Commercially available tissue microarray (TMA) slides (TH8010 and TH802a, US Biomax, Inc.) were purchased for immunohistochemistry (IHC) analysis. Specific primary antibodies against SIX1 (Sigma, USA) and EYA1 (Proteintech, China) were used for IHC with a 2-step protocol (21). Whole slide image capture was performed on the EVOS auto cell image system (Life technology, USA). For semi-quantitative evaluation of protein level in tissue, the staining intensity was graded as previously described (22). The immunohistochemical score were assessed by two experienced pathologists without knowledge of patients' characteristics. Scores were calculated on intensity and percentage of positive staining tumor cell nuclei or cytoplasm in the whole tissue stains were evaluated according to Fromowitz Standard. Briefly, the staining intensity was graded as follows: no staining, 0; weakly positive, 1; moderately positive, 2; and strongly positive, 3. The percentage of positive cells was into four grades: 0–25% staining, 1; 26–50% staining, 2; 51–75% staining, 3; and 76–100% staining, 4. The multiplication of the intensity and percentage scores was used to calculate the final staining score. For quantification, all stains were assessed at 200× magnifications and at least three fields from each core were counted.

Human papillary thyroid cancer cell lines BCPAP, NPA and TPC-1 were provided by Dr. Du (Tongji Hospital of Tongji Medical College) and cultured in recommended condition. Cell line authenticity was confirmed by Short Tandem Repeat (STR) DNA profiling (the STR profiles were shown in Supplementary Materials). All experiments were performed using cell lines from passage 6 to 25. Cells were seeded at 50% confluence in 6 cm plate on the day before transduction. HEK 293T cells were transfected with pLV vector or pLV-SIX1 expression plasmid with package plasmids by Lipofectamine™ 2000 (Invitrogen, Carlsbad CA, USA). The supernatant was collected and polybrene (1:1,000) was added into the supernatant. The mixed supernatant was applied to recipient cells for infection. The expression of SIX1 was verified by quantitative reverse transcription-PCR (qRT-PCR) and Western blot. For small interfering RNA (siRNA)-mediated downregulation of EYA1, BCPAP-Vector and BCPAP-SIX1 cells were seeded in 6-well plates and transfected with siRNA or scramble control (Ribobio Company, Guangzhou, China) duplexes using Lipofectamine™ 2000 (Invitrogen, Carlsbad CA, USA). EYA1 siRNA: (sense) 5′-CAGGAAAUAAUUCACUCACAAdTdT-3′; (antisense) 5′-UUGUGAGUGAAUUAUUUCCUGdTdT- 3′.

Cell and tissue lysates were extracted using ice-cold RIPA buffer and measured using a bicinchoninic acid (BCA) protein assay kit (Promoter, China). Proteins were resolved on 10% SDS-polyacrylamide gels and transferred to PVDF membranes. The antibodies used in Western blot were as follows: SIX1, EYA1, C-MYC (Santa Cruz, USA), STAT3 (Cell Signaling Technology, USA), p-STAT3 (Tyr705) (Cell Signaling Technology, USA), BCL-XL (Cell Signaling Technology, USA), Caspase 3 (Proteintech, China), Cleaved PARP1 (Ruiying Bio, China), Cleaved Caspase 9 (Ruiying Bio, China), β-tublin (Cell Signaling Technology, USA), GAPDH (Cell Signaling Technology, USA) and VINCULIN (Sigma, USA).

RNA was prepared from PTC cells with the TRIzol reagent (Invitrogen, USA). cDNA was reversed from 1 μg total RNA using a reverse transcription kit (TOYOBO, Japan). RT-qPCR was performed with the SYBR^® Green Real-time PCR Master Mix Kit (TOYOBO, Japan). Gene expression was normalized to GAPDH. The primer sequences for real-time RT-PCR were as follows: SIX1: (forward) 5′-ACAAGAACGAGAGCGTACTCA-3′, (reverse) 5′-CTCCACGTAATGCGCCTTCA-3′; EYA1: (forward) 5′-GTTCATCTGGGACTTGGA-3′, (reverse) 5′-GCTTAGGTCCTGTCCGTT-3′; GAPDH: (forward) 5′-CAATGACCCCTTCATTGACC-3′, (reverse) 5′-GATCTCGCTCCTGGAAGATG-3′.

For MTT assay, 2,000 cells were seeded into 96-well plates and analyzed by adding MTT (tetrazolium bromide, 5 mg/mL, GE Healthcare) as previously described (22). To measure the growth curve, 2.5 × 10³ cells were seeded in 24-well culture plates and the numbers of viable cells were serially counted for 6–7 days. Colony formation assay was performed as previously described (23). Two weeks later, cells were fixed with 4% paraformaldehyde and stained by 0.5% crystal violet for visualization and counting on the plate.

About 3 × 10⁵ cells were seeded into a 6 cm Petri dish. After 24 h incubation, the cells were collected and fixed with 75% cold ethanol at −20°C overnight. DNA was incubated with 200 μL RNase A (1 mg/mL) and 500 μL propidium iodide (PI, 100 μg/mL) for 30 min at room temperature in the dark and analyzed by using the FACSort flow cytometer (Becton, Dickinson Company, USA). The data were analyzed with ModFit LT V2.0 software (Becton, Dickinson Company, USA).

Transwell chambers (pore size 8.0 μm) (Corning Inc., USA) were either uncoated (migration assay) or coated (invasion assay) with Matrigel as previously described (23). All experiments were conducted in triplicate.

The apoptosis was determined by the Annexin V-FITC/PI apoptosis detection kit (BD Biosciences, USA). Briefly, cells were treated with 200 μM H₂O₂ (Sigma, USA) for 24 h. The cells were collected, and then resuspended in 200 μL of binding buffer. After incubation of Annexin V-FITC and PI for 15 min at room temperature, 300 μL binding buffer was added to the cells and the results were analyzed by flow cytometry (Beckman-Coulter Inc., USA). The experiment was repeated in triplicate.

EdU incorporation assay was performed with EdU assay kit (Ribobio, China). Briefly, 1 × 10³ cells per well were culture in 96-well plates for 48 h, and then 50 μM of EdU was added to each well and cultured for additional 2 h. The cells were fixed with 4% formaldehyde for 15 min and treated with 0.5% Triton X-100 for 20 min. After washing with PBS, 100 μl of 1 × Apollo reaction cocktail was added and incubated for 30 min. After staining with 100 μl of Hoechst 33342 for 30 min, the cells were visualized under EVOS cell image system. The results were analyzed by Image-Pro Plus 6.0 software (Media Cybernetics, USA). All experiments were done in triplicate and three independent repeating experiments were performed.

Cells were seeded in a 24-well plate (5 × 10⁴cells/well) and cultured for 48 h, then fixed with 4% paraformaldehyde. Next, cells were permeabilized with 0.1% Triton X-100 and blocked using 5% goat serum for 30 min. Cells were further incubated overnight with primary antibodies against SIX1 (anti-mouse, 1:100, Santa Cruz, USA), EYA1 (anti-rabbit, 1:100, Proteintech, China). Next day, cells were incubated with either Alexa Fluor 594 or Alexa Fluor 488 for 1 h. Nuclei were visualized with DAPI stain. The stained cells were examined with EVOS cell image system.

Five- to 6-week-old female NOD/SCID mice were purchased from Beijing HFK Bioscience Limited Company and maintained under the Specific Pathogen-Free (SPF) conditions at Laboratory Animal Centre. All the protocols were reviewed and approved by the Institutional Ethics Committee and performed in accordance with national guidelines. All mice were randomly divided into two groups: TPC1-shControl (n = 20) and TPC1-shSIX1 (n = 20). Cells were counted and serially diluted in 100 ml of 1:1 PBS/Matrigel, and injected subcutaneously. The mice were monitored every 3 days. At 3 weeks post-injection, all mice were euthanized and analyzed for tumor formation. Tumors were collected for further Western Blot analysis.

GraphPad Prism 6.0 software was used for the statistical analyses. The IHC scores were tested whether the data matched normal distribution or not. If it was, then the difference between groups were conducted by using parametric statistics (Student t-test), otherwise performing non-parametric statistics (Mann-Whitney test). The correlations between clinicopathological and immunohistochemical variables were calculated according to Person χ²-test. Cell culture experiments differences between the groups were evaluated by the Student t-test, including EdU assays, MTT activity, growth curve assays, colony formation assay, Transwell migration, and invasion assays, Annexin V-FITC/PI assays, PCR and western blot assays. All data were expressed as mean ± standard error. P < 0.05 were considered statistically significant.

To identify the expression pattern of SIX1 and EYA1 in human thyroid tissue as well as tumor, we performed IHC analysis by using two tissue microarrays (TH8010 and TH802a). After excluding the duplicated reports (n = 6) and detached sample (n = 9), two tissue microarrays contained 145 samples, including normal thyroid tissue (n = 10), atypical adenoma (n = 1), oncocytoma (n = 4), colloid adenoma (n = 15), fetal adenoma (n = 10), follicular adenoma (n = 10), follicular carcinoma (FTC, n = 18), medullary carcinoma (MTC, n = 5), papillary carcinoma (PTC, n = 66), and anaplastic carcinoma (ATC, n = 6). The IHC score for each type was shown in Figure 1A. In comparison with normal tissue, SIX1 and EYA1 kept low expression in benign tumor and no statistic difference was found for either SIX1 or EYA1 between groups. However, a significant increase in SIX1 and EYA1 was detected in malignant tumor, including FTC (p = 0.011, p = 0.038; respectively), MTC (p = 0.006, p = 0.043; respectively), PTC (p < 0.001, p < 0.001; respectively) and ATC samples (p = 0.002, p < 0.001; respectively) (Figure 1A). The representative images of IHC staining for normal and cancerous tissue were shown in Figure 1B. Intriguingly, SIX1 was tightly linked to EYA1 in malignant tumor. The analysis from Spearman correlation based on all malignant sample in these tissue microarrays indicated that SIX1 had a strongly positive relationship with EYA1 (Figure 1C). BRAF mutation is the most common mutation in PTC (8). To test whether there was relationship between SIX1/EYA1 and BRAF^V600E mutation, the GEO database (GSE54958) was applied to analyze. There was no difference in the expression of SIX1/EYA1 and BRAF^V600E mutation (Data not shown).

Since PTC is the most common type of thyroid carcinoma and also occupied the majority of the tissue microarrays in this study, we next examined the association of SIX1 and EYA1 expression with clinicopathological parameters. Based on the semi-quantitative criteria, we defined IHC score over 6 (>6) was higher and scores under 6 (≤6) was lower expression. We found that the levels of SIX1 and EYA1 expression were significantly associated with age, lymph node metastasis (two sample missed this information, tissue ID: Etg030430 and Etg030007) and clinical stage, but not tumor size (Table 1). Based on these data, high expressions of SIX1 and EYA1 are closely correlated with thyroid malignant tumor. SIX1 may cooperate with EYA1 and play a key role in lymph node metastasis via a BRAF-independent manner.

Since SIX1 is a DNA-specific transcriptional factor, whereas EYA1 basically acts as a co-factor to increase SIX1-dependent function (11), we mainly focused on the functional role of SIX1 in thyroid cancer for first. The base line expressions of SIX1 were detected by western blot in PTC cell lines, including BCPAP, TPC-1, and NPA, with HEK 293T cell as a positive control. The results showed that SIX1 were barely expressed in BCPAP cells, whereas highly enriched in TPC-1 cells (Figure 2A). The next step was establishing the thyroid cell lines with either overexpression or knockdown of SIX1 by employing lentivirus-mediated gene transfer system. BCPAP cells stably overexpressing SIX1 were successfully transduced by pLV-SIX1 (BCPAP-SIX1) and pLV empty vector (BCPAP-Vector). Meanwhile, lentivirus shRNA expression vector for SIX1 or scramble control were transduced into TPC-1 cells (Figure 2B).

To assess the role of SIX1 in cell cycle regulation, propidium iodide staining and flow cytometry were performed. Overexpression of SIX1 significantly decreased the percentage of cells in the G0/G1 phase (78.47 ± 2.37% vs. 67.31 ± 1.56%, p = 0.017) and increased the subpopulation in the S phases (16.33 ± 2.44% vs. 27.68 ± 1.72%, p = 0.019) in BCPAP cells (Figures 2C,D). Conversely, knockdown SIX1 in TPC-1 cells induced a significant increase of G0/G1 phase (53.73 ± 0.87% vs. 64.79 ± 1.46%, p = 0.003) and a decrease in the S-phase population (34.42 ± 0.61% vs. 22.77 ± 1.32%, p = 0.001) compared with control group (Figures 3A,B). To further evaluate the effect of SIX1 in cell cycle regulation, EdU incorporation assay was examined. In accordance with results of flow cytometry, the proportion of EdU positive cells was increased by 2.84-folds induced by SIX1 overexpressing in BCPAP cells (Figure 2E). And the number of EdU positive cells in TPC-1-shSIX1 group was reduced by 40% in comparison with TPC1-shControl (Figure 3C). Taken together, these data demonstrate that SIX1 promoted cell cycle progression in thyroid cancer cells.

To examine the effect of SIX1 on PTC cellular proliferation, MTT assays, cell growth curve, and colony formation assays were performed to assess the proliferative ability. MTT assays showed the tumor cell growth rate significantly increased after ectopic expression of SIX1 (Figure 2F) and decreased after silencing the endogenous SIX1 expression (Figure 3D), which were in accord with the findings from cell growth curve (Figures 2G, 3E). Transfecting the exogenous SIX1 gene into BCPAP cells can increase both colony number and size (Figure 2H), and SIX1 knockdown otherwise resulted in approximately 50% colony growth inhibition in the TPC-1 cells (Figure 3F). Furthermore, cell migration and invasion were significantly increased in BCPAP cells with SIX1 overexpression (Figure 2I) and decreased by half in TPC-1 cells after knockdown of SIX1 (Figure 3G). These data suggested that SIX1 induced the oncogenic properties of PTC.

Increased proliferation and reduced apoptosis are the fundamental feature of cancer cells. Next, we evaluated whether SIX1 protected PTC cell lines from apoptosis assessed by Annexin V-FITC/PI staining. Flow cytometric analysis results indicated that H₂O₂ induced apoptosis in PTC cell lines. The proportion of apoptotic cells in BCPAP-SIX1 group was significantly reduced from 13.46 to 5.82% in comparison with BCPAP-Vector (p = 0.022) (Figure 4A). The proportion of apoptotic cells in TPC1-shSIX1 group was increased significantly from 8.75 to 18.56% in comparison with TPC1-shControl (p = 0.0133) (Figure 4A). The protein expression levels of caspase 3, cleaved PARP1 and cleaved caspase 9 were detected by Western blotting. The results showed that SIX1 significantly attenuated H₂O₂-induced upregulation of cleaved Caspase 3, cleaved PARP1 and cleaved Caspase 9 protein expression (Figure 4B).

Although there is a consensus that EYA functions as a co-activator in SIX1-associated malignant properties, little is known about whether and how SIX1 modulates EYA1 expression. Immunofluorescence assay showed the positive relationship and co-localization of SIX1 and EYA1 (Figure 5A). Moreover, we carried out qRT-PCR to examine the mRNA change. However, the mRNA of EYA1 did not change when SIX1 was overexpressed in BCPAP or SIX1 was knocked down in TPC-1 cells (Figure 5B), indicating that SIX1 induces the protein abundance of EYA1 at post-transcriptional level, rather than directly involves in the transcriptional regulation of EYA1 mRNA.

Cell cycle proteins regulated by SIX1 might be responsible for the proliferation in PTC. To address the underlying mechanisms, cell cycle elements were screened by western blot. SIX1 significantly promoted the abundance of C-MYC (Figure 5C). Since C-MYC is the downstream target of STAT3 signaling and other group has reported SIX1 can induce p-STAT3 expression in human keratinocytes (24), we hypothesized that SIX1 might promote C-MYC abundance via activation of STAT3 signaling. Ectopic expression of SIX1 could significantly upregulate the expression of p-STAT3 (Try705) as well as its downstream target BCL-XL. As expected, knockdown SIX1 in TPC-1 cell showed the opposite (Figure 5C).

To further determine whether EYA1 was required for the SIX1-induced STAT3 signaling activation. siRNA was used to transiently knock down EYA1 in BCPAP-vector and BCPAP-SIX1 cells. Intriguingly, the upregulation of p-STAT3, C-MYC, and BCL-XL induced by SIX1 was abolished after silencing EYA1. Additionally, knocking down EYA1 in BCPAP-Vector cells diminished endogenous expression of STAT3 signal and its downstream targets (Figure 5D). We evaluated the effects of EYA1 knock-down in TPC-1 cells which had higher endogenous expression of SIX1. The expressions of SIX1, p-STAT3, C-MYC, and BCL-XL abundance were down regulated after silencing EYA1 by siRNA in TPC-1 cells (Figure 5E).

Next, we compared the functional significance of SIX1/EYA1 interaction. Both colony formation assay and MTT showed that the proliferation ability was significantly reduced when EYA1 was knocked down in SIX1-overexpressing BCPAP cells (Figures 6A,B,E). In agreement, SIX1-induced DNA synthesis was inhibited by knock down EYA1, as evaluated by EdU incorporation assay (Figures 6G,H). On the other side, colony formation assay and MTT demonstrated that knockdown endogenous EYA1 can inhibit TPC-1 cell growth (Figures 6C,D,F). Together, these data suggest that the activation of STAT3 signaling induced by SIX1 is dependent on the presence of EYA1 in PTC cells.

To test whether blocking the expression of SIX1 abolished the PTC cell proliferation in vivo, 2 × 10⁶ TPC1-shControl or TPC1-shSIX1 cells were injected, respectively, into the subcutaneous fat of immunodeficient mice. Downregulation of the endogenous SIX1 by shRNA in TPC-1 cells not only showed a trend of slowing down the tumor associated weight loss (Figure 7A), but also significantly reduced the volume as well as the weight of xenografted tumors (Figures 7B–D). At the time of sacrifice, xenograft tumors were harvested and examined by western blot. Consistent with these findings in vitro, the tumors in TPC1-shSIX1 group retained SIX1 lower-expression in vivo and the levels of EYA1, p-STAT3 (Try705) and C-MYC were significantly decreased compared with shControl group (Figures 7E,F). Taken together, these results highlight the potential for targeting SIX1 in the comprehensive therapy for PTC patients.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.