The human tongue carcinoma cell line Tca-8113 was provided by Jilin Cancer Hospital (Changchun, China) and CAL-27 cells were provided by the Hospital of Stomatology, Jilin University (Changchun, China). Cells were cultivated in Iscove's modified Dulbecco's medium (Gibco; Thermo Fisher Scientific, Inc., Waltham, MA, USA) containing 10% fetal bovine serum (Gibco; Thermo Fisher Scientific, Inc.) with 100 units of penicillin, maintained at 37°C in a 5% CO₂ humidified atmosphere, according to the recommendation of the supplier.

A total of 8 tongue carcinoma tissue samples were provided by the Tissue Bank of China-Japan Union Hospital, Jilin University (Changchun, China). The clinicopathological characteristics of the patients were summarized in Table I. The present study was approved by the Ethics Committee of the China-Japan Union Hospital, Jilin University (Changchun, China).

The cell lines were recovered from liquid nitrogen and cultured for 72 h. A total of 50–100 mg tissue samples were homogenized in 1 ml TRIzol reagent (Invitrogen; Thermo Fisher Scientific, Inc.). Total RNA was extracted from the cells and each tissue sample using TRIzol reagent following the manufacturer's protocol. DNase I (Sangon Biotech Co., Ltd., Shanghai, China) was used to eliminate genomic DNA contamination. The concentrations and purity of the isolated RNA were measured by NanoDrop 2000 (Thermo Fisher Scientific, Inc.).

The cDNA synthesis reaction was performed three times using the M-MuLV First Strand cDNA Synthesis kit (Sangon Biotech Co., Ltd., Shanghai, China) according to the manufacturer's protocol. The total reaction volume was 20 µl. Total RNA (1 µg), 1 µl random primer and RNase free water were mixed, incubated at 65°C for 5 min and then cooled down immediately on ice for 30 sec. The rest of the reaction reagents were added, then the mixture was incubated at 42°C for 60 min and the reaction was terminated by heating at 70°C for 10 min.

The primers of 12 putative reference genes were selected based on previous studies, and are widely used and recognized to be good reference genes (18,19). The primers were synthesized by Sangon Biotech Co., Ltd. and the sequences are listed in Table II. A Roche LightCycler 480 detection system (Roche Diagnostics GmbH, Mannheim, Germany) was used for RT-qPCR. Reactions were performed using 2xSG Fast qPCR Master Mix (Sangon Biotech Co., Ltd.) according to the manufacturer's protocol. All the samples were run in triplicate. The PCR volume was 20 µl, containing 2 µl cDNA. The following cycling conditions were used: 55°C for 5 min; 95°C for 5 min; 40 cycles of 95°C for 20 sec, 55°C for 20 sec and 72°C for 4 min. This cycle was followed by melting curve analysis, and the baseline and cycle threshold values (Cq values) were automatically determined for all the plates using Roche LightCycler 480 software v1.5.0 (Roche Diagnostics GmbH). RT-qPCR amplification products were detected by 1% agarose gel electrophoresis and melting curve to verify the specificity of the primers. A standard curve was constructed for each primer pair to determine the product specificity.

The Cq values were identified by quantitative comparison of the amplification of the candidate genes. The Cq values were calculated to relative quantities (Q) for data analysis, in view of the PCR efficiencies of the candidate genes according to the equation: Q=2^−ΔΔCq (20).

A random pool of cDNA from the samples was selected and used for 10-fold serial dilutions, ranging between 0.001 and 1X. PCR analysis was run in triplicate, as mentioned previously. PCR efficiency was calculated using the slopes of the calibration curve and by the formula: E=10⁻¹/^slope.

All the samples were divided into three groups: Cell line + tissue group, cell line group and tissue group. In order to evaluate the stability of the reference genes, three frequently used software programs, geNorm v3.5 (http://medgen.ugent.be/~jvdesomp/genorm/), NormFinder v0.953 (http://moma.dk/normfinder-software) and BestKeeper version 1 (http://www.gene-quantification.de/bestkeeper.html), were utilized. GeNorm is designed to establish reference genes for RT-qPCR and is used to analyze and determine the M-value, which refers to the stability of the reference gene expression. M is the mean pairwise variation for a given gene compared with other tested genes, following stepwise exclusion of the gene with the highest M value and calculated in order to select the most two stable genes. The default value suggested by geNorm is M=1.5. A higher M-value indicates less stable expression, and a lower M value indicates more stable expression. If M >1.5, the gene is not suitable for use as a reliable reference gene. GeNorm software was also used to analyze the pair-wise variation value of the normalization factor (V), which has a default value of 0.15. It is possible to use the value of Vn/Vn+1 to determine whether adding a novel reference gene affects the normalization factor. If the value of Vn/Vn+1 is >0.15, it is necessary to use the n+1 reference genes as internal controls. If it is <0.15, then it is not necessary to use novel reference genes. NormFinder software is a tool designed to identify the optimal reference gene among a set of candidates and it has a similar operation principle to geNorm. This program analyzes expression data, ranks the set of candidate normalization genes according to their expression stability and considers the gene with the minimum expression data as the most stable gene. It is also possible to use this software to compare the stability of inter and intragroup reference genes and propose an optimal combination of two genes. BestKeeper evaluates candidate reference gene stability based on the correlation coefficient (R-value). The genes were ranked according to their R value, with a higher R value indicating a more stable and reliable gene.

The primer sequences, corresponding length of the amplified products and PCR amplification efficiency are listed in Table II. The gel imaging system indicated that the size of the amplified fragment was consistent with the expected size, with a clear band and without primer dimers and nonspecific bands (Fig. 1A). In addition, melting curve analysis of each gene fragment amplified by qPCR revealed that all curves exhibited a single signal peak (Fig. 1B). For the candidate reference genes, the amplification efficiency range of the standard curve was 1.95–2.09 and all correlation coefficients were >0.96.

The expression level of the candidate reference genes was determined by the Cq value, which is inversely proportional to the expression level of the gene. Higher Cq values indicated smaller expression quantities. The Cq value of all the samples ranged between 7.70 and 32.93 (Fig. 2). In all groups, 18S had the smallest mean Cq values of 9.48±1.51 (cell line + tissue group; Fig. 2A), 8.38±0.43 (cell line group; Fig. 2B) and 10.30±1.52 (tissue group; Fig. 2C) and HMBS had the greatest mean Cq values of 26.41±2.37 (cell line + tissue group; Fig. 2A), 25.06±1.70 (cell line group; Fig. 2B) and 27.43±2.37 (tissue group; Fig. 2C).

Theoretically, the 12 reference genes constituted appropriate internal controls for gene expression. In the cell line + tissue group, ALAS1 and RPL29 had the lowest M-values, followed by GUSB, which suggested that these were the most stable candidate genes for studies between human tongue carcinoma cell lines and tissue. In the cell line group, B2M and RPL29, followed by TBP, were suggested as the most stable reference genes for studies between Tca-8113 and CAL-27 cell lines. In the tissue group, RPL29 and HPRT1, followed by ALAS1, were suggested as most stable reference genes for studies on human tongue carcinoma tissue (Fig. 3A). A combination of 2 and 3 reference genes were optimal in the cell line group (V2/3=0.116) and tissue group (V3/4=0.103), respectively (Fig. 3B).

In order to further evaluate the stability of the 12 reference genes, the present study also used the NormFinder program. PPIA + HMBS was the most stable reference gene combination in the cell line + tissue group, whilst GUSB and RPL29 were the most stably expressed genes in this group (Fig. 4A). In the cell line group, B2M + HMBS was the most stable reference gene combination, whilst PUM1 and GADPH were the most stably expressed genes (Fig. 4B). In the tissue group, PPIA + HMBS was the most stable reference gene combination, whilst HMBS and GUSB were the most stably expressed genes (Fig. 4C).

The BestKeeper program was also used to compare the stability of internal reference genes. As the BestKeeper program is only able to analyze 10 internal reference genes, the 2 most unstable internal reference genes indicated by the geNorm software were removed in each group. In terms of the R-value, the most stable internal reference gene in the cell line + tissue group was GUSB, followed by ALAS1 and RPL29 (Fig. 5A). In the cell line group the most stable internal reference gene was RPL29, followed by B2 M and GAPDH (Fig. 5B), and in the tissue group the most stable internal reference gene was RPL29, followed by GUSB and HPRT1 (Fig. 5C).