VPS72, a member of VPS protein family, can be used as a new prognostic marker for hepatocellular carcinoma

Abstract

Background

Vacuolar protein sorting (VPS) plays a crucial role in intracellular molecular transport between organelles. However, studies have indicated a correlation between VPSs and tumorigenesis and the development of several cancers. Nevertheless, the association between VPSs and hepatocellular carcinoma (HCC) remains unclear.

Methods

By analyzing databases such as The Cancer Genome Atlas (TCGA) and The International Cancer Genome Consortium (ICGC), we investigated the differences in VPSs expression between normal tissue and HCC transcriptomes. Furthermore, we examined the relationship between VPSs expression and overall survival (OS) in patients with HCC. Univariate and multivariate Cox analyses were employed to assess the prognostic value of VPS72 as an independent factor, and the correlation between VPS72 and the tumor immune microenvironment was also analyzed.

Results

We observed significant overexpression of 28 VPSs in HCC tissues compared to normal tissues. The mRNA expression of VPSs displayed a negative correlation with OS, while exhibiting a positive correlation with tumor grade and stage. Additionally, both univariate and multivariate Cox analyses identified VPS72 as a potential independent risk factor for HCC prognosis. Overexpression of VPS72 demonstrated a positive correlation with various clinicopathological factors associated with poor prognosis, as well as the infiltration levels of immune cells.

Conclusion

Therefore, our research shows that VPSs participate in HCC occurrence and development, especially VPS72, which may act as a potential target for HCC treatment and prognosis biomarker.

INTRODUCTION

Hepatocellular carcinoma (HCC) ranked the sixth most prevalent malignancy and the third major cause of cancer-related mortalities globally in 2020, with about 906,000 new cases and 830,000 deaths.¹ Regarding HCC treatment, surgical resection can be an option for only 5%–15% of patients at early stages, while the advanced treatment depends on transcatheter arterial chemoembolization (TACE) or oral sorafenib.² In recent years, immunotherapy and molecularly targeted therapy research have advanced rapidly, with the combined anti-PD-L1 (programmed cell death-Ligand 1) antibody atezolizumab and VEGF（vascular endothelial growth factor）-neutralizing antibody bevacizumab becoming the standard first-line therapy for HCC,³ but the effect of immunotherapy on nonalcoholic steatohepatitis (NASH) and HCC is completely unclear, so the application of immunotherapy in HCC has not achieved very satisfactory results.⁴ HCC has various molecular pathogenesis; HCC has many potential operable mutations, but these mutations have not been translated into clinical practice.⁵ Therefore, through the discovery of new biomarkers of diagnosis and treatment, we can better understand the complex relationship between HCC occurrence and development and the immune microenvironment, as well as improve patient prognosis and individualized treatment.

In eukaryotic cells, membrane-wrapped vesicles are involved in the transport of intracellular molecules along organelles, including the endoplasmic reticulum (ER), Golgi apparatus, endosome, secretory vesicles, and lysosomes.^6,7 Functionally, vacuolar protein sorting (VPS) is crucial for decarbonizing in the endocytosis, biosynthesis and secretion, exocytosis, and recovery of intracellular molecules.^8–10 Among them, VPSs play a significant role in mammalian vesicle transport.¹¹ There are 30 human VPS genes in the GeneCards database (). There is only a small amount of research on the association between VPSs genes and their products and human diseases. Many physiological processes in the human body are affected by abnormal VPS expression, such as ubiquitin disruption in the signal transduction pathway following VPS11/VPS18 overexpression.¹² In the absence of VPS13, not only the cortical ER marker Rtn1 accumulation in late endosomes is observed, but also a sharp reduction in the number of ER packaged into autophagosomes.¹³ The abnormal expression of VPS is related to some human neurodegenerative diseases. For example, VPS41 loss is linked to human progressive neurodevelopmental disorders.¹⁴ At the same time, some VPS family members' abnormal expression and prognostic values have also been discovered in tumor-related studies. In gastric cancer, VPS52 is a tumor suppressor gene, and its absence is correlated to a poor prognosis. VPS52 overexpression inhibits tumor proliferation both in vitro and in vivo.¹⁵ Wang et al. found that VPS33b is a crucial tumor suppressor in hepatocellular carcinogenesis, and its downregulation is a key step in inflammation-driven HCC¹⁶; however, the role of the human VPS gene and its products in malignancies is largely unknown. Recently, with the ongoing advancement of microarray technology and bioinformatics, many sequencing data have been produced. The analysis of thousands of gene expression or copy number variations available in the online database was used for examining the correlation between different VPSs expressions and clinical parameters, including the overall survival (OS) rate of HCC patients; moreover, VPSs role in HCC occurrence and development was investigated.

RESULTS

Transcriptional levels of VPSs family between HCC and normal tissues

Combining the analysis results of The Cancer Genome Atlas (TCGA) and International Cancer Genome Consortium (ICGC) databases revealed that VPS4A/4B/8/11/13A/13B/13C/13D/16/18/25/26A/26B/28/29/33A/33B/35/37A/37C/37D/39/41/45/52/53/54/72 was significantly overexpressed in HCC tissues more than in normal liver tissues (Figure 1A,B).

[IMAGE OMITTED. SEE PDF]

VPSs family gene mutation and correlation analysis

Thirty VPS mutations and their relationships were studied using somatic mutation and expression data from VPSs. First, based on the results of the VPSs genetic mutation, the VPS13 gene mutation rate was higher in HCC patients, primarily due to missense mutations (Figure 2A). Then, based on the results of the expression correlation (Supporting Information: Figure S1), the most significant correlation between VPS45 and VPS72 was 0.76 (Figure 2B) in TCGA and 0.83 in the ICGC database (Figure 2C). Protein–protein interaction (PPI) analysis revealed that 49 neighbor genes had a significant relation to VPSs in the STRING database (Figure 2D). Cytoscape 3.7.1 was utilized for visualizing PPI network results via the intergene connectivity, revealing that VPS25 and VPS36 exhibited a central part in this network, and there was a close relation between VPSs and CHMP families (Figure 2E). Through the Metascape database, “vacuolar transport” and “endosomal sorting complex required for transport (ESCRT)” are the main biological functions of these genes (Figure 2F,G).

[IMAGE OMITTED. SEE PDF]

Prognostic value and independent prognostic value of VPSs in HCC patients

The relation between VPSs expression and patient OS was further analyzed in TCGA and ICGC databases, revealing that based on Univariate analysis, the higher VPS16/25/26A/29/33A/33B/37A/37C/54/72 mRNA expression levels were correlated to shorter OS in HCC (Figure 3A,B and Supporting Information: Figure S2), AS well as T stage, pathological stage, and tumor status, were correlated to shorter OS in HCC (Figure 3C). Moreover, based on further multivariate analysis, the higher VPS26A/33A/37C/72 mRNA expression was correlated to shorter OS in HCC patients (Supporting Information: Table S1), so VPS26A/33A/37C/72 have the potential to be independent risk factors for the purpose of predicting the prognosis of HCC patients, with lowest p value of 0.003 of VPS72 in multivariate Cox analysis (Figure 3D).

[IMAGE OMITTED. SEE PDF]

The association between VPS72 expression and clinicopathological factors in patients with HCC

VPS72 was discovered to have the ability to be an independent prognostic factor for predicting the prognosis of HCC patients through the studies described above. The association between VPS72 expression and clinicopathological factors in HCC patients was investigated, revealing that higher VPS72 expression in patients with death (Figure 3E) and higher levels of pathological stage (Figure 3F), histologic grade (Figure 3G), T stage (Figure 3H), and alpha-fetoprotein (AFP) (Figure 3L). Significant elevated VPS72 in HCC tissues of patients with postoperative recurrence (Figure 3I) and vascular infiltration (Figure 3J), and adjacent hepatic tissue inflammation (Figure 3K).

By analyzing multiple GEO chip data, the overexpressed VPS72 in HCC tissues (Figure 4A) was examined, revealing that VPS72 was overexpressed in cancer tissues of HCC patients with multiple data sets in the ONCOMINE database. The fold change was 2.605 (Figure 4B) in Roessler Liver, 2.365 (Figure 4C) in Roessler Liver 2, and 2.342 (Figure 4D) in Chen liver. The fold change in Wurmbach liver is 2.605 (Figure 4E); based on the Wambach data set, the higher VPS72 expression was found in higher tumor grades (Figure 4F), hepatitis C virus (HCV) positive (Figure 4G), patients with satellites (Figure 4H), and vascular invasion (Figure 4I).

[IMAGE OMITTED. SEE PDF]

Identification and analysis of VPS72-related differentially expressed genes (DEGs)

For exploring the abnormal changes in the downstream signal transduction pathway resulting from VPS72 overexpression, the DEG were distinguished in HCC tissues with VPS72 overexpression and low expression according to TCGA data. Out of 2554 DEGs, 1550 were upregulated and 1004 downregulated. By generating Volcan diagrams and bar charts for the purpose of visually showing the DEGs distribution (Figure 5A,B), the heat map depicted the first 25 DEGs, which had significant upregulation and downregulation between VPS72 high- and low-expression groups (Figure 5C,D). Through Gene Set Enrichment Analysis (GSEA) enrichment analysis, we found that multiple signal pathways affecting tumor progression were affected by VPS72 (Figure 5E).

[IMAGE OMITTED. SEE PDF]

Correlation analysis between VPS72 expression and various immune infiltrates

The tumor microenvironment immune cells significantly affect the biological tumor behavior. VPS72 expression was discovered to have a correlation with the infiltration level of several immune cells in the HCC microenvironment by studying the infiltration of various immune cells (Figure 6A,B), which exhibited a positive correlation to Th2 cells, NKCD56 bright cell levels (Figure 6C), but negative correlation to innate immune cell levels such as dendritic cell and cytotoxic (Figure 6D).

[IMAGE OMITTED. SEE PDF]

DISCUSSION

HCC is caused by various factors, including metabolic syndrome, obesity, type 2 diabetes, and nonalcoholic fatty liver disease (NAFLD). It remains one of the deadliest malignant tumors worldwide.¹⁷ HCC occurrence and progression are linked to abnormal malignant cell events such as proliferation, migration, invasion, and autophagy.^18,19 Research has revealed that vesicle transport is closely related to cellular physiological processes like autophagy. VPS family proteins play an important role in vesicle transport¹¹; however, VPS family proteins' role in diseases has not been thoroughly investigated. We have the lead in conducting this study to examine the expression, mutation, and prognostic value of different VPS family members in HCC.

Herein, 28 VPSs were significantly overexpressed in HCC patients. VPS16/25/26A/29/33A/33B/37A/37C/54/72 overexpression was positively correlated with short OS. Further univariate and multivariate analyses revealed that VPS26A/33A/37C/72 might serve as independent risk factors for the prognosis of HCC patients. According to other research, VPS33A is an important subunit of the class C VPS complex. Frameshift mutations in VPS33A may result in protein truncation and alter the interaction with the tumor suppressor UVRAG.²⁰ VPS of melanoma cells with deletion of VPS33A or cappuccino protein (CNO) caused an increase in cis-diaminedichloroplatinum II nuclear localization, DNA platinization damage, and apoptosis, thus improving the sensitivity of therapy.²¹ Chae et al. found that the decrease of VPS26A is related to synaptic defects, astrocyte overactivation, and cognitive impairment in diabetic mice.²² Eastman et al. proved that VPS37C is a functional component of mammalian ESCRT-1.²³ Furthermore, Kolmus et al. found that VPS37C expression is related to cellular stress response related to ESCRT-1 instability, indicating that endocytosis is closely related to tumorigenesis and development,²⁴ but how VPS33A, VPS26A, and VPS37C play a role in tumors needs further study.

At the end of mitosis, VPS72, also known as YL1, can mediate the nuclear recombination of H2A. VPS72 or H2A.Z deficiency caused malformed and dysfunctional nuclei.^25,26 Han et al. pointed out that VPS72 expression level may influence HCC biological process.²⁷ Chris Kaiser et al. found that VPS72 is related to abnormal intracellular transport and abnormal vesicle morphology in Saccharomyces cerevisiae.²⁸ In our study, through the analysis of TCGA and ICGC databases, VPS72 is overexpressed in HCC tissues, which has a correlation to a poor prognosis in HCC patients and may serve as an independent risk factor for HCC, and VPS72 is also highly expressed in HCC tissues in GEO database and multiple ONCOMINE data sets.

Moreover, VPS72 overexpression had a significant correlation to poor prognosis, higher Pathological stage, histologic grade, T stage, AFP level, postoperative recurrence, vascular infiltration, and the degree of adjacent hepatic tissue inflammation in HCC patients; moreover, VPS72 expression in ONCOMINE database was associated with higher tumor grades, patients with satellites, HCV positive, and vascular invasion. GSEA enrichment analysis demonstrated that VPS72 was related to E2F,²⁹ G2M,³⁰ MYC,³¹ and other signal pathways that affect tumor progression. Therefore, we believe VPS72 can affect HCC occurrence and progression and serve as a target for HCC diagnosis and treatment.

Recently, increasing evidence supports that the immune microenvironment affects HCC occurrence and development, which have adverse effects on the clinical prognosis and immunotherapy efficacy.³² A further major result of this study is the association between VPS72 mRNA level and immune cell infiltration degree in HCC. The levels of DC, cytotoxic cells, neutrophils, and especially Th2 cell had a significant correlation to VPS72 expression. Essential immunomodulatory cells in the body are Th cells; under normal circumstances, Th1/Th2 ratio is in a dynamic balance state.³³ The increased Th2 cytokines secretion by patients having malignant tumors leads to the imbalance of Th1/Th2, which may be associated with the immune escape of many kinds of tumors,³⁴ such as liver cancer.³⁵ Therefore, VPS72 may influence the efficacy of immunotherapy in patients with liver cancer by affecting immune cell infiltration, but more research is required.

Our study has limitations, and more research is needed for the purpose of investigating the potential molecular mechanism of carcinogenesis of VPS family proteins. Nonetheless, we are the first to report the differential expression of the VPS protein family in HCC and its potential diagnostic and prognostic value. We observed that the overexpression of VPS protein in HCC was correlated to a poor prognosis. Further analysis displays that VPS26A/33A/37C/72 may serve as independent risk factors for determining patient prognosis. VPS72 overexpression has a positive correlation to various clinicopathological factors with poor prognosis and with the infiltration level of immune cells; hence, herein, the VPS family was found to have the ability to participate in HCC occurrence and development, especially VPS72, which may serve as a potential target for HCC treatment and prognostic biomarkers.

MATERIALS AND METHODS

Clinical samples and data collection

TCGA³⁶ synapse data portal (), LIHC (liver hepatocellular carcinoma) was accessed for downloading RNAseq and mutation data and also extracting the clinical data of the patients (Table 1). The ICGC³⁷ synapse data portal was also accessed to download RNAseq and mutation data for HCC samples.

Table 1 The basic characteristics of HCC patients.

Characteristics	HCC patient number
T stage
T1	183
T2	95
T3	80
T4	13
N stage
N0	254
N1	4
M stage
M0	268
M1	4
Pathologic stage
Stage I	173
Stage II	87
Stage III	85
Stage IV	5
Tumor status
Tumor free	202
With tumor	153
Gender
Female	121
Male	253
Age
≤60	177
>60	196
Histologic grade
G1	55
G2	178
G3	124
G4	12
Adjacent hepatic tissue inflammation
None	118
Mild	101
Severe	18
AFP (ng/mL)
≤400	215
>400	65
Albumin (g/dL)
<3.5	69
≥3.5	231
Child-Pugh grade
A	219
B	21
C	1
Vascular invasion
No	208
Yes	110

VPS family genes mutation and correlation analysis

The GenVisR³⁸ package in R software was first used for analyzing VPS gene mutations in patients with HCC in TCGA and ICGC databases. The corrplot package was then used for examining the correlation between VPS genes in HCC patients.

Construction and functional enrichment analysis of VPSs protein-related interaction network

STRING v11 ³⁹() was utilized to find 49 genes of VPS-related genes. These networks can reveal multiple direct or indirect interactions associated with VPS genes. Cytoscape version 3.7.1⁴⁰ was utilized for displaying the networks and hiding network disconnected nodes. The online analysis tool Metascape⁴¹ offers a comprehensive list of VPS-related genes, VPS genes, and resources for real-time analysis; pathway analyses were performed using Metascape.

Analysis of survival rate of VPSs

Most HCC patients have a poor prognosis, so OS was selected as the survival outcome. The survival package in R software was utilized for univariate Cox analysis, screening out the survival-related VPSs (p < .05). Furthermore, multivariate Cox analysis was used for analyzing the VPSs significantly related to OS and its relationship with clinicopathological factors.

Other database validation

For validation, HCCDB was utilized for analyzing VPS72 expression in HCC. ONCOMINE database⁴² () was then accessed to obtain VPS72 transcriptional expression between various cancer tissues and their corresponding normal control samples. Further analysis of the expression of VPS72 in different clinicopathological factors like grade, HCV positive, patients with satellites, and vascular invasion.

Functional enrichment analysis of VPS72 and its relationship with tumor immune microenvironment

Based on VPS72 expression levels, the gene expression data were categorized into high and low VPS72 groups. GSEA was conducted with the GSEA software. Spearman correlation analysis was used for evaluating the association between VPS72 expression and immune cell infiltration. Moreover, using the Wilcoxon rank sum test, the immune cell level of different VPS72 expression groups were analyzed.

Statistical analysis

R v4.1.2 () and SPSS 26.0 (SPSS Inc.) were utilized for all statistical analyses. The Student's t test or one-way analysis of variance (ANOVA) test was conducted for statistical analyses. Correlations of VPSs were evaluated with Spearman's correlation coefficient. To compare the H-scores of HCC and normal tissues, paired t tests were used. *p < .05, **p < .01, and ***p < .001.

AUTHOR CONTRIBUTIONS

Jian Huang: Data curation. Jin Gan: Formal analysis; visualization. Jian Wang: Methodology; writing—review and editing. Min Zheng: Writing—review and editing. Han Xiao: Conceptualization; data curation; formal analysis.

ACKNOWLEDGMENTS

We acknowledge TCGA and ICGC database for providing their platforms and contributors for uploading their meaningful data sets.

CONFLICT OF INTEREST STATEMENT

The authors declare no competing interests.

DATA AVAILABILITY STATEMENT

We obtained transcriptome data set and clinicopathological data set from TCGA () and ICGC () databases.

References

Sung H, Ferlay J, Siegel RL, et al. Global cancer statistics 2020: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin. 2021;71(3):209‐249. [DOI: https://dx.doi.org/10.3322/caac.21660]