Introduction

Globally, clear cell renal cell carcinoma (ccRCC) is one of the most common pathological types of renal cell carcinoma (RCC), accounting for about 75% of RCCs [1]. Although ccRCC can be detected in its early stages and surgical resection is the mainstream strategy for its mitigation, up to one-third of patients with ccRCC will still develop metastases [2]. ccRCC is a highly immune-invasive tumor. Immunotherapy-based combination therapies have now become the standard of care, so the use of immune checkpoint inhibitors (ICIs) to mitigate tumor-associated suppressed anti-cancer immune responses, such as programmed cell death-1 (PD-1) and cytotoxic T lymphocyte-associated antigen-4 (CTLA-4), has revolutionized the treatment of advanced ccRCC and significantly improved survival status in patients with this disease [3, 4]. Despite advances in treatment, patients with advanced ccRCC still have a poor prognosis, mainly due to drug resistance [5]. This response often relies on the dynamic interaction between the tumor and the tumor microenvironment (TME) [6]. However, the molecular mechanisms and pathological processes involved in cancer-causing immunity of ccRCC are unknown.

The TME, which includes tumor cells, local stromal cells, and other immune cells [7], is crucial in tumor progression and metastasis. Moreover, most non-tumor cell populations, such as stromal and immune cells, have been identified as prognostic markers of tumorigenesis. Previous studies have demonstrated that inflammation in TME, in which infiltrated diverse inflammatory cells contribute to tumorigenesis, is generally considered as a hallmark of cancer [8, 9]. Cancer immunotherapy mainly exerts effects with some proteins to enhance the function of or restore immune cells in the TME. Considering the importance of immunity in the development and the treatment of cancer, discovering novel ccRCC immune-related biomarkers may be useful for immunotherapy.

TREM-1 is involved in triggering DAP12 association and Src family kinase-mediated phosphorylation of ITAM tyrosine on cell membranes [10,11,12], followed by recruitment and phosphorylation of GRBP-2, Syk, and ZAP70, thereby initiating downstream signal transduction [10, 11, 13]. These pathways induce the mobilization of Ca2+, actin cytoskeletal rearrangement, and activation of transcription factors such as Elk1 (ETS domain protein), NFAT (nuclear factor that activates T cells), AP1, c-Fos, c-Jun, and Nf-κB, which transcribe genes encoding pro-inflammatory cytokines, chemokines, and cell surface molecules [14]. Recent studies have shown a close link between TREM-1 and cancer, as TREM-1 is selectively expressed on tumor-associated macrophages (TAMs) in non-small cell lung cancer [15], activates in the liver Kupffer cells, and promotes the progression of hepatocellular carcinoma [16]. Moreover, the increased abundance of TREM-1 in neutrophils may promote the development of colorectal cancer [17]. Nevertheless, the association of TREM-1 with immune infiltration in ccRCC has not been investigated.

In this paper, we identified the prognostic immune-related genes (IRGs) and immune-related lncRNAs (irlncRNAs) based on the TCGA database. Subsequently, differential expression analysis combined with Gene Ontology (GO), Kyoto Encyclopedia of Genes and Genomes (KEGG), and protein and protein interaction (PPI) analysis were performed to find the immune-related hub gene in high- and low-risk groups by utilizing the LASSO regression model. We have selected and compared IL-6, CXCL1, CXCL13, CXCL8, CXCL2, CXCL6, PI3, NFKBIZ, and TREM1 hub genes. Although these hub genes can secrete the release of promote inflammatory factors, affect the tumor microenvironment [18,19,20,21,22], only TREM-1 is a cell surface receptor and a member of the immunoglobulin superfamily that potently amplifies inflammatory responses by secretion of proinflammatory mediators. TREM-1 is known as an activating receptor expressed on neutrophils, monocytes, and macrophages [23]. Furthermore, we performed immune microenvironment analysis, immune cell infiltration analysis, clinical correlation analysis, and GSEA analysis for TREM-1 to determine their potential involvement in immune infiltration in ccRCC.

Methods

Data collection and pre-processing

The data of 539 ccRCC and 72 normal samples were obtained from the TCGA database (https://portal.gdc.cancer.gov/). From the ImmPort database (https://www.immport.org/home), a list of recognized 2484 IRGs was obtained. We have covered case-control studies of bioinformatics analysis methods.

Establishment of weighted co-expression network and immune prognostic signature

The weighted gene co-expression network analysis (WGCNA) was used to identify the modules associated with survival time and status. We obtained a list of irlncRNAs that were co-expressed with the IRGs through Pearson correlation analysis [18]. Then, we conducted a univariate Cox analysis of the irlncRNAs to identify the prognostic irlncRNAs.

Construction of the prognostic IRGs and irlncRNA signature model

To construct the prognostic signature model, the Lasso regression analysis was applied using the R package glmnet. The stability of the model was verified via univariate and multivariate Cox regression analysis in combination with the clinicopathological characteristics correlation analysis. Kaplan–Meier (KM) survival curves and receiver operating characteristic curves (ROC) were drawn using the survminer and timeROC R packages, respectively.

Differential gene analysis and functional enrichment analysis

Differentially expressed IRGs were analyzed through the limma R package using the following screening criteria: false discovery rate (FDR) < 0.05 and log2 |fold change| >1, which means that the difference multiplier is greater than 2 times. Functional enrichment analyses using GO and KEGG were conducted for differentially expressed IRGs using the clusterProfiler R package.

Exploring the hub gene and validation of its mRNA expression level and prognostic value

The hub gene was screened from the IRGs in the GO enrichment results using the String website (https://string-db.org/). The setting parameters of the String website: network type: full STRING network; meaning of network edges: evidence; active interaction sources: textmining, experiments, databases, co-expression, neighborhood, gene fusion, co-occurrence; minimum required interaction score: medium confidence (0.400); network display mode: interactive svg. Then, the mRNA expression levels of the hub gene in ccRCC were assessed in TIMER and TCGA database. The Kaplan–Meier (KM) survival curve, univariate, and multivariate Cox regression analyses were plotted to investigate the prognostic value using the corresponding R packages.

Exploring the relationship between hub gene expression and immunocytes and their marker genes

The ESTIMATE algorithm was applied to evaluate the tumor microenvironment scores. Then, the corresponding survival analysis was performed using the survminer R package. To evaluate the relationship between the proportion of 22 immune cell types and the hub gene, CIBERSORT, a newly developed deconvolution algorithm, was applied. Then, the correlation between TREM-1 expression and the gene markers of the main immune cells was verified in TCGA.

Gene set enrichment analysis and PPI network analysis of the hub gene

Gene set enrichment analysis (GSEA) was used to analyze the immune-related KEGG pathways of the hub gene through the R package. To further investigate the role of the hub gene, we constructed the PPI networks using the String website (https://www.string-db.org/) and GeneMANIA (http://genemania.org/).

Chemicals and antibody

All chemical reagents were purchased from Beijing Sinopharm Chemical Reagents Co. Ltd. (Beijing, China). TREM-1 antibody was from Santa Cruz Biotechnology, Inc. (Cat#: sc-9013. Texas, USA).

Kidney cancer tissue

The formalin-fixed, paraffin-embedded kidney cancer TMA used in this work was obtained from the Department of Pathology, Taizhou Hospital, Zhejiang Province from January 2021 to November. To avoid subjective differences, all cancer tissue samples were pathologically confirmed by two pathologists with 8 years of diagnostic experience in accordance with Fuhrman nuclear grading [24]. The use of human kidney cancer tissue samples was approved by the Institutional Ethic Committee of Taizhou Hospital of Zhejiang Province (K20210755).

Entry criteria: (1) patients with no metastasis found before surgery, (2) no history of major diseases or malignant tumors in the past, (3) the specimen collection method is total nephrectomy or partial nephrectomy, (4) postoperative or puncture biopsy confirmed clinicopathological confirmation as ccRCC.

Exclusion criteria: (1) patients with evidence of metastasis, who have received radiotherapy, chemotherapy, immunotherapy and molecularly targeted therapy; (2) patients with postoperative pathological confirmation of mixing two or more clinical pathological types; (3) patients with other major diseases after surgery.

Immunohistochemical

Briefly, paraffin-embedded 3-μm-thick sections were deparaffinized, rehydrated, submerged into citric acid buffer for heat-induced antigen retrieval, immersed in 0.3% hydrogen peroxide to block endogenous peroxidase activity, blocked with 3% bovine serum albumin, incubated with primary antibodies at 4°C overnight, and developed using the DAKO ChemMate Envision Kit/HRP (Dako-Cytomation, USA). The sections were then counterstained with hematoxylin, dehydrated, cleared and mounted. The tissues exhibiting brown staining in the cytoplasm, nucleus, or membrane were considered positive. The slides were scanned using Leica SCN400 slide scanner (Leica Biosystems, Wetzlar, Germany). The optical density (OD) of protein expression was quantitatively determined using Image J software.

Results

Identification of prognostic IRGs and irlncRNAs

We identified 1358 differentially expressed IRGs associated with ccRCC using the limma package in the R language. Subsequently, the clinical data of ccRCC and 1358 IRGs were analyzed and classified through WGCNA analysis using the R language (Fig. 1A). A total of 200 prognostic IRGs were obtained by univariate Cox analysis (p < 0.05). Then, we utilized co-expression analysis to obtain irlncRNAs that were co-expressed with IRGs (| r | > 0.6, p < 0.001). A total of 364 irlncRNAs were identified, of which 275 were distinguished as prognostic via univariate Cox analysis (p < 0.01).

Fig. 1
figure 1

The WGCNA analysis and the LASSO Cox regression model. A The correlation coefficient and the p value are shown in the figure. B–C The parameters of LASSO Cox regression model

Full size image

Construction of the prognostic IRGs and irlncRNAs signature model

The previously acquired 200 IRGs and 275 irlncRNAs were used to construct the LASSO Cox regression model regarding the prognosis of ccRCC (Fig. 1B–C). We integrated the expression profiles of 3 hub genes (AR, TREM-1, PTX3) and 3 hub lncRNAs (HOTAIRM1, AC003092.1, DLGAP1-AS2) involved in model construction (Fig. 2A). The risk score was calculated by multiplying the expression level of each gene or lncRNA and its corresponding coefficient. The 530 ccRCC samples were divided into low- and high-risk groups based on the median risk score (Fig. 2B–D). Kaplan Meier analysis showed that the long-term survival time of patients with ccRCC in the high-risk group was shorter than that in the low (Fig. 2E). ROC curve analysis showed that the AUC for 3-year survival was 0.718 (Fig. 2F).

Fig. 2
figure 2

Risk model construction and its related analysis. A Expression levels of the genes and lncRNAs involved in model construction, ***p < 0.001. B–C Distribution of the risk scores, survival status, and the heat maps of three hub lncRNAs and three hub mRNAs. E–F Survival curve and ROC curve analysis. G Univariate and multivariate Cox regression analyses for patients with ccRCC

Full size image

Model validation and clinical relationship assessment

Both univariate and multivariate Cox regression results showed that the tumor stage, tumor grade, and the risk score of this model were all associated with the prognosis of ccRCC (Fig. 2G). These results demonstrated that our model could be used as the independent prognostic factor for ccRCC. The relationship between risk groups and clinicopathological characteristics noted that the risk score was correlated with survival status, tumor grade, and tumor stage but not with the age or gender of the patients (Table 1).

Table 1 Relationship between risk groups and clinicopathological characteristics in 524 ccRCC patients
Full size table

GO, KEGG, and PPI analyses

We conducted differential gene expression analysis and obtained 503 DEGs, including 66 IRGs. Then, we performed GO and KEGG analysis of these DEGs (Fig. 3A). To search for immune-related DEGs, we selected genes enriched via GO involved in the immune response, including B-cell activation, B-cell mediated immunity, T-cell activation, etc. Next, we constructed the PPI network for these selected immune-related DEGs using string and cytoscape software (Fig. 3B). The number of nodes in the top 20 immune-related DRGs is shown in Fig. 3C. Eventually, we selected a significant module from the MCODE plugin analysis that included the following genes: IL-6, CXCL1, CXCL13, CXCL8, CXCL2, CXCL6, PI3, NFKBIZ and TREM-1 (Fig. 3D).

Fig. 3
figure 3

The enrichment analysis. A GO and KEGG enrichment analysis of 503 DEGs. B Construction of a PPI network that includes 36 nodes and 64 edges. C The number of nodes in the top 20 immune-related DEGs. D A significant module was selected using the MCODE plugin of Cytotype software

Full size image

Correlation between TREM-1 expression and ccRCC

We analyzed the clinical relationship between the nine immune-related DEGs and ccRCC according to the ranking order obtained from the ENCODE results and selected the hub gene TREM-1. The oncomine database analysis revealed that TREM-1 was highly expressed in different types of cancer, including ccRCC (Fig. 4A). We further validated the relationship of TREM-1 to ccRCC using the TCGA database, and the results were similar to before (Fig. 4B–C). The TREM-1 levels were gradually upregulated along with the tumor grades and stage 3 to stage 4 in ccRCC (Fig. 4D). Furthermore, the analysis of the clinical characteristics revealed that TREM-1 was correlated with survival status, grade, and TNM stage in ccRCC (Table 2). The survival analysis showed that ccRCC patients with high TREM-1 expression have a shorter survival time than those with low expression (Fig. 4E). Univariate and multifactorial Cox regression results indicated that TREM-1 is an independent prognostic factor for ccRCC (Table 3).

Fig. 4
figure 4

The relationship between TREM-1 and ccRCC in TCGA. A The TREM-1 expression profiles in different cancers, *p < 0.05, **p < 0.01, ***p < 0.001. B–D Expression levels of TREM-1 in patients with ccRCC. E Survival curve analysis

Full size image
Table 2 Relationship between TREM-1 expression level and clinicopathological characteristics in 524 KIRC patients
Full size table
Table 3 Univariate and multivariate COX regression of TREM-1
Full size table

Correlation of TREM-1 with immune infiltration and their gene markers in ccRCC

The ESTIMATE results revealed that the presence of stromal cells and immune cells in the high-TREM-1 expression group was more abundant than the group with the low expression (Fig. 5A). The survival analysis showed a negative correlation between the immune score and survival time (Fig. 5B). The CIBERSORT analysis results displayed higher contents of activated memory CD4+ T cells, monocytes, M0 and M2 macrophages, resting dendritic cells, activated resting mast cells, and neutrophils in the high-TREM-1 expression group than in the low-TREM-1 expression group. In contrast, a low expression of TREM-1 yielded higher concentrations of CD8+ T cells, resting NK cells, activated NK cells, M1 macrophages, and resting mast cells (Fig. 5C). In addition, the different abundances of the infiltrated immune cells between the high- and low-TREM-1 expression groups were further confirmed through the correlation analysis conducted between the different marker subsets of immune cells and TREM-1 [25]. These findings validate that TREM-1 is strongly linked with innate immunity since the major innate immune cell types such as dendritic cells, neutrophils, natural killer cells, and macrophages were strongly associated with TREM-1 expression (Table 4).

Fig. 5
figure 5

Tumor microenvironment scores and survival analysis. A The immune scores in TME. B The survival analysis of the immune score. C Violin plot of 22 types of immune cells

Full size image
Table 4 Relationship between TREM-1 and immune cell markers
Full size table

TREM-1 showed correlation IHC

We found that uniform cytoplasmic and membrane staining showed positive expression of TREM-1. In benign tissues, TREM-1 expression was very low, but high in tumor tissue areas. And TREM-1 expression increased with increasing Fuhrman grade. OD values: Grade 3 vs Grade 1: 0.308 vs 0.199 (Fig. 6).

Fig. 6
figure 6

Immunohistochemical analysis of TREM-1 expression in renal clear cell carcinoma

Full size image

Prediction of the function of TREM-1

GSEA analysis showed that TREM-1 was involved in various immune-related pathways, including B-cell receptor, JAK-STAT, NOD-like receptor, Toll-like receptor, and T-cell receptor signaling pathway (Fig. 7A). Moreover, the string and GeneMANIA results showed that the function of TREM-1 and its associated molecules (TYROBP, CD274, SFTPD, and VSIG4) were primarily related to the negative regulation of mononuclear cell, lymphocyte and leukocyte proliferation, and the negative regulation of lymphocyte activation (Fig. 7B-C). These results indicate that TREM-1 is closely related to immune infiltration. Flow chart of bioinformatics analysis in this paper (Fig. 8).

Fig. 7
figure 7

The GSEA and PPI enrichment. A The GSEA results. B–C The functional interactions related to TREM-1 in string and GeneMANIA

Full size image
Fig. 8
figure 8

Flow chart of bioinformatics analysis in this paper

Full size image

Discussion

The TME could affect the prognosis of patients in the primary location, as there is complicated dynamic crosstalk among neoplastic and non-neoplastic cells, including adjacent stromal cells and infiltrating immune cells. In addition, immune and stromal scores were significantly associated with the expression of TREM-1. Our findings showed that only the group with a higher immune score had a poor survival prognosis, consistent with those of Wu et al. [26]. Interestingly, unlike previous studies of gastric cancer [27], clear cell renal cell carcinoma [28], and glioblastoma [29], higher stromal and immune scores indicate better OS in patients, suggesting that TME scores may have different prognostic values in different tumor types. Subsequently, we identified that one of the immune-related hub gene responsible for this difference is TREM1 and is gradually upregulated with increasing pathologic stages and grades. Therefore, TREM1 might be identified as a novel and potential biomarker that could be of interest for further research in ccRCC.

Previous studies have shown that macrophages, which play an essential role in modulating the tumor immune microenvironment, are generally divided into two categories: M1 macrophages of the classical activation pathway and M2 macrophages of the alternative activation pathway. In this study, we found that compared to M1 macrophages, the content of M2 macrophages was more abundant in the high TREM-1 expression group. However, the functions of the two are different, with M1 macrophages inhibiting tumor progression and M2 macrophages having a tumor-promoting phenotype [30, 31]. Furthermore, Yoshihiro et al. revealed that TAMs, especially M2-polarized TAMs, contribute to the activation of STAT3, which can promote cancer cell survival, angiogenesis, and immunosuppression in the tumor microenvironment [32]. The content of Tregs is also increased in high TREM-1 group, which could produce immunosuppressive cytokines and immune-inhibitory receptors to disturb the activation of anti-tumor T cell responses, leading to a worse prognosis.

We further found that TREM-1 levels in ccRCC correlate with T cell exhaustion markers (PD-1 and CTLA-4). CTLA-4 and PD-1, which have been widely studied in malignant tumors, can be converted into immune regulation of TME [33]. PD-1 signaling inhibits Akt phosphorylation by preventing CD28-mediated activation of phosphatidylinositol 3-kinase (PI3K). The differential regulation of PI3K activation by the ligation of PD-1 and CTLA-4 led to different cell phenotypes, while TREM-1’s conduction pathway can lead to downstream signal transduction of PI3K [34]. Hence, TREM-1 might play a critical role in regulating the tumor microenvironment in ccRCC and may influence the prognosis of patients.

To explore the underlying molecular mechanism of TREM-1, we further conducted GSEA and constructed the PPI network. GSEA results revealed that the pathways involving TREM-1 were mainly enriched in the B-cell receptor, JAK/STAT, NOD-like receptor, T-cell receptor, and Toll-like receptor signaling pathways. These pathways have previously been reported to be involved in the immune response in ccRCC [35,36,37]. As reported, the effect of inhibiting the JAK/STAT signaling pathway on the suppression of immune activity appears promising and could become a strategy to prevent tumor progression [38]. The top 10 related genes that may have a similar function in ccRCC were predicted by constructing a string network. A recent study uncovered TREM2/APOE/C1Q-positive macrophage infiltration as a potential prognostic biomarker and a candidate therapeutic target [39]. Further research revealed that TYROBP, HMGB1, and TLR4 were also significantly associated with the progression of ccRCC (40).

However, this research had a few limitations. First, although we confirmed that TREM-1 was highly expressed in tumor tissue compared with benign tissue, and the expression was positively correlated with Fuhrman grade, by immunohistochemistry experiment. However, due to the limitation of the small number of tumor tissue sections, the expression and distribution of TREM-1 and the invasion of immune cells were not simultaneously investigated. Second, the bioinformatics analysis needs to be supported by future biological experiments that involve in vivo and in vitro methods. In summary, these results indicate that TREM-1 may have the potential to drive the progression of ccRCC. Therefore, this gene may serve as an independent prognostic predictor and be a candidate novel biomarker for ccRCC diagnosis.