RBX1 overexpression is correlated with poor ATC prognosis

To understand the clinical relevance of RBX1 to ATC, we measured the RBX1 expression in 30 ATC tissue samples together with corresponding adjacent tissues using western blotting and qRT-PCR. RBX1 mRNA was overexpressed in tumor samples compared with their non-neoplastic counterparts (Fig. 1A, B). Furthermore, the levels of RBX1 protein were markedly elevated in ATC tissues, which was in agreement with the qRT-PCR results (Fig. 1C, D). As shown in Fig. 1E, RBX1 is overexpressed in 62.6% of the ATC tissue specimens. These results indicated that the expression levels of RBX1 protein were significantly enhanced in ATC tissues (Fig. 1E). Immunoblotting indicated that the expression of RBX1 in the ATC cell lines was markedly higher compared with that in the PTC and thyroid cell lines (Fig. 1F, G). Next, the correlation analysis between RBX1 protein overexpression and ATC clinicopathological parameters indicated that the patients with low levels of RBX1 had a significantly longer overall survival time than those with high levels of RBX1 (Fig. 1H). Overall, these findings revealed the hypothesis that RBX1 plays a role in ATC progression and can be a potential biomarker for ATC diagnosis.

Fig. 1

RBX1 is overexpressed in ATC and closely correlated with poor prognosis in patients. A, B RBX1 mRNA expression levels in the ATC tissues together with the normal tissues adjacent to the tumor were investigated by qRT-PCR. **P < 0.01. C, D Expression levels of RBX1 protein in ATC tissues and the normal tissues adjacent to the tumor were investigated by western blotting. **P < 0.01. E RBX1 protein expression in the ATC tissues and the normal tissues adjacent to the tumor were determined using immunohistochemistry. Scale bar, 50 μm. F, G RBX1 protein and mRNA levels in ATC and PTC cell lines. ***P < 0.001. H Kaplan–Meier curves were used to visualize the overall survival of both low and high RBX1 expression of ATC patients. ***P < 0.001

RBX1 regulates the migration and invasiveness of ATC cells in vivo and in vitro

We knocked down RBX1 with subtype-specific shRNAs in two highly invasive ATC cell lines, namely, CAL62 and KMH-5M. Compared with scrambled shRNA, both shRBX1 and shRBX1 significantly reduced RBX1 expression in stable cell lines (Fig. 2A and Additional file 1: Fig. S1A, B) together with the stable overexpression of RBX1 in 8305C cell lines (Fig. 2B and Additional file 1: Fig. S1C). Transwell assays also revealed that the knockdown of RBX1 markedly inhibited the metastatic ability of KMH-5M and CAL62 cells, while the overexpression of RBX1 enhanced the metastatic ability of 8305C cells (Fig. 2C, D and Additional file 1: Fig. S1D-E). The proliferative capacity of the ATC cells was significantly suppressed by the knockdown of RBX1 in contrast to control cells. However, the overexpression of RBX1 significantly promoted their proliferative capacity (Fig. 2E, F and Additional file 2: Fig. S2). We conducted an in-depth investigation to understand the effects of RBX1 on the metastasis of ATC by constructing the tumor models in nude mice, which were classified into shRBX1 and shNC groups. Consecutive lung sections showed that the number of lung micrometastases in ATC patients was remarkably attenuated in the shRBX1 group (Fig. 2G). On the contrary, RBX1 overexpression enhanced the number of metastatic nodules in the lungs (Fig. 2H). In conclusion, these findings reflect that the stable RBX1 knockdown can suppress the metastasis together with the invasion of ATC in vivo and in vitro, and serve as a candidate oncogene in the development and metastasis of ATC.

Fig. 2

RBX1 facilitated the proliferation and migration of ATC cells. A Western blotting was performed to identify the RBX1 expression levels in CAL62 and KMH-5M cells stably transfected with shRBX1 plasmid. B RBX1 expression levels in the 8305C cells stably transfected with HA-RBX1 plasmid were measured using a western blot. C Transwell assays of CAL62 cells transfected with shRBX1 plasmid. **P < 0.01. D Transwell assays of 8305C cells transfected with HA-RBX1 plasmid. **P < 0.01. E EdU assays of CAL62 cells transfected with shRBX1 plasmid. **P < 0.01. F EdU assays of 8305C cells transfected with HA-RBX1 plasmid. **P < 0.01. G, H H&E staining of the sections of metastatic nodules in the lungs embedded with paraffin. *P < 0.05, **P < 0.01

RBX1 promotes ATC development by increasing the Warburg effect

E3 ubiquitin ligases can promote metabolic reprogramming in the development of different types of cancer [32, 33]. As the Warburg effect is characterized by a metabolic shift that is ubiquitous in the tumor cells, involving ATC, we investigated the effect of RBX1 on the glucose metabolism in ATC. Glucose-6-phosphate (G6P) levels, lactate generation, glucose consumption, and cellular levels of ATP were substantially reduced in CAL62 cells after RBX1 knockdown (Fig. 3A), whereas the overexpression of RBX1 led to the contrasting effect in 8305C cells (Fig. 3B). To further confirm the effect of RBX1 on ATC glycolysis, ECAR was determined, which indicates the overall glycolytic flux. The knockdown of RBX1 significantly attenuated the capacity and rate of glycolysis in the CAL62 cells (Fig. 3C, D), while the overexpression of RBX1 enhanced ECAR in the 8305C cells (Fig. 3E, F). As an indicator of mitochondrial respiration, OCR was enhanced in the CAL62/shRBX1 cells (Fig. 3G, H), whereas RBX1 overexpression decreased OCR in 8305C cells (Fig. 3I, J).

Fig. 3

RBX1 promotes ATC progression by enhancing the Warburg effect. A Cellular glucose consumption, G6P levels, ATP levels, and lactate generation in CAL62/shRBX1 cells. *P < 0.05. B Cellular glucose consumption, G6P levels, ATP levels, and lactate generation in 8305C/HA-RBX1 cells. *P < 0.05. C, D ECAR data showing the glycolytic rate and capacity in CAL62/shRBX1 cells. E, F ECAR data showing the glycolytic rate and capacity in 8305C/HA-RBX1 cells. G, H OCR results showing basal respiration and maximum respiration in CAL62/shRBX1 cells. I, J OCR results showing basal respiration and maximum respiration in 8305C/HA-RBX1 cells. K, L Lactate generation by 8305C/p-RBCK1 or CAL62/shRBX1 cells in the presence of 2-DG. *P < 0.05, **P < 0.01. M, N Role of 2-DG in the invasion and migration of 8305C/p-RBCK1 or CAL62/shRBX1 cells. *P < 0.05, **P < 0.01. O–Q Culturing 8305C cells in a medium with galactose but without glucose abolished the impact of RBX1 overexpression on cell invasion and migration, NS = No Significant

To investigate whether the Warburg effect led to the ATC cell development, 8305C/HA-RBX1 and CAL62/shRBX1 cells were treated with different concentrations of 2-deoxyglucose (2-DG) for one day. 2-DG remarkedly suppressed the glycolysis in 8305C/HA-RBX1 and CAL62/shRBX1 cells in a dose-dependent mode (Fig. 3K, L). Additionally, the invasive and migration ability of 8305C/HA-RBX1 and CAL62/shRBX1 cells were attenuated in a dose-dependent manner (Fig. 3M, N). To confirm that glycolysis regulates the invasion and migration of ATC, cells were cultivated in a medium containing galactose rather than glucose, thereby decreasing glycolytic flux and driving them to be dependent on oxidative phosphorylation. This significantly decreased the increase in 8305C cell invasion and migration resulting from the overexpression of RBX1 (Fig. 3O–Q). These results display that RBX1 inhibits oxidative phosphorylation while facilitating the aerobic glycolysis in ATC cells, but facilitates invasion and migration by enhancing the Warburg effect in the ATC cell lines.

PKM2 is essential for RBX1 to enhance the Warburg effect

In contrast to PKM1, higher PKM2 expression is a critical factor for promoting the Warburg effect in cancer cells compared to normal cells. Decreased expression of PKM2 might result in the suppression of the Warburg effect, thereby suppressing the tumorigenic potential [34, 35]. Thus, we determined whether RBX1 regulates PKM2 expression by initially measuring the PKM2 expression in RBX1-knockdown and -overexpressing ATC cells. Western blotting analysis revealed that the knockdown of RBX1 significantly reduced the expression of PKM2 in CAL62 cells (Fig. 4A). In contrast, the RBX1 overexpression markedly enhanced the expression of PKM2 in 8305C cells (Fig. 4B). Additionally, the enhanced PKM2 levels reversed the reduction of PKM2 expression in CAL62/shRBX1 cells (Fig. 4C). Rescue experiments indicated that the restoration of PKM2 expression could abrogate the decrease in metastatic capacity of ATC resulting from RBX1 silencing (Fig. 4D). Figure 4F shows in vivo tumor metastasis, and the ATP levels in ATC cells. Simultaneously, the knockdown of RBX1 attenuated ECAR in ATC cells, while concomitant PKM2 overexpression decreased the reduction in glycolytic capacity and rate (Fig. 4G, H).

Fig. 4

RBX1 promotes ATC progression by upregulating PKM2 expression. A, B qRT-PCR and western blotting were performed for measuring the PKM2 and RBX1 expression. C Western blotting was performed to determine the PKM2 and RBX1 expression in various groups. D Quantification of transwell assay in various groups. *P < 0.05, **P < 0.01. E Quantification and representative images of the lung metastases in various groups of nude mice (n = 6). F Cellular glucose consumption, G6P levels, ATP levels, and lactate generation in the specific groups. *P < 0.05, **P < 0.01. G, H Measurement of OCR and ECAR in the specific groups. *P < 0.05. I PKM2 and RBX1 expression levels in various groups were determined using western blotting. J. Quantification of transwell assay in diverse groups. *P < 0.05, **P < 0.01. K Quantification and representative images of the lung metastases in various groups of nude mice (n = 6). L Cellular glucose consumption, G6P levels, ATP levels, and lactate generation in the specific groups. *P < 0.05, **P < 0.01. M, N ECAR and OCR were measured in the indicated groups. *P < 0.05, **P < 0.01

Subsequently, we discussed the effect of attenuated PKM2 expression on the PKM2 and RBX1 protein levels as well as on the cell invasion and migration in 8305C cells with RBX1 overexpression. Western blotting results demonstrated that the overexpression of RBX1 significantly enhanced PKM2 expression, while the knockdown of PKM2 largely attenuated the increase in PKM2 expression caused by RBX1 in 8305C cells (Fig. 4I). At the same time, a reduction in PKM2 significantly decreased the cell invasion and migration enhanced by RBX1 (Fig. 4J). Besides, the analysis of in vivo metastasis revealed that the decrease in PKM2 reduced the incidence of pulmonary and intrahepatic metastasis in the 8305C-RBX1 group (Fig. 4K). PKM2 knockdown led to a reduction in G6PD activity, ATP levels, lactate generation, and glucose consumption mediated by RBX1 in ATC cells (Fig. 4L). The overexpression of RBX1 upregulated ECAR in ATC cells, while the concomitant PKM2 knockdown hindered the increase in glycolytic capacity and rate (Fig. 4M, N). These results display that PKM2 is a type of functional downstream target of RBX1 to modulate aerobic glycolysis and is essential for RBX1-mediated tumor development.

RBX1 regulates PKM alternative splicing to promote the PKM2-mediated Warburg effect

Cancer cells achieve the metabolic advantage over normal cells by modulating PKM alternative splicing and facilitating the PKM2 expression [19]. To further demonstrate whether RBX1 is responsible for this phenomenon, we detected the expression levels of RBX1, PKM2, and PKM1 in ATC patient samples and then compared these levels with those in the corresponding adjacent non-cancerous tissues using western blotting. PKM2 and RBX1 protein levels were significantly enhanced in ATC tissues. Besides, scatter plots exhibited a positive association between the expression of PKM2 and RBX1 (Fig. 5A, B). However, the expression of PKM1 was markedly low in the ATC tissues, and the expression of RBX1 and PKM1 were negatively correlated (Fig. 5C, D). Based on the positive relationship between the expression of PKM2 and RBX1, together with the inverse relationship between PKM1 and RBX1 observed in ATC, these findings display that RBX1 exerts a crucial role in modulating the alternative splicing of PKM.

Fig. 5

RBX1 regulates PKM alternative splicing. A Western blot was performed to determine the PKM2 and RBX1 protein expression levels in ATC tissues together with the normal tissues adjacent to the tumor. B Scatter plots of RBX1 and PKM2 protein expression in ATC. C Western blot was performed to determine the PKM1 and RBX1 protein expression levels in ATC tissues together with the normal tissues adjacent to the tumor. D Scatter plots of RBX1 and PKM1 protein expression in ATC. E qRT-PCR showing expression levels of PKM1 and PKM2 in shRBX1-CAL62 cells. F Western blotting showing expression levels of PKM1 and PKM2 in shRBX1-CAL62 cells. G qRT-PCR showing expression levels of PKM1 and PKM2 in HA-RBX1 8305C cells. H Western blotting showing expression levels of PKM1 and PKM2 in HA-RBX1 8305C cells. I Compared to the tissue samples with low RBX1 expression, the PKM2/PKM1 ratio was significantly higher in those with high RBX1 expression. J, K Relative fluorescence intensities of mCherry and eGFP were quantified using ImageJ, and the fold change in the eGFP/mCherry ratio was calculated for the HA-RBX1, shRBX1, and control groups

To investigate the effect of RBX1 on modulating PKM alternative splicing, the expression of PKM isoform was identified in CAL62 cells after the depletion of shRNA-mediated RBX1 expression by quantitative western blotting and qRT-PCR. At the level of mRNA, RBX1 knockdown in CAL62 cells led to enhanced PKM1 and attenuated expression of PKM2 (Fig. 5E). An in-depth analysis of the expression of PKM isoforms at the protein level in cells with RBX1 depletion indicated an increase in PKM1 and a decrease in PKM2, which was in accordance with the RNA expression data (Fig. 5F). Furthermore, both western blotting and qRT-PCR demonstrated that the increase in RBX1 significantly enhanced the PKM2 expression and attenuated the PKM1 expression in 8305C cells (Fig. 5G, H). Next, we analyzed the expression of PKM1 and PKM2 isoforms in the tissue samples with low or high RBX1 expression. The PKM2/PKM1 ratio was significantly higher in the tissue samples with high expression of RBX1 when compared with those with low expression of RBX1 (Fig. 5I). To further demonstrate the effect of RBX1 on the modulation of PKM alternative splicing, we produced a dual reporter PKM minigene system. The eGFP/mCherry ratio with the overexpression of RBX1 increased 5.4-fold compared with that in the control. RBX1 knockdown led to a 3.2-fold decrease in the eGFP/mCherry ratio compared with that in the control, which elucidated its effect on modulating the PKM alternative splicing (Fig. 5J, K). Together, these studies confirmed that RBX1 regulates PKM alternative splicing to promote the PKM2-mediated Warburg effect in ATC cells.

RBX1 regulates PKM alternative splicing via SMAR1/HDAC6 complex degradation

We initially attempted to identify RBX1-interacting proteins in ATC cells using mass spectrometry. As shown in Additional file 5: Table S1, we identified several reported RBX1-interacting proteins, including TLE3, FoxA1, and SHP-1, as well as a previously unreported RBX1 interactor, namely, SMAR1. These experiments validated the direct interaction of HDAC6 with SMAR1 and the coexistence of SMAR1-HDAC6 as a complex. The generation of such a complex demonstrates the molecular dynamics between the proteins in modulating SMAR1-mediated PKM alternative splicing. Therefore, we hypothesized that RBX1 regulates PKM alternative splicing via the degradation of the SMAR1/HDAC6 complex. Co-IP analysis revealed an interaction between SMAR1 and RBX1 (Fig. 6A and Additional file 3: Fig. S3A-B). We observed that purified GST-RBX1 was bound to FLAG-tagged SMAR1 in vitro (Fig. 6B). Docking analysis suggested binding interactions between SMAR1 and RBX1 (Fig. 6C). These findings indicated that RBX1 binds with SMAR1 in ATC cells. Further analysis of PKM isoform expression in shSMAR1 cells at the protein level indicated reduced expression of PKM1 and upregulation of PKM2 (Additional file 4: Fig. S4). Moreover, as shown in Fig. 6D, RBX1 knockdown upregulated the level of SMAR1 protein in CAL62 cells, whereas RBX1 overexpression significantly reduced SMAR1 expression in 8305C cells. However, the expression of SMAR1 mRNA was not influenced by RBX1 alteration in the ATC cells (Fig. 6E, F), indicating that RBX1 regulated the expression of SMAR1 after it has been translated.

Fig. 6

RBX1 regulates PKM alternative splicing via SMAR1/HDAC6 complex degradation. A Co-IP showing direct binding of endogenous RBX1 and SMAR1 in CAL62 cells. B GST pull-down assay showing direct binding of endogenous RBX1 and SMAR1. C Top-ranked docking confirmation. 3D structures of SMAR1 and RBX1, with SMAR1 and RBX1 shown in green and cyan. D Western blotting was performed to detect the expression of RBX1 and SMAR1 in different groups. E, F qRT-PCR was performed to detect the expression of RBX1 and SMAR1 in different groups. G. CAL62 cells were transfected with shRBX1 plasmid. After that, cells were exposed to 20 μmol/L cyclohexanone (CHX) at the given times, and the SMAR1 degradation was identified using western blotting. H 8305C cells were treated with 10 μM MG132, while the RBX1 expression was altered. The SMAR1 protein expression level was determined by western blotting. I CAL62 cells were treated with 10 μM MG132 while transfecting them with HA-RBX1 or shRBX1 plasmids. Subsequently, the level of ubiquitin bound to the SMAR1 protein was measured by Co-IP. J Wild-type SMAR1 or K- to -R mutations in ATC cells (mutations in all the Lys positions of SMAR1 gene) for ubiquitination. K Determination of the type of SMAR1 ubiquitination in ATC cells. L Western blotting showing expression levels of RBX1, SMAR1, HDAC6, and PKM2 in shRBX1-CAL62 cells. M Co-IP combined with western blotting showing the expression levels of SMAR1 and HDAC6 in shRBX1-CAL62 cells. N Western blotting showing expression levels of RBX1, SMAR1, HDAC6, and PKM2 in HA-RBX1-8305C cells. O Co-IP combined with western blotting showing the expression levels of SMAR1 and HDAC6 in HA-RBX1-8305C cells

Previous studies have proven that UPS regulated the polyubiquitination and degradation of SMAR1 [19]. As the E3 ubiquitin ligase, RBX1, plays a role in protein degradation and recycling, we initially determined the degeneration of SMAR1 protein in RBX1-knockdown cells after inhibition of protein synthesis using cycloheximide (CHX). As shown in Fig. 6G, silencing RBX1 significantly inhibited SMAR1 degradation in ATC cells. Moreover, we observed that the levels of SMAR1 protein were restored in the cells with RBX1 overexpression after treatment with the proteasome inhibitor MG132 (Fig. 6H). Second, RBX1 ectopic expression led to an enhanced SMAR1 ubiquitination, while the RBX1 knockdown reduced SMAR1 polyubiquitination (Fig. 6I). Eventually, these findings revealed that the mutations in all the Lys positions of SMAR1 abolished the SMAR1 polyubiquitination induced by RBX1 in ATC cells (Fig. 6J). Furthermore, the mutation of Lys48 position on the ubiquitin nearly abolished the RBX1-mediated ubiquitination of SMAR1, while the mutation of K63R on ubiquitin did not exhibit any effect (Fig. 6K). These findings revealed that RBX1 is responsible for SMAR1 degradation via the ubiquitin–proteasome pathway in ATC.

Finally, we determined whether RBX1 influenced PKM alternative splicing via the destruction of the SMAR1/HDAC6 complex. The changes in the expression of HDAC6, SMAR1, PKM2, and PKM1 as well as the changes in the SMAR1/HDAC6 complex, were determined in CAL62 cells with RBX1 knockdown. It can be observed that RBX1 knockdown in CAL62 cells significantly increased the levels of SMAR1 expression, SMAR1/HDAC6 complex, and PKM1 and decreased the expression of PKM2. However, no variations were observed in the HDAC6 protein levels (Fig. 6L, M). In contrast, the RBX1 overexpression in the 8305C cells significantly attenuated the expression of SMAR1, the SMAR1/HDAC6 complex, and PKM1, and increased PKM2 expression (Fig. 6N, O). Together, these results indicated that ATC cell aerobic glycolysis and migration resulting from the modulation of RBX1-mediated PKM alternative splicing depends on the destruction of the SMAR1/HDAC6 complex.

Oncogenic effect of RBX1 is dependent on SMAR1 destabilization

We transfected the shSMAR1 plasmid into RBX1-knockdown ATC cells and measured its effects on biological function. RBX1 knockdown significantly increased SMAR1 protein expression, whereas the SMAR1 knockdown attenuated the increased expression of SMAR1 after RBX1 knockdown in ATC cells (Fig. 7A). Rescue tests indicated that the reduced expression of SMAR1 abrogated the RBX silencing-induced reduction in the metastatic capacity of ATC cells (Fig. 7B, C). A reduction in SMAR1 resulted in a rescue of the attenuated proliferation capacity of CAL62/shRBX1 cells (Fig. 7D, E). Knockdown of SMAR1 rescued reduction in glucose consumption, G6P, ATP levels, and lactate generation mediated by shRBX1 in ATC cells (Fig. 7F). Simultaneously, shRBX1 reduced ECAR in ATC cells, while SMAR1 knockdown decreased the reduction in glycolytic capacity and rate (Fig. 7G, H). In contrast, SMAR1 upregulation largely decreased the RBX1 increased cell invasion and migration (Fig. 7I, K), thereby decreasing the proliferation capacity of the 8305C/RBX1 group (Fig. 7L, M). Furthermore, SMAR1 upregulation rescued the increase in glucose consumption, G6P, ATP levels, and RBX1-mediated lactate generation in ATC cells (Fig. 7N). The overexpression of RBX1 enhanced ECAR in ATC cells, while the simultaneous increase in SMAR1 led to a decrease in the enhanced glycolytic capacity and rate (Fig. 7O, P). In conclusion, these results demonstrated that RBX1 facilitates the migration and aerobic glycolysis in ATC through SMAR1.

Fig. 7

RBX1 promotes ATC migration and aerobic glycolysis via SMAR1. A Western blotting was performed to determine the SMAR1 and RBX1 expression in various groups. B, C Quantification of transwell assay in various groups. *P < 0.05, **P < 0.01. D, E Quantification of EdU assays in different groups. *P < 0.05, **P < 0.01. F Cellular glucose consumption, G6P levels, ATP levels, and lactate generation in specific groups. *P < 0.05, **P < 0.01. G, H Measurement of OCR and ECAR in specific groups. *P < 0.05. I SMAR1 and RBX1 expression levels in various groups were determined using western blotting. J, K Quantification of transwell assays in different groups. *P < 0.05, **P < 0.01. L, M Quantification of EdU assays in different groups. *P < 0.05, **P < 0.01. N Cellular glucose consumption, G6P levels, ATP levels, and lactate generation in specific groups. *P < 0.05, **P < 0.01. O, P ECAR and OCR were measured in the indicated groups. *P < 0.05, **P < 0.01