GINS2 promotes the progression of human HNSCC by altering RRM2 expression

2.1TCGA gene expression datasets

We downloaded the transcriptome datasets and matched clinical data from the website of The Cancer Genome Atlas (TCGA) datasets (https://tcga-data.nci. nih.gov/tcga/). In the TCGA database, we found that there were only 40 pairs of HNSCC RNAseq datasets for tumor-adjacent histologically-normal tissues and tumor tissues (S1 Table), including the oral cavity, larynx, oral tongue, the base of the tongue, and floor of the mouth. The Trimmed Mean of the M-values method was used for data standardization. Quality control was carried out by calculating the biological coefficient of variation (BCV). In S2 Fig., we could see that the Normal (next to cancer) and Cancer samples were separated, indicating that the sample data of different batches had high stability and could be used for subsequent analysis. Since we would like to screen for genes whose expression level is related to stage, we classified these datasets according to the clinical stage of HNSCC: 2 pairs in stage I, 13 in stage II, 17 in stage III, and 8 in stage IV. Due to the small amount of data in stage I, we merged the datasets of stage I and stage II for further analysis. Finally, 40 pairs of samples were classified into 3 groups: stage I + stage II, stage III, stage IV. We set the following conditions to select our candidate genes: (a) Genes that are up-regulated in the 3 groups and whose expression levels increase with the stage; (b) Genes whose expression levels in “stage III” and ” stage IV” groups were up-regulated with the stage but were not differentially expressed in “stage I + stage II” group. Afterward, we screened out 485 candidate genes that met the above conditions. Then, we set the exclusion criteria to select our candidate genes: (i) a huge number of related literature; (ii) transmembrane protein; (iii) unknown functional genes containing open reading frames; (iv) relatively few types of related diseases; (v) no related carrier is convenient to use. Finally, we screened out 61 genes in HNSCC that were differentially expressed between tumor and tumor-adjacent histologically-normal tissues, and whose expression levels were related to clinical prognosis (S3 Table). Based on the data provided by the TCGA database, we found that among the 61 genes in HNSCC, 40 genes were highly expressed.

2.2HNSCC samples and immunohistochemistry (IHC)

From August 2011 to May 2015, we collected 70 HNSCC tumor samples from Jiangsu Cancer Hospital. The demographic and clinical data were searched from electric records. All patients had not received any treatment before this research. All samples used to detect the expression level of GINS2 were collected with informed consent. The research protocol was permitted by the Ethics committee of Jiangsu Cancer Hospital. All samples were proven to be HNSCC pathologically. There were 17 cases of Nasopharyngeal cancer, 21 hypopharyngeal cancer, 18 laryngeal cancer, 14 tongue cancer. We performed GINS2 IHC staining on all samples: Fix the tissue using 4% paraformaldehyde, the sections were incubated with primary antibody at 4^∘C overnight, immunostain the DAB chromogen, followed by nuclear staining with hematoxylin. Two pathologists were blinded to the patients’ clinical information. The histopathological features were assessed and the degree of immunostaining was scored. The scores were based on the ratio of positively stained tumor cells [0 (negative), 1 (1–25% positive), 2 (26–50% positive), 3 (51–75% positive) or 4 (76–100% positive)] and staining intensity [0 (no staining), 1 (weak, light-yellow), 2 (moderate, yellow-brown) or 3 (strong, brown)]. Multiplying the two parameters resulted in an immunoreactivity score of 0 to 12 (0, 1, 2, 3, 4, 6, 8, and 12). We used the median as the cutoff value for the high or low expression. Staining index ⩾ 6 means high expression, and < 6 means low expression.

2.3Cell culture

Human HNSCC cell lines CNE-2Z, 5-8F, FaDu, Tca-8113, Hep-2 were obtained from Jiangsu Cancer Hospital. All the cell lines were approved by the Institutional Ethical Review Board of Jiangsu Cancer Hospital. In addition, the cell lines we used were authenticated by STR profile. All cells were tested negative for mycoplasma contamination. The cell culture conditions were: RPMI 1640 medium (Gibco, NY) containing 5% fetal bovine serum (Gibco, USA), 95% air, and 5% CO2 at 37^∘C.

2.4Cell transfection

Lentiviral (sh-GINS2, shRRM2, shYAP1, shSKP2, sh-Control, oe-RRM2, oe-GINS2, oe-vector) were all synthesized and packaged by Shanghai Jikai Gene Technology Co., Ltd (Shanghai, China). Cells (1 × 10⁵ cells/well) were seeded in 6-well plates; Transfection can be performed when the confluence of cells reaches 60%–70% under the microscope. Dilute an appropriate amount of Lipofectamine 3000 in centrifuge tube 1 containing Opti-MEM medium and mix well. Add an appropriate amount of siRNA or plasmid to centrifuge tube 2 containing Opti-MEM medium and mix well. After mixing the liquids from centrifuge tubes 1 and 2 thoroughly, incubate for 10 minutes. Then, evenly drop the mixture onto the cell culture plate and continue cultivation with a constant temperature medium until ready for use. The infection efficiency was observed by detecting the expression of GFP 72 h after the lentivirus infection. The lentiviral vector used is GV493; the order of the vector components is hU6-MCS-CBh-gcGFP-IRES-puromycin; the plasmid backbone commonly used in overexpression plasmid construction was SV40-firefly_Luciferase-MCS. GINS2 shRNA (target sequence: 5’-GATTAACCTGAAACAAAGA-3’), RRM2 shRNA (target sequence: 5’-UGAUCUUGCAGAGAUGAACAG-3’), YAP1 shRNA (target sequence: 5’-CUGCCACCAAGCUAGAUAATT-3’), SKP2 shRNA (target sequence: 5’-ATCTTAGCGGCTACAGAAAG-3’). The overexpression sequence is the coding region (CDS) of the corresponding gene.

2.5In vivo nude mouse xenograft tumor models

Sixteen female BALB/C nude mice (5–6 weeks), purchased from the Vital River Laboratory Animal Technology (Beijing, China), were maintained under specific pathogen-free conditions and applied for ectopic tumor construction. CNE-2Z cells (5 × 10⁶) transfected with Ctrl-shRNA or GINS2-shRNA were resuspended in 100 μL of sterile PBS and injected subcutaneously into the shoulder or back of the thigh of mice. Then tumor volume (V) was measured twice a week using the standard formula V = a*b2/2 (a = length, b = width). When the volume approached to 1500 mm³, we have obtained sufficient data and experimental results to gain a deeper understanding of tumor growth processes and treatment efficacy. Therefore, we terminate the experiment to avoid further burden and suffering to the mice, the mice were sacrificed by cervical dislocation, the xenograft tumors were excised and photographed. All animal experiments were authorized by the Animal Science Committee of the Animal Science of Nanjing, China.

2.6Quantitative real-time polymerase chain reaction (qRT-PCR)

Cells were collected in Trizol (Invitrogen, CA) to extract total RNA. Then, reverse transcription was carried out to generate cDNA using M-MLV Reverse Transcriptase (Promega, Madison, WI, USA). qRT-PCR reactions were conducted with the SYBR Green Master Mix kit (Takara, Japan) according to the manufacturer’s instructions. The following primers were used: GINS2, forward primer 5’-CAGAAATGTCGCCTGCTCC-3’ and reverse primer 5’-GGATTTCGTCTGCCTTCG-3’; GAPDH, forward primer 5’-TGACTTCAACAGCGACACCCA-3’ and reverse primer 5’-CACCCTGTTGCTGTAGCCAAA-3’.

2.7Western blot (WB) analysis

The method of detecting proteins from protein lysates has been described previously [12]. Primary antibodies were listed as follows: GINS2 (Sigma, USA), YAP1 (Cell Signaling Technology, USA), CCNE2 (Abcam, UK), MKI67 (Abcam, UK), SKP2 (Abcam, UK), RRM2 (Abcam, UK) and GAPDH (Santa-Cruz, USA).

2.8Cellomics ArrayScan infinity

Cells were seeded in triplicates into 96-well plates (2,000 cells per well), at 37^∘C, 5% CO₂ for 5 days evaluation. Load samples onto a suitable slide, ensuring proper controls and replicates. Set up the acquisition parameters, including exposure time, filter settings, and imaging modes. Fluorescent cell quantification was conducted under a Cellomics ArrayScan Infinity system to detect cell proliferation ability. Generate quantitative data and statistical analyses based on the image analysis results.

2.9Celigo cell counting assay

Cells were seeded into 96-well plates (2,000 cells per well), at 37^∘C, 5% CO₂ for 5 days, and measured daily by Celigo Imaging Cytometer (Nexcelom Bioscience, USA).

2.10MTT assay

Cells were seeded into 96-well plates (2,000 cells per well), at 37^∘C, 5% CO₂ for 5 days, and added MTT (20 μL; 5 mg/mL) (Genview, Australia) every day, and tested the absorbance at 490 nm by the microplate reader.

2.11Colony formation assay

Cells were seeded into 6-well plates (400 cells per well), at 37^∘C, 5% CO₂ for 14 days. The medium replacement was repeated every 2 or 3 days. Finally, cells were fixed with 4% paraformaldehyde, stained with Giemsa, counted, and photographed.

2.12Flow cytometric (FCM) analysis

Cells were seeded into 6-well plates (1 × 10⁵ cells per well), cultured at 37^∘C, 5% CO₂. Then, when the cell growth reached 85%, cells were harvested and stained by Annexin V-APC for apoptosis analysis.

2.13Microarray and bioinformatics

Total RNA in shCtrl and shGINS2 groups was extracted using Trizol. The RNA that passed the quality inspection would be further analyzed. P-value < 0.05 and a 2-fold change cutoff were selected. Upload the above differentially expressed genes to Ingenuity Pathway Analysis (IPA) for analysis. StarBase v3.0 was used to figure out the expression level of GINS2 in HNSCC patients, as well as the interaction between GINS2 and RRM2.

2.14Statistical analysis

Differences between groups were analyzed using Student’s t-test. The Kaplan-Meier method and log-rank test were used to estimate survival rates. All statistical analyses were performed with SPSS 19.0 software. Data were presented as the mean ± SDs. P< 0.05 was considered statistically significant.

3.1GINS2 was overexpressed in HNSCC and associated with clinical stage and poor prognosis

By analyzing the RNAseq datasets of HNSCC in the TCGA database, we have identified 40 highly-expressed genes in HNSCC for subsequent analysis. We knocked down the 40 highly-expressed genes and then verified the proliferation ability through Cellomics ArrayScan Infinity in CNE-2Z cells. Results indicated that the NC group proliferated normally, and the cell multiplication rate reached 6.21 on the 5th day. However, the proliferation of the positive control cell group significantly slowed down, and the multiplication rate was only 1.16 on the 5th day (S4 Fig.). Among the 40 highly expressed genes, those with Fold change ⩾ 2.0 were CSGALNACT2, IL1F9, TMEFF1, SPINK6, and GINS2 (Table 1, S5 Fig.). Knockdown of GINS2 inhibited proliferation most obviously with a highly Fold change of 6.73. Subsequently, we performed the quantitative analysis by IHC and found that GINS2 expression in cancer is different from that in benign tumors and inflammatory tissue (p= 0.010428) (S6 Table). GINS2 gene expression was significantly increased in cancer (S7 Fig.). Thus, the high expression of GINS2 might be important for HNSCC progression.

We performed IHC analysis on 70 HNSCC samples collected clinically to explore the clinical correlation between GINS2 expression and HNSCC progression. We noticed that the tissue expression level of GINS2 was related to the clinical stage in HNSCC patients (Table 2). With the increase of clinical-stage, the proportion of high expression of GINS2 in each group also increased. Moreover, IHC results revealed that GINS2 was almost non-existent in adjacent normal tissues, but it was highly expressed in HNSCC and was mainly located in the nucleus (Fig. 1A–H).

Figure 1.

High expression of GINS2 was correlated with poor clinical outcomes. (A) Negative staining of GINS2 (magnification 100×), (B) weak staining (100×), (C) moderate staining (100×), (D) strong staining (100×), (E) negative staining (400×), (F) weak staining (400×), (G) moderate staining (400×), (H) strong staining (400×). (I) Overall survival. GINS2, GINS complex subunit 2. P< 0.0001.

Subsequently, we used survival analysis to further clarify the impact of GINS2 overexpression on the prognosis of HNSCC patients. Through the survival curve, we found that the overall survival (OS) of patients with high GINS2 expression (n= 18) was obviously lower than patients with low expression (n= 52) (P< 0.0001) (Fig. 1I). Therefore, the above results indicated that HNSCC patients with high GINS2 expression are more likely to have a poor prognosis.

We further obtained GINS2 expression in HNSCC and normal samples from starBase v3.0 software. The results showed that the expression level of GINS2 in 502 cases of HNSCC was 7.38, while in 44 normal samples was only 2.06, with a fold change of 3.59 (P< 0.0001), which was consistent with our above results that GINS2 was upregulated in HNSCC (Fig. 2A). Then we performed qRT-PCR to obtain the GINS2 expression level in some HNSCC cells. We found that GINS2 mRNA was highly expressed (ΔCt ⩽ 12) in all 5 HNSCC cell lines (CNE-2Z, FaDu, Tca-8113, Hep-2, and 5-8F) (Table 3, Fig. 2B). Next, CNE-2Z, 5-8F, and FaDu cell lines were used in the following experiments.

Figure 2.

GINS2 expression and transfection efficiency determination in HNSCC cell lines. (A) GINS2 was up-regulated in HNSCC patients. The expressions of GINS2 in HNSCC (n= 502) and normal samples (n= 44) were found from the starBase v3.0 database with the cancer expression level of 7.38 and the normal expression level of 2.06 (fold change = 3.59, P= 1.2e-29). (B) GINS2 mRNA expression in HNSCC CNE-2Z, FaDu, Tca-8113, Hep-2, and 5-8F cell lines. (C, D) GINS2 expression was inhibited by shRNA. The mRNA expressions of GINS2 were determined by qRT-PCR, and the protein expressions were determined by WB in CNE-2Z, 5-8F, and FaDu cells. (D) The over-expressions efficiency of GINS2 was confirmed by qRT-PCR in CNE-2Z, 5-8F, and FaDu cells. Values were expressed as mean ± standard deviation. n= 3, ^****P < 0.0001. HNSCC, head and neck squamous cell carcinomas; WB, western blot; qRT-PCR, quantitative real-time polymerase chain reaction.

Figure 3.

Effects of GINS2 knockdown or overexpression on HNSCC cell proliferation and apoptosis. (A) Celigo cell counting assay results of GINS2 knockdown on cell growth. (B) MTT assay results of GINS2 knockdown on cell proliferation. (C) MTT assay results of GINS2 overexpression on cell proliferation. (D) Colony formation assay results of GINS2 knockdown on HNSCC cells. (E) Colony formation assay results of GINS2 overexpression on HNSCC cells. (F) Apoptosis assay results of GINS2 knockdown on HNSCC cells. ^{*P <} 0.05, ^**P < 0.01, ^***P < 0.001, ^****P < 0.0001.

Figure 4.

Knockdown of GINS2 inhibited the MYC signaling pathway and downregulated RRM2. (A) Heatmap showed the differentially expressed genes between shCtrl and shGINS2 groups. Up-regulated genes are red, and down-regulated genes are green. (B, C) Histogram and Heatmap showed the functional classification of GINS2 regulatory genes. (D, E) The network of molecules on MYC and its pathway. (F) The mRNA expressions of 12 genes (PPKACB, DDX18, RRM2, SKP2, KRAS, MKI67, CCNE2, DDX5, EIF4A1, MTBP, YAP1, and NUDT21) determined by qRT-PCR. (G) The protein expressions of RRM2, SKP2, MKI67, CCNE2, and YAP1 were determined by WB. ^{*P <} 0.05, ^**P < 0.01, ^***P < 0.001, ^****P < 0.0001. IPA, Ingenuity Pathway Analysis.

Figure 5.

Effects of RRM2, SKP2, and YAP1 knockdown on cell growth, and the relationship of RRM2 and GINS2 in HNSCC. (A) RRM2, SKP2, and YAP1 were knocked down successfully in FaDu cells. The mRNA expressions were determined by qRT-PCR. (B) Celigo cell counting assay results of RRM2, SKP2, and YAP1 knockdown on cell growth in FaDu cells. (C) RRM2 was knocked down successfully in CNE-2Z cells. The RRM2 mRNA expression was determined by qRT-PCR. (D) Celigo cell counting assay results of RRM2 knockdown on cell growth in CNE-2Z cells. (E) A positive correlation was found between GINS2 and RRM2. ^***P < 0.001, ^****P < 0.0001.

Figure 6.

RRM2 rescued the effect of GINS2 in HNSCC cells. (A) The protein expressions of RRM2 were determined by WB in CNE-2Z and FaDu cells. (B) MTT results of the rescue assay. (C) Apoptosis results of the rescue assay. ^****P < 0.0001.

Figure 7.

Knockdown of GINS2 suppresses tumor growth in vivo. (A) Xenograft models in nu/nu mice were generated using CNE-2Z cells transfected with Ctrl-shRNA (n= 6) or GINS2-shRNA (n= 6). (B) The volumes of subcutaneous tumors in indicated mice were measured for 34 days. (C) The average weights of excised tumors are shown. Data represents the mean ± SD (n= 6); ^*P< 0.05; ^**P< 0.01; ^***P< 0.001.

3.2GINS2 was knocked down or overexpressed in HNSCC cells

We first knocked down GINS2 in CNE-2Z, 5-8F, and FaDu cells. According to the results of qRT-PCR (P< 0.0001) (Fig. 2C) and WB analysis (Fig. 2D), GINS2 expression levels in shGINS2 groups were much lower than shCtrl groups. On the other hand, we also overexpressed GINS2 in HNSCC cells successfully, as shown in Fig. 2E (P< 0.0001).

3.3GINS2 promoted the proliferation of HNSCC cells

To explore the role of GINS2 expression in the growth of HNSCC cell lines, we conducted a series of proliferation-related experiments. The results of the Celigo cell counting assay indicated that knocking down GINS2 lowered the total number of cells as well as cell growth rate in CNE-2Z, 5-8F, and FaDu cell lines (P< 0.0001) (Fig. 3A). MTT assay results showed that knocking down GINS2 inhibited HNSCC cell proliferation (P< 0.0001) (Fig. 3B). On the contrary, as shown in Fig. 3C, when we overexpressed GINS2, the proliferation of HNSCC cells was enhanced (P< 0.01). Similarly, the colony formation assay results told us that compared with the control group, the ability of the cells to form colonies was markedly reduced in the shGINS2 group (P< 0.01) (Fig. 3D). Conversely, up-regulating the expression of GINS2 promoted the clonal formation of cells (P< 0.05) (Fig. 3E). Thus, we have demonstrated that GINS2 accelerated the tumorigenicity of HNSCC cells in vitro.

3.4GINS2 knockdown induced apoptosis of HNSCC cells in vitro

Tumor progression is generally accompanied by the occurrence of anti-apoptosis. The results of apoptosis assay in CNE-2Z, 5-8F, and FaDu cells revealed that the apoptosis rates of shGINS2 groups were higher than the control groups (P< 0.0001) (Fig. 3F), which demonstrated that GINS2 knockdown enhanced the apoptosis of HNSCC cells.

3.5GINS2 inhibition downregulated the expression of RRM2

Since GINS2 affected proliferation and apoptosis of HNSCC as demonstrated above, we checked the gene expression profiles of the GINS2 knockdown group and the control group in CNE-2Z cells, hoping to explore its possible mechanism. Compared with the shCtrl group, the shGINS2 group had 169 genes up-regulated, and 449 genes down-regulated (Fig. 4A). Then, we selected the top 500 genes whose expression levels changed significantly after GINS2 knockdown and used IPA to conduct functional classification. The analysis results prompted that these genes were notably enhanced in many biological functions, such as cell proliferation, organismal death, congenital anomaly of the musculoskeletal system, and viral infection (Fig. 4B and C). Subsequently, we again used IPA to screen the regulatory factors related to GINS2, and the data validated that the z-score for MYC was -2.158 (P< 0.01). In IPA, z-score > 2 referred to significantly activated, and z-score <-2 referred to significantly inhibited. Thus, the MYC pathway was significantly inhibited in the shGINS2 group of CNE-2Z cells.

The MYC pathway is a well-known signaling pathway closely related to cancer. The pathway contains many genes (Fig. 4D), which constitute a complex regulating network map (Fig. 4E). Then, the expression of 12 genes (PPKACB, DDX18, RRM2, SKP2, KRAS, MKI67, CCNE2, DDX5, EIF4A1, MTBP, YAP1, and NUDT21) in the pathway were verified by qRT-PCR. Except for NUDT21 (P= 0.2945), all other 11 genes were remarkably down-regulated by GINS2 inhibition in the mRNA level (P< 0.05) (Fig. 4F). As well-known genes, RRM2, SKP2, MKI67, CCNE2, and YAP1 were selected for the next WB analysis. As presented in Fig. 4G, all 5 genes in shGINS2 groups were remarkably down-regulated. In agreement, the results were in agreement with gene expression profiling, implying that GINS2 might exert an influence on HNSCC cell proliferation and apoptosis by regulating RRM2, SKP2, MKI67, CCNE2, and YAP1.

3.6RRM2, positively correlated with GINS2, its knockdown suppressed cell proliferation in HNSCC

To prove our conjecture, we chose to knock down RRM2, SKP2, and YAP1 in FaDu cells via lentivirus infection. The efficiency of transfection was confirmed by qRT-PCR (P< 0.0001) (Fig. 5A). The results of the Celigo cell counting assay indicated that knocking down RRM2 (P< 0.0001), SKP2 (P< 0.001), and YAP1 (P< 0.001) dropped the total cell number as well as cell growth rate in FaDu (Fig. 5B). Among them, the RRM2 knockdown group showed the most significant growth inhibition. Then, we knocked down RRM2 in CNE-2Z which was confirmed by qRT-PCR (P< 0.0001) (Fig. 5C), and repeated the above Celigo cell counting assay. Again, knocking down RRM2 significantly reduced the cell growth rate in CNE-2Z cells (P< 0.001) (Fig. 5D). Based on the foregoing results, after knockdown of GINS2, the expression of RRM2 was also down-regulated; and down-regulation of RRM2 also inhibited proliferation, we speculated that the two are likely to be positively correlated. We retrieved 502 HNSCC tissues through the starBase v3.0 software and found that the expression levels of the two were indeed positively correlated (r= 0.583, P< 0.0001) (Fig. 5E).

3.7RRM2 partially rescued the effect of GINS2 in HNSCC cells

Later, CNE-2Z and FaDu cells were transfected with shGINS2, or co-transfected with shGINS2 and oe-RRM2 in rescue assays. The transfection effect was tested by WB analysis (Fig. 6A). Essentially, in MTT analysis, the overexpression of RRM2 rescued the repression of shGINS2 on cell proliferation in CNE-2Z cells, as well as in FaDu Cells (Fig. 6B). Meanwhile, in apoptosis assay, the overexpression of RRM2 also partially abolished shGINS2-mediated enhancement of cell apoptosis in CNE-2Z and FaDu cells (Fig. 6C).

3.8GINS2 knockdown inhibited in vivo tumor growth

To extend the in vitro observations, we investigated the effects of GINS2 knockdown on tumor growth in vivo. CNE-2Z cells that were transfected with either GINS2-shRNA or Ctrl-shRNA lentivirus were injected into nude mice, and the tumor growth was monitored. Tumors formed by GINS2-silenced CNE-2Z cells were much smaller during the experimental period than the control tumor (Fig. 7A and B). Furthermore, the excised Ctrl-shRNA tumors weighed on an average of 4006 mg whereas GINS2-shRNA tumors averaged 1368 mg (Fig. 7C). Collectively, these results emphasize the in vivo promoting role of GINS2 in cancer progression.