Genomic characterization of a newly established esophageal squamous cell carcinoma cell line from China and published esophageal squamous cell carcinoma cell lines

CSEC216 was established from an endoscopic biopsy sample derived from a 45-year-old male patient who underwent esophageal endoscopy in the Cancer Hospital of Shantou University Medical College. The tumor located at middle thoracic esophagus, 23–30 cm from the incisor with constrictive lumen and elevated mucosa observed in the esophageal tract (Fig. 1a).

Written informed consent was received from the patient.

The endoscopic specimen was immersed in a 5 ml tube filled with culture medium (Dulbecco’s modified Eagle’s medium with 10% fetal bovine serum, 100 U/ml penicillin, 100 mg/ml streptomycin). The specimen was then washed 3 times in PBS and minced with a sterile scalpel into small fragments of 1 mm³. The tissue fragments were placed in a collagen coated T-25 flask containing 2 ml culture medium and gently placed in a 37 °C incubator containing a humidified atmosphere and 5% CO₂. Cells were grown for 7 days. After which the medium was replaced every 3 days. Stromal fibroblasts were eliminated by differential trypsinization [7]. Cells were suspended with 0.25% trypsin and 0.02% EDTA treatment and propagated at a split ratio of 1:2 or 1:3 depending on their growth rate. The cells were resuspended in freezing medium (90% FBS + 10%DMSO) and stored in liquid nitrogen from passage 2.

The CSEC216 cell line had been subcultured for 50 passages, during which the total culturing time and population doubling time(PDT)¹ were used for growth curve plotting.

For population doubling time calculation, 10,000 cells at passage 28 in 300 μl culture medium were inoculated into each well of a 24-well plate and culture medium was replaced every 3 days. Three wells of cells were harvested by trypsinization each time with 24 h intervals, the cell number of each well was determined using a Neubauer hemacytometer, then the cell numbers were averaged. Cell growth during log phase was used to calculate doubling time (Roth V. 2006 Doubling Time Computing, http://​www.​doubling-time.​com/​compute.​php 18/10/2015).

The karyotype of CSEC216 cells was analyzed by spectral karyotyping (SKY). To obtain cells at metaphase, cells at 70% confluence at passage 20 were treated with 0.03 μg/ml colcemid for 7 h then harvested by trypsinization. After neutralization by mixing the cell suspension with equal amount of culture medium and centrifugation at 1000 rpm, the supernatant was discarded. Cells were resuspended in 2 ml 0.8% sodium citrate and kept at 37 °C for 15 min. 250 μl fresh made fix solution (methanol:acetic acid = 3:1) was added to the cell suspension at 37 °C for 5 min to prefix the cells. Cells were collected by centrifugation and washed 3 times with fix solution. After the final wash, supernatant was discarded. Cells were resuspended in 500 μl fix solution and 15 μl cell suspension was dropped onto an adhesive slide, air dried and the hybridization area was marked with a diamond pen. The slide was stored at room temperature and kept from light for 7 days. SKY was performed using a SKY probe kit SkyPaint™ mixture (Applied Spectral Imaging, Migdal a’Emek, Israel), according to the manufacturer’s instruction). The karyotype was analysed by SkyView™ Imaging System (Applied Spectral Imaging) equipped with a Zeiss Axioplan-2 fluorescence microscope.

Genomic DNA of cells at passage 5 was extracted using DNA Mini Kit (QIAGEN) under manufacture’s protocol. DNA samples were sent to Genetica DNA Laboratories USA for Short Tandem Repeats(STR) profiling test. Fifteen STR loci and AMEL were tested. STR electropherogram and signed STR profiling report were shown in supplementary data. Eight STR loci and AMEL were used for online matching in the STR databases of ATCC (https://​www.​atcc.​org/​STR_​Database.​aspx 20/11/2015) and DSMZ (https://​www.​dsmz.​de/​fp/​cgi-bin/​str.​html 20/11/2015). Matching criteria for human cell line authentication was set as 0.8 recommended by the website.

For the in vitro migration assay, 10,000 cells of CSEC216 and KYSE520 [10] were resuspended in Millipore chamber (Hanging Cell Culture Inserts 8 μm PET, Millipore). The chambers were placed in 24-well plate containing 750 μl full complete culture medium and incubated for 24 h. After incubation, non-invaded cells on top of the membrane were scraped off with cotton swab. The chamber was treated with fixing/staining solution (mixture of 4% paraformaldehyde and crystal violet) at room temperature for 30 min.

For invasion assay, the same procedure as in the migration assay was followed. The only difference was that cells were suspended in a mixture of 100 μl FBS free DMEM medium and 100 μl matrigel (BD Biosciences).

ESCC cell lines were searched on Cellosaurus (https://​web.​expasy.​org/​cellosaurus/​) [11]. Keyword C4024 (NCI Thesaurus Code for ESCC) was used for searching. A list of problematic ESCC cell lines were downloaded from Cellosaurus database, all ESCC cell line data was shown in Additional file 1: Table S1. CCLE (Cancer Cell Line Encyclopedia) is a database that contains multi-omic data of over a thousand cancer cell lines, 28 ESCC cell lines included. Before took into analysis in this paper, cell lines from CCLE were checked if any problematic cell lines existed. CCLE ESCC cell lines’ background information was shown in Additional file 2: Table S2.

CSEC216 cells showed polygonal epithelial cell morphology and adherent growth as monolayer(Fig. 1b–e). Immunofluorescence staining revealed CSEC216 cells positively expressed Cytokine and E-Cadherin but negative for vimentin and the Ki-67 labeling index (LI) was 67.1% (Fig. 1f–i). These results demonstrated that CSEC216 cells were derived from epithelial cells and possessed high proliferation activity. The growth curve demonstrated that CSEC216 cells started to proliferate rapidly after the first subculture and the growth rate becomes steady after the 5th subculture (data not shown). The PDT was 29.7 h at log phase. For migration and invasion in vitro analysis, only minority cells had migrated through the membrane, the migration rate and invasion rate is 3.48% and 1.41% respectively compared with 6.05% and 2.91% of KYSE520 cell line (photo not shown).

In total 15 STR loci and AMEL were analyzed in an STR profiling test. AMEL, D5S818, D13S17, D7S820, D16S539, vWA, TH01, TP0X, CSF1PO STR loci were matched with the STR database of ATCC and DSMZ. No cell lines were matched in ATCC STR database. Cell lines with similar STR loci were matched in DSMZ STR database (data not shown). None of these cell lines was cultured during the CSEC216 establishment. Both results showed that CSEC216 was a unique cell line that was not misidentified and contaminated with other registered cell lines. Full report and electropherogram were shown in Additional file 3: Figure S1.

SKY analysis yielded representative karyotype photograph of CSEC216 cells at passage 20, which was demonstrated in Fig. 2a, karyotypic report was summarized in Fig. 2b. Main clone was hypotriploidy. The chromosome number ranged from 65 to 70. Numerical abnormalities and structural aberrations were common in CSEC216 cells.

Frequently observed chromosomal numerical abnormalities were presented by chromosome loose of Y,8,11,19 and chromosome gain of 1,3,5,17,20. Structural aberrations were complicated as deletion, translocation, isochromosome and derivation chromosome were frequently observed. Derivation chromosome of 2, 3, 13 composed by the most sophisticated translocations was observed, i.e. der (2) t (2, 8) (q2?;?)t(2;8)(?;?)t(2;8)(?;?)t(8;12)(?;p11)、der(3)t(3;?7)t(p11;?)t(?7;?10)(?;?)t(?3;?10)(?;?)、der (13) t(?8;13)(?;p11)t(?8;?10)(?;?)t(?10;?13)(?;?)t(?8;?13)(?;?). Chromosomal break points at p1 or q1 was common in chromosome 1, 2, 3, 4, 5, 7, 8, 9, 10, 12, 13, 16, 17, 18, 21, X.

The genomic DNA of CSEC216 were sequenced with 30X average depth. WGS yielded 3,290,751 SNAs, 572,345 INDELs and 1860 CNA regions. After removal of germline mutations by comparing with gnomad database file (hg19_gnomad211_genome),75527 SNAs and 61866 INDELs were obtained. The genome-wide mutation rate of SNA and INDEL were 2.52 and 2.06 mutations per megabase respectively. The density of SNAs and INDELs per megabase along the chromosomes were plotted in Fig. 3b. C:G > T:A transitions were predominant substitutions which was consistent with our previous work and other published data, accounted for 30.58% (23090/75498) of all SNAs (Fig. 4a, b) [12, 13]. Landscape of CNAs in euchromosomes were showed in Fig. 3a. Genome loss or gain were detected in every chromosome, the size of DNA fragments ranged from 1 KB to 53 MB. Large scale (> 100 kb) amplifications at every chromosomes and large scale deletions at 1p, 2q, 3p, 4p, 4q, 5q, 6q, 7q, 10p, 12q, 13q, 16q, 18q, 20q, 21q, Yq were identified. The frequent copy number changes were elsewise reflected the sophisticated genotype. For copy number gain, the highest copy number was 13 while the lowest was 2. SNA and CNA data of CSEC216 were shown in Additional file 4: Table S3, Additional file 5: Table S4.

Somatic SNAs were converted into a matrix of 96 representative mutations in trinucleotide context for mutation signature analysis (Fig. 4a,b). Mutational signature analysis revealed that the mutational spectrum of CSEC216 were most similar to COSMIC Cancer Signature.5 (COSMIC Mutational Signatures, https://​cancer.​sanger.​ac.​uk/​cosmic/​signatures_​v2), which featured by transcriptional strand bias for T > C substitution at ApTpX context, and has been identified in all kinds of cancer types with yet unknown aetiology (Fig. 4c, d).

The etiological implication of the mutation signature extracted from all somatic SNAs of CSEC216 was vague. Since 98.8% of somatic SNAs of CSEC216 were located in intron and intergenic region, the mutational context of those SNAs could over rule the mutations that literally contributed to relative etiology. To investigate the impact of genomic mutations to those genes which were implicated to play a role in the carcinogenesis of ESCC, mutations at exon, up/down stream were extracted from SNA, INDEL and CNA data. Resulted in 910 SNAs and 1310 INDELs which affected 999 genes and 14072 genes in 1166 CNA regions. A list of 44 genes related with ESCC were selected from literatures and COSMIC database (Cancer Gene Census, https://​cancer.​sanger.​ac.​uk/​cosmic/​download), of which most genes were chosen from our previous work [12, 14‐17]. Gene list were shown in Additional file 6: Table S5. ESCC related genes were mostly found with amplified gene copy number, i.e. TP53, SOX2, EGFR, MYC, CCND1, CDKN2A, NOTCH1, BRCA1, NFE2L2, etc. While only TP53 and NFE2L2 were found in SNAs (Fig. 5). Integrated mutation data of CSEC216 was shown in appendices.

After removing problematic cell lines, a total of 28 SNA and 24 CNA mutation data were downloaded from CCLE database (https://​portals.​broadinstitute.​org/​ccle/​about). CNA and SNA data of cell line were shown in Additional file 7: Tables S6, Additional file 8: Tables S7. Firstly, the mutation matrix of each cell line with 96 SNA mutations at trinucleotide context were compared (Fig. 6a). The spectrum of 96 mutation context of 28 cell lines were shown in Additional file 9: Figure S2. Globally, high rate of C > T transition at TpCpX, CpCpX, ApCpG and C > G transversion at TpCpA/C/T trinucleotide were commonly found among cell lines. However, CSEC216 showed relatively lower mutation rate at the above mentioned sites, but uniquely possessed high mutation rate of T > C transition at ApTpA trinucleotide. De novo mutation signatures were extracted from ESCC cell lines by non-negative matrix factorization (NMF) algorithm in MutationalPatterns package. Three mutation signatures were extracted (Fig. 6b). Signature A was featured by high rate of C > T transition at XpCpG trinucleotide. After comparing with COSMIC Mutational Signatures, Signature A was similar to COSMIC Cancer Signature 1, which has been commonly found among different cancer types, also proposed to be the result of spontaneous deamination of 5-methylcytosine and has a relation with aging [18, 19]. Signature B had dominant C > T transition at CpCpC and TpCpC trinucleotide, after comparison with COSMIC database, Signature 11 was most similar. Signature 11 has been found in melanoma and glioblastoma, the mutational pattern may caused by treatments with the alkylating agent temozolomide [18]. Since temozolomide is not prevalently used to treat ESCC patient, the aetiology of this Signature B remained unclear. Signature C exhibited high rate of C > G and C > T transition transversion at TpCpX trinucleotide. When comparing with COSMIC database, Signature B was similar to Signature 2, which has been identified in variety cancer types, and possibly affected by the activity of APOBEC family of cytidine deaminases, especially APOBEC1/3A/3B [18, 20].

The contribution of Signature A/B/C to the mutational patterns of each cell line was compared. ESCC cell lines demonstrated diverse composition of mutation signatures. Signature A/B/C was dominant(proportion > 50%) in 32.1% (9 of 28)/21.4% (6 of 28)/17.9% (5 of 28) cell lines respectively. Extreme cases were found in OE21 and TE15 where Signature C take 93.7% and 91.3% proportion among the 3 signatures (Fig. 6c).

For ESCC cell lines researching on Cellosaurus, 224 hits plus 2 ESCC cell lines were retrieved, 12 of which were problematic cell lines because of cell line cross-contamination. Meanwhile, information of 14 immortalized human esophageal epithelial cell lines were collected from Celloasurus and published literatures. Information of those cell lines were shown in appendices.

In order to investigate the mutation status of genes that have been reported being mutated in ESCC, genes that had somatic mutation at exonic or up/down stream region were selected from ESCC cell lines and mapped to 44 ESCC related genes. 39 genes were identified in SNA/INDEL data, TP53 undoubtedly was the most prevalent one, every cell line possessed 1 or 2 mutations in TP53, followed by LRP1B (42.9%), CSMD3 (39.3%), NOTCH1 (35.7%), ZFHX4 (32.143%), NFE2L2 (28.6%) (Fig. 7b). 31 genes were identified in CNA data, the most significantly altered genes were CCND1, which amplified in every cell line, followed by MYC, SOX2, SFRP4 (95.8%), EGFR, PIK3CA, BCL6 (91.7%). CDKN2A was found deleted in 95.8% cell lines, followed by LRP1B (83.3%), KDM6A, PCDH9 (75%), PTEN (70.8%), ARID1B, BRCA2, RB1, NFE2L2 (66.7%) (Fig. 7a).