MAGI3 is low-expressed in ccRCC, and correlates with poor prognosis of ccRCC. Metastasis is responsible for more than 90% of cancer-associated deaths in RCC patients[7]. To screen key genes and pathways associated with progression and metastasis in ccRCC, we compared the gene expression profiles of ccRCC tissues with the adjacent normal tissues (GSE12606), and metastatic ccRCC tissues with non-metastatic ccRCC samples (GSE31232) respectively. GEO2R analysis found 1827 differential expression genes (DEGs) in ccRCC and 40 DEGs in mRCC (Fig S1A), with five common DEGs identified, including MAGI3, EGF, ADH1C, EYA4 and C3orf52 (Fig S1B). These genes were further verified in TCGA-KIRC dataset, four genes including EGF, ADH1C, EYA4 and C3orf52 were significantly reduced in ccRCC tissues (Fig. S2A), however, none of them had statistically different between metastatic and non-metastatic ccRCC samples (Fig. S2B).
MAGI3 was also investigated with TCGA and another independent cohort, and it was shown that MAGI3 expression was robustly downregulated in ccRCC tissues at both mRNA (Fig. 1A) and protein levels (Fig. 1B, 1C). Compared with non-metastatic ccRCC samples, MAGI3 expression was further significantly reduced in metastatic ccRCC samples at either mRNA (Fig. 1D) or protein levels (Fig. 1E) in TCGA-KIRC datasets or clinical specimen respectively. These results indicated that MAGI3 was a novel gene identified to associate with metastasis of ccRCC. Taken together, all of these data indicated that MAGI3 was low-expressed in ccRCC, and correlated with metastasis of RCC.
In order to explore the clinical relevance of MAGI3 in ccRCC, the correlations of MAGI3 level with ccRCC stage, grade and prognosis were evaluated with several independent ccRCC patient cohorts of TCGA-KIRC or GEO. The results showed that low level of MAGI3 was correlated with advanced stage (Fig. 1F), high grade (Fig. 1G) and poor prognosis (Fig. 1H). Cox univariate and multivariate analysis indicated that MAGI3 expression level was an independent prognostic factor for OS and PFS of ccRCC patients (Table 1). These results suggested that MAGI3 might be employed as a biomarker in predicting prognosis of ccRCC.
MAGI3 inhibits migration, invasion and metastasis of ccRCC. To further verify the correlation of ccRCC metastasis with MAGI3 expression level, Gene Set Enrichment Analysis (GSEA) from TCGA-KIRC or GSE36895 dataset was conducted, the results showed that the gene signatures of metastasis-related upregulated gene-set were enriched in ccRCC specimens with MAGI3 low expression group (Fig. 2A). The metastatic process is a result of tumor cell migration and invasion. To investigate the biological impact of MAGI3 on metastasis of ccRCC, migration and invasion were examined by transwell migration and invasion assays as well as wound healing assays. Overexpression or knockdown of MAGI3 in 769-P or 786-O cells was confirmed by western blotting (Fig. 2B). Ectopic expression of MAGI3 significantly suppressed migration and invasion of 786-O and 769-P cells (Fig. 2C&E), while knockdown of MAGI3 markedly promoted migration and invasion in the both ccRCC cells (Fig. 2D&F). To verify that MAGI3 inhibits ccRCC cell metastasis in vivo, a xenograft tumor model was established by injection of 786-O/shMAGI3 or 786-O/shCtrl cells into the tail vein of NOD/SCID mice. A higher ratio of lung metastasis was observed in the 786-O/shMAGI3 group as compared with the control group, indicating that knock down of MAGI3 expression significantly enhanced 786-O cell metastasis in vivo (Fig. 2G). Taken together, these data indicated that MAGI3 could inhibit the migration, invasion, and metastasis of ccRCC cells.
MAGI3 promotes β-catenin protein phosphorylation and ubiquitin-proteasome degradation. To gain insight into the molecular mechanism of MAGI3 expression on control cell functions, we used KIRC database in TCGA to screen the gene set closely related to MAGI3 expression in ccRCC through UALCAN website, and then conducted the pathway enrichment analysis for MAGI3-related gene set by DAVID. The results indicated that the function of MAGI3 related genes was mainly enriched in "ubiquitination dependent protein degradation" and "Wnt signaling pathway"(Fig. S3), reminding MAGI3 expression may influence Wnt signaling pathway activation. To verify the effects of MAGI3 expression on Wnt signaling pathway, the GSEA was performed. The results showed that the gene signatures of Wnt/β-catenin activation were enriched in patients with MAGI3 lower levels in TCGA KIRC and GSE36895 datasets (Fig. 3Aa), which is consistent with our previously report that MAGI3 inhibited malignant phenotypes of glioma and cervical cancer through negatively regulating Wnt/β-catenin signaling[16, 17]. However, the underlying molecular mechanisms was not yet elucidated. Meanwhile, we noticed that there were no correlations between MAGI3 and β-catenin at mRNA level (Fig. 3Ab), but MAGI3 significantly negatively correlated with β-catenin at protein level in clinical specimens (Fig. 3Ac). Moreover, overexpression MAGI3 could significantly reduce the levels of β-catenin protein in ccRCC cells (Fig. 3B). On the contrary, knockdown of MAGI3 increased the levels of β-catenin protein (Fig. 3C). It is well known that β-catenin is a short half-life protein and rapidly turned over by proteasomal degradation[14]. To explore whether MAGI3 regulates β-catenin stability, we next determined the half-life of β-catenin protein by overexpression MAGI3. As shown in Fig. 3D, Overexpression of MAGI3 significantly accelerated degradation of the wild type β-catenin, but not β-catenin-T779A mutant (Fig. 3D) as T779A mutant of β-catenin can abolish their interaction[16], suggesting that MAGI3 decreased β-catenin stability dependent on the interaction between MAGI3 and β-catenin. Studies have shown that β-catenin is mainly degraded by ubiquitin-proteasome pathway[18]. Therefore, we next investigated whether MAGI3 promoted β-catenin degradation by enhancing β-catenin ubiquitination. Also, the results showed that MAGI3 indeed significantly promoted ubiquitination of the wild type β-catenin (Fig. 3E) but not β-catenin-ΔCT (β-catenin deleted C-terminus) (Fig. 3F). Taken together, these results indicated that MAGI3 promoted β-catenin ubiquitination and degradation via interaction with β-catenin.
It is well known that phosphorylation of β-catenin at Ser33/Ser37/Thr41 by glycogen synthase kinase 3β (GSK3β) targets β-catenin for ubiquitination and degradation[14]. Therefore, we next examined whether MAGI3 promoted the phosphorylation of β-catenin, and our results showed that MAGI3 overexpression increased the levels of phospho-β-catenin (Ser33/Ser37/Thr41) (Fig. 4A), while MAGI3 knockdown reduced the phospho-β-catenin (Ser33/Ser37/Thr41) in ccRCC cells (Fig. 4B). For the regulatory mechanism of MAGI3 on β-catenin phosphorylation, we proposed that MAGI3 acts as a scaffold protein to facilitate the phosphorylation of β-catenin by GSK3β. This hypothesis was further confirmed by the results of Fig. 4C&D that MAGI3 enhanced the association of GSK3β with wild type β-catenin, but not with β-catenin-ΔCT, indicating that MAGI3 promotes β-catenin phosphorylation by facilitating the association of GSK3β with β-catenin, and the C-terminus of β-catenin plays an indispensable role during this process. β-catenin protein composed three domains, including N-tail, armadillo repeat domain (ARM) and C-tail (Fig. 5A). We found that the N-tail could associate with ARM domain of β-catenin, β-catenin-ΔNT (β-catenin deleted N-tail) and β-catenin-wt. Moreover, the N-tail had a stronger binding affinity with β-catenin-ΔNT (β-catenin deleted N-tail) than β-catenin-wt (Fig. 5B lane 2&3), indicating that the intramolecular N-tail of β-catenin was able to associate with ARM domain, thereby disturb the binding between extra molecular N-tail with β-catenin-wt. Also, we found that the N-tail had a stronger binding affinity with β-catenin-ΔNT than the ARM (Fig. 5B lane 3&4), indicating that the C-tail facilitated the interaction of the N-tail with its Arm domain. Meanwhile, overexpression MAGI3 could reduce extra molecular Flag-N-tail binding with GST-ΔNT (Fig. 5C). However, when C-terminus was deleted from β-catenin-ΔNT, MAGI3 had no effect on the interaction of extra molecular Flag-N-tail binding with GST-ΔNT-ΔCT (Fig. 5D), indicating that the C-tail of β-catenin associated with MAGI3 retarded the interaction of the intramolecular N-tail with its Arm domain of β-catenin, and the free C-tail facilitated the interaction of the N-tail with its Arm domain of β-catenin to form an intramolecular conformation, such as a loop structure. Therefore, when MAGI3 bound to the C-tail, the intramolecular N-tail could be released from the ARM domain to reinforced the association between β-catenin and GSK3β, and promoted GSK3β-mediated β-catenin phosphorylation (Fig. 5E).
MAGI3 suppresses metastasis and overcomes rapalogs resistance of ccRCC cells via inhibition of β-catenin signaling. We further investigate whether MAGI3 inhibits invasion and migration of ccRCC cells through suppressing Wnt/β-catenin pathway. The results showed that knockdown of MAGI3 increased β-catenin protein level (Fig. 6A, lane 2), and promoted invasion and migration of ccRCC cells (Fig. 6B). To rescue the level of β-catenin protein with knockdown β-catenin (Fig. 6A, lane 4), the invasion and migration induced by MAGI3 knockdown of ccRCC cells were inhibited (Fig. 6B). Consistent with these results, when treated with Wnt/β-catenin inhibitor IWR-1-endo, the invasion and migration induced by MAGI3 knockdown of ccRCC cells were also inhibited (Fig. 6C). Taken together, these data suggest that MAGI3 inhibits migration and invasion in ccRCC cells through suppressing the Wnt/β-catenin pathway.
Evidences suggest that both Wnt/β-catenin and PI3K/Akt/mTOR pathways contribute to progression and poor prognosis of ccRCC, and inhibition of Wnt/β-catenin and PI3K/Akt/mTOR pathways is regarded as a promising innovative therapeutic approach for cancer treatment[19]. Surprisingly, the potential therapeutic effect of simultaneous targeting of PI3K/Akt/mTOR and Wnt/β-catenin pathways has never been explored in RCC preclinical models. Therefore, we next to investigate the effects of MAGI3 expression on the sensitivity of ccRCC cells to rapalogs (Everolimus). As shown in Fig. 7, the assay of cell viability in 786-O or 769-P cells showed that overexpression of MAGI3 to suppress Wnt/β-catenin signaling could significantly increase the sensitivity of ccRCC cells to Everolimus, while knockdown of MAGI3 to activate Wnt/β-catenin signaling could robustly decreased the sensitivity of ccRCC cells to Everolimus (Fig. 7A). Meanwhile, when activated Wnt/β-catenin signaling by CHIR, the drug sensitivity-enhanced by MAGI3 over expression was totally abolished (Fig. 7Ba, right panel); while blocked Wnt/β-catenin signaling by XAV-939, an inhibitor of Wnt/β-catenin pathway, the drug sensitivity-reduced by knockdown of MAGI3 was also rescued (Fig. 7Bb, right panel). These results demonstrated that MAGI3 conferred ccRCC cell sensitivity to Everolimus through Wnt/β-catenin pathway. Moreover, the sensitivity of ccRCC cells to Everolimus was dose-dependently increased when treated with XAV-939 (Fig. 7C). Furthermore, simultaneous targeting of PI3K/Akt/mTOR and Wnt/β-catenin pathways by using Wnt inhibitor XAV-939, and rapalogs such as Everolimus could synergistically inhibited clone formation (Fig. 7D), and the cell migration (Fig. 7E) of ccRCC respectively. Altogether, these data indicate that inhibition of Wnt/β-catenin pathway by with Wnt inhibitor or MAGI3 overexpression could synergistically potentiates the sensitivity of ccRCC cells to rapalogs.
In summary, this study provide first lines of evidences that Low MAGI3 expression promotes Wnt/β-catenin signaling activation to mediate metastasis and rapalogs resistance of clear cell renal cell carcinoma; while high MAGI3 expression suppresses Wnt/β-catenin signaling via facilitating GSK3β-mediated β-catenin phosphorylation and subsequently ubiquitination and degradation, leading to inhibit ccRCC metastasis. MAGI3 overexpression or application of Wnt inhibitor synergistically potentiates the sensitivity of ccRCC cells to rapalogs (Fig. 7F).