Background
Metastasis to the cervical (neck) lymph nodes is one of the most significant clinical factors responsible for death from oral squamous cell carcinoma (SCC). Currently, there are no satisfactory clinical, imaging, pathologic or molecular techniques that can reliably determine if neck metastases are present at the time of surgery to remove the primary tumor. Therefore, patients and physicians frequently elect to remove the cervical lymph nodes (neck dissection) at the time the tumor is excised if the chance of metastasis is > 20% based on current imperfect risk assessment capability. This surgical procedure is lengthy, complex, and risky with high morbidity due to functional deficits and disfigurement. Being able to determine which patients do not need such surgery would have a substantial, immediate clinical benefit.
We recently reported that two subtypes of oral SCC distinguished by tumor genomic aberrations differ in risk for metastasis [
1]. One subtype, the 3q8pq20 subtype, is characterized by the presence of one or more of the recurrent copy number aberrations, +3q, -8p, +8q and/or +20 and has a substantial risk of metastasis (46%). The other subtype (non-3q8pq20) lacks these copy number alterations and is associated with a low risk of metastasis (7%). These initial studies, which were replicated in a small independent cohort, indicated that non-3q8pq20 status has 93% negative predictive value (NPV),
i.e., ability to predict that these cases do not have neck metastases, and thus do not need neck dissection.
The non-3q8pq20 tumors lack chromosome level instability, which suggests that development of these tumors could be associated with other, copy number neutral, mechanisms, such as microsatellite instability or epigenetic alterations. Microsatellite instability is not common in oral SCC from western countries, whereas genome-wide alterations in methylation patterns are observed [
2,
3]. Analysis of a head and neck cancer patient cohort [
2] for which both copy number and methylation measurements were available (NCBI GEO accession numbers GSE20939 and GSE20742, respectively) found 15 loci significantly differentially methylated in 3q8pq20 compared to non-3q8pq20 tumors and normal oral tissue [
1]. To investigate whether these loci are a potential biomarker for distinguishing 3q8pq20 and non-3q8pq20 tumors, we investigated the methylation status of the loci in an oral SCC cohort in which 3q8pq20 status had been determined [
1]. The overall goal was to develop a simple assay for 3q8pq20 status utilizing a panel of differentially methylated loci that could be performed on tissue samples obtained by a non-invasive technique prior to surgery and thus guide decisions regarding the need for neck dissection.
Methods
Patients and tissue samples
The study was approved by the Institutional Review Board of the University of California San Francisco (H7867-23910-05). Formalin fixed paraffin embedded (FFPE) SCC surgical resection specimens were available from 52 cases of the previously published SCC cohort#2 [
1] and included oral cavity sites–tongue, gingiva, floor of mouth, retromolar trigone and buccal mucosa. Associated clinical data were obtained through the University of California San Francisco Oral Cancer Tissue Bank and Cancer Registry (Table
1 and Additional file
1: Table S1 of reference [
1]). Patient consent was obtained for use of all specimens.
Table 1
Patient characteristics
AB003 | Retromolar Region | N0 | Yes |
AB004 | Gingiva | N0 | Yes |
AB007 | Floor of Mouth | N+ | Yes |
AB010 | Tongue | N0 | Yes |
AB011 | Tongue | N0 | Yes |
AB014 | Retromolar Region | N+ | Yes |
AB015 | Tongue | N+ | Yes |
AB017 | Buccal Mucosa | N+ | Yes |
AB018 | Floor of Mouth | N+ | Yes |
AB019 | Floor of Mouth, tongue | N+ | Yes |
AB020 | Hard Palate | N+ | Yes |
AB021 | Tongue | N0 | no |
AB023 | Tongue | N0 | Yes |
AB025 | Gingiva | N0 | No |
AB026 | Retromolar Region | N0 | Yes |
AB029 | Floor of Mouth, tongue, buccal mucosa | N0 | Yes |
AB031 | Tongue | N0 | Yes |
AB032 | Buccal Mucosa | N0 | Yes |
AB033 | Retromolar Region | N+ | Yes |
AB034 | Buccal Mucosa | N0 | No |
AB035 | Tongue | N0 | No |
AB039 | Gingiva | N0 | Yes |
AB041 | Tongue | N0 | Yes |
AB042 | Tongue | N0 | Yes |
AB045 | Gingiva | N0 | No |
AB048 | Tongue | N0 | Yes |
AB049 | Tongue | N0 | No |
AB051 | Tongue | N+ | Yes |
AB054 | Buccal Mucosa | N+ | Yes |
AB055 | Floor of Mouth | N0 | Yes |
AB056 | Retromolar Region | N+ | Yes |
AB059 | Tongue | N+ | Yes |
AB060 | Tongue, Floor of Mouth | N+ | Yes |
AB061 | Buccal Mucosa | N0 | Yes |
AB062 | Gingiva | N0 | No |
AB063 | Tongue | N0 | Yes |
AB064 | Buccal Mucosa | N0 | Yes |
AB066 | Tongue | N0 | Yes |
AB067 | Floor of Mouth | N0 | Yes |
AB068 | Gingiva | N0 | No |
AB070 | Floor of Mouth | N0 | Yes |
AB071 | Hard Palate | N0 | Yes |
AB073 | Gingiva | N+ | No |
AB077 | Floor of Mouth | N0 | Yes |
AB079 | Tongue | N0 | No |
AB080 | Tongue | N0 | No |
AB081 | Gingiva | N+ | Yes |
AB082 | Floor of Mouth | N+ | Yes |
AB083 | Floor of Mouth, tongue, gingiva | N+ | Yes |
AB084 | Gingiva | N+ | Yes |
AB085 | Tongue | N0 | Yes |
AB086 | Floor of Mouth | N0 | No |
Tumor cell lines
Human oral tongue SCC cell lines SCC4, SCC9, SCC15 and SCC25 were obtained from the American Type Culture Collection (Manassas, VA), BICR16, H357, H103, PE/CA-PJ15, and PE/CA-PJ49 from the Health Protection Agency Culture Collections (HPA, Salisbury, UK), CAL33 from Deutsche Sammlung von Mikroorganismen und Zellculturen GmbH (DSMZ, Braunschweig, Germany), and OSC20 from the Japanese Collection of Research Biosources (Osaka, Japan). The DOK cell line, derived from a human oral dysplasia was obtained from the Health Protection Agency Culture Collections (HPA, Salisbury, UK) and HaCaT, a skin keratinocyte line was from the Deutsche Sammlung von Mikroorganismen und Zellculturen GmbH (DSMZ, Braunschweig, Germany). Cells were propagated according to the methods recommended by the suppliers.
Bisulfite conversion and pyrosequencing
The DNA concentration was quantified by Quant-iT™ dsDNA BR Assay (Life Technologies, Grand Island, NY). A total of 200 ng of each DNA sample was bisulfite converted with the EZ DNA Methylation-Direct Kit (Zymo Research, Orange, CA). EpiTect Control DNA (QIAGEN, Germantown, MD) was used as methylated and unmethylated control DNA. PyroMark assays (QIAGEN, Germantown, MD) were used to determine methylation status of HOXA9 (Hs_HOXA9_05_PM) and MT1A (HS_MT1A_02_PM). A custom assay was designed for HOXA11 (Forward: biotin-5’-AGAGGTAGGTAGGGAAGATG-3’ , Reverse: 5’-CCCCTCCCATAAACTTACTCTAAA-3’ , Sequencing: 5’-ACACTCTCTCATTCATAATC-3’). Bisulfite PCR was performed using the PyroMark PCR kit (QIAGEN, Germantown, MD) and amplification was carried out by an initial incubation at 95°C for 15 min, followed by 45 cycles of 94°C for 30 sec, 55°C for 30 sec, 72°C for 30 sec. A final incubation was carried out at 72°C for 10 min. The biotinylated PCR product was purified and subjected to pyrosequencing using the PyroMark Q24 System and PyroMark Q24 gold reagents (QIAGEN, Germantown, MD). Data were analyzed by PyroMark Q24 2.0.6 software. The HOXA9 assay (Hs_HOXA9_05_PM) includes three CpG islands. Methylation level was assigned as the mean of the three sites.
RT-PCR
Total RNA was extracted from cell lines using TRIzol® Reagent (Life Technologies, Grand Island, NY) according to the manufacturer’s instructions. RNA quantity was determined with a NanoDrop 2000 Spectrophotometer (Thermo Fisher Scientific Inc., Asheville NC, US) and RNA integrity was assessed with the Bioanalyzer™ (Agilent Technologies, Inc., average RIN for cell lines = 8.2). A fixed amount of total RNA (500 ng) per each sample was reverse transcribed with iScript™ Select cDNA Synthesis kit (Bio-Rad, Hercules CA, US).
Standard TaqMan qRT-PCR Gene Expression assays were conducted in triplicate to quantify HOXA9 expression levels relative to GUSB. Duplex PCR was performed with the FAM labeled Taqman assay for HOXA9 (Hs00365956_m1) and VIC labeled Taqman assay for GUSB (Hs00939627_m1). Reactions (10 μL per well) included 5 μL TaqMan Gene Expression Master Mix (Life Technologies, Grand Island, NY), 0.5 μL 20× Gene Expression Assay Mix, and 1 μL cDNA diluted to a final concentration of 10 ng/μL. Assay plates were run on an Applied Biosystems 7900HT detection system using standard settings (cycling program included 2 min incubation at 50°C and 10 min incubation at 95°C followed by 40 cycles of 95°C for 15 sec and 60°C for 1 min). Data values (Cycle Threshold [Ct] values) were extracted from each assay with the SDS v2.0 software tool (Life Technologies, Grand Island, NY). Gene expression values were derived from the equation: ΔCt = (Ctgene–CtGUSB) and expressed as 2-ΔCt.
For end point PCR, the primer sequences used to amplify
GUSB were 5’-TGCGCACAAGAGTGGTGCTGA-3′ and 5′-TCGACCCCATTCACCCACACGA-3′. The primers for
BRCA1 have been reported previously [
4]. Amplification reactions used the HotStarTaq Master Mix kit (QIAGEN, Germantown, MD). An initial incubation at 95°C for 15 min was followed by 45 cycles of 94°C for 30 sec, 55°C (
BRCA1) or 60°C (
GUSB) for 30 sec, 72°C for 30 sec and a final incubation at 72°C for 10 min. Nucleotide free water was used for the negative control and Universal Human Reference RNA (Agilent technologies, Santa Clara, CA) was used for the positive control.
Establishment of the SCC4 cell line with the HOXA9 inducible-expression construct
To generate stable, inducible cell lines expressing HOXA9 upon doxycycline induction, HOXA9 cDNA (BC10023, Open Biosystems) was subcloned into pLVX-tight-puro (Clontech, CA). SCC4 cells were transfected with pLVX-Tet-On and pLVX-HOXA9 using Lipofectamine LTX (Life Technologies, Grand Island, NY)) according to the manufacturer’s instructions. Three days after transfection, stable clones were selected by culturing for 2 weeks in medium supplemented with 1 μg/mL Puromycin (Sigma-Aldrich) and 300 μg/mL G418 (Roche). These SCC4-HOXA9 cells were cultured with 1 μg/mL doxycycline to induce HOXA9 expression, which was verified by qPCR and western blotting. Empty vector control SCC4 cells (SCC4-empty vector) were generated by transfecting with pLVX-Tet-On and pLVX-tight-puro and selection by culturing for 2 weeks in medium supplemented with 1 μg/mL puromycin and 300 μg/mL G418.
To assess cell proliferation, 500 SCC4-HOXA9 and SCC4-empty vector cells were seeded in 96-well culture plates. After allowing cells to attach and grow for 24 hours, 100 μL of culture medium supplemented with 1 μg/mL doxycycline was added to the cultures. Plates containing six replicate wells of each cell type were harvested over a period of nine days and proliferation was measured using the CyQUANT® NF assay (Life Technologies, Grand Island, NY).
Colony formation assays were performed by seeding cells in six-well plates at a density of 1000 cells/well in 2 mL complete medium. After culturing for two weeks, colonies were stained with 0.5% crystal violet in 2% ethanol and colonies were counted.
Statistical analysis
A Wilcoxon rank sum test was used to assess differential methylation between groups. A T-test was used to assess differences in proliferation and colony formation assays.
Discussion
The risk of metastasis in oral cancer patients is associated with the status of copy number alterations at chromosome 3q, 8p, 8q and 20. The presence of any one of these aberrations (3q8pq20 genomic subtype) is associated with a substantial risk of metastasis (46%), while absence of alterations in all of these regions (non-3q8pq20 subtype) shows greater promise as a biomarker for low risk of metastasis (93% NPV). If validated in further larger clinical studies, determination of non-3q8pq20 status prior to surgery could identify those patients at low risk of metastasis who could be spared the extra surgery of an elective neck dissection. While non-3q8pq20 status could be determined prior to surgery by profiling tumor biopsies for copy number alterations, assignment of non-3q8pq20 status might be ambiguous if no copy number changes were present on other chromosome arms. Therefore, we investigated whether differential methylation of loci could act as a surrogate and identify 3q8pq20 and non-3q8pq20 subtypes. Two of the candidate loci (HOXA11 and MT1A), selected from analysis of published data, failed to validate in our cohort. Only HOXA9 was found to be differentially methylated in 3q8pq20 compared to non-3q8pq20 tumors, as well as between node positive and node negative cases. Although the differences in methylation level reached statistical significance, they are modest and would probably not result in a robust clinical test for nodal status.
Homeobox (HOX) genes are transcription factors with roles in development, regulating patterning during embryogenesis and maintaining differentiated states. De-regulated expression of HOX genes is reported in cancers [
5]. They can be overexpressed and act as oncogenes or they can act as tumor suppressors with expression down regulated via promoter methylation. Our data are consistent with
HOXA9 acting as a tumor suppressor in oral cancer. Methylation of
HOXA9 has been reported previously in oral cavity cancer [
6], and methylation and loss of expression of
HOXA9 reported in breast [
4,
7,
8], lung [
9], ovarian [
10] and bladder cancer [
11], whereas
HOXA9 is well known to act as an oncogene in leukemia [
5]. The tumor suppressor function of
HOXA9 has been extensively investigated in breast cancer where it has been shown that
HOXA9 directly regulates
BRCA1[
4] and a number of other genes involved in invasion, growth and metastasis [
7]. While we show here that
HOXA9 also appears to positively regulate
BRCA1 expression in oral cancer cells, further studies will be required to fully understand how
HOXA9 functions as a tumor suppressor in oral cancer. Indeed, the oncogenic and tumor suppressive activities of deregulated
HOXA9 expression in different tissues highlight the importance of tissue context for the functioning of deregulated developmental genes in cancer.
Acknowledgements
We thank members of the Helen Diller Family Comprehensive Cancer Center Genome Analysis shared resource for their help with quantitative RT-PCR assays and Teruyuki Muraguchi and Mauting Lin for assistance with preparation of HOXA9 expressing SCC4 cells. Some of the pyrosequencing work was performed at the Protein and Nucleic Acid facility, Stanford University.
Funding
This work was supported by NIH grants CA118323 and CA131286 to DGA, and CA113833 to BLS. AB was the recipient of a pre-doctoral fellowship from the California Tobacco-Related Disease Research Program (18DT-0011).
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
KU, RV and BH carried out the molecular assays and cell culture experiments. AB, BLS and DGA assembled the patient cohort and associated clinical and molecular information. KU and DGA drafted the manuscript. DGA conceived of the study, and participated in its design and coordination. All authors read and approved the final manuscript.