VGLL3 is a prognostic biomarker and correlated with clinical pathologic features and immune infiltrates in stomach adenocarcinoma

Due to its poor clinical outcome, there is an urgent need to identify novel prognostic markers for stomach adenocarcinoma (STAD). Here, we aimed to explore the relationship between VGLL3 expression and clinico-pathological features, dendritic cells, macrophages, and prognosis of STAD. VGLL3 expression levels were significantly associated with histological grade, T stage, and TNM stage. VGLL3 levels and patient’s age were also independent prognostic factors of the clinical outcome of STAD. In addition, VGLL3 was associated with the abundance of macrophages and dendritic cells in tumor infiltrates, of which only VGLL3 and macrophage counts were the independent prognostic factors of immune cell infiltration in the TIMER Database. Extracellular matrix receptor interaction, focal adhesion, pathways in cancer, MAPK, JAK STAT, and WNT signaling pathways were enriched in VGLL3 high-expressing datasets as determined by Gene Set Enrichment Analysis (GSEA), while DNA replication, glyoxylate, and dicarboxylate metabolism, glutathione metabolism, homologous recombination, and glycosylphosphatidylinositol gpi banchor biosynthesis were enriched in VGLL3 low-expressing datasets. Thus, VGLL3 is a novel prognostic biomarker of both the clinical outcome and immune infiltration in STAD, and may therefore be a promising therapeutic target.

cell infiltration, tumor-associated macrophages (TAM) was widely studied in cancers including gastric cancer [18][19][20] . Therefore, we hypothesized that VGLL3 expression was associated with immune cell infiltration and tumor-associated macrophages.
To determine any potential correlation of VGLL3 expression with immune cell infiltration of STAD, the Spearman's correlation test was used to evaluate this relationship using TIMER databases. In addition, Cox regression, Kaplan-Meier survival analysis, and the log-rank test were used to determine the significance of VGLL3 expression on the prognosis of STAD with immune cell infiltration.
Furthermore, the potential underlying mechanism of VGLL3 in STAD was explored by using Gene Set Enrichment Analysis (GSEA). The gene sets were classified into VGLL3 high (VGLL3 hi ) and low expression (VGLL3 lo ) groups based on the median expression level. Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways were significantly enriched by the criterium of false discovery rate (FDR) <0.05.

Results
Patient characteristics. Tumor samples were obtained from 317 patients who were pathologically diagnosed with STAD and were aged 62. 16 Table 1. VGLL3 was associated with poor differentiation, advanced T stage, and TNM stage. To explore the correlationof VGLL3 on the progression of STAD, the Wilcox test and Kruskal test method were employed to analyze the difference in VGLL3 expression with clinico-pathological features. Our data showed that VGLL3 expression significantly correlated with the histological grade (P = 1.437e-05, Fig. 1A), T stage (P = 1.462e-05, Fig. 1B), and TNM stage (P = 0.004, Fig. 1C), but not with age, gender, N stage, and M stage (P > 0.05, Fig. 1D-G) of STAD. Taken together, these results showed that VGLL3 was highly expressed in poor differentiation (Fig. 1A), advanced T stage (Fig. 1B), and advanced TNM stage (Fig. 1C). Thus, these findings suggested that VGLL3 was www.nature.com/scientificreports www.nature.com/scientificreports/ involved in the progression of STAD. Considering that a mutation might result in loss of function, we mined the STAD mutation based on cBioPortal (www.cbioportal.org). There were only 8 missense mutation and 1 del mutation, which might result in loss of function in the 478 STAD samples (Fig. 1I). From the mutation status, no significant differences were observed in VGLL3 transcription expression between missense status and wild type status of VGLL3 (Fig. 1H). Thus, these data suggested that VGLL3 mutations were not associated with its expression.
VGLL3 was an independent unfavorable prognostic marker with clinical features. To explore the correlation of VGLL3 expression with the prognosis of STAD, the Cox proportional hazard regression model was employed to analyze prognostic factors. For the overall survival analysis, we used Kaplan-Meier survival analysis and log-rank test methods. Univariate Cox regression analysis showed that age, TNM stage, M stage, N stage, and VGLL3 levels were significantly associated with STAD prognosis (P < 0.05), however, gender and grade were not (Table 2). Multivariate Cox regression analysis indicated that only VGLL3 and age were independent prognostic factors (P < 0.01, Fig. 2A). In addition, Kaplan-Meier survival analysis showed that VGLL3 in the high expression group had a significantly poorer prognosis in the TCGA cohort (P < 0.05, Fig. 2B), which was validated in other cohorts in GEO (Gene Expression Omnibus) usingKaplan-Meier Plotter online data analysis P < 0.05, Fig. 2C).
Taken together, these results demonstrated that VGLL3 was an unfavorable prognostic factor and independent prognostic marker.
VGLL3 is an independent unfavorable prognostic marker with immune cell infiltration. To explore immune infiltrates and VGLL3 expression on the prognosis of STAD, Kaplan-Meier survival analysis and the log-rank test method were employed to analyze the overall survival. Both immune infiltrates and VGLL3 expression were divided into VGLL3 high and low levels by using the median expression. The Log-rank test and Kaplan-Meier survival analysis showed that high levels of VGLL3 (Fig. 4A) and macrophages (Fig. 4B) were significantly associated with poor survival (P < 0.05), whereas tumor purity, B cells, CD8 + T cells, CD4 + T cells, neutrophils, and dendritic cells were not (P > 0.05, data not shown).
VGLL3 is associated with the PI3K/Akt/m TOR signaling pathway, extracellular matrix receptor interaction and focal adhesion. To explore the potential underlying mechanism of VGLL3 in the progression of STAD, VGLL3 expression was first divided into high (VGLL3 hi ) and low (VGLL3 lo ) expression groups as determined by the median expression. Next, the gene set enrichment analysis (GSEA) method was employed to enrich Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways. The gene sets associated with high VGLL3 expression were enriched in extracellular matrix (ECM) receptor interaction, focal adhesion, pathways in cancer, MAPK signaling pathway, JAK STAT signaling pathway, ABC transporters, and the WNT signaling pathway. In contrast, VGLL3 low expression-related gene sets were enriched in glyoxylate and dicarboxylate metabolism, DNA replication, glutathione metabolism, homologous recombination and glycosylphosphatidylinositol gpi banchor biosynthesis (Fig. 5A,B). To further investigate the molecular mechanism underlying VGLL3 expression, a heat map of the top 31 genes was created, which showed a strong correlation with VGLL3 expression based on r > 0.7 (Fig. 5C). Protein-protein interaction (PPI) network analysis between VGLL3 expression and co-expressed genes was determined by building a regulatory network using Cytoscape software, and was analyzed using STRING (https://string-db.org/). The crosstalk between genes (Fig. 5D) further showed a significant enrichment of the PI3K/Akt/m TOR signaling pathway, ECM receptor interaction and focal adhesion (Fig. 5E). Together, these results demonstrated that VGLL3 was involved in cancer-related KEGG pathways.
It might be that VGLL3 plays a vital role in the progression of STAD by regulating the PI3K/Akt/m TOR signaling pathway, ECM receptor interaction, and focal adhesion.

Discussion
VGLL3 is an inhibitor of adipocyte differentiation and a novel Ets1 interacting partner, and regulates trigeminal nerve formation and cranial neural crest migration 7,22 . In cancer, VGLL3 has been shown to be overexpressed in a subset of soft tissue sarcomas 8 , and was associated with a tumor suppressor phenotype in epithelial ovarian cancer 6 . When studying the association of VGLL3 and the progression of STAD, we analyzed VGLL3 expression levels in STAD samples from the TCGA dataset, and found that high levels of VGLL3 positively correlated with a higher histological grade, T stage, and TNM stage of the tumor, which are indicators of greater tumor malignancy 23 . In addition, VGLL3 overexpression was also associated with worse prognosis and poor overall survival. Moreover, multivariate Cox regression analysis confirmed VGLL3 as an independent unfavorable prognostic factor of STAD. Together, these results demonstrated that VGLL3 might promote theprogression of STAD 6 . In contrast, VGLL3 acts as a tumor suppressor in epithelial ovarian cancer. Thus, an effective marker to predict www.nature.com/scientificreports www.nature.com/scientificreports/ prognosis in cancer patients is urgently needed to optimize tailored treatment in precision medicine. This will aid in individualized treatments, postoperative counseling, and result in improved survival.
STAD is a highly heterogeneous cancer, and tumor heterogeneity is instrumental in immune evasion [24][25][26][27][28] . In fact, several studies have shown that tumor infiltrating immune cells are prognostic markers for cancer progression [29][30][31] . Therefore, we hypothesized that VGLL3 expression is associated with immune infiltration in STAD tumors, and a potential marker of the tumor immune microenvironment. CIBERSORT is an analytical tool developed to estimate the abundance of member cell types in a mixed cell population 32 . Pearson correlation analysis has been performed between genes and immune cell infiltration determined by both CIBERSORT 33 and TIMER. However, for this analysis we used TIMER because of its easy to use web interfaceand its abundant incorporatation of clinical data. In our study, we found that VGLL3 expression positively correlated with tumor heterogeneity regarding immune cell infiltration, particularly with CD8+ T cells, CD4+ T cells, macrophages, neutrophils, and dendritic cells, but not B cells. This suggested that VGLL3 may activate or inhibit immune cells infiltrating the tumor, although the underlying mechanism involved is unknown. Univariate analysis showed that T cells and dendritic cells were favorable prognostic factors, while VGLL3 levels, neutrophils, B cells, macrophages, and tumor heterogeneity were detrimental prognostic factors. Therefore, it is possible that VGLL3 inhibits T cells and dendritic cells, and activates neutrophils and macrophages. Multivariate analysis showed that along with VGLL3 levels, macrophages were also an independent prognostic factor. Tumor-associated macrophages (TAM) affect several aspects of tumor physiology, including tumor cell proliferation 34 , angiogenesis 35 , invasion 36 , metastasis 37 , immunosuppression 38 , and drug resistance 39 . In a previous study, TAM were reported as a biomarker for gastric cancer 18,40 , and TAM were strongly associated with VGLL3 expression in the current study. This suggested that immunotherapy might be considered based on VGLL3 expression.
To further elucidate the molecular mechanisms underlying the role of VGLL3 in STAD, we searched for pathways enriched in VGLL3 overexpressing datasets using GSEA. We found that high VGLL3 levels significantly associated with ECM receptor interaction, focal adhesion, pathways in cancer, MAPK signaling pathway, JAK STAT signaling pathway, ABC transporters, and the WNT signaling pathway, and that low expression levels associated with glyoxylate and dicarboxylate metabolism, DNA replication, glutathione metabolism, homologous recombination, and glycosylphosphatidylinositol gpi banchor biosynthesis. Given that VGLL3 is a coactivator of mammalian TEFs, it likely regulates more genes and pathways indirectly. Although there were several gene sets www.nature.com/scientificreports www.nature.com/scientificreports/ in the GSEA, here we focused on the KEGG pathway gene set to elucidate the molecular mechanism of VGLL3, which might affect one or several KEGG signaling pathways. Therefore, we next analyzed the pathways enriched in genes that strongly correlated with VGLL3 expression (r > 0.7), and identified the PI3K/Akt/mTOR signaling pathway, ECM receptor interaction, and focal adhesion pathway. The PI3K/AKT/mTOR signaling pathway plays a significant role in tumor cell proliferation, growth and survival, in addition to modulating the tumor microenvironment and tumor-associated macrophages (TAMs) [41][42][43][44][45][46][47] . ECM and focal adhesion molecules, including PLXDC2, PDGFRB, FSTL1, TIMP2, FAP, SPARC, AEBP1, NOX4, and FBLN2 are known to regulate macrophage mobilization into tumor tissues 19,20,[48][49][50] , and are associated with TAM 51-53 . Therefore, it is possible that VGLL3 also modulates TAM via at least some of those genes.
Considering that mutations might result in loss of function, we mined the STAD mutations using the cBi-oPortal (www.cbioportal.org). There were 8 missense mutations and 1 del mutation, which might result in loss of function. The VGLL3 associated mutation rate was 1.8% in 478 samples with STAD. Due to the low mutation rate, the effect of the VGLL3 mutation status could be neglected.
To summarize, VGLL3 has a potential prognostic and therapeutic value in STAD, because its overexpression was associated with advanced tumor stage, poor differentiation, TAM infiltration, and poor prognosis. The enriched pathways and co-expressing genes identified in this study provide further insights into the molecular basis of the role of VGLL3 in STAD. Overall, our findings show that VGLL3 plays a significant role in the prognosis and clinical progression of STAD, as well as in immune cell infiltration. However, since the TCGA database contains mRNA expression data and not in situ protein expression data, our findings will have to be validated in patient's tissue samples using immunohistochemistry, and correlated to their clinico-pathological features.

Materials and Methods
RNA-sequencing data and samples. The RNA-sequencing V2 datasets and clinical data of STAD samples were downloaded using the Bioconductor/TCGA Biolinks function package from the TCGA database (http:// tcga-data.nci.nih.gov./tcga/) 54 .