Patient characteristics
The clinicopathological characteristics of the 140 ESCC patients are summarized in Table1. According to the 8th edition of the TNM classification of the American Joint Committee on Cancer (AJCC)(16), 74 patients were classified into Stage II and Stage III, and 66 cases were Stage IV. All the patients received the same regimen of concurrent CRT described above. During the study period, CR and non-CR were achieved in 36 and 104 patients, respectively. The CR rate was 25.7%. Among the 104 patients who did not achieve CR, 72 cases received adjuvant chemotherapy, and 10 cases received surgical esophagectomy. The other patients did not receive any antitumor treatments until tumor progression.
Selection of cutoff score for high expression of DAB2IP
To assess statistical significance and avoid the problems of multiple cut-point selection, X-tile program(17) was used to determine an optimal cutoff value for expression of DAB2IP protein while correcting for the use of minimum P statistics by Miller–Siegmund P-value correction. The calculated staining score of immunopositive cells ranged from 0 to 12 in all tissues. Based on the cut-point value determined by X-tile software related to survival status, we categorized the samples into low (IHC score ≤5) and high (IHC score >5) expression subgroups (Figure1C-a,b).
Reduced expression of DAB2IP protein in ESCC cell lines and ESCC biopsy tissues
In this study, the protein levels of DAB2IP were examined by Western blotting (Figure1A). All the four ESCC cell lines expressed relatively lower levels of DAB2IP than non-cancerous esophageal control tissues. In addition, EC109 cells displayed the highest levels of endogenous DAB2IP expression, whereas Kyse150 cell line showed the lowest level expression (Figure1A, left panel). In 14 paired primary tissues, 10/14 (71.4%) showed downregulated DAB2IP expression when compared with adjacent non-neoplastic esophageal tissues. The 10 representative ESCC cases with downregulated expression of DAB2IP are illustrated in Figure1A, right panels.
Next, we performed IHC to evaluate the expression of DAB2IP in the 140 ESCC biopsy specimens and 25 corresponding normal mucosal tissues. The DAB2IP staining showed weak and diffuse immunostaining in background stromal cells. Positive signals of DAB2IP protein were predominantly located in the cytoplasm of esophageal cell and also weakly stained in the nucleus. According to the criteria mentioned above, low expression of DAB2IP was observed in 105/140(75.0%) of primary ESCCs biopsy tissues compared with only 2/25 (8%) in normal esophageal mucosa (Table1).
Correlation of DAB2IP expression with clinicopathologic variables
Table 1 summarizes the detailed information about the rates of low expression of DAB2IP with respect to several standard clinicopathological features in the cohort. Results showed that no significant association was found between low DAB2IP expression and some of the clinicopathological features, such as patient’s age, gender, and tumor location(P>0.05). Interestingly, DAB2IP expression correlated closely with World Health Organization (WHO) grade (P < 0.001), Tumor size (P< 0.001), T status (P = 0.003), Node (N) status (P< 0.001), and distant lymph node metastasis (M) status (P= 0.017) (Table1).
Relationship between DAB2IP expression and CRT response
Primary CR was achieved in 25.7% (36/140) of the patients with ESCC. Moreover, using the optimal cutoff value of >5score, DAB2IP expression was also the factor that showed a significant correlation with CRT response in the cohort (rp=0.322, n=140, P <0.001) in which low expression of DAB2IP was observed more frequently in the non-CR subset than the CR subset.
Correlation between clinicopathological variables, DAB2IP expression and ESCC patient survival
Of the 140 patients with ESCC, 8 patients were lost to follow-up. The median survival time was 17.8months. The 2-year OS and DSS for the entire cohort of patients were 38.6% and 36.6%, respectively. Kaplan-Meier analysis revealed that low-level expression of DAB2IP was associated with short overall survival time (P=0.002) and disease specific survival time (P <0.001) (Figure1C-c,d).
In log-rank test, low DAB2IP expression was also found to correlate closely with poor DSS of ESCC patients (median14.77 versus 34.50months, P <0.001) (Table 2).Further univariate Cox regression analysis identified WHO grade, tumor size, T status, N status, M status, CRT response and DAB2IP expression to have a significant impact on DSS (P =0.012, 0.003, 0.001, 0.030, 0.022, <0.001 and <0.001 respectively; Table 3). Other clinicopathological variables, including age, gender and location showed no significant correlation with DSS (P =0.908, 0.317 and 0.387 respectively; Supplementary 2). The parameters that were significant in univariate analysis were further evaluated in Cox regression multivariate analysis. The results showed that the expression of DAB2IP (P <0.001), T status (P <0.001), and CRT response (P =0.004) were independent predictors of tumor prognosis ((Additional file 1: Table S1).
DAB2IP regulates the chemosensitivity of ESCC cells to cisplatin in vitro
To further investigate whether DAB2IP is capable of modulating the chemoresistance of ESCC cells in vitro, the Kyse150 cell line, which is showed the lowest endogenous expression levels of DAB2IP, was subsequently transfected with pcDNA3.1-DAB2IP or the control plasmid pcDNA3.1. As shown in Figure2A left, ectopic expression of DAB2IP in Kyse150 cells (Figure2C, left) resulted in a substantial reduction of IC50 value under cisplatin treatment compared to that of the empty vector control cells (5.52uMvs 10.97uM). Consistently, stably silencing DAB2IP by lentivirus in EC109 cells, which showed a relative high expression of DAB2IP (Figure 1A, left panel), resulted in increased IC50 value compared with the control EC109-shluc cells (1.35uMvs 2.78uM, Figure 2A right). The knockdown efficiency of shRNA lentivirus was validated by Western blot (Figure 2C, right).
To further identify the crucial role of DAB2IP in ESCCs chemoresistance, the Annexin V/PI assay was performed to evaluate the effect of DAB2IP on apoptosis. As illustrated in Figure 2, the proportion of apoptosis was dramatically increased when Kyse150-DAB2IP cells were pretreated with cisplatin(25.560±4.680%vs 38.160±5.620%, P =0.041, Figure 2B, upper panels). Accordantly, ablation of endogenous DAB2IP remarkably reduced cisplatin-induced apoptosis in EC109 cells (32.840±4.57%vs 21.280±4.890%, p=0.0403, Figure 2B, lower panels).
In agreement with the results obtained by Annexin V/PI assay, our following Western blot analysis indicated that the ectopic expression of DAB2IP in Kyse150 cells obviously increased protein levels of cleaved-caspase 3 and cleaved-PARP induced by cisplatin (Figure 2C left). On the other hand, the depletion of endogenous DAB2IP in EC109 cells led to a substantial reduction of cleaved-caspase 3 and cleaved-PARP in the presence of cisplatin (Figure 2C right). Taken together, these results provide strong evidence that DAB2IP plays a crucial role in regulating chemosensitivity of ESCC cells.
DAB2IP influences the sensitivity of ESCC cells to ionizing radiation (IR) in vitro
Since our above ESCC cohort data revealed that DAB2IP expression was significantly correlated with CRT response (Table1), we next used clonogenic assay to determine the important role of DAB2IP in ESCC radiosensitivity in vitro. As shown in Figure 3A, ectopic overexpression of DAB2IP in Kyse150 cells resulted in an obviously reduction of clonogenic survival fractions when compared with the corresponding control Kyse150-vector cells (Figure 3A left). In addition, virtually increment of clonogenic survival fractions was observed in DAB2IP repressed EC109 cells in comparison to control EC109-shluc cells. Additional non-linear curve-fitting analysis of the dose–response survival curves was performed according to the linear quadratic (LQ) model, and the parameters for each curve including α value, β value, α/β value, SF2, and sensitizing enhancement ratio (SER) were listed (Additional file 2: Table S 2). According to these values, SF2 was markedly reduced in Kyse150-DAB2IP cells compared to the control Kyse150-vecter cells (0.43 vs 0.66) with consequence that SER value was 1.52. Consistently, a substantial increase of SF2 value was observed in DABIP-silenced EC109 cells when compared with control EC109-shluc cells (0.67 vs 0.59), which indicated that the SER of DAB2IP silencing in EC109 cells was 0.87 (Additional file 2: Table S2).
Next, to obtain more evidence regarding the involvement of DAB2IP protein in regulating ESCC cell radiosensitivity, Annexin V/PI assay and immunoblots for detecting the apoptosis-related marker proteins (i.e., cleaved caspase-3 and cleaved PARP) was perfomed. As shown in Figure 3B and 3C, with the absence of IR, neither overexpression nor knockdown of DAB2IP could influence the proportion of apoptosis and the levels of apoptosis-related marker proteins. This phenomenon had changed after treating cells with 3Gy X-rays. In the presence of IR, both Annexin V/PI assay and Western blot results indicated that the apoptotic activity of DAB2IP was significantly enhanced in DAB2IP-overexpressed Kyse150 cells in contrast to control Kyse150-vector cells, whereas repression DAB2IP in EC109 cells cause a remarkable decrease of apoptotic cells (32.360±5.210%vs17.890±4.65%, P =0.023), cleaved caspase-3, and cleaved PARP when compared with corresponding control EC109-shLuc cells.
The radiosensitizing activity of DAB2IP is associated with enhanced double-stranded DNA breaks (DSB) repair in ESCC cells
It is well known that the DSB repair capacity is closely correlated with intrinsic radiosensitivity of cells.(18) DSB repair involves the rapidly phosphorylation of γH2AX and the P53 binding protein (53BP1) recruitment in site-specific DNA damage.(19) Thus, the formation of γ-H2AX and 53BP1 foci, which displayed as discrete foci at surrounding DNA double-strand breaks, is widely used to monitor radiation-induced DNA breaks and to assay DNA rejoining defects.(20)
To investigate whether DAB2IPcaninfluence DNA repair ability of ESCC cells, the dual immunofluorescence staining for 53BP1 (red) and phospho-γH2AX (green) foci was performed. The DSB repair kinetics were determined by counting the colocalized foci (yellow). As illustrated in Figure 3D upper panels, after 36h of IR treatment, the unrepaired DNA damage detected by counting colocalized foci of γH2AX and 53BP1 was significantly increased in Kyse150-DAB2IP cells when compared with control Kyse150-vector cells (97±18vs 206±22,P =0.003), while no obvious difference was observed before IR treatment (0h). Consistently, 36 hours after IR exposure, the colocalized foci number of γH2AX and 53BP1 was remarkably reduced in DAB2IP-silenced EC109 cells when compared to control EC109-shluc cells(388±36vs 246±38, p=0.038). However, without IR treatment (0h), no distinct difference was detected in these two group cells. These data indicated that the expression of DAB2IP was associated with DSB repair capability of ESCC cells.
Overexpression of DAB2IP enhances the radiosensitivity of esophageal squamous cell carcinoma cells in vivo
To further determine whether DAB2IP has similar impact on ESCC cell response to IR in vivo, Kyse150-DAB2IP and control Kyse150-vector cells were inoculated into female BALB/c nude mice. In agreement with our in vitro experiments, we found that overexpression of DAB2IP alone did not influence tumorigenicity of ESCC cells, (i.e., both the injected Kyse150-DAB2IP and control Kyse150-vector cells showed a similar efficiency and growth rate of the ESCC xenografts in nude mice)(Figure 4A).However, when the mice-bearing tumors received 6Gy of IR, the growth of the Kyse150-DAB2IP tumors was significantly inhibited (from a mean tumor volume of 180mm3 before IR treatment, increasing to 287.00±38.00mm3 at the endpoint of observation).In comparison, the increase in tumor volume of control Kyse150-vector was greater, from a mean tumor volume of 180 mm3 to 423.00±43.00mm3) (P <0.01; Figure 4A).
The stably overexpressed levels of DAB2IP measured by IHC staining in Kyse150 xenograft tumors are shown in Figure 4B. Similar to the results observed in above in vitro experiment, the in vivo radiation-induced γH2AX foci assay indicated that, 24hours after receiving 6Gy radiation, the unrepaired DNA damage was substantially increased in Kyse150-DAB2IP xenograft tumor tissues compared to parental control cell line xenograft tumors (Figure 4C).
Collectively, the in vivo and in vitro data demonstrated that DAB2IP plays a crucial role in modulating radiosensitivity of ESCC cells.
DAB2IP regulates ESCC cell radiosensitivity, possibly through enhancing IR-induced activation of ASK1-JNK signaling
Apoptosis is an important mechanism by which IR exerts its therapeutic response and faulty apoptosis is a known mechanism leading to resistance to radiation therapy.(21) In previous literature, DAB2IP has been demonstrated to be involved in several apoptotic pathways, including Ras/Raf/MEK/ERK (MAPK) signaling, PI3K/Akt signaling and ASK1-JNK signaling.(22-24) Therefore, to address which signaling pathways were critically involved in DAB2IP regulating ESCC cells radiosensitivity, we initially performed Western blot to evaluate the phosphorylation levels of AKT, JNK, and ERK, which represent the activated degree of these three signaling pathways, respectively. As shown in Figure 5A, in the present of 3Gy IR treatment, stably silencing DAB2IP by lentivirus in EC109 cells obviously increased the protein levels of p-JNK. No changes were observed in p-ERK and p-AKT levels. This result implied that ASK1-JNK signaling pathway might be an important mechanism in DAB2IP regulating ESCC cells radiosensitivity.
DAB2IP facilitates IR-induced dephosphorylation at Ser-966 of ASK1 accompanied by reinforce dphosphorylation of JNK
Next, to confirm whether ASK1-JNK pathway was involved in DAB2IP-inducedregulation of ESCC cells radiosensitivity, the stably ectopic DAB2IP-overexpressed Kyse150 cells were established and followed by 3Gy X-ray irradiation. The dephosphorylated levels of ASK1 at Ser966 and phosphorylated levels of JNK, both triggered by X-ray irradiation, were determined by Western blot. As illustrated in Figure 5B, elevated expression of DAB2IP alone in Kyse150 cells did not alter p-ASK1ser966 and p-JNK levels in the absence of IR treatment. However, when treated cells with 3Gy IR, ectopic overexpression of DAB2IP remarkably reduced the levels of p-ASK1-ser966, concomitant with a substantially increased levels of p-JNK. These observations were further supported by our results from knockdown experiments. As anticipated, silencing DAB2IP by lentivirus alone in EC109 cells did not affect the protein levels of ASK1 p-Ser966 and p-JNK in the condition without IR. However, in the presence of IR, we observed dramatically increased levels of p-ASK1 Ser966 concurrently with virtual reduction of p-JNK in DAB2IP-depleted EC109 cells when compared with our control EC109-shluc cells (Figure 5C).
DAB2IP enhances IR-induced ASK1 activation by facilitating dissociation of ASK1 from its inhibitor 14-3-3 but not from its other inhibitor, thioredoxin (Trx)
It has been well documented that 14-3-3 is one of the most important inhibitors of ASK1.The dephosphorylation at Ser-966 of human ASK1 and consequent dissociation of 14-3-3 with ASK1 has been thought to be a crucial activation mechanism of ASK1-JNK pathway.(25-28) In addition, it was further reported that DAB2IP could enhance TNF-α–induced ASK1 activation by facilitating dissociation of inhibitor 14-3-3 from ASK1.(29) Thus, we posited that DAB2IP might be capable of intensifying IR-induced ASK1 activation through a similar mechanism. To test this premise, Kyse150 cells were firstly transfected with vector pcDNA3.1 or pcDNA3.1-DAB2IP plasmid, then cells were treated with or without IR, and the association of ASK1 with 14-3-3 was determined by immunoprecipitation with anti-ASK1 followed by Western blot with anti-14-3-3. As expected, the interaction between ASK1 and 14-3-3 was substantially reduced in response to IR treatment, whereas this reduction was reinforced by elevated expression of DAB2IP (Figure 5D).Accordingly, the radiation-induced disruptive effect on the ASK1-14-3-3 complex was also observed in EC109 cells, and this disruptive effect was markedly attenuated by knocking down of DAB2IP (Figure 5E).
Besides, we also determined whether DAB2IP could affect IR-induced dissociation of ASK1 with Trx, another important negative regulator of ASK1. Intriguingly, our coimmunoprecipitation assay showed that neither knockdown nor overexpression of DAB2IP could impact on IR-induced disruption of ASK1-Trx complex (Figure 5D and 5E). This important result has excluded the possibility that DAB2IP regulated IR-induced ASK1 activation by enhancing the dissociation of ASK1 with Trx.
The enhanced activity of DAB2IP on IR-activated JNK signals dependent on the presence of ASK1
In order to further identify whether the sensitizing effect of DAB2IP in response to IR was through ASK1-JNK pathway, Kyse150 cells were firstly stably knocked down ASK1 by lentivirus, and subsequently, the control pcDNA3.1or pcDNA3.1-DAB2IP plasmid was transfected into these cells to determine whether DAB2IP can still activate JNK signal in the absence of ASK1. As shown in Figure 5F, in the presence of IR, the enhanced activity of DAB2IP toward JNK signal was virtually totally prevented in ASK1-depleted Kyse150 cells. These data suggested that the presence of ASK1 was required and might be a predominant mediator for the enhanced activity of DAB2IP toward JNK signal.