Genes and Genomes (KEGG) (https://www.genome.jp/kegg (accessed on 28 June 2024)).

KEGG	Term	Count	p-Value
Turquoise	Pathways in cancer	63	3.51 × 10−6
	PI3K-Akt signaling pathway	47	7.03 × 10−6
	MAPK signaling pathway	37	2.28 × 10−4
	Hippo signaling pathway	21	0.002751
	mTOR signaling pathway	19	0.013614
Yellow	Metabolic pathways	123	0.029441
	Pathways in cancer	60	6.99 × 10−5
	EGFR tyrosine kinase inhibitor resistance	14	2.31 × 10−3
	NF-kB signaling pathway	14	0.022381
	HIF-1 signaling pathway	14	0.029444
Green	Pathways in cancer	58	5.47 × 10−4
	Human papillomavirus infection	35	0.013714
	Human cytomegalovirus infection	30	8.63 × 10−4
	Viral carcinogenesis	24	0.014283
	Viral myocarditis	16	5.64 × 10−5

,

Table A3. Top GO and pathway analysis results for the blue module, obtained from the Gene Ontology database (http://www.geneontology.org, accessed on 28 June 2024).

Category	Term	Count	p-Value
BP	GO:0045944~positive regulation of transcription by RNA polymerase II	152	2.35 × 10−14
	GO:0045893~positive regulation of DNA-templated transcription	94	4.34 × 10−11
	GO:0006338~chromatin remodeling	69	3.39 × 10−17
	GO:0000398~mRNA splicing, via spliceosome	47	2.93 × 10−14
	GO:0008380~RNA splicing	38	5.65 × 10−9
CC	GO:0005829~cytosol	656	6.62 × 10−79
	GO:0005634~nucleus	627	2.76 × 10−50
	GO:0005737~cytoplasm	550	9.12 × 10−34
	GO:0005654~nucleoplasm	544	2.40 × 10−85
	GO:0000785~chromatin	120	1.89 × 10−8
MF	GO:0005515~protein binding	1116	7.48 × 10−69
	GO:0003723~RNA binding	248	7.34 × 10−46
	GO:0042802~identical protein binding	169	3.16 × 10−7
	GO:0061629~RNA polymerase II-specific DNA-binding transcription factor binding	31	4.44 × 10−6
	GO:0140297~DNA-binding transcription factor binding	30	3.22 × 10−6

,

Table A4. Top GO and pathway analysis results for the brown module, obtained from the Gene Ontology database (http://www.geneontology.org, accessed on 28 June 2024).

Category	Term	Count	p-Value
BP	GO:0045944~positive regulation of transcription by RNA polymerase II	115	4.74 × 10−9
	GO:0007165~signal transduction	114	4.41 × 10−9
	GO:0000122~negative regulation of transcription by RNA polymerase II	81	2.91 × 10−5
	GO:0045893~positive regulation of DNA-templated transcription	68	2.60 × 10−6
	GO:0008380~RNA splicing	38	5.65 × 10−9
CC	GO:0005886~plasma membrane	389	5.98 × 10−14
	GO:0005829~cytosol	363	2.61 × 10−8
	GO:0005737~cytoplasm	345	6.44 × 10−5
	GO:0005654~nucleoplasm	245	5.81 × 10−4
	GO:0070062~extracellular exosome	188	1.76 × 10−12
MF	GO:0005515~protein binding	790	8.86 × 10−14
	GO:0042802~identical protein binding	133	2.09 × 10−5
	GO:0004672~protein kinase activity	27	3.93 × 10−3
	GO:0005178~integrin binding	22	1.55 × 10−4
	GO:0004713~protein tyrosine kinase activity	17	1.26 × 10−4

and

Table A5. Top GO and pathway analysis results for the red module, obtained from the Gene Ontology database (http://www.geneontology.org, accessed on 28 June 2024).

Category	Term	Count	p-Value
BP	GO:0045944~positive regulation of transcription by RNA polymerase II	118	7.69 × 10−11
	GO:0007165~signal transduction	103	7.45 × 10−7
	GO:0030154~cell differentiation	58	3.70 × 10−5
	GO:0045893~positive regulation of DNA-templated transcription	56	0.0013
	GO:0008284~positive regulation of cell population proliferation	50	1.10 × 10−5
CC	GO:0005737~cytoplasm	368	1.72 × 10−9
	GO:0005829~cytosol	351	1.53 × 10−7
	GO:0005654~nucleoplasm	240	5.72 × 10−4
	GO:0048471~perinuclear region of cytoplasm	54	6.57 × 10−3
	GO:0032991~protein-containing complex	49	8.41 × 10−3
MF	GO:0005515~protein binding	747	4.49 × 10−8
	GO:0042802~identical protein binding	129	4.40 × 10−5
	GO:0106310~protein serine kinase activity	34	1.76 × 10−3
	GO:0045296~cadherin binding	33	8.03 × 10−4
	GO:0043565~sequence-specific DNA binding	32	1.00 × 10−4

, and the top terms are presented in Figure 5. Blue, red, and brown modules are mainly involved in transcription regulation. These modules are active in the cytosol, plasma membrane, and cytoplasm and are all associated with protein binding. KEGG pathway analysis indicated that these modules participate in pathways related to ccRCC through inflammatory pathways.

3.4. Protein–Protein Interaction (PPI) Networks and Hub Genes Analysis

PPIs of the identified genes were obtained from the STRING database. STRING was used to construct PPI networks for each preserved module with a high-confidence score (>0.7). The top 20 hub genes in each module were determined using degree centrality, closeness centrality, betweenness centrality, and MNC topology metrics. Figure 6 shows these hub genes, with red having the highest interaction scores. The high score of interaction reveals that these hub genes have a very significant association with ccRCC. Moreover, the genes overlapped between these metrics were defined as final hub genes (Figure 7) and used for further analysis.

3.5. Survival Analysis

All the hub genes were screened for survival analysis using GEPIA, but only six genes yielded statistically significant results. Their Kaplan–Meier survival plots are presented in Figure 8. MAPK8, BCL2, SMAD3, MAPK1, CREBBP, and EP300 are associated with much better survival rates since their analysis showed lower chances of death and recurrence of the disease. On the other hand, SOX2 is associated with poor OS, and its role in disease recurrence is less certain. Collectively, these findings indicate that these genes have the potential to be biomarkers in ccRCC.

3.6. Drug Repurposing Analysis

GEO2R was used to classify the top hub genes as upregulated or downregulated genes based on the positive and negative values of fold-change (FC). The DRE webserver analyzed these gene profiles to identify promising drug candidates. Lower false discovery rates (FDR) and more negative Tau values provide more reliable results and stronger effects on the target genes [24]. The top drug candidates identified for upregulated hub genes are ziprasidone, etazolate, trequinsin, dicycloverine, and decitabine. On the other hand, the leading candidates of drugs in downregulated hub genes include fentiazac, asenapine, KU-60019, pazopanib, and niacin. Detailed information about these drugs is included in

Table 2. Top candidates of upregulated and downregulated hub genes.

	Genes	Drug	Mechanism	Tau	FDR
Upregulated	ATM, CDC4, GSK3B, ITGA2B, MAPK8, MED1, PPARG, RPL5, CD44, HNRNPC, BCL2, CALML3, EGFR, GRB2, HDAC1, HSP90AA1, INS, PRKACG, SOX2, SRC, SMAD3, MAPK1, and TGFB1	Ziprasidone	Dopamine receptor antagonist, Serotonin receptor antagonist	−99.0039	0.00884
		Etazolate	Phosphodiesterase inhibitor	−98.9247	0.00207
		Trequinsin	Phosphodiesterase inhibitor	−97.5632	0.00557
		Dicycloverine	Acetylcholine receptor antagonist	−97.26467	0.00936
		Decitabine	DNA methyltransferase inhibitor	−96.3448	0.00267
Downregulated	PTK2, BDNF, CREBBP, DLG4, EP300, H3C12, and ESR1	Fentiazac	Cyclooxygenase inhibitor	−99.6497	0.001254
		Asenapine	Dopamine receptor antagonist, serotonin receptor antagonist	−99.2415	0.007037
		Ku-60019	ATM kinase inhibitor	−99.1959	0.007115
		Fr-180204	MAP kinase inhibitor	−98.2394	0.000190
		Niacin	NAD precursor, vitamin B	−98.2211	0.002544

.

4. Discussion

4.1. Gene Expression Modules and Pathway Analysis across the Datasets

ccRCC is the most common type of renal cancer. It progresses rapidly and often metastasizes to other parts of the body. This makes the prognosis difficult to determine [31,32]. Most cases of ccRCC are generally diagnosed in their later stages, so standard treatments are less effective; this is why new diagnostic methods are needed and being developed [32,33]. To address these problems, the present study carried out WGCNA analysis on samples of ccRCC from S1 to S4. This research used the S1 dataset to study the molecular and genetic characteristics of ccRCC and identify the key genes and pathways present from S1 to S4. Following this, the identified targets were analyzed to know if they may serve as potential biomarkers or targets for ccRCC. The KEGG pathway analysis of highly preserved gene modules (Table A1) suggests that these modules are linked to inflammatory pathways that may drive tumor development (Figure 9).

The analysis of S1 and S4 datasets revealed that specific pathways are present at certain stages of ccRCC (Figure 10). In S1, the HIF-1, TGF-β, and PI3K-Akt pathways suggest early tumor adaptation. HIF-1, mediated by the oxygen-sensitive HIF-1α and HIF-1β, helps cancer cells adapt to low oxygen. It is also linked with inflammation since it can regulate inflammatory components [34,35], thus promoting tumor development in ccRCC [36,37]. Enhanced TGF-β signaling can foster a proinflammatory microenvironment [14,38,39,40], reducing the effectiveness of anti-cancer therapies [41]. The PI3KT-AkT pathway has been known to stimulate inflammation processes and interact with other inflammatory pathways [42], which also accelerates tumor progression [43,44,45,46] and chemotherapy resistance [47]. The Ras and MAPK pathways were found to be present in S2 tumors, which may indicate heightened aggression and immune response in ccRCC [35,36,37,38]. Dysregulated RAS promotes inflammatory cytokine secretion [48,49], influencing ccRCC. The dysregulated MAPK pathway in ccRCC [43,50] leads to the release of cytokines such as IL-6, IL-8, and TNF-α [44,45,50].

Hippo signaling was detected in S3. Hippo signaling shows an inverse correlation with the proliferation and invasion of tumors [46,51,52]. Abnormal Hippo can increase cytokine production and activate other cancer pathways like JAK-STAT [47,51,53]. In S4, the NF-κB pathway and viral carcinogenesis are present. The NF-κB activated by IL-1 and TNFα connects inflammation with tumorigenesis [54,55,56]. NF-κB drives HIF-1α [57,58]. Hippo, MAPK, PI3K-AKT, and NF-κB pathways interact with each other, supporting inflammatory signals, cancer cell growth, and therapy resistance [44,47,59,60]. Additionally, viral infections like HPV and CMV associated with inflammation may significantly affect cancer treatment and growth [42,61,62,63]. The findings suggest that these pathways are involved in ccRCC S1–S4, indicating their role as drivers of ccRCC development.

The S1 dataset as a reference set helps track genes and pathways that may be present from S1 to S4. This may facilitate the identification of potential targets associated with ccRCC. The blue, brown, and red co-expression modules were preserved across datasets. Functional annotation linked these modules to inflammation-driven cancer pathways (Table A3, Table A4 and Table A5. Studying pathways present at different/certain stages of ccRCC may provide additional knowledge regarding ccRCC, which can help identify potential targets for specific stages of the disease.

4.2. Module Key Hub Genes and Their Protein Functions

4.2.1. Involvement of Hub Genes in Inflammation-Related Pathways Associated with ccRCC

The identified hub genes were found to be connected in these pathways. KEGG pathway analysis confirmed the involvement of hub genes in RCC, since RCC was obtained in the analysis (Table A1). Upregulated genes EGFR, CD44, and SOX2 in the red module are linked with the HIF-1 signaling pathway. EGFR, overexpressed in ccRCC and activated by IL-6 and TNF-α, is less effective as a target due to the lack of mutations or gene amplification [64,65]. Its membranous expression in RCC tumors [66], compared to its cytoplasmic location in normal kidney tissue, is associated with poorer prognosis and advanced stages [67]. HIF-1 also drives CD44 [68,69] upregulation in ccRCC [70,71,72] and correlates with poor survival and higher tumor grade [73,74]. SOX2, upregulated by HIF-1, enhances pro-inflammatory signaling and cancer stem cell-like properties, contributing to therapy resistance [75,76,77] with poorer overall survival (Figure 8E). In ccRCC, the NF-κB pathway mediates cytokines and chemokines [55,78,79,80,81,82]. Loss of VHL amplifies NF-κB activation, thus upregulating BCL2 [14,79,80]. BCL2 upregulated in the brown and red modules showed that it is associated with better survival and reduced risk of death and recurrence in kidney cancer (Figure 8A), establishing its role as potential marker. Venetoclax, a BCL2 inhibitor, was found to be effective in cancers such as leukemia, but continuous use of this agent can lead to drug resistance [83,84]. ATM enhances NF-κB, and its expression level could serve as a biomarker for ccRCC [65,85]. ATM is also involved in resistance to radio- and chemo-therapeutic treatments [65].

Upregulated SRC was found in the brown and red modules. Studies suggest that targeting SRC may disrupt pathways in ccRCC [66,86]. Chemokine can also activate the RAS signaling pathway. Cytokine-based therapies and immune checkpoint inhibitors are being developed to target chemokine signaling to address cancer cells immunity and address their resistance [87,88,89,90,91]. The Hippo pathway, linked with RCC [52], involves SMAD3 [53,92]. SMAD3 activates inflammation and interacts with Hippo effector YAP/TAZ [14,93]. SMAD3 also showed good survival analysis with reduced risk of death and recurrence (Figure 8D), thus targeting SMAD3 in Hippo may effectively inhibit ccRCC [47,59].

The MAPK pathway, often activated in cancer [94], upregulates MAPK1 and MAPK8 [95,96] found in the blue and red modules. The survival analysis also indicated that high levels of MAPK1 and MAPK8 are associated with better survival outcomes (Figure 8E,F). Research shows that downregulated MYPT1 in ccRCC inhibits MAPK8 activation, reducing metastasis [97]. The PI3K-Akt pathway regulates genes such as the upregulated GSK3B found in the red and brown modules. This pathway is frequently dysregulated in RCC since it interacts with the VHL/HIF [98,99,100], and a study found that targeting PI3K/Akt, along with its component, with inhibitors like LY294002 and wortmannin reduces inhibit RCC [101]. Overall, these pathways and genes are found to be associated with RCC [17,100].

4.2.2. Protein Chaperones and Epigenetic Regulators in Inflammation and Tumor Development in ccRCC

Molecular chaperones are known to be involved in protein folding but are now recognized in inflammation [102,103] and, under certain conditions, may contribute to tissue pathology [104,105]. Additionally, epigenetic regulators are also linked to inflammation [106,107,108]. Upregulated protein chaperones like HSP90AA1 were observed in the blue module, while epigenetic regulators such as HDAC1 and CREBBP were observed in the blue, brown, and red modules. HSP90AA1 maintains the stability of proteins related to signal transduction [109,110,111]. Epigenetic alterations in ccRCC influence inflammation through histones [107,112,113]. HDACs such as HDAC1 modulate the production of cytokines [107,112,114,115,116]. CREBBP is associated with inflammation [117]. Studies suggest that dysregulation of CREBBP, along with its partner EP300, drives the development of cancer and chemoresistance of several malignancies and affects tumor immune responses. Thus, this makes CREBBP/EP300 promising targets in cancer studies, as supported with the survival analysis (Figure 8B,C) [118]. Despite their potential as a target or biomarker, further research is still needed to validate their roles in ccRCC.

4.2.3. Metabolic Reprogramming in ccRCC

The constructed gene network suggests that it may be associated with the metabolic reprogramming of ccRCC supported by the PI3K/Akt and HIF-1 pathways. The yellow module (Table A2) highly represented the role of metabolic pathways. Several upregulated genes like INS, PRKACG, and CALML3 found in the brown and red modules were involved in these pathways. Metabolic reprogramming has been associated with tumor development and progression in ccRCC. The Warburg effect involves conditions of normal oxygen, resulting in the generation of lactate, along with an acidic microenvironment that boosts the migratory capabilities of the cancer cells. Glutamine metabolism is elevated in ccRC cells. As in most cases of ccRCC, mutation of the VHL gene disrupts the degradation of HIFs, initiating metabolic changes [119,120].

NMR studies have identified metabolites like lactate and glutamate in RCC [121]. In ccRCC, glutamate influences the PI3K/AKT pathway, with glutamate derived from glutamine via glutaminase [121,122]. The downregulation of GLUD1, which converts glutamate to α-ketoglutarate, is linked to increased malignancy and poor prognosis by enhancing PI3K/Akt/mTOR. Thus, GLUD1 is a potential target in ccRCC [122,123,124,125]. Another study also identified increased levels of creatine, alanine, lactate, and pyruvate using 1H-NMR. Metabolites findings with transcriptomic data from ccRCC tumor stem cells showed changes with HIF-α. Pyruvate, betaine, and creatinine were also found in HIF-1 in ccRCC. Pyruvate is an intermediate in glycolysis and thereby reflects increased glycolytic activity by HIF-1α through the upregulation of genes related to glycolysis. Betaine intersects with HIF-1α involved in methionine and choline metabolism. Creatinine, known for renal function and dysfunction, is indirectly related to HIF-1α activity [126]. Moreover, lactate stabilizes HIF-1α, making it a potential indicator of the HIF-1α pathway in ccRCC [34,126].

4.3. Viral Infections Associated with ccRCC

Viral infections significantly contribute to global cancer rates [63,127,128,129]. One study showed that the modules are linked to viral infections, including HPV and CMV [130]. These viral infections in ccRCC are known to cause inflammation [61,131,132]. CMV initiates inflammatory responses through CD14 and TLR2 [62]. CMV infection can be more severe in ccRCC patients compared to those with healthy immune systems, and cancer treatment increases the risk of infections like CMV, which can cause serious health issues [133]. The link between HPV and renal cancer is not fully understood yet. Some studies find no significant association, while others suggest HPV may contribute to RCC development or progression [131].

4.4. Drug Repurposing Based on Gene Signatures

New chemotherapeutic drugs have advanced tumor treatment but their development process is expensive and time consuming. Also, not all developed drugs are effective in clinical trials. Thus, repurposing existing drugs may offer a quicker and more economical solution, as these drugs have already undergone safety testing and are used to treat other diseases [134,135]. Among the upregulated hub genes, ziprasidone (ZIP) is the top candidate repurposed drug for ccRCC (Table 2). ZIP is an antipsychotic medication [136]. The effects of ZIP on inflammatory pathways like NF-κB and HIF-1 are not well established but studies found that it may inhibit several cancers such as pancreatic cancer [137] and breast cancer [138]. For instance, ZIP inhibits GOT1 to disrupt metabolic processes and reduce pancreatic cancer cell growth and has shown anti-cancer properties in xenograft models in preclinical studies [137]. Through in vitro studies, it was found that ZIP selectively inhibits aromatase with less toxicity than existing inhibitors in breast cancer [138].

Antipsychotic drugs such as haloperidol, chlorpromazine, thioridazine, and penfluridol are now being studied for their potential anti-cancer properties [134]. Thioridazine has also been studied in the PI3K/AKT and NF-κB pathways [134,139,140,141]. Thus, the findings suggest that ZIP could be repurposed as a potential treatment for ccRCC since it was already tested in other cancer treatments and may inhibit pathways where other antipsychotic drugs were already being studied.

For downregulated hub genes, fentiazac, a nonsteroidal anti-inflammatory drug (NSAID), was the top candidate drug that may be repurposed for ccRCC (Table 2). The connection of inflammation and cancer has led to interest in NSAIDs for cancer treatment. NSAIDs like aspirin, ibuprofen, diclofenac, celecoxib, tepoxalin, naproxen, and indomethacin have demonstrated anti-cancer properties, and some are already in clinical trials [142,143,144,145,146,147,148]. Fentiazac has not yet been documented for cancer treatment, particularly ccRCC, but it may be repurposed to target ccRCC by inhibiting COX-1 and COX-2 [149], like other NSAIDs studied for cancer drug repurposing. As a COX-1 inhibitor, it may also inhibit pathways associated with ccRCC, such as MAPK and NF-κB, targeting genes like EGFR, MAPK1, and BCL2 [150,151]. These considerations suggest that fentiazac could be a potential candidate for ccRCC treatment.

Other drug candidates associated with upregulated hub genes are etazolate, trequinsin, dicycloverine, and decitabine. On the other hand, the candidates for downregulated hub genes are asenapine, KU-60019, Fr-180204, and niacin (Table 2). Neuroactive drugs such as etazolate (phosphodiesterase inhibitor) and asenapine (dopamine and serotonin receptor antagonist) may target ccRCC by the MAPK, PI3K/Akt, and HIF-1α pathways [152,153,154]. Trequinsin, a phosphodiesterase inhibitor, and dicycloverine (phosphodiesterase inhibitor) commonly used for gastrointestinal cramps [155] may reduce side effects. However, their relationship with ccRCC and other cancers has not been documented yet. Decitabine (DNA methyltransferase inhibitor) induces G2/M cell cycle arrest in ccRCC cells by inhibiting the p38/NF-κB pathway [156]. EX-527 (a SIRT inhibitor) can be repurposed to impact sirtuin pathways [157,158] that are associated with ccRCC. KU-60019 (an ATM kinase inhibitor) may improve cancer treatment by inhibiting DNA damage response and reducing AKT phosphorylation. Its ability to radiosensitize cancer cells supports its potential [159]. Fr-180204 (a MAP kinase inhibitor) can disrupt MAPK pathways, while cefotiam, an antibiotic [160], and niacin, a vitamin B precursor [161], do not directly target ccRCC, but they may have indirect roles by affecting inflammation and metabolic processes related to ccRCC.

These potential drug candidates may help manage symptoms and inhibit ccRCC through several pathways. However, the results are based solely on theoretical data and hypotheses, highlighting a significant limitation of this study. Further experimental research, including wet lab experimentation, is necessary to validate these findings and determine their practical applicability. Nonetheless, our study provides a starting point for further investigation into correlating hub genes in ccRCC.

5. Conclusions

In the study, three significant highly preserved modules across different ccRCC stages obtained from GSE53757, GSE22541, GSE66272, and GSE73731 were identified. With WGCNA, the top hub genes were identified for each module and were used to determine functional annotation, pathways, and potential drug candidates that can be repurposed. The analysis provides insights about ccRCC through signaling pathways associated with inflammation. The top drug candidates identified for repurposing are ziprasidone (dopamine and serotonin receptor antagonist) and fentiazac (cyclooxygen-ase inhibitor). Other candidates are phosphodiesterase inhibitor, acetylcholine receptor antagonist, DNA methyltransferase inhibitor, an ATM kinase inhibitor, a MAP kinase inhibitor, and NAD precursors. These findings may inhibit ccRCC through the identified inflammatory pathways. While our study is limited by its in silico approach, the theoretical data provided can be assist in developing potential drug candidates or markers for ccRCC. The hub genes identified may also offer an understanding regarding the mech-anism of ccRCC. Moreover, further analysis is recommended such as using RNA-seq data to analyze ccRCC and compare it with DNA microarray. Additionally, experiments and wet lab research are recommended to confirm and apply these findings.