FBXL6 is an independent risk factor for aggressive HCC and drives HCC lung metastasis in vivo significantly more strongly than Kras mutation, p53 loss or Tsc1 loss
A previous study indicated that FBXL6 was highly expressed in HCC tissues, but whether FBXL6 protein expression is an independent risk factor in metastatic HCC remains unclear. Here, we first examined the expression and localization of FBXL6. Our IHC staining of liver tissues from patients demonstrated that FBXL6 was expressed mainly in HCC cells (Fig. 1a). To extend this work, we analyzed the relationship between FBXL6 expression and clinical prognosis in HCC patients. IHC staining of 108 paired HCC samples (tumor tissues and adjacent normal tissues from the same patients) revealed that FBXL6 was overexpressed in 60.2% (65/108) of the HCC tissues compared to the paired adjacent tissues and was positively associated with advanced TNM stage, vascular thrombosis, and metastasis (Fig. 1b, Supplementary Table 1). In line with this finding, FBXL6 exhibited high expression in 94% of HCC tissues compared to the matched adjacent normal tissues in The Cancer Genome Atlas (TCGA) datasets (Supplementary Fig. 1a). Furthermore, we investigated the mutation profiles for the FBXL6-overexpressing HCC cases and found frequent TP53 or CTNNB1 mutation in this subset of HCC patients (Supplementary Fig. 1b). Univariate and multivariate analyses showed that FBXL6 was strongly associated with overall survival (OS) in HCC patients, supporting the idea that FBXL6 expression is an independent risk factor in HCC (Fig. 1c, Supplementary Table 2). Thus, these results indicate that elevated FBXL6 expression is positively associated with metastasis in HCC patients.
To address the role of FBXL6 in HCC metastasis in vivo, we generated Fbxl6LSL-fl/+ mice by inserting the CAG promoter-LoxP-STOP-LoxP-mouse Fbxl6 cDNA cassette into the ROSA26 locus in C57BL/6 N mice by CRISPR/Cas9-mediated genome editing. Fbxl6;Alb-Cre mice were obtained by crossing Fbxl6LSL-fl/+ mice with Alb-Cre mice (Supplementary Fig. 2a), and their Alb-Cre (wild-type) littermates were monitored for tumorigenesis and metastasis for 310 days, at which time they were imaged. The mouse genotype was validated by PCR (Supplementary Fig. 2b). We found that elevated expression of FBXL6 in hepatocytes was very strongly related to hepatocarcinogenesis and lung metastasis. The promotive effects on metastasis in these mice were much stronger than those in mice with Kras mutation (KrasLSL-G12D/+;Alb-Cre), p53 haploinsufficiency (p53+/–) or Tsc1 insufficiency (Tsc1fl/fl;Alb-Cre), as indicated by the tumor number, largest tumor size, liver/body weight ratio, and rate of lung metastasis (Fig. 1d–h). Furthermore, there were no significant differences between female and male mice in the number of FBXL6-induced tumors, largest tumor size and liver/body weight ratio, suggesting that the oncogenic activity of FBXL6 is independent of sex (Supplementary Fig. 2c). H&E staining showed that in Fbxl6;Alb-Cre mice that developed spontaneous liver tumors, the tumors formed within blood vessels, while in Alb-Cre mice, no such liver tumors formed (Supplementary Fig. 3a). The presence of abnormal cells around the vessels was significantly increased in Fbxl6;Alb-Cre mice compared with Alb-Cre mice (Supplementary Fig. 3b). Importantly, FBXL6 mice with elevated FBXL6 expression developed spontaneous liver tumors and lung metastasis (Fig. 1i, j), as indicated by the number of metastases in FBXL6-overexpressing mice compared to wild-type mice (Fig. 1k). In line with this finding, the mRNA expression of metastasis markers (Icam1, Vcam1, Upa, Ccl2, and Mmp9)30 was elevated in the liver tumors of mice with high Fbxl6 expression compared with the liver tissues of wild-type mice (Fig. 1l). H&E staining of lung metastases and IHC staining of the hepatocyte marker Lipase C (LIPC) further validated the FBXL6-driven lung metastasis of HCC (Fig. 1m). In line with these findings, FBXL6 promoted HCC cell migration and upregulation of metastasis markers (Icam1, Vcam1, Upa, Ccl2, and Mmp9) (Supplementary Fig. 3c, d), while silencing FBXL6 blocked these effects (Supplementary Fig. 3e–g). Taken together, these findings indicate that via elevated expression, FBXL6 acts as a superoncogene in driving HCC development in vivo and in vitro.
Characterization of the proteome and ubiquitinome in response to FBXL6 overexpression
To systematically identify the substrates of FBXL6, we carried out mass spectrometry-based label-free quantitative proteomics and ubiquitomics on FBXL6-high HCC tumor tissues and adjacent tissues from Fbxl6;Alb-Cre mice (Supplementary Fig. 4a). Through these analyses, we identified 4384 proteins and 3198 quantifiable proteins. We defined candidate proteins with significant differential expression (p < 0.05 by Student’s t test) as those with a fold change of at least 1.2 between FBXL6-induced HCC tumors and adjacent tissues. With this criterion, 851 downregulated proteins and 1123 upregulated proteins in HCC tumors with high FBXL6 expression were identified (Supplementary Fig. 4b).
FBXL6 overexpression caused a global increase in protein ubiquitination after normalization to the total protein content (Supplementary Fig. 4c). In total, 1710 ubiquitinated sites were quantified for the ubiquitinomics analysis. Based on a criterion of a ≥1.5-fold increase in ubiquitination in FBXL6-overexpressing HCC tumors compared to adjacent tissues, a total of 500 significantly upregulated ubiquitin sites in 296 proteins were identified in FBXL6-overexpressing HCC tumors. A summary of the results of these analyses is shown in Supplementary Fig. 4c.
FBXL6 triggers K63-linked TKT ubiquitination at Lys16(K16) and Lys319 (K319), leading to TKT activation and HCC metastasis
To further screen the candidate FBXL6 substrates, we reanalyzed protein expression and ubiquitination in HCC tumors and adjacent tissues using proteomics and ubiquitomics analyses, respectively. A total of 231 proteins were ubiquitinated in HCC tumors but not in the adjacent tissues (a cutoff of a 1.5-fold change in the ubiquitination level). A total of 1974 proteins showed significant differential expression (a cutoff of a ≥1.2-fold change in the protein level) between HCC and adjacent tissues. Furthermore, 180 proteins were identified as candidate interacting proteins with FBXL6 in a pulldown assay using an anti-FBXL6 antibody in HCC cells. We made a chart showing the overlapping proteins to identify potential FBXL6 substrates and identified 2 overlapping proteins, HNRNPF (heterogeneous nuclear ribonucleoprotein F) and TKT (transketolase). Both of these proteins exhibited high levels of ubiquitination and changes in protein expression (Fig. 2a), suggesting that they might be substrates of FBXL6. TKT, a key enzyme in the nonoxidative branch of the pentose phosphate pathway (PPP), was upregulated 2.262-fold in HCC tissues compared to adjacent normal tissues (Fig. 2a)2,22, whereas HNRNPF was upregulated only 1.363-fold in HCC tissues compared to adjacent normal tissues. Further, compared to knockdown of HNRNPF, knockdown of TKT dramatically inhibited cell proliferation and migration (Supplementary Fig. 5a–c), suggesting that TKT but not HNRNPF may play a critical role in FBXL6-driven HCC cell migration. To test the interaction between FBXL6 and TKT, we performed a co-IP experiment and found that FBXL6 was able to interact with TKT, whereas the previously reported potential substrates of FBXL6 (HSP90, CCNA2, and VDAC2) had low or no binding affinity compared with that of TKT in HCC cells (Fig. 2b, Supplementary Fig. 6a, b). These findings suggest that TKT binds to FBXL6. To further confirm whether FBXL6 ubiquitinates TKT, we cotransfected HA-TKT and his-ubiquitin with Flag-FBXL6 or Flag-FBXL6ΔF (a mutant with deletion of the F-box domain) into 293 T cells and then treated the cells with MG132 for 3 h before performing an in vivo ubiquitination assay. Overexpression of FBXL6 increased the level of ubiquitinated TKT, whereas no such effect was observed after overexpression of the F-box deletion mutant of FBXL6 (FBXL6ΔF) (Fig. 2c). To strengthen these findings, we performed domain analyses to further characterize the FBXL6-TKT interaction and TKT ubiquitination. We reconstructed FBXL6 mutants with deletion of the leucine-rich repeat (LRR) domain: FBXL6(ΔLRR219-244), FBXL6(ΔLRR383-408) and FBXL6(ΔLRR497-528) (Supplementary Fig. 6c). The results of the co-IP experiment indicated that deletion of LRR497-528 (ΔLRR497-528) in FBXL6 significantly attenuated the interaction between TKT and FBXL6 compared with that in cells expressing wild-type FBXL6 or the other deletion mutants (ΔLRR383-408 and ΔLRR219-244), suggesting that LRR497-528 in FBXL6 is responsible for the FBXL6-TKT interaction (Supplementary Fig. 6d). As expected, compared with wild-type FBXL6 and the other deletion mutants, FBXL6(ΔLRR497-528) lost the ability to promote TKT polyubiquitination, a finding that further indicates that FBXL6 ubiquitinates TKT (Supplementary Fig. 6e). SCF (Skp1-Cullin-F box) family members usually catalyze Lys63-linked polyubiquitination of a target protein to induce its activation31, whereas Lys48- or Lys11-linked polyubiquitination induces proteasomal degradation of the target protein32,33. We thus generated three ubiquitin mutants, ubiquitin-K63R, ubiquitin-K48R, and ubiquitin-K11R (in which K63, K48, or K11 was substituted with arginine), to determine the type of ubiquitin linkage involved in FBXL6-mediated TKT ubiquitination. Mutation of K63 (K63R) in ubiquitin nearly completely abolished FBXL6-mediated ubiquitination of TKT, whereas the K48R and K11R mutations in ubiquitin, like wild-type ubiquitin, significantly promoted TKT polyubiquitination, indicating that FBXL6 catalyzes K63-linked polyubiquitination of TKT (Fig. 2d).
Whether posttranslational regulation of TKT is associated with its activity is unknown. Our ubiquitomics analysis indicated that TKT ubiquitination at Lys16 (K16) and Lys319 (K319) was significantly increased in HCC tumors (Fig. 2e). We further examined the TKT protein sequence and found that these two lysine sites were evolutionarily conserved (Fig. 2f). To confirm that K319 and K16 in TKT are ubiquitinated by FBXL6, we generated two mutants, namely, K319A and K16A, to disrupt the ubiquitination of TKT at K319 and K16, respectively. Interestingly, the ubiquitination-disrupting K319A and K16A mutants exhibited a significant loss of ability to be ubiquitinated by FBXL6, suggesting that FBXL6 promotes K63-linked TKT ubiquitination at K319 and K16 (Fig. 2g). To confirm whether these two lysine residues in TKT are related to its activation, we analyzed TKT activity using a TKT-specific activity assay kit. Wild-type TKT had high activity, whereas the ubiquitination-disrupting mutants (K319A and K16A) exhibited a complete loss of activity in both human and mouse hepatocellular carcinoma cells (Fig. 2h, i). Consistent with these findings, both the TKT-K319A and TKT-K16A mutants significantly attenuated NADPH production compared with wild-type TKT, as indicated by the NADPH/NADP+ ratio (Supplementary Fig. 7a). These results demonstrated that the K319 and K16 residues are associated with TKT activation. Previous studies reported that TKT promotes cancer cell proliferation and migration. As expected, the ubiquitination-disrupting mutants (K319A and K16A) lost the ability to promote the proliferation and migration of both human and mouse hepatocellular carcinoma cells (Fig. 2j–l). To determine whether TKT-mediated cell proliferation and migration depend on FBXL6, we cotransfected FBXL6 with TKT or a TKT mutant (K319A or K16A) into Huh7 and Hepa1-6 cells. FBXL6 significantly upregulated wild-type TKT activity and promoted TKT-mediated cell proliferation and migration but did not affect the TKT mutants (K319A and K16A) (Supplementary Fig. 7b–e), suggesting that FBXL6 activates TKT via K63-linked polyubiquitination at K319 and K16. Furthermore, wild-type TKT was localized mainly in the cytoplasm, whereas its mutants (K319A and K16A) accumulated in the nucleus (Supplementary Fig. 8a, b), suggesting that ubiquitinated TKT localizes mainly in the cytoplasm.
To further study the effect of TKT on FBXL6-triggered HCC migration, we reduced TKT expression in FBXL6;Alb-Cre and Alb-Cre primary hepatocytes and FBXL6-overexpressing HCC cells using TKT siRNAs. FBXL6 promoted HCC cell proliferation and migration, and knockdown of TKT dramatically attenuated these effects (Supplementary Fig. 9a–c). Furthermore, TKT-mediated HCC cell proliferation and migration were dependent on FBXL6, as indicated by the finding that knockdown of FBXL6 attenuated the oncogenic effects of TKT (Supplementary Fig. 9d, e). Collectively, these results indicate that FBXL6 triggers the K63-linked polyubiquitination, cytoplasmic localization, and activation of TKT, leading to HCC metastasis.
TKT phosphorylation at Thr287 is critical for the activation and cytoplasmic localization of TKT and the interaction of TKT with the E3 FBXL6
FBXL6 is an F-box SCF-E3 that forms a functional complex with Cullin-1, Skp1, and Rbx1 proteins. F-box proteins usually specifically recognize and bind potential substrates in a manner dependent on phosphorylation34. Therefore, it is reasonable to hypothesize that the TKT protein needs to be phosphorylated before it can be recognized by FBXL6. To investigate this possibility, the changes in phosphorylated TKT in FBXL6-overexpressing tumors were analyzed using a phosphoproteomics approach. An increase in TKT phosphorylation at Thr287 was observed in liver tumors (Supplementary Fig. 10a). Previous literature indicated that AKT-mediated TKT phosphorylation at Thr382 is associated with its activation in mice fed a lysine-deficient diet35, whereas we found that only Thr287 in TKT was phosphorylated in HCC tumors with high FBXL6 expression, indicating that TKT phosphorylation at Thr287 may play a critical role in FBXL6-mediated TKT ubiquitination. Furthermore, this site was highly evolutionarily conserved in various species (Fig. 3a). To investigate whether this site is responsible for the FBXL6-TKT interaction, we cotransfected Huh7 cells with a TKT mutant (TKT-Thr287A) and FBXL6 and performed a co-IP experiment. Interestingly, FBXL6 pulled down wild-type TKT, but this interaction was completely disrupted by Thr287 mutation in TKT, suggesting that phosphorylation of TKT at Thr287 is critical for the interaction between FBXL6 and TKT (Fig. 3b).
To determine whether Thr287 phosphorylation is associated with the cellular localization of TKT, we detected TKT by immunofluorescence staining and found that the TKT T287A mutant was localized mainly in the nucleus, whereas wild-type TKT was localized in the cytoplasm (Fig. 3c–e). To determine whether Thr287 phosphorylation is responsible for TKT activation, we analyzed the activity of the TKT T287A mutant using the TKT activity assay kit. The Thr287 mutant exhibited reduced TKT activity compared with wild-type TKT (Fig. 3c, f). In line with this finding, TKT T287A blocked cell proliferation and migration compared with wild-type TKT (Fig. 3g–i). Taken together, these findings indicate that Thr287 is the critical amino acid for the interaction between TKT and FBXL6 and for TKT activation.
VRK2 is important for TKT phosphorylation at Thr287 and facilitates FBXL6-mediated K63-linked ubiquitination of TKT and TKT activation
We next examined whether Thr287 is involved in FBXL6-mediated TKT ubiquitination. The results of the in vitro ubiquitination assay showed that FBXL6 promoted TKT ubiquitination, and this effect was attenuated by the TKT T287A mutation (Fig. 4a), suggesting that Thr287 plays an important role in FBXL6-mediated TKT polyubiquitination. Subsequently, we investigated the potential kinases involved in Thr287 TKT phosphorylation using GPS 2.0 software36. VRK2 and JNK2 were identified as the top two candidates that may phosphorylate TKT at T287 (Supplementary Fig. 10b). The in vivo ubiquitination assay results showed that inhibition or knockdown of VRK2 but not JNK2 significantly attenuated FBXL6-mediated TKT polyubiquitination (Fig. 4b). This effect was similar to that of the TKT T287A mutation (Fig. 4b). To further confirm whether TKT is phosphorylated at T287 by VRK2, we analyzed the effect of VRK2 inhibition or knockdown on TKT phosphorylation by Western blotting. We found that inhibition or knockdown of VRK2 significantly blocked TKT phosphorylation at T287 (Fig. 4c). Conversely, overexpression of VRK2 promoted phosphorylation of TKT at Thr287 (Fig. 4d). These findings indicated that TKT Thr287 is a VRK2 phosphorylation site. Inhibition of VRK2 or overexpression of the TKT mutant (T287A) dramatically blocked FBXL6-mediated K63-linked TKT polyubiquitination (Fig. 4e). We next examined the role of VRK2 in TKT-mediated cell proliferation and migration. Inhibition of VRK2 significantly attenuated TKT-mediated cell proliferation and migration (Fig. 4f, g). Taken together, these data demonstrate that VRK2 is involved in the phosphorylation of TKT at T287, which is required for FBXL6-mediated TKT K63-linked polyubiquitination and activation.
FBXL6 upregulates PD-L1 and VRK2 expression via a TKT-dependent decrease in ROS accumulation and mTOR activation, leading to immune evasion and HCC metastasis
The immune microenvironment plays an important role in tumorigenesis and metastasis. Here, we found that PD-L1 was highly expressed in FBXL6-high tumors compared with normal tissues (Fig. 5a), suggesting that PD-L1 may be a downstream target of the FBXL6-TKT axis. To validate this hypothesis, we cotransfected FBXL6 with a siRNA targeting TKT in Huh7 and Hepa1-6 cells and found that FBXL6 upregulated PD-L1 expression and that knockdown of TKT blocked this effect (Fig. 5b). A previous study reported that TKT reduced intercellular ROS accumulation. Here, we found that the TKT-mediated decrease in ROS accumulation was significantly blocked by knockdown of FBXL6 (Supplementary Fig. 11a–c). This finding further confirms that TKT activity depends on FBXL6. Intercellular ROS scavenging can activate the mTOR pathway37,38. We found that treatment with the ROS inhibitor VAS2870 significantly increased the protein levels of PD-L1 and p-mTOR (whose downstream targets are p-S6K and p-4EBP1) and attenuated the TKT knockdown-mediated decreases in cell proliferation and migration (Fig. 5c, Supplementary Fig. 11d, e). Next, we sought to determine whether there is a positive correlation between mTOR activation and PD-L1 expression. Inhibition of mTOR with its inhibitors (rapamycin and sapanisertib) significantly reduced PD-L1 expression and TKT-mediated HCC cell proliferation and migration (Fig. 5d–g, Supplementary Fig. 11f). Taken together, these findings indicate that the FBXL6-TKT-ROS-mTOR axis positively regulates PD-L1 expression.
High expression of VRK2 was observed in FBXL6-overexpressing liver tumors (Fig. 5a). VRK2 is often highly expressed in various types of cancers and is involved in cancer pathogenesis, but the underlying mechanisms remain unclear. We sought to determine whether the FBXL6-TKT-mediated ROS-mTOR pathway regulates VRK2 expression. To this end, Huh7 and Hepa1-6 hepatocellular cancer cells were cotransfected with FBXL6 with or without a siRNA targeting TKT. FBXL6 transfection upregulated VRK2 expression, and silencing TKT attenuated VRK2 expression (Fig. 5b). Furthermore, treatment with the ROS inhibitor VAS2870 upregulated VRK2 expression, and treatment with the mTOR inhibitors significantly reduced VRK2 expression (Fig. 5c–e). In line with these findings, mTOR and VRK2 signaling was upregulated in FBXL6 tumors compared with normal tissues (Fig. 5h). Inhibition or knockdown of VRK2 significantly blocked the FBXL6-mediated cell proliferation and migration of HCC cells (Fig. 5i, j, Supplementary Fig. 11g, h). Taken together, these findings demonstrate that VRK2 is a downstream target of the FBXL6-TKT-ROS-mTOR axis and facilitates FBXL6-mediated HCC tumorigenesis and metastasis.
Targeting or interfering with TKT suppresses HCC tumorigenesis and lung metastasis triggered by FBXL6 in vivo
To study whether inhibition of TKT blocks FBXL6-triggered HCC lung metastasis, FBXL6 primary cells were cultured, transplanted into the hypodermis of nude mice, and then retransplanted into the livers of 16 male C57BL/6 N mice. The process for establishing this orthotopic HCC model in mice and the pharmacological treatment plan are shown in Fig. 6a. Compared to vehicle treatment, treatment with N3PT (25 mg/kg, i.p., every other day), a TKT inhibitor39, significantly reduced the tumor weight and tumor volume (Fig. 6b, c). Consistent with this finding, knockdown of TKT significantly attenuated tumor growth in Fbxl6;Alb-Cre mice, as characterized by the decreased tumor number and tumor size (Supplementary Fig. 12a-c). Importantly, N3PT treatment significantly blocked HCC lung metastasis, as indicated by the decreased rate of lung metastasis in the N3PT group compared with the control group (Fig. 6d, e). To confirm the distant lung metastasis of HCC, H&E and IHC analyses of metastatic nodules were performed. The number of lung metastatic nodules labeled with the hepatocyte marker lipase C was significantly increased in the control group compared with the N3PT treatment group (Fig. 6f). In line with these findings, knockdown of TKT significantly reduced HCC lung metastasis in Fbxl6; Alb-Cre mice, as indicated by the decreases in the lung metastasis rate and lipase C expression in metastatic nodules (Supplementary Fig. 12d, e). Notably, PD-L1 expression was significantly reduced after N3PT treatment, indicating that inhibition of TKT attenuates immune evasion (Fig. 6f, g). Consistent with this finding, the expression of metastasis markers (Icam1, Vacm1, Upa, Ccl2, and Mmp9) was significantly reduced in the N3PT treatment group compared to the control group (Fig. 6g). To extend this work, the effect of N3PT on ROS accumulation in these two groups was examined. N3PT significantly attenuated the FBXL6-mediated decrease in ROS accumulation in liver tumors (Fig. 6h). Furthermore, N3PT reduced the expression of the proliferation biomarkers Ki67 and CCNB2 and the metastasis biomarker MMP9 in liver tissues (Supplementary Fig. 13a). To further confirm the efficacy of N3PT in a mouse model, we tested the effect of the TKT inhibitors N3PT and oxythiamine (OT) on HCC cells in vitro. N3PT and OT significantly blocked the FBXL6-mediated proliferation of HCC cells (Huh7 and Hepa1-6) (Supplementary Fig. 13b). Taken together, these results indicate that inhibition of TKT constitutes a potential therapeutic strategy for HCC tumorigenesis and lung metastasis driven by abnormally high FBXL6 expression.
The activated TKT protein level correlates positively with the protein levels of FBXL6 and VRK2 and predicts unfavorable survival in HCC patients
Based on our finding that VRK2 phosphorylation promotes TKT activation in mice, we investigated the expression level of VRK2 in HCC patients. VRK2 was highly expressed in HCC tumors (Fig. 7a). Increased VRK2 expression levels were positively associated with advanced tumor stage, recurrence, vascular thrombosis, and metastasis (Fig. 7b, c; Supplementary Table 3). In line with this finding, a higher VRK2 expression level correlated with a shorter overall survival (OS) time in 121 HCC patients (Fig. 7d; hazard ratio = 0.3634, p < 0.001) and was identified as an independent risk factor in HCC patients (Supplementary Table 4). Next, we tested the possible correlation between the activated TKT (pThr287-TKT antibody staining) and VRK2 protein levels. The level of activated TKT (p-Thr287) correlated positively with the VRK2 protein expression level (Fig. 7a, e; p < 0.0001). Furthermore, activated TKT (p-Thr287) was negatively associated with overall survival in HCC and correlated positively with TNM stage, recurrence, and metastasis (Fig. 7f–h, Supplementary Table 5). These data suggest that VRK2 and activated TKT (p-Thr287) may accurately predict clinical outcomes.
We next evaluated the association between activated cytoplasmic TKT and FBXL6 in HCC using patient tissues. First, we used IHC staining to evaluate TKT expression in HCC tissues. The level of cytoplasmic TKT was high in 54.6% (59/108) of the HCC tissues (Fig. 7i). High expression of FBXL6 was also observed in HCC tissues (Fig. 7i). A strong positive correlation between FBXL6 expression and the cytoplasmic TKT level was observed in 51.85% (56/108) of the HCC tissues (Fig. 7j). Importantly, high coexpression of FBXL6 and TKT in HCC tissues was associated with a worse prognosis than was low coexpression of these two proteins (Supplementary Fig. 14a). Consistent with this pattern, high coexpression of FBXL6 and TKT was associated with high-grade HCC, poorly differentiated tumors, and a high frequency of recurrence and metastasis compared with low coexpression of these two proteins (Supplementary Fig. 14b–e).
In conclusion, our study identifies a superoncogene named FBXL6 that plays a critical role in promoting HCC tumor formation and lung metastasis in mice and humans. Our results suggest that elevated FBXL6 expression promotes VRK2 phosphorylation-dependent TKT polyubiquitination and activation and thereby enhances ROS-mTOR-mediated upregulation of VRK2 and PD-L1 expression and immune evasion, ultimately promoting HCC metastasis (Fig. 7k). Targeting TKT could significantly block hepatocytic FBXL6-driven HCC metastasis.