Study population
Patients with suspected oral lesions of OSCC were enrolled. After signing the informed consent, the subjects were interviewed using a standard questionnaire that requested information about socio-demographic, medical, and lifestyle factors. The patients from Department of Head and Neck surgery of The National Cancer Institute,Dental school of Universidad de Valparaíso,Dental school of Universidad del Desarrollo (Chile), The Hospital Lencinas and the Servicio de Estomatología y Medicina Bucal Dental school, Universidad Nacional de Cuyo (Argentina); received a routine intraoral examination and oral mucosal biopsies were taken and classified according to thediagnosis and POI in three groups:oral epithelial dysplasia, less invasive OSCC (POI type 1 and 2) and invasive OSCC (POI type 3, 4 and 5)group. Seventy-one cases of primary OSCC diagnosed over a period of 2 years (2017–2019) were included in the study. None of the patientshad received any tumor specific therapy (chemotherapy or radiotherapy) before the resection.Twenty cases diagnosed as inflammatory lesions and histologically confirmed with normal mucosal margins from the resection specimens were included as control group in the qPCR analysis. The Ethics Committee of the School of Medicine of Universidad del Desarrollo (FM-UDD CAS), National Cancer Institute of Chile and Medicine School of Universidad Nacional de Cuyo (FCM-UNCuyo) approved this study according to Declaration of Helsinki to experimentation with human subjects.
Histopathologicalanalysis
TheOral biopsieswere fixed in 10% buffered formalin (Merck, USA), embedded in paraffin (Merck), and sectioned. Tissue sections of 4 μm were deparaffinized with Neoclear (Merck), rehydrated with graded alcohols, stained with hematoxylin–eosin (H&E, Merck), and visualized with a light microscope (DM2000; Leica, Germany). Images were captured with a digital camera (DFC295; Leica). Samples were classified according to the revised criteria given by the World Health Organization (2005). Three independent observers performed histological analyses blind; one of them is a pathologist expert in oral diseases(42, 43).
Immunohistofluorescenceanalysis
Tissue sections of 4 μm were deparaffinized, rehydrated, blocked with 5% FBS (Gibco, USA) dissolved in PBS 1X (Gibco, USA) and incubated overnight at 4 °C with a dilution 1:50 of antibodies foranti-SPINK7 (Abcam, ab122326, USA), anti-HER2(BD PharmigenTM, #554299, USA), anti-p53 (Abcam, (PAb 1801 ab28, USA) andanti-pRB (8516S, Cell signaling, USA). Then, samples were washed with PBS 1X and incubated two hours at room temperature with a dilution 1:400 of Alexa488-conjugated goat anti-mouse IgG or Alexa 555-conjugated rabitt anti-mouse IgG (Cell Signaling, USA). Cross-reactivity of the secondary antibody was testedincubating samples without the primary antibody. Nuclei were counterstained with a dilution 1:1500 of DAPI (Sigma, Aldrich) in PBS 1X. Samples were embedded in fluorescence mounting medium S3023 (Dako cytomation, USA) and scanned in a confocal microscope (Olympus). Five representative optical sections by sample (n=6/group) were photographed using 60X magnification. The images obtained per field of each sample, were processed with the same conditions and the positive protein signal (pixels intensity) was analyzed and quantified using Fiji Image J software (NIH, USA)(42, 43).
Confocal microscopy analysis
A Gaussian filter of 1 was applied anda constant background value of 150 wassubtracted for each image. The same threshold value was set for each channel including the structures of interest and the corresponding masks were obtained. The yellow pixels (red and green pixels overlap) versus the total pixels were quantified and the colocalization was measured with Coloc2 plugin(FijiImageJ)(44).
Gene expression analysis
Total RNA was isolated from theoral biopsies. The mRNA was purified using RNEasy PlusMini Kit (Qiagen, Germany). Contaminating genomic DNA was degraded with 1 U of DNAse RQ1 (Promega). One μg of RNA was reverse transcribed for 60 min at 42 °C using 200 U M-MLV reverse transcriptase (Invitrogen) and 0.5 μM oligo-dT primers (Invitrogen). Real time PCR was performed in a final volume of 10 μL containing 50 ng of cDNA, Power SYBR Green PCR master mix (Life Technologies, Grand Island, NY) and 0.5 μM of each specific primer, using the Step One Plus PCR system (Life Technologies). Controls without reverse transcriptase were included. Amplicons were analyzed according to their size and melting temperature (Supplementary Table 1, S1). To normalize data, 18S RNA and β-actin were used as reference genes. The RNA level of a target gene was calculated using the 2ΔCt method and graphed as fold change(45).
Geneexpressions TCGA profile
The data studied was programmatically extracted from the publicly available data set of OSCC from The Cancer Genome Atlas Project (TCGA) on May, 2019 using the recount2 platform (https://jhubiostatistics.shinyapps.io/recount/). Non-standardized RNASeq gene expression levels from 548 samples were downloaded. Samples from oral cavity were selected obtaining a final subset of 332 tumor samples and 32 non-tumoral tissue samples. RNA expression levels were evaluated for 6 genes (NFKB1, RB1, TP53, ERBB2, CYP4B1, SPINK7). Crude counts were scaled by the total coverage of the sample (area under the curve, ‘AUC’) and differential gene expression analysis (DGE) was performed using the generalized linear model method of the EdgeR R package comparing non-tumor versus tumor samples(46). Log2 Fold change values were obtained associated with exact p-values and False Discovery Rate values (FDR). To evaluate gene expression correlation, data was transformed using Voom conversion from the R limma package, allowing normal linear modeling of the RNA counts. Afterwards, pairwise Pearson’s product-moment correlation analysis was performed for the aforementioned genes and p-values were calculated(47).
Gene mutationsTCGA profile
The mutational analysis of OSCC, data was programmatically downloaded using the TCGAbiolinks package of Bioconductor (48). Mutation Annotation Format (MAF) files with aggregated mutation information generated from whole-exome sequencing were downloaded.From 546 samples of Head and Neck cancer, 329 samples of OSCC were obtained. The maftools Bioconductor package was used to analyze and visualize the MAF files(49). An Oncoplot was drawn showing the variants (SNP) of the 15 most mutated genes in OSCC, followed by 5 genes of interest (RB1, ERBB2, NFKB1, CYP4B1 and SPINK7)(47).
Statistical analysis
The population distribution of the samples from our patientswas non-parametric.Comparisonsof gene and protein expression among the groups were performed using One-way Kruskal-Wallis test and Dunn's test as post-test. Stat Graph Prism 5.0 software was used for statistical analysis. Data are presented as median ± SEM, andp<0.05 was considered statistically significative.