The Gene Expression Omnibus database (https://www.ncbi.nlm.nih.gov/gds/?term=GSE47074) provided the GSE47074 dataset, which included four primary colon cancer samples (tumor group) and four standard samples (control group). In addition, The Cancer Genome Atlas (TCGA) (https://tcga-data.nci.nih.gov/tcga/) database provided 414 colon adenocarcinoma (COAD) samples (tumor group) and 41 standard samples (control group). Differentially expressed genes (DEGs) in the two datasets were obtained separately utilizing the limma package in R studio (version 4.0, http://www.bioconductor.org/packages/release/bioc/html/limma.html). Genes with fold change (FC) $>$2 were classified as upregulated, whereas genes with FC 0.5 were considered downregulated, with significance established at p 0.05. The ggplot2 package (https://cran.r-project.org/package=ggplot2) in R was utilized to visualize the DEGs. Then the up- and downregulated DEGs in TCGA-COAD and GSE47074 were cross-analyzed using the Bioinformatics and Evolutionary Genomics (http://bioinformatics.psb.ugent.be/webtools/Venn/) tool to obtain the genes in the intersection.

Next, protein–protein interaction (PPI) network analysis was conducted on overlapping genes utilizing the Search Tool for the Retrieval of Interacting Genes (STRING) (https://string-db.org/) database. Subsequently, the Cytohubba plugin in Cytoscape (version 3.8.2, IBM, Armonk, NY, USA) was used to apply the maximum clique centrality (MCC), maximum neighborhood component (MNC), and degree algorithms to identify the top 20 genes of three key modules. Lastly, we performed intersection analysis of the genes from these three network modules to pinpoint the essential intersection genes critical for further investigation.

To investigate the 15 essential intersection genes in the TCGA-COAD dataset, we implemented the least absolute shrinkage and selection operator (LASSO) Cox regression algorithm using the glmnet package in R (version 4.0, http://cran.r-project.org/web/packages/glmnet/index.html). A LASSO Cox regression model was built with 10-fold cross-validation, plotting the relationship between the logarithm of lambda ($\lambda$) and the partial likelihood deviation. The final coefficients for the characteristic genes were determined, which facilitated samples to be classified as high or low risk according to the expression of crucial ferroptosis genes in the TCGA-COAD database. Then a risk model was established to analyze the survival status, survival time, and gene cluster distribution of these groups. The survival package in R was utilized to create Kaplan-Meier (KM) survival curves and calculate overall survival (OS) probabilities. The 95% confidence intervals (CIs) and hazard ratios (HRs) were computed using the log-rank test and univariate Cox proportional hazard regression. To evaluate prognostic power, receiver operating characteristic (ROC) curves were constructed for 1-, 3-, and 5-year survival rates, with the area under the curve [22] used to assess diagnostic accuracy. Lastly, prognosis-related gene expression was examined in the tumor and control groups from the TCGA-COAD and GSE47074 datasets using the Sangerbox platform (version 3.0, SangerBox Bioinformatics, Beijing, China; available at http://vip.sangerbox.com/home.html), leading to identification of the hub gene in this study.

The association of AURKB and MAD2L2 expression levels was investigated utilizing the Gene Expression Profiling Interactive Analysis (GEPIA) database (http://gepia.cancer-pku.cn/) to learn more about the link between the two. The correlation coefficient across the two genes was obtained. p 0.05 was considered statistically significant.

CRC cell lines (LoVo, HCT-116, and SW480) and the NCM460 normal human colon mucosal epithelial cell line were purchased from the American Type Culture Collection (ATCC, Shanghai, China). Cells were cultured in Dulbecco’s Modified Eagle Medium supplemented with 10% fetal bovine serum, 100 U/mL penicillin, and streptomycin (all from Sangon, Shanghai, China) at 37 °C in a humidified incubator with 5% CO₂. All cell lines were validated by STR profiling and tested negative for mycoplasma. Detailed validation results are provided in Supplementary Fig. 1.

CRC cells were plated in 6-well dishes at a density of 5 $\times$ 10⁵ cells per well. Once the cells were 70–80% confluent, they were transfected with 10 µL Lipofectamine™ 2000 (Invitrogen, Shanghai, China) according to the manufacturer’s instructions. For AURKB knockdown, two distinct small interfering RNAs (siRNAs) targeting AURKB were transfected into the cells. Non-targeting negative control small interfering RNA (si-NC) was utilized as a negative control. To overexpress MAD2L2, a MAD2L2 expression vector was used to transfect the cells, with empty vector as the control. Two days after transfection, cells were collected for further analysis. The sequence of si-AURKB-1 was: 5^′-GGAGGAGGAUCUACUGAUGAUUCUAGA-3^′ (forward), 5^′-UCUAGAAUCAAGUAGAUCCUCCUCC-3^′ (reverse); the sequence of si-AURKB-2 was 5^′-GAUCAUGGAGGAGGAGUUGGCAGAUGCU-3^′ (forward), 5^′-AGCAUCUGCCAACUCCUCCAUGAUC-3^′ (reverse).

Total RNA was extracted from CRC cells using TRIzol reagent (9109, Takara, Dalian, China), and complementary DNA (cDNA) was synthesized using the PrimeScript™ RT Reagent Kit (RR047A, Takara, Dalian, China). SYBR Green (RR820A, Takara, Dalian, China)-based quantitative PCR (qPCR) was conducted on the CFX96 Real-Time PCR Machine (1855195, Bio-Rad, Hercules, CA, USA) according to standard qPCR procedures. The 2^{-Δ⁢Δ⁢Ct} method was employed to quantify the gene expression levels, with GAPDH serving as an internal control. The range of primer sequences utilized in qPCR is summarized in Table 1.

Proteins were extracted from cultured cells with Radioimmunoprecipitation Assay (RIPA) buffer (P0013B, Beyotime, Shanghai, China) containing phosphatase and protease inhibitors. Then the cell lysate was centrifuged for 20 min, and proteins were quantified with a bovine serum albumin (BSA) protein assay kit (P0010, Beyotime, Shanghai, China). Equal protein concentrations were resolved by sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) and electrotransferred to PVDF membranes (Beyotime, shanghai, China). Then the membranes were incubated overnight at 4 °C with the following primary antibodies (all from Abcam, Shanghai, China): AURKB (1:1000, ab2254), cyclin D1 (1:10,000, ab134175), p27 (1:5000, ab32034), p21 (1:1000, ab109520), B-cell lymphoma 2 (Bcl-2; 1:2000, ab182858), Bcl-2-associated X protein (Bax, 1:1000, ab32503), caspase-3 (1:5000, ab32351), pyruvate kinase M2 (PKM2; 1:10,000, 15822-1-AP), lactate dehydrogenase A (LDHA; 1:5000, ab52488), hexokinase 2 (HK2; 1:1000, ab209847), glucose transporter 1 (GLUT1; 1:200, ab150299), MAD2L2 (1:1000, ab180579), p53 (1:1000, ab32049), H2A histone family member X (H2A.X; 1:5000, ab140498), and GAPDH (1:5000, ab8245). After three washes with Tris-Buffered Saline with Tween 20 (TBST), membranes were incubated with horseradish peroxidase conjugated anti-mouse (1:1000, A0216, Beyotime, Shanghai, China) and anti-rabbit (1:1000, A0208, Beyotime, Shanghai, China) antibodies, and placed in an overnight incubation at 4 °C. Proteins were detected using enhanced chemiluminescence (ECL; Beyotime) and quantified with ImageJ software (version 1.8.0, National Institutes of Health, Bethesda, MD, USA).

After centrifugation of HCT-116 and LoVo cells, 500 µg protein from each sample was incubated overnight at 4 °C with 2 µg of either anti-AURKB (1:30; ab45145, Abcam), anti-MAD2L2 (1:40; ab180579, Abcam), or IgG control (1:200; ab200699, Abcam) . Then the samples were incubated for 2 h with Protein A/G agarose beads (Santa Cruz Biotechnology, Dallas, TX, USA). Boiling in SDS-PAGE loading buffer allowed bound proteins to be eluted from the beads after they had been cleaned with RIPA buffer (P0013B, Beyotime, Shanghai, China). Following SDS-PAGE and electrotransfer to PVDF membranes, the proteins were detected by Western blotting using anti-AURKB and anti-MAD2L2 antibodies. Protein detection was performed using an ECL detection system (P0018S, Beyotime, Shanghai, China).

To measure cell viability, CRC cells were trypsinized and plated in 96-well plates at a density of 2 $\times$ 10³ cells per well. After 0, 1, 2, 3, and 4 days of culture, 10 µL/well of Cell Counting Kit-8 (CCK-8) reagent was added to the cells and incubated for 1 h. The absorbance at 450 nm was determined with a microplate reader (Model 680, Bio-Rad, Hercules, CA, USA).

A Transwell assay was conducted to measure the invasive capabilities of CRC cells. The cells were seeded in the upper chamber of Transwell (Corning, Shanghai, China) that were pre-coated with 80 µL Matrigel (BD Biosciences, Shanghai, China). Then 600 µL complete medium was added to the lower chamber, and 200 µL serum-free media containing 1 $\times$ 10⁴ CRC cells was added to the upper chamber. Utilizing a cotton swab, non-invasive cells on the upper membrane surface were removed during a 48 h incubation at 37 °C. After fixing in 4% paraformaldehyde, cells were stained for 15 min with DAPI (C1002, Beyotime, Shanghai, China). Cells were observed and recorded with an optical microscope (BX53, Olympus, Tokyo, Japan). The same procedure used for the invasion test was applied to the CRC cells plated to the upper chamber, which was not coated with Matrigel for the migration experiment.

Flow cytometry was employed to evaluate cell cycle distribution. After washing cells with cold phosphate-buffered saline (PBS), CRC cells were fixed in 70% ethanol overnight at 4 °C. Then the cells were washed, stained with propidium iodide (PI) solution, and incubated for 30 min at room temperature in the dark. The FACSCalibur™ flow cytometer (342973, BD Biosciences, San Jose, CA, USA) was utilized for flow cytometry, and data were evaluated with FlowJo software (version 10.0, BD Biosciences, San Jose, CA, USA) to quantify the cell cycle phases (G1, S, G2).

Apoptosis was assessed using an Annexin V-FITC/PI staining kit (556547, BD Biosciences, San Jose, CA, USA). After washing with cold PBS, CRC cells (1 $\times$ 10⁶ cells/mL) were resuspended in Annexin V binding buffer. Utilizing the FACSCalibur™ flow cytometer, cells were stained with 5 µL Annexin V-FITC (556420, BD Biosciences, San Jose, CA, USA) and 5 µL PI, and then incubated in the dark at room temperature for 15 min. Apoptotic cells were quantified as Annexin V-positive/PI-negative (early apoptosis) and Annexin V-positive/PI-positive (late apoptosis) using FlowJo software (version 10.0, BD Biosciences, San Jose, CA, USA).

To measure the levels of malondialdehyde (MDA) (cat.no. S0131S), LDH (cat.no. C0016), and superoxide dismutase (SOD) (cat. no. S0101S), detection assay kits were used (Beyotime, Shanghai, China). Briefly, CRC cells (5 $\times$ 10³ cells/well) were seeded in 96-well plates and allowed to adhere. Then reaction mixtures for LDH, MDA, and SOD detection were added to the wells and incubated for 30 min according to the manufacturer’s instructions. Finally, the cells were gently washed with PBS, and the respective detection kits were used to quantify the levels of LDH, superoxide dismutase (SOD), and malondialdehyde (MDA).

The levels of intracellular reactive oxygen species (ROS) of CRC cells were measured using a ROS detection kit (E004-1-1, Nanjing Jiancheng Bioengineering Institute, Nanjing, China). Briefly, HCT-116 and LoVo cells were plated at a density of 2 $\times$ 10⁴ cells/well in 96-well plates. Then the CRC cells were incubated for 30 min in the dark with 10 µM 2^′,7^′-dichlorofluorescein diacetate (DCFH-DA, P0011, Beyotime Biotechnology, Shanghai, China). After incubation, the cells were rinsed three times in serum-free medium, and ROS levels were quantified with a microplate reader.

Lactate levels and glucose uptake levels in CRC cells were assessed using lactate (cat.no. A019-2-1) and glucose assay kits (cat.no. F006-1-1), respectively, according to the manufacturer’s instructions (Nanjing Jiancheng Bioengineering Institute, Nanjing, China). Briefly, 200 µL culture medium per well of 96-well plates were used to seed 2 $\times$ 10⁴ cells. Following 48 h of culture, the supernatant culture medium was collected, and lactate and glucose levels were assessed using a colorimetric assay according to the manufacturer’s instructions. Lactate production rates and glucose consumption were normalized to cell count to ensure accurate sample comparisons.

Intracellular ATP levels were measured using an ATP assay kit (S0026, Beyotime, shanghai, China). Briefly, after collecting the cells, they were lysed with 200 µL lysis buffer and centrifuged at 12,000 rpm at 4 °C; the supernatant was removed as the test sample. Subsequently, 100 µL ATP detection reagent and 10 µL test sample were mixed following the manufacturer’s instructions, and the luminescence was measured using a NanoDrop spectrophotometer (ND-2000, Thermo Fisher Scientific, Waltham, MA, USA). A standard curve was used to calculate the relative ATP levels, which then were normalized to cell number or protein concentration to ensure accurate comparisons.

R was used as the programming language for data analyses. Two groups were compared using the unpaired Student’s t-test. One-way analysis of variance and Tukey’s post hoc test were used for multiple group comparisons. p 0.05 was considered statistically significant. Each experiment, unless otherwise noted, was carried out in triplicate.

Initially, we conducted differential expression analysis on the GSE47074 and TCGA-COAD datasets. Fig. 1A shows 1521 upregulated and 1171 downregulated DEGs in the TCGA-COAD dataset, whereas Fig. 1B shows 568 upregulated and 343 downregulated DEGs in the GSE47074 dataset. Subsequently, the DEGs in the two datasets were analyzed, identifying 319 intersecting upregulated DEGs and 171 downregulated DEGs (Fig. 1C). Subsequently, the intersection genes were further explored by PPI network analysis. Cytoscape illustrated the top 20 genes per MCC, MNC, and Degree algorithms. The MCC network contained 20 points and 190 edges (Fig. 1D), and the MNC network and Degree network contained 20 points and 189 edges (Fig. 1E,F). The top 20 genes of the three algorithms were analyzed again, and 15 essential CRC genes were obtained (Fig. 1G).

Fig. 1.

Selection of differentially expressed genes and PPI network analysis based on TCGA-COAD and GSE47074 datasets to screen essential intersection genes. (A) Volcano plot of DEGs in the TCGA-COAD dataset, showing 1521 upregulated and 1171 downregulated DEGs. (B) Volcano plot of DEGs in the GSE47074 dataset, with 568 upregulated and 343 downregulated DEGs. (C) Cross-analysis Venn diagram of 1521 upregulated DEGs of TCGA-COAD, 1171 downregulated DEGs of TCGA-COAD, 568 upregulated DEGs of GSE47074, and 343 downregulated DEGs of GSE47074. (D–F) PPI network analysis of intersecting genes using the MCC (D), MNC (E), and Degree (F) algorithms, visualizing the top 20 genes in each network. (G) Venn diagram showing the overlapping regions of the three different network centers of MCC, MNC, and Degree, showing 15 essential intersection genes. TCGA-COAD, The Cancer Genome Atlas-Colon Adenocarcinoma; DEGs, differentially expressed genes; PPI, protein-protein interaction; MCC, maximum correlation criterion; MNC, maximum neighborhood component.

We performed LASSO Cox regression analysis for 15 intersecting genes to identify the critical genes connected with COAD prognosis. The LASSO coefficient spectrum of the candidate genes is shown in Fig. 2A. The optimal value of the regularization parameter ($\lambda$) was obtained via 10-fold cross-validation based on the minimum partial likelihood deviation (Fig. 2B). Six genes (cyclin A2 [CCNA2], DNA topoisomerase II alpha [TOP2A], exonuclease 1 [EXO1], AURKA, AURKB, and BUB1 mitotic checkpoint serine/threonine kinase B [BUB1B]) were selected to construct the predictive risk model. Next, each patient’s risk score was determined by calculating the expression of these six genes. The median risk score divided patients into high- and low-risk categories. The risk scores, clinical survival time, and gene expression z-scores of all samples were visualized, and the findings revealed that the clinical survival time of high-risk patients was considerably worse (Fig. 2C). KM survival analysis showed that individuals at high risk had a substantially lower overall survival (OS) than those at low risk (log-rank p = 0.0433, HR = 1.452, 95% CI: 1.011, 2.084; Fig. 2D). In addition, the ROC curve was used to evaluate the predictive performance of the risk model, with 1-, 3-, and 5- survival rates yielding area under the curve (AUC) values of 0.629, 0.575, and 0.592, respectively (Fig. 2E).

Fig. 2.

Construct a prognostic-related risk model through LASSO regression analysis. (A) Processes of LASSO Cox model fitting. Each curve represents a CRC-related gene. (B) Selection of the optimal $\lambda$ value in the LASSO model using 10-fold cross-validation. The red dots represent the partial likelihood of deviance for each value of $\lambda$, with the vertical dotted line indicating the optimal $\lambda$ corresponding to the minimum deviance. (C) Distribution of risk scores in the high- and low-risk groups. The top panel displays the risk scores, the middle panel indicates the survival status (alive or dead), and the bottom panel shows a heatmap of the z-scores of gene expression levels for the six selected genes. (D) KM survival curves comparing OS between the high- and low-risk groups. Patients in the high-risk group had significantly worse OS (log-rank p = 0.0433, HR = 1.452, 95% CI: 1.011, 2.084). The median survival time for the high-risk group was 5.6 years. (E) Time-dependent ROC curves for the predictive model, showing AUC values of 0.629, 0.575, and 0.592 for 1-, 3-, and 5-year survival predictions, respectively. CRC, colorectal cancer; LASSO, least absolute shrinkage and selection operator; COAD, colon adenocarcinoma; KM, Kaplan-Meier; OS, overall survival; HR, hazard ratio; CI, confidence interval; ROC, receiver operator characteristic; AUC, area under the curve; CCNA2, cyclin A2; TOP2A, DNA topoisomerase II alpha; EXO1, exonuclease 1; BUB1B, BUB1 mitotic checkpoint serine/threonine kinase B.

TCGA-COAD and GSE47074 datasets were used to examine the expression of these six genes (BUB1B, AURKB, AURKA, EXO1, TOP2A, and CCNA2), and it was discovered that all of the genes had greater expression levels in the tumor groups of both datasets (Fig. 3A,B). Among these, AURKB was an interesting gene due to its function in controlling mitosis, a critical stage of the cell cycle. Disruption of mitosis by DNA damage can severely impair cell division, and aerobic glycolysis is known to supply the energy necessary for the DDR [23]. Given its potential involvement in the aerobic glycolysis of CRC cells, AURKB was determined as the core gene in this investigation, warranting further investigation into its role in CRC progression.

Fig. 3.

Expression of six prognosis-related genes. (A) Box plot of the expression levels of six prognostic genes (BUB1B, AURKB, AURKA, EXO1, TOP2A, and CCNA2) in COAD versus normal tissues in TCGA-COAD dataset. (B) Box plot of the expression levels of six prognostic genes (BUB1B, AURKB, AURKA, EXO1, TOP2A, and CCNA2) in normal and tumor samples in the GSE47074 dataset. *p 0.05, *** p 0.001. TCGA-COAD, The Cancer Genome Atlas-Colon Adenocarcinoma.

To investigate the function of the hub gene AURKB in CRC, qPCR was carried out to assess AURKB expression in NCM460 cells and three CRC cell lines. The findings revealed that the AURKB mRNA levels in the three CRC cells were significantly higher than those in normal colon epithelial cells, of which the expression of AURKB in LoVo and HCT-116 cells was the highest (Fig. 4A). This finding was corroborated by Western blot analysis, as illustrated in Fig. 4B,C. Subsequently, the expression of AURKB was knocked down in CRC cells, and the findings demonstrated a significant reduction in expression levels following AURKB knockdown compared to the control group (Fig. 4D–F). CCK-8 assays showed that the proliferation of CRC cells was reduced following AURKB silencing (Fig. 4G,H). A transwell assay was carried out to assess the invasion and migration of LoVo and HCT-116 cells following AURKB knockdown. The results showed that knockdown of AURKB significantly inhibited the metastasis of CRC cells (Fig. 4I,J).

Fig. 4.

Knockdown of AURKB inhibits CRC cell proliferation, invasion, and migration. (A) The qPCR data of AURKB mRNA expression in NCM450, HCT-116, LoVo, and SW480 cells. (B,C) Western blot analysis of AURKB protein expression levels in NCM450, HCT-116, LoVo, and SW480 cells. (D) qPCR data of AURKB mRNA expression in HCT-116 and LoVo cells after AURKB knockdown. (E,F) Western blot analysis of AURKB protein expression levels in HCT-116 and LoVo cells after AURKB knockdown. (G,H) CCK-8 assay detected the proliferation of HCT-116 (G) and LoVo (H) cells on days 0–4 after knockdown of AURKB. The X-axis is time (days), and the Y-axis is the optical density value when the absorbance was 450 nm. (I,J) Transwell assay detected the invasion and migration abilities of HCT-116 (I) and LoVo. Scale bar: 50 µm. (J) Cells after AURKB knockdown. Scale bar: 50 µm. *p 0.05. qPCR, quantitative PCR; CRC, colorectal cancer; CCK-8, cell counting kit-8; AURKB, aurora kinase B; si-NC, small interfering RNA negative control.

After AURKB knockdown, flow cytometry results showed that the proportion of cells in the G1 phase significantly increased, whereas the proportion of cells in the S phase significantly decreased. These results suggest that AURKB knockdown blocked cells in the G1 phase by inhibiting the G1/S transition (Fig. 5A,B). The mRNA expression of key cell cycle regulators, including cyclin D1, p27, and p21, were further examined by qPCR. The results showed that AURKB knockdown significantly downregulated cyclin D1 and upregulated p27 and p21 expression in CRC cells (Fig. 5C,D). Western blot analysis confirmed these findings at the protein level, showing decreased protein expression of cyclin D1 and increased p27 and p21 expression after AURKB knockdown in both cell lines (Fig. 5E–G). These findings indicate that AURKB knockdown causes G1 arrest in CRC cells, possibly through downregulating cyclin D1 and upregulating cell cycle inhibitors p27 and p21.

Fig. 5.

Effects of AURKB knockdown on cell cycle distribution and cell cycle regulators in CRC cells. (A,B) Flow cytometry analysis of cell cycle distribution in HCT-116 (A) and LoVo (B) cells transfected with control siRNA (si-NC) or AURKB-specific siRNA (si-AURKB-1 and si-AURKB-2). The bar graph on the right summarizes the percentage of cells in each cell cycle phase (G1, S, G2). (C,D) qPCR analysis of cyclin D1, p27, and p21 mRNA expression levels in HCT-116 (C) and LoVo (D) cells after AURKB knockdown. (E–G) Western blot analysis of protein expression of cycle-related proteins cyclin D1, p27, and p21 in HCT-116 and LoVo cells after AURKB knockdown. *p 0.05. qPCR, quantitative PCR; CRC, colorectal cancer; siRNA, small interfering RNA; AURKB, aurora kinase B; si-NC, small interfering RNA negative control.

The effect of AURKB knockdown on apoptosis of colon cancer cells was analyzed by flow cytometry, and an apoptosis rate of approximately 8% was observed (Fig. 6A,B). In addition, qPCR and western blotting were conducted to evaluate the expression changes of key apoptotic markers in CRC cells. Consistent with the flow cytometry results, knockdown of AURKB enhanced the expression of caspase-3 and Bax and decreased the expression of Bcl-2 (Fig. 6C–G). We observed significant changes in the expression patterns of apoptosis marker genes. This discrepancy between the relatively low apoptosis rate and the significant changes in expression of apoptosis-associated genes may stem from the complexity of the apoptosis pathway. Apoptosis is a complex process that involves multiple signaling pathways and feedback mechanisms. Knockdown of AURKB may have affected multiple apoptosis-associated genes and proteins, but these changes did not all directly lead to cell death. In this process, some genes may play a regulatory role mainly in the apoptotic pathway rather than directly inducing cell death. This complex regulatory network may lead to discrepancies between significant changes in gene expression patterns and the apoptosis rates detected by flow cytometry. Our findings suggest that knockdown of AURKB induces apoptosis in CRC cells by regulating the expression of critical apoptotic regulators. Since si-AURKB-1 showed more pronounced effects in inhibiting cell growth and promoting apoptosis, we selected it for subsequent analyses.

Fig. 6.

AURKB knockdown induces apoptosis in CRC cells. (A,B) Flow cytometry analysis of apoptosis in HCT-116 (A) and LoVo (B) cells transfected with control siRNA (si-NC) or AURKB-specific siRNA (si-AURKB-1 and si-AURKB-2). The bar graph on the right summarizes the percentage of apoptotic cells in each group. (C,D) qPCR data of mRNA expression of apoptosis-related markers Bax, Bcl-2, and caspase-3 in HCT-116 and LoVo after AURKB knockdown. (E–G) Western blot analysis of protein expression of apoptosis-related proteins Bax, Bcl-2, and caspase-3 in HCT-116 and LoVo cells after AURKB knockdown. *p 0.05. qPCR, quantitative PCR; CRC, colorectal cancer; AURKB, aurora kinase B; siRNA, small interfering RNA; Bcl-2, B-cell lymphoma 2; Bax, Bcl-2-associated X protein.

To assess the effect of AURKB knockdown on oxidative stress in CRC cells, we used detection kits to measure LDH, ROS, MDA, and SOD expression levels in LoVo and HCT-116 cells. As shown in Fig. 7A, knockdown of AURKB significantly increased LDH activity in CRC cells, indicating increased cell membrane damage. Similarly, ROS levels were markedly elevated following AURKB knockdown in both cell lines (Fig. 7B), suggesting increased oxidative stress. Additionally, MDA levels, a marker of lipid peroxidation, were significantly higher in AURKB knockdown cells than controls (Fig. 7C). Conversely, SOD activity, which serves as an antioxidant defense mechanism, was significantly reduced in cells with AURKB knockdown (Fig. 7D). These findings indicate that AURKB knockdown in CRC cells leads to increased oxidative stress, cellular damage, and lipid peroxidation, which may be involved in the induced apoptosis.

Fig. 7.

Knockdown of AURKB affects oxidative stress in CRC cells. (A) Relative LDH activity in HCT-116 and LoVo cells transfected with control siRNA (si-NC) or AURKB-specific siRNA (si-AURKB-1). (B) Relative ROS levels in HCT-116 and LoVo cells following AURKB knockdown. (C) Relative MDA levels in HCT-116 and LoVo cells after AURKB knockdown. (D) Relative SOD levels in HCT-116 and LoVo cells after AURKB knockdown. *p 0.05. CRC, colorectal cancer; LDH, Lactate dehydrogenase; ROS, Reactive oxygen species; MDA, Malondialdehyde; SOD, Superoxide dismutase; AURKB, aurora kinase B; siRNA, small interfering RNA; si-NC, small interfering RNA negative control.

Through kit testing, it was found that AURKB knockdown significantly reduced ATP production, glucose uptake, and lactate production in CRC cells compared with the control group (Fig. 8A–C). qPCR was conducted to assess the expression levels of glycolysis-related genes (PKM2, LDHA, HK2, GLUT1) in CRC cells. Compared to the control group, the expression of glycolysis-related genes was significantly downregulated after AURKB knockdown (Fig. 8D,E). Western blot analysis confirmed that AURKB knockdown led to decreased protein expression of PKM2, LDHA, HK2, and GLUT1 in CRC cells (Fig. 8F–H). These results indicate that AURKB knockdown impairs glycolysis and energy production in CRC cells by downregulating critical glycolytic enzymes and glucose transporters, thereby decreasing ATP levels.

Fig. 8.

Effects of AURKB knockdown on glycolysis and ATP production in CRC cells. (A) Relative lactate production in HCT-116 and LoVo cells transfected with control siRNA (si-NC) or AURKB-specific siRNA (si-AURKB-1). (B) Relative glucose uptake in HCT-116 and LoVo cells following AURKB knockdown. (C) ATP levels in HCT-116 and LoVo cells after AURKB knockdown. (D,E) qPCR data of mRNA expression of glycolysis-related genes PKM2, LDHA, HK2, and GLUT1 in HCT-116 and LoVo cells after AURKB knockdown. (F–H) Western blot analysis of protein expression of glycolysis-related proteins PKM2, LDHA, HK2, and GLUT1 in HCT-116 and LoVo cells after AURKB knockdown. *p 0.05. CRC, Colorectal cancer; qPCR, Quantitative PCR; AURKB, Aurora kinase B; siRNA, Small interfering RNA; PKM2, Pyruvate kinase M2; LDHA, Lactate dehydrogenase; HK2, Hexokinase 2; GLUT1, Glucose transporter 1.

The GEPIA database analyzed the expression correlation between AURKB and MAD2L2. Bioinformatics analysis revealed a significant positive relationship between the expression levels of AURKB and MAD2L2, with a correlation coefficient of 0.415 (Fig. 9A). The connection between AURKB and MAD2L2 was further elucidated by co-immunoprecipitation experiments, which further confirmed that AURKB can interact with MAD2L2 in LoVo and HCT-116 cells (Fig. 9B). Subsequently, qPCR was used to detect the significant overexpression efficiency of MAD2L2 in CRC cells, and Western blotting experimental results also demonstrated effective knockdown (Fig. 9C–E). The CCK-8 assay demonstrated that MAD2L2 overexpression significantly reversed the proliferation of CRC cells inhibited by AURKB knockdown (Fig. 9F,G). In addition, Transwell experiments performed in CRC cells showed that MAD2L2 overexpression rescued the inhibitory effects of AURKB knockdown on cell behavior (Fig. 9H,I).

Fig. 9.

MAD2L2 overexpression rescues the inhibitory effects of AURKB knockdown on CRC cell proliferation, migration, and invasion. (A) Correlation analysis between AURKB and MAD2L2 expression in CRC detected by the GEPIA database. (B) Co-IP demonstrated the interaction of AURKB and MAD2L2 in HCT-116 and LoVo cells. (C) qPCR data of MAD2L2 mRNA expression in HCT-116 and LoVo after MAD2L2 overexpression. (D,E) Western blot analysis of MAD2L2 protein expression levels in HCT-116 and LoVo cells after MAD2L2 overexpression. (F,G) CCK-8 assay was used to detect the proliferation of control, si-AURKB-1, and si-AURKB-1+over-MAD2L2 treated HCT-116 and LoVo cells from day 0 to day 4. (H,I) Transwell assay detected the invasion and migration abilities of Control, si-AURKB-1, and si-AURKB-1+over-MAD2L2 treated HCT-116 and LoVo cells. Scale bar: 50 μm. *p 0.05. CRC, colorectal cancer; Co-IP, co-immunoprecipitation; qPCR, quantitative PCR; CCK-8, cell counting kit-8; MAD2L2, mitotic arrest deficient 2 like 2; GEPIA, gene expression profiling interactive analysis.

To understand the impact of MAD2L2 overexpression on the CRC cell cycle and DDR in the context of AURKB knockdown, we evaluated the expression of key cell cycle proteins (cyclin D1, p21, and p53) and DNA damage marker H2A.X in CRC cells. qPCR analysis showed that AURKB knockdown significantly reduced cyclin D1 mRNA expression while increasing p53, p21, and H2A.X expression in CRC cells (Fig. 10A,B). However, overexpression of MAD2L2 partially reversed these effects, resulting in the restoration of cyclin D1 levels and decreased expression of p53, p21, and H2A.X. Western blot analysis confirmed these findings at the protein level (Fig. 10C–E). These results suggest that MAD2L2 overexpression can mitigate the impact of AURKB knockdown on DDR and cell cycle progression, highlighting the potential compensatory mechanism in CRC cells. We also observed a decrease in the mRNA and protein expression of MAD2L2 after knockdown of AURKB, suggesting that AURKB may have regulatory effects on the expression level of MAD2L2. However, we also found that overexpression of MAD2L2 did not restore the decrease in AURKB mRNA and protein expression caused by AURKB knockdown. This finding suggests that MAD2L2 may not be directly involved in regulating AURKB expression despite the effect of AURKB on MAD2L2 expression.

Fig. 10.

MAD2L2 overexpression mitigates the effects of AURKB knockdown on cell cycle regulators and the p53 DDR pathway. (A,B) qPCR was used to detect the mRNA expression levels of AURKB, MAD2L2, cyclin D1, p53, p21, and H2A.X in the control group, si-AURKB-1, si-AURKB-1+over-MAD2L2 treated HCT-116 and LoVo cells. (C–E) Western blot analysis was used to detect the protein expression levels of AURKB, MAD2L2, cyclin D1, p53, p21, and H2A.X in the control group, si-AURKB-1, si-AURKB-1+over-MAD2L2 treated HCT-116 and LoVo cells. * p 0.05. qPCR, quantitative PCR; DDR, DNA damage response; AURKB, aurora kinase B; MAD2L2, mitotic arrest deficient 2 like 2; H2A.X, H2A histone family member X.

Next, ATP levels, glucose uptake, and lactate production in CRC cells were assessed. In contrast to the control, AURKB knockdown inhibited ATP levels, lactate production, and glucose uptake, while MAD2L2 overexpression reversed this inhibitory phenomenon (Fig. 11A–C). The expression of key glycolytic enzymes at the mRNA and protein levels was further evaluated. qPCR analysis showed that AURKB knockdown significantly reduced the mRNA levels of these glycolysis-related genes in CRC cells (Fig. 11D,E). Western blot analysis showed that PKM2, LDHA, HK2, and GLUT1 protein levels were significantly reduced after AURKB knockdown (Fig. 11F). Notably, MAD2L2 overexpression reversed these effects and restored PKM2, LDHA, HK2, and GLUT1 levels in AURKB knockdown cells (Fig. 11G,H). These results indicate that MAD2L2overexpression can restore glycolytic activity and ATP production impaired by AURKB knockdown, highlighting its role in maintaining CRC cell metabolism.

Fig. 11.

Effects of MAD2L2 overexpression on glycolysis and ATP production in AURKB knockdown CRC cells. (A) Lactate production levels in control, si-AURKB-1, si-AURKB-1+over-MAD2L2 treated HCT-116 and LoVo cells. (B) Glucose uptake levels in control, si-AURKB-1, si-AURKB-1+over-MAD2L2 treated HCT-116 and LoVo cells. (C) ATP levels in control, si-AURKB-1, si-AURKB-1+over-MAD2L2 treated HCT-116 and LoVo cells. (D,E) qPCR was used to detect the mRNA expression levels of glycolysis-related genes PKM2, LDHA, HK2, and GLUT1 in the control group, si-AURKB-1, si-AURKB-1+over-MAD2L2 treated HCT-116 and LoVo cells. (F–H) Western blot analysis was used to detect the protein expression levels of glycolysis-related proteins PKM2, LDHA, HK2, and GLUT1 in the control group, si-AURKB-1, si-AURKB-1+over-MAD2L2 treated HCT-116 and LoVo cells. *p 0.05. qPCR, quantitative PCR; AURKB, aurora kinase B; MAD2L2, mitotic arrest deficient 2 like 2; PKM2, pyruvate kinase M2; LDHA, lactate dehydrogenase, HK2, hexokinase 2; GLUT1, glucose transporter 1.