1. Overexpression of NRAGE induces radioresistance of ESCC cells in 2D culture
Our previous studies indicated that NRAGE was upregulated in ESCC radioresistant cells and extremely likely to be an RT-related critical factor [11, 19, 24]. Inadequately, there was lack of direct evidence to confirm the effect of NRAGE on resistance-promoting to IR. To verify the association between NRAGE and ESCC radioresistance, the expression of NRAGE in three types of ESCC cells, TE13, Kyse170, and Eca109, were compared (Fig 1a, b). Moreover, Eca109 cells with the least NRAGE expression was selected to stably overexpress NRAGE (Fig 1c, d). First, compared with Eca109-vector cells (indicated below as E), we aimed to identify the cellular changes resulting from expression of NRAGE in Eca109 cells (indicated below as E/N). It was visibly different in morphological distinction with more irregular, elongated spindle-shaped cells and disappearance of polarity (Fig 1e). Additionally, cell proliferation and radiosensitivity were tested through CCK-8 and clone formation assay. E/N cell exhibits its super growth ability and radiation-hardened effect (Fig 1f, g, h). Before exposure to IR, both E and E/N cells showed vigorous multiplication without difference during the first 4 days. From the fifth day, E/N cells displayed enhanced proliferation ability. However, in the IR group, the significant difference between them was observed early at the fourth day (Fig 1f). Moreover, E and E/N cells were exposed to different doses of radiation for colony formation. E/N cells showed relatively higher colony survival rates and increased radiobiological parameter, SF2 (E vs EN=0.518 vs 0.636), D0 (E vs EN=1.443 vs 1.901), and Dq (E vs EN=1.338 vs 1.603), under a series of doses of 0, 2, 4, 6, 8, and 10 Gy (Table S1; Fig 1g, h). These results suggest that NRAGE overexpression induces radioresistance of ESCC cells in 2D culture.
2. Overexpression of NRAGE inhibits cell migration, invasion, cell cycle progression and apoptosis after IR in 2D culture
To further define the mechanism of NRAGE in ESCC radioresistance, cell migration, invasion, cell cycle progression, and apoptosis in E/N cells were analyzed apart from the detection of proliferation. Wound healing assays were performed in E and E/N cells with or without 5 Gy X-ray radiation and then imaged at 12 h and 24 h. Results showed that ESCC cells with upregulated NRAGE had significantly faster migration ratio than E cells, especially after IR (Fig 2a, b). Transwell assays also showed that the number of invasion cells through the membrane regardless of the presence of IR was significantly larger in E/N cells (Fig 2c, d). It revealed that NRAGE may enhance invasion and migration of ESCC cells after IR led to more resistive effect to IR. Cell apoptosis assays showed that the rate of spontaneous apoptosis in E/N cells was significantly decreased (E vs E/N, 7.99%±0.50% vs 4.16%±0.15%, p=0.0057). After RT, E/N cells had lower apoptosis than E cells (E vs E/N, 24.29%±1.12% vs 34.63%±1.83%, p<0.0001) (Fig 2e, f). Furthermore, we analyzed cell cycle progression of E and E/N cells with or without 5 Gy IR (Fig 2g, h). It was found that, before IR exposure, NRAGE overexpression was associated with an increased percentage of cells in the S phase (33.23±1.78 vs 25.69±1.70, p=0.01), the most radioresistant cell stage, and a lower ratio in the most radiosensitive cell stage G2/M (18.87±0.46 vs 27.91±0.81, p=0.0018). After treatment with 5 Gy IR, cell cycle distributions were rearranged to a greater extent with more arrested cells in the S phase (26.46±5.61 vs 16.27±2.71, p=0.0005) and G0/G1 phase (45.50±4.95 vs 35.21±0.96, p=0.0004) and downregulated cells in the G2/M phase (28.04±0.67 vs 48.52±1.77, p<0.0001). These revealed that NRAGE overexpression could reduce cell apoptosis and change cell cycle division of ESCC, affecting cellular radioresistance.
3. NRAGE overexpression activates canonical Wnt signaling pathway in ESCC cells with 2D culture
Previous studies indicated that NRAGE had a potential association with β-catenin in the formation of radioresistance in ESCC[11], so we detected protein expression in canonical Wnt signaling pathway, including β-catenin, phosphorylation of β-catenin (p-β-catenin), Gsk-3β, phosphorylation of Gsk-3β (p-Gsk-3β), and CyclinD1. Compared with E, an increase in β-catenin (p=0.009) level, followed by increased p-Gsk-3β levels (p=0.002), led to the increased expression of cyclin D1 (p=0.032), a targeting gene of β-catenin in E/N cells, while downregulation of p-β-catenin (p<0.001) and Gsk-3β (p=0.016) was detected (Fig 3a, b). These data indicated that, after NRAGE overexpression, the canonical Wnt signaling pathway was overall activated, which may be a switch-induced radioresistance.
4. 3D bioprinted ESCC cell-laden system cultured in vitro
To the best of our knowledge, there is considerable difference on cancer cell morphology genetic profile and tumoral heterogeneity in 2D cultures. To focus more on the tumor cell growth environment and microenvironment in vitro, we selected culture cells in the 3D bioprinting system to identify the function of NRAGE in radioresistance of ESCC cells. A gelatin-alginate blend (10% gelatin and 3% sodium alginate) was used as the 3D bioprinted material. Hydrogel seeded with E and E/N cells were extruded at variable pressure (0.31 MPa), needle type (cylindrical), and needle diameter (340 µm) to study cell characteristics directly after printing. Extruded gelatin-alginate blend was stained with live/dead dye and imaged (Fig 4a, b). Most cells remained viable (green), and only a small number of dead cells (red) were observed. Subsequently, the result of the analyses showed that dead or live cells were counted to quantify cell survival at > 80% and E/N cell-laden 3D-scaffolds exhibited stronger survivability. After printing, images were obtained, followed by crosslinking in 3% CaCl2 and incubation at 37°C to allow the gelatin to leach out. The cell-laden 3D-scaffolds had a grid-like structure arranged in multiple layers, and cells were uniformly distributed in porous scaffolds with tight order, exhibiting good cytocompatibility (Fig 4c). At the first week, the printed scaffolds did not display obvious proliferation. Then, the cell growth rate accelerated slowly over time. After 3 weeks, cells began to grow into spheroids and pushed the surrounding hydrogels aside to occupy a larger space. Especially for E/N cell-laden hydrogels, the phenomena were highlighted. It was extremely biomimetic to the solid tumor growth in vivo. SEM observations revealed the spheroids bulged out over the scaffolds surface, and the trace of cells squeezed the surrounding hydrogel, which showed that E/N cell-laden scaffolds were apt to spheroiding. It was also suggested that E/N cells in the 3D group have a significantly higher secretion of growth hormone than E cells, and the difference gradually became pronounced over time (Fig 4e). After culture for 7 days, as shown in HE staining, individual cells scattered in printed scaffolds were observed (Fig 4f). During the culture period, natural gelatin began to degrade gradually via hydrolysis in the culture medium and then provided space for cells proliferating in clusters at 14 and 21 days. Furthermore, more and larger cell clusters were observed in E/N cell-laden 3D scaffolds. Fig. 4G showed the results of IHC staining for NRAGE that E/N cells in the 3D culture system had a distinct positive staining in nuclear cells with larger clusters. These results implied that cells with NRAGE overexpression in the 3D culture system are more suitable for survival and cloud growth, which hinted that ESCC cells with NRAGE overexpression exerted greater adaptability to survive and multiply.
5. NRAGE overexpression enhanced the proliferation and radioresistance of ESCC cells in 3D bioprinted hydrogels
The obvious advancement and characteristic of the 3D-printed model showed more similar growth environment and microenvironment in vivo. To identify the effect of NRAGE overexpression in ESCC cells cultured in the 3D-printed model on proliferation of tumor cells, alamarBlue assays were selected to compare cell viability between 3D-printed and 2D groups. As shown in Fig 5a, regardless of the culture condition (2D or 3D), E/N cells had considerably higher survival percentages than E cells. More interestingly, there was a faster proliferation rate in E/N cells in the 2D group in the first 20 days, whereas the 3D-printed group showed a significantly higher proliferation rate of cells after 20 days. Similarly, an apparent trend that E cells in the 3D-printed group would proliferate faster than those in the 2D group after 21 days was observed (Fig 5a). Moreover, the difference in responses to IR between E and E/N cells in the 3D-printed model was identified by alamarBlue assays after 5 Gy IR. Evidently, E/N-3D cells had an absolute dominance of survival on resistance to IR compared with E-3D cells (Fig 5b). Deeply, the protein expression of Ki-67, a marker for cell proliferation activity, between E-3D and E/N-3D cells with or without 5 Gy IR, was evaluated by IHC, and both of them had a positive expression in relative individual cells scattered in hydrogel at 7 days. However, after 14 days, more positive staining in larger clusters in E/N-3D cells appeared. Furthermore, this different trend was also found after 5 Gy IR (Fig 5c). To verify whether the β-catenin expression change in ESCC cells with NRAGE overexpression was consistent from 2D to 3D groups, IHC staining was performed. Similarly, in the first 7 days, both E-3D and E/N-3D cells had a higher β-catenin expression levels with larger cell clusters. After culture for 14 days, larger E/N-3D cell clusters were stained positively by β-catenin antibody compared with those in E cells. Unsurprisingly, there was a more obvious distinction between groups after 5 Gy IR. The results confirmed further that accelerated NRAGE expression in ESCC cells activate β-catenin expression to regulate radiosensitivity.
6. NRAGE is upregulated in patient samples with EC following radical RT and correlated with poor prognosis
We analyzed NRAGE expression in published profiles of patients with ESCC and found that it was upregulated in ESCC samples (182 cases) compared with adjacent normal tissue samples (286 cases) (p<0.05, Fig 6a) (match TCGA normal and GTEx data, http://gepia.cancer-pku.cn/). Additionally, to thoroughly explore the role of NRAGE on radioresistance of ESCC and relationship with β-catenin, we further analyzed the expression of NRAGE and β-catenin in a total of 44 paraffin-embedded, ESCC tumor tissues receiving definitive RT (Table s2). As shown in Fig. 6A, the 1-, 3-, and 5-year overall survival rate of 44 patients were 69%, 36%, and 21%, respectively (Fig 6b). According to the evaluation criteria of RT curative effect, 36 patients were classified in the efficacy group (complete response, CR, 26 patients, and partial response, PR, 9 patients) and 8 patients were classified in the inefficacy group (No response, NR, 8 cases) (Fig 6d-f). There were statistically significant differences between the two groups in the 1-, 3-, and 5-year OS rate: 81%, 45%, and 26% for the efficacy group and only 15% of 1-year OS rate in the inefficacy group were achieved (p=0.0001) (Fig 6c). Compared to the inefficacy group in which NRAGE and β-catenin were expressed at low levels (Fig 6g and i), NRAGE and β-catenin were overexpressed, especially for positive nuclear expression, in efficacy group specimens (Fig. 6H and J). According to the analysis of relationship between staining score and short-term effect of RT, NRAGE protein expression was dramatically upregulated in the RT efficacy group (CR + PR) tumor tissues compared with the NR group (p=0.015) (Fig 6k). Unsurprisingly, more NRAGE nuclear protein expression was detected in efficacy group (p=0.0021) (Fig 6l). Additionally, there was higher β-catenin total protein expression in the efficacy group than in the NR group (p=0.081) (Fig 6m). However, the difference in β-catenin nuclear protein expression between the two groups was significant (p=0.0037) (Fig 6n).
Routinely, we analyzed the association between NRAGE total/nuclear protein or β-catenin total/nuclear protein expressions and clinicopathological features of 44 patients with ESCC by Spearman analysis. It was revealed that the expression of NRAGE total protein, especially for NRAGE nuclear protein, was strongly associated with curative efficacy (p=0.0023, p=0.006). However, regardless of NRAGE total protein or NRAGE nuclear protein, there was no association with age (p=0.656, p=0.277), gender (p=0.734, p=0.277), clinical stage (p=0.932, p=0.759), tumor size (p=0.121, p=0.488), LNM (p=0.153, p=0.148), synchronous chemotherapy (p=0.906, p=0.862), and events (p=0.135, p=0.528) (Table 1). As shown in Table 1, no correlation between β-catenin expression and age (p=0.288, p=0.231), sex (p=1.000, p=0.0.358), clinical stage (p=0.824, p=0.986), tumor size (p=0.168, p=0.263), LNM (p=0.221, p=0.587), synchronous chemotherapy (p=0.099, p=0.459), and events (p=0.754, p=0.296) was found. A significant correlation could not be found between β-catenin total protein expression and clinicopathological features (p=0.143), but a strong association between β-catenin nuclear protein expression and curative efficacy was observed (p=0.006). Kaplan-Meier survival curves exhibited no association between OS in definitive RT and NRAGE or β-catenin total protein expression (Fig 6o and q, p=0.198, p=0.504), but a strong positive NRAGE nuclear protein expression was significantly shorter than those with positive and weak positive NRAGE expression (Fig 6p, p<0.0001). Additionally, there was a correlated trend between β-catenin nuclear protein expression and OS (Fig 6r, p=0.081). Moreover, we analyzed the association between NRAGE and β-catenin nuclear protein expressions and confirmed their linear correlation (χ2=4.106, p=0.043) (Table s3). These results indicate that NRAGE expression, especially NRAGE nuclear expression, in patients with ESCC receiving radical RT was correlated with poor survival and may be linked to heightened β-catenin nuclear accumulation. Furthermore, univariate and multivariate analyses were used to determine whether NRAGE could be a risk factor in patients with ESCC receiving radical RT. Log-rank test in the univariate analysis showed that synchronous chemotherapy (p=0.037), curative efficacy (p=0.000), and strong positive NRAGE nuclear protein expression (p=0.000) were associated with a significantly increased risk of death in patients with ESCC receiving radical RT (Table 2). Multivariate Cox regression analysis revealed that NRAGE nuclear protein could be a factor for predicting poor survival when it has strong positive expression (RR=14.536, p=0.000). Synchronously, clinical stage (RR=2.995, p=0.024) and synchronous chemotherapy (RR=0.354, p=0.019) were included as factors (Table 2). Collectively, all these indicated that NRAGE overexpression occurred during nuclear translocation after IR and stimulated β-catenin expression in the cytoplasm to increase the nuclear localization of β-catenin, which activated the Wnt/β-catenin signaling pathway and then induced the radioresistance in ESCC. A flowchart of the possible mechanism is shown in Fig 6s.