Introduction
Kidney cancer is a common malignancy of the urinary system that originates from the renal tubular epithelium. The most common histological subtype of kidney cancer is renal cell carcinoma (RCC), accounting for approximately 90% [
1,
2], including kidney renal clear cell carcinoma (KIRC, 70%), papillary RCC (10–15%), and chromophobe RCC (5%) [
2]. The Global Cancer Statistics (2020) stated that RCC accounted for 2.2% of newly diagnosed cancers annually, of which 25 to 30% of patients were diagnosed as advanced or metastatic with a 5-year survival rate of 10%, and 20% to 30% of patients having a propensity for recurrence and metastasis after local operation [
3‐
5].
Depending on tumor immunogenic characteristics, the systemic treatment of RCC has witnessed significant changes in the last 20 years [
6]. Traditional immunotherapy was predominantly based on interferon (IFN)-α and interleukin (IL)-2; however, IFN-α exhibits poor efficacy and IL-2 displays high toxicity [
7,
8]. The subsequent application of vascular endothelial growth factor (VEGF), tyrosine kinase inhibitors (TKIs), and mammalian target of rapamycin (mTOR) inhibitors has improved the efficacy and safety of RCC systemic treatment [
6]. Recently, immunotherapy agents targeting programmed death-1 (PD-1)/programmed death-ligand 1 (PD-L1) axis alone or a combination with anti-cytotoxic T lymphocyte-associated protein 4 (CTLA-4) monoclonal antibodies or antiangiogenic agents has greatly expanded the systemic treatment options for RCC [
6,
9].
Although the application of immune checkpoint inhibitors (ICIs) as tumor therapeutics has led to major improvements in the RCC systemic treatment, most patients fail to achieve a durable complete response (CR). This could be because RCC is significantly different from other solid tumors in immunogenic features, has a high mutational burden and CD8
+ T cell infiltration, and is associated with poor prognosis [
10,
11]. CD8
+ T cells constitute the major anti-tumor effective cells in the tumor microenvironment (TME) and exert cytotoxic effects. However, their function is impaired by immunosuppressive cells or molecules in the TME [
12]. Up-regulation of co-suppressor molecules, including PD1 and CTLA-4, on the surface of CD8
+ T cells bind to relevant ligands, ultimately causing CD8
+ T cell exhaustion [
13,
14]. Therefore, blocking PD-1-mediated inhibitory signaling by monoclonal antibodies could reverse CD8
+ T cell exhaustion, thereby hindering tumor progression. However, this contradicted CD8
+ T cell characteristics in RCC [
10,
11,
15]. Recent studies have reported that the timing of PD-1 inhibition could negatively affect T-cell priming and memory CD8
+ T cell formation, thus contributing to more appropriate timings in RCC immunotherapy [
16,
17].
No relevant studies are available to explore the function of CD8
+ T cell-associated gene sets in the KIRC and its TME. The high accuracy and specificity of single-cell RNA sequencing allows analysis of the functional status of CD8
+ T cells and the expression of its associated genes using the single-cell sequencing data of immune-infiltrated KIRC (GSE121636 [
18]) and normal kidney (GSE131685 [
19]). In the present study, we developed a CD8
+ T cell-associated immunological risk prognostic model using CD8
+ T cell-associated marker genes obtained from single-cell sequencing analysis to predict the survival status, tumor immune microenvironment, and immunotherapy responsiveness of KIRC patients, thereby providing a potential target and predictive evidence for immunotherapy.
Materials
Single-cell data filtration, pre-processing, and cluster identification
The single-cell sequencing data of KIRC GSE121636 [
18] and normal kidney GSE131685 [
19] were screened using the Gene Expression Omnibus (GEO) database (
https://www.ncbi.nlm.nih.gov/geo/). See Table
1 for details. The Seurat package was used to generate objects and filter out poor-quality cells. The standard data pre-processing processes were performed and percentages of gene numbers, cell counts, and mitochondrial sequencing counts were calculated. The filtering criteria were genes with less than only three cells detected and disregarded cells with less than 200 detected gene numbers. Cells with less than 200 or more than 2500 genes detected and those with high mitochondrial content (> 10%) were filtered out as well. We scaled the UMI counts using a scale factor of 10,000 to normalize the library size effect of each cell. Following the log transformation of data, other factors, including “percent.mt,” “nCount_RNA,” and “nFeature_RNA” were corrected for variation regression using the ScaleData function in Seurat (v3.0.2). The corrected normalized data metric was applied to standard analysis. The top 50 variable genes were extracted for principal component analysis (PCA). The top 10 principal components were retained for UMAP visualization and clustering. Cell clustering was performed using the FindClusters function (resolution = 0.5) included in the Seurat R package.
Differential expression and survival prognosis analysis
The “Survival R package” was used to analyze differential expression, overall survival (OS), disease-specific survival, and progression-free survival of CD8+ T cell-associated genes based on single-cell sequencing and assays. Furthermore, we established the correlation between key genes and clinicopathological features of KIRC and constructed a prognostic nomogram and calibration curve in the TCGA–KIRC cohort.
Consensus clustering
A cluster analysis was performed using the “Consensus Cluster Plus R Package” [
20] and agglomerative PAM clustering with a 1-Pearson correlation distances and resampling 80% of the samples for 10 repetitions. The optimal number of clusters was determined using the empirical cumulative distribution function plot.
Identification of differentially expressed genes (DEGs)
Differential mRNA expression between subtypes was studied using the Limma package (version: 3.40.2). The screening criteria were adjusted as p < 0.05 and |fold change| > 2.
Functional enrichment analysis and gene set enrichment analysis (GSEA)
The “clusterProfiler” package was used to evaluate the Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) analyses of different CD8
+ T cell-associated subtypes, with
q-value and
p-value thresholds of < 0.05. Moreover, the difference in gene sets between high and low expression of CD8
+ T cell-associated genes was further assessed using the gene set enrichment analysis (GSEA) software (
http://www.broadinstitute.org/gsea/index.jsp).
Somatic mutation and immune landscape analysis
The KIRC somatic mutation data were obtained from the TCGA GDC data portal, and waterfall plots were created using the “Maftools” package. The relative percentages of different immune cell types were determined using the ESTIMATE and xCell algorithms, and the relative percentages of immune cell types between the two subtypes were compared by landscape plots. Moreover, TISIDB (
http://cis.hku.hk/TISIDB/index.php) was used to further evaluate the correlation between two subtypes and different immune indicators, including lymphocytes, immune inhibitors, immunostimulators, major histocompatibility (MHC) molecules, chemokines, and chemokine receptors.
Construction of CD8+ T cell-associated risk signature
Regression analysis was performed using the Lasso–Cox method via the glmnet package. A tenfold cross-validation was set up to obtain the optimal model. The prognostic significance of genes involved was assessed using the Cox method, and the relationship between different risk scores and patient follow-up time, events, and changes in the expression of individual genes was analyzed.
Cell lines and cell culture
Human renal cancer cell lines (786-O and ACHN) were cultured in Dulbecco’s modified Eagle medium (DMEM; Gibco Thermo Fisher Scientific, USA), containing 10% fetal bovine serum (Lonsera, Uruguay), and 1% penicillin–streptomycin solution (Keygen, China). All cell lines were purchased from the Shanghai Institutes for Biological Sciences and incubated in 95% humidified air at 37 °C and 5% CO2.
RNA extraction and RT-PCR
The RNA was extracted using the RNA extraction kit (Takara Kusatsu, Japan), and the Hiscript II First-Strand cDNA Synthesis Kit was used to synthesize the complementary DNA (Vazyme, China). Reverse transcriptase-polymerase chain reaction (RT-PCR) was performed using the MonAmpTM SYBR Green qPCR Mix (Monad Biotech, China).
Small interfering RNA
The small interfering RNAs (siRNAs) against the LAG3 gene were designed and synthesized by GenePharma Co. (China).
Cell proliferation and colony formation assays
For the cell proliferation assay, 1000 cells were seeded into 96-well plates for 0 h to 120 h, and 10 µL of the cell counting kit-8 (Keygen, China) solution was added per well. After a 2 h incubation at 37 °C, optical density at 450 nm (OD 450 nm) was measured using a microplate reader (Bio-Tek, USA). For the colony formation assay, cells were seeded into 6-well plates at a density of 1–2 × 10 [
3] cells/well and incubated for 10 to 14 days at 37 °C. Next, the cells were washed using phosphate-buffered saline, fixed with 4% polyformaldehyde (Service Bio, China), and stained with 0.1% crystal violet solution (Keygen, China). Colonies containing > 50 cells were counted using the ImageJ 2X software 2.1.4.7 (Rawak Software Inc., Germany).
Wound healing and transwell assay
For the wound healing assay, cells were inoculated into 6-well plates and treated with si-/nc-LAG3. A straight scratch was created on the plate with a sterilized needle tip when the cell density was approximately 70%. The cell wound edge was marked and imaged under a microscope at the starting time point. After 0 h to 24 h, the migrated distance was measured and the wound closure percentage was calculated. For transwell assays, cells were inoculated into a 24-well transwell cell apical chamber containing the matrix gel (BD, USA) for evaluating invasion and gel-free for migration. The bottom and upper chambers contained the RPMI medium and fetal bovine serum-free medium, respectively. Cells invading the bottom chambers were fixed with 4% polyformaldehyde, stained with 0.1% crystal violet solution, counted, and imaged under a microscope.
Tissue samples and tissue microarrays
Formalin-fixed and paraffin-embedded prostate cancer tissue samples were collected from patients who underwent radical nephrectomy in the affiliated Zhongda Hospital of Southeast University, China, from April 2020 to November 2021. The study samples were from patients with KIRC, and the pathological diagnosis was confirmed by at least two pathologists. With the tumor as the center, normal tissues adjacent to the tumor were used as study materials, and two pairs of tissue microarrays were created with a 0.6 mm diameter.
Immunohistochemistry
Formalin-fixed and paraffin-embedded tissue was dewaxed and dehydrated using xylene and serially-diluted ethanol. The tissue sections were incubated at 121 ℃ in an autoclave for 5 min to extract the antigen, following which these were incubated with anti-LAG3-monoclonal antibody at 4 ℃ overnight, and the bound antibody (Proteintech) was incubated at 37 ℃ for 30 min.
Statistical analysis
The statistical analysis was performed using the R software (version 4.0.2). Multivariate Cox regression analyses were used to evaluate the prognostic significance. When p < 0.05 or log-rank p < 0.05, the difference was significant.
Discussion
Kidney cancer constitutes the 14th most common malignancy worldwide with 431,288 new diagnosed cases and 179,368 new deaths in 2020 [
3]. The major etiologies of kidney cancer include hypertension, obesity, and smoking [
22]. The ICI-based combination therapy demonstrated excellent clinical efficacy in several large clinical trials and is now the first-line care standard for patients with advanced or metastatic renal cancer with a low OS at first diagnosis [
23‐
26]. Despite the pivotal function of the PD-1/CTLA-4 axis in the treatment of RCC has greatly improved clinical outcomes compared to previous treatment options, the majority of patients with RCC did not achieve durable clinical benefits after ICI-based combination therapy [
23‐
26]. Therefore, it is highly essential to investigate the tumor immune microenvironment and explore novel immunotherapeutic targets, and ultimately optimize systemic treatment for RCC.
In response to the immune microenvironment of renal cancer, Finke [
6] and Stein [
15] provided reviews on the immunology and immunobiology of renal cancer, respectively. In most solid tumors, the degree of CD8
+ T cell infiltration was positively correlated with good prognosis for tumor patients [
27]. CD8
+ T cells exerted a direct cytotoxic effect on target cells and performed a critical role in anti-cancer immunity. However, the expression of suppressive molecules (PD-1 and CTLA-4) on the surface of CD8
+ T cells increases in response to sustained stimulation by tumor-specific antigens, and their function decreases and eventually reaches the exhausted state, as demonstrated in multiple cancer models [
13,
14,
28]. Therefore, blocking their inhibitory signaling using anti-PD1/CTLA-4 antibodies could rejuvenate exhausted CD8
+ T cells, enhance their cytotoxicity, promote tumor cell lysis, and restrict tumor metastasis [
29]. However, kidney cancer had a distinctive immune profile such that the degree of CD8
+ T cell infiltration is positively correlated with poor prognosis, and the specific mechanism was unclear. Several hypotheses have been proposed to explain this paradoxical phenomenon. First, the activation status and virulence potential of CD8
+ T cells were highly specific in kidney cancer, where stem-like TCF1
+ or PD-1
+ TIM3
− LAG3
− CD8 + T cell subsets contribute to the anti-cancer immune effect [
30‐
32]. Second, the low density of tertiary lymphoid structures generated numerous immature DC cells, causing the infiltration of polyclonal CD8
+ T cells that could not recognize tumor-associated antigens [
11,
27,
32]. Third, the absence of tumor-specific genes, such as the relative absence of
PBRM1 mutations in highly CD8
+ T cells RCC, which was often associated with a better prognosis [
33]. Finally, specific metabolic dysregulation of CD8
+ T cells in RCC restricted CD8
+ T cell activation and did not recover through the PD-1 axis inhibition [
34].
The high significance of CD8
+ T cells in the immunotherapy of kidney cancer and the results of single-cell sequencing analysis (CD8 + T cells were significantly differentially expressed in KIRC and kidney tissue) can be used to construct a prognostic model of CD8
+ T cell-associated genes to guide clinical decisions in KIRC. First, the top 10 DEGs in CD8
+ T cells of KIRC were obtained by cluster identification in single-cell analysis, including
GZMK,
CD27,
CCL4L2,
FXYD2,
LAG3,
RGS1,
CST7,
DUSP4,
CD8A, and
TRBV20-1. Subsequently, the above gene expression was divided into two subtypes by cluster analysis, that is, CD8
+ T cell-associated gene low expression (C1) and high expression (C2) subtypes. The cluster analysis results were used for grouping, and the DEGs, pathway enrichment, and mutated genes between C1 and C2 subtypes were comprehensively studied. The up-regulated genes were largely enriched for glutamate and leukotriene activity, whereas the down-regulated genes were enriched for immune-related activities. Furthermore,
VHL and
STED2, the most commonly mutated genes in primary KIRC, displayed remarkably high mutation rates in the C2 subtype compared to the C1 subtype, predicting a poor prognosis such as metastasis or drug resistance in the C2 subtype [
35,
36]. We assessed the KIRC immune microenvironment in different subtypes and revealed that the stromal and immune scores were significantly higher in the CD8
+ T cell-associated gene high expression subtype than in the C1 subtype, with significant differences between the two subtypes in immune infiltrating cells and immune-related molecules. Therefore, LASSO regression and Cox univariate analyses were used to construct the CD8
+ T cell-associated risk prognostic model: RiskScore = − 0.291858656434841*GZMK − 0.192758342489394*FXYD2 + 0.625023643446193*LAG3 + 0.161324477181591*RGS1 − 0.380169045328895*DUSP4 − 0.107221347575037*TRBV20-1, which will assist the clinicians to assess the prognosis for survival, immune status, and drug selection in patients with KIRC. Relevant studies were investigated to identify CD8
+ T cell-related genes in kidney cancer. Genomics, radiology, and artificial intelligence modalities can be used to identify renal cancer differentiation more easily and earlier, and predict its Fuhrman grade and responsiveness to immunotherapy response, thus assisting the clinicians in defining the risk of stratification of the disease, treatment choices, follow-up strategies, and prognosis [
42,
43].
Using tumor single-gene related studies as the reference [
44], we performed single-gene analysis of the above genes and finally identified
LAG3 as the most valuable CD8
+ T cell-associated gene in KIRC. Lymphocyte activating gene 3 (LAG3) or CD223 is highly expressed in different T cells, CD8
+ T cells, CD4
+ T cells, and Tregs, to maintain homeostasis [
37]. However, persistent tumor-associated antigen stimulation causes its chronic expression, ultimately promoting T-cell exhaustion in cancers [
37,
38]. Therefore, LAG3 serves as the third clinical checkpoint in case of limited or even no response in 60 to 80% of cancer patients in PD1/CTLA-4 axis immunotherapy [
39]. Currently, several clinical trials are being conducted on immunotherapies targeting LAG3 in combination with PD1/CTLA-4 axis inhibitors to treat cancers including kidney cancer
40,
41. In the current study, we analyzed for the first time the distribution of LAG3 in renal cancer and renal tissue using single-cell analysis and investigated its expression, prognostic significance, immune microenvironmental relevance, and pathway enrichment using bioinformatics techniques. We confirmed that LAG3 promotes the progression and metastasis of renal cell carcinoma and is positively correlated with CD8
+ T cells using cell phenotype studies and immunohistochemistry. Presently, immunotherapy targeting LAG3 is largely used for melanoma, pancreatic cancer, and hematological tumors, with only a few studies on renal cancer. We elucidated the function of LAG3 in KIRC. We believe our findings will provide a preliminary basis and direction for LAG3-targeted immunotherapy and even CAR-T therapy in patients with kidney cancer.
The present study had certain limitations. First, heterogeneity obtained in retrospective studies needs to be verified by conducting prospective studies. Second, we only applied the top 10 CD8
+ T cell-associated genes to construct the risk prognosis model, which lacked comprehensiveness and extensiveness. Third, we only validated the function of LAG3 in KIRC at the in vitro level and lacked in vivo experiments, as well as the expression mechanism and CD8
+ T cell activity warrant further exploration. In conclusion, more basic and large clinical trials are required to validate these findings (Table
1).
Table 1
Single-cell sequence dataset information
GSE121636 | Single-cell sequencing of peripheral blood and tumor-infiltrating immune cells in renal clear cell carcinoma [5' RNA expression sequencing] | 3 tumor-infiltrating immune cells | GSE3440844, GSE3440845, GSE3440846 |
GSE131685 | Single-cell RNA sequencing of human kidney | 2 primary kidney samples | GSE4145204, GSEE4145205 |
Conclusion
The top 10 CD8+ T cell-associated genes were obtained by single-cell analysis, including GZMK, CD27, CCL4L2, FXYD2, LAG3, RGS1, CST7, DUSP4, CD8A, and TRBV20-1. These genes were divided into low- and high-expression subtypes by cluster analysis, and DEGs, pathway enrichment, mutant genes, and KIRC immune infiltration in different subtypes were studied. The best risk prognosis model was constructed (RiskScore = − 0.291858656434841*GZMK − 0.192758342489394*FXYD2 + 0.625023643446193*LAG3 + 0.161324477181591*RGS1 − 0.380169045328895*DUSP4 − 0.107221347575037*TRBV20-1). Finally, the single-gene analysis identified LAG3 as the most valuable CD8+ T cell-associated gene in KIRC, which was further confirmed by cell phenotype studies and immunohistochemistry.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.