1 Introduction

Since its emergence in 2019, coronavirus disease-2019 (COVID-19) caused by SARS-CoV-2 has wreaked havoc globally, inflicting profound tolls on public health, economies, and societies [1]. SARS-CoV-2 exhibits various mutants such as Alpha, Beta, Gamma, Kappa, Delta and Omicron [2, 3]. Clinical trials are underway to identify effective treatments for COVID-19. Notably, a randomized controlled trial incorporating lopinavir and ritonavir in severe COVID-19 cases yielded no substantial clinical symptom improvement or reduced mortality compared to standard supportive care [4]. Among the gravest concerns is the heightened susceptibility of cancer patients to COVID-19, resulting in exacerbated symptoms and elevated mortality risk [5]. This predicament is particularly worrisome for cervical squamous cell carcinoma and endocervical adenocarcinoma (CESC) patients, who constitute a significant cohort of cancer cases [6]. Cervical cancer, the fourth most prevalent cancer among women, claims the lives of over 300,000 women annually [7]. According to the statistic, nearly 11 women were dying from cervical cancer every day [8]. The majority of cervical cancer cases are associated with human papillomavirus (HPV) infection. HPV-negative cervical cancer had a worse survival [9]. As a highly vulnerable group, cancer patients are at increased risk for SARS-CoV-2 virus infections due to longer hospital stays and weaknesses in their immune systems [10,11,12]. COVID-19 infection causes treatment delays and interruption for cervical cancer patients. Given the vulnerability of cancer patients, particularly in the context of the ongoing pandemic, identifying effective therapeutic strategies for cervical cancer patients infected with COVID-19 is imperative.

Quercetin, a polyphenolic flavonoid prominently employed in traditional Chinese medicine, exhibits promising chemo-preventive properties [13, 14]. Its role in modulating signal transduction pathways and augmenting antioxidant defenses has garnered interest in disease prevention [15]. Additionally, quercetin has emerged as an adjuvant in anti-cancer regimens, enhancing treatment sensitivity and minimizing collateral damage to healthy cells [16, 17]. Furthermore, in conjunction with vitamin C, quercetin has demonstrated synergistic potential in augmenting antiviral therapies for COVID-19 [18]. Despite these insights, comprehensive investigations into quercetin’s targets and mechanisms for treating cervical cancer patients with COVID-19 are limited.

In the realm of medicinal research, web-based pharmacology approaches have emerged as robust tools for comprehending the action mechanisms of Chinese medicines [19, 20]. Leveraging network pharmacology, which characterizes drug actions and mechanisms at the molecular level, we aimed to elucidate quercetin’s potential as an adjunctive therapy for cervical cancer patients with COVID-19. The ensuing study delineates our network pharmacology-based analysis, as depicted in Fig. 1.

Fig. 1
figure 1

The graphical abstract illustrates the study’s methodology employing network pharmacology and computational bioinformatics to elucidate quercetin’s mechanism against CESC/COVID-19.

Full size image

2 Materials and methods

2.1 Screening for core genes associated with COVID-19 and potential differentially expressed genes of CESC

The National Center for Biotechnology Information (NCBI) and the GeneCards Database were employed to identify genes associated with COVID-19 [21]. Transcriptomic profiles of CESC and normal cervical samples were retrieved from The Cancer Genome Atlas (TCGA) on November 05, 2021 Supplementary transcriptomic profiles from the Genotype-Tissue Expression (GTEx) database was included for normal cervical samples [22]. Differential gene expression analysis for CESC was performed using the ‘DEseq2’ package in the R environment, considering genes with false discovery rate (FDR) < 0.05 and |log fold change (FC)| > 1 as differentially expressed. Visualization was facilitated using the ‘EnhancedVolcano’ and ‘ComplexHeatmap’ packages [23, 24]. Overlapping genes between COVID-19-related genes and CESC DEGs were identified.

2.2 Identification of quercetin-related genes and enrichment analysis

Potential quercetin targets were sourced from Comparative Toxicogenomics Database (CTD) [25], DrugBank [26], Swiss Target Prediction [27], TargetNet [28], Kyoto Encyclopedia of Genes and Genomes (KEGG) [29] database. Subsequently, gene ontology (GO) and KEGG enrichment analyses were performed using g:Profiler for enrichment and the ‘ggplot2’ and ‘GOplot’ packages for visualization [30, 31].

2.3 Construction of protein–protein interaction networks

The STRINGv11.5 [32] was employed to construct protein–protein interaction (PPI) networks for quercetin-related genes associated with anti-CESC/COVID-19 effects. Visual representation of results was generated using Cytoscape software (version 3.6.1) [33]. Hub genes were identified using the CytoHubba plugin based on the maximal clique centrality algorithm.

2.4 Clinical analysis for CESC/COVID-19-related genes

Clinical information of CESC patients was obtained from TCGA. Univariate Cox analysis was conducted to identify genes associated with overall survival (OS). Subsequently, multivariate Cox regression analysis were performed using the “glmnet” package to establish a risk score formula based on normalized gene expression values. Patients were categorized into high- or low-risk subgroups using the median risk score. Receiver operating characteristic (ROC) curves (including 3-year, 5-year, and 10-year survival) were plotted to evaluate predictive value using the ‘timeROC’ package.

2.5 Gene set enrichment analysis and immune cell infiltration analysis

Differentially expressed genes between high- and low-risk subgroups were identified using the ‘limma’ package Gene set enrichment analysis (GSEA) was conducted utilizing packages such as “org.Hs.eg.db”, “dplyr”, “enrichplot”, and “clusterProfiler”. Immune cell enrichment scores for 22 immune-related cells were calculated using the CIBERSORT algorithm. Immune cell infiltration content between high- and low-risk subgroups was visualized using the ‘ggpubr’ package. Additionally, the TIMER database [34] was utilized to assess the association between prognostic risk model genes and immune cell infiltration in CESC.

2.6 Prediction of immune checkpoint blockade treatment response

To gauge the risk score’s potential in predicting immune checkpoint blockade treatment response, clinical features and expression data from the IMvigor210 cohort were obtained (http://research-pub.gene.com /IMvigor210CoreBiologies) [35]. The risk score was utilized to predict OS and immune checkpoint blockade therapeutic responses [including progress disease (PD), stable disease (SD), partial remission (PR), and complete remission (CR)].

2.7 Molecular docking

Autodock 4.2 software [36] was employed for semi-flexible molecular docking. Crystal structures of COVID-19 main protease (6LU7) and spike glycoprotein (6VYB) were obtained from the Protein Data Bank (PDB) [37]. The grid box was set to contain all receptor region with the original ligand coordinates. The binding energy was got by autogrid4 and autodock4 function. Molecular docking outcomes were observed using PyMol software [38].

3 Result

3.1 Differentiation analysis of CESC

A comprehensive cohort of 306 CESC samples and 3 normal cervical tissue samples from TCGA, along with 19 cervix samples from GTEx, formed the basis for our investigation. Applying stringent criteria, we identified 4986 differentially expressed genes (DEGs) between CESC patients and controls, with 3085 genes exhibiting up-regulation and 1901 genes down-regulation (Fig. 2).

Fig. 2
figure 2

Differential Expression of CESC Genes. A Heat map depicting 4986 DEGs in CESC, including 3085 up-regulated and 1901 down-regulated genes. B Volcano plot showcasing DEGs in CESC, with red and green dots denoting up-regulated and down-regulated genes, respectively

Full size image

3.2 Identification of quercetin targets for CESC/COVID-19

Incorporating data from NCBI and the Genecard database, we retrieved 492 COVID-19-related genes. A total of 90 CESC/COVID-19-related genes were identified, comprising 62 up-regulated and 28 down-regulated genes in CESC. Further analysis revealed 4237 quercetin-related genes across five databases, including CTD, DrugBank, Swiss Target Prediction, TargetNet and KEGG. Our integration efforts highlighted 45 quercetin targets with implications for CESC/COVID-19, illustrated in Fig. 3A. The resultant PPI network, governed by these 45 intersection genes, was visualized via STRING and Cytoscape (Fig. 3B).

Fig. 3
figure 3

Functional Enrichment Analysis of Quercetin Targets. A Venn diagram depicting the intersection of quercetin, CESC, and COVID-19 genes. B Protein–protein interaction (PPI) network was visualized using Cytoscape, indicating the gene associations. C GO analysis showing the top 10 biological processes (BP), cellular components (CC), and molecular functions (MF) of intersecting genes. D KEGG pathway analysis of quercetin targets for CESC/COVID-19.

Full size image

3.3 GO enrichment analysis and KEGG pathway analysis

To illuminate the functional underpinnings of quercetin’s action in the context of CESC/COVID-19, we conducted Gene Ontology (GO) and KEGG pathway analyses on the 45 intersect genes. GO analyses revealed that these genes were predominantly involved in biological regulation, response to stimulus, and cellular response to stimulus. Cellular component enrichment centered around cytoplasm, extracellular region, and extracellular space, while molecular function analysis underscored signaling receptor binding, molecular function regulation, and cytokine receptor binding (Fig. 3C). KEGG pathway analysis unveiled significant enrichment in coronavirus disease-COVID-19, necroptosis, cytokine-cytokine receptor interaction, viral protein interaction with cytokine and cytokine receptor, and an array of signaling pathways (Fig. 3D).

3.4 Independent prognostic signatures and risk analysis

Our exploration into overlapping genes’ impact on survival encompassed both univariate and multivariate Cox regression analyses. Eight genes emerged from univariate Cox regression, including TNF (tumor necrosis factor), IFNG (interferon gamma), IL1A (interleukin-1), F2 (coagulation factor II), NRP1 (neuropilin 1), TYK2 (tyrosine kinase 2), IDO1 (indoleamine 2,3-dioxygenase 1), and PLA2G7 (phospholipase A2 group VII) (P < 0.05, Fig. 4A). Subsequent multivariate Cox regression identified five independent prognostic factors-TNF, F2, NRP1, TYK2, and PLA2G7 (P < 0.05, Fig. 4B). The risk score was calculated as follows: risk score = (0.079934737 * TNF exp.) + (0.754296176 * F2 exp.) + (0.077712628 * NRP1 exp.) − (0.052768486 * TYK2 exp.) − (0.130543819 * PLA2G7 exp.). The derived risk score formula enabled the stratification of patients into high- and low-risk groups, with high-risk associated with poorer OS (Fig. 4C). ROC analysis showed robust predictive performance, with AUCs of 0.733, 0.738, and 0.614 at 3, 5, and 10 years, respectively (Fig. 4D). Survival scatter plots demonstrated the superior survival of low-risk patients (Fig. 4E, F). Notably, TNF, NRP1, TYK2, and PLA2G7 exhibited differential expression between high- and low-risk groups. The expression of TNF and NRP1 was lower in low-risk group, while the expression of TYK2 and PLA2G7 was higher in low-risk group (Fig. 5A). Additionally, NRP1 and PLA2G7 displayed heightened expression in grade 3/4 compared to grade 1/2 (Fig. 5A–C).

Fig. 4
figure 4

Construction and validation of prognostic signatures. A Forest plot presenting HR and P-values from univariate Cox regression analysis. B Multivariate Cox analysis revealing independent prognostic signatures. C Kaplan-Meier survival analysis indicated significantly shorter OS in the high-risk group (P < 0.001). D Time-dependent ROC curves assessing prognostic signature accuracy. E Risk score distribution among CESC patients. F Scatterplot depicting the correlation between risk scores and survival time/status

Full size image
Fig. 5
figure 5

Differential expression of independent prognostic signatures. A Correlation between gene levels and risk scores. Expression levels of PLA2G7 (B), NRP1 (C), and their association with clinical characteristics. *P < 0.05; **P < 0.01; ***P < 0.001

Full size image

3.5 Functional enrichment analyses

The high-risk group was linked to responses to immobilization stress, lipid catabolic processes, substrate adhesion-dependent cell spreading, vasoconstriction, and hormone activity in the c5.go dataset (Fig. 6A, B). Conversely, the low-risk group exhibited significant enrichment of immune-related biological processes such as adaptive immune response, activation of the immune response, and myeloid leukocyte immunity (Fig. 6C, D). In the c2.cp.kegg dataset, the high-risk group displayed associations with ribosome, PPAR signaling pathway, tyrosine metabolism, nitrogen metabolism, and metabolism of xenobiotics by cytochrome P450 (Fig. 6E, F). On the other hand, the low-risk group was enriched in cytokine receptor interaction, chemokine signaling pathway, cell adhesion molecules, T cell receptor signaling pathway, and natural killer cell-mediated cytotoxicity (Fig. 6G, H).

Fig. 6
figure 6

GO and KEGG Pathway Enrichment Analysis. Enrichment plot illustrating the running sum of enrichment scores (left curve), gene position within pathways (middle part), and metric distribution (bottom part) in high and low-risk groups. The upSet plot displays the metric distribution of core enrichment genes. A, B GO analysis in the high-risk group. C, D GO analysis in a low-risk group. E, F KEGG pathway enrichment in the high-risk group. G, H KEGG pathway enrichment in a low-risk group

Full size image

3.6 Tumor immune microenvironment and immunotherapeutic response

Application of CIBERSORT unveiled variations in the tumor immune microenvironment between high- and low-risk groups in CESC. The high-risk group exhibited lower immune cell infiltration levels, particularly CD8+ T cells, M1 macrophages, M2 macrophages, regulatory T cells (Tregs), resting mast cells, activated memory CD4+ T cells, and dendritic cells. Conversely, infiltration of M0 macrophages, resting memory CD4+ T cells, active NK cells, and dendritic cells was significantly elevated in the high-risk group (Fig. 7A). Employing the IMvigor210 immunotherapy cohort, we found that low-risk patients enjoyed notable survival benefits from anti-PD-L1 immunotherapy (P = 0.005) (Fig. 7B). The risk model demonstrated fair predictive accuracy for immunotherapy responsiveness with AUC values of 0.644, 0.586, and 0.554 at 1, 3, and 5 years, respectively (Fig. 7C). Further analysis established that low-risk patients were more responsive to immunotherapy (P = 0.0027) (Fig. 7D).

Fig. 7
figure 7

Immune Microenvironment and Therapeutic Response. A Comparison of immune cell infiltration between high-risk and low-risk groups. B Kaplan-Meier survival curves for IMvigor210 cohort based on risk scores. C ROC curves assessing predictive accuracy of the risk score in the IMvigor210 cohort. D Comparison of risk scores in different anti-PD-L1 treatment response groups. ****P < 0.0001, **P < 0.01, *P < 0.05

Full size image

3.7 Association of genes with tumor immune infiltrates

Exploring the correlation between the expression of TNF, F2, NRP1, TYK2, and PLA2G7 and immune cell infiltrates revealed positive associations between higher expression levels and increased immune cell infiltration in tumors, including B cells, CD8+ T cells, CD4+ T cells, macrophages, neutrophils, and dendritic cells (Fig. 8A–E). Moreover, copy number variations (CNVs) of these genes exhibited notable impacts on immune cell infiltration levels. Specifically, deep deletion, arm-level deletion, arm-level gain, and high amplification of these genes significantly influenced immune cell infiltration in CESC (Fig. 9A–E).

Fig. 8
figure 8

Correlation of Gene Expression with ESTIMATE Score. Correlation between the expression of each gene (A: TNF; B, F2; C: NRP1; C: TYK2; E: PLA2G7) and ESTIMATE Score in CESC.

Full size image
Fig. 9
figure 9

Effect of copy number variation on immune infiltration. Impact of copy number variation of each gene (A: TNF; B: F2; C: NRP1; D: TYK2; E: PLA2G7) on immune cell infiltration levels

Full size image

3.8 Docking results

Molecular docking assays revealed robust binding interactions between quercetin and the crystal structures of COVID-19 main protease and spike glycoprotein, with favorable docking energies and hydrogen bond formation (− 5.88 kcal mol−1 for main protease and − 3.42 kcal mol−1 for spike glycoprotein). The quercetin formed hydrogen bonds with residues GLU-166 and THR-190 of the main protease and residue LEU-977 of spike glycoprotein (Table 1). Moreover, the docking analysis indicated favorable binding activities between quercetin and the five independent prognostic signatures with docking energy less than − 5 kcal mol−1 and two or more hydrogen bonds formed (Fig. 10; Table 2).

Table 1 Molecular docking results of quercetin with COVID-19 proteins
Full size table
Fig. 10
figure 10

Molecular docking analysis. Diagram of quercetin binding to the spike glycoprotein of COVID-19 (PDB ID: 6VYB), assessed through molecular docking analysis

Full size image
Table 2 Docking scores and hydrogen bonds of five independent prognostic signatures with quercetin
Full size table

4 Discussion

Evidence has shown potential therapeutic strategies for colon adenocarcinoma (COAD) and COVID-19 by quercetin. A total of 105 gene were potential targets of quercetin for anti-COAD/COVID-19, which were correlated with apoptosis, immune, oxidative stress. IL-17, HIF-1 and TNF were involved in quercetin-mediated treatment in COAD/COVID-19 patients [39]. Our investigation corroborates previous studies showcasing quercetin’s synergistic anti-CESC effects when combined with cisplatin [40, 41]. Employing network pharmacology, we pinpointed 45 target genes through which quercetin exerts its action against CESC and COVID-19. Enrichment analysis illuminated the therapeutic mechanisms, primarily centered around necroptosis, cytokine-cytokine receptor interaction, and viral protein interaction with cytokine and cytokine receptor pathways. Importantly, we identified that five independent prognostic signatures, including TNF, F2, NRP1, TYK2, and PLA2G7, via multivariate Cox regression could be biomarkers for prognosis.

TNF is a cytokine with potent pro-inflammatory effects [42]. Previous studies have shown that TNF had antitumor activity [43], playing an important role in tumor proliferation, migration, and invasion [44]. In CESC, it could induce apoptosis by activating the mitochondrial caspase-9 death signaling pathway. Moreover, recent study found that serum TNF level in patients with CESC were usually significantly elevated than this in non-CESC patients [45]. After surgical treatment, serum TNF level will gradually return to normal in CESC patients. Consistent with previous researches, TNF was considered as a risk factor for prognosis of CESC in our study. The expression level of TNF was significantly higher in high-risk group and tumor tissue than that in low-risk group and normal tissue.

Besides, current studies have proved that TNF-308 gene GA and AA polymorphisms were associated with the risk of cervical cancer [46, 47]. Moreover, TNF-308 AA and IL-10-592 CA/AA polymorphisms are linked to an increased risk of cervical cancer [48]. F2 was considered as an important determinant of thrombin generation [49]. A number of studies discovered that complement activation following SARS-CoV-2 infections generates thrombin and produces thrombosis [50, 51]. NRP1 is a semaphorin III receptor that is the foundation of neurosynapses [52]. NRP1 has been reported to induce tumor-associated macrophage activation and exert pro-tumor effects in cervical cancer under hypoxia [53]. Previous study have shown that NRP1 was associated with cervical cancer progression and poor survival, suggesting that NRP1 could be an independent prognostic factor in cervical cancer [54]. High expression of NRP1 in cervical cancer patients was correlated with shorter OS [55]. Moreover, preoperative chemoradiation therapy reduced Treg and Nrp1 + Treg levesl in lymph nodes of cervical cancer patients [56]. Soluble NRP1 (sNRP) in circulating and NRP1 proteins were associated with cervical cancer stages. In addition, sNRP exhibited a possible diagnostic biomarker for cervical cancer [57]. Transportin-1 (TNPO1)-induced nuclear import of FUBP1 (Far upstream element binding protein 1) led to tumor immune evasion via upregulation of NRP1 in cervical cancer [58]. And high expression level of NRP1 was associated with poor 5-year survival rates. Similarly, NRP1 was confirmed as an oncogene in this study.

In COVID-19 patients, a significant decrease in TYK2 level was observed in male patients compared to male controls [59]. The down-regulation of TYK2 was proposed as a molecular mechanism causing SARS-CoV-2 to be incapable of inducing a competent interferon response. Castineira et al. [60] demonstrated that there is an association between life-threatening diseases in COVID-19 and high level of TYK2. PLA2G7 is a calcium-independent lipoprotein-binding phospholipase, involved in cell signaling and metabolism [61]. On the one hand, some studies have shown that high level of PLA2G7 was positively correlated with aggressiveness in cancer [62,63,64]. PLA2G7 promoted cell migration and invasion in prostate cancer [63]. Meanwhile, Morigny et al. demonstrated that secretion and expression of PLA2G7 are positively correlated with cancer cachexia [65]. Depletion of PLA2G7 reduced intestinal polyposis and tumorigenesis in APC (Min/+) mice [66]. PLAG7 exhibited protective function in breast cancer via negative regulation of the Wnt signaling pathway [67]. On the other hand, Liao et al. discovered that high PLA2G7 protein level was associated with significantly longer OS than low protein level of PLA2G7 in ovarian cancer patients. The protective character of PLA2G7 was speculated to be mediated by negatively regulating the Wnt/β-catenin pathway [68]. Similarly, our analysis showed that PLA2G7 was associated with a better prognosis in CESC.

Quercetin has been reported to influence SARS-CoV-2 infection and COVID-19-associated cancer progression via suppression of HIF-1a and mTOR [69]. Quercetin influenced several signaling pathways, including TNF, TRAIL and FASL, and induced cell apoptosis in cervical cancer [70]. There are several limitations in this study. For example, a comprehensive cohort of 306 CESC samples and 3 normal cervical tissues from TCGA was used. Only 3 normal cervical tissues were included in this study. In addition, this work lacks cell line experiments and animal studies. It is worthy to note that in vitro experiments and in vivo mouse models are necessary to validate the function of quercetin for the treatment of COVID-19 with CESC.

5 Conclusion

Based on the results above, we believe that adjuvant therapy with quercetin may contribute to the treatment of COVID-19 or COVID-19 with CESC. The findings from this study identified molecules involved in the link between CESC and COVID-19, and five key genes of quercetin for CESC/COVID-19 treatment, including PLA2G7, TNF, TYK2, F2, NRP1. This finding can help the treatment of patients with CESC/COVID-19.