Clinicopathological significance of Axin2 and Snail expression in patients with OSCC
In the present study, Axin2 expression was found in the cytoplasm of cancer cells in 168 (77.4%) patients with OSCC, and immunoreactivity against Axin2 was high in 101 (high-Axin2, 46.5%) OSCC tissue samples and low in 116 (low-Axin2, 53.5%). Cancer cells demonstrated cytoplasmic and nuclear Snail expression in 186 (85.7%) patients with OSCC, and immunoreactivity against Snail was high in 107 (high-Snail, 49.3%) OSCC tissue samples and low in 110 (low-Snail, 50.7%). Significant association was found between Axin2 and Snail expression in patients with OSCC (p=0.006) (Fig. 2A). High-Axin2 or high-Snail expression both showed significant association with T stage (p<0.001 and p=0.031), lymph node metastasis (both p<0.001), vascular invasion (p=0.01 and p=0.019), and with bone invasion (p<0.001 and p=0.028) of OSCC in the present study (Table 2). Moreover, patients with high-Axin2 or high-Snail expression demonstrated increased vessel density (p<0.001 and p=0.002) and higher desmoplastic reaction (both p<0.001) than patients with low-Axin2 or low-Snail expression in our cohort (Fig. 2B and 2C). For identify the risk factors for prognosis of OSCC, multivariate analysis was performed in 179 patients who with follow up more than 5 years. Results showed that when using age, sex, lesion site, T stage, lymph node metastasis, histologic grade, vascular invasion, perineural invasion, bone invasion, status of desmoplasia, status of angiogenesis, Axin2, and Snail expression as cofactors, lymph node metastasis, status of desmoplasia, Axin2 expression, and Snail expression were independent risk factors for OSCC prognosis, with hazard ratios of 3.424 (95% confidence interval, 1.466–7.998; p=0.004), 2.491 (95% confidence interval, 1.240–5.004; p=0.01), 2.488 (95% confidence interval, 1.358–4.559; p=0.003), and 1.984 (95% confidence interval,1.097–3.588; p=0.024), respectively (Table 3).
Axin2 knockdown had a strong influence on the biological behavior of OSCC cells
Consistent with the results of a previous study , we found that Snail expression was predominantly decreased in both CA9-22△Axin2 and HSC-2△Axin2 cells compared to that in related control cells (Supplementary Fig. 1A, i and iv).
Proliferative ability was significantly reduced after Axin2 knockdown in both CA9-22 and HSC-2 cells. Compared with CA9-22 Mock, decreases of 1.4-, 2.1-, and 2.2-fold in cell number were found in CA9-22△Axin2 cells after 24h, 48h, and 72h of culture (all p=0.008). Similarly, HSC-2△Axin2 cells also showed decreases of 1.4-, 1.6-, and 2.3-fold in number compared to the HSC-2Mock cells (all p=0.008) (Supplementary Fig. 1A, ii and v). Likewise, Ki67 expression was also significantly decreased in Axin2 knockdown cells compared to that in the controls in both HSC-2 and CA9-22 cells (both p=0.002) (Supplementary Fig. 1A, iii and vi). In addition, cell motility was decreased 1.5- and 1.8-fold, respectively, in Axin2 knockdown cells compared to the control CA9-22 and HSC-2 cells (both p=0.002) (Supplementary Fig. 1B, i-iv). Moreover, 2.3- and 1.6-fold decreases in numbers of invading cells were found in Axin2 knockdown cells compared to CA9-22 and HSC-2 control cells, respectively (both p=0.002). Axin2 may have oncogenic activity in OSCC cells (Supplementary Fig. 1C, i-iv). Interestingly, compared to the related control cells, expression of Snail-related cytokines IL8, CCL2, and CCL5 was 3.5-fold, 2.8-fold, and 3.3-fold decreased in CA9-22△Axin2 cells (all p=0.002), and 2.6-fold, 1.8-fold, and 1.5-fold decreased in HSC-2△Axin2 cells (all p=0.002) (Supplementary Fig. 1D, i-ii).
Cytokines related to Axin2-Snail axis reveal strong influences on the biological behavior of CAFs
To evaluate the effect of those cytokines on the biological behavior of CAFs, both CAF1 and CAF2 cells were treated with different does (0, 2, 5, and 10ng/ml) of human recombinant proteins (IL8, CCL2, and CCL5), after which the proliferation and invasion abilities of CAFs in each group were comparatively investigated. Strong influence of IL8 or CCL5 on biological behavior of CAFs was found from dose of 2ng/ml in the present study. We found that CAF1 showed increases of 2.5- and 2.0-fold in numbers of the cells after treatment with 2ng/ml IL8 or CCL5, compared to untreated controls (both p=0.002). Similar results were obtained using CAF2, with IL8 or CCL5 treatment leading to increases in cell numbers (both p=0.002). No significant differences were observed in the proliferative ability of CAFs after CCL2 treatment in this study (Fig. 3B, i-ii).
We also found that the invasion ability of CAF1 and CAF2 cells was 2.2- and 2.3-fold increased after IL8 (2ng/ml) treatment compared to that of untreated control cells (both p=0.002). Supportively, MMP-2 expression was 2.3-fold and 2.2-fold increased after IL8 (2ng/ml) treatment in both CAF1 and CAF2 cells compared to untreated controls (both p=0.002). No significant difference was found in MMP-9 expression in CAFs after IL8 treatment in our study (Fig. 3C, i-vi).
Tumor progression and bone invasion depends on Axin2 expression in tumor cells in vivo
As shown in Fig. 4, tumor volume was significantly decreased in mice injected with Axin2 knockdown cells compared to controls for both CA9-22 and HSC-2 cells (i-ii). In the micro-CT imaging analysis, extensive osteolytic lesions were observed in the calvaria from CA9-22Mock or HSC-2Mock cell-bearing mice when compared to those from the related Axin2 knockdown cell-bearing mice (iii-iv). Moreover, in the tissue sections, the area of tumor-associated stroma was predominantly increased at the tumor-bone interface in CA9-22Mock or HSC-2Mock cell-bearing mice compared to that in the related Axin2 knockdown cell-bearing mice (v-vi).