CD20+FCRL4+ B Cells Activate CD8+ T Cells via MHC‐I Restriction in Nasopharyngeal Carcinoma Anti‐Tumor Immunity
Benjian Zhang, Xiaotian Yuan, Kelei Gao, Bo You, Yaxuan Wang, Lai Meng, Zirong Chen, Shaobing Xie, Yunqing Liu, Zijian Dong, Shumin Xie, Ruohao Fan, Fengjun Wang, Junyi Zhang, Zhihai Xie, Yongzhen Mo, Hua Zhang, Weihong Jiang
TL;DR
A specific type of B cell in nasopharyngeal cancer helps activate T cells to fight tumors, possibly improving immunotherapy outcomes.
Contribution
Identifies CD20+FCRL4+ B cells as activators of CD8+ T cells via MHC-I in nasopharyngeal carcinoma, revealing a novel antitumor mechanism.
Findings
CD20+FCRL4+ B cells co-localize with CD8+ T cells and enhance their activation and cytotoxicity.
IFNγ and WDFY4 upregulation improve B cell delivery of MHC-I epitopes to T cells.
Higher infiltration of CD20+FCRL4+ B cells correlates with better response to anti-PD-1 therapy.
Abstract
Nasopharyngeal carcinoma (NPC) is a malignant tumor characterized by extensive immune cell infiltration. However, the function and significance of B cells in NPC have been overlooked. Exploring B cells and their interactions with other immune cells will provide deeper insights into the immune microenvironment of NPC and theories of immunotherapy.We utilized single‐cell sequencing data to identify characteristic B cell subtypes of NPC. Subsequently, the presence of the CD20+FCRL4+ B cell subpopulation was validated in NPC samples using immunohistochemistry and flow cytometry. The interaction between this B cell subpopulation and CD8+ T cells was investigated by establishing an in vitro and in vivo co‐culture system.Our analysis revealed a subset of CD20+FCRL4+ B cells that may interact with CD8+ T cells through the MHC‐I pathway. Furthermore, we observed a co‐localized distribution of…
Genes, proteins, chemicals, diseases, species, mutations and cell lines named across the full text — each resolved to its canonical identifier and authoritative record.
Click any figure to enlarge with its caption.
FIGURE 1
FIGURE 2
FIGURE 3
FIGURE 4
FIGURE 5
FIGURE 6
FIGURE 7
FIGURE 8
FIGURE 9- —National Natural Science Foundation of China10.13039/501100001809
- —Central South University Graduate Students Independent Exploration and Innovation Project
- —Natural Science Foundation of Hunan Province10.13039/501100004735
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsCancer Immunotherapy and Biomarkers · Immunotherapy and Immune Responses · Immune Cell Function and Interaction
Introduction
1
Nasopharyngeal carcinoma (NPC) is a squamous cell carcinoma originating from the nasopharyngeal epithelium [1]. It exhibits significant geographic clustering, with over 70% of cases occurring in southern China and Southeast Asia [2, 3]. Currently, the first‐line treatment choice for NPC is a combination of radiotherapy and chemotherapy [1]. However, approximately 30%–40% of patients experience distant metastasis and local recurrence, resulting in suboptimal outcomes [4, 5]. In recent years, immunotherapy has brought about a revolution in the clinical management of various tumors. The notable immune infiltration and relatively low degree of differentiation observed in NPC set it apart from other malignancies [6], suggesting a favorable response to immune checkpoint inhibitors (ICIs) targeting the programmed cell death protein 1/programmed death‐ligand 1 (PD‐1/PD‐L1) pathway blockade. Indeed, ICIs based on PD‐1/PD‐L1 blockade have demonstrated clinical benefits in patients with NPC. In 2021, they were approved in China as first‐line agents in combination with gemcitabine and cisplatin (GP) for refractory recurrence and/or metastatic NPC (R/M NPC) [7]. However, the effective rate of PD‐1 therapy in patients with NPC is only 20%–30% [8], emphasizing the unmet clinical needs and the urgent requirement for a deeper understanding of the tumor microenvironment (TME) in NPC. This understanding would facilitate the identification of therapeutic targets and reliable biomarkers for risk stratification [9, 10].
Over the past few decades, researchers have uncovered that tumor‐infiltrating lymphocytes (TILs), considered the most critical constituents of the TME, play a significant role in immune regulation in NPC research [11]. In recent years, high‐density TILs have been observed within heterogeneous tumor cells and other stromal cells in NPC biopsy tissues, with single‐cell resolution, and this observation has been associated with favorable survival outcomes in patients with NPC [12, 13]. Consequently, an increasing number of subsets of tumor‐infiltrating immune cells have been recognized as promising biomarkers and potential therapeutic targets. While previous studies on NPC‐related immune cells have primarily concentrated on T cells, which can elicit high levels of antigen‐specific cytotoxic responses [14], other immune cell populations, such as lymphocytes, myeloid cells, and fibroblasts residing within the NPC microenvironment, have not yet received full recognition and characterization. Recent evidence indicates that B cells are emerging factors that influence the efficacy of immunotherapy in various malignancies [15]. However, the phenotype and function of infiltrating B cells in NPC remain unclear.
Among all the subtypes of infiltrating stromal cells in the TME, B cells have received the least attention due to their relatively low infiltration in many malignant tumors. However, NPC exhibits distinct differences in the TME compared to other head and neck tumors, which may contribute to varying prognoses. For instance, in head and neck squamous cell carcinoma (HNSCC), B cells constitute approximately 10% of immune cells, whereas in NPC, they make up about 40% [16, 17]. As the second most abundant TILs in NPC, B cells play a critical role in modulating the tumor immune response through various pathways of immune regulation [18]. In addition to their traditional immune functions, recent studies have indicated that B cells can aggregate at the tumor periphery and form complex tumor‐associated immune structures known as tertiary lymphoid structures (TLSs), which exert anti‐tumor effects and are associated with a favorable prognosis. However, tumor‐infiltrating B lymphocytes themselves rarely participate in the anti‐tumor response; they are typically found in close association with T cells, myeloid cells, and other immune cell types. Most studies have focused on the interaction between T and B cells. Numerous studies have shown that B cells interact with CD4^+^ T cells, promoting their activation, differentiation, and proliferation. In turn, CD4^+^ T cells provide signals to B cells, influencing their class switching and functional activities [19]. However, the interaction between B cells and CD8^+^ T cells remains poorly understood. Some researchers have suggested that the CXCL13‐CXCR5 axis may explain the interplay between these two cell types. In high‐grade serous ovarian cancer, tumors with high CXCL13 expression exhibit increased activation and infiltration of CXCR5‐expressing CD8^+^ T cells, which correlates with CD20^+^ B cell clusters. In the presence of CXCL13, CD20^+^ B cells are associated with improved patient prognosis [20]. Additionally, in a study on metastatic melanoma, the co‐localization of tumor‐infiltrating CD8^+^ T cells and CD20^+^ B cells was linked to improved survival, independent of other clinical variables [21]. However, the interaction between CD8^+^ T cells and B cells in NPC remains unclear and warrants further investigation. This study aims to preliminarily explore the distribution and significance of B cells and CD8^+^ T cells in NPC.
Results
2
CD20+FCRL4+ B cells With High Expression of Antigen Presentation and Co‐Stimulatory Molecules are Present in NPC
2.1
We integrated 21 primary NPC (pNPC) samples (Figure S1A, B) and performed quality control and initial clustering (Figure S1C, D). Cluster analysis of CD19 and CD79A‐high‐expressing cells, representing B cells, was conducted (Figure 1A), and the proportion of cell types in each sample was visualized (Figure 1B). Functional genes related to B cells were compared (Figure 1C). CD20^+^BCL6^+^ B cells showed high expression of germinal center‐related genes. CD20^+^FCRL4^+^ B cells exhibited high expression of antigen presentation molecules and co‐stimulatory molecules. CD20^+^NOP53^+^ B cells exhibited a gene expression profile similar to that of CD20^+^FCRL4^+^ B cells but with elevated expression of specific B cell‐related transcription factors. CD20^+^IGHD^+^ B cells showed elevated IGHD expression, suggesting that they are naïve B cells. IGHG^+^ plasma cells exhibited high immunoglobulin heavy chain gene expression and low CD20 expression, indicating their identity as plasma cells.
Single‐cell transcriptomic landscape of CD20+FCRL4+ B cells in NPC. (A) UMAP plot depicting the B cell subpopulations. (B) Proportions of different B cell subpopulations in scRNA‐seq samples from NPC, with the y‐axis indicating the percentage and the x‐axis denoting the data source. Color annotations correspond to panel A. (C) Expression levels of marker genes for distinct B cell subpopulations. The shading of circles represents average expression, and the size of the circles indicates the proportion of positive cells. (D) Differentiation trajectory of CD20+ B cells, with each point representing a cell. Color annotations are consistent with panel A. (E) Pseudotime analysis of CD20+ B cell differentiation, where darker colors indicate early differentiation and lighter colors indicate late differentiation. The trajectory progresses from left to right, branching into two cell fates from the pre‐branch point. (F) Flow cytometry gating strategy for CD20+ B cells in samples derived from surgically resected or biopsied NPC. (G) Histogram showing FCRL4 expression in CD20+ cells from PBMC, primary NPC (pNPC), and recurrent NPC (rNPC) samples, with the proportion of FCRL4− and FCRL4+ cells annotated. (H) Connected line plot comparing FCRL4 expression levels in CD20+ and CD20− cells in each NPC sample. *** p<0.001.(I) Box plot comparing the proportion of FCRL4+ cells within CD20+ cells in NPC and PBMC. * p<0.05.
We conducted pseudotime analysis on CD20^+^ B cells, revealing two differentiation directions (Figure 1D,E). Cell fate 1 transitioned to CD20^+^BCL6^+^ B cells, upregulating IgG heavy chain and proliferation genes, possibly reflecting germinal center B cell maturation toward plasma cells. Cell fate 2 shifted to CD20^+^FCRL4^+^ B cells, showing increased expression of antigen presentation and co‐stimulatory molecules (Figure S2A). Gene set enrichment analysis (GSEA) analysis of differentially expressed genes (DEGs) in CD20^+^FCRL4^+^ B cells versus other CD20^+^ B cells was performed (Figure S2B). The antigen presentation pathway, actin assembly and regulation, and cell adhesion pathways were activated, while the humoral immune and immunoglobulin synthesis pathways were inhibited. We compared gene sets that were differentially expressed between cell fates 1 and 2 (Figure S2C). Results suggested the expression of interferon‐induced genes, implying that the activation of the interferon signaling pathway may play a pivotal role in CD20^+^ B cell differentiation toward cell fate 2.
Subsequently, we extracted the top 10 marker genes with the highest log2FC values from five B cell subgroups (File S1) for single‐sample GSEA (ssGSEA) scoring (Figure S2D). In a survival cohort, using the median as a cutoff, we predicted the relationship between B cell subgroups and progression‐free survival (Figure S2E; File S2). The results suggested that higher levels of B cell infiltration may indicate a better prognosis, with CD20^+^IGHD^+^ B cells, CD20^+^FCRL4^+^ B cells, and CD20^+^NOP53^+^ B cells meeting the significance threshold (P<0.05). Next, we collected NPC samples from 13 patients with NPC (Figure S4; File S3). We compared the proportion of FCRL4^+^ B cells within CD20^+^ B cells in NPC samples and in peripheral blood mononuclear cell (PBMC) samples from four paired patients with NPC (Figure 1F,G; Figure S4A,B). In NPC, the proportion of FCRL4^+^ cells was markedly higher in CD20^+^ cells than in CD20^−^ cells, indicating that FCRL4 is predominantly expressed in B cells (Figure 1H). The results also indicated that FCRL4^+^ cells constituted a notably higher proportion of CD20^+^ cells in NPC compared to PBMCs (Figure 1I), with substantial variation between different patients (Figure 1G; Figure S4A, B). Flow cytometry analysis revealed a slightly higher proportion of FCRL4^+^ cells within CD20^+^ cells (46.2% ± 29.5%) compared to single‐cell sequencing results (36.2% ± 25.6%).
Potential High‐Frequency Interactions Between CD20+FCRL4+ B Cells and CD8+CD69+ T Cells in NPC
2.2
We performed secondary clustering of representative T cell clusters with high CD3E expression, revealing a total of 14 subpopulations. We annotated these subpopulations based on characteristic marker genes (Figure 2A, Figure S5A,B). Pseudotime analysis (Figure 2B–D) showed that CCR7^+^ naïve T cells, characterized by high expression of CCR7, LEF1, SELL, and TCF7, represent the starting point of CD8^+^ T cell differentiation. TNF^+^ T apoptosis (Tap) subpopulations also appeared early in pseudotime. Although TNF^+^ Tap subpopulations exhibited high TNF and mitochondrial gene expression, their proportion was relatively low (Figure 2E), and their significance remains uncertain. Subsequently, CD8^+^ T cells bifurcated into two differentiation trajectories (Figure 2C). One trajectory gave rise to GNLY^+^ effective T (Teff) cells expressing high levels of the cytotoxic molecules GNLY, GZMA, and GZMB, as well as the chemokine receptor CX3CR1. The other differentiation pathway led to tissue‐resident CD69^+^ memory T (Trm) cells with high expression of CD69, ITGAE, ITGA1, and CXCR6 (Figure 2F) and partially specialized into CXCL13^+^ Tem cells with high expression of CXCL13 and the cytotoxic molecule GZMK. We correlated CD69^+^ Trm cell pseudotime differentiation with CD8^+^ T cell immunological memory gene scores from two independent sources (Figure 2G), indicating a significant increase in the immunological memory gene score during CD69^+^ Trm cell differentiation. Moreover, we compared the functional characteristics of CD8^+^ T cell subpopulations (Figure S6). CD69^+^ Trm and CXCL13^+^ Tem cells exhibited higher levels of TCR signaling, proliferation, and migration capacity, as well as immunological memory compared to other subpopulations, indicating their tissue‐resident nature and immune activation status. Furthermore, CD69^+^ Trm cells displayed cytotoxicity levels similar to those of CX3CR1^+^ Teff and GNLY^+^ Teff, suggesting their active killing function. Developmental trajectory analysis also hinted at potential interconversion between CX3CR1^+^ Teff and CD69^+^ Trm cells (Figure 2B). Additionally, CD69^+^ Trm and CXCL13^+^ Tem cells showed higher exhaustion scores compared to other subpopulations.
Single‐cell transcriptomic landscape of CD8+ T cells in NPC. (A) Expression levels of marker genes for different CD8+ T cell subpopulations. The shading of circles represents average expression, and the size of the circles indicates the proportion of positive cells. (B) UMAP plot of CD8+ T cell subpopulations, with each point representing the predicted cell differentiation status by Slingshot, and lines indicating the differentiation trajectory. (C) Pseudotime predicted by Cytotrace annotated on the UMAP plot, with different colors representing various differentiation stages. (D) Boxplots showing pseudotime values for each CD8+ T cell subpopulation, with color annotations consistent with panel B. (E) Proportions of various CD8+ T cell subpopulations in scRNA‐seq samples from NPC, with the y‐axis indicating the percentage and the x‐axis denoting the data source. (F) Feature plot depicting the expression of tissue‐resident marker genes, where darker colors indicate higher expression levels. (G)‐(H) Scatter plots showing the correlation between ssGSEA scores of CD8+ T cell immune memory gene sets (M3049 and M3039 from MsigDB) and pseudotime of CD69+CD8+ T cells. (I) Box plots showing ssGSEA scores of functional gene sets for different CD8+ T cell subtypes.
In terms of the total number of interactions and interaction weights between T cell subpopulations and B cell subpopulations (Figure 3A), CD20^+^FCRL4^+^ B cells exhibited more frequent interactions with CD8^+^CD69^+^ Trm cells. The signals sent from T cells to B cells primarily involve chemotactic and CD40 signals (Figure S7A). CD20^+^FCRL4^+^ B cells exhibit a markedly higher predicted interaction strength with T cells based on CCL and CXCL compared to other B cell subgroups, while CD40 signaling primarily originates from CD20^+^BCL6^+^ B cells. When considering signals sent from B cells to T cells, the main components likely include adhesion molecules, co‐stimulatory signals, and antigen presentation signals. CD20^+^FCRL4^+^ B cells are predicted to be the strongest senders of these signals (Figure 3B–D; Figure S7B). Subsequently, we predicted receptor–ligand pairs that may facilitate interactions between the subpopulations of B and CD8^+^ T cells (Figure 3E,F). CD20^+^FCRL4^+^ B cells were predicted to send extensive adhesion signals (e.g., ITGB1 and ITGA4‐VCAM1), antigen presentation signals (e.g., HLA‐A‐CD8A), and co‐stimulatory signals (e.g., CD86‐CD28) to CD8^+^CD69^+^ Trm and CD8^+^CXCL13^+^ Tem cells. Additionally, CD8^+^CD69^+^ Trm cells were predicted to send unique chemotactic signals (e.g., CCL5‐CCR1 and CCL3‐CCR1) to CD20^+^FCRL4^+^ B cells. Notably, CD8^+^CD69^+^ Trm cells, exhibiting the highest IFNγ expression levels among CD8^+^ T cell subpopulations (Figure 2A), may interact with CD20^+^FCRL4^+^ B cells via the IFNG‐IFNGR1 pathway, potentially playing a pivotal role in B cell differentiation into CD20^+^FCRL4^+^ B cells (Figure S2C).
Communication patterns between B cells and T cells in NPC. (A) Circle plot displaying the total communication counts (left) and communication strength (right) between CD20+FCRL4+ B cells and other B cell and T cell subpopulations. (B)‐(D) Heatmap displaying the communication intensity between B cell subtypes as signal senders (rows) and T cell subtypes as signal receivers (columns) in different signaling pathways. (E)‐(F) Bubble plot illustrating the signal strength from CD20+FCRL4+ B cells to CD8+ T cell subpopulations (E) and from CD8+CD69+ T cells to B cell subpopulations (F). The color intensity of circles represents the probability of communication, with darker colors indicating higher probabilities. (G) Communication pattern map between CD20+FCRL4+ B cells and CD69+CD8+ T cells.
We summarized the potential interaction patterns and significance between CD8^+^CD69^+^ Trm and CD20^+^FCRL4^+^ B cells in a graphical overview (Figure 3G). Initially, chemokines secreted by CD8^+^CD69^+^ Trm cells, including CCL3, CCL5, and CXCL13, are received by chemokine receptors on the surface of CD20^+^FCRL4^+^ B cells, leading to cytoskeletal rearrangement and cellular migration to the same spatial region. Subsequently, adhesion molecules ICAM/VCAM on their surfaces interact with ITGA/TIGB, resulting in cell adhesion between CD8^+^CD69^+^ Trm cells and CD20^+^FCRL4^+^ B cells. CD20^+^FCRL4^+^ B cells present antigens to CD8^+^CD69^+^ Trm cells through MHC‐I molecules and co‐stimulatory molecules CD80/CD86, further activating CD8^+^CD69^+^ Trm cells. CD8^+^CD69^+^ Trm cells, upon receiving antigen stimulation, gain or enhance their cytotoxicity and secrete IFNγ. IFNγ, as a key cytokine for B cell differentiation into CD20^+^FCRL4^+^ B cells, promotes the formation of more CD20^+^FCRL4^+^ B cells. Therefore, it is inferred that CD20^+^FCRL4^+^ B cells may participate in NPC cell immunity or the formation of CD8^+^ T cell immune memory through the MHC‐I presentation of exogenous tumor antigens (also known as cross‐presentation).
NPC‐Derived CD20+FCRL4+ B Cells Co‐localize with CD69+ CD8+ T Cells and Exhibit Potential B‐T Cell Interactions
2.3
Initially, the spatial reconstruction of cell distribution in scRNA‐seq data was performed using a receptor–ligand weighted network (Figure 4A). A co‐enrichment pattern was observed between CD20^+^FCRL4^+^ B cells and CD8^+^ T cells (Figure 4B). Consistent findings were corroborated through multiple immunofluorescence experiments, revealing that CD20^+^FCRL4^+^ B cells and CD8^+^CD69^+^ T cells in both primary and recurrent NPC tissues tended to co‐localize in similar regions, with some cells located in close proximity (Figure 4C,D). The majority of tissue‐infiltrating CD8^+^ T cells expressed CD69 (Figure 4C,D). FCRL4 was predominantly expressed in tissue‐infiltrating CD20^+^ B cells, although a subset did not express FCRL4 (Figure 4C). Comparative spatial analysis of CD20^+^FCRL4^+^ and CD20^+^FCRL4^−^ cells with CD8^+^ T cells revealed that CD20^+^FCRL4^+^ B cells were in closer spatial proximity to CD8^+^ T cells (Figure S6A–D). Regions confirmed as TLS by H&E staining exhibited minimal expression of CD8 and FCRL4, in contrast to areas distant from TLS (Figure 4E). Additionally, FCRL4^+^ cells exhibited markedly higher expression of MHC‐I compared to other cells (Figure S6E–H).
Tissue localization characteristics of CD20+FCRL4+ B cells and CD8+ T cells in NPC. (A) 3D scatter plot showing the spatial distribution predicted by CSOmap for different cell types, with blue outlines representing CD8+ T cells and red outlines representing FCRL4+ memory B cells. (B) Heatmap illustrating the co‐enrichment status of different cell subtypes predicted by CSOmap, where red indicates co‐enrichment, blue indicates mutual exclusion, and green indicates no apparent spatial relationship. (C)‐(D) Immunofluorescence images illustrating the spatial localization of CD20+FCRL4+ B cells and CD69+CD8+ T cells in pNPC (C) and rNPC (D). Yellow arrows indicate CD20+FCRL4− cells, and white arrows point to clearly co‐localized FCRL4+ and CD8+ cells. Different panels are selected from the same field of view. (E) On the left, an H&E‐stained image of an area containing TLS, and on the right, an immunofluorescence image showing the expression of FCRL4 and CD8 in the same region. The dashed circle outlines the TLS region.
In addition, FCRL4^+^ B cells have been reported in the tonsils, adenoids, and various chronic inflammatory tissues. To gain preliminary insight into whether FCRL4^+^ B cells in NPC may differ functionally from those in non‐malignant inflammatory contexts, we obtained resected samples from one patient with pNPC and one patient with chronic tonsillitis, isolated FCRL4^+^ B cells, and performed T–B cell co‐culture and cytotoxicity assays (Figure S7A). This analysis represents a limited, small‐sample comparison and should be interpreted with caution. In this preliminary assessment, FCRL4^+^ B cells derived from NPC appeared to promote CD8^+^ T‐cell expansion and tended to increase the proportion of CD45RO^+^ CD8^+^ T‐cell subsets, whereas FCRL4^+^ B cells from tonsillar tissue did not show a comparable effect (Figure S7B–E). Consistently, CD8^+^ T cells co‐cultured with NPC‐derived FCRL4^+^ B cells demonstrated higher cytotoxic activity, while those co‐cultured with tonsil‐derived FCRL4^+^ B cells did not exhibit a similar enhancement (Figure S7F). Together, these observations—while based on very limited sampling—raise the possibility that FCRL4^+^ B cells in NPC may possess distinct functional properties compared with their counterparts in chronic inflammatory tissues.
IFNγ and Antigenic Substances Promote the Formation of the CD20+FCRL4+ B Cell Phenotype
2.4
In in vitro experiments, we utilized the ARH‐77 cell line to simulate B cell responses to various stimuli (Figure 5A). ARH‐77 cells express CD19 and CD20 but do not express CD38. Their cellular background falls between naïve B cells and B lymphoblasts, with their B cell receptor (BCR) still responsive to antigens, possibly resembling a precursor state of CD20^+^FCRL4^+^ B cells. Single‐cell sequencing predicted the role of IFNγ signaling in the differentiation pathway of B cells into FCRL4^+^ B cells (Figure S2C). Therefore, we first stimulated ARH‐77 cells with 5 ng/mL of IFNγ, which selectively upregulated the expression of MHC‐I molecules (Figure 5B,E) without affecting the levels of MHC‐II molecules (Figure 5B,F) or altering FCRL4 expression levels (Figure S8A).
*Stimulation of B lymphoblastoid cell lines to differentiate into CD20+FCRL4+ B cells in vitro. (A) Flow cytometry gating strategy for in vitro cultured cells. (B) Histograms showing the expression of functional molecules in B cell lines under different treatments (IFNγ 5 ng/mL; CD40L 100 ng/mL). (C)‐(D) Histograms showing FCRL4 expression in B cell lines under different treatments (LPS 5 µg/mL; SPA 5 µg/mL) with the proportion of FCRL4− and FCRL4+ cells annotated. (E)‐(F) Bar chart comparing the expression levels of MHC‐I (E) and MHC‐II (F) molecules between the control group and the IFNγ‐treated group. nsP≥0.05; **p<0.01. (G) Bar chart comparing the expression of FCRL4 molecules between the control group and the CD40L intervention group. p<0.05. (H) Connected line plot comparing the expression levels of MHC‐I molecules between FCRL4+ and FCRL4− cells in the same sample after co‐treatment with IFNγ and CD40L. ** p<0.01. (I) Bar chart showing the proportion of FCRL4+ cells in B cells in the control group, LPS‐treated group, and SPA‐treated group. * p<0.05; *** p<0.001; **** p<0.0001. (J) The bar chart compares the signal intensity of FCRL4 expression in ARH‐77 cells in the blank control group, HK‐1 co‐culture group, and HK‐1 lysate stimulated group. nsP≥0.05; ** p<0.01; *** p<0.001.
Subsequently, stimulation with 100 ng/mL of CD40L induced the upregulation of both MHC‐I and MHC‐II molecule expression (Figure S8B,C) and increased FCRL4 expression (Figure 5B,G) in ARH‐77 cells. Co‐stimulation with IFNγ and CD40L also resulted in the upregulation of MHC‐I and MHC‐II molecules, along with FCRL4 expression (Figure S8D–F). Comparison of MHC‐I molecule expression between FCRL4 low and FCRL4 high cells showed significantly higher MHC‐I expression in FCRL4 high cells (Figure 5B,H).
We also stimulated ARH‐77 cells with Staphylococcus aureus protein A (SPA) and lipopolysaccharide (LPS). SPA and LPS activate B cells through different mechanisms, with the former inducing B cell activation through cross‐linking of the BCR signaling and the latter directly binding to Toll‐like receptor 4 to activate B cells. Our results demonstrated that both SPA and LPS partially induced the differentiation of B cells into FCRL4^+^ cells, with SPA exhibiting a stronger effect (Figure 5C,I). This suggests that FCRL4 expression may be a response to antigenic stimulation. Thus, we co‐cultured ARH‐77 with HK‐1 or treated it with lysates from HK‐1 to observe FCRL4 expression. The results showed a significant upregulation of FCRL4 expression when ARH‐77 was cultured with HK‐1 lysates (Figure 5D,J).
We also subjected B cells sorted from PBMCs of normal individuals to the aforementioned stimulation with IFNγ and HK‐1 lysates (Figure S9A). The results showed that HK‐1 lysates significantly upregulated the expression of FCRL4, while IFNγ downregulated FCRL4 expression (Figure S9B), contrasting with the observations in ARH‐77. Both IFNγ and HK‐1 lysates upregulated the expression of MHC‐I but did not reach statistical significance in this assay (Figure S9C). However, upon combined treatment with HK‐1 lysates and IFNγ, MHC‐I was significantly upregulated (Figure S9C).
IFNγ can Induce B Cells to Present Exogenous NPC‐Related Antigens
2.5
We used a reported method to synthesize epitope peptide probes that can bind to MHC‐I [22]. The 426‐434 segment CLGGLLTMV (Figure S10A) encoded by Epstein–Barr virus (EBV) latent membrane protein 2 is a linear peptide epitope (epitope ID: 6568). Predictions suggest its binding affinity to MHC‐I HLA‐A*02 allele products, but with low affinity for MHC‐II molecules (Figure S10B). The predicted reliability of CLGGLLTMV exceeds 70%, and its binding affinity with HLA‐A is significantly lower than the threshold for spontaneous binding, which is ‐5 kcol/mol (Figure S10C, D).
We investigated the uptake and subcellular localization of the probe (CLGGLLTMV attached to fluorescein isothiocyanate [FITC]) under different stimulation conditions in ARH‐77 and HMy2.CIR cells. HMy2.CIR cells are a rapidly growing mutant of the ARH‐77 cell line, generated through gamma irradiation, and they do not express products from the HLA‐A and B loci. Initially, we assessed the expression and localization of MHC‐I in B cells through immunofluorescence staining (Figure 6A). Both cell types demonstrated the ability to take up the probe regardless of stimulation with IFNγ, LPS, or the absence of stimulation (Figure 6B). In the presence of IFNγ, with or without LPS, the probe consistently aggregated at the cell membrane of ARH‐77 cells, mirroring the cellular location of MHC‐I. In contrast, unstimulated ARH‐77 cells primarily retained the probe intracellularly (Figure 6B). HMy2.CIR cells, which have defective MHC‐I expression, did not exhibit translocation of the probe to the cell membrane, even when stimulated with IFNγ (Figure 6B).
CD20+FCRL4+ B cells activate CD8+ T cells by presenting exogenous MHC‐I‐restricted antigen both in vitro and in vivo. (A) Immunofluorescence staining shows the localization of MHC‐I in ARH‐77 cells, with MHC‐I stained red and nuclei stained blue using DAPI. (B) Demonstration of B cells from different treatment (IFNγ 5 ng/m:; LPS 5 µg/mL) groups capturing and presenting the CLGGLLTMV epitope. The green fluorescence indicates the cellular location of the fluorescein isothiocyanate (FITC)‐labeled CLGGLLTMV epitope, and the blue fluorescence marks the cell nucleus. (C) Flow cytometry gating strategy for CD20+ B cells and CD8+ T cells in the ARH‐77 or HMy2.CIR and CD8+ T cell co‐culture system. The stimulation concentration of IFNγ was 5 ng/mL, consistent with previous descriptions. (D) Scatter plots demonstrate the expression of CD20 and CD8 in co‐culture systems across different treatment groups, with the proportions of negative and positive cells labeled. (E) Contour plots illustrate the expression of CD69 and CD45RO in CD20−CD8+ cells within co‐culture systems across different treatment groups, with the proportions of negative and positive cells labeled. (F) Bar graphs compare the proportion of CD8+ cells among CD20− cells in co‐culture systems across different treatment groups. nsP≥0.05; * p<0.05; **** p<0.0001. (G) Bar graphs compare the proportion of CD45RO+ cells among CD8+ cells in co‐culture systems across different treatment groups. nsP≥0.05; **** p<0.0001. (H) A bar plot compares the percentage of surviving HK‐1 cells after tumor killing experiments in different treatments within the co‐culture system. nsP≥0.05; ** p<0.01; *** p<0.001; **** p<0.0001. (I) A bar plot compares the cytotoxicity rate of HK‐1 cells after tumor killing experiments in different treatments within the co‐culture system. nsP≥0.05; * p<0.05; *** p<0.001; **** p<0.0001; **** p<0.0001. (J) Schematic of the animal experiment timeline, with day 0 as the day of tumor implantation, and tumor measurements taken on days 7, 14, and 21. (K) Tumors excised on day 21 post‐sacrifice, with the scale shown in centimeters (cm). (L) Tumor growth curve showing tumor volumes recorded on days 7, 14, and 21, with statistical analysis performed on day 21 tumor volumes. nsP≥0.05; ** p<0.01; **** p<0.0001.
IFNγcan Induce B Cells to Activate CD8+ T Cells in an MHC‐I‐Dependent Manner, Leading to Cytotoxicity Against NPC Cells
2.6
After co‐culturing ARH‐77 or HMy2.CIR cells, which had been treated with various combinations of stimuli, with CD8^+^ T cells for 24 h, we analyzed the changes in the CD8^+^ T cell population (Figure 6C). During the co‐culture of B cells and CD8^+^ T cells, B cells were pre‐treated with IFNγ or HK‐1 lysate for 24 h, then washed before being added to the CD8^+^ T cell culture system for another 24 h. This procedure ensured that the stimuli did not directly affect the CD8^+^ T cells. To rule out potential residual interference, two control groups were set up: CD8^+^ T cells cultured alone for 24 h without any treatment, and CD8^+^ T cells treated with a combination of IFNγ and HK‐1 lysate. Our observations indicated that pre‐stimulation of ARH‐77 or HMy2.CIR cells with HK‐1 lysate (with or without IFNγ) resulted in the expansion of CD8^+^ T cells (Figure 6D,F). This effect was most pronounced in the co‐culture group where ARH‐77 cells were pre‐stimulated with both IFNγ and HK‐1 lysate. In contrast, the co‐culture group with MHC‐I‐deficient HMy2.CIR cells pre‐stimulated under the same conditions showed only a mild expansion of CD8^+^ cells (Figure 6D,F). Additionally, we found that in the co‐culture groups of ARH‐77 or HMy2.CIR cells pre‐stimulated with HK‐1 lysate (with or without IFNγ), varying proportions of CD45RO^+^ cells within the CD8^+^ cell population were observed (Figure 6E,G). The proportion of CD45RO^+^ cells among the expanded CD8^+^ cells in the co‐culture group of ARH‐77 cells pre‐stimulated with both IFNγ and HK‐1 lysate was significantly higher than that in the co‐culture group of HMy2.CIR cells pre‐stimulated under the same conditions and the co‐culture group of ARH‐77 cells pre‐stimulated with HK‐1 lysate alone (Figure 6E,G). Furthermore, we generated an ARH‐77 cell line with β2‐microglobulin (B2M), a key component of MHC class I molecules, knocked out (Figure S11A). Following stimulation as described above, the cells were co‐cultured with CD8^+^ T cells to assess tumor‐killing activity (Figure S11B–F). The results demonstrated that B2M deletion markedly impaired the ability of ARH‐77 cells to promote CD8^+^ T‐cell expansion (Figure S11D). In addition, the capacity of B2M‐deficient ARH‐77 cells to induce the expansion of the CD45RO^+^ CD8^+^ T‐cell subset was significantly reduced (Figure S11C,E). Moreover, their ability to enhance CD8^+^ T‐cell‐mediated tumor cell killing was also substantially diminished (Figure S11F).
We co‐cultured CD20^+^ B cells derived from PBMCs and CD8^+^ T cells using the same method (Figure S12A). The results indicated that only the co‐culture of CD20^+^ B cells pre‐stimulated with both IFNγ and HK‐1 lysate led to an expansion of CD8^+^ T cells (Figure S12B, D). In contrast to the results from the cell line experiments, there were no significant changes in the proportion of CD45RO^+^ cells in any of the groups (Figure S12C,E). Based on this observation, we speculated that PBMC‐derived CD20^+^ B cells may require additional CD40L signaling to achieve full functional competence. Therefore, we incorporated CD40L stimulation along with IFNγ and HK‐1 lysate treatment (Figure S12G). The results showed that, although CD40L stimulation did not markedly alter the expansion of CD8^+^ T cells (Figure S12H,I), the proportion of CD45RO^+^ subsets within CD8^+^ T cells was significantly increased after CD40L treatment compared with the IFNγ and HK‐1 lysate group (Figure S12H, J). Moreover, CD40L‐stimulated CD20^+^ B cells also significantly enhanced the cytotoxic activity of CD8^+^ T cells (Figure S12K).
We performed ELISA to measure the levels of IFNγ released in the culture supernatants of the PBMC‐derived CD20^+^ B and CD8^+^ T cell co‐culture system. We observed that the co‐culture groups pre‐stimulated with both HK‐1 lysate and IFNγ exhibited the highest levels of IFNγ release (Figure S13A), whereas no significant release of TNFα was observed between groups (Figure S13B). Significant releases of Gzms‐A and Gzms‐B were observed in groups co‐stimulated with HK‐1 lysate and IFNγ (Figure S13C,D).
After pre‐stimulation with different combinations, we conducted cytotoxicity assays using HK‐1 cells to simulate potential anti‐tumor immune responses in vivo. The results showed that co‐culture of ARH‐77 cells with CD8^+^ T cells, pre‐stimulated with HK‐1 lysate (with or without IFNγ), led to an increase in cytotoxic activity against HK‐1 cells (Figure 6H). The co‐culture group pre‐stimulated with both HK‐1 lysate and IFNγ exhibited the highest level of cytotoxicity (Figure 6H). The co‐culture systems of PBMC‐derived CD20^+^ B cells and CD8^+^ T cells showed similar results (Figure S12F). In agreement with these observations, the LDH release assay yielded comparable results (Figure 6I). CD8^+^ T cells co‐cultured with fully stimulated ARH‐77 cells exhibited maximal cytotoxicity, while MHC‐I–deficient B cells significantly attenuated this response. Finally, we injected either unstimulated CD8^+^ T cells or CD8^+^ T cells co‐cultured with B cells into the peritumoral and intratumoral regions of tumor‐bearing BALB/c‐nu mice (Figure 6J; Figure S14A). Human CD45^+^ cells in the spleen were evaluated on days 0, 8, 15, and 21 throughout the experimental course. The results showed that the proportion of human CD45^+^ cells peaked at over 30%, indicating successful immune reconstitution (Figure S14B). The results showed that tumor growth was significantly restricted in the mice treated with CD8^+^ T cells co‐cultured with ARH‐77 cells compared to those receiving CD8^+^ T cells alone (Figure 6K–L). Mice treated with CD8^+^ T cells co‐cultured with HMy2.CIR cells also exhibited mild tumor growth restriction, although the difference was not statistically significant (Figure 6K–L). In addition, we observed that in mice treated with CD8^+^ T cells co‐cultured with ARH‐77 cells, only small, scattered areas of residual tumor tissue were present, surrounded by abundant fibrotic and necrotic regions (Figure S15). Moreover, a large number of immune cells were found infiltrating both the tumor parenchyma and the surrounding areas (Figure S15).
IFNγ Enhances MHC‐I Restricted Cross‐Presentation of Exogenous Antigens by B Cells Through the Upregulation of the Beige and Chediak‐Higashi (BEACH) Protein and WD Repeat and FYVE Domain‐Containing Protein 4 (WDFY4) Expression
2.7
In single‐cell sequencing analysis, we found that CD20^+^FCRL4^+^ B cells significantly overexpress WDFY4 (Figure 7A). After stimulating ARH‐77 cells with 5 ng/mL IFNγ for 24 h, we observed a significant upregulation of WDFY4 protein and mRNA levels, an effect that could be inhibited by WDFY4 siRNA (Figure 7B–D). Next, antigen presentation experiments showed that knocking down WDFY4 could inhibit the interferon‐induced presentation of antigen peptide probes by ARH‐77 cells but did not affect uptake (Figure 7E). We also conducted co‐culture experiments of WDFY4‐knockdown ARH‐77 cells with CD8^+^ T cells (Figure 7F). The results showed that, compared to the fully stimulated group (ARH‐77 cells pre‐stimulated with IFNγ and HK‐1 lysate), WDFY4‐knockdown ARH‐77 cells that were fully stimulated exhibited a significantly reduced ability to promote the expansion of CD8^+^ T cells (Figure 7G,I).
IFNγ promotes B‐cell cross‐presentation of antigen by upregulating WDFY4 expression. (A) WDFY4 expression in B cell subpopulations in scRNA‐seq, with color intensity indicating expression levels and bubble size representing the percentage of positive cells. (B) WDFY4 protein expression in ARH‐77 cells before and after IFNγ stimulation, as well as following WDFY4 knockdown. The stimulation concentration of IFNγ was 5 ng/mL, consistent with the description above. (C) Quantification of WDFY4 protein expression grayscale values. nsP≥0.05; * p<0.05; ** p<0.01. (D) WDFY4 mRNA expression in ARH‐77 cells before and after IFNγ stimulation and following WDFY4 knockdown. nsP≥0.05; *** p<0.001. (E) Localization of MHC‐I expression in ARH‐77 cells and probe uptake after WDFY4 knockdown, with MHC‐I shown in red fluorescence and the probe in green fluorescence. (F) Flow cytometry analysis strategy for the co‐culture system of ARH‐77 cells after WDFY4 knockdown with CD8+ T cells. (G) Scatter plots demonstrate the expression of CD20 and CD8 in co‐culture systems across different treatment groups, with the proportions of negative and positive cells labeled. (H) Contour plots illustrate the expression of CD69 and CD45RO in CD20−CD8+ cells within co‐culture systems across different treatment groups, with the proportions of negative and positive cells labeled. (I) Bar graphs compare the proportion of CD8+ cells among CD20− cells in co‐culture systems across different treatment groups. nsP≥0.05; ** p<0.01. (J) Bar graphs compare the proportion of CD45RO+ cells among CD8+ cells in co‐culture systems across different treatment groups. nsP≥0.05; * p<0.05; ** p<0.01. (K) Bar plot compares the percentage of surviving HK‐1 cells after tumor killing experiments across different treatments within the co‐culture system. nsP≥0.05; *** p<0.001.
Additionally, the proportion of CD45RO^+^ cells among CD8^+^ T cells decreased (Figure 7H,J). In cytotoxicity assays, the co‐culture system of fully stimulated WDFY4‐knockdown ARH‐77 cells with CD8^+^ T cells showed a significantly reduced ability to kill HK‐1 cells (Figure 7K). In the in vivo experiments, we reconstructed HK‐1 tumor–bearing mice with CD8^+^ T cells together with ARH‐77 cells in which WDFY4 was knocked down (Figure 7B), FCRL4 was knocked out (Figure S16A), or an empty vector was introduced after applying the same full stimulation described above (Figure S16B). The results showed that, compared with the control reconstitution group, tumor growth was not suppressed in the WDFY4‐knockdown reconstitution group (Figure S17C–E), whereas tumor growth was markedly suppressed, to a degree similar to the control reconstitution group, in the FCRL4‐knockout reconstitution group (Figure S17C–E).
Moreover, we investigated the role of canonical IFNγ downstream transcription factors in regulating WDFY4. Based on the DNA‐binding motifs of STAT1 and STAT2, we initially screened three putative binding sites within the upstream promoter region of WDFY4 (Figure S17A‐B). ChIP assays showed that STAT1 bound to two of these sites, whereas STAT2 did not exhibit any positive enrichment, suggesting that STAT1 is likely the transcription factor mediating IFNγ‐dependent regulation of WDFY4 (Figure S17C). Next, we examined the signaling cascade from IFNγ to STAT1. Following stimulation of ARH‐77 cells with IFN γ, we observed increased phosphorylation levels of JAK1 and STAT1, as well as elevated expression of WDFY4, while total protein levels showed no notable change (Figure S17D–E). Additionally, treatment with a JAK1 inhibitor (JAKi) reduced JAK1 and STAT1 phosphorylation and concomitantly decreased WDFY4 expression (Figure S17D‐E). Together, these results indicate that IFNγ may regulate WDFY4 expression through the JAK1–STAT1 pathway.
The Differentiation of CD20+FCRL4+ B Cells in NPC May Indicate Improved Immunotherapy Outcomes
2.8
We performed gene set scoring based on marker genes for B cell subpopulations (File S1) in GSE102349 to reflect the abundance of each subpopulation and found a strong positive correlation between CD20^+^FCRL4^+^ B cell scores and PD1 expression (Figure 8A). Next, we constructed a correlation matrix for the marker genes of CD20^+^FCRL4^+^ B cells and IGHG^+^ plasma cells, which represent terminally differentiated cell types (Figure 1D), in GSE102349 (Figure 8B). The results showed that the marker genes for each terminally differentiated cell type clustered independently. We extracted the expression matrix of the most closely clustered genes (marked in red and green in Figure 8B) and performed unsupervised clustering, identifying four patient types (Figure 8C): high abundance of CD20^+^FCRL4^+^ B cells (Polarization toward FCRL4^+^ cells), high abundance of IGHG^+^ plasma cells (Polarization toward IGHG^+^ cells), high abundance of both (equilibrium polarization), and low abundance of both (inconspicuous polarization). Additionally, we defined samples with a high abundance of either cell type as exhibiting terminal differentiation, while those with low abundance of both were classified as lacking terminal differentiation (Figure 8C). Survival analysis revealed no significant association between specific differentiation into CD20^+^FCRL4^+^ B cells or IGHG^+^ plasma cells and patient survival, but patients with a lack of terminal differentiation had significantly worse prognosis (Figure 8D). This suggests that both CD20^+^FCRL4^+^ B cells and IGHG^+^ plasma cells contribute positively to antitumor immunity in NPC, and polarization toward either type is beneficial for patient prognosis. We then compared the immune status of the four patient types and found that the immunotherapy score (easier score), PD1 expression, interferon score, and TLS were all elevated in patients with polarization toward FCRL4^+^ cells, while the cytotoxicity score was highest in patients with polarization toward IGHG^+^ cells (Figure 8E). In immunohistochemical staining of NPC samples, we also observed a significant positive correlation between the abundance of FCRL4^+^ cells and PD1^+^ cells in the tissue (Figure 8F–H).
Differentiation States of B Cells in NPC and Immune Characteristics in Differentiation States. (A) Correlation Between Gene Signatures of B Cell Subsets and PD1 Molecular Expression. Scatter plots showing the correlation between ssGSEA scores of five B cell subsets in GSE102349 and PD1 expression. Pearson's R values and P‐values are calculated. (B) Correlation heatmap showing the co‐expression relationships of marker genes for the two terminal differentiation states of B cells in GSE102349. The hierarchical clustering method is used to cluster genes based on the correlation between each gene, where blue indicates a correlation close to 1, and red indicates a correlation close to ‐1. (C) Heatmap displaying ssGSEA scores for the two terminal differentiation states in GSE102349. Samples are clustered using hierarchical clustering. (D) K‐M curves for samples from GSE102349 grouped according to different B cell differentiation states, with P‐values from log‐rank tests. (E) Box plot comparing immune feature scores among different B cell differentiation states. (F) IHC images illustrating the expression of FCRL4 or PD1 in pNPC and rNPC. (G)‐(H) The R×C table statistically analyzed the relationship between FCRL4 expression levels and PD1 expression levels in pNPC and rNPC, along with the chi‐square test P‐values.
Next, we examined the potential link between CD20^+^FCRL4^+^ B cell abundance and combined chemoimmunotherapy. We collected MRI results and tissue samples from three patients with recurrent NPC before and after receiving combined chemoimmunotherapy (Figure 9A; File S4). In the Neo1 patient, the tumor enhancement area decreased by more than 50% after treatment, with postoperative pathology showing large areas of necrosis and scattered atypical cells (Figure 9B). This patient exhibited extensive infiltration of CD20^+^FCRL4^+^ B cells and CD8^+^ T cells both before and after treatment. In the Neo2 patient, the tumor enhancement area decreased by approximately 50% after treatment, with postoperative pathology revealing large areas of necrosis and small to moderate clusters of atypical cells (Figure 9C). This patient had minimal immune cell infiltration before treatment, but large numbers of CD20^+^FCRL4^+^ B cells and CD8^+^ T cells appeared after treatment. In contrast, the Neo3 patient showed almost no change in tumor size before and after treatment, and postoperative pathology revealed sheets of atypical cells (Figure 9D). This patient had abundant CD8^+^ T cell infiltration both before and after treatment, but no CD20^+^FCRL4^+^ B cells were present. We subsequently analyzed the changes in the intratumoral density of CD20^+^FCRL4^+^ B cells before and after neoadjuvant therapy in 13 patients, comprising 6 cases of stable disease (SD) and 7 cases of partial response (PR). The pre‐treatment density of CD20^+^FCRL4^+^ B cells appeared comparable between SD and PR patients (Figure 9E). In contrast, post‐treatment CD20^+^FCRL4^+^ B cell density tended to be higher in PR patients relative to those with SD (Figure 9F). Furthermore, while CD20^+^FCRL4^+^ B cell density in SD patients showed no clear change before and after treatment (Figure 9G), PR patients displayed a notable increase in this subset following therapy (Figure 9H). These observations may suggest a potential association between elevated post‐treatment CD20^+^FCRL4^+^ B cell density and better therapeutic response.
Imaging, pathology, and tissue immunofluorescence staining of patients with recurrent NPC receiving neoadjuvant chemoimmunotherapy. (A) Specimen collection process: For patients suspected of NPC relapse, imaging studies and biopsy using forceps are first performed after hospital admission to confirm the diagnosis, collecting the initial imaging data and tissue samples (pretherapy). After completing one cycle of treatment with gemcitabine, cisplatin, and toripalimab, imaging evaluation is conducted, followed by surgical resection to collect the second set of imaging data and tissue samples (post‐treatment). (B)‐(D) MRI T1‐weighted contrast‐enhanced imaging, H&E staining of tissue pathology, and immunofluorescence staining for FCRL4 and CD8 in Neo1‐3 patients, respectively. (E)‐(F) Changes in CD20+FCRL4+ B‐cell density before (E) and after (F) treatment. nsP≥0.05; * p<0.05. (G)‐(H) Changes in CD20+FCRL4+ B‐cell density before and after treatment in SD (G) and PR (H) patients. nsP≥0.05; * p<0.05.
Discussion
3
In this study, a subset of B cells identified as CD20^+^FCRL4^+^ B cells was discovered in NPC, suggesting their potential role as specialized antigen‐presenting cells (APCs) that interact with CD8^+^ T cells. Classical studies suggest that B cells, through MHC‐II molecules, can present antigens to CD4^+^ T cells, thereby inducing T cell responses [23]. It has been confirmed that B cells in various human tumors often upregulate the functional molecules of APCs, including MHC‐II, and co‐stimulatory molecules such as CD80, CD86, and ICOS ligand (ICOSL) [24, 25, 26]. A recent in vitro study found evidence of interaction between CD4^+^ T cells and B cells in NPC [27]. Specifically, a subset of PD‐1^+^CXCR5‐CD4^+^ Th‐CXCL13 cells was identified that could drive B cells into the TLS of NPC, providing in vitro evidence for the interaction between CD4^+^ T cells and B cells in NPC. While we were conducting this study, Zhang et al. also revealed the critical role of a class of FCRL4‐expressing B cells across various cancers, as well as their spatial proximity to CD4^+^ T cells [28]. Although direct evidence of an interaction between B cells and CD8^+^ T cells is lacking, some studies suggest potential connections. For instance, in ovarian cancer, the presence of CD20^+^ B cells is associated with significantly improved prognostic outcomes for CD8^+^ T cells, indicating that interactions between these lymphocyte subtypes lead to more effective anti‐tumor immunity [29]. A study based on a mouse model revealed that tumor‐specific B cells presenting antigens could enhance the responses of both CD4^+^ Tfh and CD8^+^ T cells to tumors [30]. Another study demonstrated that B cells transfected with recombinant vaccinia virus encoding the tumor‐associated antigen NY‐ESO‐1 could present antigen peptides to CD8^+^ T cells in an MHC‐I‐dependent manner [31]. Additionally, direct co‐localization of CD20^+^ B cells and CD8^+^ T cells has been observed in various solid tumors, including breast cancer and melanoma [29, 32, 33]. These studies collectively indicate a close relationship between TME B cells and CD8^+^ T cells, potentially established through an antigen cross‐presentation pathway involving MHC‐I molecules. This process involves a series of steps, including B cell uptake of exogenous antigens, cross‐presentation, activation of CD8^+^ T cells, and tumor killing. However, no study has been able to fully simulate this process in vivo or in vitro. Our research successfully simulated and confirmed this process through in vitro experiments, demonstrating that B cells can take up exogenous NPC antigens and present them to CD8^+^ T cells, ultimately mediating the killing of tumors by CD8^+^ T cells. By comparing ARH‐77 and HMy2.CIR (ARH‐77 lacking expression of HLA‐A and HLA‐B), we preliminarily confirmed that B cell activation of CD8^+^ T cells is MHC‐I restricted and dependent on the presence of IFNγ. However, the transformation of naïve B cells from peripheral blood into CD20^+^FCRL4^+^ B cells with an APC phenotype is more complex than previously thought. While IFNγ and exogenous antigen stimulation may be important initial steps, co‐culture experiments with CD20^+^ B cells and CD8^+^ T cells derived from PBMCs showed an expansion of CD8^+^ T cells without any change in the expression of molecules such as CD45RO. This suggests that naïve B cells entering the TME may require more complex induction and selection in TLS to obtain a complete APC phenotype.
FCRL4 is an immune regulatory receptor generally believed to be expressed on the human memory B cell subset in mucosa‐associated lymphoid tissues, and it has been confirmed to bind with IgA [34]. Research has employed soluble CD40 ligand stimulation to simulate T cell‐dependent activation, successfully inducing the generation of FCRL4^+^ B cells in vitro from purified memory B cells [35], consistent with the observations made in our study. However, in this study, we observed that antigenic substances possess a stronger capability to induce FCRL4 expression. Stimulation of ARH‐77 cells with LPS or SPA resulted in the expansion of a subset of FCRL4^+^ cells. Conversely, when using the more complex antigenic component, HK‐1 lysate, most ARH‐77 cells exhibited FCRL4 expression. Existing literature indicates that the activation of FCRL4 and its downstream signaling pathway plays a role in inhibiting BCR signal transduction [36]. Therefore, we hypothesize that the high expression of FCRL4 on B cells in NPC may serve as a form of negative feedback regulation in response to BCR activation by external antigenic substances. The expression of FCRL4 on B cells in the NPC microenvironment may serve as an indicator of the strength of NPC tumor antigenicity. Furthermore, FCRL4^+^ B cells have also been identified in various non‐neoplastic infectious and inflammatory diseases, and their significance in the context of malignancy warrants deeper consideration. First, there is a notable similarity in spatial distribution between these contexts. In inflammatory conditions such as tonsillitis and autoimmune diseases, prominent lymphoid follicular hyperplasia is frequently observed. These follicles are composed of secondary lymphoid structures, and FCRL4^+^ B cells are typically located in the perifollicular regions surrounding them. In our study, we found that NPC tissues frequently develop TLSs that closely resemble secondary lymphoid tissue. CD20^+^FCRL4^+^ B cells were predominantly distributed around these TLSs. In contrast, NPC samples lacking TLS formation rarely contain CD20^+^FCRL4^+^ B cells. These observations suggest that the presence of FCRL4^+^ B cells in inflammatory lesions and CD20^+^FCRL4^+^ B cells in NPC is dependent on the existence of organized lymphoid architecture, such as secondary lymphoid organs or TLS. However, these two cell populations differ functionally. In contexts like tonsils, HIV infection, and systemic lupus erythematosus, FCRL4^+^ B cells are characterized by the expression of inhibitory receptors (including FCRL4 itself), attenuated BCR signaling, and an association with chronic antigenic stimulation and inflammatory milieus. The presence of FCRL4^+^ B cells in these settings is thought to limit excessive humoral immune activation. In contrast, CD20^+^FCRL4^+^ B cells in NPC exhibit high expression of MHC class I and costimulatory molecules (CD80/CD86), indicating an immunocompetent rather than functionally exhausted phenotype. These cells possess the capacity for cross‐presentation of exogenous NPC antigens. This feature likely represents the fundamental distinction between FCRL4^+^ B cells found in chronic inflammation and those in NPC.
The Role of IFNγ and WDFY4 in Antigen Cross‐Presentation by B Cells: It is well established that IFNγ can upregulate the expression of both MHC‐I and MHC‐II molecules [37, 38]. However, in our study, IFNγ primarily upregulated MHC‐I expression in B cells. There have been no reports on whether IFNγ can initiate cross‐presentation by APCs. In this study, IFNγ was found to significantly upregulate the expression of WDFY4. WDFY4 is predicted to act upstream of or within antigen processing and presentation, as well as the cellular response to viruses. It has been shown to influence the pathogenesis and progression of various autoimmune diseases [39, 40]. This role is supported by findings that WDFY4 is essential for the cross‐presentation function of classical dendritic cells and may regulate vesicle transport pathways to allow exogenous antigens to enter the MHC‐I restricted antigen presentation pathway [41]. While B cells have been demonstrated to perform antigen cross‐presentation [42], there is no report on whether WDFY4 plays a similarly crucial role in B cells as it does in dendritic cells (DCs). Additionally, the role of WDFY4 in tumorigenic diseases has not been reported. In summary, this study reveals that IFNγ can initiate cross‐presentation by B cells, and this process is dependent on WDFY4.
Before the advent of single‐cell sequencing technology, the role of B cells in anti‐tumor immunity in NPC received minimal attention. Some studies even suggested that infiltrating B cells in NPC might serve as host cells for EBV, leading to persistent infection and subsequent carcinogenesis [43]. In recent years, researchers have used single‐cell sequencing technology to dissect the complex composition of functionally heterogeneous B cell subpopulations in NPC. These B cells have been found to engage in extensive interactions with various immune cells, including T cells, DCs, and others [44, 45, 46, 47]. A recent important study has revealed that the combination chemotherapy of GP activates innate‐like B cells (ILBs) to lead to an anti‐tumor immune response [24]. Chemotherapy‐induced DNA fragments activate the STING type I interferon‐dependent pathway, leading to the induction of ILBs. Subsequently, ILBs further amplify follicular helper cells and type 1 helper T cells through the ICOSL‐ICOS axis, thereby enhancing cytotoxic T cells in the TLS post‐chemotherapy. This study unveils the central role of B cells in the anti‐tumor immune response. Our study complements this B cell‐centric anti‐tumor immune response, suggesting that B cells, through MHC‐I‐restricted cross‐presentation of exogenous NPC antigens, may serve as an essential mechanism for activating cytotoxic T cells in NPC.
Moreover, the significance of B cells in NPC may extend beyond conventional radiotherapy and chemotherapy. While we were conducting this study, a research was published in which a single‐cell sequencing analysis of patients with NSCLC undergoing anti‐PD‐1 therapy combined with chemotherapy identified a class of memory B cells similar to the CD20^+^FCRL4^+^ B cells described in our study, namely FCRL4^+^ memory B cells [48]. These cells exhibited high expression of FCRL4 and co‐stimulatory molecules, and the increase in these cells suggested a favorable response to PD‐1 blockade combined with chemotherapy. In a recent pan‐cancer study conducted concurrently with ours, tumor‐associated atypical B cells characterized by FCRL4 were predicted to be potentially linked to immunotherapy response in various cancers, including NPC [28]. In our study, we proposed two potential differentiation pathways for B cells in NPC based on inferred single‐cell developmental trajectories and the classification of patients with NPC using public data. We found that patients whose B cells differentiate toward CD20^+^FCRL4^+^ B cells may benefit more from anti‐PD1 therapy. At our Nasal Skull Base Surgery Center, we observed a possible direct link between CD20^+^FCRL4^+^ B cells and treatment response in some neoadjuvant chemoimmunotherapy cases for patients with recurrent NPC. One patient who lacked CD20^+^FCRL4^+^ B cells both before and after treatment exhibited minimal response to combined chemoimmunotherapy. Conversely, patients who had CD20^+^FCRL4^+^ B cells present before treatment or who developed these cells afterward displayed notable responses. This suggests that the presence of CD20^+^FCRL4^+^ B cells, or their potential for differentiation, may play a key role in the effectiveness of chemotherapy or immunotherapy. However, based on the three patients included in this study, using CD20^+^FCRL4^+^ B cells or their markers as an indicator for pre‐treatment evaluation may not yet be reliable. If future research provides stronger evidence confirming the link between CD20^+^FCRL4^+^ B cells and chemoimmunotherapy, patients with the potential to differentiate into this cell type may benefit from such treatments. Unfortunately, due to the limited number of patients in this single‐center study, sufficient data could not be obtained. Future multi‐center studies and animal experiments may provide deeper insights into the function and significance of CD20^+^FCRL4^+^ B cells.
Finally, it is necessary to acknowledge the limitations of our study. First, because NPC is primarily treated with radiotherapy, large tumor specimens are difficult to obtain ethically, making it challenging to isolate sufficient numbers of CD20^+^FCRL4^+^ B cells directly from NPC tissues. Consequently, when comparing the functional effects of CD20^+^FCRL4^+^ B cells derived from different sources on CD8^+^ T cells, we were only able to include one patient per group. The NPC sample used in this assay was obtained from a typical exophytic lesion with relatively soft tissue, which incidentally yielded an unusually large amount of material during biopsy. This naturally limits the representativeness and generalizability of the experiment.
Second, the cohort used to explore the association between CD20^+^FCRL4^+^ B cells and immunotherapy response was small and derived from a single center. Therefore, the conclusions drawn from these observations remain hypothesis‐generating rather than definitive. Future multicenter studies with enlarged cohorts will be essential to strengthen these findings.
Third, the representativeness of ARH‐77 as an experimental model warrants caution. The cellular background of ARH‐77 remains a matter of debate: although initially described as exhibiting B‐lymphoblastoid characteristics, subsequent studies suggest that it has undergone immunoglobulin class switching and functions as a mature IgG‐secreting plasma cell, while other reports propose that it represents a transitional state between lymphoblasts and plasma cells. In our study, ARH‐77 retained robust antigen uptake and presentation capabilities, but it is not an ideal model for fully recapitulating the biology of tissue‐derived CD20^+^FCRL4^+^ B cells. In vivo, antigen presentation and other interactions between B cells and T cells typically occur within lymphoid organs, such as lymph nodes and tertiary lymphoid structures. In this study, combined stimulation of ARH‐77 cells in vitro enabled them to exert a certain degree of antigen‐presenting function; however, this environment still differs significantly from the actual extracellular matrix or cellular microenvironment in vivo. While there is sufficient evidence that ARH‐77 cells activate CD8+ T cells through antigen presentation, the exact mechanisms by which CD8+ T cells exert their cytotoxic effects—whether through non‐specific or antigen‐specific targeted killing—remain to be further elucidated. Therefore, the use of ARH‐77 cells inherently carries limitations and can only support very limited investigation into specific potential functions. Lastly, the absence of immunocompetent animal models that faithfully recapitulate NPC poses a major obstacle to the field of NPC immunology. The development of suitable animal models will therefore be essential for advancing mechanistic and translational research.
Materials and Methods
4
Collection of Clinical Samples
4.1
The human tissue and tumor specimens utilized in this study were collected following the guidelines established by the International Council for Harmonization of Medical Sciences and the World Health Organization. Ethical approval was granted by the local ethics committee at Xiangya Hospital of Central South University (202404083), and informed consent was obtained from all patients who participated in this study.
Acquisition and Preprocessing of Bulk RNA‐seq Data
4.2
The bulk RNA sequencing dataset GSE102349, comprising 88 paired primary NPC patients with progression‐free survival (PFS) information, was retrieved from the Gene Expression Omnibus (GEO, https://www.ncbi.nlm.nih.gov/geo/) database. To ensure comparability of expression values across various samples, we converted the raw expression matrix into transcripts per million (TPM) units.
Acquisition and Preprocessing of Single‐Cell RNA Sequencing (scRNA‐seq) Data
4.3
We obtained two single‐cell RNA sequencing datasets (GSE150825 and GSE162025) from GEO. The preprocessing steps were performed using the Seurat V322 package. Quality control procedures were conducted to exclude low‐quality cells based on the following criteria: 1) Gene expression in ≥3 cells; 2) Cells with ≥300 and <3000 features; 3) Cells with mitochondrial gene reads <10% of the transcriptome; and 4) Cells with a count of <10,000. SCTransform was used to normalize the data, addressing batch effects and technical noise across cells. DoubletFinder, utilizing the top 20 principal components (PCs) from principal component analysis (PCA), identified and removed doublets in accordance with the projected doublet rate provided in the Chromium Single‐Cell 30 v2 reagents kit user guide (10X Genomics).
Clustering and Annotation
4.4
Initially, we selected 3000 Highly Variable Genes (HVGs) with significant expression variability. Principal Component Analysis (PCA) was employed to determine the optimal principal components (PCs), which were visualized using an elbow plot. We then applied the Harmony tool, utilizing experimental batches as grouping variables, to correct the PC values and eliminate batch effects in subsequent analyses. The batch‐corrected PC values were used for K‐Nearest Neighbors (KNN) clustering to group similar cells. Subsequent Wilcoxon rank‐sum tests were conducted to compare gene expression between clusters. Genes with a p‐value < 0.05 and a log2 fold change (log2FC) ≥ 0.25 were considered differentially expressed and served as pivotal markers for cell clusters. To identify cell types, we cross‐referenced these markers with the CellMarker database (http://xteam.xbio.top/CellMarker/) and integrated them with previous NPC single‐cell sequencing studies.
Developmental Trajectory Analysis
4.5
To project specific cell population trajectories, we utilized the monocle package. Marker genes were used to define cellular development. However, the monocle algorithm assumes continuous pseudotime and monotonous cell state transitions. To address this limitation, Cytotrace modeled subpopulation relationships as a connected graph, ordering cells based on their developmental sequence. Nevertheless, it lacked detailed insights into differentiation. Therefore, we employed Slingshot to capture nonlinear cell state relationships.
Enrichment Analysis
4.6
Single‐sample Gene Set Enrichment Analysis (ssGSEA) was employed to assess predefined gene set enrichment within the samples. Scores were computed based on gene expressions and then standardized across all samples, using the GSVA package for this purpose. Additionally, gene set enrichment analysis (GSEA) was performed using the clusterProfiler package. For the calculation of log2 fold change (log2FC) values between subtypes, the limma package was utilized. These log2FC values, along with filtered gene sets obtained from the Molecular Signatures Database (MsigDB, https://gsea‐msigdb.org/gsea/msigdb/), were utilized for GSEA. Enrichment terms were compared between subtypes, and significance was determined by a p‐value < 0.05 and a false discovery rate (FDR) < 0.05.
Cell Communication
4.7
Cell communication and interactions between different cell types were analyzed using the cellchat26 tool, which is based on gene co‐expression and cell‐cell interaction networks. A cellchat S4 object was constructed using the quality‐controlled and filtered transcriptome matrix, along with cell type annotations. The ‘CellChatDB.human’ database served as the source for receptor‐ligand interactions to quantify cell communication probabilities, allowing us to deduce significant communication events. Overexpressed ligands, receptors, and interactions were identified within each cell group. The resulting communication network between cell subtypes was revealed, incorporating communication weights/counts, signal pathway strength, and receptor‐ligand interaction probabilities.
Spatial Reconstruction
4.8
CSOmap integrates scRNA‐seq data to visualize and interpret cell spatial patterns. It calculates correlations in gene expression of ligand‐receptor pairs and sets interaction weights to 1. Finally, by utilizing neighborhood relationships and dimensionality reduction, a nonlinear manifold captures the spatial distribution of cells.
Immunophenotype Prediction
4.9
The easier package, in conjunction with various prior knowledge sources, leveraged the patient's RNA sequencing TPM matrix to identify systemic markers in the tumor microenvironment, evaluate immune cell composition, and quantify immune attributes. The initial analysis encompassed immune response hallmarks, immune cell populations, pathway and transcription factor activities, ligand‐receptor interactions, and cell‐cell communications. Subsequently, the integration of these features enabled the prediction of immune therapy response in HNSCC tissue.
Epitope Peptide Prediction
4.10
The Immune Epitope Database (IEDB, https://iedb.org/) predicted and validated epitopes through T and B cell assays as well as MHC binding assays. Only continuous linear epitopes were taken into consideration. Candidate peptides underwent analysis using NetMHC28 to assess their MHC affinity. Alphafold229 was used to predict the spatial structure of peptides, while the molecular structure of HLA‐A was obtained from the Protein Data Bank (PDB, https://rcsb.org/). Preprocessing of HLA‐A was performed using PyMOL 2.6.0, which included the removal of ligands and water molecules as well as hydrogenation. HLA‐A's docking activity pocket was predicted using Yinfotek Cloud Computing (https://cloud.yinfotek.com). Semi‐flexible docking of peptides and HLA‐A was conducted using AutoDock VINA 1.5.6 software with a binding energy threshold of ‐5 kcal/mol. Docking results and peptide structures were visualized using PyMOL.
Peptide Synthesis and Modification
4.11
The peptides employed in this experiment were synthesized by ChinaPeptides (QYAOBIO) Co., Ltd. (Shanghai, China). The peptide sequence CLGGLLTMV was modified with fluorescence by coupling the fluorescent dye FITC to the cysteine (C) residue at the peptide's N‐terminus. The synthesis was followed by purification using high‐performance liquid chromatography (HPLC). The purity of the synthesized peptides utilized in this study was confirmed to be above 95% through mass spectrometry (MS) analysis. The fluorescently modified peptides exhibited absorption and emission wavelengths of 488 and 525 nm, respectively.
Flow Cytometry
4.12
Peripheral blood samples underwent Ficoll (cytiva) gradient separation, resulting in the isolation of peripheral blood mononuclear cells (PBMCs). NPC tissue was processed into single‐cell suspensions. Cultured cells were counted, adjusted to a concentration of ∼10^6^ cells/mL. The above single‐cell suspension was blocked through Fc receptor blocking, incubated with fluorescent antibodies, and then subjected to formaldehyde fixation before being analyzed by flow cytometry. Finally, the results were analyzed using FlowJo software. When cells were flow‐sorted, they were resuspended in serum‐free RPMI 1640 medium (Gibco), maintained aseptic and not fixed with formaldehyde. The antibodies used included CD20‐Pacific Blue (BioLegend, clone 302328), MHC‐I‐APC (eBioscience, catalog number 17‐9958‐42), FCRL4‐PE/Cyanine7 (BioLegend, clone 340208), MHC‐II‐FITC (Proteintech, 65218), and Zombie NIR (BioLegend, clone 423105), CD69‐PE (Invitrogen, 12‐0699‐41), CD8‐ eFluor 506 (Invitrogen, 69‐0088‐42), CD45RO‐FITC (Invitrogen, 11‐0457‐42).
CCK‐8 Assays
4.13
The medium was removed from the cell culture system of 96‐well plates, and after washing three times with PBS, 90 µL of serum‐free DMEM medium was added to each well. Add 10 µL CCK‐8 solution to each well. The incubation was continued in a cell incubator with 5% CO2 at 37°C for 2 h. A 96‐well plate was placed on a microplate reader at 450 nm to determine the optical density (OD) of each well. Cell survival rate was calculated as follows: cell viability = OD(450 nm) of the test group/OD(450 nm) of the blank control group.
ELISA
4.14
The IFNγ (absin, abs510007), TNFα (absin, abs510006), Gzms‐A (abcam, ab255728), Gzms‐B (Proteintech, KE00121) ELISA assay was performed using the corresponding ELISA kit following the manufacturer's instructions. Culture supernatants were collected, centrifuged at 1000 g for 15 min to remove particles. Antibody working solution, streptavidin‐HRP working solution, and chromogenic substrate working solution, as well as a gradient‐diluted standard, were prepared according to the kit instructions. To initiate the assay, 300 µL of washing solution was added to each well, left to soak for 30 s, and then discarded. Different concentrations of standard samples, experimental samples, or quality control samples were added to the corresponding wells, each with 100 µL. The reaction wells were sealed with a plate sealer and incubated at room temperature for 2 h. The liquid in the wells was aspirated, and the plate was washed and air‐dried. Next, 100 µL of detection antibody was added to each well, sealed, and incubated at room temperature for 2 h. After washing the plate, 100 µL of SA‐HRP was added to each well, and incubated at room temperature for 20 min. Following another wash, 100 µL of chromogenic substrate was added to each well and incubated at room temperature for 30 min. Finally, 50 µL of stop solution was added to each well, and the absorbance at 450 nm was measured using a microplate reader.
Preparation of HK‐1 Tumor Cell Lysate
4.15
Nasopharyngeal carcinoma cell line HK‐1 identified by short tandem repeat (STR) profiling was obtained from the cell bank of Hunan Key Laboratory of Otorhinolaryngology Major Diseases. Adherent HK‐1 cells were grown to 70% to 80% confluence and dissociated using trypsin. After centrifugation, the mixture was resuspended in 1 mL of PBS and placed in an ultrasonic crushing machine at a frequency of 80hz until the solution became clear. The insoluble fraction was precipitated by centrifugation at 16,000 g at 4°C. The supernatant was collected as tumor lysate, and its protein concentration was determined at 280 nm by the outer diameter.
Cell Culture and Transfection
4.16
The human B lymphoblast cell lines ARH‐77 and HMy2.CIR, obtained from Abiowell Biotechnology Co., Ltd. (Changsha, China), were authenticated for species origin using STR profiling. PBMC‐derived CD20^+^B cells and CD8^+^T cells were obtained from healthy donors. The cells were cultured in DMEM high‐glucose medium (glucose < 4500 mg/L), supplemented with 10% heat‐inactivated fetal bovine serum and penicillin‐streptomycin. The cell concentration was adjusted to 2×10^6^ cells/mL for seeding in a 24‐well plate. Following 24 h of stimulation with the following reagents, either individually or in combination: 5 ng/mL human IFNγ (Beyotime, P5664), 100 ng/mL soluble CD40L (CST, #32621), 20 ng/mL LPS (Abiowell, AWH0796), 20 ng/mL SPA (Sigma‐Aldrich, P6031), and 10 µg/mL NPC‐lysate, flow cytometry analysis or an antigen peptide uptake experiment was conducted. Following the manufacturer's instructions, Lipofectamine 3000 (Thermofisher) was used to transfect cells with WDFY4 siRNA for 24 h. Then complete medium and subsequent interventions were delivered according to the experimental design. The siRNA were purchased from Sangon Biotech (Wuhan, China). WDFY4 siRNA sequence: 5’‐TGTCGGGAAAGACAAGTTATT‐3’.
Antigen Peptide Uptake and Presenting Experiment
4.17
For the antigen peptide uptake experiment, cells were centrifuged and suspended in serum‐free DMEM medium, and the cell density was adjusted to 2×10^6^ cells/mL before seeding. Fluorescently labeled peptides (1 µg/mL) in PBS, with the prior IFNγ and LPS concentrations, were added. After 24 h of incubation, the cells were washed, placed on a poly‐L‐lysine‐coated glass slide, air‐dried, sealed, and observed under a fluorescence microscope.
T‐B Cell Co‐Culture and Cytotoxicity Assay
4.18
ARH‐77, HMy2.CIR and PBMC‐derived CD20^+^B cells were cultured as previously mentioned. The cell concentration was adjusted to 1×10^6^ cells/mL for seeding in a 6‐well plate. Following 24 h of stimulation with the following reagents: 5 ng/mL human IFN, 10 µg/mL NPC‐lysate. The cells were centrifuged and washed three times with PBS. CD8^+^T cells obtained by flow sorting were then mixed with stimulated ARH‐77 or HMy2.CIR or CD20^+^B cells obtained by flow sorting at a ratio of 1:1. After washing by centrifugation, the pooled cell suspension was resuspended in DMEM medium without serum and inoculated into 6‐well plates, with 1×10^6^ cells/mL of each cell type. Flow cytometry or cytotoxicity assays were performed after 24 h of co‐culture.
For the cytotoxicity assay, HK‐1 cells grown to 70%–80% confluence in 96‐well plates were first washed and replaced with serum‐free DMEM medium. T‐B cells co‐cultured in 6‐well plates for 24 h were centrifuged and washed, resuspended in DMEM, and added to the HK‐1 cell culture system to adjust the cell density to 1×10^6^ cells/mL. After 24 h of action, the supernatant and suspended cells were removed by aspiration, and after washing three times, the cell activity was measured by CCK‐8 assays.
Animal Experiment
4.19
The animal experiments were approved by the Animal Ethics Committee of Central South University (Approval No: CSU‐2024‐0145). Logarithmic phase HK‐1 cells were digested, centrifuged, and resuspended in PBS to create a cell suspension with a concentration of 2×10^7^ cells/ml. A 100 µl aliquot of the suspension was injected subcutaneously into the right axilla of 6‐week‐old BALB/c‐nu mice. On days 7 and 14, 50 µl containing approximately 1×10^6^ cells of either PBMC‐derived CD8^+^ T cells or CD8^+^ T cells co‐cultured with B cells were injected peritumorally and intratumorally. Tumor size was measured on days 7, 14, and 21, and tumor volume was calculated using the formula: Volume = 0.5 × length × width [2].
Immunofluorescence and Immunohistochemistry of Tissues
4.20
Freshly excised tissues were fixed in 10% paraformaldehyde for over 24 h and then paraffin‐embedded. Pathological sections were deparaffinized, followed by antigen retrieval using Tris‐EDTA solution (Servicebio). For immunofluorescence, sections were blocked with 20% BSA at room temperature for 30 min, incubated overnight at 4°C with primary antibodies in PBS, and treated with fluorescently labeled secondary antibodies after washing. Immunohistochemistry used horseradish peroxidase‐conjugated secondary antibodies, followed by DAB chromogen and hematoxylin counterstaining.
The following antibodies were applied: CD69 (Proteintech, 10803‐1‐AP), CD8 (Abcam, ab237709), FCRL4 (Absin, abs133895), CD20 (Proteintech, 60271‐1‐Ig), PD1 (Proteintech, 18106‐1‐AP), and MHC‐I (Proteintech, 15240‐1‐AP). Imaging was conducted using Pannoramic MIDI II‐3Dhistech, and analysis was performed using CaseViewer and ImageJ2 Software. Cell positivity was evaluated based on the proportion of labeled lymphocytes among all lymphocytes: > = 20% (++), > = 5% (+), or <5% or no labeled lymphocyte infiltration (+/‐).
Western‐Blot
4.21
The treated cells were collected, lysed in RIPA buffer, and incubated on a rocker at 4°C for 15 min. The protein concentration of the lysates was measured using a bicinchoninic acid protein assay kit (Beyotime, P0009) and normalized to equal amounts of protein. The protein bands were separated by 6% polyacrylamide SDS‐PAGE, transferred to a PVDF membrane, and probed with the WDFY4 primary antibodies (ABclonal, A4897). Then, the blots were incubated with goat‐anti‐rabbit HRP‐conjugated secondary antibodies (Proteintech, RGAR001). The immunoreactive bands were visualized by enhanced ECL chemiluminescence (Beyotime, P0018S). The same membranes were then reprobed with antibodies against GAPDH (Proteintech, 60004‐1) to confirm equal loading of the samples.
Q‐PCR
4.22
Cells were collected in 1.5 mL EP tubes, and after washing with PBS, 1 mL of Trizol (Invitrogen) was added at room temperature for 5 min. To each tube, 200 µL of chloroform was added, followed by inversion and shaking for 30 s until the mixture turned milky. The tubes were left at room temperature for 5 min. The upper colorless liquid was aspirated and transferred to new 1.5 mL EP tubes, with each tube containing approximately 400–500 µL. An equal volume of isopropanol was added, mixed thoroughly, and left at room temperature for 10 min. The mixture was centrifuged at 4°C and 12 000 rpm for 10 min, resulting in a white precipitate at the bottom of the tube. The precipitate was washed twice with 75% ethanol, air‐dried on ice, and then dissolved in 10 µL of RNase‐free water. The concentration and purity were measured at wavelengths of 260 nm/280 nm using a spectrophotometer. The SweScript RT II First Strand cDNA Synthesis Kit (Servicebio) was used to remove genomic DNA and reverse transcribe mRNA into cDNA. q‐PCR was performed using Fast SYBR Green qPCR Master Mix (Servicebio). The reaction program was set as follows: initial denaturation at 95°C for 30 s, followed by denaturation at 95°C for 10 s and annealing/extension at 60°C for 30 s for 40 cycles. Primers were synthesized by Sangon Biotech (Shanghai). The forward primer for WDFY4 was 5’‐GCCTCATCCCCTCCAAG‐3’, and the reverse primer was 5’‐CTGCCTCTCATTCACCGA‐3’.
Bioinformatics Analysis and Statistical Analysis
4.23
Matlab was utilized to facilitate the spatial reconstruction analysis with CSOmap. The R language was employed to conduct a wide range of analyses, including single‐cell transcriptome workflows, cell communication assessments, developmental trajectory analysis, single‐cell enrichment evaluations, and standard transcriptome analyses. Survival analysis was carried out using the survival package and survminer, involving the generation of survival curves and log‐rank tests. Correlation analysis was performed using ggscatter for visualization and coefficient calculations, while ggplot2 and ggsignif were used for statistical tests and mean/median value comparisons. Correlation coefficients and heatmaps were generated with corrplot. Heatmaps and hierarchical clustering were created using pheatmap and its built‐in algorithms.
For statistical tests, t‐tests were used for normally distributed and homoscedastic independent samples, Wilcoxon rank‐sum tests were applied for non‐normally distributed data, and paired t‐tests were utilized for paired samples. One‐way ANOVA was employed for comparing multiple samples, followed by pairwise comparisons using the SNK‐Q test. Statistical significance was defined as p < 0.05 (^^), p < 0.01 (^^), p < 0.001 (^^), and p < 0.0001 (^****^).
In the bioinformatics analyses, we used published computational tools, including Seurat V3 [49], Monocle [50], Cytotrace [51], Slingshot [52], CellChat [53], Easier [54], NetMHC [55], and Alphafold2 [56].
Author Contributions
Weihong Jiang, Benjian Zhang, Hua Zhang, Yongzhen Mo, and Xiaotian Yuan designed the study. Xiaotian Yuan, Benjian Zhang, Yunqing Liu, and Keilei Gao performed the experiments. Benjian Zhang, Lai Meng, Bo You, and Zirong Chen interpreted the pathology slides and MRI images. Yaxuan Wang, Benjian Zhang, Shaobing Xie, Shumin Xie, Keilei Gao, Ruohao Fan, Fengjun Wang, Junyi Wang, and Zhihai Xie collected the specimens. Xiaotian Yuan, Yuanqing Liu, and Zijian Dong were responsible for bioinformatics and data analysis. Xiaotian Yuan, Benjian Zhang, and Kelei Gao wrote the final manuscript.
Funding
This research was funded by the National Natural Science Foundation of China (82171118, 82403801), the Central South University Graduate Students Independent Exploration and Innovation Project (1053320222011), and the Natural Science Foundation of Hunan Province (2021JJ41027, 2024JJ6631). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Conflicts of Interest
The authors declare no conflict of interest.
Declarations—Ethics Approval and Consent to Participate
The studies involving human participants were reviewed and approved by the Xiangya Hospital Research Ethics Committee of Central South University (No. 2025010747). The patients/participants provided their written informed consent to participate in this study. Written informed consent was obtained from the individual(s) for the publication of any potentially identifiable images or data included in this article.
Supporting information
Supporting File 1: advs73501‐sup‐0001‐SuppMat.docx.
Supporting File 2: advs73501‐sup‐0002‐Data.zip.
Supporting File 3: advs73501‐sup‐0003‐Data.zip.
The reference list from the paper itself. Each links out to its DOI / PubMed record.
- 1L. Gong , D. L.‐W. Kwong , W. Dai , et al., “Comprehensive Single‐cell Sequencing Reveals the Stromal Dynamics and Tumor‐specific Characteristics in the Microenvironment of Nasopharyngeal Carcinoma,” Nature Communications 12 (2021): 1540.10.1038/s 41467-021-21795-z PMC 794380833750785 · doi ↗ · pubmed ↗
- 2Y.‐P. Chen , A. T. C. Chan , Q.‐T. Le , P. Blanchard , Y. Sun , and J. Ma , “Nasopharyngeal carcinoma,” The Lancet 394 (2019): 64–80.10.1016/S 0140-6736(19)30956-031178151 · doi ↗ · pubmed ↗
- 3K. C. W. Wong , E. P. Hui , K.‐W. Lo , et al., “Nasopharyngeal Carcinoma: An Evolving Paradigm,” Nature Reviews Clinical Oncology 18 (2021): 679–695.10.1038/s 41571-021-00524-x 34194007 · doi ↗ · pubmed ↗
- 4H. S. Eddington , M. Mc Leod , A. W. Trickey , N. Barreto , K. Maturen , and A. M. Morris , “Patient‐reported distress and age‐related stress biomarkers Among colorectal cancer patients,” Cancer Medicine 10 (2021): 3604–3612.33932256 10.1002/cam 4.3914 PMC 8178484 · doi ↗ · pubmed ↗
- 5B. B. Y. Ma , W.‐T. Lim , B.‐C. Goh , et al., “Antitumor Activity of Nivolumab in Recurrent and Metastatic Nasopharyngeal Carcinoma: An International, Multicenter Study of the Mayo Clinic Phase 2 Consortium (NCI‐9742),” Journal of Clinical Oncology 36 (2018): 1412–1418.29584545 10.1200/JCO.2017.77.0388 PMC 5941615 · doi ↗ · pubmed ↗
- 6S. Obeng‐Gyasi , C. Ong , and E. S. Hwang , “Contemporary Management of Ductal Carcinoma in Situ and Lobular Carcinoma in Situ,” Chinese Clinical Oncology 5 (2016): 32.27197512 10.21037/cco.2016.04.02 · doi ↗ · pubmed ↗
- 7J.‐Y. Xu , X.‐L. Wei , Y.‐Q. Wang , and F.‐H. Wang , “Current Status and Advances of Immunotherapy in Nasopharyngeal Carcinoma,” Therapeutic Advances in Medical Oncology 14 (2022): 17588359221096214.35547095 10.1177/17588359221096214 PMC 9083041 · doi ↗ · pubmed ↗
- 8W. Fang , Y. Yang , Y. Ma , et al., “Camrelizumab (SHR‐1210) Alone or in Combination With Gemcitabine plus Cisplatin for Nasopharyngeal Carcinoma: Results From Two Single‐arm, Phase 1 Trials,” The Lancet Oncology 19 (2018): 1338–1350.30213452 10.1016/S 1470-2045(18)30495-9 · doi ↗ · pubmed ↗