Skip to main content
NCBI home page
Skin Research and Technology logoLink to Skin Research and Technology
. 2024 Aug 2;30(8):e13900. doi: 10.1111/srt.13900

Establishment of a CD8+ T cells‐related prognostic risk model for acral melanoma based on single‐cell and bulk RNA sequencing

Wenwen Wang 1,2, Pu Liu 3,4, Jie Ma 4,✉, Jun Li 1,✉, Ling Leng 2,✉
PMCID: PMC11296306  PMID: 39093712

Abstract

Background

CD8+ T cells have been recognized as crucial factors in the prognosis of melanoma. However, there is currently a lack of gene markers that accurately describe their characteristics and functions in acral melanoma (AM), which hinders the development of personalized medicine.

Methods

Firstly, we explored the composition differences of immune cells in AM using single‐cell RNA sequencing (scRNA‐seq) data and comprehensively characterized the immune microenvironment of AM in terms of composition, developmental differentiation, function, and cell communication. Subsequently, we constructed and validated a prognostic risk scoring model based on differentially expressed genes (DEGs) of CD8+ T cells using the TCGA‐SKCM cohort through Lasso‐Cox method. Lastly, immunofluorescence staining was performed to validate the expression of four genes (ISG20, CCL4, LPAR6, DDIT3) in AM and healthy skin tissues as included in the prognostic model.

Results

The scRNA‐seq data revealed that memory CD8+ T cells accounted for the highest proportion in the immune microenvironment of AM, reaching 70.5%. Cell–cell communication analysis showed extensive communication relationships among effector CD8+ T cells. Subsequently, we constructed a prognostic scoring model based on DEGs derived from CD8+ T cell sources. Four CD8+ T cell‐related genes were included in the construction and validation of the prognostic model. Additionally, immunofluorescence results demonstrated that ISG20 and CCL4 were downregulated, while LPAR6 and DDIT3 were upregulated in AM tissues compared to normal skin tissues.

Conclusion

Identifying biomarkers based on the expression levels of CD8+ T cell‐related genes may be an effective approach for establishing prognostic models in AM patients. The independently prognostic risk evaluation model we constructed provides new insights and theoretical support for immunotherapy in AM.

Keywords: acral melanoma, CD8+ T cells, prognostic model, single‐cell RNA sequencing

1. INTRODUCTION

Melanoma is a highly lethal tumor, accounting for 5% of all malignant skin cancers, but it is responsible for over 75% of deaths related to malignant skin cancers. 1 , 2 Over the past two decades, the incidence of melanoma has continued to rise, with a 170% increase in newly diagnosed cases worldwide from 1990 to 2019. 3 Globally, melanoma causes approximately 55,500 deaths annually. 4 Acral melanoma (AM) is a subtype of melanoma, with over 60% of AM cases occurring on the palms, soles, or under the nails. 5 , 6 Due to the subtle nature of early‐stage AM lesions, most patients are diagnosed at an advanced stage. 7 , 8 For patients with advanced melanoma, the efficacy of surgical resection or chemotherapy is very limited. 9 The 5‐year survival rates for patients with stage I–IV melanoma are 94.1%, 44.0%, 38.4%, and 4.6%, respectively. 10 Although recent developments in immunotherapy have somewhat improved the survival outcomes for melanoma patients. 11 AM patients generally exhibit a relatively low response rate to these treatments. 12 Therefore, identifying new prognostic biomarkers is crucial for improving the survival outcomes of AM patients.

The tumor immune microenvironment (TIME) is primarily composed of a series of immune cells that possess both anti‐tumor immune and pro‐tumor immune functions. 11 As a “double‐edged sword” in the tumor microenvironment (TME), the quantity and quality of immune cells in the TIME play a crucial role in tumor metastasis. Studies have indicated that there are changes in monocyte subsets within the microenvironment before and after central nervous system metastasis of melanoma, along with differential gene expression in tumor‐infiltrating monocytes. 13 Additionally, immune cells in the TME can be utilized to predict cancer patient prognosis and analyze drug resistance. Studies have indicated that the infiltration of CD8+ T cells is associated with improved survival rates in melanoma. 14 − 16 AM exhibits significant intra‐tumoral and inter‐tumoral heterogeneity, as well as a highly immune‐suppressive TME and a complex intercellular communication network. 17 , 18 Despite the recognized importance of CD8+ T cells in tumors, our current understanding of them remains limited, particularly in melanoma, where there is a lack of gene markers that accurately describe the characteristics and functions of CD8+ T cells. Furthermore, there is currently no effective gene signature related to CD8+ T cells to predict the prognosis of melanoma patients or guide their treatment decisions. These limitations hinder the development of personalized medical strategies for melanoma. The individual heterogeneity of malignant melanoma, its complex pathological classification, and unclear pathogenesis have led to difficulties in assessing the prognosis of melanoma.

In this study, we combined single‐cell RNA sequencing (scRNA‐seq) data and bulk RNA sequencing data to explore the differences in immune cell composition within AM. We constructed a prognostic evaluation model based on differentially expressed genes (DEGs) in CD8+ T cells at the single‐cell level, comprehensively characterizing the TIME of AM in terms of composition, developmental differentiation, function, and cell communication. Finally, we validated the expression of genes used in the prognostic model in both AM and healthy skin tissues using immunofluorescence staining. Overall, our results indicate that identifying biomarkers based on the molecular characteristics of CD8+ T cells may be an effective approach for establishing prognostic models for AM patients. The independently prognostic risk assessment model we developed provides new theoretical support for AM immunotherapy, offering key insights for improving the precision and efficacy of immunotherapy and for the development of more effective treatment methods.

2. METHODS

2.1. Data sources and preprocessing

The single‐cell transcriptomic data of AM were obtained from the GEO database (GSE215120), while the single‐cell transcriptomic data of healthy skin were sourced from GSE228421. Samples were subjected to quality control using the Seurat package in R software (version 4.1.2) with the following filtering criteria: nUMI > 500, 6000 > nGene > 250, percent_mito < 20, log10GenesPerUMI > 0.8. High‐quality cells selected through this process were used for further analysis. 19 The transcriptomic expression matrix and clinical information of melanoma patients were retrieved from the TCGA‐SKCM cohort using the R package TCGAbiolinks, and normalization was performed using the TPM (transcripts per kilobase million) method. Patients with incomplete clinical information and a survival time of less than 30 days were excluded, resulting in a final cohort of 95 patients for the study. Cell type annotation was performed using known marker genes and singleR.

2.2. Single‐cell data analysis

The data were normalized using the NormalizeData function and scaled with the ScaleData function. The FindVariableFeatures function was employed to identify highly variable genes within the samples. Non‐linear dimensionality reduction was visualized using t‐distributed stochastic neighbor embedding (t‐SNE) with a resolution parameter set to 0.5. 20 Batch effects across different patients were mitigated using the Harmony package in R. DEGs for each cluster were identified using Wilcoxon tests and the FindMarkers function, with thresholds set at logfc.threshold > 0.25 and p < 0.05. 21

2.3. Enrichment analysis

Gene ontology enrichment analysis (GOEA) is a bioinformatics method used to identify significantly overrepresented gene ontology (GO) terms within a specific set of genes or proteins. GO provides a standardized classification system for describing gene functions, biological processes, and cellular components. It is structured as a directed acyclic graph (DAG) with three main categories: molecular function, biological process, and cellular component. GOEA helps us understand the common features and functions of these genes or proteins. In this study, we employed the R package “clusterProfiler” to investigate the pathways and biological processes most significantly enriched among the DEGs. This analysis allows us to gain insights into the biological themes and functional characteristics associated with our gene sets. 22

2.4. Single‐cell trajectory analysis

Pseudo time analysis is a method used to infer and simulate the temporal order of cell development or dynamic processes. It involves sorting individual cells according to their temporal sequence to construct a trajectory of cell development. In our study, we performed pseudo time analysis on all CD8+ T cell subtypes to explore the different states of cells during the developmental process. This step was accomplished using the Monocle2 software package. 23 The “reduceDimension” function was used to reduce the dimensionality of the DEGs (method = “DDRTree”, max_components = 2). The “plot_cell_trajectory” function was employed to order and visualize the cells.

2.5. Cell−cell communication analysis

Cell−cell communication analysis is a method used to study the process of information transmission between cells through the secretion of signaling molecules, such as cytokines, hormones, and growth factors. In our study, cell−cell communication analysis was performed using CellPhoneDB. 24 CellPhoneDB is a database that focuses on ligands and their interactions. Unlike other databases, CellPhoneDB considers the subunit structures of ligands and accurately represents heteromeric complexes. It helps us understand how different cell types communicate through signal transduction. CellPhoneDB is an important tool in single‐cell transcriptomic research.

2.6. Survival analysis and protein−protein interaction (PPI)

The R package glmnet was used to construct the Lasso‐Cox regression model. We set L1 regularization and performed 10‐fold cross‐validation, using a stepwise regression method to select the features for the model. Patients were divided into low‐risk and high‐risk groups based on the median risk score. Survival analysis was conducted using the Kaplan−Meier method, and the log‐rank test was employed to compare survival differences between the high‐risk and low‐risk groups. A time‐dependent receiver operating characteristic (ROC) curve was generated. Finally, univariate and multivariate Cox regression analyses were performed on the risk scores and other clinical characteristics to confirm the independence of the risk score as a prognostic factor. PPI analysis was conducted using STRING (string‐db.org/).

2.7. Human subjects

This study was approved by the Institutional Review Board of Peking Union Medical College Hospital, with the approval number ZS‐2556. Written informed consent was obtained from all patients and donors. The design and implementation of the study complied with the Declaration of Helsinki. The study included two male patients with AM, who underwent skin biopsies. Tissue samples from AM lesions were transferred to the laboratory within 1 h for histopathological examination. Foreskin tissues from two individuals undergoing circumcision were obtained as normal control tissues.

2.8. Immunofluorescence staining

Skin tissue samples were fixed overnight in 4% formaldehyde at 4°C, followed by dehydration gradient processing. After paraffin embedding, the tissues were sectioned into 4 µm thick slices for immunofluorescence staining. For staining, the sections were deparaffinized and subjected to antigen retrieval by microwaving in antigen retrieval buffer for at least 12 min, followed by cooling at room temperature for 1–2 h and washing with PBS. The sections were then blocked with normal horse serum in PBS for 1 h and incubated with primary antibodies overnight at 4°C. Subsequently, the sections were incubated with secondary antibodies for 1 h, counterstained with DAPI, and mounted with a fluorescent mounting medium for imaging. Negative control samples were incubated with secondary antibodies alone. Images were captured using a ZEISS LSM880 high‐resolution laser confocal microscopy system at 20×/60× magnifications.

2.9. Statistical analysis

DEGs were identified using the Wilcoxon rank‐sum test, with a p‐value of less than 0.05 considered statistically significant. All analyses were performed using R software (version 4.1.2).

3. RESULTS

3.1. Heterogeneity of the immune microenvironment in AM

This study included five unmatched AM skin tissue samples and five normal skin tissue samples. After stringent quality screening, we obtained a total of 88 903 high‐quality cells from the cancer and normal skin tissue samples for further analysis. Batch effects were removed using Harmony, and t‐distributed stochastic neighbor embedding (t‐SNE) results revealed that the 88,903 cells were divided into 29 clusters (Figure 1A). Subsequently, based on known marker genes, the cells were categorized into three major types: epithelial cells (marker genes: EPCAM, KRT10), immune cells (marker genes: CD3D, CD68, PTPRC), and stromal cells (marker genes: PECAM1, VIM, COL1A1) (Figure 1B). We then performed subclustering of the immune cells and used SingleR for automatic annotation. The results indicated that the proportion of memory CD8+ T cells increased from 25.3% in normal skin samples to 70.5% in tumor samples, whereas macrophages decreased from 48.0% to 6.9%, granulocytes decreased from 20.4% to 16.6%, and HSC/MPP cells increased from 1.4% to 4.8% (Figure 1C and D, Table S1). The significant differences in cell composition between AM and normal skin tissue samples highlight the heterogeneity of the microenvironments in AM and normal skin tissues. We used receptor–ligand interaction analysis to infer intercellular communication relationships. The results showed that neutrophils, HSC/MPP cells, macrophages, and effector CD8+ T cells exhibited extensive communication relationships, which may play a crucial role in shaping the immune microenvironment of AM (Figure 1E). To further explore the developmental differentiation characteristics of CD8+ T cells in AM, we conducted pseudo‐time series analysis on development‐related CD8+ T cells using Monocle2. The pseudo‐time analysis revealed that HSC/MPP cells are located at the beginning of the trajectory path, while memory CD4+ T cells are situated at the terminal state (Figure 1F–H).

FIGURE 1.

FIGURE 1

Open in a new tab

(A) T‐SNE clustering of cells in all samples. (B) T‐SNE visualization of cell types in all samples. (C) Proportions of immune cells in AM and normal skin samples. (D) T‐SNE visualization of cell types in AM and normal skin samples. (E) Cell–cell communication heatmap. (F) Trajectory analysis results using Monocle2. (G) Monocle2 trajectory analysis results colored by different cell types. (H) Cell density changes over pseudo‐time.

3.2. Identification of DEGs from CD8+ T cells

Given the widespread increase in CD8+ T cells within tumor samples, we decided to further explore the differential genes of CD8+ T cells between tumor and adjacent non‐tumor samples. Ultimately, we identified 263 DEGs in CD8+ T cells (Table S2). GO enrichment analysis revealed that, in the biological process (BP) category, these genes were primarily enriched in adaptive immune response, immune response‐activating cell surface receptor signaling pathway, and immune response‐activating signal transduction (Figure 2A). In the cellular component (CC) category, they were predominantly enriched in immunoglobulin complex, external side of plasma membrane, and vacuolar membrane (Figure 2B). In the molecular function (MF) category, these marker genes were mainly enriched in antigen binding, immunoglobulin receptor binding, and organic acid binding (Figure 2C).

FIGURE 2.

FIGURE 2

Open in a new tab

(A) Enrichment results of DEGs in GO (BP). (B) Enrichment results of DEGs in GO (CC). (C) Enrichment results of DEGs in GO (MF). GO, gene ontology; DEGs, differentially expressed genes; BP, biological process; CC, cellular component; MF, molecular function.

3.3. Construction and validation of a prognostic model based on DEGs of CD8+ T cells

We downloaded the gene expression matrix and clinical information of patients from the TCGA‐SKCM cohort to construct a prognostic model. Differential genes of CD8+ T cells identified from single‐cell data were subjected to univariate Cox regression analysis to determine potential prognostic DEGs in the TCGA‐SKCM cohort (p < 0.05). Subsequently, Lasso regression analysis was performed with stepwise regression to reduce the number of DEGs in the final risk model. Ultimately, four genes—ISG20, CCL4, LPAR6, and DDIT3—were used to construct the prognostic risk score model. The risk score was calculated using the following formula: risk score = expression level of ISG20 * (−0.222) + expression level of CCL4 * (−0.532) + expression level of LPAR6 * (0.236) + expression level of DDIT3 * (0.467). Based on the median risk score, all patients were divided into high‐risk and low‐risk groups. Survival curves indicated that patients in the high‐risk group had worse overall survival (OS) compared to those in the low‐risk group (Figure 3A−D). Furthermore, the study demonstrated that the risk score exhibited good performance in predicting OS at 365 days, 548 days, and 730 days in the TCGA cohort (AUC: 0.814, 0.828, and 0.826, respectively) (Figure 3E).

FIGURE 3.

FIGURE 3

Open in a new tab

(A) Expression levels of four molecules in the TCGA‐SKCM cohort based on the Lasso‐Cox model. (B) Correlation of expression among the four molecules in the TCGA‐SKCM cohort. (C) Distribution of the risk scores in the TCGA‐SKCM cohort. (D) Kaplan−Meier survival analysis based on the risk scores in the TCGA‐SKCM cohort. (E) ROC curve and AUC of the risk score predictions for the TCGA‐SKCM cohort.

3.4. Development and validation of a model incorporating clinical features and corresponding nomogram

Next, we carried out univariate and multivariate Cox analyses to determine whether the risk score could serve as an independent prognostic factor for AM patients compared to other common clinical indicators. Univariate Cox regression analysis revealed that the risk score could act as an independent prognostic factor for AM and was positively correlated with overall survival (OS) (HR: 2.768, 95% CI: 1.699−4.510, p < 0.001) (Figure 4A). The results of multivariate Cox analysis, which included clinical characteristics, showed that the prognostic risk score was significantly associated with OS in AM patients and could serve as an independent prognostic factor (p < 0.01). Additionally, ROC curves indicated that the AUC for predicting OS at 365 days, 548 days, and 730 days were 0.840, 0.852, and 0.842, respectively (Figure 4B). Finally, we constructed a nomogram for clinical use (Figure 4C). To elucidate the key functions performed by the molecules used to construct the prognostic model in AM patients, we further conducted a PPI analysis and identified their protein interaction relationships. CCL4 and ISG20 showed a close connection, with proteins interacting with CCL4 and ISG20 being mostly related to immune system function, antiviral mechanisms, and inflammatory responses. DDIT3 and LPAR6 had no direct connections; proteins interacting with LPAR6 were primarily associated with skin and hair structure, lipid metabolism, and cell signal transduction, while those interacting with DDIT3 were related to endoplasmic reticulum stress, cell differentiation, proliferation, and inflammatory responses (Figure 4D).

FIGURE 4.

FIGURE 4

Open in a new tab

(A) Hazard ratios of risk scores and clinical characteristics. (B) ROC curve of the Lasso‐Cox model combined with clinical data. (C) Nomogram combining risk scores and clinical data. (D) PPI network of the four molecules in the Lasso‐Cox model. ROC, receiver operating characteristic; PPI, protein−protein interaction.

3.5. Immunofluorescence validation of four prognostic signatures

Lastly, we analyzed the expression of the four key genes in the prognostic model. Compared to normal skin tissue, the expression levels of ISG20 and CCL4 (Figure 5A and B) were downregulated in AM tissue, while LPAR6 and DDIT3 were upregulated in AM tissue (Figure 6A and B). The staining results provide support for the rationality of the regression coefficients in our model.

FIGURE 5.

FIGURE 5

Open in a new tab

(A) Immunofluorescence staining of ISG20 in melanoma tissue and normal skin tissue. (B) Immunofluorescence staining of CCL4 in melanoma tissue and normal skin tissue. The left rectangular image has a scale bar of 100 µm; the right square image has a scale bar of 10 µm. ISG20 is labeled in red, CCL4 is labeled in red, HMB‐45 is labeled in green, and DAPI is labeled in blue.

FIGURE 6.

FIGURE 6

Open in a new tab

(A) Immunofluorescence staining of LPAR6 in melanoma tissue and normal skin tissue. (B) Immunofluorescence staining of DDIT3 in melanoma tissue and normal skin tissue. The left rectangular image has a scale bar of 100 µm; the right square image has a scale bar of 10 µm. LPAR6 is labeled in red, DDIT3 is labeled in red, HMB‐45 is labeled in green, and DAPI is labeled in blue.

4. DISCUSSION

AM is more prevalent in Asian populations and exhibits more aggressive biological behavior along with poorer responses to immunotherapy. Early diagnosis and treatment could potentially bring clinical benefits to patients with AM. Consequently, there is an urgent need to identify reliable and effective prognostic biomarkers for these patients. The advent of scRNA‐seq technology has introduced new methods for exploring the molecular characteristics of immune cells within the TME. With the clinical application of immunotherapy, genes characteristic of immune cells has shown promising prognostic performance. Screening tumor prognostic features based on these immune cells provides new avenues for personalized immunotherapy in AM, ultimately improving patient clinical outcomes. In this study, we utilized scRNA‐seq data to characterize the immune microenvironment of AM and identified differential genes in CD8+ T cells to develop a prognostic model. Additionally, we reconstructed the developmental trajectory of CD8+ T cells. A novel prognostic model was created using the TCGA‐SKCM cohort, with AUCs for overall survival at 365 days, 548 days, and 730 days being 0.814, 0.828, and 0.826, respectively. It was confirmed that the risk score could serve as an independent prognostic factor.

CD8+ T lymphocytes play a crucial role in cancer immunity and non‐tumor immunity. 25 , 26 These cells are among the primary immune effector cells of the body and can directly recognize and kill tumor cells. CD8+ T cells identify specific antigens on the surface of tumor cells through their T‐cell receptors (TCR) and launch targeted attacks. 27 They can directly release cytotoxic molecules such as perforin and granzymes to disrupt tumor cell membranes and induce apoptosis, and they also secrete cytokines like gamma‐interferon to enhance the anti‐tumor activity of other immune cells, such as macrophages and natural killer cells. 28 , 29 Moreover, in cancer immunotherapy, activating and enhancing the function of CD8+ T cells is a key strategy. For instance, immune checkpoint inhibitors (such as PD‐1/PD‐L1 and CTLA‐4 inhibitors) work by lifting the inhibition on CD8+ T cells, restoring their ability to effectively attack tumor cells. 30 , 31 CAR‐T cell therapy also involves genetically engineering CD8+ T cells to more effectively recognize and kill tumor cells. 32 In melanoma, CD8+ T cells play a critical role as well. 33 Research has shown that CD8+ T cells can recognize and attack melanoma cells. When the immune system functions properly, CD8+ T cells can target melanoma cells by recognizing specific antigenic proteins on their surface, ultimately leading to the death of these cells. This process, known as immune clearance, is crucial for controlling the growth and spread of melanoma. 33 However, melanoma cells can employ various mechanisms to evade CD8+ T cell attacks, such as modulating antigen presentation and reducing MHC‐I molecule expression. These mechanisms can increase the resistance of melanoma cells to the immune system, allowing the tumor to survive and continue growing. 34

We have identified four CD8+ T cell hub genes: ISG20, CCL4, LPAR6, and DDIT3. In the prognostic feature model, we found that LPAR6 and DDIT3 were associated with poor outcomes in AM patients, while ISG20 and CCL4 were found to have protective effects on the prognosis of AM patients. Lysophosphatidic acid (LPA) is a lipid molecule involved in tumor proliferation, and LPAR6, one of its receptors, is the most recently identified G protein‐coupled receptor (GPCR) in the LPA family. 35 , 36 This receptor has been linked to various types of cancer, including colorectal cancer, prostate cancer, pancreatic cancer, and liver cancer. 37 − 40 However, the role of LPAR6 remains controversial. In colorectal cancer, LPAR6 acts as a tumor suppressor and inhibits tumor migration, while in other cancers, it appears to promote tumor development. 40 These findings suggest that the protein encoded by LPAR6 may play a significant role in cancer biology. Nevertheless, the relationship between LPAR6 and tumor progression, as well as the underlying mechanisms, are still not fully elucidated. DNA damage‐inducible transcript 3 (DDIT3) encodes a protein that that is part of the CCAAT/enhancer‐binding protein family of transcription factors. This protein acts as a transcription factor, regulating gene expression, cell growth, differentiation, and energy metabolism. DDIT3 is pivotal in apoptosis induced by endoplasmic reticulum (ER) stress. 41 It manages cell proliferation under ER stress by forming dimers with other transcription factors that have basic leucine zipper (bZIP) structural features. In stressful situations, the transcription of DDIT3 and other transcription factors is regulated through a complex network involving bZIP transcription factors, where DDIT3 and ATF3 inhibit each other's transcription. 42 , 43 Overexpressing DDIT3 can cause cell cycle arrest and apoptosis, highlighting its central role in stress responses. 44 − 46 Furthermore, DDIT3 is frequently overexpressed in various tumor types, often due to amplification of the chromosomal region 12q13. 47 High levels of DDIT3 expression in CD8+ T cells infiltrating tumors are associated with poor clinical outcomes in ovarian cancer patients. Mechanistically, it has been demonstrated that increased DDIT3 expression diminishes tbx21 transcription in tumor‐infiltrating CD8+ T cells, hindering their effector immunity and suggesting a regulatory effect on their antitumor activity. 48 Lin et al. observed significantly elevated DDIT3 levels in gastric cancer. 49 Zhang et al. found that DDIT3 also contributes to prostate cancer progression, including invasion capability and cell proliferation. Modulating DDIT3 expression in prostate cancer tissues could be a promising therapeutic target for both prostate cancer and castration‐resistant prostate cancer. 50 Additionally, Lin et al. identified that DDIT3 is involved in cancer stem cell regulation. Overexpression of DDIT3 enhances stemness in gastric cancer by regulating CEBPβ. 49

CCL4, also referred to as macrophage inflammatory protein 1β (MIP‐1β), serves as a pivotal pro‐inflammatory chemokine essential for initiating immune responses in humans. 51 Through interaction with its specific receptor CCR5, CCL4 collaborates with related but distinct chemokines like CCL3 and CCL5 to elicit diverse effects on both immune and non‐immune cells. Its actions include the recruitment of dendritic cells, neutrophils, monocytes, macrophages, NK cells, and T cells to sites of inflammation. 51 The escalation in the number of immune cells prompted by CCL4 renders it a potentially significant component in cancer immunotherapy. 52 − 54 Many evidence indicates that CCL4 may contribute to tumor development and progression by recruiting regulatory T cells and pro‐tumor macrophages, and by influencing other cells present in the TME, such as fibroblasts and endothelial cells, thus enhancing their pro‐tumor capabilities. Conversely, in certain contexts, CCL4 can enhance tumor immunity by recruiting cytotoxic lymphocytes and phagocytic macrophages. 55 In tumors, active immune responses lead to the production of CCL4 by B cells and basophils. 56 , 57 The chemokines produced can serve as chemoattractants for CCR5‐positive anti‐cancer tumor‐infiltrating lymphocytes (TILs). 58 − 61 Currently, no reports indicate that CCL4 recruits tumor‐associated macrophages (TAMs); instead, they bolster the anti‐cancer functions of monocytes, 62 suggesting they do not attract TAMs but rather facilitate the infiltration of anti‐cancer M1 macrophages. Interferon‐stimulated exonuclease gene. 20 (ISG20) is a gene targeted by IFN‐γ, and is situated on human chromosome 15q26. It encodes a 181‐amino acid protein with RNase activity that combats viruses. 63 The protein sequence produced by ISG20 is highly preserved across various species. 64 ISG20 plays a vital role in the degradation of RNA and DNA. 65 In the realm of cancer research, ISG20 mRNA expression levels were notably elevated in 11 different types of cancer, such as adrenocortical carcinoma, cervical squamous cell carcinoma, diffuse large B‐cell lymphoma, glioblastoma multiforme, kidney renal clear cell carcinoma, kidney renal papillary cell carcinoma, liver hepatocellular carcinoma, pancreatic adenocarcinoma, skin cutaneous melanoma, testicular germ cell tumors, and uterine corpus endometrial carcinoma. Among these, heightened ISG20 expression correlated with prolonged overall survival in cervical squamous cell carcinoma and melanoma, indicating ISG20's potential as a promising biomarker for cancer progression in patients with these types of cancer. 66 Furthermore, we observed that ISG20 serves as a protective gene for survival and is associated with T cells and dendritic cells. We hypothesize that ISG20 might influence overall survival in AM cancer patients by activating T cells or dendritic cells and thus participating in immune responses. 67

This study has several limitations. First, the predictive ability of the prognostic model based on DEGs in CD8+ T cells need to be further validated in large‐scale prospective clinical studies. Second, the mechanisms by which the prognostic genes exert their effects were not explored in this study.

5. CONCLUSION

In this study, we constructed a comprehensive map of the immune microenvironment in AM by integrating scRNA‐seq and bulk RNA sequencing data. We characterized the AM immune microenvironment in terms of composition, developmental differentiation, function, and cellular communication. Furthermore, we developed a prognostic regression model at the single‐cell level and validated the independent prognostic value of the risk scores. Finally, we confirmed the expression of four CD8+ T cell‐derived genes through immunofluorescence staining of both AM samples and normal skin samples. Our findings provide novel insights into the prognosis of AM patients and highlight the critical role of CD8+ T cells in AM. Additionally, we suggest that therapeutic targets for AM can be developed through pathways associated with CD8+ T cell‐related genes. AM patients may benefit from further screening of anticancer drugs that are sensitive to high and low groups of CD8+ T cell‐related genes.

Supporting information

Supporting information

SRT-30-e13900-s001.xlsx (9.8KB, xlsx)

Supporting information

SRT-30-e13900-s002.xlsx (28.4KB, xlsx)

ACKNOWLEDGMENTS

We thank the bioinformatics platform and imaging facility of National Center for Protein Sciences (Beijing) for providing the computational resource for single‐cell RNA sequencing data analysis and imaging analysis assistance. This work was supported by the CAMS Innovation Fund for Medical Sciences (CIFMS) (2021‐I2M‐1‐052, 2023‐I2M‐3‐002, and 2023‐I2M‐QJ‐001) and the National High Level Hospital Clinical Research Funding (2022‐PUMCH‐B‐061 and 2022‐PUMCH‐A‐022).

Wang W, Liu P, Ma J, Li J, Leng L. Establishment of a CD8+ T cells‐related prognostic risk model for acral melanoma based on single‐cell and bulk RNA sequencing. Skin Res Technol. 2024;30:e13900. 10.1111/srt.13900

Wenwen Wang and Pu Liu contributed equally to this study.

Contributor Information

Jie Ma, Email: majie@ncpsb.org.cn.

Jun Li, Email: lijun35@hotmail.com.

Ling Leng, Email: zhenlinger@126.com.

DATA AVAILABILITY STATEMENT

The data that support the findings of this study are available in The Cancer Genome Atlas Program at https://www.cancer.gov/ccg/research/genome‐sequencing/tcga

REFERENCES

  • 1. Gajón JA, Juarez‐Flores A, De León Rodríguez SG, et al. Immunotherapy options for acral melanoma, a fast‐growing but neglected malignancy. Arch Med Res. 2022;53(8):794‐806. [DOI] [PubMed] [Google Scholar]
  • 2. Garg M, Couturier DL, Nsengimana J, et al. Tumour gene expression signature in primary melanoma predicts long‐term outcomes. Nat Commun. 2021;12(1):1137. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 3. Arnold M, Singh D, Laversanne M, et al. Global burden of cutaneous melanoma in 2020 and projections to 2040. JAMA Dermatol. 2022;158(5):495‐503. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 4. Schadendorf D, van Akkooi ACJ, Berking C, et al. Melanoma. Lancet (London, England). 2018;392(10151):971‐984. [DOI] [PubMed] [Google Scholar]
  • 5. Wu XC, Eide MJ, King J, et al. Racial and ethnic variations in incidence and survival of cutaneous melanoma in the United States, 1999–2006. J Am Acad Dermatol. 2011;65( Suppl 51):S26‐37. [DOI] [PubMed] [Google Scholar]
  • 6. Huang K, Fan J, Misra S. Acral lentiginous melanoma: incidence and survival in the United States, 2006–2015, an analysis of the SEER registry. J Surg Res. 2020;251:329‐339. [DOI] [PubMed] [Google Scholar]
  • 7. Gumaste PV, Fleming NH, Silva I, et al. Analysis of recurrence patterns in acral versus nonacral melanoma: should histologic subtype influence treatment guidelines? J Natl Compreh Cancer Netw. 2014;12(12):1706‐1712. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8. Chang AE, Karnell LH, Menck HR. The National Cancer Data Base report on cutaneous and noncutaneous melanoma: a summary of 84,836 cases from the past decade. The American College of Surgeons Commission on Cancer and the Am Cancer Soc. Cancer. 1998;83(8):1664‐1678. [DOI] [PubMed] [Google Scholar]
  • 9. Hartman RI, Lin JY. Cutaneous melanoma – a review in detection, staging, and management. Hematol Oncol Clin N Am. 2019;33(1):25‐38. [DOI] [PubMed] [Google Scholar]
  • 10. Chi Z, Li S, Sheng X, et al. Clinical presentation, histology, and prognoses of malignant melanoma in ethnic Chinese: a study of 522 consecutive cases. BMC Cancer. 2011;11:85. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11. Luke JJ, Flaherty KT, Ribas A, Long GV. Targeted agents and immunotherapies: optimizing outcomes in melanoma. Nat Rev Clin Oncol. 2017;14(8):463‐482. [DOI] [PubMed] [Google Scholar]
  • 12. Shoushtari AN, Munhoz RR, Kuk D, et al. The efficacy of anti‐PD‐1 agents in acral and mucosal melanoma. Cancer. 2016;122(21):3354‐3362. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13. Liu T, Liu JP. Integration analysis of single‐cell transcriptome reveals specific monocyte subsets associated with melanoma brain and leptomeningeal metastasis. Skin Res Technol. 2024;30(4):e13710. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14. Gartrell RD, Marks DK, Hart TD, et al. Quantitative analysis of immune infiltrates in primary melanoma. Cancer Immunol Res. 2018;6(4):481‐493. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15. Cabrita R, Lauss M, Sanna A, et al. Tertiary lymphoid structures improve immunotherapy and survival in melanoma. Nature. 2020;577(7791):561‐565. [DOI] [PubMed] [Google Scholar]
  • 16. Wong PF, Wei W, Smithy JW, et al. Multiplex quantitative analysis of tumor‐infiltrating lymphocytes and immunotherapy outcome in metastatic melanoma. Clin Cancer Res. 2019;25(8):2442‐2449. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 17. Zhang C, Shen H, Yang T, et al. A single‐cell analysis reveals tumor heterogeneity and immune environment of acral melanoma. Nat Commun. 2022;13(1):7250. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 18. Huang R, Shen G, Ren Y, et al. Prognostic value of genetic aberrations and tumor immune microenvironment in primary acral melanoma. J Transl Med. 2023;21(1):78. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19. Giorgi FM, Ceraolo C, Mercatelli D. The R language: An Engine for Bioinformatics and Data Science. Life (Basel, Switzerland). 2022;12(5). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20. Zhou B, Jin W. Visualization of single cell RNA‐Seq data using t‐SNE in R. Methods Mol Biol (Clifton, NJ). 2020;2117:159‐167. [DOI] [PubMed] [Google Scholar]
  • 21. Ntranos V, Yi L, Melsted P, Pachter L. A discriminative learning approach to differential expression analysis for single‐cell RNA‐seq. Nat Methods. 2019;16(2):163‐166. [DOI] [PubMed] [Google Scholar]
  • 22. Yu G, Wang LG, Han Y, He QY. clusterProfiler: an R package for comparing biological themes among gene clusters. Omics J Integr Biol. 2012;16(5):284‐287. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 23. Trapnell C, Cacchiarelli D, Grimsby J, et al. The dynamics and regulators of cell fate decisions are revealed by pseudotemporal ordering of single cells. Nat Biotechnol. 2014;32(4):381‐386. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 24. Efremova M, Vento‐Tormo M, Teichmann SA, Vento‐Tormo R. CellPhoneDB: inferring cell−cell communication from combined expression of multi‐subunit ligand‐receptor complexes. Nat Protoc. 2020;15(4):1484‐1506. [DOI] [PubMed] [Google Scholar]
  • 25. van der Leun AM, Thommen DS, Schumacher TN. CD8(+) T cell states in human cancer: insights from single‐cell analysis. Nat Rev Cancer. 2020;20(4):218‐232. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 26. Liu L, Xue Y, Wang J, Shen C, Li Y, Huang Y. Transcriptome analysis of frontal fibrosis alopecia revealed involvement of immune cells and ferroptosis. Skin Res Technol. 2024;30(2):e13608. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 27. Bossi G, Trambas C, Booth S, Clark R, Stinchcombe J, Griffiths GM. The secretory synapse: the secrets of a serial killer. Immunol Rev. 2002;189:152‐160. [DOI] [PubMed] [Google Scholar]
  • 28. Anikeeva N, Somersalo K, Sims TN, Thomas VK, Dustin ML, Sykulev Y. Distinct role of lymphocyte function‐associated antigen‐1 in mediating effective cytolytic activity by cytotoxic T lymphocytes. Proc Natl Acad Sci U. S. A. 2005;102(18):6437‐6442. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 29. Le Floc'h A, Jalil A, Vergnon I, et al. Alpha E beta 7 integrin interaction with E‐cadherin promotes antitumor CTL activity by triggering lytic granule polarization and exocytosis. J Exp Med. 2007;204(3):559‐570. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 30. Tran E, Turcotte S, Gros A, et al. Cancer immunotherapy based on mutation‐specific CD4+ T cells in a patient with epithelial cancer. Science (New York, NY). 2014;344(6184):641‐645. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 31. Fourcade J, Kudela P, Sun Z, et al. PD‐1 is a regulator of NY‐ESO‐1‐specific CD8+ T cell expansion in melanoma patients. J Immunol. 2009;182(9):5240‐5249. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 32. Chmielewski M, Hombach AA, Abken H. Antigen‐specific T‐cell activation independently of the MHC: chimeric antigen receptor‐redirected T cells. Front Immunol. 2013;4:371. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 33. Rosenberg SA, Yang JC, Restifo NP. Cancer immunotherapy: moving beyond current vaccines. Nat Med. 2004;10(9):909‐915. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 34. Restifo NP, Smyth MJ, Snyder A. Acquired resistance to immunotherapy and future challenges. Nat Rev Cancer. 2016;16(2):121‐126. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 35. Shimomura Y, Wajid M, Ishii Y, et al. Disruption of P2RY5, an orphan G protein‐coupled receptor, underlies autosomal recessive woolly hair. Nat Genet. 2008;40(3):335‐339. [DOI] [PubMed] [Google Scholar]
  • 36. Takahashi K, Fukushima K, Onishi Y, et al. Lysophosphatidic acid (LPA) signaling via LPA(4) and LPA(6) negatively regulates cell motile activities of colon cancer cells. Biochem Biophys Res Commun. 2017;483(1):652‐657. [DOI] [PubMed] [Google Scholar]
  • 37. Sokolov E, Eheim AL, Ahrens WA, et al. Lysophosphatidic acid receptor expression and function in human hepatocellular carcinoma. J Surg Res. 2013;180(1):104‐113. [DOI] [PubMed] [Google Scholar]
  • 38. Ketscher A, Jilg CA, Willmann D, et al. LSD1 controls metastasis of androgen‐independent prostate cancer cells through PXN and LPAR6. Oncogenesis. 2014;3(10):e120. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 39. Ishii S, Hirane M, Fukushima K, Tomimatsu A, Fukushima N, Tsujiuchi T. Diverse effects of LPA4, LPA5 and LPA6 on the activation of tumor progression in pancreatic cancer cells. Biochem Biophys Res Commun. 2015;461(1):59‐64. [DOI] [PubMed] [Google Scholar]
  • 40. Rhodes DR, Kalyana‐Sundaram S, Mahavisno V, et al. Oncomine 3.0: genes, pathways, and networks in a collection of 18,000 cancer gene expression profiles. Neoplasia (New York, NY). 2007;9(2):166‐180. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 41. Li Y, Guo Y, Tang J, Jiang J, Chen Z. New insights into the roles of CHOP‐induced apoptosis in ER stress. Acta Biochim Biophys Sin. 2014;46(8):629‐640. [DOI] [PubMed] [Google Scholar]
  • 42. Chen BP, Wolfgang CD, Hai T. Analysis of ATF3, a transcription factor induced by physiological stresses and modulated by gadd153/Chop10. Mol Cell Biol. 1996;16(3):1157‐1168. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 43. Fawcett TW, Martindale JL, Guyton KZ, Hai T, Holbrook NJ. Complexes containing activating transcription factor (ATF)/cAMP‐responsive‐element‐binding protein (CREB) interact with the CCAAT/enhancer‐binding protein (C/EBP)‐ATF composite site to regulate Gadd153 expression during the stress response. Biochem J. 1999;339(Pt 1):135‐141. [PMC free article] [PubMed] [Google Scholar]
  • 44. Jauhiainen A, Thomsen C, Strömbom L, et al. Distinct cytoplasmic and nuclear functions of the stress induced protein DDIT3/CHOP/GADD153. PloS One. 2012;7(4):e33208. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 45. Barone MV, Crozat A, Tabaee A, Philipson L, Ron D. CHOP (GADD153) and its oncogenic variant, TLS‐CHOP, have opposing effects on the induction of G1/S arrest. Genes Dev. 1994;8(4):453‐464. [DOI] [PubMed] [Google Scholar]
  • 46. Wang XZ, Ron D. Stress‐induced phosphorylation and activation of the transcription factor CHOP (GADD153) by p38 MAP Kinase. Science (New York, NY). 1996;272(5266):1347‐1349. [DOI] [PubMed] [Google Scholar]
  • 47. Persson F, Olofsson A, Sjögren H, et al. Characterization of the 12q amplicons by high‐resolution, oligonucleotide array CGH and expression analyses of a novel liposarcoma cell line. Cancer Lett. 2008;260(1‐2):37‐47. [DOI] [PubMed] [Google Scholar]
  • 48. Cao Y, Trillo‐Tinoco J, Sierra RA, et al. ER stress‐induced mediator C/EBP homologous protein thwarts effector T cell activity in tumors through T‐bet repression. Nat Commun. 2019;10(1):1280. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 49. Lin H, Liu S, Gao W, Liu H. DDIT3 modulates cancer stemness in gastric cancer by directly regulating CEBPβ. J Pharm Pharmacol. 2020;72(6):807‐815. [DOI] [PubMed] [Google Scholar]
  • 50. Zhang X, Lu Z, Ren X, et al. Genetic comprehension of organophosphate flame retardants, an emerging threat to prostate cancer. Ecotoxicol Environ Saf. 2021;223:112589. [DOI] [PubMed] [Google Scholar]
  • 51. Menten P, Wuyts A, Van Damme J. Macrophage inflammatory protein‐1. Cytok Growth Factor Rev. 2002;13(6):455‐481. [DOI] [PubMed] [Google Scholar]
  • 52. Luo X, Yu Y, Liang A, et al. Intratumoral expression of MIP‐1beta induces antitumor responses in a pre‐established tumor model through chemoattracting T cells and NK cells. Cell Mol Immunol. 2004;1(3):199‐204. [PubMed] [Google Scholar]
  • 53. Schaller TH, Batich KA, Suryadevara CM, Desai R, Sampson JH. Chemokines as adjuvants for immunotherapy: implications for immune activation with CCL3. Exp Rev Clin Immunol. 2017;13(11):1049‐1060. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 54. Williford JM, Ishihara J, Ishihara A, et al. Recruitment of CD103(+) dendritic cells via tumor‐targeted chemokine delivery enhances efficacy of checkpoint inhibitor immunotherapy. Sci Adv. 2019;5(12):eaay1357. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 55. Mukaida N, Sasaki SI, Baba T. CCL4 signaling in the tumor microenvironment. Adv Exp Med Biol. 2020;1231:23‐32. [DOI] [PubMed] [Google Scholar]
  • 56. Griss J, Bauer W, Wagner C, et al. B cells sustain inflammation and predict response to immune checkpoint blockade in human melanoma. Nat Commun. 2019;10(1):4186. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 57. Sektioglu IM, Carretero R, Bulbuc N, et al. Basophils promote tumor rejection via chemotaxis and infiltration of CD8+ T cells. Cancer Res. 2017;77(2):291‐302. [DOI] [PubMed] [Google Scholar]
  • 58. Iida N, Nakamoto Y, Baba T, et al. Tumor cell apoptosis induces tumor‐specific immunity in a CC chemokine receptor 1‐ and 5‐dependent manner in mice. J Leuk Biol. 2008;84(4):1001‐1010. [DOI] [PubMed] [Google Scholar]
  • 59. González‐Martín A, Gómez L, Lustgarten J, Mira E, Mañes S. Maximal T cell‐mediated antitumor responses rely upon CCR5 expression in both CD4(+) and CD8(+) T cells. Cancer Res. 2011;71(16):5455‐5466. [DOI] [PubMed] [Google Scholar]
  • 60. Allen F, Rauhe P, Askew D, et al. CCL3 enhances antitumor immune priming in the lymph node via IFNγ with dependency on natural killer cells. Front Immunol. 2017;8:1390. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 61. Zhao N, Dang H, Ma L, et al. Intratumoral γδ T‐cell infiltrates, chemokine (C‐C motif) ligand 4/chemokine (C‐C motif) ligand 5 protein expression and survival in patients with hepatocellular carcinoma. Hepatol (Baltimore, Md). 2021;73(3):1045‐1060. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 62. Nath A, Chattopadhya S, Chattopadhyay U, Sharma NK. Macrophage inflammatory protein (MIP)1alpha and MIP1beta differentially regulate release of inflammatory cytokines and generation of tumoricidal monocytes in malignancy. Cancer Immunol Immunother: CII. 2006;55(12):1534‐1541. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 63. Chattopadhyay A, Joseph JP, Jagdish S, et al. High throughput screening identifies auranofin and pentamidine as potent compounds that lower IFN‐γ‐induced Nitric Oxide and inflammatory responses in mice: DSS‐induced colitis and Salmonella typhimurium‐induced sepsis. Int Immunopharmacol. 2023;122:110569. [DOI] [PubMed] [Google Scholar]
  • 64. Liao KL, Watt KD. Mathematical modeling for the combination treatment of IFN‐γ and anti‐PD‐1 in cancer immunotherapy. Math Biosci. 2022;353:108911. [DOI] [PubMed] [Google Scholar]
  • 65. Cai J, Zhong M, Xu J, Cheng H, Xu S. Codelivery of triptolide and IFN‐γ to boost antitumor immunity for triple‐negative breast cancer. Int Immunopharmacol. 2023;120:110346. [DOI] [PubMed] [Google Scholar]
  • 66. Cheng J, Fu J, Tan Q, et al. The regulation of ISG20 expression on SARS‐CoV‐2 infection in cancer patients and healthy individuals. Front Immunol. 2022;13:958898. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 67. Qu X, Shi Z, Guo J, Guo C, Qiu J, Hua K. Identification of a novel six‐gene signature with potential prognostic and therapeutic value in cervical cancer. Cancer Med. 2021;10(19):6881‐6896. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supporting information

SRT-30-e13900-s001.xlsx (9.8KB, xlsx)

Supporting information

SRT-30-e13900-s002.xlsx (28.4KB, xlsx)

Data Availability Statement

The data that support the findings of this study are available in The Cancer Genome Atlas Program at https://www.cancer.gov/ccg/research/genome‐sequencing/tcga

Articles from Skin Research and Technology are provided here courtesy of International Society of Biophysics and Imaging of the Skin, International Society for Digital Imaging of the Skin, and John Wiley & Sons Ltd

RESOURCES