The copy number data were obtained from our study. The tumor and the matched adjacent non-tumor samples were recruited from Shanxi Cancer Hospital of Shanxi Medical University. The patients were without preoperative chemotherapy, radiotherapy, and other treatments before operation, and written consent was obtained from all of them. Hematoxylin and eosin (H&E) staining was used to diagnose these tissues, and the diagnosis was performed by at least two pathologists independently. The ESCC individuals were staged according to the American Joint Commission for Cancer (AJCC)/International Union Against Cancer (UICC) TNM staging system (eighth edition). The study was approved by the Institutional Reviewing Board (IRB) and the Research Committee of Shanxi Medical University.

ESCC cell lines KYSE150, KYSE180, KYSE450, and TE-1 and immortal embryonic esophageal epithelium cell lines NE3 and HET-1A used in the research were purchased from the Cell Bank of Type Culture Collection of the Chinese Academy of Sciences. The cell line 293T was from our lab. The cell lines KYSE150, KYSE180, KYSE450, and TE-1 were cultured in HyClone™ RPMI-1640 medium, and the cell lines HET-1A and 293T were cultured in HyClone™ DMEM/High Glucose medium (GE Healthcare Life Sciences, HyClone Laboratories, Logan, UT, USA). The culture was with 10% fetal bovine serum (FBS; Gibco; Thermo Fisher Scientific, Inc., Waltham, MA, USA). The cell line NE3 was cultured in a 1:1 mixture of EpiLife medium (Cascade Biologics, Inc., Portland, OR, USA) and defined keratinocyte serum-free medium (dKSFM; Gibco; Thermo Fisher Scientific, Inc., Waltham, MA, USA). All of the cell lines were cultured at 37°C, 5% CO₂. The culture medium was replaced according to the cell state. Subculture was carried out when the cell fusion was about 80%–90%.

SiRNAs or plasmids were transfected into the cells at the logarithmic growth phase using Lipofectamine 2000 reagent (Invitrogen, Carlsbad, CA) according to the instructions of the manufacturer. For knockdown of endogenous CDCA7, we used vectors containing the sequence 5′-GCCCTCAGAGAATTCTGTGACTGAT-3′ (CDCA7-si1) and 5′-CATCCGTGACCCTTCCGCATATAAT-3′ (CDCA7-si2). These shRNAs were cloned into the vector pHBLV-U6-Scramble-ZsGreen-Puro vector. For stable overexpression, the coding sequence (CDS) region of the CDCA7 gene was cloned into pHBLV-CMV-MCS-3FLAG-EF1-ZsGreen-T2A-PURO. The recombinant plasmids and the packaging plasmids (Hanbio Biotechnology Co., Ltd., Shanghai, China) were co-transfected into 293T cells. The lentivirus supernatant was used to infect the KYSE150, KYSE450, and KYSE180 cell lines. The negative control was the corresponding empty vectors. The KYSE150 and KYSE450 knockdown stable cell lines and the KYSE180 overexpression stable cell line were screened out for 7–14 days with 1.0, 1.0, and 0.8 µg/ml puromycin (Invitrogen; Thermo Fisher Scientific, Inc.), respectively. The efficiency of knockdown and overexpression was determined by real-time PCR and Western blot assay. We used small interference RNA (siRNA) for CCNA2 knockdown, and the siRNA sequence information is as follows: CCNA2-si1, 5′-CTATGGACATGTCAATTGT-3′; CCNA2-si2, 5′-GAGTGTTAATGAAGTACCA-3′. The CDS of CDCA7 and CCNA2 genes was cloned into the pcDNA3.1 vector with a V5 tag and 6*His tag.

The MTT assay was performed using a 96-well plate with 5 × 10³ transfected cells each well and cultured for 24–120 h. A 20-µl MTT solution (5 mg/ml) was added to a 200-µl culture medium each well for 4 h at 37°C. The MTT formazan crystals that remained after removing the medium were then solubilized in dimethyl sulfoxide (DMSO) for 15–20 min. The absorbance was measured by a spectrophotometer at 490 nm to show the relative number of surviving cells in each well indirectly.

A total of 1,000 cells/well were seeded into six-well plates and incubated at 37°C and 5% CO₂ for 10–15 days. Polyformaldehyde (4%) was used to fix these cells and 1% crystal violet was used to stain these cells subsequently. The numbers of colonies containing more than 10 cells were counted.

Cells collected were fixed with 70% alcohol and stored overnight at −20°C. Propidium iodide (PI) was used to stain the collected cells according to the instructions of the manufacturer. The stained cells were analyzed using a flow cytometer (BD Company, USA).

KYSE150, KYSE180, and KYSE450 cells were transfected with CDCA7-V5 plasmid and empty vector, respectively. Formaldehyde (4%) was used to fix the cells for 10 min. BSA (1%) was used to incubated the cells for 1 h to block non-specific protein–protein interactions after permeabilized by 0.1% Triton X-100. The cells were incubated with the primary antibody rabbit anti-V5 (Abcam, Cambridge, UK, 2 µg/ml) overnight at 4°C. Alexa Fluor^® 594 goat anti-rabbit IgG antibody (Thermo Fisher, Carlsbad, USA, 1:1,000) was used for 30 min at room temperature after washing four times in PBS. DAPI at a concentration of 0.5 µg/ml was used to stain the cell nuclei.

KYSE150 cells were transfected with the V5-tagged CDCA7 plasmid for the chromatin immunoprecipitation (ChIP) assay. The assay was performed according to the instructions of the manufacturer (Millipore, Burlington, MA, USA). The DNA fragments were enriched by anti-V5 antibody (Abcam, Cambridge, UK), and the isotype IgG (Abcam, Cambridge, UK) was used as a negative control. CHIP-seq was performed by Novogene (Beijing, China). Screening and quality control of the CHIP-seq were based on standard protocol. The sequences of primers used for the amplification of CCNA2 genome regions containing a putative CDCA7 binding site are listed in Table S1 .

The cells were lysed for 1 h with RIPA buffer containing protease and phosphatase inhibitors (Thermo Fisher Scientific) on ice. The components of RIPA buffer are as follows: 1% Triton X-100, 50 mM Tris–HCl, pH 7.6, 150 mM NaCl, 1% sodium deoxycholate, and 0.1% SDS. The lysates were centrifuged at 12,000g at 4°C for 30 min, and the total protein concentrations of supernatant were determined by the Bradford method. Fifty micrograms of protein was separated by 10% SDS-PAGE and then transferred onto polyvinylidene fluoride (PVDF) membranes (Millipore, USA). The membrane was incubated with special antibodies, including CDCA7, CCND1, CCNA2, CCNE1, and GAPDH, at 4°C overnight. The IRDye 800CW secondary antibody (Abcam, Cambridge, UK) was used to detect the blot. A relative amount of protein was normalized to GAPDH level. The antibodies used in this experiment are shown as follows: CDCA7 (Sigma, USA), CCND1 (Proteintech, Rosemont, IL, USA), CCNA2 (Proteintech, Rosemont, IL, USA), CCNE1 (Proteintech, Rosemont, IL, USA), and GAPDH (Proteintech, Rosemont, IL, USA).

Total RNA of ESCC cells was purified using RNAiso plus (Takara, Dalian, China). Two micrograms of total RNA was used for complementary DNA (cDNA) synthesis using a PrimeScript^® RT reagent kit with gDNA Eraser (Takara). TB Green^® Premix Ex Taq^® II kit (Takara) was used in real-time PCR according to the instruction of the manufacturer. All real-time PCR reactions were performed in triplicate with an Applied Biosystems Step One Plus (ABI, Foster City, CA, USA). The relative expression levels of the target genes were normalized to endogenous GAPDH. Quantification of the expression levels of target genes was calculated using the 2^−ΔΔCt formula. The primers synthesized by Thermo Fisher are listed in Table S2 .

According to the results of ChIP-seq, we cloned the CDCA7 DNA-binding fragment with CCNA2 from the KYSE150 cell line genomic DNA. The cloned DNA fragment was constructed into the reporter plasmid of pGL3-promoter (Promega, Madison, WI, USA). Then, we divided this DNA fragment into four segments and constructed them into the pGL3-promoter vector. The different DNA fragments were cloned using the primers listed in Table S3 . Six motif sequences obtained from ChIP-seq were also constructed into pGL3-promoter vector. Cells (3 × 10⁴) were cultured in triplicate in 24-well plates for 12–24 h. Then, the pGL3 reconstruction reporter plasmids were transiently co-transfected with the pRL-TK plasmid into KYSE150 and KYSE150 knockdown cells using Lipofectamine 2000 reagent (Invitrogen, Carlsbad, CA). After transfection for 48 h, luciferase and Renilla signals were measured according to the instruction of Dual-Luciferase Reporter Assay Kit (Promega, Madison, USA).

The effects of CDCA7 on tumorigenesis and growth in vivo were detected via mouse xenograft assay. We used 20 5- to 6-week-old female NU-Foxn1^nu nude mice (Vital River Laboratory Animal Technology Co., Ltd., Beijing, China) for the mouse xenograft assay. A total of 3 × 10⁶ KYSE150-NC cells or CDCA7-knockdown stable KYSE150 cells were used to inject into the right or left oxter of female NU-Foxn1^nu nude mice, respectively. Tumor size and weight were determined with calipers and balance twice a week. The mice were executed and the tumors were removed after 28 days. The formula V = (W ² × L)/2 was used to calculate the tumor volume. V is the tumor volume, W is the tumor width, and L is the tumor length. Tumor size was presented as mean ± standard deviation (SD).

The isolated xenograft tumor tissues were fixed using formalin and embedded by paraffin for immunohistochemical staining. Briefly, xylene and a series of grades of alcohol were used to deparaffinize and rehydrate these sections, and the sections were then soaked with 3% H₂O₂ 15 min. Sodium citrate buffer (pH 6.0) or Tris-EDTA buffer (pH 9.0) were used for antigen retrieval for 4 or 3 min in a pressure cooker, followed by incubation with primary antibody at 4°C overnight. The slides were incubated with second antibody at 37°C for 20 min after washing with PBS and then stained with DAB and counterstained with hematoxylin. The expression of CDCA7, Ki-67, and CCNA2 was quantitatively analyzed with Aperio Cytoplasma 2.0 software by immunohistochemistry. The antibodies used in this experiment are shown as follows: CDCA7 (Sigma, USA, 1:200 dilution), Ki-67 (Abcam, Cambridge, UK), and CCNA2 (Proteintech, Rosemont, IL, USA, 1:2,000 dilution).

The mRNA expression data and the clinical information of 179 ESCC tissues and paired non-tumor tissues by microarray analysis were from a previous study by Li et al. (26) and downloaded from the GEO database (GSE53625). The copy number data were obtained from our other study.

The copy number data of CDCA7 in varied cancer types, including ESCC, ECA, lung squamous cell carcinoma (LUSC), head and neck squamous cell carcinoma (HNSC) collected from the TCGA database, were downloaded via cBioPortal for Cancer Genomics (https://www.cbioportal.org/) (27, 28). The expression data of CDCA7 and CCNA2 in different cancer types such as ESCC, LUSC, HNSC were downloaded from TCGA via Xena Browser (https://xenabrowser.net/heatmap/).

Each of the experiment in the study was performed in triplicate, and data were presented as the mean ± SEM. Statistical Package for Social Science for Windows (SPSS, version 20.0; IBM Inc., USA) was used to analyze the experimental data. The means of two groups and more than two groups were compared using Student’s t-test and one-way ANOVA, respectively. P-value of <0.05 was considered to be statistically significant. GraphPad Prism software was used to analyze the correlations between CDCA7 and CCNA2 using non-parametric correlation (Spearman).

In our previous study, CDCA7 was identified as one of the copy number amplification genes in ESCC (8). Here, we analyzed the copy number amplification data from TCGA through cBioPortal and found that the copy number amplification of CDCA7 existed in various kinds of tumor. Its alteration frequency was much higher in ESCC than in other tumors ( Figure 1A ). Furthermore, we analyzed the mRNA expression data of 179 pairs of ESCC tumors and adjacent normal tissues via microarray analysis. The mRNA expression data and the clinical information of 179 ESCC tissues and paired non-tumor tissues by microarray analysis were from a previous study by Li et al. and downloaded the from the GEO database (GSE53625) (26). We observed that CDCA7 showed statistically higher expression levels in most of the individuals compared with that of normal tissues ( Figure 1B ). After analyzing the copy number amplification and expression of CDCA7 in 95 ESCC patients in the TCGA database, we found that there was a correlation between the expression of CDCA7 and the copy number amplification, indicating that the CDCA7 copy number amplification may cause to increase its expression ( Figure 1C ).

The cohort of 179 patients was divided into two groups according to the expression level of CDCA7. The top 25% of patients were defined as the patients with a higher level (named as CDCA7 _high) and the remaining 75% were defined as the patients with a lower level (named as CDCA7 _low) according to the expression level of CDCA7 from high to low. Then, we analyzed the correlation between the expression of CDCA7 and the clinical variables in ESCC. The results in Table 1 show that the expression of CDCA7 was related to the grade of ESCC patients (P = 0.0083). The patients with CDCA7 _high had a poor grade compared with the CDCA7 _low patients. Furthermore, the patients with CDCA7 _high had a worse survival than those with CDCA7 _low (log rank P = 0.01, Figure 1D ) using Kaplan–Meier survival analysis. The multivariate analysis showed that TNM stage [hazard ratio (HR) = 2.662, 95% CI: 1.593–4.449, P < 0.001], location (lower vs. upper) (HR = 2.718, 95% CI: 1.336–5.530, P = 0.006), age (HR = 1.657, 95% CI: 1.074–2.558, P = 0.022), and CDCA7 expression (HR = 1.999, 95% CI: 1.241–3.218, P = 0.004) were independent predictive factors for overall survival ( Figure 1E ). Furthermore, CDCA7 was related with the survival status in patients in the male group (P = 0.001), age <60 group (P = 0.02), drinking group (P < 0.001), smoking group (P = 0.014), T1+T2 group (P = 0.019), N0+N1 group (P = 0.026), and TNM stage = III group (P = 0.025) ( Figures S1, S2 ). Hence, we speculate the copy number amplification and high expression level of CDCA7 may promote the occurrence and development of ESCC.

To verify the biological roles of CDCA7 in ESCC, we first analyzed the mRNA and protein expression levels in immortal embryonic esophageal epithelium cell lines NE3 and HET-1A and ESCC cell lines including KYSE150, KYSE180, KYSE450, and TE-1 via quantitative real-time PCR (q-RTPCR) and Western blot ( Figure S3 ). In all these cell lines, we selected KYSE150 and KESE450 as relatively high endogenous CDCA7 level cell lines for knockdown experiments. KYSE180 was selected as low endogenous CDCA7 level cell line for overexpression. The efficiency of overexpression and knockdown were verified by Western blot, respectively ( Figures 2A, B ). Then, we detected the changes in cell phenotypes, including proliferation, colony formation, and cell cycle. The results showed that CDCA7 silencing significantly inhibited the ability of cell proliferation and colony formation in KYSE150 and KESE450 ( Figures 2C, E ), while overexpression of CDCA7 increased the ability of cell proliferation and colony formation markedly ( Figures 2D, F ). Meanwhile, the results of flow cytometry indicated that CDCA7 overexpression decreased the proportion of G1-phase cells and increased the proportion of G2 + M-phase cells ( Figure 2H ). On the contrary, CDCA7 silencing significantly increased the proportion of G1-phase cells and decreased the proportion of G2 + M-phase cells ( Figure 2G ).

To confirm the oncogenic role of CDCA7 in vivo, we established a subcutaneous transplantation tumor model in female NU-Foxn1nu nude mice using stable CDCA7-knockdown KYSE150 and KYSE150 cells. Four weeks later, tumors were stripped after the mice were sacrificed. The tumor growth rate of the KYSE150 group was significantly faster than that of the CDCA7-knockdown group ( Figure 2I ). The results found that the mean tumor volume of the CDCA7-knockdown group and the control group was 158.74 ± 24.83 and 488.41 ± 35.84 mm³, respectively (t-test, P < 0.001, Figure 2I ). The mean tumor weight of the control group and the CDCA7-knockdown group was 209.61 ± 108.84 and 54.23 ± 19.39 mg, respectively (t-test, P < 0.001, Figure 2I ).

Immunofluorescence assay was performed in KYSE150, KYSE450, and KYSE180 cells to affirm whether CDCA7 expresses in the nucleus as CDCA7 was found to be a DNA-binding protein and can serve as a transcription regulator (9, 11, 14). The results showed that CDCA7 was located in both the cytoplasm and the nucleus ( Figure S4 ).

Since CDCA7 may act as a transcription regulator, chromatin immunoprecipitation sequencing (ChIP-seq) technology was applied to screen a wide range of DNA fragments interacting with CDCA7. Genome-wide mapping of CDCA7-binding profile by ChIP-seq identified 14,930 binding events (P < 10⁻³), associated with 11,897 unique genes following a nearest gene annotation. As shown in Figure 3A , most (12,462/14,930) of the binding events occur at a distance about 2,000 bp from the transcriptional start site (TSS) of genes, which is generally considered to be the gene promoter region and activation region. These results suggested that CDCA7 may play a role as a transcription factor or transcription regulatory factor. Next, we performed a KEGG pathway enrichment analysis on the 11,897 unique genes which were associated with the DNA fragments obtained from ChIP-seq. The KEGG pathway enrichment analysis showed that target genes were enriched in the pathways including pathways in cancer, cell cycle pathway, PI3K–Akt signaling pathway, MAPK signaling pathway, Ras signaling pathways, and Hippo signaling pathways, which may contribute to ESCC cell proliferation and tumorigenesis ( Figure 3B ). Among the DNA fragments obtained from ChIP-seq, some of them were located in the promoter region of the cell cycle related genes, i.e., 3,000 bp before the transcription initiation site. These genes include CCND1, CCNE1, CCNA2, etcetera ( Figure 3C ). Bdg files, such as the CDCA7_V5.bdg and Control.bdg shown in Figure 3C , are the corresponding bedgraph format track files provided by the company, which are convenient to view the position distribution of reads on the genome under different resolution conditions. CDCA7_V5.bdg represents the DNA fragments that can bind to CDCA7 detected by the anti-V5 antibody, and Control.bdg is the DNA fragments of the input group. When we opened the bdg files in the UCSC database and compared CDCA7_V5.bdg with Control.bdg, the location of the peak is the binding site of CDCA7 with the three cyclins. As shown in Figure 3C , CDCA7 binds with CCND1 at the position of −3,732 to −2,502 bp, binds with CCNA2 at the position of −74 to 734 bp, and binds with CCNE1 at the positon of −53 to 925 bp from each transcription start site, respectively. At the same time, when the binding of H3K4Me1 and H3K21Ac was displayed on the genome in the UCSC database, we found that apart from CCNE1, CDCA7, and H3K4Me1, H3K27Ac shared the same binding position in CCND1 and CCNA2 genomes. These findings once again suggested that CDCA7 may play a role as a transcription factor or transcription regulator to regulate the expression of cyclins.

Since the results of ChIP-seq showed that CDCA7 may act as a transcription regulator to regulate the expression of cyclins, we detected the mRNA and protein levels of the three cyclins in CDCA7 overexpression and knockdown stable cell lines. We found that the expression level of the CCNA2 was significantly decreased when CDCA7 was knocked down and vice versa ( Figures 4A, B ). Meanwhile, we verified the correlation between CDCA7 and the three cyclins using the mRNA data of 96 ESCC tissues in the TCGA database. The results ( Figure 4C ) showed that there was a weak positive correlation between CDCA7 and CCND1 (r = 0.2121, P = 0.038) and a strong positive correlation between CDCA7 and CCNA2 (r = 0.6527, P < 0.0001), while there was no correlation between CDCA7 and CCNE1 (r = −0.0528, P = 0.6116). Next, we analyzed the mRNA expression data of ESCC (n = 358) in GSE53625. Based on the mRNA expression data of ESCC (n = 358), CDCA7 was positively correlated with CCNA2 (r = 0.7047, P < 0.0001). Interestingly, when the mRNA expression data of other types of SCC in the TCGA database were analyzed, we found that CDCA7 was positively correlated with CCNA2 in LUSC (n = 501, r = 0.4995, P < 0.0001) and HNSC (n = 502, r = 0.5771, P < 0.0001) ( Figure 4D ). Immunohistochemistry was further used to detect the tumor tissue stripped from nude mice with anti-CDCA7, anti-CCNA2, and anti-Ki-67 antibodies. The results showed that the staining intensity of CCNA2 and Ki-67 in the CDCA7 knockdown group was obviously weaker than that of the KYSE150 group ( Figure 4E ). The H-score of CCNA2 and Ki-67 in CDCA7-knockdown group (181.344 ± 17.549 and 7.84 ± 0.200) was significantly lower than that in the control group (91.19 ± 9.07 and 58.59 ± 0.626) (t-test, P < 0.001, Figure 4E ).

These results revealed that the CDCA7 probably affected the cell cycle progression, occurrence, and development of cancers through regulating the expression of CCNA2.

The binding region of CDCA7 on CCNA2 started from position −74 to 734 bp relative to the TSS, and we constructed the −90 to 809 bp into the pGL3 promoter vector for the dual-luciferase assay subsequently. The reason why we expanded the region of the DNA fragment is the high GC content of the DNA sequence near 734 bp and there is no way to design a suitable pair of PCR primers. To explore the binding domain of CDCA7, the interval from −83 to 809 bp was divided into four segments randomly. The four segments were −90 to 130, 113–292, 275–476, and 456–809 bp. Dual-luciferase assay indicated that the 456–809-bp region of the CCNA2 was the core element regulated by CDCA7 ( Figure 5A ). To understand the molecular mechanism for the activity of CDCA7 in regulating gene transcription, a de novo search for DNA-enriched motifs was performed within the binding fragments and five predicted motifs were obtained. Each of the motifs corresponded to some transcription factor at different degrees ( Figure 5B ). To demonstrate whether CDCA7 acts its role as a transcription factor or a transcription regulator factor by binding with DNA through these motifs, we constructed the sequences of the motifs into pGL3-promoter vectors for dual-luciferase reporting experiments, the results revealed that motif-1 (5′-TAGACAAGAGTT-3′), motif-2 (5′-GTGATCAGTGCAGA-3′), motif-3 (5′-CTGGAACAGCAC-3′), motif-4 (5′-GTGTGTGTGTGT-3′), and motif-5 (5′-AGTAGTAGTA-3′) might be the functional binding sites of CDCA7 with DNA ( Figure 5D ). Next, we compared these motifs with the sequences from −74 to 734 bp and found five DNA fragments highly similar to the five motifs located in the 456–809-bp region ( Figure 5C ). We inferred that the five DNA segments may be the functional binding sites of CDCA7 with CCNA2. In order to further research whether CDCA7 directly binds to CCNA2 through these sites, ChIP-PCR was performed. We found that CDCA7 bound to the DNA fragment from 484 to 495, 641 to 654, 670 to 679, and 711 to 722 bp in the CCNA2 genomic region ( Figure 5E ). To identify whether CDCA7 regulates CCNA2 through these binding sites, we constructed a dual-luciferase reporter plasmid with the segments of CCNA2 from −90 and 809 bp which deleted the 484–495, 641–654, 670–679, and 711–722 bp. The results showed that CDCA7 could not activate the expression of downstream reporter gene when the four binding sites were knocked out ( Figure 5F ). It proved again that CDCA7 may regulate the transcription and expression of CCNA2 by binding with these four binding sites.

To confirm whether CDCA7 promotes cell cycle through CCNA2, we carried out the interference and rescue experiment of CCNA2. The results showed that forced overexpression of CCNA2 in CDCA7 knockdown ESCC cells ( Figure 6A ) was performed, and a series of phenotype changes had been identified in ESCC cells. The results showed that CCNA2 overexpression promoted the proliferation and colony formation ( Figure 6B ) induced by CDCA7 knockdown in KYSE450 cells. Meanwhile, when we silenced its expression in CDCA7 overexpression ESCC cells ( Figure 6C ), the results of cell phenotype experiments showed that knockdown of CCNA2 can reduce cell proliferation and colony formation ability ( Figure 6D ) induced by CDCA7 overexpression in KYSE180 cells. These results suggested that the acceleration effect of CDCA7 on cell cycle may depend on its transcription regulation of CCNA2, and CCNA2 inhibition may partially reverse the cell proliferation progression induced by CDCA7 overexpression.

The results indicated that CDCA7 gene might act as a tumor promoter in ESCC and its copy number amplification or increased expression may accelerate the cell cycle process and promote cell proliferation by binding to the genome of CCNA2 functional domain and increasing its expression in ESCC. When CDCA7 is knocked down or decreased, its transcription regulation effect is attenuated, and the cell cycle process and the cell proliferation of ESCC are inhibited as the expression of CCNA2 is depressed. Furthermore, the mechanism that CDCA7 acts as an oncogene possibly through regulation of cell proliferation might be applied in various types of SCC.