TRIM proteins in hepatocellular carcinoma

The tripartite motif (TRIM) protein family is a highly conserved group of E3 ligases with 77 members known in the human, most of which consist of a RING-finger domain, one or two B-box domains, and a coiled-coil domain. Generally, TRIM proteins function as E3 ligases to facilitate specific proteasomal degradation of target proteins. In addition, E3 ligase independent functions of TRIM protein were also reported. In hepatocellular carcinoma, expressions of TRIM proteins are both regulated by genetic and epigenetic mechanisms. TRIM proteins regulate multiple biological activities and signaling cascades. And TRIM proteins influence hallmarks of HCC. This review systematically demonstrates the versatile roles of TRIM proteins in HCC and helps us better understand the molecular mechanism of the development and progression of HCC.


Introduction
The tripartite motif (TRIM) protein family is a highly conserved group of RING-type E3 ligases with 77 members known in the human, most of which consist of a RING-finger domain, one or two B-box domains, and a coiled-coil domain [1]. Dysregulation of TRIM proteins has been found and shown crucial roles in different types of diseases including inflammation, viral infection, and cancer [2][3][4].
Liver cancer is the fourth leading cause of cancerrelated death globally, and hepatocellular carcinoma (HCC) represents approximately 90% of primary liver cancer [5,6]. In HCC, TRIM proteins have impacts on cell proliferation, apoptosis, cancer metastasis, metabolic reprogramming, stemness, carcinogenesis, immunogenicity, and resistance to cancer therapies. Furthermore, targeting TRIM proteins showed its potential effects on HCC. In this review, we summarize the roles of TRIM proteins in HCC as ubiquitin ligases or non-ubiquitination roles. We systematically demonstrate the biological functions of TRIM proteins in HCC and summarize the signaling cascades affected by TRIM proteins.

Structural classification and domain functions of TRIM proteins
The tripartite motif (TRIM) protein family is named for their highly conserved RING domain, B-box domains, and the coiled-coil (CC) region at N-terminal. Unlike N-terminal domains, C-terminal domains of TRIM proteins vary in different subtypes, and TRIM proteins can be classified into subfamily C-I to C-XI according to distinctive C-terminal domains [1]. In detail, C-terminal domains of TRIM proteins including COS domain, Fibronectin type-III domain (FN3), PRY domain, B30.2/ SPRY domain (SPRY), acid-rich region (ACID), filamintype I domain (FIL), NHL domain, PHD domain, bromodomain (BRD), Meprin and TRAF-homology domain (MATH), ADP-ribosylation factor family domain (ARF), and transmembrane region (TM). Another subfamily called UC refers to 8 TRIM proteins without RING domain (Fig. 1).
N-terminal and C-terminal domains exert functions cooperatively or independently during the biological process in HCC. The SPRY helps the nuclear translocation of TRIM22 [13]. The SPRY also mediates the interaction of tyrosine-protein kinase Src (SRC) with TRIM7, the intracellular part of interferon alpha/beta receptor 1 (IFNAR1) with TRIM10, and pleckstrin homology domain leucine-rich repeats protein phosphatase 1 (PHLPP1) with TRIM11, to promote their substrate degradation [14][15][16].
Notably, TRIM14 has no RING domain, but it is still able to mediate ubiquitination degradation of NS5A through SPRY [17]. The NHL domain helps with the recognition between TRIM71 and a structural RNA stem-loop motif within the 3'-untranslated region (UTR) of CDKN1A mRNA [18]. Functions of other C-terminal domains need further research in HCC.
Apart from genetic variations, expressions of TRIM proteins are also modulated via epigenetic mechanisms, including DNA methylation, mi-RNA, circRNA, and long non-coding RNAs (lncRNA). TRIM21 is down-regulated by methylation in its 5′-UTR [50]. TRIM33 is reduced through aberrant CpG methylation at its promoter [22].
Furthermore, combinations of TRIM proteins with other proteins show higher efficiency in clinical assessments. Our study found that the combination of TRIM33 and phosphorylated SMAD2 is more efficient in predicting recurrence and OS in HCC [22]. TRIM28/minichromosomal maintenance complex component 6 (MCM6) is a novel marker for diagnosing HCC [66]. Zinc finger protein 354C (ZNF354C)/ TRIM28/HDAC6 and TRIM35/pyruvate kinase isoform M2 (PKM2) are more effective prognostic factors for HCC [23,67].

Sustaining proliferation
Sustaining proliferation is a hallmark of cancer. Aberrant expression of TRIM proteins leads to abnormal cell cycle progression and sustaining proliferation.
Mutation of myc is significantly deleterious to HCC development, abnormal activation of myc-related signaling is crucial for the proliferation of HCC. Mitogenactivated protein kinase kinase kinase 13 (MAP3K13) promotes phosphorylation and suppresses proteasomal degradation of TRIM25. TRIM25 mediates the ubiquitin degradation of F-box/WD repeat-containing protein 7α (FBXW7α), which is the main E3 ligase that down-regulates c-myc [88]. TRIM56 is associated with up-regulated c-Myc and activated β-catenin [26]. TRIM71 inhibits functions of miR-let-7 and up-regulated down-stream c-Myc, Lin-28B, HMGA2 and type 1 insulin-like growth factor receptor (IGF1R) [42].
Anoikis is defined as the detached from the extracellular matrix (ECM) -induced apoptosis, and resistance to anoikis is a hallmark of cancer [98]. TRIM31 targets TP53 proteasomal degradation to over-activates the AMP-activated protein kinase (AMPK) pathway to promote anoikis [99]. Oppositely, TRIM50 down-regulates SNAIL through ubiquitin degradation, therefore reverses EMT and inhibits anoikis resistance [24].

Metabolic reprogramming of HCC cells
The Warburg effect refers to the metabolic reprogramming in cancer that energy is generated through aerobic glycolysis instead of mitochondrial oxidative phosphorylation [108]. PKM2 is a rate-limiting enzyme in glycolysis, whose phosphorylation provides extra metabolic advantages for HCC cells. TRIM35 competes with fibroblast growth factor receptor 1 (FGFR1) to interact with PKM2 and consequently inhibits the phosphorylation of PKM2 [23,109]. TRIM28/melanoma-associated antigen (MAGE)-A3/MAGE-C2 complex promotes the Warburg effect through ubiquitin degradation of fructose-1,6-bisphosphatase 1 (FBP1), which is the ratelimiting enzyme in gluconeogenesis [110]. TRIM11 is significantly induced upon glucose deprivation. TRIM11 down-regulates AMPKβ2 through ubiquitin degradation to suppress AMPK pathway and leads to starvation-induced autophagy [111].
Furthermore, TRIM proteins are involved in lipid, hormone, and biliary acid metabolism. Liver-specific TRIM28-knockout mice exhibit aberrant androgen receptor stimulation, biliary acid disturbances, and significantly altered gut microbiota such as Prevotella, Akkermansia muciniphila, and Bacteroides uniformis, which are species predominantly associated with metabolic dysfunction and inflammation. Notably, this abnormality can be completely abolished under axenic conditions [72]. Liver-specific TRIM28-knockout results in sexual dimorphic metabolic syndrome through activating the ERK1/2-MAPK pathway. Loss of TRIM28-dependent epigenetic silencing results in activation of fat-specific protein 27 (FSP27), glutathione S-transferase, Cyp2d9, Cyp2a, Cyp2b, and Cyp3a gene clusters, and thereby leads to male-predominant liver steatosis and adenoma [112].
PML-knockout mice show increased white fat initially, but exhibit weight loss and white fat browning in end-stage HCC with the metabolic reprogramming from glycogen storage to lipolysis [113]. PML-deficient HBsAg-transgenic mice showed obvious oxidative phosphorylation and fatty acid metabolism impairments and encountered early steatosis-specific liver tumorigenesis [114].

Initiation of HCC
Somatic hepatocyte-specific inactivation of TRIM24, TRIM28, or TRIM33 all promotes spontaneous HCC [74,117]. TRIM24 forms quantities of dimers with TRIM33, and a few trimers with TRIM28 and TRIM33. Liver TRIM24-knockout induced HCC is significantly promoted by further loss of TRIM33, and is slightly hindered by further loss of TRIM28 [74]. Mechanistically, TRIM24 attenuates RARα-mediated transcription through chromatin remodeling as mentioned above. Thus, TRIM24 deficiency activates downstream targets of the RA pathway such as Cyp26a1, protein-glutamine gamma-glutamyl transferase 2 (TGM2), RBP1, and receptors for retinol uptake STRA6 (STRA6). Notably, deletion of a single allele of RARα is sufficient to restore the phenotype of TRIM24-knockout mice [117]. TRIM24 also binds to the RARE of STAT1 promoter to inhibit STAT1 expression and promotes expressions of tumor-suppressive factors such as p21, Bmyc, and hepatocyte nuclear factor 6 (HNF6) [76]. Tumor initiation cells (TICs) in HCC are a subset of HCC cells with stem cell features and influence the initiation, cell growth, drug resistance, and recurrence of tumors [118]. Arsenite treatment represses PML expression, which down-regulates Oct4, Sox2, and Klf4 expressions. As a result, it reduces viability and stemness of CD133+ CD13+ TICs and enhances sensitivity to pirarubicin in HCC [119].

Anti-HBV effect
HBV-and HCV infection are leading risk factors for HCC worldwide, and dysregulated responses to the infection of HBV or HCV fuel the progression of HCC [120][121][122].
Multiple TRIM proteins are identified inhibiting HBV replication in HCC, including TRIM5, TRIM6, TRIM11, TRIM14, TRIM25, TRIM26, TRIM31, and TRIM41 [123]. TRIM25 is an interferon-stimulated gene (ISG) augmented by IFN and IL-27, which mediate lysine 63-ubiquitination of RIG-I to suppress HBV replication [124]. PML is significantly associated with genomic instability and DNA repair in HBV-related HCC [125,126]. PML is negatively correlated with HBsAg because of proteasomal degradation or translocation of HBsAg to the nucleus [113,114,125,126]. PML suppresses the early phase of HCC since it enhances DNA repair and induces resistance to IFN-α or DNA damage-induced apoptosis (Fig. 3A), but turns out to be oncogenic in the end stage. It enhances a metabolic shift from glycogen storage to lipolysis, which implicates more energy available for driving HCC progression (Fig. 3B) [113,125].
HBV regulatory protein X (HBx) stimulates HBV gene expression from the covalently closed circular (cccDNA) and is involved in HCC development [127]. TRIM14 is a STAT1-dependent type-I ISG. The TRIM14 SPRY domain interacts with the C-terminal of HBx to inhibit the formation of Smc-HBx-damage-specific DNA-binding protein 1 (DDB1) complex [128]. TRIM5γ is another type-I ISG which mediates HBx ubiquitin degradation through the B-box domain. TRIM31 is recruited by TRIM5γ and can also mediate HBx ubiquitin degradation [129]. Another study supplemented that TRIM31 is a type-III ISG and can be induced upon HBV replication [130].
HBx also regulates TRIM expressions. TRIM22 can be strongly stimulated by IFN-α and IFN-γ through IFN regulatory factor-1 (IRF1) in HCC. TRIM22 suppresses HBV core promoter by its nuclear-located RING domain, whose translocation is mediated by the SPRY domain (Fig. 3C) [13]. However, HBx protein down-regulates the transcription of TRIM22 through a single CpG methylation in its 5'-UTR to inhibit the binding between the promoter and IRF1, thereby inhibiting the IFN-stimulated induction of TRIM22 and resulting in HCC (Fig. 3D) [50]. And HBx protein promotes expressions of TRIM7 or TRIM52 [43,131].

Prospect
Currently, the poor prognosis and low percentage of patients responding to systemic therapies are characteristics of HCC, and new therapeutic methods for targeting HCC are urgently needed. As TRIM proteins exert functions mainly through the ubiquitin system (UPS), it seems feasible that use proteasomal inhibitors to block TRIM proteins to ameliorate HCC. Proteasome inhibitors like bortezomib, ixazomib, and carfilzomib have shown effectiveness in some cancer, but their applications are unsatisfying in HCC, as bortezomib in HCC in phase II trial lacked activity [150][151][152]. Carfilzomib and gankyrin inhibitors are far from clinical applications [153]. Several factors account for the ineffectiveness in common. Bortezomib may not inhibit the UPS in the liver as expected, or the dose and schedule need further modulations. Alternatively, the crosstalk of intertwined signaling pathways may counteract each other. Side effects of proteasome inhibitors like neuropathy also restrain their application [154].
Recently, proteolysis-targeting chimeras (PROTACs) give novel insight into applications of TRIM proteins. PROTACs technology employs E3 ligase ligands and fuses target protein with E3 ligase by a flexible chemical bond, to elicit ectopic ubiquitination and degrade specific target proteins [155]. PROTACs entered clinical research for cancer therapies in 2019 [156]. And the first oral PROTAC ARV-110 has shown effectiveness in prostate cancer [157]. TRIM proteins have promising applications in HCC through two aspects of PROTACs. TRIM proteins can be direct targets of PROTACs. For instance, dTRIM24 can recruit VHL E3-ligase to elicit potent and selective degradation of TRIM24 [158]. And dTRIM24 has successfully degraded TRIM24 in human metaplastic breast cancer patient-derived xenografts to decrease tumor cell viability [159]. TRIM proteins may also become mediums in PROTACs, which means recruiting TRIM proteins to specifically down-regulate some oncoproteins to alleviate HCC. But the design of new PRO-TACs ligand compounds is challenging since they need to conjugate the "right binding site" for limited ubiquitin sites as well as for reserving enough space to elongate the ubiquitin chain.
Notably, many virus proteins enable to hijack host E3 ligases to antagonize anti-viral factors, which may enlighten the development of PROTACs of TRIM proteins. These proteins seem natural and prototypical PROTACs [156,160]. The HPV E6 oncoprotein employs ubiquitin-specific protease 15 (USP15) to degrade TRIM25 [161]. Murine gamma herpesvirus 68 induces proteasomal-dependent degradation of PML by the virion tegument protein ORF75c [162].
In addition, the C-VI subfamily (TRIM24, TRIM28, and TRIM33) might be the best transitional therapeutic targets in the future. This subfamily has powerful influences on the progression of HCC. Homozygous deletion of any of them leads to spontaneous HCC, and they regulate epigenetic silencing through chromatin remodeling. They affect diverse signaling pathways including the RA pathway, β-catenin pathway, TGF-β pathway, et.al. On the other hand, the PROTAC dTRIM24 has been invented. RA may also restrain the function of TRIM24. This may also enlighten our next-step transitional research.
Besides, there are still some unsolved problems in the TRIM family. The distribution is tightly linked to protein functions. We collect the intracellular location of TRIM proteins from UniProt (Additional file 2: Table S3) [163]. But few researchers concerning about the sublocation of TRIM in HCC. Another shortness is the relationship between TRIM and first-or second-line therapy drugs for HCC. It is worth further investigating whether TRIM may benefit our current therapies in HCC. As lncRNAs are essential in regulating expressions and functions of TRIM proteins, it seems better to investigate relationships between lncRNA and TRIM on m6A regulations or chemotherapy resistance in HCC.

Conclusion
Growing clinical research has revealed that expressions of TRIM proteins are frequently altered and significantly associated with clinical indexes and prognosis in HCC. Some TRIM proteins are novel tumor markers and independent prognostic factors for HCC, indicating their potential in early diagnoses, prognosis assessments, and clinical therapies. In HCC, TRIM proteins regulate their proliferation, apoptosis, metastasis, metabolic reprogramming, immune responses, and resistance to cancer therapies. Mechanistically, TRIM proteins regulate levels and functions of downstream proteins through ubiquitination-dependent and independent mechanisms, and specific members of TRIM proteins regulate the activity of TGF-β/Smad, MAPK, PI3K-AKT, Wnt/β-catenin, cell cycle, STATs, and RA signaling cascades in HCC (Fig. 5). Targeting TRIM proteins showed therapeutic potential in HCC.
Additional file 1: Fig. S1. Oncoprint plot of all somatic mutations of TRIM proteins in TCGA-LIHC using cbioportal database.
Additional file 2: Table S1. Differential gene expression under different mutation status in HCC through TIMER2.0. The logFC with statistical significance (p<0.05) are colored in the table. Table S2. Cox analysis for all TRIM proteins based on TCGA-LIHC patients with OS more than a month. Table S3. Intracellular location of TRIM proteins from uniprot.
Additional file 3: Fig. S2. A univariate cox analysis of every 75 TRIM proteins based on TCGA-LIHC dataset with p < 0.05.