Pan-cancer analysis reveals interleukin-17 family members as biomarkers in the prediction for immune checkpoint inhibitor curative effect

Xiaying Han; Jianxin Ye; Runzhi Huang; Yongai Li; Jianpeng Liu; Tong Meng; Dianwen Song

doi:10.3389/fimmu.2022.900273

. 2022 Sep 8;13:900273. doi: 10.3389/fimmu.2022.900273

Pan-cancer analysis reveals interleukin-17 family members as biomarkers in the prediction for immune checkpoint inhibitor curative effect

Xiaying Han ^1,^2,^†, Jianxin Ye ^1,^†, Runzhi Huang ^3,^4,^†, Yongai Li ¹, Jianpeng Liu ^5,^*, Tong Meng ^1,^*, Dianwen Song ^1,^*

PMCID: PMC9493092 PMID: 36159856

Abstract

Background

The interleukin-17 (IL-17) family contains six homologous genes, IL-17A to IL-17F. Growing evidence indicates that dysregulated IL-17 family members act as major pathogenic factors in the early and late stages of cancer development and progression. However, the prevalence and predictive value of IL-17 for immune checkpoint inhibitor (ICI) therapeutic effectiveness in multiple tumor types remain largely unknown, and the associations between its expression levels and immunotherapy-associated signatures also need to be explored.

Methods

The pan-cancer dataset in The Cancer Genome Atlas (TCGA) was downloaded from UCSC Xena (http://xena.ucsc.edu/). The immunotherapeutic cohorts included IMvigor210, which were obtained from the Gene Expression Omnibus database and included in a previously published study. Other datasets, namely, the GEO dataset and PRECOG, GEO, and METABRIC databases, were also included. In 33 TCGA tumor types, a pan-cancer analysis was carried out including their expression map, clinical risk assessment, and immune subtype analysis, along with their association with the stemness indices, tumor microenvironment (TME) in pan-cancer, immune infiltration analysis, ICI-related immune indicators, and drug sensitivity. RT-PCR was also carried out to verify the gene expression levels among MCF-10A and MCF-7 cell lines.

Results

The expression of the IL-17 family is different between tumor and normal tissue in most cancers, and consistency has been observed between gene activity and gene expression. RT-PCR results show that the expression differences in the IL-17 family of human cell (MCF-10A and MCF-7) are consistent with the bioinformatics differential expression analysis. Moreover, the expression of the IL-17 family can be a sign of patients’ survival prognosis in some tumors and varies in different immune subtypes. Moreover, the expression of the IL-17 family presents a robust correlation with immune cell infiltration, ICI-related immune indicators, and drug sensitivity. High expression of the IL-17 family is significantly related to immune-relevant pathways, and the low expression of IL-17B means a better immunotherapeutic response in BLCA.

Conclusion

Collectively, IL-17 family members may act as biomarkers in predicting the prognosis of the tumor and the therapeutic effects of ICIs, which provides new guidance for cancer treatment.

Keywords: IL-17 family, pan-cancer, TCGA, systematic analysis, immunotherapeutic effects

Introduction

The interleukin-17 (IL-17) family consists of a subset of cytokines involved in acute and chronic inflammatory responses. These members play crucial roles in host defense against microorganisms and the development of inflammatory diseases, which have gained a lot of attention recently (1, 2). Since the discovery of IL-17A in 1993 (3), other IL-17A homologous genes, namely, IL-17B, IL-17C, IL-17D, IL-17E (also known as IL-25), and IL-17F, have been screened and identified (4). As early as 1863, Virchow considered that chronic inflammation and the incidence of cancer were indeed closely related (5). In detail, chronic inflammation can trigger a series of molecular events, leading to malignant transformation of differentiated cells and antitumor immunosuppression, ultimately leading to tumor occurrence and metastasis (6–9). Thus, IL-17 has been intensively studied in the context of cancer development and progression. Its dysregulation is regarded as a major pathogenic factor involved in the early and late stages of cancer development (2). If targeted respectively, the IL-17/L-17R axis could serve as a potential novel immunotherapeutic target in cancer (10, 11). For example, IL-17 has been reported to be involved in the initiation and development of pancreatic precursor lesions of pancreatic ductal adenocarcinoma (PDAC) (12). The interaction between IL-17 and IL-17RA, which are overexpressed in the epithelium upon Kras activation, promotes a stemness signature in premalignant lesions, and IL-17 blockade increases immune checkpoint inhibitor (ICI) [cytotoxic T lymphocyte antigen 4 (CTLA-4) and programmed death 1 (PD-1)] sensitivity (13, 14). However, the predictive value of the IL-17 family for ICIs’ curative effect in multiple tumor types remains largely unknown, and it is also necessary to explore the relationship between its expression level and immunotherapy-related signatures.

Novel immune checkpoints are expressed on immune cells, can regulate the degree of immune activation, and play an important role in preventing the occurrence of autoimmunity. The abnormal expression and function of immune checkpoints are one of the important reasons for the occurrence of many diseases. ICIs could significantly prolong the overall survival (OS) of cancer patients who were sensitive to ICI therapy (15, 16), and a majority of patients fail to benefit from ICI treatment (17, 18). Therefore, there is a substantial unmet clinical need in identifying suitable patients who may respond to treatment with ICIs (19–21).

In this study, the IL-17 expression landscape of 33 different cancers was presented and the underlying effect of IL-17 expression in tumor prognosis, tumor immune microenvironment, and tumor immunotherapy has been explored. Many important immune modulators and dynamic immunological biomarkers, such as Tumor Immune Dysfunction and Exclusion (TIDE) value, PD-L1 expression, tumor mutational burden (TMB), and microsatellite instability (MSI), were investigated in this research. Furthermore, the correlation between IL-17 expression and ICI treatment was explored. Overall, this work provided evidence to elucidate the immunotherapeutic role of IL-17 in cancer, which could provide new guidance for cancer treatment.

Methods and materials

The mRNA expression of IL-17A–F in single cell type and normal human tissues

The Human Protein Atlas (https://www.proteinatlas.org/) was used to explore the expression of IL-17A–F in single cell types and normal human tissues.

Data download and preprocessing

The pan-cancer dataset in The Cancer Genome Atlas (TCGA) was downloaded from UCSC Xena (http://xena.ucsc.edu/). Firebrowse (http://firebrowse.org/) was used to analyze the basic features of the pan-cancer dataset. For the therapeutic cohort, a systematic search was performed to identify the ICI cohorts, which could be publicly retrieved and reported with complete clinical information. An immunotherapeutic cohorts were finally employed in this study: advanced urothelial cancer with atezolizumab intervention (IMvigor210 cohort downloaded from previously published research) (22). The full names and abbreviations of 33 types of tumors in the GDC TCGA from the UCSC Xena database are shown in Table S1 .

Differential expression, coexpression, and gene activity of IL-17A–F in tumor and normal tissues

For all types of TCGA tumors, “ggpubr” R package was used for differential expression analysis between tumor and adjacent tissues (Wilcox test). The difference in gene expression of the IL-17 family in pan-cancer showed log2 fold change (log2FC) in heatmap. The Corrplot R software package was used to analyze the coexpression of IL-17A–F, and the protein–protein interaction (PPI) network between these genes and their receptor genes was also constructed by applying STRING database (https://string-db.org/) (23) and Cytoscape (Version: 3.8.2). IL-17B, IL-17C, and IL-17D gene activity was generated by single sample gene set enrichment analysis (ssGSEA). First, the average values of the gene activity of 33 cancers were calculated and arranged. After that, the differences in gene activity in tumors and normal tissue were explored.

Clinical correlation analysis

The differences in OS results between patients with high and low expression of IL-17A–F were analyzed; Kaplan–Meier plots of IL-17 genes in pan-cancer were generated using the R package (24, 25). The phenotype information and survival outcome data of 33 TCGA cancer types were downloaded from the GDC TCGA dataset in the UCSC Xena database. According to the median expression levels of IL-17A–F, patients were divided into a high-expression group and a low-expression group. In addition, we also performed Cox proportional hazard regression to analyze the hazard ratios of IL-17A–F in each TCGA tumor type. Moreover, the differences in IL-17 expression levels in different stages of KIRP and BRCA were analyzed. Variations with a p-value < 0.05 were considered significant.

Immune subtype analysis

Tumor microenvironment (TME) had an important therapeutic and prognostic significance in antitumor therapy. Based on five representative immune signals, researchers identified six immune subtypes of TCGA tumors, which provided resources for the analysis of TME in some specific tumors. Six stable and repeatable immune subtypes were classified according to five immune expression characteristics (26): macrophages/monocytes (27), total lymphocyte infiltration (mainly T and B cells) (28), TGF-β response (29), IFN-γ response (30), and wound healing (31). The six subtypes may play a key role in predicting disease outcomes, rather than relying solely on the characteristics of individual cancer types (26). In order to detect the mRNA expression levels of IL-17A–F of six different immune subtypes in TCGA tumor types, we used Kruskal–Wallis test for differential expression analysis.

Stemness indices and TME in pan-cancer

Solid tumor tissue is composed of tumor cells and non-tumor cells such as immune cells, stromal cells, and vascular cells. To study the association between the expression level of IL-17 and the proportion of immune and stromal cells in TCGA tumor types, ESTIMATE (estimation of immune scores and stromal scores) (32) was utilized to reckon the proportion of these two TME components. The ESTIMATE scores were calculated according to the characteristics of gene expression, which reflected the purity of the tumor to a certain extent. Therefore, the Spearman correlation between the expression level of IL-17 genes and stromal scores was analyzed by using the ESTIMATE package and the limma package.

In order to further analyze the correlation in pan-cancer between the IL-17 family and stemness characteristics, one-class logistic regression (OCLR) algorithm was used and Spearman correlation analysis was performed based on gene expression and stemness indices (33). Stemness indices described the potential renewal ability of cells, which could lead to tumor metastasis and occurrence. Here, we used the DNA methylation-based stemness index (DNAss) and the mRNA expression-based stemness index (RNAss), which were two types of stemness indices.

For STAD and THCA, specifically, we utilized RNAss, DNAss, stromal scores, immune scores, and ESTIMATE scores to analyze the correlation with IL-17 transcriptional expression.

Immune infiltration analysis

The composition and abundance of immune cells in the TME strongly influence the progress of tumors and the effect of immunotherapy (34). TIMER2.0 no longer uses only one algorithm (http://timer.cistrome.org/) but uses the six most advanced algorithms, namely, TIMER, CIBERSORT, quanTIseq, xCell, MCP-counter, and EPIC algorithms, to provide more reliable immune infiltration levels for TCGA or tumor profile by using the estimated infiltration level (35, 36). The correlation of the expression of IL-17A/B/F genes and the level of immune infiltration of different tumor types was explored. The scores of six immune infiltrating cells of 33 tumors were obtained in the TIMER2.0 database, the correlation between IL-17A/B/F gene expression and these immune infiltrating cells was analyzed, and the correlation between IL-17B transcription level and the purity of BLCA, CESC, KIRC, and PCPG was also obtained at the same time.

Cytotoxic T-lymphocyte infiltration level and survival analysis

Predicting tumor response to immune checkpoint blockers (ICBs) requires an understanding of how tumors escape the immune system. We used the web application TIDE (http://tide.dfci.harvard.edu) to analyze the related work (37). The cancer datasets with both patient survival durations and tumor gene expression profiles from the TCGA (38), PRECOG (39), GEO, (40) and METABRIC (41) databases (37). The PRECOG database provided only survival duration information without other clinical factors. METABRIC is a breast cancer cohort, and all tumors were split according to the PAM50 subtypes (luminal A, luminal B, HER2, basal, and triple-negative) (37). The GEO dataset included GSE1427, GSE12417_GPL96, GSE9893, GSE5123, GSE5828, GSE50081, GSE31245, GSE1993, GSE71014, GSE11969, GSE39582, GSE4475, GSE37745, GSE3149, GSE16131, GSE54236, GSE50081, GSE49997, GSE63885, GSE1037, GSE32062, GSE71187, GSE65858, GSE54236, GSE10245, GSE42127, and GSE18521.

To predict each tumor’s potential to escape T cell-mediated killing, we first classified each tumor into CTL (cytotoxic T lymphocytes)-high or CTL-low categories through the CTL marker expression levels (CD8A, CD8B, GZMA, GZMB, and PRF1). The correlation between the expression of IL-17A–F and the CTL infiltration level was explored in different tumors. The association between the CTL level and OS was computed through the two-sided Wald test in the Cox-PH regression. Samples were classified into two groups for each Kaplan–Meier plot: “High CTL” samples have above-average CTL values among all samples, while “Low CTL” samples are below average. Samples were split into a high-expression group and a low-expression group according to the IL-17A–F expression to show the association between the CTL level and survival outcome.

T-cell dysfunction and exclusion analysis

Here, both T-cell dysfunction and exclusion signatures were used to model immune escape in tumors with different CTL levels (42, 43). Some tumors have a high level of infiltration by cytotoxic T cells, but these T cells tend to be in a dysfunctional state. Patients with dysfunctional T-cell infiltration are more likely to be resistant to ICB reprogramming. In other tumors, immunosuppressive factors may exclude T cells from infiltrating tumors such as impaired priming of tumor-specific T cells or suppressive cells prohibiting T-cell infiltration into the tumor (42–44). The average expression profiles of cancer-associated fibroblasts (CAFs), myeloid-derived suppressor cells (MDSCs), and the M2 subtype of tumor-associated macrophages (TAMs) reported to restrict T-cell infiltration in tumors were to model T-cell exclusion (43). In Jiang’s study, it was described that “the Pearson correlation with either T cell dysfunction (in CTL-high tumors) or exclusion (in CTL-low tumors) signatures was defined as the TIDE prediction score (37)“. All tumors were ranked by their TIDE scores to predict their ICB response. A higher tumor TIDE prediction score is associated not only with worse ICB response, but also with worse patient survival under anti-PD1 and anti-CTLA4 therapies (37). TIDE (http://tide.dfci.harvard.edu) was used to get the related results. The dataset from the TCGA database, E-MTAB-179, GSE12417_GPL570 from the GEO E-MTAB-179 dataset, METABRIC database, Braun2020_PD1 (45), Gide2019_PD1 (46), Hugo2016_PD1 (47), Lauss2017_PD1 (48), Liu2019_PD1 (49), Mariathasan2018_PDL1 (50), Miao2018_ICB (51), Nathanson2017_CTLA4 (52), Riaz2017_PD1 (53), VanAllen2015_CTLA4 (54), and Zhao2019_PD1 (55) were included. The correlation between IL-17A–F gene expression and T dysfunction value in the core dataset, the normalized z score calling from Cox-PH regression in the immunotherapy dataset, and the normalized expression value from immune-suppressive cell types were explored.

Regulation of immune molecule analysis

The TISIDB database (http://cis.hku.hk/TISIDB) integrates 988 reported genes related to antitumor immunity through the analysis of high-throughput screening and genomic analysis data to pre-calculate the association between TCGA cancer-type genes and immune characteristics, including lymphocytes, immunomodulators, and chemokines (56). In our study, we used the TISIDB database to analyze the association between the expression level of IL-17D and immunomodulators. We used the TCGA WES data to calculate the TMB levels in various tumors through the R software package “maftools”, and the MSI scores for each TCGA cancer case were derived from (57), using radar charts to illustrate the relationship between them. Finally, for STAD, we performed gene set enrichment analysis (GSEA) to describe the significant differences observed between the high-expression IL-17 group and the low-expression IL-17 group in KEGG and GO databases. The selected pathway criteria are p < 0.05 and the top five highest normalized enrichment scores.

Immunotherapeutic response analysis

Complete remission (CR), partial remission (PR), progressive disease (PD), and stable disease (SD) are the four outcomes of immunotherapy. In this study, CR and PR were classified as responders, and PD and SD were classified as non-responders. CR, PR, and SD were classified as clinically harmless groups, and PD was assigned to another group. The Wilcoxon test was used to study differences in gene expression of IL-17 members between responders and non-responders in the related independent immunotherapy cohorts (IMvigor 210).

Drug sensitivity analysis in pan-cancer

The IC₅₀ of 265 small molecules in 860 cell lines and its corresponding mRNA gene expression were collected from Genomics of Drug Sensitivity in Cancer (GDSC) (https://www.cancerrxgene.org/) (58–60). The mRNA expression data and drug sensitivity data were merged. Pearson correlation analysis was performed to obtain the correlation between gene mRNA expression and drug IC₅₀. p-value was adjusted by FDR. The IC₅₀ of 481 small molecules in 1,001 cell lines and its corresponding mRNA gene expression were also collected from the Cancer Therapeutics Response Portal (CTRP) (https://portals.broadinstitute.org/ctrp/). The mRNA expression data and drug sensitivity data were merged. Pearson correlation analysis was performed to obtain the correlation between gene mRNA expression and drug IC₅₀. p-value was adjusted by FDR. A bubble plot was provided to summarize the correlations between inputted genes and drugs. A gene will be obtained only when it is associated with at least one drug. Also, a drug will be obtained only when it is associated with at least one gene.

Quantitative real-time polymerase chain reaction assay

Total RNA was extracted from MCF-10A and MCF-7 cell lines with the TRIzol kit (Invitrogen, Carlsbad, CA, USA), the obtained RNA was reverse transcribed into cDNA with a reverse transcription kit (EnzyArtisan, Shanghai, China), and a SYBR Green qPCR Kit (EnzyArtisan, Shanghai, China) was picked to amplify the target transcript to analyze gene expression. The human primer sequences for IL-17B, IL-17C, IL-17D, and IL-25 (IL-17E) are listed in Table S2 .

Results

The mRNA expression of IL-17A–F in single-cell type and normal human tissues

Figures S1 , S2 show the mRNA expression of IL-17A–F in single-cell type and normal human tissues. IL-17A is mainly expressed in T cells including naive CD4 T-cell activated, memory CD4 T-cell Th17, naive CD8 T-cell activated, memory T-reg, memory CD4 T-cell TFH, dendritic cells, Langerhans cells, macrophages, and gastric mucus-secreting cells. Among all organs, it is mainly expressed in bone marrow and lymphoid tissues, kidney and urinary bladder, liver and gallbladder, the gastrointestinal tract, and the proximal digestive tract. IL-17B is mainly expressed in breast myoepithelial cells, Sertoli cells, peritubular cells, theca cells, smooth muscle cells, late spermatids, cardiomyocytes, fibroblasts, early spermatids, Langerhans cells, Kupffer cells, breast glandular cells, basal keratinocytes, and extravillous trophoblasts. Among all organs, it is mainly expressed in male/female tissues, muscle tissues, the proximal digestive tract, the gastrointestinal tract, endocrine tissues, bone marrow, and lymphoid tissues. IL-17C is mainly expressed in basal respiratory cells, club cells, ionocytes, respiratory ciliated cells, cholangiocytes, and exocrine glandular cells. Among all organs, it is mainly expressed in male/female tissues, kidney and urinary bladder, liver, and gallbladder. IL-17D is mainly expressed in astrocytes, late spermatids, oligodendrocyte precursor cells, Muller glia cells, early spermatids, excitatory neurons, oligodendrocytes, inhibitory neurons, Leydig cells, skeletal myocytes, melanocytes, spermatocytes, fibroblasts, extravillous trophoblasts, rod photoreceptor cells, and microglial cells. Among all organs, it is mainly expressed in brain and skeletal muscle. IL-25 is mainly expressed in cardiomyocytes. Among all organs, it is mainly expressed in testis and muscle tissues. IL-17F is mainly expressed in T cells (naive CD8 T cells). Among all organs, it is mainly expressed in bone marrow and lymphoid tissues, kidney, and urinary bladder.

Differential expression, coexpression, and gene activity of IL-17A–F in tumor and normal tissues

The outline of our research is shown in Figure 1 . We used datasets from TCGA, and the basic information is shown in Figure S3 . The gene expression of IL-17A–F is displayed in Figure 2A . Wilcox test was used to analyze the differential expression of five IL-17 family genes in tumor and adjacent tissue ( Figure 2B ). Obviously, the expression of IL-17D was the highest among the IL-17 family in major tumors. Across the heatmap, we could find that the expression of IL-17C was mostly upregulated in most types of tumors, while the expression of IL-17B and IL-17D was low in some tumors. Compared to normal tissues, the expression of IL-17B was downregulated in most tumors except CHOL, GBM, and LIHC (p < 0.01), and the expression of IL-17C was upregulated in all tumors except KICH (p < 0.01). Moreover, IL-17D was the only gene in the IL-17 family whose expression was downregulated in LUAD (p < 0.01). In the three gastrointestinal cancers including ESCA, COAD, and READ, the expression of IL-17F was all upregulated (p < 0.01). IL-17B was significantly downregulated in BRCA compared to normal tissues. Furthermore, we once again performed in vitro experiments to investigate the expression differences of IL-17B, IL-17C, IL-17D, and IL-25 (IL-17E) of human cell (MCF-10A and MCF-7) ( Figure 2C ), which was consistent with the differential expression analysis mentioned above. Coexpression analysis showed a strongly positive coexpression relationship (r = 0.71) between IL-17A and IL-17F ( Figure 2E ), further proved by the PPI network ( Figure 2D ). The contact could also be revealed between IL-17A and IL-17C (r = 0.2). On the contrary, there was a negative coexpression relationship between IL-17D and IL-17C (r = −0.1). As regards gene activity, the difference in gene activity between normal tissues and adjacent tissues was basically consistent with the difference in gene expression ( Figure 3 ). For example, the gene activity of IL-17B in BRCA was also lower compared with normal tissues.

Differential expression analysis. **(A)** The box plot showing the transcriptional expression levels of the IL-17 family based on the TCGA dataset. **(B)**The heatmap exhibiting the transcriptional level of the IL-17 family in TCGA tumor types compared to adjacent normal tissues; the gradient colors represent the log fold change (logFC) value. (Red points indicate high expression, while blue points indicate low expression.) **(C)** Level of IL-17B, IL-17C, IL-17D, and IL-25 (IL-17E) mRNA in MCF-10A and MCF-7. **(D)** The protein–protein interaction (PPI) network constructed among IL-17A, IL-17B, IL-17C, IL-17D, IL-25, IL-17F, and their receptors. The core gene was marked red. **(E)** Co-expression analysis between each two genes is presented. (Red points indicate positive correlation, while blue points indicate negative correlation.) (∗∗∗∗p < 0.0001; ∗∗∗p < 0.001; ∗∗p < 0.01; ∗p < 0.05.).

The different activity analyses based on the TCGA dataset. **(A–C)** The mean activity of IL-17B, IL-17C, and IL-17D in 33 cancers (from high to low). **(D–F)** The different activity analysis between tumors and normal groups of IL-17B, IL-17C, and IL-17D in 33 cancers (from high to low). (∗∗∗∗p < 0.0001; ∗∗∗p < 0.001; ∗∗p < 0:01; ∗p < 0.05.).