SAMD13 serves as a useful prognostic biomarker for hepatocellular carcinoma

UALCAN database is an open-access web for analyzing cancer OMICS data based on tumors and normal samples from TCGA, MET500, CPTAC, and CBTTC [14]. Here, the UALCAN was used to compare the expression levels of SAMD13 in pan-cancer, including HCC and clinical data, as well as the associations between them. GEPIA and GEPIA2 is a web tool for gene expression profiling and interactive analyses based on TCGA and The Genotype-Tissue Expression (GTEx) data [15]. This database was used to access overall survival (OS) and disease-free survival (DFS) based on the expression of SAMD13, and the further association between this gene and the expression of immune-related marker genes was verified.

The TCGA–LIHC (HCC) data set was used as the training set, while three other independent data sets (GSE22058, GSE25097, and GSE45436) and Gene Expression database of Normal and Tumor tissues 2 (GENT2) (which is web-based tool from public gene expression data sets) were used as validation sets. Gene expression profiling data sets, including GSE22058 (97 non-tumor and 100 tumor), GSE25097 (6 normal, 40 cirrhosis, and 268 tumor), and GSE45436 (39 non-tumor and 95 tumor), and chemo-resistant gene profiling data sets, including GSE54175 (cisplatin- or doxorubicin-resistant subline), GSE93595 (anti-angiogenic therapy (JNJ-28841072)-resistant subline), GSE121153 (sorafenib-resistant subline), GSE125180 (doxorubicin-resistant subline), GSE109211 (clinical parameters from responders and non-responders to sorafenib in patients with HCC) were downloaded from the gene expression omnibus (GEO, https://​www.​ncbi.​nlm.​nih.​gov/​geo/​). cBioportal for Cancer Genomics database which analyzes multi-omics data from The Cancer Genome Atlas was used for analyzing genetic alteration for SAMD13 gene in a liver study (TCGA, PanCancer Atlas) [16]. The KM–plotter database was used for survival curves in various cancers [17]. Hazard ratios (HR) and p values (from the log-rank test) were calculated online.

Human Liver Browser and Single-cell Atlas in Liver Cancer (scAtlasLC) were used to characterize SAMD13 expression of single-cell transcriptomic profiles in HCC [18, 19]. The Tumor Immune Estimation Resource database (TIMER 2.0) used analysis and visualization of immune infiltrates from TCGA in TIMER Database [20]. In addition to exploring the associations of HCC-infiltrating immune cells with SAMD13 gene expression, the degree of immune infiltration of B cell, CD8 + T cell, CD4 + T cell, macrophages, neutrophil, and dendritic cell (DC) was analyzed in TIMER database. The strength of correlations evaluated by the purity-adjusted partial Spearman’s rho value and estimated statistical significance in TIMER 2.0.

MethSurv database could provide the initial assessment of DNA methylation biomarkers using TCGA [21]. In this study, single CpG methylation of SAMD13 and Heatmap analysis were performed in patients with HCC. Next, Shiny Methylation Analysis Resource Tool (SMART) [22] was used to analyze differential methylation by each SAMD13 probe and Spearman’s correlation between methylation level and mRNA level. The CpG-aggregated methylation values were measured by mean (β values). To characterize the methylation patterns, the significant CpGs were classified according to their functional roles in genomic locations, such as promoters within 1,500 bps of a transcription start site (TSS) (TSS1500); within 200 bps of a TSS (TSS200); 5’ untranslated regions (5'UTR); first exon (1stExon); body (non-promoter); 3'UTR (non-promoter).

The 22 genes with the strongest correlation with SAMD13 were selected using Pathway Commons [23] and this allowed the creation of a protein–protein interaction (PPI) network for SAMD13 genes as well as binding and targeted genes. To determine the functional and pathway meaning, cellular component (CC), molecular function (MF), biological process (BP), and reactome pathway (RP) for 22 selected genes forming a cluster with SAMD13 was performed using geneontology.org [24]. Functional enrichment analyses of Gene Ontology (GO) terms were performed using Fisher's Exact test and adjusted by a False Discovery Rate (FDR) correction for multiple testing in pantherdb.org (http://​pantherdb.​org/​).

SAMD13 protein expression levels in normal and HCC tissues were reviewed in the Human Tissue Atlas which is an online tool for genome-wide analysis (http://​www.​proteinatlas.​org/​) [25].

Huh7, HepG2, SK-Hep1 cells were purchased from the American Type Culture Collection. SNU475 and SNU449 cells were Korean Cell Line Bank (Seoul, Republic of Korea). Huh7, HepG2, SK-Hep1 Cells were maintained in Dulbecco’s modified Eagle’s medium containing 10% FBS (GIBCO, Grand Island, NY, USA).SNU475 and SNU449 cells were maintained in RPMI1640 medium containing 10% FBS in a humidified incubator with 5% CO₂ at 37 °C. HepaRG cells were cultured in William’s E medium supplemented with 10% of FBS, 5 µg/mL insulin, 2‐mM Glutamax, 1% penicillin–streptomycin, and 50 µM hydrocortisone hemisuccinate (Sigma-Aldrich).

Total RNA was isolated using a TRIzol reagent-based kit (Intron Biotech, Seongnam-Si, Republic of Korea). Reverse transcription was performed using the SuperScript IV First-Strand Synthesis System for RT-PCR (Thermo Fisher Scientific, Waltham, MA, USA) according to the manufacturer’s protocol. cDNA was amplified with specific primers and SYBR Premix Ex Taq (Takara Bio, Otsu, Shiga, Japan and Agilent Technologies, Santa Clara, CA, USA). The SAMD13 mRNA levels were normalized to the amount of GAPDH mRNA. qPCR was performed according to the manufacturer’s instructions (Applied Biosystems, Foster City, CA, USA, and Agilent Technologies). Relative quantification of gene expression was performed using the 2-ΔΔCT method. The primer pair used for SAMD13 (forward: 5’-CAAGGAAAATGGCTCTGTCGGTG-3’; reverse: 5’-AGCTTGCTCCTCAAATCCCACG-3’) and GAPDH (forward: 5’-AACGGGAAGCTTGTCATCAATGGAAA-3’; reverse: 5’- GCATCAGCAGAGGGGGCAGAG-3’).

The statistical analysis was calculated automatically based on the online database above. Student's t test implemented by GraphPad Prism (Ver7).Correlations were analyzed by Spearman and Pearson’s correlation. p < 0.05 was considered statistically significant.

To explore the expression pattern of SAMD13between tumor and normal tissues in various types of cancer, we examined the SAMD13 differences using the UALCAN database. The comparison of expression level between each type of normal and cancer cells revealed that SAMD13 increased in breast invasive carcinoma, cholangiocarcinoma, glioblastoma multiforme, kidney renal papillary cell carcinoma, liver hepatocellular carcinoma, lung adenocarcinoma, prostate adenocarcinoma, and stomach adenocarcinoma, while decreased in colon adenocarcinoma, head and neck squamous cell carcinoma, kidney chromophobe, rectum adenocarcinoma, thyroid carcinoma, and uterine corpus endometrial carcinoma (Fig. 1A). To validate these findings, we further examined the expression levels of SAMD13 using the GENT2 portal. As shown in Additional file 1: Figure S1, SAMD13 was upregulated in blood, brain, liver, and bone cancers, while downregulated in colon, breast, kidney, head and neck, tongue, thyroid, adrenal gland, pharynx, testis, and endometrial cancers. To assess the prognostic significance of SAMD13 expression in various types of cancers, log-rank test was performed using pan-cancer RNA-seq in KM-plotter. Among these 21 types of cancer, four were significantly associated with the prognosis of all cancers (Fig. 1B). High mRNA expression level of SAMD13 were observed to be significantly associated with worse overall survival in lung adenocarcinoma (HR, 1.51; 95% CI 1.15–2.07; p = 0.004) and hepatocellular carcinoma (HR, 2.15; 95% CI 1.15–3.06; p < 0.001), whereas thymoma (HR, 0.23; 95% CI 0.05–1.12; p = 0.048) and kidney renal clear cell carcinoma (HR, 0.011; 95% CI 0.49–0.91; p = 0.011) were significantly associated better prognosis. In addition, these observations were validated by GEPIA2 in pan-cancer database. As shown in Fig. 1C, high SAMD13 expression is associated with a poorer overall survival LIHC and LUAD, while low expression of SAMD13 indicated better prognosis in THYM. When analyzing disease-free survival, the high expression of SAMD13 is associated with a worse prognosis in LGG, LICH, and LUAD, while the low expression of SAMD13 indicated better prognosis in COAD, GBM, KIRC, and THCA.

Since SAMD13 gene expression was not only upregulated in various carcinomas but also linked to worst prognosis in HCC, we further performed validation on additional three independent GEO data sets. As expected, these GEO data sets also revealed the significant augmentation of SAMD13 mRNA expression in the tumor group compared to their normal/non-tumor group. Interestingly, increased mRNA expression of SAMD13 was also revealed in the cirrhosis group at GSE25097 (Fig. 2A). Next, this trend was evaluated at the protein level between normal liver and HCC tissues, and as a result, the immunohistochemical data show that SAMD13 expression was higher than that compared to normal tissue (Fig. 2B).To further validate the differential expression of SAMD13, mRNA expression of SAMD13 was conducted on 1 immortalized hepatic cell line (HepaRG), 2 well-differentiated HCC cell lines (Huh7 andHepG2), and 3 poorly differentiated HCC cell lines (SK-Hep-1, SNU475, and SNU449). As shown Fig. 3C, the SAMD13 expression were increased in poorly differentiated HCC cells relative to well-differentiated HCC cells suggesting that SAMD13 may be acted HCC tumors to more aggressive tumor phenotypes. To better understand the relationship between SAMD13 expression and prognosis in HCC, we investigated the correlation between SAMD13 expression and overall survival (OS), disease specific survival (DSS), disease free interval (DFI), and progression free interval (PFI). As shown in Fig. 2D, high expression of SAMD13 was significantly associated with shorter OS (log-rank test = 11.6; p = 0.0006605), DSS (log-rank test = 7.32; p = 0.006818), and PFI (log-rank test = 6.127; p = 0.01331); meanwhile, DFI was correlated with shorter tendency (log-rank test = 3.343; p = 0.0675).

We then determined a relationship between clinicopathological features and SAMD13 expression using TCGA data through UALCAN in HCC, and summarized in Table1. According to tumor stage, SAMD13 expression were significantly upregulated in HCC group classified as stage I to stage III compared to the corresponding normal group. Based on tumor grade, a significant increase in SAMD13 levels was observed in the HCC group in all tumor grades. In addition, upregulation of SAMD13 expression was observed in patients with HCC and fibrolamellarcarcinoma types. In addition, SAMD13 expression was statistically significant in all other categories of patients with HCC, including gender, age, and race. These results suggest that SAMD13 was significantly correlated with clinical parameters in patients with HCC.

To further assess the effect of SAMD13 on the TME, we investigated the association between SAMD13 and expression profiling of TME in HCC using the Human Liver Browser and scAtlasLC data sets. It was revealed that SAMD13 was mainly expressed in scar-associated macrophages (SAMs), T cells, tumor-associated macrophages (TAMs), tumor-associated endothelial cells (TECs), and carcinoma cells, while moderately expressed in hepatocytes and cholangiocytes (Fig. 3A, B). Then, we further analyzed between SAMD13 and immune infiltrates in HCC using TIMER database. The results showed that the expression of SAMD13 was significantly correlated with B cell (r = 0.159, p = 3.19e − 03),CD8 + T cells (r = 0.106, p = 5.03e-02), CD4 + T cells (r = 0.282, p = 1.02e − 07), macrophage (r = 0.291, p = 4.32e-08), neutrophils (r = 0.148, p = 5.93e − 03), and dendritic cells (r = 0.249, p = 3.32e-06)in HCC (Fig. 3C). To study the relationship between SAMD13 and infiltrated immune cells, TIMER and GEPIA database were use. Immune gene markers of B cell, T cell (general), CD8 + cell, CD4 + cell, tumor-associated macrophage (TAM), M1 and M2 macrophage, neutrophil, natural killer cell (NK) cell, DC, Th1/2/fh/17cell, Treg, and T cell exhaustion in HCC were assessed. The results revealed that SAMD13 was a significant positive relationship including all immune infiltrates, including B cell (CD17,CD79A), T cell general (CD3D, CD3E, CD2), CD8 + T cell (CD8A, CD8B), CD4 + T cell (CD4), TAM (CD68, IL10), M1 macrophage (IRF5, PTGS2), M2 macrophage (CD163, VSIG4, MS4A4A), neutrophil (ITGAM,CCR7), DC (HLA-DPB1, HLA-DRA, HLA-DPA1, CD1C, NRP1, ITGAX), Th1 (STAT4, STAT1, IFNG, TNF), Th2 (GATA3, STAT6, STAT5A), Th17 (STAT3), Treg (CCR8, TGFB1, STAT5B), and T cell exhaustion (PDCD1, CTLA4, HAVCR2). However, NK cell and Tfh was not significantly associated between SAMD13 and immune infiltrates (Table 2). In addition, the comprehensive prognostic value analysis was performed on SAMD13 expression and immune cell infiltration in HCC. The results of Fig. 3D showed that low expression of SAMD13 along with low immune infiltration of neutrophil, macrophage, macrophage M0, macrophage M2, and myeloid-derived suppressor cells (MDSC) had better prognosis than the high expression of SAMD13 group in HCC. However, there were no significantly changed inCD8 + T cells, CD4 + T cells, B cell, DC, and NK cells (data not shown). These results indicated that SAMD13 is not only closely related to immune cell infiltration, but also correlated with prognosis in HCC.

To prove into the role of SAMD13 methylation in HCC, we studied the correlation between SAMD13 methylation state and its expression levels. Based on SMART database, SAMD13 methylation level is significantly lower in HCC tissues compared to normal liver tissues (Fig. 4A) and specific methylation sites were presented in heatmaps using MethSurv database (Fig. 4B). According to the heatmap, there were 15 CpG sites in SAMD13 and we found that average methylation of all CpG sitesof SAMD13 (Overall; Aggregation, p = 2.4e-05), including N_Shore (Aggregation, p = 0.00048), S_Shore(Aggregation, p = 0.058), S_Shelf (Aggregation, p = 0.039), and Open_Sea (Aggregation, p = 6.3e-14) was significantly lower in tumor tissues than in the normal counterpart (Fig. 4C). At the same time, we performed survival analyses using methylation profiling of SAMD13 CpGs in HCC. The results indicated that CpGs of SAMD13 gene containing cg15103960 (HR, 0.71; 95% CI, 0.51-1; p = 0.052), cg02041547 (HR, 0.46; 95% CI, 0.32-0.64; p < 0.001), and cg23086720 (HR, 0.51; 95% CI, 0.36-0.72; p < 0.001) was associated with better prognosis, while cg23694882 (HR, 1.52; 95% CI, 1.08-2.14; p = 0.016), cg15089272 (HR, 1.5; 95% CI, 1.07-2.12; p = 0.019), and cg23925111 (HR, 1.62; 95% CI, 1.15-2.27; p = 0.006) showed opposite (Table 3). We also found that cg20426713, cg11555919, cg23694882, cg13305246, cg22529952, and cg22529952 probes were significantly lower in tumor tissues than in the normal counterpart, while cg23086720 probe showed higher in tumor compared to normal control (Additional file 1: Figure S2). Given the close relationship between the methylation of SAMD13 and SAMD13 expression, we found a positive correlation between the methylation level and the expression of SAMD13 (Overall; Aggregation, p = 1.1e-05), including N_Shore (Aggregation, p = 0.012), S_Shore (Aggregation, p = 0.00023), and Open_Sea (Aggregation, p = 2.8e-09) (Fig. 4D and Additional file 1: Figure S3). Moreover, we examined the copy number alterations and revealed that SAMD13 mRNA had higher copy number gain compare to diploid or shallow deletion (Fig. 5A). Interestingly, later tumor stages including T3, T3A, T3B, and T4 showed a relatively high tendency for deletion or gain of copy number alterations, although absolute copy number alterations were decreased (Fig. 5B). Since the relationship between copy number amplification and individual DNA methylation patterns affects gene regulation, we next explored the relationship between methylation value and copy number amplification in SAMD13. Interestingly, we found that the average methylation value of CpGs of SAMD13 was significantly negative correlation with copy number amplification in HCC (Aggregation, p = 2.3e-05) (Fig. 5C). In methylation value on clinical relevance, we found that cg23694882 was much hyper-methylated in stage-dependent manner (p = 0.0022), while cg23086720 showed hypo-methylated in stage dependent manner (p = 0.00073). These results indicated that methylation value and copy number alterations of SAMD13 gene plays a crucial role in the prognosis that might reflect the complexity/heterogeneity with HCC patients in SAMD13.

To further investigate the biological role in SAMD13, genes were found to be interacted with SAMD13 based on Pathway Commons and STRING (Fig. 6A, B), and list of genes was presented in Table 4 and 5, respectively. The functions of SAMD13 and interacted genes were classified into each functional group: molecular function, biological process, reactome pathway, and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway. Among functional and pathway enrichment analysis, SAMD13 was significantly associated with DNA binding on molecular function, including “chromatin binding”, “histone binding”, “methyl-CpG binding”, “general transcription initiation factor binding”; regulation of nucleic acid metabolic process on biological process, including “metaphase/anaphase transition of mitotic cell cycle”, “histone ubiquitination and/or methylation”, “regulation of transcription involved in G1/S transition of mitotic cell cycle”, “hepatocyte apoptotic process”; cell cycle regulation on reactome pathway, including “inactivation of APC/C via direct inhibition of the APC/C complex”, “aberrant regulation of mitotic exit in cancer due to RB1 defects”, “G1 phase”, “oncogene induced senescence”; proteolysis on KEGG pathway, including “ubiquitin mediated proteolysis” and “cell cycle” (Additional file 2: Table S1 and Additional file 3: Table S2). To further explore the potential network co-expressed with SAMD13 gene in HCC, the GEPIA2 database was employed to evaluate the relationship between SAMD13 and co-expressed genes. In the TCGA-HCC dataset, 13 genes were identified with significant correlation with SAMD13 from the tumor group only and a total of six genes including FOXM1, JUN, JARID2, BRE, BUB1B, and PHC2 were shown to significantly predict poor overall survival in log-rank tests in HCC (Fig. 6C). According to the high/low expression groups among these genes combined with SAMD13, we divided the TCGA-HCC samples into four combinations based on the median of gene expression for all six pairs of genes. In survival analysis of the six pairs of gene, the group with high expression of SAMD13 and FOXM1, JUN, JARID2, BRE, BUB1B, or PHC2 had the poorest prognosis, while the groups of low expression of SAMD13 with low expression of all six genes had the best prognosis, respectively (Fig. 6D).

Since the biological role of SAMD13 enriches the “cell cycle” and “nucleic acid metabolic process”, we further examined the chemosensitivity to cisplatin, doxorubicin, sorafenib, and JNJ-28841072 which have been shown to potently inhibit the cell cycle used for the treatment of HCC from three independent GEO data set, including GSE54175 (identification of chemo-resistant genes in human metastatic HCC), GSE121153 (sorafenib-resistant HCC), GSE125180 (doxorubicin-resistant HCC), and GSE93595 (anti-angiogenic drug resistant HCC). As shown Fig. 7A, SAMD13 mRNA level was elevated in almost all cell lines with conventional chemotherapy, based on analyses from the GEO databases. Typically, SAMD13 was positively correlated with the elevation of gene expression for cisplatin-resistant MHCC97L cells (GSE54175), sorafenib-resistant Huh7 cells (GSE121153), doxorubicin-resistant 834 cells (GSE125180), and JNJ-28841072-resistant Huh7 cells which had acquired resistance under long-term anti-angiogenic therapy from serial transplantation in immunocompromised mice (GSE93595). To further verify the clinical significance of SAMD13 expression for assessing response of chemotherapy, drug-cancer response data set was investigated. One cohort (GSE109211) treated with placebo or sorafenib was retrieved, which contained 140 patients randomized to receive 98 non-responders and 42 responders. Compared to placebo- or sorafenib-treatment, there was no significant differences between responders and non-responders (Fig. 7B), as well as recurrence of HCC (Additional file 1: Figure S4). Since the role of SAMD13 in ICB treatment of HCC has not been reported, we further assessed the TCGA–LIHC data to explore the correlation between SAMD13 and some ICB related genes. The results demonstrated that SAMD13 was significantly positive correlation with ICB genes, including CD276, CD86, ICOS, HAVCR2, LAIR1, CD44, TIGIT, CD80, CTLA4 (Fig. 7C). Collectively, these results suggest that SAMD13 expression might not be involved in chemotherapy response to cancer treatment, but rather a target for ICB treatment that can counter acquired chemical resistance.