Correlations between TBL1XR1 and recurrence of colorectal cancer

More than 25% localized CRC patients died from post-operative metastasis, and risk of metastasis varies among individuals due to the high heterogeneity of CRC. Therefore, figuring out potential biomarkers for disease recurrence would be invaluable to improve the follow-up efficiency and clinical treatment. Transducin (β)-like 1 X-linked receptor 1 (TBL1XR1) is a core component of the nuclear receptor corepressor complex, which functions as a repressive coregulatory factor for multiple transcription factors. The clinical significance of TBL1XR1 in CRC hasn’t been fully elucidated. In this study, we investigated the expression of TBL1XR1 in primary CRC tissues and liver metastases from TNM stage IV CRC patients, and found that its expression in primary tumor tissues was an independent prognostic factor for tumor recurrence. Thus, we enrolled another cohort including TNM stage I-III patients to further evaluate the relationship between TBL1XR1 expression and disease recurrence. Accordingly, high TBL1XR1 expression indicates poor disease-free survival of stage I-III CRC patients. Furthermore, we confirmed the importance of β-catenin signaling pathways in TBL1XR1-mediated CRC cell oncogenicity by clinical and cellular results. Our results emphasize the necessity of individual therapy decisions based on clinical biomarkers, especially for localized CRC patients who are not routinely treated with adjunctive chemotherapy.

Scientific RepoRts | 7:44275 | DOI: 10.1038/srep44275 Therefore, we further enrolled another cohort which included stage I-III patients to evaluate whether TBL1XR1 can be helpful in predicting metastasis and recurrence. Univariate and multivariate analyses showed that high expression of TBL1XR1 is an independent risk factor for post-operative recurrence of localized and regional CRC patients. Furthermore, we performed cellular studies combined with gene overexpression and knock-down methods, and biofunctional studies revealed that β -catenin was the key molecular in regulating TBL1XR1-mediated cell proliferation and invasion.

Results
Characteristics of patients in cohort I and their correlations with TBL1XR1 expression. The selection criteria for CRC patients were showed in Fig. 1. Briefly, we selected 47 stage IV CRC patients with synchronous liver metastasis (CRCLM) as cohort I to investigate the possible relationship between TBL1XR1 and CRC metastasis. Among them, 39 patients (83.0%) underwent R0 resection for both primary tumors and liver metastases, whereas the other 8 patients underwent R1 resection. Thirty-seven cases (78.7%) were treated with systematic adjunctive chemotherapy before detectable metachronous metastasis. Detailed clinicopathological characteristics of CRCLM patients were shown in Table 1.
TBL1XR1 was identified as high expression in 27 primary CRC tissues (57.4%) and 34 liver metastases (72.3%), which was predominantly localized in the nucleus with slight immunoreactivity in cytoplasm of CRC cells ( Fig. 2A-H). High expression of TBL1XR1 in primary tumor tissues was correlated with increasing number of liver metastases (P = 0.020, Table 1). However, we didn't find any significant relationships between patients' clinical features and TBL1XR1 levels in liver metastases. One possible explanation is that the protein patterns and functions in primary CRC cells may be more principle in determining the metastatic malignancy. Also, the microenvironment of tumor cells in liver metastases can be different compared with those in primary colorectal locations, which will significantly alter the expression and modifications of proteins. TBL1XR1 expression in primary tumor tissues is an independent predictive factor for DFS of patients in cohort I. We retrospectively achieved the information about tumor recurrence of the patients in cohort I. Totally 35 cases occurred metachronous metastasis, and the 3-year disease-free survival (DFS) was 45.8%. According to the Kaplan-Meier survival curves and log-rank test, we identified that high expression of TBL1XR1 in either primary tumor tissues (P = 0.014, Fig. 2I) or liver metastases (P = 0.041, Fig. 2J) was correlated with poor DFS. In addition, bilobar distribution of liver metastases (P = 0.043), R1 resection margin (P = 0.031) and absent of adjunctive chemotherapy (P = 0.007) were all unfavorable factors for DFS of CRCLM patients (Table 2).
Clinicopathological features of patients in cohort II. TBL1XR1 expression was shown to be correlated with the disease recurrence of the stage IV CRC patients according to the results from cohort I study. Therefore, we were interested to further investigate the possible role of TBL1XR1 in predicting metachronous metastasis for localized and regional CRC patients, which was critical in improving clinical decisions.
The cohort II patients included 19 cases with TNM stage I, 59 cases with stage II and 58 patients with stage III (Fig. 1). None of the stage I-II patients received any chemotherapy before detectable recurrence, whereas 46 stage III cases underwent routinely chemotherapy with fluorouracil and cisplatin after primary surgery resection. Totally, 8 cases occurred LN metastasis, 2 cases occurred lung metastasis, and 29 cases occurred liver metastasis during our follow-up. Other patients' characteristics were shown in Table 4.  Expression of TBL1XR1 in cohort II patients and its clinical significance. The expression pattern of TBL1XR1 in cohort II patients were revealed by IHC ( Fig. 3A-F). The high expression ratio of TBL1XR1 was 21.1%, 34.8% and 53.4% for stage I-III CRC patients, respectively (Fig. 3G). Moreover, the mean immunoreactivity score (IRS) of TBL1XR1 was positively correlated with clinical stage (Fig. 3H), indicating that patients with more advanced TNM stage showed higher TBL1XR1 expression level.
We also performed the survival analysis for cohort II patients, which demonstrated that high expression of TBL1XR1 can indicate high recurrence risk of stage I-III CRC patients (Fig. 3I, Table 5) from both univariate and multivariate analyses (HR = 3.716, 95% CI 1.858-7.433, P < 0.001). Similarly, TBL1XR1 was significant in predicting metachronous metastasis for stage I-II subgroup patients (P < 0.001, Fig. 3J, Supplemental Table 1).
However, for the stage III subgroup, TBL1XR1 showed no statistical significance in DFS evaluation (P = 0.266, Supplemental Table 2). This perhaps due to the fact that most of the stage III patients underwent chemotherapy (46/58 cases, 79.3%), which was critical in inhibiting post-operative metastasis. To eliminate the effects of chemotherapy, we further evaluate the DFS patterns of the patients without adjunctive chemotherapy in both cohort I and cohort II. The Kaplan-Meier survival analysis showed that high expression of TBL1XR1 can significantly increase the recurrence risk of non-chemotherapy treated patients (P < 0.001, Fig. 3K, Supplemental Table 3). TBL1XR1 promotes cell proliferation and invasion by up-regulating the transcription activity of β-catenin. Interestingly, we found that TBL1XR1 expression was positively correlated with the Cyclin D1 (P = 0.035, Fig. 4A-C) and c-Myc levels (P = 0.047, Fig. 4D-F) in CRC tissues, both are well-known β -catenin-regulated proteins. In contrast, E-cadherin level was negatively correlated with TBL1XR1 level (P < 0.001, Fig. 4G-I). Thus, we further studied the cellular functions of TBL1XR1 in regulating CRC oncogenicity.
TBL1XR1 showed higher expression level in SW480 cells than that in normal colon epithelial NCM480 cells (Fig. 5A). To elucidate the underlying mechanisms of TBL1XR1 in promoting CRC development, we tested the effect of TBL1XR1 overexpression and knock-down in SW480 CRC cells. Luciferase assay demonstrated that overexpression of TBL1XR1 can significantly enhance the transcription activity of β -catenin (Fig. 5B).
The mutation and continuous activation of β -catenin in CRC has been well-investigated [14][15][16] , therefore, the role of TBL1XR1 in regulating Wnt pathway may be mediated by cross-talk with β -catenin. Our results showed that TBL1XR1-siRNA can down-regulate the level of EMT proteins, such as Vimentin, Snai1, Slug, Twist1, and N-cadherin (Fig. 5C). Meanwhile, the expression of Cyclin-D1 and c-Myc were also inhibited under TBL1XR1-siRNA. On the other hand, overexpression of TBL1XR1 in SW480 cells can up-regulate the expression of Vimentin, Slug, c-Myc, etc. We next transfected the inactive β -catenin mutant, TCF-4-dn, to better verify the participate of β -catenin in regulating TBL1XR1 signaling pathway. Western Blot showed that both β -catenin-target proteins and EMT proteins were less expressed in the cells co-transfected with TBL1XR1 and TCF-4-dn, compared with the cells only overexpressed TBL1XR1 (Fig. 5C). In addition, overexpression of TBL1XR1 can dramatically enhance cell proliferation and invasion capacities, while co-transfection with TCF4-dn can almost abolish the oncogenic effects of TBL1XR1 (Fig. 5D,E).

Discussion
Most the CRC patients in the localized stages (TNM stage I and II) can be treated effectively by surgical resection with the 5-year OS up to 95% for stage I and 60-80% for stage II, respectively. However, the majority of patients with metastatic colorectal cancer in the advanced stages (TNM stage III and IV) have quite a poor prognosis since the 5-year OS drop dramatically to 35% with regional stage (stage III) and to 10% with distant stage (stage IV) 17,18 . Moreover, up to 25% of localized CRC patients ultimately die due to post-operative metastasis 19 . Thus, metastasis is the decisive and the most lethal event during this disease course. Molecular biomarker for cancer may be helpful for prediction of metastasis or therapeutic treatment 20 .
In this study, we explored the expression pattern of TBL1XR1 in CRC patients with different clinical stages (TNM I-IV), including both the primary tumor tissues and liver metastases. Consequently, we found that TBL1XR1 expression in primary CRC tissues can significantly help to predict disease relapse and recurrence   Most of the stage I-II patients, which are lymph-node negative, don't receive any post-operative adjunctive chemotherapy in clinical practice. Therefore, endogenous protein alterations are the predominant factors affecting the prognosis for those localized CRC patients. Liver is the most frequent metastasis localization for stage IV patients due to the near distance and sufficient circulation drainage. Although the predominant prognostic factor for stage IV CRC patients can be the status of metastasis, the biological changes in primary CRC determine the metastasis and recurrence pattern 21 . In contrast, stage III patients are more likely to suffer from metachronous metastasis or recurrence. However, the recurrence rate of stage III patients is not higher than that of stage II CRC patients during our follow-up due to the application of systematic post-operative chemotherapy. In fact, stage III patients were advised for routinely chemotherapy while stage I-II patients were not. Therefore, we suspect that current chemotherapy strategies may partly inhibit the TBL1XR1-regulated tumor aggression because we didn't find clinical significance of TBL1XR1 in stage III patients. On the other hand, it was reported that TBL1XR1 can suppress cisplatin sensitivity in nasopharyngeal carcinoma 22 . Whether TBL1XR1 is involved in drug resistance for CRC patients and its potential in novel chemotherapy development remain to be further investigated. Anyway, we verified that TBL1XR1 was correlated with high recurrence risk for those patients who didn't receive systematic adjunctive chemotherapy after surgery. Taken together, our clinical findings emphasized the clinical significance of TBL1XR1 expression in CRC patients with different clinical stages.
Although the involvement of TBL1XR1 in various malignancies has been identified, the functional study about how it participates in metastasis of CRC is poorly understand. Our cellular studies revealed that TBL1XR1 can up-regulate the transcription activity of β -catenin. Taking into consideration that SW480 cells possess APC mutation and high basal β -catenin level 23 , we transfected the inactive mutant of β -catenin (TCF-4-dn) to competitively inhibit its activity. Overexpression of TBL1XR1 can promote EMT and cell oncogenicity, while inhibition of β -catenin impaired its effect. Therefore, the tumor-promoting effect of TBL1XR1 is at least partially mediated through regulating the transcription activity of β -catenin.

Conclusion
Our study not only identified TBL1XR1 as an independent prognostic factor for the recurrence of CRC patients, but also emphasized the potential of biomarker identification as the guidance for chemotherapy treatment of early stage CRC patients.

Patients and Methods
Patients selection. As shown in Fig. 1, we initially selected 49 CRC patients with synchronous liver metastasis who underwent surgical resection for both primary tumor tissues and liver metastases in Qilu Hospital, Qianfoshan Hospital, Shanxian Central Hospital, or Yuncheng People's Hospital between 2000 to 2013. Among them, 2 patients occurred disease relapse with in the first 6 months after surgery and were excluded. Finally, 47 stage IV CRCLM patients were enrolled as cohort I in this study. Another 162 stage I-II CRC patients were collected at the beginning. However, we later found that 56 patients among them were resected with less than 12 lymph nodes, which may increase the false negative possibility of lymph node metastasis. The World Congress of Gastroenterology proposed examination of a minimum of 12 lymph nodes for classification of tumors as Stage II 24 . Another 18 patients among them received post-operative chemotherapy. Therefore, both the 56 patients and 18 patients were excluded to better evaluate the clinical significance of TBL1XR1. In addition, 60 cases of stage III CRC patients were collected while later 2 patients were identified with disease relapse within 6 months after surgery, who were also excluded from our study. Finally, the cohort II group included 88 stage I-II and 58 stage III CRC patients (Fig. 1).
This study was approved by the Ethics Committee of Shandong University. Written informed consents were obtained from all the patients. The design of this study is accordance with the guidelines of CONSORT Statement (http://www.consort-statement.org/).
Luciferase assay. pcDNA3.0-TBL1XR1 or TBL1XR1-siRNA transfected SW480 cells were seeded in 24-well plates. After 24 h culture, β -catenin responsive firefly Luciferase reporter plasmids and Renilla reporter plasmids (pRL-TK) were co-transfected into the cells. After another 16 h culture, the cells were stimulated with Wnt3a (200 ng/mL) for 6 h. Then lysed the cells and measured both Luciferase and Renilla activities with Dual-Luciferase Reporter Assay kit (Promega) according to the manufacturer's instruction. The final Luciferase activity was normalized to the Renilla activity. Each experiment was performed in triplicate and repeated at least three times.
Invasion assay. Transfected SW480 cells were subjected to a Transwell migration assay. 1 × 105 cells were planted onto the upper chamber and cultured with DMEM containing 10% FBS for 4 h, then replaced the medium with FBS-free DMEM. The lower chamber was filled with DMEM containing 10% FBS. After incubation for 24 h, cells in the upper surface of membrane were carefully removed with a cotton swab, and cells that invaded to lower surface were fixed and stained with 0.5% crystal violet. The number of invading cells was counted from five randomly selected visual fields.
Statistics. Disease free survival (DFS) was defined as the period from tumor resection to the date of recurrence (including local recurrence, regional recurrence and distant recurrence) or till the date of our last follow-up. Data analyses were performed with SPSS 20.0 and Graphpad Prism 5.0 software. The correlations between TBL1XR1 expression with patients' clinicopathological characteristics were evaluated with chi-square test. The quantitively IRS comparison among patients with different TNM stages were carried out by one-way ANOVA analysis. Kaplan-Meier survival analysis and log-rank test were used to identify the prognostic value of various parameters, and the factors with statistical significance were further subjected to multivariate analysis by Cox regression hazard model. A P < 0.05 was considered as statistical significant (two-tailed).