Ectopic ATF5 expression is correlated with bladder cancer recurrence.
To investigate the significance of ATF5 in BLCA as well as its potential oncogenic mechanism, ATF5 mRNA expression was preliminary analyzed between BLCA tissues and paratumor tissues in an available BLCA data from The Cancer Genome Atlas (TCGA) database. Expression levels of the ATF5 gene in BLCA were markedly elevated than those of adjacent normal bladder tissues (Fig. 1A). Its gene expression was negatively correlated with overall survival outcomes of BLCA patients (P = 0.039, Fig. 1B). Furthermore, there was a negative association between ATF5 gene expression and relapse-free survival outcomes (P = 0.013, Fig. 1C), indicating that ATF5 might be a predictor of recurrent risk in BLCA.
To assess the correlation of ATF5 expression and clinicopathological characteristics of BLCA, 140 BLCA specimens (102 non-relapse BLCA tissues and 38 recurrent BLCA tissues after treatment) were stained using a human ATF5 antibody. ATF5 expression in BLCA specimens was markedly elevated than in paratumor tissues (Fig. 1D). In addition, ATF5 expression levels in recurrent BLCA specimens were markedly elevated than those of non-relapse specimens (Fig. 1D). Statistic analysis of the 140 BLCA samples revealed that ATF5 expressions were associated with clinical grade (P = 0.034), stage (P = 0.014), tumor multiplicity (P = 0.004), recurrence (P = 0.022) as well as progression (P = 0.016), however, they were not correlated with gender, tumor size or age (Table 4). Importantly, Kaplan-Meier as well as log-rank analyses demonstrated that ATF5 expressions were negatively related to overall survival outcomes (P = 0.012, Fig. 1E) and relapse-free survival outcomes (P = 0.019, Fig. 1F). These results showed that ectopic ATF5 expression was closely related to recurrence, which predicted a poor prognosis of BLCA.
Table 4
Correlation between ATF5 Expression and Clinicopathologic Charateristics of 140 Bladder Cancer.
Characteristic | ATF5 Protein level | Chi-square test |
High (65) | Low(75) | P value |
Age (> 60 versus ≤ 60 years) | 36/29 | 42/33 | 0.942 |
Gender (M versus F) | 46/19 | 60/15 | 0.238 |
Grade (Low versus High) | 42/23 | 61/14 | 0.034* |
Stage (Ta-1 versus T2-4) | 45/20 | 65/10 | 0.014* |
Tumor Size (> 3cm versus ≤ 3cm) | 21/44 | 14/61 | 0.079 |
Multiplicity (Single versus Multiple) | 37/28 | 60/15 | 0.004** |
Recurrence (Yes versus No) | 24/41 | 14/61 | 0.022* |
Progression (Yes versus No) | 18/47 | 8/67 | 0.016* |
Elevated ATF5 expression in bladder cancer cells enhanced a tumor initiating cells (TIC)-like phenotype.
GSEA analysis was conducted to further assess the potential biological roles of ATF5 in cancers. GSEA analysis of the public BLCA database from TCGA showed a positive correlation between ATF5 expression and stemness signatures (Fig. 2A), which was tightly associated with cancer recurrence. Moreover, qRT-PCR and WB showed that the expression levels of stemness-associated biomarkers, such as OCT4, ABCG2 and SOX2 were markedly elevated in ATF5-overexpressed SW780 and UM-UC-3 cells, whereas reduced in ATF5-silenced bladder cancer cells (Fig. 2B-2D). Subsequently, the tumor sphere formation test was carried out to investigate the role of ATF5 in self renewal of spherogenic bladder cancer cells. ATF5-knockdown cells formed fewer spheres with fewer cells relative to the control group (Fig. 2E), whereas ATF5-overexpressing cells formed more spheres with more cells (Fig. 2F). These findings indicated that ATF5 upregulation enhanced tumorigenic capability of bladder cancer cells in vitro.
To evaluate the oncogenic significance of ATF5, through the subcutaneous route, bladder cancer cells were inoculated into the inguinal area of nude mice. Tumor formation in ATF5-overexpressing SW780 cells was markedly greater, relative to the control group (Fig. 2G). Besides, ATF5-knockdown cells formed significantly smaller tumors with lower tumorigenic capability (Fig. 2G). This data demonstrated that ATF5 upregulation enhanced bladder cancer cell tumorigenic capability in vivo.
ATF5 upregulation stimulates Wnt/b-catenin signaling pathway.
To investigate the mechanism underpinning ATF5-mediated TIC-like phenotype in BLCA, GSEA of TCGA-BLCA datasets was performed to analyze the potential correlation of ATF5 expressions and genes regulated by multiple signaling signatures. ATF5 gene expression was positively related to Wnt-related and beta-catenin-related gene signatures (Fig. 3A), implying that ATF5 stimulated Wnt/β-catenin signaling. Moreover, TOP/FOP assay demonstrated that ATF5 upregulation significantly promoted, whereas ATF5 knockdown repressed the transcriptional stimulation of TCF/LEF in the SW780 as well as UM-UC-3 cells (Fig. 3B). Consistently, WB anslysis demonstrated that ATF5 upregulation significantly elevated the active-β-catenin signals, while ATF5 attenuation decreased them (Fig. 3C). Expressions of downstream genes of the Wnt/β-catenin signaling, such as CD44, TCF1, JUN and CCND1, were elevated in ATF5-overexpressing cells, whereas these targets were reduced in ATF5-knockdown cells (Fig. 3D). Overall, Elevated ATF5 could stimulate Wnt/β-catenin pathway.
ATF5 binds, and activates the promoter of DVL1 gene.
JASPAR analysis was conducted to identify the potential downstream targets of ATF5 in the Wnt/β-catenin pathway. DVL1 gene promoters were predicted to include ATF5 binding sites (Fig. 4A). ChIP test validated that ATF5 could directly bind to DVL1 promoters (Fig. 4A-4B). Moreover, ATF5 upregulation enhanced DVL1 promoter-driven luciferase activity (Fig. 4C) and DVL1 expression (Fig. 4D-4E) in bladder cancer cells, while ATF5 knockdown exhibited the opposite effects (Fig. 4C-4E). These data further validated that ATF5 could directly bind to and positively stimulate DVL1. Subsequently, we explored whether DVL1 stimulation was necessary for ATF5 to promote the TIC-like phenotype of bladder cancer cells. Knockdown of DVL1 in ATF5-overexpressing cells (Fig. 4F) inhibited the expression level of stemness-related markers, including OCT4, ABCG2, and SOX2 (Fig. 4G), and expressions of downstream genes of Wnt/β-catenin pathway (Fig. 4H). Importantly, DVL1 knockdown decreased the sphere-formation capability of ATF5-overexpressing cells (Fig. 4I). These data demonstrated that DVL1 stimulation was necessary for ATF5 to enhance a TIC-like phenotype of bladder cancer cell.
Altogether, this result showed that ATF5 promoted tumorigenic capability of bladder cancer cells by directly binding and promoting DVL1 to stimulate the Wnt/β-catenin pathway.
Clinical correlation of ATF5/DVL1/beta-catenin axis in bladder cancer.
Finally, qRT-PCR and WB analyses (Fig. 5A) were performed to check if the Wnt/ATF5/DVL1 axis in bladder cancer cells is of clinical significant. There was a positive association between ATF5 protein and the DVL1 gene (P = 0.011, R2 = 0.734), DVL1 protein (P = 0.002, R2 = 0.826) and the active-β-catenin expression (P = 0.008, R2 = 0.722) in eight fresh BLCA samples (Fig. 5B-5E). In summary, these results showed that ATF5 overexpression in BLCA was positively related to DVL1 expression, which in turn stimulated Wnt/β-catenin pathway to promote tumorigenic capability of bladder cancer cell.
Aberrant ATF5 amplification is involved in ATF5 overexpression in bladder cancer cells, and is correlated with poor prognostic outcomes.
In various cancers, the ATF5 locus on chromosome 19q13.33 is generally amplified. The copy number variation (CNV) of ATF5 and DVL1 in TCGA-BLCA datasets were shown in Fig. 6A and 6B. ATF5 and DVL1 mRNA levels were tightly related to the CNV (Fig. 6C-6D). The amplification in CNV of ATF5 and DVL1 predicted a higher gene expression, while the deletion in CNV of AFT5 and DVL1 predicted a lower gene expression (Fig. 6E-6F). However, the CNV of ATF5 was not directly associated with DVL1 gene expression (Fig. 6G), indicating that ATF5 CNV does not contribute to the ectopic expression of DVL1 gene. Although aberrant CNV of DLV1 contributed to ectopic expression of DVL1 gene, differences in survival outcomes between DVL1 amplification or deletion patients were insignificant (P = 0.379, Fig. 6H). Notably, bladder cancer with ATF5 amplification was found to be a predictor of worse survival outcomes than those without ATF5 amplification (P = 0.014, Fig. 6I). These results indicated that aberrant ATF5 amplification plays a role in ATF5 overexpression in bladder cancer, which conferred a poor prognosis.