PMC:5337621 / 6731-22029 JSONTXT

{"project":"2_test","denotations":[{"id":"28218735-24240108-79088311","span":{"begin":6884,"end":6886},"obj":"24240108"},{"id":"28218735-19111882-79088312","span":{"begin":6888,"end":6890},"obj":"19111882"},{"id":"28218735-18353650-79088313","span":{"begin":8086,"end":8088},"obj":"18353650"},{"id":"28218735-26782714-79088314","span":{"begin":9828,"end":9829},"obj":"26782714"},{"id":"28218735-15860560-79088315","span":{"begin":11264,"end":11266},"obj":"15860560"},{"id":"28218735-15991339-79088316","span":{"begin":12336,"end":12338},"obj":"15991339"},{"id":"28218735-21886100-79088317","span":{"begin":12340,"end":12342},"obj":"21886100"},{"id":"28218735-23792361-79088318","span":{"begin":12908,"end":12910},"obj":"23792361"}],"text":"Results\n\nAURKA expression correlates with poor prognosis in gastric cancer\nWe first assessed the relationship between AURKA and gastric cancer prognosis by immunohistochemical analysis. To this end, we studied the expression of AURKA protein in a large cohort of 240 gastric cancer patients and found that 172 (71.7%) subjects had high AURKA expression. To further validate our immunohistochemistry results, we performed western blot analysis of paired samples from gastric tumors and their adjacent normal tissues. Consistently, we found a higher AURKA expression in gastric tumor tissues compared with their adjacent normal tissues (Figure 1a). Clinicopathologic correlation analysis showed that AURKA positivity was strongly correlated with a number of gastric cancer clinical features. The AURKA staining was positively correlated with tumor stage (P=0.006), clinical stage (P=0.002), lymph node metastasis (P=0.007) and distant metastasis (P\u003c0.0001). We did not find a significant correlation between AURKA overexpression and gender, age, sex, tumor size, tumor site and grading (Supplementary Table 1).\nMoreover, gastric cancer patients with high AURKA showed significantly poorer disease free survival (DFS, P=0.001) and overall survival (OS, P\u003c0.001) compared with low AURKA expressing gastric cancer patients (Figure 1b). Univariate analysis showed that survival time also decreased with tumor size (\u003e4 cm) (DFS, P\u003c0.001; OS, P\u003c0.001), higher tumor stage (pT) classification (DFS, P\u003c0.0001; OS, P\u003c0.0001), lymph node metastasis (DFS, P\u003c0.0001; OS, P\u003c 0.0001) and advanced pTNM stage (DFS, P\u003c0.0001; OS, P\u003c0.0001). There was no statistically significant correlation between DFS or OS and age, sex or tumor grade (Supplementary Table 2). In multivariate, analysis both lymph node metastasis (DFS, P=0.002; OS, P=0.007) and tumor stage (DFS, P\u003c0.0001; OS, P\u003c0.0001) were significant independent prognostic factors for survival (Figure 1c). In addition, multivariate analysis also indicated a significant correlation between AURKA expression and survival (DFS, P=0.030; OS, P=0.016). Thus, we showed that AURKA level, as an independent prognostic factor, was adversely associated with clinical prognosis, suggesting that the poor chemotherapy response in gastric cancer patients might be related to high AURKA expression (Figure 1c).\nImportantly, TUNEL assay showed that overexpression of AURKA reversed cell apoptotic death caused by doxorubicin in gastric cancer BGC823 cells (Figure 1d; Supplementary Figure 1). Conversely, AURKA depletion by siRNA effectively increased sensitivity to doxorubicin (Figure 1e), indicating that AURKA could override DNA damage checkpoint to promote cell survival.\n\nAURKA regulates Survivin expression in gastric cancer cells\nTo confirm our hypothesis that AURKA modulates Survivin expression in gastric cancer cells, we depleted endogenous AURKA in gastric cancer cells and examined AURKA and Survivin protein expression. The western blot analysis indicated Survivin expression dramatically decreased in AURKA-depleted AGS and BGC823 gastric cancer cells (Figure 2a), suggesting that AURKA positively regulated Survivin expression. In agreement, AURKA ectopic overexpression in both cells induced Survivin expression levels (Figure 2b), indicating AURKA was an upstream regulator of Survivin expression. It was noteworthy that silencing of AURKA had little or no effects on mRNA levels in AGS and BGC823 cells (Supplementary Figure 2), suggesting that AURKA regulated Survivin at the post-transcriptional level. Further, AURKA small molecule inhibitor VX-680 treatment also resulted in a marked decreased in Survivin protein level in a kinase-dependent manner (Figure 2c), indicating AURKA kinase activity was required for the regulation of Survivin expression. To confirm further that AURKA mediated Survivin upregulation, we performed immunohistochemistry of 62 pairs of gastric cancer specimens and found that AURKA expression was positively correlated with Survivin expression (r=0.402; P\u003c0.01; Figure 2d; Supplementary Table 3). Considering that co-overexpression of AURKA and Survivin was associated with poor prognosis of gastric carcinomas, we assessed the effect of Survivin expression on drug sensitivity. To achieve this, BGC283 cells were transfected with Survivin cDNA expressing plasmid or siRNA sequences and treated with doxorubicin followed by TUNNEL assay. As shown in Figures 2e and f, elevated Survivin expression was correlated with decreased sensitivity to doxorubicin while Survivin depletion exacerbated cell death compared with control cells in response to the DNA-damaging agent. Together, these data suggested that AURKA regulated Survivin protein expression and upregulation of Survivin levels was the critical mechanism by which AURKA caused gastric cancer cell proliferation and drug resistance to doxorubicin.\n\nVX-680 synergistically enhances the cytotoxic effect of doxorubicin by suppression of survivin\nWe next sought to determine whether the small molecule AURKA kinase inhibitor VX-680 might potentiate the genotoxic effects of doxorubicin. To this end, BGC823 cells were treated with VX-680, doxorubicin, or a combination of both, and the proliferation of the cells analyzed by MTT assay. The combination of these two agents strongly reduced cell proliferation. The inhibition rate of VX-680 and 0.5 μg/ml doxorubicin combination were 42.4% (q=1.19) and 49.7% (q=1.35), while VX-680 and 1 μg/ml doxorubicin combination were 59.7% (q=1.39) and 67.4% (q=1.51), respectively (Figure 3a). Western blot showed that combination of these two agents strongly reduced AURKA activity, with an increase in cleaved PARP expression. Most importantly, Survivin expression was induced by doxorubicin but was strongly reduced by the combination treatment (Figure 3b). Moreover, VX-680 induced cell death was significantly (P\u003c0.001) rescued by ectopic expression of Survivin (Figure 3c), indicating that Survivin was one of the key molecules targeted by AURKA signaling. Our results provided evidence that AURKA expression correlated with cell response to chemotherapy and inhibition of AURKA might potentiate the efficacy of chemotherapeutic agent, such as doxorubicin for gastric cancer therapy.\n\nAURKA suppresses Survivin polyubiquitylation and proteasomal degradation\nGiven that AURKA had little influence on Survivin mRNA levels, we next explored the underlying mechanism for AURKA-mediated Survivin expression. Depletion of AURKA led to increased Survivin protein degradation following treatment with the translation inhibitor cycloheximide (Figures 4a and b). Indeed, depletion of AURKA reduced Survivin protein half-life, indicating that AURKA stabilized Survivin and regulated Survivin levels via post-translational mechanisms.\nAs AURKA has been shown to regulate protein ubiquitylation and proteasomal degradation,10, 26 we next asked whether AURKA suppressed Survivin polyubiquitylation and proteasomal degradation. We first tested the effects of proteasome inhibitor MG132 in conjunction with AURKA kinase inhibitor or siRNA depletion on Survivin expression. Western blot analysis indicated that AURKA kinase inhibition or siRNA depletion alone reduced Survivin expression, whereas the addition of MG132 could rescue its expression level (Figures 4c and d), suggesting that the Survivin protein downregulation in response to AURKA kinase inhibition or AURKA depletion was at least partially, due to ubiquitin-proteasomal degradation. These findings were corroborated by the in vivo Survivin ubiquitylation assay which showed that AURKA knockdown decreased Survivin stability and this was associated with an increase in its polyubiquitylation (Figure 4d).\n\nAURKA modulates SCFFBXL7 to suppress Survivin degradation\nProtein proteasomal degradation precedes orchestrated events involving a series of enzymatic reactions comprising of E1 ubiquitin activating enzyme, E2 ubiquitin conjugating enzymes, and E3 ubiquitin ligases. The E1 and E2 are common enzymes, while E3 ubiquitin ligases are highly substrate specific.27 To delineate the E3 ligase responsible for Survivin degradation controlled by AURKA in gastric cancer cells, we performed mass spectrometry analysis of Survivin immunoprecipitates after treatment with MG132 (data not shown) and identified a number of putative Survivin-interacting E3 ubiquitin ligases. To confirm further our mass spectrometry-based data, we selected seven Survivin-interacting proteins with potential E3 ligase activity, and had the genes cloned and transiently expressed in BGC823 cells (Figure 5a). Amongst these E3 ligases, only FBXL7 when overexpressed reduced Survivin expression at the protein level. We then examined Survivin and FBXL7 interaction by transiently co-expressing His-tagged Survivin and Flag-tagged FBXL7 in HEK293T cells followed by coimmunoprecipitation (Figure 5b). We found the Flag-tagged FBXL7 coprecipitated with His-tagged FBXL7 and vice versa, but not with the control IgG, suggesting that Survivin interacted with FBXL7 either directly or as part of a larger complex. Next, we transiently transfected AGS and BGC823 cells with increasing amounts of Flag-FBXL7 plasmid (Figure 5c) and found Survivin protein levels decreased in a dose-dependent manner in response to FBXL7 ectopic overexpression. To confirm the specificity of FBXL7 for Survivin, we purified all components of ubiquitylation reaction using the prokaryotic protein expression system and performed in vitro ubiquitylation analysis. As indicated in Figure 5d, SCFFBXL7 promoted the generation of polyubiquitylated Survivin species, thus confirming that Survivin was a specific substrate of the SCFFBXL7 E3 ligase complex.\nRecent evidence demonstrated that AURKA could regulate c-Myc expression through cooperating with hnRNP K,8 we hypothesized that AURKA regulated FBXL7 expression in order to control Survivin steady state. To test this hypothesis, we depleted AURKA in both AGS and BGC823 cells and analyzed FBXL7 expression. As shown in Figure 6a, AURKA depletion significantly increased FBXL7 protein levels. Consistent with the Western blot results, FBXL7 mRNA levels also increased in AURKA knockdown cells (Figure 6b), indicating AURKA regulates FBXL7 expression at the transcriptional level. It is noteworthy that endogenous FBXL7 mRNA and protein levels were low in AURKA wild-type gastric cancer cells; however, both FBXL7 mRNA and protein levels reached detectable levels only after AURKA depletion. To examine AURKA-mediated FBXL7 transcriptional activity, we cloned a 1.5 kb fragment of FBXL7 upstream of the transcriptional start site into the pGL3 basic vector and tested for its promoter activity. As shown in Figures 6c and d, both total AURKA depletion and kinase activity inhibition significantly increased FBXL7 promoter activity, suggesting that AURKA suppressed FBXL7 promoter activity in a kinase-dependent manner.\n\nFBXL7 suppression by AURKA is dependent on FOXP1 transcriptional activity\nTo further explore the possible mechanism by which AURKA regulates FBXL7 expression, we examined putative transcription factor binding sites in −1.5 kb to +1 region of the FBXL7 promoter region using the MatInspector module of the Genomatix database28 (Supplementary Table 5). We also performed in vitro GST pulldown assay to determine AURKA-interacting transcription factors. For that purpose, we purified GST and GST-AURKA proteins using prokaryotic expression system and incubated them with nuclear and cytoplasmic fractions from AGS cells followed by in vitro GST pulldown. The pulldown assay showed that AURKA bound to about 35 proteins with transcription factor activity (Supplementary Table 6). Interestingly, SOX30 and FOXP1 were two factors detected by the pulldown assay and the MatInspector promoter-binding transcription factor analysis (Figure 7a). To confirm the in vitro GST-pulldown assay results, we transiently transfected HEK293T cells with His-FOXP1 and AURKA plasmids and performed coimmunoprecipitation studies and found that both AURKA and FOXP1 could pull down each other, but the control antibody did not show any binding (Figure 7b), indicating AURKA and FOXP1 indeed interacted with each other. Furthermore, we confirmed protein–protein interaction between AURKA and FOXP1 using PRISM database,29, 30 which predicts the binding site of two proteins using known template interfaces (Figure 7c). PRISM analysis results further revealed that residues 507–519 of FOXP1 located in forkhead domain might be involved in interacting with residues 127–209 of AURKA (data not shown). Moreover, mass spectrometry analysis of in vitro AURKA phosphorylated FOXP1 (data not shown) identified S83, S104 and S440 FOXP1 (Supplementary Figure 3d) as putative AURKA phosphorylated sites on FOXP1. Intriguingly, one of them (S440) was located within the transactivation domain of FOXP1.31 It was therefore conceivable that AURKA might target FOXP1 for phosphorylation and transcriptional repression.\nWe then investigated the in vivo recruitment of FOXP1 to the FBXL7 promoter by chromatin immunoprecipitation (ChIP). As shown in Figure 7d, both FOXP1 and AURKA antibodies could effectively immunoprecipitate the FBXL7 promoter DNA, indicating both FOXP1 and AURKA, are recruited to the FBXL7 promoter directly or as part of a protein complex. Next, we examined whether FOXP1 binding to FBXL7 promoter enhanced FBXL7 gene expression. To this end, we performed luciferase reporter assay in HEK293T cells using full-length wild-type FBXL7 promoter (Figure 7e). The reporter assay showed that AURKA expression did not affect FBXL7 promoter activity, while FOXP1 increased considerably the FBXL7 promoter activity. Notably, the expression of AURKA also significantly repressed the induction of FBXL7 promoter activity by FOXP1, suggesting AURKA modulated FOXP1 activity to restrict FBXL7 expression, whereas in the absence of AURKA, FOXP1 served as an activator of FBXL7 expression. To explore further the ability of FOXP1 to transactivate the FBXL7 promoter, we generated FBXL7 promoter 5′-truncation constructs to test their ability to be transactivated by FOXP1 (Figure 7f). The promoter/reporter assays showed that FOXP1 induced the promoter activity of all 5′ deletion constructs, confirming that FOXP1 could induce FBXL7 expression through its promoter. Moreover, the results also strengthened the MatInspector database findings that FOXP1 bound to multiple sites on the FBXL7 promoter. In addition, we also determined whether these transcriptional effects were mediated at the translational level. We transfected AGS cells with the empty pcDNA vector, AURKA, FOXP1, or FOXP1 and AURKA together, and investigated the FBXL7 protein levels by immunoblotting. As shown in Supplementary Figure 3a, ectopic expression of AURKA significantly downregulated FBXL7 protein levels, while FOXP1 alone upregulated FBXL7 expression. However, a combination of AURKA and FOXP1 again suppressed FBXL7 expression, indicating that AURKA negatively regulated FBXL7 expression through modulating FOXP1. Together, these results suggest that AURKA targets FOXP1 to negatively regulate the expression of FBXL7, which can in turn negatively regulate Survivin protein levels in gastric cancer cells.\n"}