Patients and tissue samples
At the Hospital of Stomatology, Sun Yat-Sen University, we collected 91 oral squamous cell carcinoma tissue samples and 10 adjacent non-tumor tissue samples between September 2016 and January 2020. The cancer patients were selected based on their clinical and pathological information. Patients who had undergone preoperative radiotherapy or chemotherapy were excluded. We took two identical tissue samples from each patient. One was stored at -80℃ for RNA extraction, and the other was fixed in formalin and embedded in paraffin for immunohistochemical analysis. All patients signed informed consent forms, and the study was approved and supervised by the Institutional Research Ethics Committee of the Hospital of Stomatology, Sun Yat-Sen University.
Cell lines, Cell Culture, and Transfection
The human OSCC cell lines HN6, HSC3, CAL33, SCC15, and SCC25 were purchased from The America Type Culture Collection. HSC3 and CAL33 cells were cultured in Dulbecco’s modified Eagle’s Serum (DMEM, Gibco, USA) supplemented with 10% fetal bovine serum (FBS, Gibco, USA). HN6, SCC15, and SCC25 cells were grown in DMEM/Hams F12 medium (DMEM/F12, Gibco, USA) containing 10% FBS. NOK cells were grown in keratinocyte SFM medium mixed with growth factor. All cells were cultured at 37℃ in a humidified incubator with 5% CO2.
RNA interference and transfection were conducted to test the function of SHMT2 in OSCC. The short-interfering RNA was synthesized by Ruibo Biotechnology Co., Ltd. (Guangzhou, China). The sequences specific to SHMT2 were:
HN6 and HSC3 cells were cultured in six-well plates. After the cell density reached 80%, the cells were washed twice with PBS. The transfection reagent was then prepared. For mixture one, 5 µl lipofectamine RNAi-MAX Transfection Reagent (Invitrogen; Thermo Fisher Scientific Inc.) was added to 125 µl Opti-MEM (Invitrogen; Thermo Fisher Scientific Inc.) medium according to the manufacturer’s instructions. For mixture two, 5 µl siRNA was diluted with 125 µl Opti-MEM. Mixture one and mixture two were mixed in one tube and incubated for 10 min at room temperature. After incubation, cells were grown with 250 µl of the mixture and 750 µl complete medium at 37℃ with 5% CO2 for 12 h. Then 1 ml of complete medium was introduced into the transfection medium. Cells were collected 48 h after transfection for subsequent experiments.
Immunohistochemistry (IHC) was performed as follows. The paraffin-embedded tissues were cut into 4.0 mm sections for dewaxing and dehydration. Staining was carried out using antigenic retrieval and goat serum block (AR0009, Boster Bio, Wuhan, China). The sections were incubated with primary antibody against SHMT2 (1:500, GeneTex), Ki-67 (1:200, ab16667, Abcam), p21Cip1 (1:50, 12D1, Cell Signaling), p27Kip1 (1:50, ab32034, Abcam), and CyclinD1 (1:50, E3P5S, Cell Signaling), CDK4 (1:500, D9G3E, Cell Signaling) at the recommended dilution overnight at 4℃. Staining was then developed using diaminobenzidime (DAB, GK600510, Gene Tech, China), and the cell nuclei were stained with hematoxylin (D006, Nanjing, Jiancheng, Biotech, China). The staining score was evaluated by multiplying the staining intensity and the percentage of positive cells. The intensity of staining was scored as 0 = negative, 1 = weakly positive, 2 = positive, 3 = strongly positive. The percentage of positive cells were defined as 0% = 0; <10% = 1; ³10 to <50% = 2; ³50% to <75% = 3; ³75% = 4. The total IHC scores ranged from 0–12 and tissues with a score ³6 were assigned to the high expression group, while those with scores <6 were placed in the low expression group.
Western blot analysis
Cells were lysed in RIPA buffer (CW2333S, CWBIO, China) supplemented with protease inhibitor cocktail set 1 (539131, Millipore, USA) and phosphatase inhibitor. Protein concentrations were measured using BCA protein assay kits (CW0014S, CWBIO). Equal amounts of total cell lysate were separated using 10% SDS-PAEG (P0014B, CWBIO) and transferred to PVDF membranes (ISEQ00010, Millipore, USA) blocked with 5% bovine serum albumin (0332, Ameresco) for 1 h at room temperature. Then the blots were incubated with the primary antibodies against SHMT2 (1:1,000, GeneTex), E-cadherin (1:1,000, 24E10, Cell Signaling), ZO1 (1:1000, D7D12, Cell Signaling), N-cadherin (1:1,000, D4R1H, Cell Signaling), Vimentin (1:1,000,D21H3,Cell Signaling，b-catenin (1:1,000, D10A8, Cell Signaling), p21Cip1 (1:1000, 12D1, Cell Signaling), p27Kip1 (1:1000, D69C12, Cell Signaling), Cyclin D1(1:1000, E3P5S, Cell Signaling), CDK4(1:1000, D9G3E, Cell Signaling), b-actin (1:1000,13E5, Cell Signaling), and GAPAH (1:1,000,D16H11, Cell Signaling) at 4℃ overnight. Membranes were incubated in HRP-conjugated secondary antibody (1:2000, 7074S, Cell Signaling) for 1 h at room temperature. The protein bands were exposed using a chemiluminescent substrate (WBKLS0500, Millipore, USA), and quantitatively analyzed using ImageJ.
Quantitative reverse transcription polymerase chain reaction (qRT-PCR)
Total RNA was extracted from the cultured cells and tissues using RNAzolâ RT (RN190, Molecular Research Center, USA) according to the manufacturer’s protocol. The quality and quantity of the RNA were examined using a NanoDropâ ND-1000 Spectrophotometer (Thermal Fisher, Wilmington, DE, USA). RNA was reverse transcribed into cDNA using the PrimeScript RT reagent Kits (RR036A, Takara Bio, Japan). All samples were tested using SYBR Master Mix (11201ES208, Yeasen, China). The quantity of mRNA was calculated using method 2−DDCt and normalized to GAPDH. The prime sequences were as follows:
The proliferation of OSCC cell lines was detected using cell counting Kit-8s (CCK-8, 40203ES80, Yeasen, China), at 0, 24, 48, 72, 96 h. In brief, 2 × 103(HN6) or 3 × 103 (HSC3) cells/well were seeded in 96-well culture plates, in triplicate. At the end of the experiment, 100 ml of 10% CCK-8 reagent (10 ml CCK-8 and 90 ml serum-free media) was added to each well, and the plates were incubated for 1 h at 37℃. Absorbance was measured at 450 nm using a microplate reader (Bio-Read, USA).
Migration and invasion assays
Cell motility was evaluated by migration and invasion assays using Matrigel (10 mg/ml, 354234, Corning, China) and trans-well filters (pore size, 8µm; BD Biosciences). And for invasion assay, 0.33 mg/ml Matrigel was coated on the upper chamber of Transwell assay plates 2h in advance. While for migration assay, Matrigel was not required. Briefly, 3 × 104 HN6 cells or 4 × 104 HSC3 cells were seeded into the upper chambers with 200 µl serum-free medium, while the lower chambers were supplemented with 600 µl complete medium. After 24 hours, the cells on the upper filter were removed, while those on the lower chamber were fixed with paraformaldehyde and stained with 0.1% crystal violet. The cells in five random fields were subsequently counted under the microscope to assess the number of nuclei.
Wound healing assay
Cell migration and repair ability were assessed using wound healing assays. OSCC cells were seeded in six-well plates and cultured to cover 90% of each well. Scratches in the cell layer were made with a 1 ml sterile pipette tip, and cells were washed twice with phosphate-buffered saline (PBS). After that, cells were cultivated with serum-free medium. Cells were photographed under a light microscope at 0 h, 24 h, and 48 h. The ImageJ software was used to quantify the distance between cell edges, and to calculate the wound healing rate.
Flow cytometry analysis
Following centrifugation, 2 × 105–2 × 106 cells were collected, and the supernatant removed. The cells were resuspended in PBS, centrifuged again, and the supernatant discarded. Then 1 ml DNA staining solution and 10 µl permeabilization solution (Multi Sciences) was added to each tube, and cells were incubated at room temperature without light. Finally, cells were measured using Beckman flow cytometry and data was analyzed using FlowJo (https://www.flowjo.com/) Software.
KFluor488 Click-it EdU imaging detection kits (KeyGen Biotech, Jiangsu, China) were used to assess the cell proliferation. Cells were grown in 24-well plates, and were incubated with 50 uM EdU for 2 h at temperature. Cells were fixed with 4% paraformaldehyde in PBS and osmosis was promoted with 0.5% Triton X-100 in PBS. Click-iT reaction mixture was prepared for EdU detection, and Hoechst 33342 was used to re-stain DNA. Finally, the cells were photographed with an inverted fluorescence microscope and counted using ImageJ software, based on the stained nuclei.
Lentiviral Short Hairpin RNA (shRNA) Transfection
A stable knockdown of SHMT2 expression using shRNA was established, to detect the function of SHMT2 in vivo. The lentivirus packaging process was carried out by Cyagen US Inc. Lentivirus carrying shRNA targeting SHMT2 was used to achieve stable knockdown of SHMT2 expression. The sequence of the targeting vector was:
HN6-shSHMT2-1: 5′-CTTCGAGTCTATGCCCTATAACTCGAGTTATAGGGCATAGACTCGAAG -3′
Female five-to six-week-old NOD/SCID mice were randomly divided into two groups, the sh-SHMT2 group and the control group, with eight mice in each group. Then, 2 × 106 HN6 cells suspended in 200 ml PBS were subcutaneously inoculated into the left inner thighs of mice. Transplanted tumors were observed weekly, and the tumor xenografts were harvested, photographed, and fixed for the IHC four weeks after injection. Tumor volume was calculated using the following formula: volume = (width2 × length)/2. All animal experiments were approved by Sun Yat-Sen University’s Animal Experiment Ethics Committee and conducted in accordance with their principles of animal welfare.
Data collection and analysis
OSCC samples (313) and normal oral mucosal tissue samples (30) were selected from the head and neck squamous cell carcinoma datasets in the Cancer Genome Atlas (TCGA) database (https://www.cancer.gov/tcga), according to the OSCC classification criteria. Carcinomas originated from the tongue, floor of the month, buccal mucosa, hard palate, alveolar ridge, and oral cavity. Samples with complete clinical information were selected for further analysis. The 313 OSCC samples selected were divided into an SHMT2 low-level group and a high-level group, according to the median expression level of SHMT2. The R software (http://www.R-project.org/) was used to perform survival analysis, weighted gene co-expressed network analysis (WGCNA) and gene set enrichment analysis (GSEA). The results of the bioinformatic analysis of the TCGA database were validated using GSE30784 from the Gene Expression Omnibus database (https://www.ncbi.nlm.nih.gov/).
Weighted gene co-expressed network analysis
The top 50% genes with the highest variance were selected among 10215 genes for WGCNA analysis using the R “WGCNA” package. A gene expression similarity matrix was constructed with which to analyze the cooperative expression between genes. An adjacency matrix was built based on the above matrix, and an appropriate soft threshold was introduced to establish a scale free network. A topological overlap matrix (TOM) was constructed by defining a measure of node dissimilarity. Network modules were then identified using dynamic hierarchical tree clustering. Finally, we selected the most relevant module with which to further analyze the correlation between modules and clinical parameters in OSCC using Gene Ontology annotation (GO)[22, 23] and the Kyoto Encyclopedia of Genes and Genomes pathway (KEGG) enrichment analysis.
Gene set enrichment analysis
GSEA software (http://software.broadinstitute.org/gsea/index.jsp) was used to conduct the enrichment analysis on the groups with high expression of SHMT2 and low expression of SHMT2 to evaluate the trend of distribution of SHMT2 in the gene table of OSCC biological phenotypic relevance sequencing, and to determine its contribution to the phenotype. A P-value < 0.05 and a false discovery rate (FDR) <0.05 were considered as credible enrichment.
All in vitro experiments were repeated at least three times. The mean and the standard deviation of data were analyzed by GraphPad Prism 8.0 software. The difference between the two groups was analyzed using two-tailed Student’s t-tests, and ANOVAs were used for the multiple groups. The correlations between SHMT2 expression and the clinicopathological features of the OSCC patients were analyzed using chi-square tests or Fisher’s exact tests. Survival curves were constructed using the Kaplan-Meier method and tested with log-rank. SPSS 25.0 was applied to establish a Cox proportional hazard regression model for univariate and multivariate analysis. The differences were considered statistically significant at p < 0.05 (* p < 0.05, ** p < 0.01, *** p < 0.001).