The KIF2C expression of tumor and adjacent normal tissues profile was analyzed using the TIMER 2.0 database (https://cistrome.shinyapps.io/timer/), a bioinformatics tool for analyzing and visualizing TCGA (Li et al., 2020a). The 33 TCGA cancer types analyzed are shown in Table 1. The gene expression levels are represented as log2 transcript per million (TPM) values. For some tumor tissues without normal tissues, normal expression comparison analyses, the TPM of the given genes from TCGA, and matched GETx (Genotype-Tissue Expression) normal tissue RNA sequencing (RNA-seq) datasets were visualized by GEPIA2 (http://gepia2.cancer-pku.cn/#analysis) (Tang et al., 2017). The cutoff for significance was chosen at p = 0.01. The tumor pathological staging map was also obtained from GEPIA2. TPM values were calculated, and the expression levels of genes were presented using the Log2 (TPM + 1) scale.

Protein expression of KIF2C was evaluated using the UALCAN online analysis software (http://ualcan.path.uab.edu/index.html), which is a comprehensive web resource for analyzing cancer-omics data including Clinical Proteomic Tumor Analysis Consortium analysis in various tumor types (Chen et al., 2019).

The correlation between KIF2C messenger RNA (mRNA) expression and overall survival (OS) and disease-free survival (DFS) was predicted using the PrognoScan database (http://www.abren.net/PrognoScan/) and GEPIA2. PrognoScan database is also a collection of publicly available cancer microarray datasets and a tool for assessing the biological relationships between gene expression and prognosis, providing a convenient platform to evaluate potential tumor markers and therapeutic targets (Mizuno et al., 2009). The threshold was set as a Cox p-value < 0.05.

Genomic alteration data were analyzed through the cBio Cancer Genomics Portal (http://cbioportal.org) (Cerami et al., 2012; Gao et al., 2013), which collects multidimensional cancer genomics dataset. We observed the alteration frequency and mutation type of all 33 tumors. We also obtain the data on the overall disease-free, progression-free, and DFS differences for TCGA cancer cases with or without KIF2C genetic alteration.

HPA provides tissue and cell distribution information of 26,000 human proteins (Uhlén et al., 2015). In this database, researchers used highly specific antibodies and immunoassay techniques to detect the expression of each protein in 64 cell lines, 48 human normal tissues, and 20 tumor tissues in detail. In this study, we conducted HPA to observe the localization of KIF2C in cells and compare the protein expression of KIF2C between normal tissue and carcinoma tissue.

The LinkedOmics database 8 (http://linkedomics.org) (Vasaikar et al., 2018) is an online open-access powerful bioinformatics platform, which includes multi-omics information and clinical data involving 11,158 patients and 32 cancer types in TCGA project. We used the Linkedomics platform to analyze the relationship between the expression level of KIF2C in TCGA database and the clinical–pathological characteristics of patients.

Tissue specimens were obtained from patients who underwent hysterectomy and pathological diagnosis in the Department of Pathology of The Third Affiliated Hospital of Zhengzhou University (Zhengzhou, China) from 2017 to 2021. Paraffin-embedded histological specimens of 35 cervical squamous tissue samples, 25 cervical adenocarcinoma tissue samples, and 40 normal cervical tissues samples were sectioned (thickness 5 μm). All the patients had not received any antitumor therapy before their diagnoses. The project was approved by the Medical Ethics Committee of the Third Affiliated Hospital of Zhengzhou University (ethics approval number: 2021-080-01) and performed in accordance with the principles of the Declaration of Helsinki.

Immunohistochemistry and scoring criteria were performed as described (Soumaoro et al., 2004). Briefly, the tissue sections were deparaffinized in xylene, rehydrated in graded ethanol, and soaked in distilled water. After heat-mediated citric acid antigen retrieval, the tissue sections were incubated with the primary antibody for KIF2C (DF7348, rabbit antihuman, 1:100, Affinity) and Ki-67 (ab92742, rabbit antihuman, 1:500, Abcam) at 4°C overnight and incubated with horseradish peroxidase-conjugated antibody for 60 min at room temperature. Staining was developed by incubating with diaminobenzidine. Finally, these processed tissue sections were observed under a microscope and then counterstained with hematoxylin. The results were scored independently by two pathologists. A total score of >3 points was considered a high expression, and ≤3 points was considered low expression.

Cervical cancer cell lines SiHa, HeLa, C33a, and Caski and normal cervical cell line HCerEpic were borrowed from the relevant research group with our laboratory. All cells were cultured in Dulbecco’s modified Eagle’s medium (Solarbio, SuoLaibao Technology Co., Ltd.) supplemented with 10% fetal bovine serum (Gibco, Thermo Fisher Scientific, Inc.) and incubated at 37°C in a moist atmosphere containing 5% CO₂.

When C33a and SiHa cells reached 20–30% confluence on the day of transfection, two lentiviral stocks (named GV248-shKIF2C and GV248-shCon, purchased from Genechem, Shanghai, China) were transduced into the cells with HitransG A, respectively. Stable cell lines were screened with 0.25 μg/ml puromycin. The cells were collected when 80–90% of the cells with green fluorescence were grown under a fluorescence microscope, then quantitative real-time polymerase chain reaction (qRT-PCR) and Western blot analysis were used to analyze the level of KIF2C expression.

CRISPR/Cas9 p53 knockdown and its control plasmids were kindly provided by Dr. Qi Zhang; then, the plasmids were transfected into SiHa-sh-KIF2C and SiHa-sh-NC cervical cancer cells, respectively. The cells were transfected using JetPrime transfection reagent (Polyplus transfection, Illkirch, France) according to the manufacturer’s protocol; then, Western blot analysis was used to analyze the level of p53 expression.

For overexpression, the plasmid (vector name: GV657) for human KIF2C and negative control were synthesized by Genechem (Shanghai, China), then transfected into Caski cell using JetPrime transfection reagent (Polyplus transfection, Illkirch, France) according to the manufacturer’s protocol using 1=µl of JetPrime for 1 µg of plasmid DNA. After being transfected with 24 h, the protein was collected, and Western blotting was used to verify the transfection efficiency.

According to the supplier’s instructions, synthesis of complementary DNA by extracting 1 μg total RNA with Trizol (Cwbio, Beijing, China) and SYBR reverse transcription reagent (US Everbright Inc., Suzhou, China). The ratio of target glyceraldehyde 3-phosphate dehydrogenase expressed the relative expression of target gene mRNA. The sequences for the primers are as follows:

GADPH, Forword, 5′-TGA​CTT​CAA​CAG​CGA​CAC​CCA-3′;

The relative gene expression was normalized to control by the 2^−ΔΔCT method.

Total proteins from cells were extracted with radioimmunoprecipitation assay lysis buffer (Solarbio, SuoLaibao Technology Co., Ltd.). Western blot analysis was carried out by first electrophoresing the proteins through a sodium dodecyl sulfate–polyacrylamide gel electrophoresis gel and subsequently transferring to 0.45-µm polyvinylidene fluoride membrane (Millipore, Ireland). Immunoblotting was performed using antibodies against indicated proteins: KIF2C (DF7348, rabbit antihuman, 1:1,000, Affinity), p21 (10355-1-AP, rabbit antihuman, 1:1,000, Proteintech), p53 (sc126, mouse antihuman, 1:, Santa), Bax (50599-2-Ig, rabbit antihuman, 1:5,000, Proteintech), bcl-2 (60178-1-Ig, mouse antihuman, 1:2,000, Proteintech), and GADPH (bs2188R, rabbit antihuman, 1:5,000, Bioss). Glyceraldehyde 3-phosphate dehydrogenase was used as an endogenous control. The transferred membranes were blocked with 5% milk for 2 h and then incubated with the indicated primary antibody overnight at 4°C. Then, the membranes were incubated with the appropriate secondary antibodies (1:9,000) at room temperature for 1 h. Finally, the electrogenerated chemiluminescence solution (A:B = 1:1) was added onto the membrane, and the front of the membrane was placed on the glass plate. The GelView 6000Plus was used to process the pictures after exposure.

The successfully transfected cervical cancer cells (C33a, SiHa, and Caski) were inoculated in 96-well plates with 4 × 10³ cells per well. A 10-µl CCK-8 solution (US Everbright Inc., Suzhou, China) was added to each well after plating for 12, 24, 48, 72, and 96 h. When the cells were further incubated for 3 h, a microboard reader (Thermo Scientific) was used to measure the spectrometric absorbance at the wavelengths of 450 nm.

After transfection, the cells were inoculated in six-well plates with 600 per well and cultured 2 weeks. Colonies were fixed with methanol for 20 min and subsequently stained with 0.1% crystal violet for 30 min. Colonies with more than 60 cells were defined as positive.

For this assay, serum-free C33a, SiHa, and Caski cells were seeded to the upper chamber with 5 × 10⁴ cells in 100 µl, and the medium containing 20% fetal bovine serum was used as a chemoattractant in the lower chamber. After incubating for 36 h, migrated cells were fixed by formaldehyde and stained with crystal violet. For invasion assay, cells were also suspended in serum-free medium and counted and seeded in the upper chamber at a concentration of 5 × 104 cells in 100 µl; the only difference was spreading the Matrigel Basement Membrane Matrix (BD Bioscience, United States) on the upper chamber before adding the cells. The number of migrated and invaded cells was captured and calculated using a microscope.

Total RNAs were extracted from successfully transfected cervical cancer cells using Trizol reagent (Cwbio, Beijing, China). After a primary test for RNA quality and integrity, the sequencing samples were finally sequenced on Novaseq 6000 sequencer (Illumina) with PE150 model. The raw sequencing data were mapped to the human genome using STRA 2.5 software. Reads mapped to the exon regions of each gene were counted by featureCounts, and then reads per kilobase of transcript per million reads mapped was calculated. Differentially expressed (p-value cutoff of 0.05; the fold-change cutoff of 2) genes between groups were identified using the edgeR package. For further precise analysis, Gene Ontology (GO) analysis and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analysis for differentially expressed genes were conducted by KOBAS 2.1 software (p-value < 0.05; false discovery rate < 0.05). The RNA sequencing analysis and data analysis were performed with the help of Seqhealth cooperation (Wuhan, China). RNA-seq data have been deposited to National Center for Biotechnology Information Sequence Read Archive (Sequence Read Archive accession: PRJNA767113).

Four- to 5-week-old female BALB/c nude mice (SPF Biotechnology Co., Ltd., Beijing, China) were used in this study. The control and experimental cervical cancer cells (2 × 10⁶) were transplanted to establish the subcutaneous xenograft model. Tumor volumes were measured every 7 days. After 4 weeks, the mice were killed, and tumor weight was examined. The tumor volumes were calculated by the formula: tumor volume = length × width²/2. Animal research was approved by the Animal Care and Use Committee of Zhengzhou University and was conducted in accordance with the Animal Care Guidelines of Zhengzhou University.

Statistical analysis of all data was done using SPSS 26.0 software. The measurement data with normal distribution was exhibited as the mean ± standard deviation. The statistical significance between the two groups was compared with the two-tailed Student’s t-test. Pearson’s chi-square test analyzed the association between KIF2C expression and clinicopathological features. p < 0.05 was considered statistically significant. All cell experiments had three repetitions.

In the TIMER database, we investigated the expression of KIF2C in pan-cancer. The GEPIA2 database supplemented part of cancers without normal samples. The results are shown in Figures 1A,B. As expected, KIF2C expression was significantly increased in most tumor tissues compared with normal tissues, including TCGT and LAML. However, there was no significant difference between tumor and normal tissues in ACC and LGG. As there were no normal samples studied in either dataset, an analysis could not be performed for MESO and UVM. In addition, the protein expression level of KIF2C was also shown in a similar manner. Clinical Proteomic Tumor Analysis Consortium analysis showed that the protein expression of KIF2C was higher in primary tissues of BRAC, OV, COAD, KIRC, UCEC, and LUAD compared with those in normal tissues (Figure 1C). All the results were statistically significant (p < 0.05). Next, we used the UALCAN database to analyze the correlation between KIF2C mRNA expression and clinicopathological stages. Expression of KIF2C was highly associated with advanced clinical stage (Figure 1D) of ACC (p < 0.001), BRCA (p < 0.001), KICH (p < 0.001), KIRC (p < 0.001), KIRP (p < 0.001), SKCM (p = 0.04), LUAD (p = 0.001), and LIHC (p < 0.001) but not others.

Next, we investigated KIF2C prognostic value using PrognoScan and GEPIA2 database. In PrognoScan, high KIF2C expression is associated with worse OS [hazard ratio (HR) = 1.45, Cox p = 0.000964] and disease-specific survival (DSS, HR = 2.2, Cox p = 0.000025) in bladder cancer (Figures 2A,B), with worse OS (HR = 4.35, Cox p = 0.000473) in brain cancer (Figure 2C), with worse OS (HR = 1.39, COX p = 0.047079), distant metastasis-free survival (HR = 4.98, COX p = 0.003309), DFS (HR = 6.19, COX p = 0.001105), and recurrence-free survival (RFS, HR = 4.98, COX p = 0.000259) in breast cancer (Figures 2D–G), with worse distant relapse-free survival (HR = 2.33, COX p = 0.000006) in soft tissue cancer (Figure 2H), with worse OS (HR = 3.6, COX p = 0.042603), RFS (HR = 2.41, COX p = 0.000001) in lung cancer (Figures 2I,J) and with worse OS (HR = 9.01, COX p = 0.000058) in skin cancer (Figure 2K). Conversely, KIF2C high expression is associated with good DFS (HR = 0.53, COX p = 0.013566) in colorectal cancer (Figure 2L). For blood cancer (Figures 2M,N), KIF2C presented the opposite manner in OS (HR = 0.53, COX p = 0.017) and DSS (HR = 2.53, COX p = 0.000363). For other cancers, no statistically significant associations were observed.

To further explore the prognostic value of KIF2C, we performed the survival analysis using Kaplan–Meier curves from the GEPIA2. As shown in Figure 3A, higher KIF2C expression was associated with poor OS in ACC (p < 0.001), KICH (p = 0.018), KIRC (p = 0.006), KIRP (p = 0.0017), LGG (p < 0.001), LIHC (p < 0.001), LUAD (p = 0.016), MESO (p < 0.001), PAAD (p = 0.02), SKCM (p = 0.01), and PRAD (p = 0.049), whereas higher KIF2C expression was linked to better prognosis in THYM (p = 0.034). As for DFS (Figure 3B), high KIF2C expression was significantly associated with worse DFS in ACC (p < 0.001), KIRC (p = 0.0018), KIRP (p < 0.001), LGG (p < 0.001), PAAD (p < 0.001), LIHC (p < 0.001), PRAD (p = 0.00011), THCA (p = 0.03), and MESO (p = 0.016). These results suggested that higher KIF2C was associated with worse prognosis.

KIF2C mutations and copy number alterations were investigated using the cBioPortal database. The genetic alteration includes mutation, amplification, deep deletion, and multiple alterations. Notably, the highest mutation frequency (>6%) was observed in UCEC (Figure 4A). The amplification type of CNA was the major type in the OV, which showed an alteration frequency of ∼4% (Figure 4A). The predominant mutation type in PGPC was a deep deletion that resulted in a frameshift (Figure 4A). All COAD and LAML cases with genetic alteration had a mutation of KIF2C (Figure 4A). In addition, the amplification of KIF2C was present in all SARC, MESO, and LIHC cases (Figure 4A). The information of KIF2C genetic alteration is presented in Figure 4B. The missense mutation was the only type of genetic alteration of KIF2C. In addition, we also explored the potential association between KIF2C genetic alteration and clinical survival rate in various cancers. The CESC patients with KIF2C genetic alteration had poor prognosis in OS (Figure 4C, p = 0.047), progression-free survival (Figure 4C, p < 0.001), DFS (Figure 4C, p < 0.001), and DSS (Figure 4C, p = 0.0167). We also used the HPA to investigate the intracellular location and tissue protein expression levels of KIF2C. We found that KIF2C was mostly localized in both cellular nucleus and cytosol (Figure 4D), and it was only detected in CESC tissues (Figure 4D). These results suggested that KIF2C could be a potential therapeutic target in CESC.

To explore the possibility of KIF2C as a therapeutic target of cervical cancer, we first detected its expression in human tissues. Findings showed that KIF2C was upregulated in cervical cancer tissues compared with normal tissues (Figures 5A,C,E). Not only that, the expression of KIF2C was higher in squamous cervical carcinoma than cervical adenocarcinoma. All three differences were statistically significant (p < 0.05, Figure 5G). Besides, we also found that the expression of KIF2C has a similar expression pattern with Ki-67 in cervical tissues (Figures 5B,D,F). This suggested that the expression of KIF2C may be related to the active proliferation of cancer cells.

To investigate the clinical relevance of KIF2C in cervical cancer, 304 cervical cancer cases from TCGA were analyzed. The results showed that KIF2C expression was correlated with the type of cervical cancer (p < 0.05), whereas it was not directly correlated with age, tumor–nodes–metastases stage, 2018 International Federation of Gynecology and Obstetrics stage, lymphatic metastasis, and distant metastasis (Table 2).

Because we found that KIF2C was clinically relevant in cervical cancer in the discussion earlier, we examined the KIF2C expression in various cervical cancer and normal cell lines, including C33a, HeLa, SiHa, Caski, and HCerEpic cell lines. We analyzed mRNA and protein levels of KIF2C using qRT-PCR and Western blot assay in those cells. The results showed that the mRNA or protein levels of KIF2C in C33a, HeLa, and SiHa cells were higher than those in HCerEpic cells (Figures 6A–C). The two cell lines (SiHa and C33a) with the highest KIF2C expression and Caski cell, which had a relatively low level of KIF2C expression, were used for subsequent experiments. To further investigate the role of KIF2C in cervical cancer, we established the stable KIF2C knockdown cell lines using the short hairpin RNA lentiviral system. We confirmed the knockdown efficiency using Western blot, qRT-PCR (Figures 6D–I), and green fluorescent protein (Figure 6J) in SiHa and C33a cells. For overexpression, the Caski cell was transfected with KIF2C plasmid, and the transfection efficiency was verified with Western blot (Figure 6K,L).

To explore the role of KIF2C in cervical cancer cells, we examined the cell viability using CCK-8 assay and clone formation assay in SiHa or C33a cells. The viability of sh-KIF2C SiHa or C33a cells was significantly decreased compared with that of control cells (Figure 7A). These results indicated that the KIF2C knockdown inhibited the proliferation of cervical cancer cells. Next, we investigated the role of KIF2C on in vitro tumorigenicity in SiHa and C33a cells. The clonal colonies of sh-KIF2C SiHa and C33a cells were also significantly decreased compared with those of control cells (Figure 7B). We then examined whether KIF2C knockdown could affect the migration and invasion in SiHa and C33a cells. Results showed that the numbers of migrated or invaded cells in KIF2C knockdown groups were significantly lower than those in control groups (Figure 7C). These results indicated that KIF2C knockdown suppresses the migration and invasion of cervical cancer cells.

To further probe the role of KIF2C in cervical cancer cells, we transfected the KIF2C overexpression plasmid into the Caski cell line. CCK-8 assay and clone formation experiment were used to detect the cell viability and clone formation ability. For the CCK-8 assay, the results showed the cell viability was significantly increased in the KIF2C overexpression groups compared with the control groups (Figure 8A). For the clone formation experiment, the clonogenic ability of Caski cells, which overexpressed KIF2C, was significantly increased (Figure 8B). All findings indicated that KIF2C overexpression could promote cell proliferation. By the same token, a transwell assay was performed to examine cell migration and invasion ability, and the results demonstrated that KIF2C overexpression can remarkably improve the migration and invasion ability of cervical cancer cells (Figure 8C).

To gain a deeper insight into the mechanism of KIF2C expression, we used RNA-seq analysis. We screened the differentially expressed genes using sh-KIF2C SiHa and sh-NC SiHa. Under the screening criteria, which were false discovery rate < 0.05 and |log2FC| >1.5, a total of 349 differentially expressed genes (104 upregulated and 245 downregulated genes) were analyzed (Figure 9A). The results were presented as a heat map, in which blue was the downregulated genes and red was the upregulated genes (Figure 9B). Then, we subjected the 349 genes to GO and KEGG analyses. The results of the GO analysis showed the main pathways in which the upregulated genes were involved in protein transport into membrane raft, protein transport within the lipid bilayer, negative regulation of signal transduction, negative regulation of signaling, etc. (Figure 9C). In addition, it also showed the main pathways in which the downregulated genes were involved in xenobiotic glucuronidation, negative regulation of cellular glucuronidation, regulation of glucuronosyltransferase activity, immune system process, and other signaling pathways (Figure 9D). The results of the KEGG analysis revealed that the upregulated genes were mainly enriched in mucin-type O-glycan biosynthesis, leukocyte transendothelial migration, cell adhesion molecules (CAMs), phagosome, etc. (Figure 9E), whereas the downregulated genes were mainly enriched in drug metabolism-other enzymes, ascorbate and aldarate metabolism, pentose and glucuronate interconversions, and other related pathways (Figure 9F).

To further experimentally validate the specific mechanism by which KIF2C regulates cervical carcinogenesis, we deeply analyzed the KEGG pathway and found that its differentially expressed genes were enriched in the p53 signaling pathway (Supplementary Table S1). There have been some studies showing that alteration of the p53 signaling pathway is involved in the development of cervical cancer. Given that, we further analyzed the expression of proteins involved in the p53 signaling pathways, including p21, p53, bcl-2, and bax. In SiHa and C33a cells with KIF2C knockdown, the expression of p21, p53, and bax proteins was significantly increased, whereas the expression of bcl-2 protein significantly decreased (Figures 10A–D). However, the expression of the proteins mentioned earlier exhibited an opposite trend in the presence of the overexpressed KIF2C (Figures 10E,F). These results indicated that KIF2C might promote cervical cancer progression through the p53 signaling pathway.

For additionally exploring the associations of KIF2C with p53 signaling pathways, the p53 CRISPR/Cas9 knockout plasmid and control CRISPR/Cas9 plasmid were transfected into the SiHa cell with KIF2C stable knockdown. Transfection efficiency was verified using Western blot assay (Figures 11A,B). The outcome revealed that p53 knockdown reversed the prohibiting effect of KIF2C-silenced on cell proliferation (Figures 11C–E), migration, and invasion (Figures 11F–H). Besides, we also found that the tendency of p21 and Bax showed the same trend with p53; the level of bcl-2 showed the opposite results (Figures 11I,J). All these data detailed that KIF2C might induce cervical cancer cells growth, invasion, and migration by regulating the p53 signaling pathway.

To assess the role of KIF2C in mouse xenograft models, we subcutaneously inoculated SiHa and C33a cells in the flanks of nude mice. We continuously monitored the tumor growth and measured the volume and weight of tumors for 4 weeks. We found that tumor growth of the sh-KIF2C group was significantly decreased compared with that of the control group (Figure 12A). In addition, the volume and weight of tumors of the sh-KIF2C group were lower than those of the control group (Figures 12B,C). These results indicated that KIF2C knockdown played a tumor-suppressive role in vivo.