SPP1 is a Prognostic Related Biomarker and Correlated with Tumor-Inltrating Immune Cells in Hepatocellular Carcinoma

Background: Secreted phosphoprotein 1 (SPP1) functions as a tumor promoter in varies tumors, but little is known whether it is an actual player on driving immune inltration in hepatocellular carcinoma. Methods: In this study, we identied the expression of SPP1 by Oncomine, GEPIA and TIMER databases, and assessed SPP1 immumohistochemical staining analysis by The HPA database. We evaluated the clinical outcomes between SPP1 expression and hepatocellular carcinoma patients via Kaplan-Meier Plotter. We also tested the relationship between SPP1 and critical oncogenes by TIMER and GEPIA databases. Then we explored immune inltration analyses using TIMER and TISIDB datasets. In addition, we performed functional enrichment analyses with Metascape and GeneMANIA databases. Results: We found that SPP1 overexpressed in hepatocellular carcinoma tissues and high SPP1 expression was correlated with shorter OS and PFS survivals in hepatocellular carcinoma patients. SPP1 expression is positive correlation with critical oncogenes related stemness associated genes, cell cycle and proliferation, therapeutic resistance, metastasis, and tumor angiogenesis in hepatocellular carcinoma. Importantly, SPP1 expression was positively correlated with inltrating levels of CD4+ T cells, CD8+ T cells, macrophages, neutrophils, and dendritic cells. Furthermore, SPP1 expression showed strong correlations with diverse immune hallmark sets in hepatocellular carcinoma. Notably, functional enrichment analysis suggested that SPP1 strong related with immune response. Conclusions: These ndings imply that SPP1 is correlated with prognosis and immune cell inltrating, offering a new potential immunotherapeutic target in hepatocellular carcinoma. we further investigated the relationship between SPP1 expression and TIL abundance via TIMER database and TISIDB database. We analyzed the relation between SPP1 expression with the levels of inltrating B cells, CD8+ cells, CD4+ cells, macrophages, neutrophils and dendritic cells in hepatocellular carcinoma. We observed that SPP1 was signicantly associated with tumor purity (r=0.177, p=9.73e-04), and the levels of inltrating B cell (r=0.474, p=1.24e-20), CD8+ cells (r=0.284, p=9.30e-08), CD4+ cells (r=0.378, p=3.84e-13),

liver cancer [10][11][12] . SPP1 has been widely implicated in cancer invasion and metastasis, and its expression is associated with poor prognosis in many types of cancer. SPP1 promote cancer invasion and metastasis through angiogenesis, degradation of (ECM), the formation of lamellipodia, and switching phenotypes towards either cancerassociated broblasts or stem cells. In head and neck, lung, colorectal, and breast cancer, SPP1 has been implicated in treatment resistance [13][14][15][16] . SPP1 is secreted by HCC cells and serum SPP1 increases in HCC patients. The plasmatic SPP1 concentration was referred not as a potential biomarker for hepatocellular carcinoma diagnosis, but as an independent prognostic parameter for survival [17][18][19] . Yet, the underlying mechanisms are still unclear and it is urgent to study the depth pro le. SPP1 can regulate the host immune system via upregulating IL-12 and IFNγ in mouse macrophages and NK cells which indicates that SPP1 may act as a potential role in host immunity [20] . SPP1 is upregulated in human gliomaassociated macrophages, and mediate macrophage polarization and facilitate immune escape by upregulating PD-L1 in lung adenocarcinoma [21,22] . It is reported that SPP1 knockdown could regulate M2 macrophage polarization via upregulating insulin-like growth factor 1 and leukemia inhibitory factor [23] . However, the molecular mechanisms of SPP1 by modulating immune in ltration cell and prognosis of hepatocellular carcinoma were still not fully elucidated.
In our present study, we comprehensively analyzed SPP1 expression and correlation with prognostic value of HCC patients in databases including Oncomine, GEPIA, TIMER, HPA and Kaplan-Meier plotter. We found SPP1 expression is positive correlation with critical oncogenes in hepatocellular carcinoma. We demonstrated the association of SPP1 with tumor in ltration immune cells in the HCC microenvironments via TIMER and TISIDB. Moreover, functional enrichment analysis suggested that SPP1 strong related with immune response. Our ndings in this report highlight the vital role of SPP1 in HCC and further offer a probable relationship and underlying mechanisms between SPP1 and tumor-immune interactions.

Assessment of SPP1 expression levels in different cancers
In order to explore the SPP1 expression levels in different cancers and the corresponding normal tissues, three different online databases were used in our study. The Oncomine database showed that the expression of SPP1 signi cantly higher in cancer samples than normal tissues in most datasets (Fig. 1a). The similar results were showed in GEPIA database (Fig. 1b). The expression of SPP1 was distinctly up-regulated in bladder cancer (BLCA), brain and CNS cancer, breast cancer (BRCA), cervical cancer (CESC), colorectal cancer (COAD and READ), esophageal cancer (ESCA), glioblastoma multiforme (GBM), stomach cancer (STAD), head and neck cancer (HNSC), kidney renal papillary cell carcinoma(KIRP), brain Lower Grade Glioma(LGG), liver hepatocellular cancer (LIHC), lung cancer (LUAD and LUSC), lymphoma, melanoma, ovarian cancer (OV), and pancreatic adenocarcinoma PAAD , skin cutaneous melanoma (SKCM), testicular germ cell tumors(TGCT), thyroid carcinoma(THCA), uterine corpus endometrial carcinoma(UCEC) and uterine carsinosarcoma(UCS) than in corresponding normal tissues. We next evaluate the expression of SPP1 between different tumors and matched normal samples in TIMER dataset with RNA-seq data from TCGA (Fig. 1c). The results are mostly consistent with the above two online databases.
Interestingly, SPP1 expression in the metastasis was higher than that in the primary tumor tissue in SKCM. However, the expression levels of SPP1 in some cancers were controversial. SPP1 was signi cantly lower expressed in kidney renal clear cell carcinoma (KIRC), kidney chromophobe (KICH) than in control normal samples in GEPIA and TIMER database. Taken together, these results demonstrated that SPP1 was up-regulated in multiple cancers suggested that SPP1 may play a crucial biological role in tumor progression.
Elevated expression of SPP1 correlated with poor clinical outcomes of hepatocellular carcinoma To clarify the relationship between SPP1 expression and prognosis of patients with hepatocellular carcinoma, we rst evaluated the expression of SPP1 in hepatocellular carcinoma and normal liver tissue using GEPIA database.
The results showed that SPP1 was signi cantly overexpressed in hepatocellular carcinoma (Fig. 2 a). However, we found that there was no relationship between SPP1 expression and tumor stage (Fig. 2b). In addition, we examined the expression of SPP1 using IHC via HPA database. We found that SPP1 existed in both cell membrane and cytoplasm of hepatocellular cancer cell, and about 45.8% (167/365) hepatocellular cancer patients with SPP1 high expression, while the expression of SPP1 was not detected in normal liver tissue ( Fig. 2c and 2d).
We next investigated the correlation between SPP1 expression and clinical outcomes of hepatocellular carcinoma.
We observed that higher SPP1 expression was correlated with overall survival (OS), progression free survival (PFS) and with stage, grade, T stage, vascular invasion, gender, race, alcohol consumption and Hepatitis virus ( Table 1).
Kaplan-Meier analysis showed that the PFS and OS of SPP1 high patients were shorter than those of SPP1 low patients in hepatocellular cancer (HR=1.59, p=0.0017; HR=2.27, p=3.5e-06; Fig. 3a (Fig. 3c-3f). Taken together, all results implied that SPP1 is a potential prognostic factor of hepatocellular cancer. SPP1 expression is positive correlation with critical oncogenes in hepatocellular carcinoma.
It was reported that SPP1 participated in maintaining cancer stem cell phenotype [24] , cell cycle progression [25] , chemoresistance [26] , cancer metastasis [27] , and tumor angiogenesis [28,29] . So we analyzed the association between  Table 2). The similar correlation analysis results in LIHC were found in GEPIA database (Supplementary Fig. 1a-1f). These results implied that SPP1 played an important role in regulating malignant phenotypes in hepatocellular carcinoma.
Emerging reports indicated that SPP1 functions in the tumor microenvironment through regulating macrophages and T cells [34,35] . Based on this, we further investigated the relationship between SPP1 expression and TIL abundance via TIMER database and TISIDB database. We analyzed the relation between SPP1 expression with the levels of in ltrating B cells, CD8+ cells, CD4+ cells, macrophages, neutrophils and dendritic cells in hepatocellular carcinoma. We observed that SPP1 was signi cantly associated with tumor purity (r=0.177, p=9.73e-04), and the levels of in ltrating B cell (r=0.474, p=1.24e-20), CD8+ cells (r=0.284, p=9.30e-08), CD4+ cells (r=0.378, p=3.84e-13), macrophages (r=0.436, p=3.22e-17), neutrophils (r=0.374, p=7.02e-13) and dendritic cells (r=0.453, p=1.38e-18). (Fig.   4a). We then detected the correlation between in ltrating cell and SPP1 expression by Kaplan-Meier plots using TIMER database. However, we didn't observe that in ltrating immune cells were signi cantly related to the prognosis of hepatocellular cancer (Fig. 4b). In addition, we compared the immune cell in ltration levels with different somatic copy number alterations of SPP1 in hepatocellular carcinoma. We found that the in ltration levels of CD4+ T cells was correlated with SPP1 arm-level deletion (Fig. 4c).

Correlation analysis between SPP1 and immune marker expression
To further explore the effects of SPP1 expression on tumor in ltration immune cells, we used TIMER, TISIDB and GEPIA online database. The heatmap of relationship between SPP1 expression and TILs in various cancers was showed in Fig. 5a). We observed that there was a strong correlation between SPP1 expression and abundance of 20 TILs types in hepatocellular carcinoma ( Fig. 5b- These results suggested that SPP1 may participate the regulation of macrophage polarization and DC in ltration.
Taken together, these ndings indicated that SPP1 expression signi cantly correlated with immune microenvironment and may promote tumor immune escape process in hepatocellular carcinoma.

Functional enrichment analysis of SPP1 in patients with hepatocellular carcinoma
To better understand the interplay functions of SPP1 and neighboring genes, we accessed SPP1 networks using GeneMANIA online dataset. The results demonstrated that the extracellular matrix gene FN1, integrin family gene ITGA5, ITGA8, ITGAV, ITGA9, ITGB8, apoptosis genes CASP3, CASP8, extracellular matrix disassembly gene MMP7, integrin-mediated signaling pathway genes MAP3K1, MAP3K14, leukocyte migration genes PDLIM7, SYK and oncogenes BRCA1, RIMS4, ETV4 and DSEL were closely associated with SPP1 (Fig. 6a). Among them, ITGA5 and F2 were found as the top two signi cant hallmarks in the PPI network relating to SPP1. enrichment items were classi ed into three functional groups: 10 items of biological process group, 5 items of molecular function group and 5 items of cellular component group. Consistent with our preceding analysis the results showed strong relationship with immune response. Top enriched ontology clusters of SPP1 and its neighboring genes included immune response-activating signal transduction, immune system process, immuneregulatory interactions between a lymphoid and a non-lymphoid cell, regulation of cell activation. Moreover, all the pathways achieved from the KEGG analysis were related with immune response (Fig. 6b-6e).

Discussion
In view of high prevalence of HBV, liver cancer is the leading incidence cancer in China. And survival outcome of HCC patients is still poor receiving with existing therapies [2] . Therefore, the determination of molecular markers has attracted much attention in the treatment and prognosis of hepatocellular cancer. SPP1 is a secreted arginine glycine aspartic acid containing phosphorylated glycoprotein overexpressed in various malignant neoplasms, and it is involved in various functions, such as in cell adhesion and migration, apoptosis, metastasis, tumor angiogenesis, and initiating cell self-renewal. SPP1 is often overexpressed in multiple cancers including pancreatic cancer, lung cancer, gastric cancer, breast cancer and colon cancer [10,36,37] . SPP1 mRNA and protein increase in HCC. SPP1 is secreted by HCC cells and serum SPP1 increases in HCC patients [17,18] . As the combination of AFP and SPP1 signi cantly improved diagnosis performance when compared with AFP alone, SPP1 was considered as a promising serological biomarker for hepatocellular carcinoma diagnosis. [38] And SPP1 was also referred as an independent prognostic parameter for survival in HCC [19] . Previous study implied SPP1 promotes progression and cancer stem cell-like phenotype in hepatocellular carcinoma cells via the αvβ3-NF-κB-HIF-1α pathway [39] . However the underlying mechanisms are still unclear and need be fully elucidated.
Recently, tumor immunotherapy such as anti-PD-1/PD-L1/CTLA-4 monoclonal antibody has extensively focused as a monotherapy, or an important part of combined therapy. Immunotherapy is fundamentally different from targeted therapy or chemotherapy [40] . Instead of directly targeting cancer cells, it recruits and activates core immune guardian T cells to recognize and eliminate cancer cells through antigen antibody response [41] . Unfortunately, only about 20% patients respond to immunotherapy, especially in hepatocellular carcinoma [42] . Therefore, it is urgent to identify new potential targets for immune-related therapy. It is reported that there is a positive correlation between the SPP1 and PD-L1 expression, and SPP1 expression and TAM in ltration in tumor tissues from patients with HCC [17] .
To gain more detailed insights into the potential immune functions of SPP1 in hepatocellular carcinoma and its regulatory network, we performed the bioinformatics analysis of public data to guide future research in hepatocellular carcinoma.
TIMER database revealed that SPP1 was high expressed in BLCA, BRCA, CHOL, COAD, ESCA, HNSC, KIRP, LIHC, LUAD, LUSC, PRAD, READ, STAD, THCA and UCEC relative to normal tissues, whereas expression was decreased in KIRP and KIRC than in control samples (Fig. 1c). HPA database indicated that SPP1 existed in both cell membrane and cytoplasm, and about 45.8% hepatocellular carcinoma patients with SPP1 high expression, while SPP1 expression was not detected in normal liver tissue (Fig. 2c-2d). Kaplan-Meier plotter database found elevated SPP1 was associated with poor outcomes (Fig. 3a-3f). The above results together imply that SPP1 may have an important value as a prognostic biomarker of hepatocellular carcinoma.
These results referred that SPP1 played an important role in regulating malignant phenotypes in hepatocellular carcinoma.
Immune in ltrating cells in the tumor microenvironment (TME) have been implied to play a vital role in tumor progression and in uence clinical outcomes in cancer patients [45] . In our report, we found that SPP1 expression was correlated with TILs abundance. We demonstrated that SPP1 could recruit CD8+ cells, CD4+ cells, macrophages, neutrophils and dendritic cells in hepatocellular carcinoma (Fig. 4a). However, we didn't observe that in ltrating immune cells were signi cantly related to the prognosis of hepatocellular cancer (Fig. 4b). Then, we found that the in ltration levels of CD4+ T cells was correlated with SPP1 arm-level deletion (Fig. 4c). In addition, we observed that there was a strong correlation between SPP1 expression and abundance of 20 TILs types in hepatocellular carcinoma (Fig. 5a-5g), Supplementary Table 1), and a signi cant positive correlation between SPP1 expression and immunoinhibitors, such as PDCD1 (PD-1), CTLA4, and TIGIT ( Fig. 5h-5n, Supplementary Table 2), suggesting that SPP1 may play an important role in immune escape of hepatocellular carcinoma. Taken together, these ndings indicated that SPP1 expression signi cantly correlated with tumor immune microenvironment and may promote tumor immune escape process. Enrichment analysis of target gene sets can help reveal important networks of transcription factors, target genes and pathway hallmarks. Our study suggested that neighboring gene network of SPP1 was associated with extracellular matrix, integrin, apoptosis, integrin-mediated signaling pathway and leukocyte migration (Fig. 6a). Functional enrichment analysis suggested strong relationship with immune response including immune response-activating signal transduction, immune system process, immune-regulatory interactions. (Fig. 6b-6e). The ndings further highlighted that SPP1 was closely related with immune response. Conclusions SPP1 may be a promising prognostic biomarker and an important regulator of tumor immune cell in ltration for hepatocellular carcinoma patients, offering a new potential immunotherapeutic target in hepatocellular carcinoma.

The Human Protein Atlas database
The SPP1 immumohistochemical (IHC) staining analysis was assessed by The Human Protein Atlas (HPA) database (https://www.proteinatlas.org/) [48] . We evaluated the protein expression in HCC and normal liver tissue, separately.
The SPP1 antibody was HAP027541.

GEPIA database analysis
Gene Expression Pro ling Interactive Analysis (GEPIA) (http://gepia2.cancer-pku.cn/#index) is an online database to analyze the RNA sequencing expression data of 9,736 tumors and 8,587 normal samples from the The Cancer Genome Atlas (TCGA) and the Genotype-Tissue Expression (GTEx) database [49] . We use this dataset to evaluate SPP1 expression levels in hepatocellular carcinoma and we also evaluate SPP1 expression in different tumor stages of HCC.

Kaplan-Meier plotter analysis
The clinical outcomes between SPP1 mRNA level and HCC patients were evaluated with Kaplan-Meier Plotter (www.kmplot.com) [50] . RNA-Seq ID of SPP1 is 6696 in Start KM Plotter for 364 HCC patients with OS data and 370 HCC patients with PFS data. The overall survival (OS) and Progression Free Survival (PFS) of patients with HCC were determined by dividing the patient samples into two groups based on best cutoff (high vs. low expression). P-value 0.05 was considered a statistical signi cance.

Spearman relationship analysis
Using Tumor Immune Estimation Resource (TIMER) database (https://cistrome.shinyapps.io/timer/ ) and GEPIA database, we analyzed Spearman relationship between SPP1 and critical oncogenes related stemness associated genes, cell cycle and proliferation, therapeutic resistance, metastasis, and tumor angiogenesis in hepatocellular carcinoma. Pvalue 0.05 was considered a statistical signi cance.

Immune in ltration analysis
The expression of SPP1 in HCC and the abundances of B cell, CD8+ T cell, CD4+ T cell, macrophage, neutrophil and dendritic cell were evaluated by TIMER database (https://cistrome.shinyapps.io/timer/ ) [51] . TISIDB dataset (http://cis.hku.hk/TISIDB/ ) , a web portal for tumor and immune system interaction, was used to assess the correlations between SPP1 expression and tumor in ltration lymphocytes (TILs) of HCC. Spearman correlations between expression of SPP1 and immunoinhibitors across HCC were also performed by TISIDB dataset [52] .

Functional enrichment analysis
The GeneMANIA project is a biological network integration for gene prioritization and predicting gene function (http://genemania.org/ ). In this study, protein-protein interaction (PPI) network of SPP1 was analyzed with the GeneMANIA [53] .We used the Metascape database (https://metascape.org) to perform Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways analyses of SPP1 [54] . Terms with a p-value < 0.01, a minimum count of 3, and an enrichment factor > 1.5 are collected and grouped into clusters based on their membership similarities.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.