1. Screening TOP2A from four GEO datasets in ovarian cancer
DEGs (P < 0.05, log2-fold change > 2) were identified from four GEO datasets (GSE36668, GSE18520, GSE14407, and GSE16709), and their volcano plots are shown (Additional file 2: Fig. S1a). 36 DEGs were screened from four databases, of which 14 were greatly upregulated and 22 were downregulated in ovarian cancer (Fig. 1a). The upregulated genes were CP, TOP2A, FOLR1, CD24, C1orf106(INAVA), KIAA0101(PCLAF), EPCAM, MAL, PRAME, EHF, FAM83D, PSAT1, C12orf56, and MPZL2. The downregulated genes were ABCA8, WNT2B, EFEMP1, TCEAL7, DCN, AOX1, MGARP, TCEAL2, RNASE4, NDN, RNF128, CELF2, GHR, CNTN1, NAP1L3, ZFPM2, BAMBI, MAF, DIRAS3, ANTXR2, TBX3, and ITM2A (Fig. 1b).
The biological function of these DEGs was further analysed. The KEGG analysis revealed that upregulated genes were especially enriched in ferroptosis, platinum drug resistance and metabolism (Fig. 1c, Additional file 3). The downregulated genes were particularly enriched in the TGF-beta signalling pathway, Wnt signalling pathway, Jak-STAT signalling pathway, mTOR signalling pathway, Hippo signalling pathway and PI3K-AKT signalling pathway (Additional file 2: Fig. S1b). Moreover, GO biological process (GO-BP), GO cellular component (GO-CC) and GO molecular function (GO-MF) analyses indicated that the upregulated genes were mainly enriched in water soluble vitamin metabolic process, apoptotic chromosome condensation, vitamin metabolic process, regulation of DNA duplex unwinding, DNA topoisomerase activity, cadherin binding involved in cell adhesion, peptidase activate or activity involved in apoptotic and drug binding (Fig. 1d-f). The GO analysis of the downregulated genes is shown in Additional file 2: Fig. S1c-e.
We mainly focused on the upregulated genes. We further investigated the association between the expression of upregulated genes and survival in ovarian cancer. As shown in Additional file 4: Fig. S2a, the expression of CP (P = 0.23), FOLR1 (P = 0.085), CD24 (P = 0.054), C1orf106 (P = 0.082), KIAA0101 (P = 0.12), CO-17A (P = 0.18), PRAME (P = 0.065) and PSAT1 (P = 0.41) had no significant correlations with the overall survival (OS) of patients with ovarian cancer. However, low EHF (P = 0.009) and MPZL2 (P = 0.024) expression were linked to poor OS in ovarian cancer (Additional file 4: Fig. S2b). High expression of MAL (P < 0.001), FAM83D (P < 0.001), C12orf56 (P = 0.017), and TOP2A (201291-s-at, P < 0.001; 201292-at, p = 0.0039) was associated with poor OS in ovarian cancer (Additional file 4: Fig. S2c-d). Additionally, high TOP2A (P = 0.0069) expression was related to poor progression-free survival (PFS) in ovarian cancer (Additional file 4: Fig. S2d). Importantly, the KEGG analysis results showed that TOP2A is associated with platinum drug resistance (Additional file 3). Therefore, TOP2A was selected for further study.
2. TOP2A is overexpressed in ovarian cancer
We examined TOP2A expression in normal tissues and ovarian cancer tissues. The labelled TOP2A is shown in the volcano maps of each GEO dataset (Additional file 2: Fig. S1a). Using the online GEPIA database, we found that the expression levels of TOP2A mRNA in ovarian cancer was higher than that in normal tissues (Fig. 2a). GTEx combined with TCGA or ICGC data analysis showed that TOP2A expression in ovarian cancer was also significantly higher than that in normal tissues (Fig. 2b-c). Moreover, TOP2A mRNA expression in different ovarian cancer cell lines is shown in Fig. 2d.
Then, we investigated the expression of TOP2A protein in ovarian cancer. As shown in Fig. 2e, the TOP2A protein was mainly located in the nucleoplasm and nucleoli. We found that total TOP2A protein expression was greatly increased in ovarian cancer tissues by using the UALCAN database (Fig. 2f). We further contrasted phospho-TOP2A expression in normal and tumour tissues. The phospho-TOP2A (S1213, S1106, S1393, S1247) protein expression was also greatly increased in ovarian cancer tissues (Fig. 2g, Additional file 5: Fig. S3a). The immunohistochemistry (IHC) results further demonstrated that TOP2A protein expression was enhanced in ovarian cancer tissues (Additional file 5: Fig. S3b). Overall, these results illustrated that the expression of TOP2A mRNA and protein were both enhanced in ovarian cancer tissues.
3. Construction of the upstream lncRNA/circRNA-miRNA regulatory network
Numerous studies have demonstrated that long noncoding RNAs (lncRNAs) and circular RNAs (circRNAs) can regulate tumorigenesis by sponging miRNAs[33, 34]. Consequently, we proposed that the lncRNA/circRNA-miRNA network might regulate TOP2A expression in ovarian cancer. First, we selected the miRWalk, miRmap, miRabel, TargetScan and ENCORI databases to identify the potential target miRNAs of TOP2A mRNA, and only hsa-miR-330-3p appeared in these five programs (Fig. 3a). Therefore, the candidate hsa-miR-330-3p was included for subsequent analyses. Figure 3b shows the binding site between TOP2A and hsa-miR-330-3p.
Based on the data in the ENCORI and LncBase databases, there were 11 candidate lncRNAs that both appeared in two databases, including AC016747.1, FGD5-AS1, AC005076.1, KCNQ1OT1, NEAT1, MALAT1, OIP5-AS1, GABPB1-IT1, GABPB1-AS1, AC087284.1 and XIST (Fig. 3c, e). In the same way, six candidate circRNAs appeared in two databases (ENCORI and circBank), including hsa-circ-0036565, hsa-circ-0036566, hsa-circ-0075953, hsa-circ-0001731, hsa-circ-0089761 and hsa-circ-0001800 (Fig. 3d-e). Cytoscape software was utilized to make a lncRNA/circRNA-miRNA-TOP2A regulatory network (Fig. 3f).
Finally, using the lnCAR database, we revealed that lncRNA KCNQ1OT1 (P = 0.0451), NEAT1 (P = 0.0267) and AC087284.1 (P = 0.0102) expression was increased in ovarian cancer tissues (Fig. 3g-i). A positive association was found between KCNQ1OT1 expression levels and ovarian cancer grades (P = 0.0491) and stages (P = 0.0109) (Additional file 6: Fig. S4a-b). Interestingly, KCNQ1OT1 expression in the platinum drug-resistant group (P = 0.0150) was higher than that in the platinum drug-sensitive group (Additional file 6: Fig. S4c). AC087284.1 (P = 0.0066) expression was positively related to ovarian cancer grade (Additional file 6: Fig. S4d). Using the UALCAN database, we found that KCNQ1OT1, NEAT1, GABPB1-AS1 and AC087284.1 expression were positively related to ovarian cancer grade (Additional file 6: Fig. S4e-h). Furthermore, using the lnCAR database, we also identified that high expression of KCNQ1OT1 was associated with poor OS (P = 0.0112), PFS (P = 0.0201) and relapse-free survival (RFS) (P = 0.0143), while high NEAT1 expression was only related to RFS (P = 0.0039) in ovarian cancer (Additional file 6: Fig. S4i-l). These results implied that the lncRNA/circRNA-miRNA-TOP2A regulatory network may dysregulate the expression of TOP2A and mediate the tumorigenesis of ovarian cancer.
4. N6-methyladenosine (m6A) analysis of TOP2A in ovarian cancer
Strong evidence has demonstrated that dysregulation of RNA epigenetics plays a role in disease pathogenesis, including cancer. The effect of RNA modification depends on writers, readers and erasers. The types of RNA modification include m6A, RNA5-methylcytosine (m5C), and N1-methyladenosine (m1A) [35]. Hence, we analysed RNA modification of TOP2A in ovarian cancer using the SangerBox database. The results showed that the writers (TRMT6, TRMT61A), erasers (ALKBH1, ALKBH3), and readers (YTHDC1, YTHDF1) of m1A; writers (NSUN3, DNMT1) and readers (ALYREF) of m5C; and writers (ZC3H13, METTL3), erasers (ALKBH5), and readers (HNRNPA2B1, YTHDF1) of m6A were positively correlated with TOP2A expression (Fig. 4a).
M6A modification is the most abundant RNA modification in eukaryotic cells [36, 37]. Increasing evidence shows that m6A modification of RNA is connected with the occurrence, development or prognosis of ovarian cancer [38–40]. Thus, we were concerned with the m6A modification of TOP2A in ovarian cancer. We studied the association between TOP2A and m6A-related genes in 303 ovarian cancer samples. Most m6A-related genes were strongly positively correlated with TOP2A expression (Fig. 4b). The m6A modification sites of TOP2A were predicted by using the SRAMP database. As shown in Fig. 4c, most of the m6A modification sites were located at the 3000–5000 site, and there were very high confidence sites at the 3000–3500 site of TOP2A. Then, we used the ENCORI, RMBase, and RBPsuite databases to predict proteins bound to TOP2A mRNA, and 65 proteins appeared in all three databases. The FMR1, HNRNPC, IGF2BP1 and IGF2BP3 proteins which were involved in m6A modification were selected (Fig. 4d-e). Furthermore, we analysed the possibility of binding of these four proteins to TOP2A mRNA using the RPISeq database. The results showed that all interaction probabilities > 0.5, indicating that these four proteins can potentially bind with TOP2A mRNA (Fig. 4f). RNA modification often relies on consensus motifs to form secondary structures that bind to RNA-binding proteins such as writers/readers/erasers. The consensus motifs of these four proteins were showed by using the RBPsuite database (Additional file 7: Fig. S5a). Then, we used the RBPsuite database to explore the binding sites of the four proteins to TOP2A. When the RNA binding protein (RBP) has a verified motif file, for the segments with a score greater than 0.5, RBPsuite marks the possible motifs in red according to the position on the segment (Additional file 7: Fig. S5b, Additional file 8). The results showed that IGF2BP1 most likely binds to the AACCAUCCACU, AAACAUGACUU and AAGGACAAGUU motifs of TOP2A at the TOP2A_50, TOP2A_13 and TOP2A_10 positions. FMR1 most likely binds to the UGACAAG motif of TOP2A at the TOP2A_23 position. Although RBPsuite does not mark the possible motifs of HNRNPC and IGF2BP3, they have possible motifs with a score greater than 0.5. Therefore, they also have the possibility to bind with TOP2A.
Collectively, these results demonstrated that TOP2A expression was positively associated with most RNA modification-related genes, especially m6A-related genes in ovarian cancer. Particularly, the FMR1, HNRNPC, IGF2BP1 and IGF2BP3 proteins may bind with TOP2A to mediate its m6A modification.
5. Alterations of the TOP2A gene
Several malignancies grow and progress as a result of genetic alterations, such as gene mutations, deletions or amplifications[41]. Information on TOP2A genetic alterations in ovarian cancer samples from TCGA datasets was examined through the cBioPortal. We revealed that the “mutation” type dominated genetic alterations in TOP2A, with the “deep deletion” type coming in second (Fig. 5a). Additional file 9 shows the types, sites and case numbers of the TOP2A genetic alteration. We discovered that missense mutations were most common in ovarian cancer, and the most common mutation site was the T215P site in the HATPase-c domain (Fig. 5b). The T215P site is shown in the 3D structure of the TOP2A protein (Fig. 5c). Moreover, the relationship between the clinical prognosis of ovarian cancer patients and TOP2A genetic alterations was further investigated. Ovarian cancer patients with unaltered TOP2A presented a better prognosis in terms of overall (P = 0.0115), progression-free (P = 2.432e-3) and disease-specific (P = 2.842e-3) survival (Additional file 10: Fig. S6a-c). Therefore, we speculated that alterations of the TOP2A gene were associated with the progression of ovarian cancer.
Next, the relationship between TOP2A expression and somatic mutations of TCGA dataset was explored. The TOP2A high group (n = 154) displayed high somatic mutation rates in the TP53 (98%), TTN (43%), FLG2 (28%), and MUC16 (25%) genes, and the TOP2A low group (n = 154) displayed high rates of mutations in the TP53 (91%), TTN (40%), MUC16 (28%), and AHNAK (21%) genes (Fig. 5d-e). Obviously, somatic mutation rates of TP53 in the TOP2A high group were higher than those in the TOP2A low group. We also found that most of the genomic alterations and higher phosphoprotein expression levels were enriched in the altered TOP2A group (Additional file 10: Fig. S6d-e). The high expression of mRNAs was mainly enriched in the unaltered TOP2A group (Additional file 10: Fig. S6f).
Overall, these results showed that alterations in the TOP2A gene were associated with the progression of ovarian cancer patients. Depending on the expression level of TOP2A, there are different somatic mutations in ovarian cancer tissues. However, whether the alterations of the TOP2A gene induce the occurrence of ovarian cancer and whether the high expression of TOP2A regulates alterations in other genomes and proteomics need further investigation.
6. TOP2A-related gene enrichment analysis in ovarian cancer.
Based on the STRING database, 10 TOP2A-binding proteins were identified, including NDC80, KIF4A, NCAPG, UBE2C, TPX2, BUB1, DLGAP5, KIF20A, CDK1 and PBK, as displayed in the PPI network (Fig. 6a). We also found that these genes were positively correlated with TOP2A, including NDC80 (R = 0.52), KIF4A (R = 0.6), NCAPG (R = 0.61), UBE2C (R = 0.76), TPX2 (R = 0.66), BUB1 (R = 0.6), DLGAP5 (R = 0.6), KIF20A (R = 0.57), CDK1 (R = 0.67) and PBK (R = 0.54) (all P < 0.001; (Additional file 11: Fig. S7a). Then, we further used the GeneMANIA database to search for the interaction network of TOP2A, CSNK2A1, TOP2B, DFFB, LIN52, LIN9, LIN54, FOXM1, and LIN37. (Additional file 11: Fig. S7b).
To further study the functions of TOP2A in ovarian cancer, we performed KEGG, GO-BP and GO-CC enrichment analyses on TOP2A-related genes through the LinkedOmics online database. The top 50 significant genes with positive and negative correlations with TOP2A expression is displayed as a heatmap (Fig. 6b-c). KEGG analysis of positively correlated genes demonstrated that TOP2A was correlated with the cell cycle, homologous recombination, DNA replication, mismatch repair, p53 signalling pathway and platinum drug resistance. Genes negatively correlated with TOP2A were enriched in antigen processing and presentation, Th17-cell differentiation, NOD-like receptor signalling pathway and Toll-like receptor signalling pathway (Fig. 6d). GO-BP analysis of positively correlated genes identified that TOP2A was associated with cell cycle G2/M phase transition, cell cycle checkpoint, DNA replication, double-strand break repair, cell cycle G1/S phase transition and signal transduction in response to DNA damage. Genes negatively correlated with TOP2A were primarily involved in interleukin-8 production, macrophage activation, and the immune response-regulating signalling pathway (Additional file 11: Fig. S7c). The GO-CC analysis results are shown in Additional file 11: Fig. S7d.
In summary, these results demonstrated that the expression of TOP2A was related to the cell cycle, DNA damage repair, platinum drug resistance and immunity. Consequently, we then deeply investigated the association between TOP2A and immunity and the effect of TOP2A on sensitizing ovarian cancer to platinum drugs.
7. TOP2A regulates tumour-infiltrating immune cells (TIICs) in ovarian cancers.
TIICs are significant elements of the tumour microenvironment that control tumorigenesis, development and invasion according to a number of studies [42, 43]. Thus, we investigated the correlation between the components of TIICs and TOP2A expression levels in ovarian cancer. First, using the HPA database, we discovered that the expression of TOP2A was higher in regulatory T (T-reg) cells, natural killer (NK) cells, and memory B cells (Fig. 7a). TOP2A expression was linked to immune subtypes of ovarian cancer (Fig. 7b).
Next, using the ImmuCellAI portal, we explored the association between TOP2A expression and the abundance of TIICs in ovarian cancer. Based on TCGA data, the results demonstrated that high expression of TOP2A was positively associated with the abundance of CD8 naive cells, induced Treg cells (iTreg cells), T helper 1 cells (Th1 cells) and neutrophils. Cytotoxic T cells, T helper 17 cells (Th17 cells), follicular helper T cells (Tfh cells), monocytes, macrophages, NK cells, gamma delta T cells (Tgd cells) and CD8 T cells were highly abundant in the low TOP2A expression group in ovarian cancer (Fig. 7c). Based on ICGA data, high expression of TOP2A was positively related to a high abundance of CD8 naive cells and mucosal-associated invariant T cells (MAITs), but Th1 cells and B cells were abundant in the low TOP2A expression group in ovarian cancer (Additional file 12: Fig. S8a). Additionally, we also revealed that TOP2A expression was negatively linked with almost all TIICs using the SangerBox database (Additional file 12: Fig. S8b).
The proportion of immune cells and stromal cells that infiltrated the tumour tissues was evaluated by the immune score and the stromal score, respectively. The ESTIMATE score, representing the tumour purity and immune microenvironment, is the sum of the stromal and immune scores. Our analysis showed that the TOP2A mRNA expression level was negatively linked to the ESTIMATE score, immune score and stromal score (Fig. 7d-f). This demonstrated that the high expression of TOP2A was related to the decreasing infiltration of immune and stromal cells, leading to high tumour purity. However, the expression of TOP2A was strongly positively correlated with tumour purity or levels of infiltrating T cells CD8+ naive and Tregs in ovarian cancer based on the algorithm from TIMER (Additional file 12: Fig. S8c). Additionally, The Additional file 13 Fig. S9 showed heatmaps of the TIIC infiltration levels in ovarian cancer.
8. TOP2A is a potential biomarker of the tumour immunotherapeutic response in ovarian cancer
Antitumor immunity, which is linked to MSI, reflects the efficacy of tumour immunotherapy in the tumour microenvironment. Because they regulate immune cell infiltration into the tumour microenvironment, ICP-blocking proteins are the most promising targets for cancer immunotherapeutic therapies [49]. ICP inhibition medications are effective in treating tumours with high MSI[44]. Therefore, we investigated the association of TOP2A expression with MSI and the expression levels of ICP genes to determine whether TOP2A was a forecaster of the immunotherapy response in ovarian cancer. Interestingly, in ovarian cancer, TOP2A expression and MSI were positively linked (P = 0.0067) (Fig. 8a). Additionally, TOP2A expression was positively associated with the ICP genes ADORA2AP, BTNL2, CD160, CD276, CD28, LCOSLG and NRP1 but negatively related to CD40, TMIGD2, TNFRSF14, TNFSF4 and VSIR (Fig. 8b).
The costimulatory and coinhibitory signals constitute immune checkpoints that precisely regulate the function of immune cells. Cancer cells avoid being killed by immune cells by inhibiting immune checkpoint signalling [45]. Three types of molecules (MHC molecules, immunostimulators and immunostimulators) are involved in tumour escape. MHC molecules are downregulated to avoid recognition by T cells. Immunostimulators are downregulated to avoid immune destruction. Immunoinhibitors are upregulated to enable tumour escape[46]. Therefore, we explored the association between immunostimulators, immunoinhibitors, MHC molecules and TOP2A expression by using the TISIDB database. As shown in Fig. 8c-d, TOP2A expression was positively related to immunostimulatory ULBP1 and MICB but negatively correlated with C10orf54, TMEM173, TNFRSF14 and TNFSF13. TOP2A expression was negatively related to immunoinhibitory CSF1R, TGFB1 and CD244 but positively correlated with TGFBR1 (Fig. 8e). Moreover, TOP2A expression levels showed a negative correlation with almost all MHC molecules in ovarian cancer (Fig. 8f).
Collectively, our results indicated that TOP2A might be a forecaster of the tumour immunotherapy response and a potential immunotherapy biomarker.
9. TOP2A promotes proliferation, migration and invasion of ovarian cancer cells and activates the AKT/mTOR and MAPK pathways.
In addition to online analysis, the protein expression levels of TOP2A in ovarian cancer tissues were detected through WB assay. Consistently, TOP2A protein expression were significantly enhanced in ovarian cancer tissue compared with adjacent normal tissues (Fig. 9a).
To verify whether TOP2A influenced the proliferation of ovarian cancer cells, three siRNA (si-TOP2A#1, #2 and #3) sequences were transferred into the cells. RT‒qPCR and WB assays demonstrated that the mRNA and protein expression of TOP2A were significantly suppressed in TOP2A-siRNA-transfected HeyA8 and SKOV3 cells compared with those in control cells. SiRNA#1 and siRNA#2 showed higher efficiency and were used in subsequent analyses (Fig. 9b, Additional file 14: Fig. S10a). CCK8 and EdU assays demonstrated that the growth of HeyA8 and SKOV3 cells was markedly suppressed after TOP2A knockdown (Fig. 9c-d, Additional file 14: Fig. S10b-c). Additionally, the migration and invasion abilities of HeyA8 and SKOV3 cells significantly decreased after TOP2A knockdown (Fig. 9e-f, Additional file 14: Fig. S10d-e).
To investigate the mechanism of the effect of TOP2A on ovarian cancer cell proliferation, invasion and migration, we detected the expression levels of several cell cycle-specific and epithelial-mesenchymal transition (EMT) proteins. As shown in Fig. 9g and Additional file 14: Fig. S10f, the protein expression levels of CDK2 and CDK6 were remarkably suppressed in HeyA8 and SKOV3 cells with TOP2A knockdown, but P21 was significantly increased. Additionally, E-cadherin was significantly increased in HeyA8 and SKOV3 cells with TOP2A knockdown, but N-cadherin and Vimentin were significantly decreased (Fig. 9h, Additional file 14: Fig. S10g). The expression level of p-H2AX represents the degree of DNA damage. The p-H2AX expression levels were remarkably enhanced in HeyA8 and SKOV3 cells with TOP2A knockdown, suggesting that TOP2A knockdown may induce DNA damage (Fig. 9i, Additional file 14: Fig. S10h). Since the Akt/mTOR[47] and MAPK[48] signalling pathways have been implicated in the pathogenesis of ovarian cancer, we investigated whether TOP2A can activate these signalling pathways. WB assays demonstrated that p-Akt, p-mTOR, p-ERK1/2, p-P38 and p-JNK1/2/3 were decreased after TOP2A knockdown (Fig. 9j-k, Additional file 14: Fig. S10i-j).
Collectively, our findings suggested that TOP2A might promote the proliferation, migration and invasion of ovarian cancer cells and activate the Akt/mTOR and MAPK signalling pathways.
10. TOP2A knockdown inhibits chemoresistance in ovarian cancer cells
GSEA showed that ovarian cancer samples with high TOP2A expression could be enriched in platinum resistance pathway (Fig. 10a). Importantly, TOP2A protein expression in CBPt-treated ovarian cancer cells was significantly increased (Fig. 10b, Additional file 15: Fig. S11a). Thus, we examined whether TOP2A could influence the sensitivity of ovarian cancer cell to CBPt. We assessed the viability of si-TOP2A cells with different concentrations of CBPt. The results showed that TOP2A knockdown effectively restored sensitivity of HeyA8 and SKOV3 cells to CBPt, and the 50% inhibitory concentration (IC50) values were significantly decreased (Fig. 10c, Additional file 15: Fig. S11b). CCK8 and EdU assays demonstrated that compared to cells treated with CBPt or si-TOP2A alone, the proliferation of HeyA8 and SKOV3 cells was inhibited in TOP2A knockdown cells treated with CBPt. (Fig. 10d-e, Additional file 15: Fig. S11c-d). Additionally, CBPt or TOP2A knockdown alone inhibited migration and invasion in ovarian cancer cells, and the inhibitory effect of combination treatment was more obvious (Fig. 10f-g, Additional file 15: Fig. S11e-f). Consistently, after treatment with CBPt, the apoptotic/necrotic ratio of TOP2A knockdown cells was greatly increased (Fig. 10h-i, Additional file 15: Fig. S11g-h).
Platinum drugs have been found to damage cancer cells mainly through the DNA damage pathway in previous studies[49]. We next examined whether TOP2A knockdown can enhance the DNA damage conferred by CBPt. TOP2A knockdown cells treated with CBPt had significantly higher p-H2AX levels than cells treated with CBPt or TOP2A knockdown alone (Fig. 10j, Additional file 15: Fig. S11i). As expected, the combination of TOP2A knockdown and CBPt treatment presented an enhanced antitumor effect compared with CBPt alone.
Together, these results suggested that TOP2A might mediate the resistance to CBPt in ovarian cancer cells. The combination of TOP2A knockdown and CBPt treatment significantly inhibited the proliferation, migration and invasion of HeyA8 and SKOV3 cells.