Comprehensive profiling of the TRIpartite motif family to identify pivot genes in hepatocellular carcinoma

Abstract Introduction TRIpartite motif (TRIM) proteins are important members of the Really Interesting New Gene‐finger‐containing E3 ubiquitin‐conjugating enzyme and are involved in the progression of hepatocellular carcinoma (HCC). However, the diverse expression patterns of TRIMs and their roles in prognosis and immune infiltrates in HCC have yet to be analyzed. Materials Combined with previous research, we used an Oncomine database and the Human Protein Atlas to compare TRIM family genes' transcriptional levels between tumor samples and normal liver tissues, as verified by the Gene Expression Profiling Interactive Analysis database. We investigated the patient survival data of TRIMs from the Kaplan–Meier plotter database. Clinicopathologic characteristics associations and potential diagnostic and prognostic values were validated with clinical and expressional data collected from the cancer genome atlas. Results We identified TRIM28, TRIM37, TRIM45, and TRIM59 as high‐priority members of the TRIMs family that modulates HCC. Low expression of TRIM28 was associated with shorter overall survival (OS) than high expression (log‐rank p = 0.009). The same trend was identified for TRIM37 (p = 0.001), TRIM45 (p = 0.013), and TRIM59 (p = 0.011). Multivariate analysis indicated that the level of TRIM37 was a significant independent prognostic factor for both OS (p = 0.043) and progression‐free interval (p = 0.044). We performed expression and mutation analysis and functional pathways and tumor immune infiltration analysis of the changes in TRIM factors. Conclusion These data suggested that TRIM28, TRIM37, TRIM45, and TRIM59 could serve as efficient prognostic biomarkers and therapeutic targets in HCC.


| INTRODUCTION
Hepatocellular carcinoma (HCC) is the sixth most common cancer in the world and was the third leading cause of cancer-related death in 2020. 1 The number of newly diagnosed HCC cases is increasing worldwide. This is a common aggressive malignancy in developing countries. Liver resection and transplantation are treatment options in selected patients, but surgery is often followed by HCC recurrence. 2 The 5-year rate of HCC recurrence is as high as 70%. 3 The majority of patients are not eligible for this radical curative therapy, however, which has a median survival of <1 year. 4 While multitargeted kinase inhibitors received FDA approval for HCC treatment, there are still no reliable biomarkers for accurate patient' prognosis due to tumor heterogeneity. There is an urgent need for novel molecular markers to effectively enhance prognosis.
The TRIpartite motif (TRIM) family of proteins, previously designated as RBCC for the presence of a Really Interesting New Gene (RING)-finger domain, zinc-finger domains named B-box motifs, and the associated Coiled-Coil region were first identified through a functional genomic approach as 37 proteins sharing common functions and cellular compartments. 5,6 Similar proteins have been discovered, including more than 80 TRIM protein genes in humans. Among these, only eight TRIM proteins are RING-less. 7 The RING-finger domain containing-TRIM protein family and the homologous to E6-AP COOH terminus (HECT) family constitute the two parts of the E3 ubiquitin ligases, which catalyzes the ubiquitin conjugation with E1 ubiquitin-activating enzyme and E2 ubiquitin-conjugating enzymes. 8 Although they differ in the C-terminus and overall domain structure, TRIM proteins participate in diverse cellular mechanisms and biologic processes, including gene expression, cell cycle progression, apoptosis, signal transduction, developmental processes, and carcinogenesis. 9 Evidence has demonstrated that a variety of TRIM protein family members play vital roles in promoting proliferation, migration, and invasion of HCC cells. A close association between TRIM proteins and poor survival has been noted. [10][11][12][13] Some TRIM proteins are of value in HCC and the explicit roles of TRIM family gene' expression on patient' diagnosis, disease progression, and immune regulation in a tumor microenvironment must be clarified.
No systematical analysis has been performed to identify the role of TRIMs in HCC. Here we performed a comprehensive bioinformatics analysis with genomesequencing technology and explored various public databases, to determine whether TRIM proteins have potential as therapeutic targets and prognostic biomarkers in HCC.
Our study suggests that TRIM28, TRIM37, TRIM45, and TRIM59 may be useful to stratify the prognosis in HCC patients. These four TRIM family members hold promise for HCC diagnosis and treatment.

| Oncomine analysis
A meta-analysis of TRIMs in cancer expression was performed using an online cancer microarray database Oncomine (www.oncom ine.org). 14 The fold change and p value cutoffs were defined as 2 and 0.01, respectively. The heatmap figures were generated for patient annotated gene expression.

| Protein atlas analysis
The Human Protein Atlas (HPA) provides qualitative antibody-based annotation of protein expression levels across 76 different cell types in 44 human tissue categories. Samples with immunohistochemistry scoring intensity in HCC tissues were included.

| GEPIA data set
The online Gene Expression Profiling Interactive Analysis (GEPIA) platform (http://gepia.cance r-pku.cn/) is an analysis tool containing the RNA sequence expression and mutation analysis and functional pathways and tumor immune infiltration analysis of the changes in TRIM factors. These data suggested that TRIM28, TRIM37, TRIM45, and TRIM59 could serve as efficient prognostic biomarkers and therapeutic targets in HCC.

K E Y W O R D S
hepatocellular carcinoma, immune microenvironment, TRIpartite motif (TRIM) family, tumor prognosis data of 9736 tumors and 8587 normal tissue samples. 15 In this study, we conducted a differential expression analysis for the mRNA of HCC and normal liver tissues as well as clinicopathologic analyses of TRIMs with GEPIA. The "Multiple Gene Comparison" module of the "LIHC" data set was used to conduct multiple gene comparison analysis of TRIMs. The p value was assessed by Student's t test and the cutoff was 0.05.

| The Kaplan-Meier plotter
The prognostic values of TRIMs mRNA expression were evaluated using the Kaplan-Meier plotter database (www.kmplot.com), which contains gene expression data and survival information for HCC patients. To analyze four indicators, the overall survival (OS), progress-free survival (PFS), relapse-free survival (RFS), and disease-specific survival (DSS) of patients with HCC, clinical samples were split according to median expression and estimated by a Kaplan-Meier survival plot. Forest plots were used to present the Cox regression models.

| cBioPortal
The cBioPortal (http://cbiop ortal.org) provides researchers to explore and analyze multidimensional genomics data. 16 According to the TCGA database, genomic profiles were obtained from cBioPortal with the threshold of ±2.0.

| Functional enrichment and TRIMsrelated pathway analysis
The KEGG pathway and GO term analysis have been previously described. 17 To better analyze and visualize the functional profiles of TRIMs, we used the R package clusterProfiler (3.8.0). Next, to prepare for the gene set enrichment analysis (GSEA), the ratio between the TRIMshigh and -low groups in HCC was calculated. The false discovery rate (FDR) q value of <0.25, the adjusted p value of <0.05, and the absolute normalized enrichment score (NES) of >1 were used to select the most significantly enriched signaling pathways.

| Coexpression network analysis
The correlations of the four TRIM members with each other were obtained by analyzing mRNA expressions (RNA-sequencing [RNA-seq] version (v.)2 RSEM) from the XIANTAO platform (www.xiant ao.love) from the TCGA project according to the online instructions.

| TIMER 2.0
The TIMER 2.0 database (http://timer.cistr ome.org/) is a comprehensive tool that quantifies tumor-infiltrating immune cells and provides analysis of immune infiltration with cancer molecular profiles. 18 The website is enabled with an intuitive interface, convenient visualization of the analysis results, and a flexible selection of particular methodologies. In the present study, the "gene module" was used to estimate the interrelationship between TRIM expression and immune cell infiltration. The "survival module" was used to evaluate the interrelationship between the assessment of clinical outcomes and the infiltration of immune cells and TRIM levels. We adopted the Wilcoxon-rank sum test and Spearman correlation.

| Statistical analysis
Data and figures were mainly processed using SPSS 23.0 and the GraphPad Prism 8.0 software. Values are displayed as the mean ± SD. The association between TRIM expression and clinicopathologic features in HCC patients was analyzed with the Wilcoxon signed-rank sum test and Spearman's or Pearson's test. Survival analyses were conducted with the log-rank test and the Cox proportional hazard regression model. The cutoff value of the TRIM levels was dependent on the median value. A probability value <0.05 was considered a statistically significant result. Significance was defined at ***p < 0.001, **p < 0.01, and *p < 0.05. Figure 1 shows a flowchart of the screening process for the targeted TRIM family genes correlated with HCC. A total of 76 TRIM factors were manually retrieved using the Oncomine database. We first compared the transcriptional levels of TRIMs in HCC with those in normal liver tissues (Table 1; Figure 2). Based on Oncomine, the mRNA expression levels of TRIM6, TRIM11, TRIM16, TRIM24, TRIM28, TRIM31, TRIM37, TRIM45, TRIM52, and TRIM59 in HCC tissues were significantly elevated in HCC versus normal liver tissue, and TRIM55 was remarkably downregulated in normal liver tissues.

| Expression levels of TRIM family genes and comparison between HCC and normal liver tissues
Specifically, the mRNA level of TRIM6 was overexpressed in patients with HCC, with a fold change of 2.178 in Wurmbach's data set (Table 1). Meanwhile, Wurmbach et al 19 also showed that TRIM11 was increased in those with HCC (fold change = 2.230 and p value = 7.26E-7). Chen et al 20 found a statistically significant upregulation of TRIM11 in HCC (fold change = 1.702 and p = 1.00E-18) as well as in focal nodular hyperplasia (fold change = 1.614 and p = 0.026). Three data sets suggested that TRIM16 expression was elevated in HCC 19,21,22 compared with normal tissues (with the fold change of 5.726, 1.779, and 2.601, respectively).
No statistical significance was reported between cirrhosis and liver samples. 21 A total of five data sets and TCGA liver statistical data reported the overexpression of TRIM24 in HCC. In Roessler's data set, TRIM24 was significantly elevated in HCC with a fold change of 2.046. 22 Chen et al also indicated that the level of TRIM24 (p = 8.45E-16) in HCC was remarkably elevated with a fold change of 1.957, yet no statistical significance was found in focal nodular hyperplasia compared with normal liver tissues (p = 0.167). 20 Similarly, the results of Wurmbach, 19 Mas, 21 and Guichard 23 all implied that TRIM24 increased significantly in HCC tumor tissues compared with normal samples, which is consistent with the results of TCGA liver statistics. The transcriptional levels of TRIM28 in HCC were remarkably higher than in normal liver tissues in the Roessler (fold change = 1.802 and p = 2.53E-41), Chen (fold change = 1.529 and p = 7.36E-10), and Wurmbach (fold change = 1.712 and p = 2.53E-4) data sets. 19,20,22 The fold change of TRIM31 expression in HCC was 3.010 (p = 1.92E-6), 1.286 (p = 4.27E-7), 1.369 (p = 3.15E-8), 1.058 (p = 3.35E-9), and 1.397 (p = 4.03E-17) in the data sets of Wurmbach, 19 Mas, 21 Chen, 20 Guichard,23 and Roessler,22 respectively. In TCGA (Table 1), TRIM31 was also overexpressed in HCC (fold change = 1.114 and p = 8.06E-9) compared with normal liver tissues. TCGA liver statistics reported a significant upregulation of TRIM37 in HCC (fold change = 1.098 and p = 2.02E-8). Similarly, the results of Roessler, 22 Guichard,23 and Wurmbach 19 all suggested that TRIM37 increased significantly in HCC tumors compared with normal liver samples (with fold change of 2.183, 1.039, and 1.358, respectively). Chen et al 20 showed upregulation of TRIM45 in HCC (fold change = 2.292 and p = 1.33E-10). The same observation was found in the Roessler data set 22 (fold change = 1.172 and p = 6.03E-17).

TRIMs in HCC
To validate TRIM family expressions at the protein level, we assessed immunohistochemical results obtained from the HPA database and compared the outcomes from the TCGA and GTEx databases. Representative images are shown in Figure 3.
TRIM6 protein levels showed negative staining and intensity based on the HPA, yet a significantly higher level of TRIM6 was detected at the transcriptional level (p = 2.2E-13) ( Figure 3A,L). TRIM11 demonstrated low staining and weak intensity in normal liver samples and as medium staining and moderate intensity in   Negative staining and intensity were detected from TRIM31 in both HCC and normal liver tissue, while significantly higher expression was found in the TCGA and GTEx databases (p = 9.2E-26) ( Figure 3F,Q). Protein and mRNA levels of TRIM37, TRIM52, TRIM55, and TRIM59 were increased in HCC tissues compared with normal liver samples by immunohistochemical experiments (p = 4.6E-28, p = 0.07, p = 1.2E-04 and p = 1.3E-17, respectively) ( Figure 3G,I-K,R,T-V). TRIM45 was detected with no significant difference based on the HPA, but with a higher expression in the HCC tissues than in the normal samples at the transcriptional level (p = 1.6E-24) ( Figure 3H,S).

| Influence of TRIM family gene mRNA expressions on patients' survival
The association between TRIM family expressions and patients' survival was analyzed with the Kaplan-Meier plotter online database. The survival curves of OS, PFS, RFS, and DSS were plotted by GraphPad Prism (Figure 4).
Briefly, the Kaplan-Meier survival curve and logrank test analyses suggested that the increased TRIM24, TRIM28, TRIM37, TRIM45, and TRIM59 mRNA levels were significantly correlated with poor OS (p = 0. and DSS (p = 0.0045 and 0.033, respectively). Moreover, the OS rate was significantly higher in TRIM6-low HCC patients (p = 0.041), whereas the PFS rate was higher in the TRIM52-low group (p = 0.0039). However, there was no significant difference detected in survival analyses of TRIM31.

| Verification of TRIMs with prognostic signature in HCC
A univariate logistic regression analysis showed that high expression of TRIM28, TRIM37, TRIM45, and TRIM59 were associated with a worse OS  (Tables S1 and S2). To further explore potential clinical biomarkers associated with the survival of HCC patients, a multivariate Cox regression analysis was used to assess gender, age, TNM stage, pathologic stage, Child-Pugh grade, AFP, albumin levels, and prothrombin time. Overexpressed TRIM37 was an independent prognostic factor associated with poor OS (HR: 1.663; CI: 1.016-2.720; p = 0.043) and PFI  (Tables S1 and S2).

| Genetic alteration, coexpression, and pathway and interaction analyses of TRIMs in patients with HCC
We used the cBioPortal online tool for HCC to comprehensively analyze the selected family members of TRIM alterations. As shown in Figure 7A, TRIM28, TRIM37, TRIM45, and TRIM59 were altered in 1.4%, 2.8%, 0.3%, and 0.9% of the queried HCC samples, respectively. Amplification and missense mutations were the most common changes. Interestingly, the To assess the role of the selected TRIMs in biologic processes, GO function and KEGG pathway analysis were calculated with the clusterProfiler package using the TCGA database. As expected, ubiquitin-protein-related activity and negative regulation of viral-related processes were significantly regulated by the TRIM alterations in HCC ( Figure 7C,D; Table S4). The GeneMANIA online tool was used to construct the network for the selected TRIMs with a network-based gene-ranking algorithm. The results showed that the four TRIM members were closely associated with genes involving tumor necrosis factor receptor super-family binding, ubiquitin-like protein ligase binding, cytokine receptor binding, and I-kappaB kinase/NF-kappB signaling ( Figure 7E).

| Associations between TRIM family genes, immune infiltration, and immune regulators
Accumulating evidence indicates that the immune microenvironment is key to modulating tumor progression and immunotherapy. Immunotherapy has revolutionized cancer therapy. Hence, we conducted further study of the relationships between TRIMs and immune cell infiltration level. In Figure 8A Figure 8B). Similar results were found in TRIM45 and TRIM59 genes. We found positive correlations between TRIM45 and TRIM59 and the infiltration of Th2 cells, T follicular helpers, and NK CD56bright cells and significantly negative associations with neutrophils, Treg, Mast cells, Th17 cells, cytotoxic cells, DC, and pDC ( Figure 8C,D). All the statistically significant numbers are bold.

| GSEA identifies TRIMs-related signaling pathways in ubiquitination and deubiquitination
The results showed that the HCC groups with high expression of TRIM28, TRIM37, TRIM45, and TRIM59 were significantly

| DISCUSSION
To date, human HCC is characterized as the most common visceral neoplasms with a high mortality. 24 Elucidating the molecular mechanism relevant to HCC carcinogenesis may help establish effective therapeutic targets or promising molecular biomarkers. However, the role of TRIMs in HCC warrants further investigation. The present study is the first to comprehensively explore the expression levels and potential prognostic values of members of the TRIM family in HCC. Through step-by-step exploration, we found that four genes were differentially overexpressed in HCC and were also statistically associated with various survival indicators. The data suggest that TRIM28, TRIM37, TRIM45, and TRIM59 were playing significant roles in HCC and may serve as promising prognostic biomarkers and therapeutic targets. Among the TRIMs, several have been reported to have an antitumor effect in HCC (Table 3). Previous studies have elucidated that TRIM21, TRIM26, TRIM35, TRIM50, and TRIM56 acted as tumor suppressors by inhibiting tumor cell proliferation, and TRIM3, TRIM16, TRIM21, TRIM25, and TRIM26 negatively regulated migration and invasion in HCC cells. The results were reversed when the The network for TRIMs and the 20 most functionally similar genes as well as their pathways using genomics and proteomics data genes were knocked down or knocked out. [25][26][27][28][29][30][31] Low expression levels of TRIM21, TRIM26, and TRIM35 had significantly shorter OS, and both TRIM3 and TRIM55 were independent factors positively affecting the prognosis of HCC patients. 26,28,[32][33][34] Different molecular mechanisms were proposed to participate in tumor cell regulation, including metabolic reprogramming, cell cycle arrest, phosphorylation, and ubiquitination.
However, more TRIMs were also stimulatory and cancerogenic factors in HCC ( Table 4). The TRIM11, TRIM31, TRIM32, TRIM52, and TRIM59 expression levels were significantly upregulated in human HCC and acted as active regulators in promoting cell proliferation and in vivo tumorigenesis. 10,11,13,[35][36][37][38] As expected, downregulation of the above TRIMs markedly suppressed in vitro HCC cell proliferation and decreased the volume and weight of tumors in vivo. Except for the TRIM family members mentioned above, TRIM29, TRIM37, TRIM44, and TRIM65 facilitated HCC migration and invasion abilities in vitro. [39][40][41][42] Consistent with previous results TRIM11, TRIM29, TRIM32, TRIM44, and TRIM59 gene expression results, higher levels were significantly associated with poorer OS and other clinical outcomes.
Similar observations were reported in TRIM7 and TRIM25. 12,43 The expression levels of TRIM28, TRIM44, and TRIM65 were independent factors used to predict the survival rate of HCC patients. 41,42,44 Moreover, further experiments were performed to assess the relationships between chemotherapy resistances and TRIM family member expression. The upregulation and prognostic potential outcomes in TRIM28, TRIM37, and TRIM59 were also observed in our study. Since TRIM32 was found to accelerate oxaliplatin resistance and TRIM44 to enhance the doxorubicin resistance of HCC cells in 2016, accumulating evidence has supported the causative drug-resistance role in TRIM members and suggested targeting TRIM family proteins to augment the sensitivity of HCC toward chemotherapy. 12,36,42,45 Although they may have similar biologic characteristics, the underlying biologic mechanisms differ from one TRIM family member to another and remain poorly understood. The contribution of regulatory in cell signal transduction, ubiquitylation, phosphorylation, and other regulatory mechanisms need further verification.
Although immunotherapy has been a breakthrough in HCC treatment, immune checkpoint inhibitors benefit only a small percentage of patients. 46 The role of the F I G U R E 9 Enrichment plots from the gene set enrichment analysis (GSEA). Pathways and biological processes related to ubiquitination and deubiquitination were differentially enriched in (A-C) TRIM28, (D-F) TRIM37, (G-I) TRIM45 and (J-L) TRIM59 in HCC immune system in cancer growth must be clarified and reliable clinical markers for predicting immunotherapy responses explored. Tumor-infiltrating lymphocytes (TILs) play key roles in tumorigenesis and neoplasm. 47 TILs establish a complex intracellular network, maintaining and improving the immunosuppressive microenvironment, and promoting tumor immune escape, eventually leading to tumor progression. 48 In the present manuscript, all four members selected from the TRIMs family were closely related to TILs. There were more TAMNK cells and fewer Th17 cells in the TRIM28-, TRIM37-, TRIM45-, and TRIM59-high HCC groups compared with the low expression groups, indicating that the downregulation of adaptive immunity contributed to an increasing in innate immunity. Previous studies have shown that TRIM28 is a negative immune regulator in response to various immune stimuli, and inhibition of TRIM28 leads to the synthesis of immunostimulatory cytokines and activation of anticancer responses. 49,50 Consistently, TRIM28 knockout in melanomas was sufficient to activate immune infiltration. 51 The negative correlation between TRIM28 and immune infiltration has been reported in lung adenocarcinoma based on the Estimation of STromal and Immune cells in MAlignant Tumor tissues using Expression data (ESTIMATE) algorithm in regulating the levels of diverse chemokines and molecular signaling pathways. 52 Based on the involvement of TRIM37 in T lymphocyte derangement in MUL syndrome, targeting TRIM37 will reveal the importance of T cell function, particularly in CD4 + lymphocytes. 53 Collectively, the data provide a rational to develop TRIMs-targeted small molecules to enhance therapeutic effects.
To further investigate the biologic functions of TRIMs selected as pivotal genes in HCC, we performed KEGG, GO, and GESA analyses. The results showed that 88.1% of the significantly enriched pathways were common among the four TRIM members, notably, ubiquitination and deubiquitination. Interleukin signaling, growth factor receptor signaling, cell cycle regulation pathway, MAPK family signaling, Wnt signaling, and nuclear receptor pathway were enriched in the TRIM28-, TRIM37-, TRIM45-, and TRIM59-high phenotype. The RINGfinger domain of the TRIM family members harbors E3 ubiquitin ligase activity and confers the ubiquitous activations to their target proteins. 9 As HCC suppressor genes, TRIM16 regulates ZEB2 proteins through ubiquitn-proteasome pathway to inhibit EMT behavior and the number of distant metastasis tumors, which could be completely abrogated by proteasome inhibition. 27 Through immunoprecipitation and a ubiquitination assay, it was verified that TRIM50 could induce poly-ubiquitination of the SNAIL protein in both the nuclei and cytoplasmic compartments. The RING domain deleted mutant significantly rescued the negative degradation of SNAIL. 31 Similar mechanisms were also detected in TRIM7 reducing Src protein. 54 As for the TRIM oncogenes in HCC, TRIM65 triggered βcatenin signaling via ubiquitylation of the Axin1 protein. TRIM52 significantly modulated the ubiquitination of PPM1A in HCC cells to exert carcinogenic functions. 11,41 Using integrated investigation systems, including cellular models, animal models, and clinical liver cancer specimens, TRIM31 promoted HCC progression by inducing K48-linked poly-ubiquitous degradation of the tuberous sclerosis complex (TSC), the 1-TSC2 complex, and the p53-AMPK axis to mediate anoikis resistance. 10,55 Furthermore, a ubiquitous analysis showed that TRIM25 poly-ubiquitinated the metastasisassociated 1 protein (MTA1) and participated in Keap1 degradation by the ubiquitin-proteasome pathway. 12,30 The latter could be reversed by a series of truncated mutants of TRIM25, including three truncated mutants and Glu9 and Glu10 mutations. However, the mechanisms underlying the E3 ubiquitin ligase activities require further investigation.
We comprehensively analyzed the expression and prognostic value of TRIM family proteins in liver cancer and provided a thorough understanding of the heterogeneity of HCC's molecular biologic properties. Our results indicated that the increased expression of TRIM28, TRIM37, TRIM45, and TRIM59 in liver cancer tissues likely plays a pivot role in HCC. High expression of these four could also serve as biomarkers to identify high-risk subgroups and potential treatment targets to enhance the prognosis of HCC patients.