A novel human in vitro papillomavirus type 16 positive tonsil cancer cell line with high sensitivity to radiation and cisplatin

The study was approved by Lund University regional ethical review board (LU 376–01), and the samples were collected, after informed consent was obtained, at Skåne University Hospital, Lund, Sweden.

Tumor samples were rinsed consecutively in phosphate buffered saline (PBS) and R10 medium (RPMI 1640 with stable glutamine supplemented with 1 mmol/L sodium pyruvate, 1 × MEM non-essential amino acids, 20 μg/mL gentamicin and 10% fetal bovine serum (FBS), all from GE Healthcare (Piscataway, NJ, USA). The tumor samples were cut in pieces and transferred to cell culture flasks with R10 medium and left to attach and grow at 37 °C under a humidified atmosphere with 5% CO₂. The cells used in the described experiments were passaged less than 18 times after establishment. Mycoplasma tests were performed on a regular basis by GATC Biotech AG (Konstanz, Germany) and were negative.

For cell doubling determination, cells were seeded in eight 96-well plates. The plates were analyzed by the sulphorhodamine B (SRB) method as previously described [16] consecutively at the indicated time points. Error bars indicate standard error of the mean. The data points were fitted to a logarithmic equation using the GraphPad Prism (5.04) software package (GraphPad Software, La Jolla, CA, USA).

Single tandem repeat (STR) analysis was performed by Uppsala Genome Center using the AmpFlSTR Identifiler polymerase chain reaction (PCR) amplification kit (Applied Biosystems, Carlsbad, CA, USA). Separation of DNA fragments was performed with capillary electrophoresis on the ABI3730XL DNA analyzer (Applied Biosystems, Foster City, CA, USA) and allele scoring was performed with the GeneMapper 4.0 software package.

Experiments on mice were approved by the local ethics committee (488/2017). We used in-house bred, athymic 5- to 8-week-old BALB/c nude (nu/nu) mice. The mice were given food and water ad libitum and treated according to regulations issued by the Swedish Board of Agriculture (SJVFS 2017:40). Cells were injected subcutaneously into the flanks of non-anesthetized animals. To determine the doubling time, tumors were retransplanted to 10 nude mice under ketamine anesthesia, two tumors per mouse. The length and width of the tumors were measured with calipers three times per week and the relative tumor size calculated in relation to day 0 (set as the day when all tumors had shown growth over 3 consecutive time points). The data points were fitted to a logarithmic equation in GraphPad Prism. After the measuring period, the mice were euthanized using cervical dislocation and the tumors removed for histological sectioning.

Xenografts and cell pellets were sectioned in 3 μm slices and stained with hematoxylin and eosin, and the following antibodies: the CINtec p16-kit (E6H4, cat. no. 725–4713), anti-pan keratin, (AE1/AE3/PCK26, cat. no. 760–2595), and anti-p63 (4A4, cat. no. 790–4509) all from Ventana Medical Systems (Tucson, AZ, USA), anti-cytokeratin 5 (XM26, cat. no. NCL-L-CK5) from Leica Biosystems (Nussloch, Germany), and p40 (BC28, cat. no. ACI3066C) from Biocare Medical (Pacheco, CA, USA), All stainings were performed using the BenchMark Ultra platform (Ventana Medical Systems).

Three HNSCC cell lines, previously established in our laboratory, were used for comparison in several experiments: LU-HNSCC-4, LU-HNSCC-5 and LU-HNSCC-6 [17]. They were propagated in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% FBS, 100 units/mL penicillin, and 100 units/mL streptomycin sulfate, and used within 15 passages after thawing. Single tandem repeat analysis was performed before the experiments as detailed above showing no cross-contamination between the cell lines or with other common contaminants. Tests for mycoplasma infection prior to the experiments were negative.

A 2–3 mm biopsy was immersed in 1 mL saline solution and immediately transferred to the laboratory. The saline was removed and the biopsy was digested in lysis buffer and the DNA extracted with the Total NA-kit using MagNA Pure LC [18] (Roche, Pleasanton, CA, USA). Sample adequacy was assessed by PCR for the human beta-globin gene [19]. Identification of 39 genital HPV types was carried out by modified general primer polymerase chain reaction (MGP-PCR) and subsequent Luminex analysis (Luminex, Austin, TX, USA) [18, 20, 21].

Samples from subsequent passages in cell cultures were immersed in 1 mL GITS (4 mol/L guanidinium thiocyanate, 22 mmol/L sodium citrate, and 5% N-lauroylsarcosine sodium salt) and stored at − 20 °C. After thawing, 200 μL was used for DNA extraction with the Total NA-kit using MagNA Pure LC (100 μL output).

Briefly, the number of viral genomes per cell was quantified by carrying out two separate real-time PCR tests, including appropriate standard curves, to amplify the HPV16 E7 gene and the human beta-globin gene as previously described [18]. The samples were analyzed in duplicate.

As a marker for the presence of mixed or episomal forms of HPV16, samples were analyzed for the ratio of E2/E7 gene copy numbers. HPV16 was classified as mixed status when complete E2 was evident from complete sequencing of the HPV16 genome and the E2/E7 ratio was < 0.9, and pure episomal status in the presence of complete E2 open reading frame (ORF) and E2/E7 ratio of > 0.9. The E7 copy numbers were retrieved from the viral load measurement as described above. For quantification of the E2 copy numbers we used a modified version of the PCR described by Peitsaro et al. [18], where E2/E6 ratios greater than 1 indicate predominance of the episomal form [22]. Quantification was extrapolated from a linear regression standard curve that was included in each batch assay, as previously described [18]. The samples were analyzed in duplicate.

The complete HPV16 genome was amplified in a 25 μL solution with 2.5 μL (15.5 ng) of extracted nucleic acid from passage 2, 0.2 μmol/L of each primer (HPV16 7465F, 5′-ATGCTTTTTGGCACAAAATGTG, HPV16 7464R 5′-GCAACCGAATTCGGTTGAAG), 200 μmol/L of each dNTP, 1 × PrimeStar GXL Buffer, and 0.625 U PrimeStar GXL DNA Polymerase (TaKaRa Bio, Shiga, Japan). The PCR was performed at 98 °C for 10 s, 60 °C for 10 s, and 68 °C for 8 min for 45 cycles. Amplified DNA was separated by electrophoresis in a 1.0% agarose gel (agarose Type I-A, A0169, Sigma-Aldrich, Stockholm, Sweden) with 1 × Gel RedTM Nucleic Acid Gel Stain (VWR International, Lund, Sweden) in 0.5 × TBE buffer [23] and visualized by UV light.

The complete HPV16 genome was obtained by next-generation sequencing (NGS) in the MiSeq system (Illumina, San Diego, CA, USA) from passage 2.

Briefly, extracted DNA (5.6 ng/μL in a total volume of 20 μL) was whole genome amplified (WGA) with multiple displacement amplification using the Illustra™ Ready-To-Go™ GenomiPhi™ DNA Amplification Kit (GE Healthcare) based on the manufacturer’s protocol, but with some modifications. Five microliter of extracted DNA was diluted with 20 μL PCR-grade water and 25 μL denaturation buffer, incubated at 95 °C for 3 min, cooled on ice, added to the lyophilized cake, incubated at 30 °C for 7 h and inactivated at 65 °C for 10 min. The WGA product was diluted 1:2 in PCR-grade water and quantified with QuantiFluor-ST (Promega, Madison, WI, USA). A DNA library with 2 unique indices was prepared from 50 ng of the WGA product using the Nextera DNA Sample Preparation kit, user guide revision B (Illumina) and validated by fragment size using 2100 Bioanalyzer High sensitivity DNA assay (Agilent Technologies, Santa Clara, CA, USA) and quantified with QuantiFluor-ST (Promega). The library was normalized to 4 nmol/L in EB-buffer (Qiagen, Hilden, Germany), denatured, diluted to 1.8 pmol/L, spiked with 1% PhiX control and paired-end sequenced by 151 plus 151 cycles using NextSeq 500 High Output reagent kit (Illumina) as described in the user guides Denaturing and Diluting Libraries for the NextSeq 500, revision A and NextSeq 500 kit Reference Guide, revision F.

Papillomavirus-related contigs having > 90% identity with each other were clustered. The total number of sequencing reads from each cluster was calculated and the longest contig was selected as a representative to remove redundancy. A de novo HPV-sequence was obtained by the use of clustering analysis, conducted separately for contigs longer and shorter than 700 bp length. For the HPV type, the total number of reads was calculated. All analyses were performed using in-house R (http://​www.​r-project.​org/​), python (http://​www.​python.​org) and bash (http://​www.​gnu.​org/​software/​bash/​) scripts that ran on a high performance (40 core, 2 TB RAM) Linux server.

The obtained sequence was then analyzed by the use of BioEdit v7.0 [24] for presence of the expected ORFs and in cases of ambiguities, re-sequencing from the passage 2 was performed to resolve the ambiguity.

The primary de novo sequence had two segments of N-ambiguities of 72 bases and of 129 bases at positions 4633–4704 and 7150–7278, respectively. In order fill these gaps, two PCRs were performed, each with 50 μL PCR mixture that contained 5 μL of template of extracted DNA from passage 2, 0.3 μmol/L of each forward and reverse primer, 200 μmol/L of each dNTP (Roche), 1.5 U AmpliTaq Gold, 1 × reaction buffer and 3.0 mmol/L MgCl₂ (all from Applied Biosystems). The following primer pairs were used to generate amplicons of 887 bp with HPV16 forward primer at position 3992 (5′- TATTGTGGATAACAGCAGCCTCTG) and reverse primer at position 4878 (5′- GGTGTGCTACTAGTTACTGTGTTAGGG), and of 316 bp with HPV16 forward primer at position 7048 (5′-AAGCAGGATTGAAGGCCAAAC) and reverse primer at position 7363R (5′- AATAACCACAACACAATTAGTAGGTGTTG). We also separately sequenced a fragment of 250 bp in order to verify the guanosine at position 1376, by a HPV16 forward primer at position 1250 (5′- GCGAAGACAGCGGGTATGG) and a reverse primer at position 1489 (5′- TGCTAACATTGCTGCCTTTGC).

PCR was carried out in an automated thermocycler (Eppendorf) programmed for tube setting: with the following parameters: 94 °C for 10 min; 45 × (95 °C for 15 s, 55 °C for 30 s, and 72 °C for 1 s). Five microlitres of the amplicons were separated by electrophoresis in a 2.0% agarose gel (agarose Type I-A, A0169, Sigma-Aldrich) with 1 × Gel RedTM Nucleic Acid Gel Stain (VWR International, Lund Sweden) in 1 × TBE buffer [23] and visualized by UV light. The remaining amplified material (45 μL) was purified by Illustra MicroSpin S-300 HR Columns (GE Healthcare) and sequenced with the HPV16 primers described above (Eurofins, Ebersberg, Germany). The sequences were then used to fil the gaps of the de novo sequence and the final sequence was compiled.

Cells were immersed in GITS and used for mRNA extraction with the Oligotex Direct mRNA Mini Kit (Qiagen) [18]. The quantitative mRNA PCR was performed as previously described [18].

Total RNA was extracted using TRI Reagent (Sigma-Aldrich) and Direct-zol RNA MiniPrep (ZYMO Research, Irvine, CA, USA) according to the manufacturer’s protocol. One μg of total RNA was reverse transcribed in a 20 μL reaction at 37 °C by using M-MLV Reverse Transcriptase and random primers (both from Invitrogen, Carlsbad, CA, USA) according to the protocol of the manufacturer. One microliter of cDNA was subjected to PCR amplification with primers indicated in Fig. 3 and listed in Table 1.

Cells were lysed in radioimmunoprecipitation assay buffer (50 mmol/L Tris, pH 7.8, 150 mmol/L NaCl, 1% sodium deoxycholate, 0.1% sodium dodecyl sulphate, 1% Triton X-100, 1 mmol/L dithiothreitol and protease inhibitors), subjected to SDS-PAGE, transferred to nitrocellulose membrane, and stained with antibodies against HPV16 L1 (ab30908, Abcam, Cambridge, UK), L2 (sc-65,708, Santa Cruz Biotechnology, Dallas, TX, USA), and epidermal growth factor receptor (EGFR) (no. 2232, Cell Signaling Technology, Danvers, MA, USA). Proteins were detected using chemiluminescence. For quantification of EGFR, loading control was performed using Coomassie R-350 as previously described [25].

Primers for all exons of TP53 were designed using the Primer3 software (http://​primer3.​ut.​ee/​); primer sequences are available on request. PCR was done on genomic DNA according to standard methods and amplified fragments were sequenced using the BigDye v1.1 Cycle Sequencing Kit on an ABI-3130 Genetic Analyzer (Applied Biosystems). ChromasLite 2.1.1 free online software (http://​technelysium.​com.​au/​) was used to analyze the data.

For cytogenetic analysis, cells were cultured in DMEM/F12 medium with 10% fetal bovine serum and antibiotics until the mitotic index was high, and then exposed to colchicine (0.03 μg/mL) for 4–5 h. The cells were detached by trypsin-EDTA, treated with 0.06 mol/L KCl for 35 min, and then fixed three times with methanol/acetic acid 3:1. Chromosome preparations were incubated at 60 °C overnight and then treated with 0.3 mol/L sodium chloride, 30 mmol/L sodium citrate, pH 7.0, at 60 °C for 4 h. G-banding was done using Wright’s stain and karyotypes were described according to ISCN 2009 [26].

For cisplatin, cells were seeded and, after 48 h, treated with increasing concentrations of cisplatin (Sandoz AS, Copenhagen, Denmark) for one hour. After 5 days at 37 °C under a humidified atmosphere with 5% CO₂, the cell amounts were measured by the SRB assay.

For determination of cetuximab sensitivity, cells were treated with increasing concentrations of cetuximab (Merck AB, Solna, Sweden) for 5 days, after which the cell amounts were determined by the SRB assay.

Radiosensitivity was analyzed by irradiating the cells in a Gamma Cell-3000 Elan from MDS Nordion (Ottawa, ON, Canada) equipped with a Cs-137 source. After irradiation, the cells were seeded and left to grow until the untreated controls were approximately 90% confluent, and then analyzed by the SRB assay.

For all experiments the data were fitted to a sigmoidal dose-response equation with variable slope using non-linear regression, and the half maximal inhibitory concentration (IC₅₀), inhibitory dose (ID₅₀) or maximum inhibition (E_max) calculated in GraphPad Prism.

A total of 27 tumor samples (23 HPV-positive), were collected and treated as detailed in the Materials and Methods section. In a single case, we were able to establish a cell line. The original tumor tissue was obtained from a 48 year old non-smoking man with moderate daily consumption of alcohol, and, besides exercise induced asthma, no other intercurrent diseases. The initial symptom was left-sided pain in the throat. He had no weight loss. Three days after his first symptom he was examined by an ENT-specialist who noted a prominent left sided tonsil without mucosal ulceration. No nodes could be felt. A fine-needle aspirate was performed, with negative result. Two and a half weeks later he was examined under anesthesia and biopsied through intact mucosa. This biopsy confirmed a non-keratinizing, p16 (CDKN2A) positive squamous cell tonsil carcinoma with HPV16 detected in the original biopsy. After further investigation with CT and CT-PET scan, the tumor was classified as stage T2N0M0. The patient was treated with full dose radiotherapy (Intensity-modulated radiation therapy) without concomitant chemotherapy leading to full remission, and was relapse-free after four and a half years. The primary tumor received 64 Gy (2.0 Gy/fraction) and the neck 54.4 Gy (1.6 Gy/fraction).

Approximately 19 months (8 passages) after the initial seeding, the cell line was visibly free from fibroblasts and displayed a stable growth rate (Fig. 1a) and was denominated LU-HNSCC-26. The doubling time was determined to be 84 h (Fig. 2a).

For future identification and authentication of the cell line, STR profiling was performed using 16 loci (Table 2). No exact matches were found when comparing with other analyzed cell lines in the DSMZ repository [27].

Cell pellets were sectioned and stained for p16 and epithelial markers. The cells were p16-positive (cytoplasmic and nuclear staining) (Fig. 3a). They were also positive for CKAE1/AE3 and CK5 (cytoplasmic staining) (Fig. 3b and c), and p63 and p40 (strong nuclear staining) (Fig. 3d and e), establishing the squamous epithelial origin of the cells.

The LU-HNSCC-26 cells were injected subcutaneously in nude mice resulting in xenografts that could be successfully retransplanted to new mice. The doubling time of tumor growth was determined to be 177 h (Fig. 2b). The histology of the resulting tumors was compared with that of the original tumor. Both tumors showed non-keratinizing morphology, growing in solid sheets and trabeculae. Scattered mitoses and apoptotic cells was seen in both, whereas prominent tumor necrosis was only is observed in the xenograft (Fig. 1b and c). Both xenograft and original tumor were strongly positive for p16 (Fig. 1d and e).

At passage two and eight the viral DNA load was 14.8, coefficient of variation (CV%) 6.6, and 19.5 (CV% 3.7) copies per cell, respectively. Long range PCR for detection of the complete HPV16 genome from passage two and eight demonstrated amplicons of approximately 8 kb (Fig. 4). Episomal HPV16 was present in the cell line with E2/E7 ratios of 0.8 (CV% 1.7) and 1.1 (CV% 2.6) from passage two and eight, respectively. The viral expression manifested 1.0 (CV% 0.1) E7 mRNA copies per HPV16 genome of passage two.

To investigate if correctly spliced HPV16 mRNAs were produced, we extracted total RNA (Fig. 5b) and performed RT-PCR with various HPV16 specific primers (Fig. 5a). We found production of mRNAs encoding full-length E6, E6*I, E6*II and E7 proteins. The E7 mRNA spliced from SD226 to SA409 was the most abundant of the early mRNAs (Fig. 5c).). Both E2 mRNAs and mRNA spliced from SD880 to SA3358 were detected (Fig. 5c), confirming that episomal copies of the HPV16 genome were present and transcriptionally active. We detected mRNAs spliced from SD880 to the exclusively late 3′-splice site SA5639, suggesting that L1i mRNAs were produced (Fig. 5c), and L2 mRNAs were also detected. These results suggested that HPV16 late proteins L1 and L2 could be produced in the LU-HNSCC-26 cell. We therefore performed a Western blot which confirmed the presence of L1 and L2 proteins in the LU-HNSCC-26 cells (Fig. 5d), albeit at very low levels.

From passage two, the complete DNA sequence of HPV16 was obtained by NGS with a minimum nucleotide coverage of 134 (Fig. 6). The HPV16 genome comprised 7904 bp, GenBank accession KY994539. The genomic organization showed the characteristic features shared by HPV types. The HPV16 genome showed closest identity (99%, 7900/7906) to the HPV16 strain CU4 (GenBank accession FJ610149), originally isolated from a uterine cervical cytology sample with CINI. In comparisons to this isolate, our HPV16 demonstrated a deletion of two bp in the second non-coding region between the E5 and L2 ORFs. Furthermore, within the E1 ORF at nucleotide position 1376, a guanosine was present instead of adenosine, translating to cysteine instead of tyrosine in the putative E1 protein.

The karyotype was determined to be 43–47,XY,del(2)(q33), del(3)(p21),der(4)t(4;5)(q35;q31), add(5) (q35),-7,-14,add (15) (p11),?i(16)(q10),-21,+mar[cp6]/44–47,idem,i(8)(q10)[cp3] (see karyogram in Additional file 1).

The gene TP53 was wild-type but homozygous for a single nucleotide polymorphism chr17:7579472G > C, giving a proline to arginine substitution at position 72 (see partial sequencing results in Additional file 2).

The sensitivities to commonly used treatments were assessed in vitro and compared to three HPV-negative HNSCC cell lines. LU-HNSCC-26 was considerably more sensitive to cisplatin than the other HNSCC cell lines tested (Fig. 7a and Table 3). The IC₅₀ for LU-HNSCC-26 (0.99 μmol/mL) was 10–20 times lower than those of the other cell lines.

For cetuximab, we compared the maximum inhibition (E_max) of the drug instead of the IC₅₀. LU-HNSCC-26 was only inhibited to a maximum of 18% (Fig. 7b and Table 3), which was less than the other cell lines [28]. This difference could not be correlated with the EGFR expression as assessed by Western blotting showing a similar expression in all four cell lines (Fig. 7d).

As for cisplatin, LU-HNSCC-26 was considerably more sensitive to radiation compared to the other cell lines (Fig. 7c and Table 3). The ID₅₀ for LU-HNSCC-26 (0.90 Gy) was 18–26 times lower than those of the other cell lines.