Eukaryotic expression vector of pE-Xb which only expressed XAGE-1b protein was constructed from pEGFP-N1, and pEGFP-XP-GFP constructed from pEGFP-N1 is a fusion protein expression vector combining with GFP. All the primer sequences are shown in Table I. The shRNA expression vector especially interfering with XAGE-1b expression was designed and constructed to pGCsi-U6/Neo/RFP. All the single oligo nucleotide sequences are shown in Table II. The primers used for detecting XAGE-1b and GAPDH expression by RT-PCR are shown in Table III.

To acquire cDNA of XAGE-1b from ACC-M cells and to amplify the fragments with the primers of GN-1B-F and GN-1B-R, the fragments were inserted into pEGFP-N1 plasmid pE-Xb expression vector without GFP fusion. XAGE-1b vectors with GFP fusion were constructed with the primers, including XAGE-1b, GN-1B-F and GN-1B-R (G), named pEGFP-Xb-GFP. Negative control vector of pEGFP-N1 was named pE-Negative. The shRNA for specially interfering with XAGE-1b was designed and synthesized and constructed into pGCsi-U6/Neo/RFP vector by Jikai (Shanghai, China). There are three interfering vectors, including Sh-a, Sh-b and Sh-c, and the positive control sequence named pG-Positive is designed specially for interfering with GFP expression, the negative control sequence (anything except for sequences against any human gene), was named pG-Negative.

The nude mice (BALB/c nu/nu) were purchased from the Animal center of Shanghai, and cultivated in the animal experimental center of the Second Military Medical University (Shanghai, China). The human salivary ACC-2, ACC-M, SPC-A1 and 293T cells used in the study were obtained from American Type Culture Collection (ATCC, Manassas, VA, USA). ACC-2, ACC-M, SPC-A1 cells were cultured in RPMI-1640 from Gibco (Langley, OK, USA) supplemented with 10% fetal bovine serum (FBS) from Sigma (St. Louis, MO, USA) at 37ºC in a humidified atmosphere of 5% CO₂ in air. 293T cells were cultured in DMEM from Gibco (Carlsbad, CA, USA) supplemented with 10% FBS. All chemicals needed for cell culture were purchased from Gibco. The effective fragments of RNA interfering were chosen based on screening RNA interfering exogenous expression in 293T cells and SPC-A1 cells.

Reverse transcription-PCR. Total RNA isolated from above culture cells was prepared with TRIzol (Invitrogen, Carlsbad, CA, USA) according to the manufacturer’s instructions. RNA quantification was done using spectrophotometry. Semi-quantitative RT-PCR analysis for shRNA interfering expression and the internal control GAPDH was carried out using ABI PE9600 PCR System. The PCR products were electrophoresed on 2% agarose gel and stained with ethidium bromide (Pierce, Rockford, IL, USA). Quantitative reverse transcription PCR analysis was carried out to understand the efficacy of the shRNA interfering expression.

ACC-2 cells were collected into 1.5 ml Eppendorf tubes on ice, centrifuged at 800 rpm for 5 min at 4ºC, resuspended in 100 μl cell lysis buffer containing proteinase inhibitor, and incubated on ice for half an hour. The lysates were centrifuged at 14000 rpm for 30 min at 4ºC, the supernatants were collected and stored at −70ºC for electrophoresis. The concentration of protein sample was determined by the Lowry method. Proteins in conditioned media were separated by electrophoresis on 15% SDS-PAGE gels and transferred electrophoretically onto nitrocellulose membranes. After being blocked for 2 h with 5% skim milk, blots were incubated with 1:1,000 diluted sheep anti-human antibody XAGE-1b and 1:3000 diluted antibody GAPDH at room temperature overnight. After four washes with PBST, the membrane was separately incubated with rabbit anti-sheep antibody at a dilution of 1:2,000 and sheep anti-mouse antibody GAPDH at a dilution of 1:4,000 for 1 h and 45 min, respectively. Finally, the protein bands were visualized with an ECL kit. Quantitative analysis of the blots was performed with an imaging densitometer and GAPDH as the control. GAPDH served as an internal control for total cDNA content.

MTT assay was used for detecting cell proliferation. The stably expressing cell lines of ACC-2 and ACC-M were cultured to logarithmic growth phase and synchronized with serum-free medium for 24 h. All the cells were inoculated on a 96-well plate and measured four times every 24 h continuously. MTT (10 μl) (Sigma) was added into the plate, 4 h later, 100 μl DMSO (Sigma) was added after emptying the cell plate and the OD₅₇₀ values were measured with the SpectraMax M5 (Molecular Devices Corp., Concord, Ontario, Canada). Then the proliferation curve was drawn.

The stable expression cell lines of ACC-2 and ACC-M were cultured to logarithmic growth phase and synchronized with serum-free medium for 24 h. The cells were inoculated onto a 6-well plate ~500 cells per well with three repeat wells, and cultured with the G418-free medium. After 14 days, the medium was removed and washed two times with PBS. Crystal violet (500 μl) was added into the well and stained for 5 min, then the dye was removed and washed two times again with PBS.

The transmembrane experiment in vitro with ACC-2 cell line was carried out according to the manufacturer’s instructions of the QCMTM 24-Well Cell Invasion Assay kit (Chemicon, Billerica, MA, USA). The transmembrane rate was calculated according to the experimental results.

Experimental groups, and amount of cells are list in Table IV. The nude mice were divided into three groups. Each cell line was cultured in a bottle as 2×10⁶ cells, and the activity of cells was measured by staining with 0.2% Trypan Blue. The cells were synchronized by replacing with fresh serum-free and double-anti medium overnight before the inoculation. Cells were digested with 0.25% trypsin and counted using an automated counter before centrifugation. The concentration was adjusted to 1×10⁷ cell/ml by serum-free medium and prepared for injection. Cell suspension (100 μl) as 1×10⁶ cells was injected s.c. into the back or the tail vein of 4- to 6-week-old female BALB/C nu/nu nude mice. The condition of tumor and tumorigenesis time was observed and recorded, the long and short diameter of tumor were measured after inoculation every three days during the first two weeks and every other day during the third to the forth week. The formula for calculating the tumor is 1/2ab², a being the maximum diameter and b the minimum diameter of tumor. The nude mice were photographed and the growth curve was drawn. The mice were sacrificed 4 weeks later and photographed again, and weighed, and all were dissected, and tumor tissues were removed. The tumor and lung tissues with formalin-fixed and paraffin-embedded were preserved for identification of RNA and DNA.

H&E staining: All the tumor tissues were fixed in formalin and embedded in paraffin. Sections (5 μm thick) were obtained and used for H&E staining. Each section was photographed at ×200 magnification. IHC staining: Antigen retrieval was performed on tumor tissue of the XAGE-1b subcutaneous expression group with formalin-fixed and paraffin-embedded, and stained with the antibody XAGE-1. Follow-up staining of the sections were performed by the Ultra Sensitive S-P kit Goat system and photographed at ×200 and ×400 magnification, respectively.

Counting of the microvessel density of the tumor: the tumor tissue slides were stained by IHC with antibody CD31 (Zhongshan Co., Beijing, China), and there are specific red-brown particles precipitated on the surface of vascular endothelial cells. Five different levels were stained, and two areas at ×200 magnification were randomly selected and photographed. The number of the stained microvessels and the average number of the microvessels in five levels represent the microvessel density of tumor tissue.

Data are presented as the means ± SD. Multiple group comparisons were analyzed by one-way ANOVA after Bonferroni’s correction. Non-parametric data between two groups was computed by Chi-square test or Fisher Exact test. The difference of two groups was determined by Student’s t-test. A P-value of <0.05 was considered statistically significant.

The expression vector, including pE-Xb and pEGFP-Xb-GFP of XAGE-1b, and the empty vector pE-Negative were transfected into ACC-2 cells. There are two monoclones after transfection with pEGFP-Xb-GFP which expressed the fusion protein of GFP and XAGE-1b. The green fluorescence focused in the nucleus with the distribution of small points, and the monoclone ACC-2-pE-Negative, obtained from ACC-2 cells transfected with the empty vector as the negative control. The monoclonal cell lines express green fluorescence protein stably and the positive rate reaches 93% (Fig. 1).

The stable cell clones were obtained by G418 screening after pE-Xb transfection. The nineteenth stable expression clone showed the maximum value based on XAGE-1b expression by western blot assay and was named ACC-2-Xb-19 (Fig. 2A, upper image). The residual G418-resistant cells without cloning were continuously cultured and named ACC-2-Xb-M. XAGE-1b expression in the cell lines is different by the semi-quantitative RT-PCT assay, and XAGE-1b expression in ACC-2-Xb-19 is maximal among the cell lines, XAGE-1b expression in ACC-2-pE-Negative is minimal (Fig. 2A, lower image), and the expression in ACC-2-Xb-M is in between the two groups. The above three cell lines and ACC-2 cells were applied for studying the proliferation in vivo and in vitro. Based on the MTT assay, the number of ACC-2-Xb-19 cells increased greatly compared with ACC-2-pE-Negative after two days of culture (p<0.01, Fig. 2B). The number of mixed expression ACC-2-Xb-M cells showed the difference in the negative control at the beginning of the fourth day (p<0.05, Fig. 2B). There are differences among the three cell lines of the cloning number, and the number of ACC-2-Xb-19 is almost two times that of the negative control (Fig. 2C).

The effects of XAGE-1b on the subcutaneous tumor in ACC-2 cells were detected in vivo of the nude mice. All cells with the counts of 1×10⁶, including ACC-2-Xb-19, ACC-2-Xb-M, ACC-2-pE-Negative and ACC-2, were injected s.c. into the left back of mice, and the mice were observed and measured once every three days. The tumors formed in ACC-2-Xb-19 and ACC-2-Xb-M groups grew rapidly compared with ACC-2-pE-Negative and ACC-2 at the beginning of the fourteenth day, and the results showed the difference among the cell lines (Fig. 2D). We found that the average tumor volume of ACC-2-Xb-19, and ACC-2-Xb-M, was 4–8 times higher than the other two groups, the ACC-2-pE-Negative and ACC-2 groups (Fig. 2E), and the tumors weight was ~7–11 times that of the control (Fig. 2F). The mice were sacrificed three or four weeks later and the tumor tissues were weighed with record and photographed in detail (Fig. 3A). H&E staining of tumor tissues confirmed t the typical histological features of ACC (Fig. 3B). XAGE-1b protein with IHC staining 2was expressed much stronger in ACC-2-Xb-19 and ACC-2-Xb-M cells (Fig. 3C). The average MVD of a tumor is 19.8±5.57/field and it is higher than the control, 10.75±1.15/field (p<0.05, Fig. 3D and E). At ×640 magnification it is very clear that XAGE-1b staining focused in the nucleus. We detected the CD31 staining and calculated the microvessel density (MVD) of tumor tissues in each group to verify the correlation of enhancement between tumorigenesis and angiogenesis (Fig. 4).

We detected the transmembrane ability of ACC-2 cells with XAGE-1b protein expression and other cells by a kit simulating transendothelium stroma to verify the hypothesis of tumor metastasis promotion of ACC with XAGE-1b overexpression. The standard curve was drawn to reflect the correlation between the amounts of transmembrane cells and the fluorescence intensity (Fig. 5A). Cells of four groups were inoculated in three wells, respectively. The amount of transmembrane cells in ACC-2-Xb-19 is three times that of ACC-2 (p<0.01, Fig. 5B) and the amounts of transmembrane cells in ACC-2-Xb-M is ~1.5 times that of ACC-2 (p<0.01, Fig. 5B). There is no significant difference between ACC-2-pE-Negative and ACC-2. The data showed that XAGE-1b overexpression promoted the invasion ability of ACC-2 cells in vitro.

We established the nude mouse model of lung mestastasis to study the ability of invasion and metastasis promoted by XAGE-1b in vivo and tested the hypothesis applying the metastasis rate analysis. All cells, including ACC-2-Xb-19, ACC-2-Xb-M and ACC-2-pE-Negative, were injected into the tail vein of the nude mice. The nude mice were sacrificed 56 days later, and the lungs were dissected. The evident tumor nodules were observed on the lung of two mice in ACC-2-Xb-19 and one in ACC-2-Xb-M, but there are no nodules in the control (Fig. 5C, the first and second line). The average metastasis rate was calculated according to the PCR results. The lung migration based on the molecular level could be detected whether there was any visualized tumor nodules or not in ACC-2-Xb-19 and ACC-2-Xb-M, and the average metastasis rate was 0.0475±0.114 (n=6) and 0.0223±0.0491 (n=5), respectively. There was no migration in the control group of ACC-2-pE-Negative, that is to say, there was no human gene in the lung tissue of nude mice, and the average migration rate was ~0.0119±0.0251. The nodules formed in the lung of ACC-2-Xb-19 and ACC-2-Xb-M were confirmed to be of the typical tumor tissue features of ACC by H&E staining (Fig. 5C, the third line), and it proved the XAGE-1b protein expression in the lung migration by IHC staining (Fig. 5C, the forth and fifth line).

We detected RNA interfering effects on the exogenous expression of green fluorescence fusion protein by several candidate vectors. RNA interfering effects were observed through co-transfecting 293T cells with pG-Sh-a, b, c and pEGFP-Xb-GFP vector, Lipofectamine 2000 group was regarded as the control without transfection, pG-Negative and pG-Positive vector interfering GFP groups were regarded as the positive and negative control group co-transfecting with pEGFP-XAGE-GFP vector. The results were observed through the fluorescence microscopy and photographed 36 h later (Fig. 6). When the interference with the targeting plasmid ratio is 4:1 and 10:1, the fluorescence intensity of pG-Sh-a group was observed to decrease greatly close to the positive control group level, pG-Sh-b group decreased evidently but no more than pG-Sh-a group and pG-Sh-c groups, the green fluorescence expression is the same as negative pG-Negative.

Interference on endogenous expression of XAGE-1b in SPC-A1 cells was studied in detail. Semi-quantitative RT-PCR was applied and we found that XAGE-1b expression transfected with the interfering vector sh-a, and sh-b decreased evidently and was less than the negative control, empty control and no-transfected vector group, and the interfering effects were not obvious in the sh-c group (Fig. 7A). For confirming the interference rate and selecting the best interfering vector, we quantitatively and calculated the interfering rate by real-time PCR method (Fig. 7B). The results showed that the interference rate of sh-a vector is the highest among three groups and reach 94%, the others are 73.2% and 10% in the sh-b and sh-c group, respectively.

The pG-Sh-a vector was chosen to establish the stable interfering cell strains for the highest interference rate, the negative control group of plasmid pG-Negative was also screened. Because of the interfering plasmid with the red fluorescence protein and linearized operation before transfection, the expression of red fluorescence protein could be used for characterizing the integration of the transfected fragment and the successful expression of the interference sequence, and the cell clones with the bright red fluorescence protein were chosen as the stable interfering cell strains. There are two resistance cell clones with 100% expression of red fluorescence protein in each group respectively, and one was chosen as the ultimate stable cell strain named ACC-M-pG-Sh-a and ACC-M-pG-Negative to be used for the research on proliferation in vitro and in vivo of nude mice (Fig. 8A). The proliferation of ACC-M-pG-Sh-a cells was observed to be less than ACC-M-pG-Negative at the beginning of the second day and less than ACC-M at the beginning of the third day according to the MTT assay (p<0.01, Fig. 8B). In the cloning study, the amount of cell cloning of ACC-M-pG-Sh-a is less than ACC-M-pG-Negative and ACC-M group, and about 75% of the control (Fig. 8C). The tumor volume in ACC-M-pG-Sh-a is less than ACC-M-pG-Negative and ACC-M group, and the growth began to slow at the beginning of the fifteenth day with the evident difference (Fig. 8D). The mouse weight in ACC-M-pG-Sh-a is 4–8 times less than the ACC-M-pG-Negative and ACC-M groups (Fig. 8E) and the tumor volume is 7–11 times less than the control group (Fig. 8F).

The effects on the ability of subcutaneous tumors with XAGE-1b interference in vivo of nude mice was studied. Cells in different groups, including ACC-M-pG-Sh-a and ACC-M-pG-Negative and ACC-M cell lines, were observed with the above method (Fig. 9A). The tumors were found to give off the bright red fluorescence under the exogenous excitation when two mice of ACC-M-pG-Negative group were observed with live fluorescence before the sacrifice on day 24, it showed the method could be used for the quantitation of tumor volume and tracer of tumor cells (Fig. 9B). H&E staining displayed the characteristics of ACC (Fig. 9C) and it confirmed the difference of XAGE-1b expression by IHC staining (Fig. 9D). The microvessel density of the tumors was also calculated. The microvessel density of ACC-M-pG-Sh-a is 12.375±3.94, less than the value of 14.25±3.83 in ACC-M-pG-Negative group without significant statistical difference (Fig. 9E and F).