You are here

Article Text

Article menu
Basic tumor immunology
Original research
Intratumoral tertiary lymphoid structure (TLS) maturation is influenced by draining lymph nodes of lung cancer
  1. Miao He1,
  2. Qihua He1,
  3. Xiuyu Cai2,
  4. Jun Liu1,
  5. Hongshen Deng1,
  6. Feng Li1,
  7. Ran Zhong1,
  8. Yi Lu1,
  9. Haoxin Peng1,
  10. Xiangrong Wu1,
  11. Zisheng Chen1,
  12. Shen Lao1,
  13. Caichen Li1,
  14. Jianfu Li1,
  15. Jianxing He1,3 and
  16. Wenhua Liang1,4
  1. 1Department of Thoracic Surgery and Oncology, The First Affiliated Hospital of Guangzhou Medical University, China State Key Laboratory of Respiratory Disease & National Clinical Research Center for Respiratory Disease, Guangzhou, Guangdong, China
  2. 2Department of General Internal Medicine, Sun Yat-sen University Cancer Center, Guangzhou, Guangdong, China
  3. 3Department of surgery, Southern Medical University, Guangzhou, Guangdong, China
  4. 4Department of Medical Oncology, The First People Hospital of Zhaoqing, Zhaoqing, Guangdong, China
  1. Correspondence to Professor Wenhua Liang; liangwh1987{at}163.com

Abstract

Background Tertiary lymphoid structure (TLS) is an organized infiltration of immune cells, showing features of germinal center (GC) commonly seen in secondary lymphoid organs. However, its relationship with tumor-draining lymph nodes (TDLNs) has not been studied and we hypothesized that TDLN may influence maturation of intratumoral TLS in non-small cell lung cancer (NSCLC).

Methods Tissue slides of 616 patients that had undergone surgeries were examined. Cox proportional hazard regression model was used to assess risk factors of patients’ survival, and logistic regression model was used for their relationship with TLS. Single-cell RNA-sequencing (scRNA-seq) was employed to explore transcriptomic features of TDLNs. Immunohistochemistry, multiplex immunofluorescence and flow cytometry were performed to analyze cellular composition. Cellular components of NSCLC samples from The Cancer Genome Atlas database were inferred with Microenvironment Cell Populations-counter (MCP-counter) method. Murine NSCLC models were used to dissect underlying mechanisms for relationship between TDLN and TLS maturation.

Results While GC+ TLS was associated with better prognosis, GC− TLS was not. TDLN metastasis reduced the prognostic relevance of TLS, and was associated with less GC formation. Primary tumor sites showed reduced B cell infiltration in TDLN-positive patients, and scRNA-seq revealed diminished memory B cell formation in tumor-invaded TDLNs, together with an emphasis on weakened interferon (IFN)-γ response. Murine NSCLC models revealed that IFN-γ signaling is involved in memory B cell differentiation in TDLNs and GC formation in primary tumors.

Conclusions Our research emphasizes the influence of TDLN on intratumoral TLS maturation and suggests a role of memory B cells and IFN-γ signaling in this communication.

  • lung neoplasms
  • B-lymphocytes
  • tumor microenvironment

Data availability statement

Data are available on reasonable request. The raw sequence data reported in this paper has been deposited into CNGB Sequence Archive (CNSA) of China National GeneBank DataBase (CNGBdb) with accession number CNP0002869 and is publicly accessible at: https://db.cngb.org/cnsa/. Detailed information will be available from the corresponding authors on reasonable request.

http://creativecommons.org/licenses/by-nc/4.0/

This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See http://creativecommons.org/licenses/by-nc/4.0/.

https://doi.org/10.1136/jitc-2022-005539

Statistics from Altmetric.com

    Request Permissions

    If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

    WHAT IS ALREADY KNOWN ON THIS TOPIC

    • Tertiary lymphoid structure (TLS) is associated with increased antitumor immunity in solid tumors and shows similar features with secondary lymphoid organs, but the role of draining lymph nodes in the antitumor immune response of TLS is unknown.

    WHAT THIS STUDY ADDS

    • Our study shows that intact tumor-draining lymph nodes (TDLNs) of lung cancer promote maturation of intratumoral TLS, and that memory B cells regulated by interferon (IFN)-γ signaling may be the underlying mechanism.

    HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

    • Our study emphasizes the relationship between TDLN and intratumoral TLS maturation, and suggests the roles of memory B cells and IFN-γ signaling in this communication.

    Background

    Non-small cell lung cancer (NSCLC) is a prevalent type of cancer worldwide and the leading cause of cancer-related fatalities.1 With the proliferation of cancer screening methods, a growing number of patients with early to mid-stage NSCLC have been detected, who are potential candidates for surgical resection. Nonetheless, recurrence would still occur in large numbers of patients, and the long-term disease-free survival (DFS) rates are subpar.2 In recent years, adjuvant therapies, including chemotherapy, targeted treatments, and immunotherapy have demonstrated considerable efficacy in extending NSCLC survival.3–5 The success of immunotherapy in NSCLC underscores the necessity of comprehending the underlying mechanisms of effective antitumor immunity.6 Tumor-infiltrating lymphocytes (TIL), such as cytotoxic CD8+ T cells, possess potent antitumor capabilities and are correlated with enhanced survival outcomes in patients with NSCLC.7 Recent studies have highlighted the role of tertiary lymphoid structures (TLS), which serve as a hub for antitumor immunity with its organized arrangement of TILs.8

    Based on the examination of H&E staining of tissue slides, TLS is typically defined as an accumulation of infiltrating lymphocytes in non-hematopoietic organs, prompted by chronic inflammation such as infectious diseases, autoimmune diseases, and cancer.9 Its role in antitumor immune responses may encompass the enhancement of effector and memory T cells and plasma cells.10 11 The existence of TLS is positively correlated with a reduced risk of recurrence after surgical resection of solid tumors and regarded as a predictive indicator of a more favorable response to immunotherapies.12–14 Furthermore, recent studies have unveiled that the antitumor impact of TLS is contingent on an efficient germinal center (GC) response, which also determines its prognostic significance.15 16 As a manifestation of TLS maturation, the GC reaction involves the activation of lymphocytes through antigen binding, followed by their differentiation into effector cells that can eradicate tumor cells.17 However, the process of TLS maturation in NSCLC and the potential mechanisms underlying GC formation are still largely unknown.

    In addition to the conventional perspective that effective adaptive antitumor responses occur in secondary lymphoid organs (SLOs), such as lymph nodes (LNs), where major histocompatibility complex-peptide complexes are presented to T cells by mature dendritic cells (DCs) and B cells are activated on antigen binding, TLS has been demonstrated to reinforce antitumor immune defense directly within the tumor.12 However, considering their shared features in antitumor immune responses, the possibility of a connection between TLS and TDLNs cannot be ruled out. Apart from serving as an immune reaction platform, TDLNs can also be a substantial provider of TILs,18 which migrate into primary tumors and participate in the formation of intratumoral TLS. Recent investigations of the impact of immune checkpoint blockers (ICBs) as neoadjuvant therapy for NSCLC have also indicated that LN metastasis is linked to reduced treatment efficacy.19–21 With the significance of TLS in patients’ responsiveness to immunotherapy,14 22 it would be possible that TDLNs, acting as SLOs, may to some extent influence the function of intratumoral TLSs.

    While TLS is composed of multiple types of cells, current studies pointed to a central role of B cells in its formation and maturation.22 Histological evaluation confirms the presence of B cells within TLS and clonal expansions of B cells have been observed in responders to immunotherapy.22 23 On one hand, in TDLNs, B cell receptor (BCR) affinity maturation can occur to generate effector plasma cells and long-lived memory B.24 On the other hand, tumor-infiltrating B (TIL-B) cells can undergo GC reaction in primary lesions, forming the typical light and dark zones of mature TLSs. As TDLNs can be an important source of TIL-B cells, communications between TDLNs and primary tumor sites are very much likely to exist.25 However, evidence for this communication is still lacking, and we do not know much whether the differentiation of B cells in TDLNs influences the GC response in primary lesions of lung cancer.

    In this research, through a morphological examination of a series of 616 patients that have undergone surgical resection, we first studied the prognostic significance of intratumoral TLS maturation and focused on the effects that TDLNs may exert. Then single-cell RNA sequencing (scRNA-seq) of TDLNs and murine models of lung cancer, together with immunohistochemical (IHC) staining, multiplex immunofluorescence (MIF) staining and flow cytometry analysis of primary tumor lesions and TDLNs were employed to explore the possible cellular mechanisms underlying the roles that TDLNs may play in the maturation of intratumoral TLS.

    Methods

    Patient cohort

    From 2009 to 2011, consecutive series of 681 patients with stage I–III NSCLC that have undergone standard surgical resection at the First Affiliated Hospital of Guangzhou Medical University (HGMU) were collected as candidates for this study. Among these cases, 616 cases with a median age of 60 years while being admitted were with qualified tissue slides that can be used for TLS examination and thus recruited into the study cohort; 57.79% of the patients were male and 42.21% of them were female. Tissue slides of resected samples were examined by experienced pathologists to identify tumors in the primary sites and draining LN.

    The inclusion criteria of this study were: (1) patients with single primary NSCLC; (2) stage between tumor, node, metastases (TNM) IA and IIIB; (3) patients that have undergone anatomical resection in combination with lymphadenectomy (systematic LN sampling or systematic LN dissection); (4) patients with resected tissues and LNs that have undergone pathological examination; and (5) patients with sufficient resected tissues for IHC and MIF tests. The exclusion criteria were: (1) patients with multiple lesions; (2) patients with small cell lung cancer or non-invasive lung cancer; (3) patients with diagnostic biopsy before surgical resection; and (4) patients with preoperative neoadjuvant therapy.

    Immunohistochemical and multiplex immunofluorescence examination

    The tumor tissues were fixed in 10% formalin, embedded in paraffin, and serially sectioned; 4 μm thick section was cut from formalin-fixed paraffin-embedded lung cancer tissues for each staining. The slides were deparaffinized, rehydrated, and subjected to epitope retrieval by boiling in Tris-EDTA (pH=9.0) for 20 min at 97°C. Endogenous peroxidase was then blocked by incubating in Antibody Diluent/Block (PerkinElmer #72424205, Massachusetts, USA) for 10 min, and protein was then blocked with 0.05% Tween solution containing 0.3% bovine serum albumin (BSA) for 30 min at room temperature.

    For IHC staining, sections were incubated with the primary antibodies in phosphate-buffered saline (PBS) with 3% BSA, which was followed by incubations with secondary antibodies at room temperature for 2 hours. TLS markers used for IHC staining included CD3 (ab16669, Abcam), CD20 (ab9475, Abcam), CD21 (ab75985, Abcam), CD23 (ab135386, Abcam), and DC-LAMP (ab271053, Abcam). For MIF staining, only one antigen was detected in each round, which includes primary antibody incubation, secondary antibody incubation, tyramine signal amplification (TSA) visualization, followed by labeling the next antibody after epitope retrieval and protein blocking as before. For primary tumor samples, CD4 (ZM0418, Zsbio), CD20 (ab9475, Abcam), CD38 (ZM-0422, Zsbio), CD66b (ab214175, Abcam), and FoxP3 (ab20034, Abcam) for panel 1, and CD8 (ZA-0508, Zsbio), CD68 (ZM-0060, Zsbio), CD133 (ab19898, Abcam), CD163 (ZM-0428, Zsbio), and programmed cell death protein ligand-1 (PD-L1) (13684s, Cell Signaling Technology) for panel 2 were sequentially detected. For selected LN and primary tumor samples, CD19 (ab134114, Abcam), CD27 (ab268144, Abcam), NCR1 (ab199128, Abcam), interferon (IFN)-γ (ab231036, Abcam) were also examined through immunofluorescence (IF) staining.

    Primary antibodies for CD8, CD20, CD38, CD66b, CD68, CD163, PD-L1, FoxP3, CD19, CD27, NCR1, and IFN-γ were incubated for 1 hour at room temperature, CD4, CD133, CD3, CD21, CD23, and DC-LAMP were incubated for overnight at 4°C. Antirabbit/antimouse horseradish peroxidase (HRP) antibodies (Zsbio # PV-6002 or PV-8000) were used as secondary antibody and incubated at 37°C for 10 min for IHC staining. For MIF of primary tumor samples, TSA visualization was performed with the Opal 7-Color multiplex IHC kit (NEL797B001KT, PerkinElmer), containing fluorophores (4’,6-diamidino-2-phenylindole (DAPI)), Opal 520 (CD20 and CD163), Opal 540 (CD38), Opal 570 (PD-L1 and CD4), Opal 620 (CD8), Opal 650 (CD66b and CD133), Opal 690 (CD68 and FoxP3), and TSA Coumarin system (NEL703001KT, PerkinElmer). After labeling all antigens for each panel, microwave treatment was performed to remove the TSA-antibody complex with Tris-EDTA buffer for 20 min at 97°C. All IF slides were counterstained with DAPI for 5 min and enclosed in Antifade Mounting Medium (NobleRyder #I0052, Beijing, China). Fresh whole-tissue section cuts from normal human tonsils were included in each staining batch as the positive control. Slides were then scanned using the PerkinElmer Vectra (Vectra 3.0.5; PerkinElmer). Cell infiltration levels were calculated as the number of immunopositive cells×100% divided by total number of cells in the image scan.

    Identification of TLS and GC in lung cancer

    For each case, slides were reviewed by two independent pathologists who were specialized in lung cancer and blind to the patients’ clinical data. TLS was analyzed under a 10× high-power field (HPF) in all tumor-containing diagnostic sections of the HGMU cohort and murine tumor samples through the examination of dense aggregates of lymphocytes. The existence of intratumoral TLS was assessed morphologically using previously published scales.15 16 For patient tumor samples, in a randomly selected HPF of intratumoral region from the H&E-stained slide, tumors with at least one intratumoral TLS were classified as TLS-positive (TLS+), and tumors without any TLS were classified as TLS-negative (TLS−); tumors would be considered as GC-positive (GC+) if at least one intratumoral TLS showed the characteristic morphology of proliferating centroblasts, and as GC-negative (GC−) if none of the TLSs show the GC features. Also, TLS density in all intratumoral region with lymphocytic aggregates were examined; 0.1 TLS/mm2 was used as a threshold to classify TLS_Low and TLS_High tumors. GC positivity score was also calculated based on the percentage of GC after counting at least 5–10 TLS, and patients with GC percentage larger or equal to 25% were classified as GC_High. To better characterize the maturation of TLS, multiplex IHC staining for CD3, CD20, CD21, CD23, and DC-LAMP were also carried out, in which dense lymphocytic aggregates with CD21 and CD23 signal were designated as the formation of GC. For murine tumor samples, total numbers of GC formation were also counted and compared between different experimental groups.

    Preparation of single-cell samples

    After surgical resection, eight TDLN samples from four patients were transported in ice-cold RPMI1640 (Gibco, Life Technologies) culture media immediately. The LN tissues were rinsed with PBS (Gibco, Life Technologies), minced into 1 mm3 cubic pieces, and were digested with 0.25% trypsin (Gibco, Life Technologies). After being terminated by culture media supplemented with 10% fetal bovine serum (Gibco, Life Technologies), 10 mL of digestion medium containing collagenase IV (100 U/mL; Gibco, Life Technologies) and dispase (0.6 U/mL; Gibco, Life Technologies) was added to the samples. The digested samples were then filtered through a 70 µm nylon mesh, and filtered cells were collected through centrifuging at 120× g and 4°C for 5 min. Cell pellets were resuspended with ice-cold red blood cell lysis buffer and further subjected to a 40 µm nylon mesh. Following centrifugation, the cells were then collected with PBS, and numbers of live cells were counted using an automated cell counter. Throughout this procedure, cells were maintained on ice whenever possible, and the entire process was completed in <1 hour.

    Droplet-based single-cell sequencing

    Single-cell suspensions were converted to barcoded scRNA-seq libraries by using the Single Cell 3′ Library and Gel Bead Kit (10X Genomics) and Chromium Single Cell A Chip Kit (10X Genomics). According to the manufacturer’s instructions, the cell suspension was loaded onto the chromium single-cell controller (10X Genomics) to generate single-cell gel beads in the emulsion (GEMs). Approximately 10 000 cells were added to each channel, and approximately 6000 cells were recovered. After cell lysing, the released RNA was barcoded through reverse transcription reaction in individual GEMs, and complementary DNA was generated. scRNA-seq libraries were then constructed using the Single Cell 3′ Library and Gel Bead Kit V3 and sequencing was done using an Illumina NovaSeq 6000 with a paired-end 150 base pair (PE150) reading strategy (performed by CapitalBio Technology, Beijing, China).

    scRNA-seq data processing

    Raw expression matrices were generated for each sample with the Cell Ranger (V.6.0.1) pipeline using human reference version GRCh38, and the output-filtered gene expression matrices were analyzed by the Seurat package (V.4.0.3).26 Cells were classified as low-quality ones if they met the following criteria: (1) <500 unique molecular identifiers (UMIs), (2) >5000 or <200 genes, or (3) >10% UMIs derived from the mitochondrial genome. After the removal of low-quality cells, the NormalizeData function was employed to normalize the gene expression matrices, and the FindVariableFeatures function was used to calculate 2000 features with high cell-to-cell variation. After linear transformation with default parameters by the ScaleData function to generate scaled data, the RunPCA function was applied to reduce the dimensionality of the datasets. ElbowPlot and JackStrawPlot were also done to identify the dimensionality of each dataset. Last, cells were clustered using the FindNeighbors and FindClusters functions, and non-linear dimensional reduction was performed with the RunUMAP and RunTSNE function. To compare cell types and proportions between samples, we used the integration methods included in the Seurat package to assemble multiple datasets and reduce batch effects. In brief, after the identification of 2000 features with high cell-to-cell variation, anchors between individual datasets were generated with the FindIntegrationAnchors function. IntegrateData function was then applied with the anchors to create a batch-corrected expression matrix of cells from different datasets that will be used for further analysis.

    Cell type annotation and cluster marker identification

    After non-linear dimensional reduction, cells from TDLNs were projected into a two-dimensional space by the method of uniform manifold approximation and projection (UMAP) and t-distributed stochastic neighbor embedding (tSNE). In this way, cells sharing common features can be clustered together. The FindAllMarkers function in the Seurat package was then used to find markers for each of the identified clusters. Annotations of the clusters were based on the expression of canonical and widely accepted markers of the corresponding cell types.

    Defining cell state scores

    Cell scores, which were based on the average expression of the genes from certain predefined gene sets, were calculated to evaluate the degree to which individual B lymphocyte belongs to the corresponding cell state. The AddModuleScore function in Seurat was used to implement this method, and the naïve, memory, and GC scores for each B lymphocyte were measured. In this analysis, the naïve markers (BACH2, ZBTB16, APBB2, SPRY1, TCL1A, and IKZF2), memory markers (CD27, CD86, RASSF6, TOX, TRERF1, TRPV3, POU2AF1, RORA, TNFRSF13B, CD80, and FCRL5), and GC markers (EBI3, BCL2A1, LMO2, GMDS2, PRPSAP2, SERPINA9, MARCKSL1, CD38, BCL6, SUGCT, EZR, ISG20, and AICDA) were used to define naïve, memory, and GC scores.

    Inference of the developmental trajectory for B lymphocytes

    The cell state transitions for B lymphocytes were estimated using the Monocle (V.2) algorithm.27 First, single cells clustered as B lymphocytes were selected from the whole dataset. The gene expression matrix in the scale of UMI counts was used as input to Monocle, and the newCellDataSet function was applied to construct the monocle cell dataset. The top 10 markers of each B cell cluster were chosen as the ordering genes to define cell progress, and the DDRTree method was called on to reduce cell dimensionality. Separate trajectories for different B cell clusters, as well as tumor-free and tumor-invaded TDLNs, were then inferred with the orderCells function.

    Putative interactions between cell types

    Putative interactions between different cell types were inferred through the receptor-ligand analysis via the CellChat algorithm,28 which can quantitatively analyze intercellular communication networks from scRNA-seq data. To infer cell-cell communication networks in TDLNs, the standard workflow is followed and normalized counts were loaded into CellChat. Preprocessing functions including identifyOverExpressedGenes, identifyOverExpressedInteractions, and projectData were applied with default parameters. For the databases in use, we selected the secreted signaling pathways and the cell-cell contact. We also used the precompiled human protein-protein interactions as a priori network information. For the main analyses, core functions including computeCommunProb, computeCommunProbPathway, and aggregateNet were called on for data from tumor-free and tumor-invaded TDLNs, respectively, which were then merged using the mergeCellChat function. To deduce potential induced or reduced communications with B lymphocytes as signal receivers, rankNet function was applied to compare the overall information flow of each signaling pathway, and netVisual_bubble function was employed to depict the communication probabilities mediated by ligand-receptor pairs from certain cell groups to other ones.

    Deconvolution of the cellular composition with MCP-counter

    Gene expression and clinical data for The Cancer Genome Atlas (TCGA)-LUAD and TCGA-LUSC datasets were downloaded from UCSC Xena (http://xena.ucsc.edu/). Then normalized log2-transformed FPKM expression matrix was loaded into the MCP-counter package (V.1.1.0),29 which produces the absolute abundance scores for eight major immune cell types (CD3+ T cells, CD8+ T cells, cytotoxic lymphocytes, natural killer (NK) cells, B lymphocytes, monocytic lineage cells, myeloid DCs, and neutrophils), endothelial cells (ECs), and fibroblasts. The deconvolution profiles were then hierarchically clustered and compared across the samples from LN-negative and LN-positive patients.

    Gene set variation and gene signature enrichment analysis

    Pathway analyses were performed on the 50 hallmark pathways as described in the molecular signature database.30 To assign pathway activity estimates to individual cells, gene set variation analysis (GSVA) was conducted with standard settings implemented in the GSVA R package (V.1.40.1).31 The differential activities of pathways between B lymphocytes from tumor-free and tumor-invaded TDLNs were calculated using Limma R package (V.3.48.3).32 Significantly disturbed pathways were identified with Benjamin-Hochberg-corrected p value of ≤0.01.

    Differential gene expression for B lymphocytes from tumor-free and tumor-invaded TDLNs was analyzed using the FindAllMarkers function with Benjamin-Hochberg false discovery rate (FDR) correction. Genes were regarded significantly different if an FDR <0.05, an average log fold change >0.1, and the gene was detected in >20% of cells in the corresponding group. Gene Ontology (GO) enrichment analyses were then performed with clusterProfiler R package (V.4.0.5).33

    Murine model of lung cancer

    C57BL/6J mice were purchased from GemPharmatech (Nanjing, China), and were housed under specific pathogen-free conditions. Recombinant mouse IFN-γ (485-MI-100) was purchased from R&D Systems (Minneapolis, Minnesota, USA). Antimouse IFN-γ antibody (XMG1.2) and antimouse NK1.1 antibody (PK136) was purchased from BioLegend (San Diego, California, USA). Fingolimod (FTY720) was purchased from APExBIO (Houston, Texas, USA).

    For orthotopic xenograft model of lung cancer, 1×106 logarithmically growing Lewis lung carcinoma (LLC1, American Type Culture Collection (ATCC) CRL1642) cells were suspended in 50 μL Matrigel (Becton Dickinson, California, USA) and injected into the left lateral thorax of mice as described previously.34After 14 days of tumor development, mice were randomized into three groups: (1) control group (PBS treatment), (2) IFN-γ group (IFN-γ 1 μg/mouse intravenously, every other day), (3) IFN-γ+FTY720 group (IFN-γ 1 µg/mouse intravenously and FTY720 0.1 mg/kg intraperitoneally, every other day). We performed the experiment with five mice per group. After 28 days of tumor development, the mice were sacrificed and tumor samples were collected.

    For footpad xenograft model of lung cancer, 5×105 logarithmically growing LLC1 cells were suspended in 50 μL culture media and injected into the left footpad of mice and afferent lymphatic vessels of popliteal LNs were ligated 1 week later to prevent spontaneous LN metastasis.35 After 14 days of tumor development, mice were randomized into four groups: (1) control group (PBS treatment), (2) anti-IFN-γ group (antimouse IFN-γ antibody 20 μg/mouse intraperitoneally, every other day), (3) tumor-invaded TDLN group (injection with 1×105 LLC1 cells suspended with 20 μL culture media into left inguinal LN), (4) tumor-invaded TDLN plus IFN-γ group (1×105 LLC1 cells injected into left inguinal LN, plus IFN-γ 1 μg/mouse intravenously every other day). We performed the experiment with five mice per group. After 28 days of tumor development, the mice were sacrificed, tumors and draining LNs were collected. For the NK cell depletion experiment, 250 μg of anti-NK1.1 (clone PK136) is injected intravenously on days −8 and −4 before harvest of LN samples.

    Cell lines

    The LLC1 cells were purchased from ATCC, which performs cell line authentication by short tandem repeat DNA profiling. Newly bought cell lines were expanded to the third passage, stored in aliquots in liquid nitrogen, and were used for no more than 4 months after resuscitation from cryopreservation. Cells were maintained under standard culture conditions and tested for Mycoplasma every 6 months using the Universal Mycoplasma Detection Kit (ATCC).

    Protein extraction and western blot analysis

    Proteins were collected by lysing of cells in radioimmunoprecipitation assay buffer after the collection of the LN samples. Equal amount of protein from the control and NK cell-depleting group was loaded into sodium dodecyl sulfate-polyacrylamide gel electrophoresis, which were then separated and transferred onto polyvinylidene fluoride (PVDF) membranes (Millipore, Billerica, Massachusetts, USA). The membranes were blocked with skim milk, incubated with appropriate dilutions of specific primary antibodies, and with HRP-conjugated secondary antibodies. The β-actin (C4) and IFN-γ (XMG1.2) antibodies were purchased from Santa Cruz Biotechnology (Dallas, Texas, USA). Visualization was done through the ECL system (Thermo Fisher Scientific, Rochester, New York, USA).

    Flow cytometry analysis

    The single-cell suspensions of TDLNs were suspended in Fluorescence Activated Cell Sorting (FACS) buffer (PBS, 1% BSA) and stained with Fixable Viability Stain 510 (Cat. No. 564406, BD Biosciences) for 15 min under room temperature. The cells were then blocked with antimouse CD16/32 antibodies (Mouse BD Fc Block, BD Biosciences) for 10 min before staining with the specific flow cytometry antibodies. The staining antibodies are listed in online supplemental table S10. Flow cytometry data were acquired on a FACSVerse flow cytometer (BD Biosciences, BD FACSDiVa Software V.8.0.1.1) and analyzed with FlowJo V.10.6.2 software (TreeStar).

    Supplemental material

    Supplemental material

    Statistical analysis

    Continuous variables were expressed as medians with IQRs, and categorical variables were expressed as numbers and proportions. Pearson’s χ2 tests were used to analyze differences of categorical variables between exposure groups. Wilcoxon and Kruskal-Wallis test were used to compare the distribution of immunopositive cells in IHC and MIF staining. Student’s t-tests were used to compare GC formation in H&E staining and flow cytometry data. The Kaplan-Meier survival curve and Cox proportional hazards regression analysis were then used for assessing the relationship between possible risk factors and DFS, and a logistic regression model was used for their relationship to TLS formation and maturation. To control for the potential selection bias, propensity score-matched (PSM) analysis was performed. We estimated the propensity score for each patient with or without LN metastasis using a multivariate logistic regression model. The variables in the model included age, gender, histology, tumor size, pleural invasion, and vessel tumor embolism for the HGMU cohort and age, gender, histology, and tumor size for the TCGA cohort. We performed 1:1 nearest-neighbor matching based on the propensity score and confirmed the success of the propensity matching procedure by recomparing the LN-negative and LN-positive groups. scRNA-seq data analysis was done by R software (V.4.1.1), meta-analysis was done with the STATA V.12.0 software (StataCorp, College Station, Texas, USA) and other statistical analyses were performed using SAS V.9.4 (SAS Institute, Cary, North Carolina, USA). Two-tailed p values <0.05 were considered statistically significant for all tests.

    Results

    Maturation of intratumoral TLS is associated with better survival after surgical resection

    As is shown in online supplemental table S1, the HGMU cohort consists of 616 patients with NSCLC who underwent surgical resection between 2009 and 2011, with 458 cases being adenocarcinoma and 158 cases being squamous cell carcinoma. During a median follow-up of 55.1 months, 251 recurrences were identified. As is shown in online supplemental table S2, together with female gender, earlier TNM staging, no visceral pleural invasion, and larger number of resected LNs, intratumoral TLS formation (HR=0.75, p=0.033) was associated with a reduced risk of recurrence. In a multivariate analysis that included age, gender, histology, TNM stage, visceral pleura invasion, vessel tumor embolism, and number of resected LNs as covariates, the HR of intratumoral TLS formation was 0.79 with a p value of 0.070.

    Recent studies suggested that the GCs formed within intratumoral TLS might determine the strength of antitumor immunity.15 To further investigate this, we classified the intratumoral TLS into GC+ TLS and GC− TLS (figure 1A, online supplemental figure S1A–C). As is shown in figure 1B and online supplemental table S3, while GC− TLS shows no prognostic significance in patients with NSCLC after surgery, cases with GC+ TLS were associated with a significantly lower risk of recurrence in both the univariate (HR=0.52, p=0.002) and multivariate (HR=0.58, p=0.012) analysis. We also classified TLS formation based on TLS density with a threshold of 0.1/mm2. The GC positivity score was then calculated based on the percentage of GC formation after counting at least 5–10 TLS, and patients with GC percentage ≥25% were classified as GC_High.

    Supplemental material

    Figure 1
    Figure 1

    Association between TLS and survival after surgical resection. (A) Representative images of H&E staining of cases with no intratumoral TLS, intratumoral GC− TLS, and intratumoral GC+ TLS. (B) Kaplan-Meier survival curves for DFS of patients with and without intratumoral TLS (left); and patients with intratumoral GC+ TLS compared with those with GC− TLS or no TLS (right). Log-rank (Mantel-Cox) test was applied to compare survival data in figure 1B. GC, germinal center; DFS, disease-free survival; TLS, tertiary lymphoid structure.

    As is shown in online supplemental figure S1D,E, patients with larger TLS density and a higher GC percentage showed better prognosis, which were similar to the results of figure 1B. As is shown in online supplemental figure S1F,G, tumors classified as TLS+ showed higher TLS density, and GC+ TLS cases showed larger GC percentage compared with GC− TLS ones. These results indicated that both methods can provide reasonable evaluation of TLS and GC. Taken together, the above results suggested that the prognostic relevance of intratumoral TLS depends on GC formation, and the degree of TLS maturation has a major impact on NSCLC survival after surgical resection.

    TDLN metastasis influenced maturation of intratumoral TLS

    To explore the mechanism of intratumoral TLS maturation, we first stratified clinical features of TLS+ patients from HGMU cohort by GC status and the results revealed that LN metastasis correlates with decreased GC formation in primary tumor sites (online supplemental table S4). Moreover, as is shown in online supplemental figure S1H–I, while smaller percentage of GC formation was found in primary tumors with LN metastasis, the density of TLS revealed no difference. We then did a stratification analysis for the association between intratumoral TLS and NSCLC prognosis based on clinical features. As is shown in figure 2A, intratumoral TLS formation is associated with better survival in patients with adenocarcinoma (HR=0.71, p=0.029), and those with no TDLN metastasis (HR=0.58, p=0.010), but its predictive capability was lost in patients with LN invasion (HR=0.98, p=0.883).

    Figure 2
    Figure 2

    Influence of TDLN metastasis on maturation of intratumoral TLS. (A) DFS in patient subgroups. (B) Representative images of H&E staining of footpad xenograft of NSCLC from control and tumor-invaded TDLN group (n=5 for each group). Cox proportional hazard regression analysis was used for assessing the relationship between possible risk factors and DFS in figure 2A. Student’s t-test was used to compare the means between groups in figure 2B and results were presented as mean±SD, *p<0.05. Ctrl, control; GC, germinal center; DFS, disease-free survival; NSCLC, non-small cell lung cancer; TDLN, tumor-draining lymph node; TLS, tertiary lymphoid structure; TNM, tumor, node, metastases.

    Based on a pooled cohort of 97 NSCLC cases that received neoadjuvant immunotherapy from published studies which reported pathological status of TDLNs,19–21 patients without regional LN metastases also showed a higher rate of major pathological response in the primary lesions (78.46% vs 18.75%, OR=7.75, p=0.021; online supplemental figure S2). The compromised prognosis and immunotherapy response collectively indicated that LN metastases might negatively regulate antitumoral immunity in the primary lesions.

    Supplemental material

    In this sense, we further explored the possible impact of tumor invasion of TDLNs on TLS maturation. As is shown in online supplemental table S5, the distribution of clinical features between LN-negative and LN-positive patients was relatively uneven, we first did a PSM, after which factors affecting LN metastasis were comparable. Then we tested the effect of LN metastasis on intratumoral TLS, and results revealed that, while the relationship to overall TLS formation (OR=0.90, p=0.533) and GC− TLS (OR=1.25, p=0.247) were insignificant, it was associated with remarkably reduced GC formation in both univariate and multivariate analysis controlling age, gender, histology, and tumor size as covariates (OR=0.57, p=0.027; OR=0.56, p=0.021, respectively) (online supplemental table S6). Furthermore, we employed the murine footpad xenograft model of NSCLC, in which the afferent lymphatic vessels of popliteal LNs were ligated 1 week postxenograft implantation to prevent possible spontaneous LN metastasis. For the experimental group, tumor cells were also injected into the ipsilateral inguinal LNs which drained the lymphatic flow of the corresponding lower limb region, and H&E staining showed a decrease of GC formation in primary tumor sites of tumor-invaded TDLNs compared with the control group (figure 2B). Collectively, the above analyses suggested that TDLN metastasis impacts the prognostic relevance of intratumoral TLS, potentially through its effects on TLS maturation.

    Single-cell transcriptomic profiling of TDLNs of lung cancer

    To uncover the underlying mechanism for the connection between TDLN and intratumoral TLS, we did scRNA-seq of TDLNs from patients with NSCLC, which provided a detailed landscape of the immune-microenvironment. Droplet-based scRNA-seq (10X Genomics) was performed on a total of eight surgically resected TDLN samples from four patients (online supplemental tables S7 and S8). After quality control, approximately 0.2 billion unique transcripts were obtained from 40 845 cells. Among these cells, 16 978 cells (41.56%) were from tumor-free TDLNs, and 23 867 cells (58.44%) were from tumor-invaded TDLNs. With the correction for reading depth, cells were integrated into a comparable dataset adjusted for batch effects. Then we performed dimensionality reduction through principal components analysis, UMAP, and tSNE, and visualized the distribution of cells among positive and negative LNs, LN stages, and patients in figure 3A.

    Figure 3
    Figure 3

    Single-cell transcriptomic profiling of TDLNs. (A) UMAP of the 40 845 cells profiled, with each cell color coded for (left to right): its sample type of origin (tumor-free or tumor-invaded TDLNs), LN station (N1 or N2), the corresponding patient, the associated cell type. (B) Canonical cell markers to label clusters by cell identity as represented in the UMAP plot. (C) Dot plot of mean expression of canonical marker genes for the nine major cell lineages. (D) Percentages of the nine major cell types across tumor-free and tumor-invaded TDLNs. Error bars indicate the 95% CI for the calculated relative frequencies, and *p<0.05 using χ2 test of independence. NK, natural killer; TDLN, tumor-draining lymph node; UMAP, uniform manifold approximation and projection.

    We also identified nine major cell types including T lymphocytes, B lymphocytes, myeloid cells, NK cells, epithelial cells, plasma cells, ECs, fibroblasts, and mast cells, based on the expression of canonical gene markers (figure 3A–C).

    As is shown in figure 3D, while the relative abundance of endothelial, mast, NK, plasma and T cells in tumor-invaded TDLNs were comparable to that in tumor-free TDLNs, epithelial cells were significantly enriched. As there are few epithelial cells in LNs under normal conditions, increase of epithelial cells is consistent with the invasion of malignant cells into TDLNs. Moreover, in contrast to the significant increase of fibroblasts and myeloid cells in positive TDLNs, a marked decrease of B lymphocytes was also shown. Given that B cells can circulate from TDLNs to primary tumor lesions, they are likely to play a role in the maturation process of intratumoral TLS. Collectively, our results suggested a change in the multicellular microenvironment from tumor-free to tumor-invaded TDLNs, specifically, reduced number of B lymphocytes.

    B lymphocyte infiltration of primary tumor lesions is reduced in LN-positive patients

    To delve deeper into the possible underlying mechanism of the relationship between TDLN and intratumoral TLS maturation, we conducted IHC and MIF staining of cell markers associated with antitumor immunity, including CD4, CD8, CD20, CD68, CD163, CD66b, CD38, FoxP3, PD-L1, and CD133 (figure 4A and online supplemental figure S3A). As depicted in online supplemental table S9, figure 4B, and online supplemental figure S3B, in patients with LN metastasis, the presence of CD20+ cells and CD66b+ cells were significantly reduced, signifying a suppression of B lymphocyte and neutrophil infiltration at their corresponding primary tumor sites. While CD20+ cells correlated well with TLS maturation, CD66b+ cells did not (figure 4C, online supplemental figure S3C). Given that CD20+ cells play a crucial role in GC formation,36 tumor-invaded TDLN may thus hinder B lymphocytes in primary tumor sites and impact the maturation of intratumoral TLS.

    Supplemental material

    Figure 4
    Figure 4

    Immune cell infiltration of primary tumor lesions. (A) Representative images of IHC and MIF staining of CD20, CD4, CD66b, CD38, and FoxP3 in tissue slides of the primary tumor with tumor-free TDLNs and tumor-invaded TDLNs. (B–C) Percentage of CD20+ compared between LN-negative and LN-positive patients (B), and TLS statuses (C). (D) MCP-counter scores of the corresponding cell types in patients with NSCLC from TCGA database. (E) Scatter diagram of B cell signature and the corresponding TLS signature of patients with NSCLC from TCGA database. For figure 4B–D, error bars indicate the IQRs, and Wilcoxon and Kruskal-Wallis rank-sum tests were used to compare the distribution of cells. *P<0.05, **p<0.01, ***p<0.001, ns, not significant. For figure 4E, rank correlation method was used to examine the correlation between B cell and TLS signature, and r denotes the Spearman’s correlation coefficient. GC, germinal center; IHC, immunohistochemistry; MIF, multiplex immunofluorescence; NSCLC, non-small cell lung cancer; TCGA, The Cancer Genome Atlas; TDLN, tumor-draining lymph node; TLS, tertiary lymphoid structure.

    To shed further light on this effect, we gaged signatures of key cell types through the MCP-counter devolution method, including the B lymphocytes, CD8 T cells, cytotoxic lymphocytes, endothelial cells, fibroblasts, monocytes, myeloid DCs, neutrophils, NK cells, and T cells, through analyzing the PSM-adjusted NSCLC patient samples from the TCGA database. As is shown in figure 4D and a significant reduction of B cell lineage in primary tumor sites was indeed observed in those with tumor-invaded TDLNs. In agreement with previous reports,22 the B cell signature also correlates well with the TLS signature (r=0.603, p<0.001), which includes the gene expression of CCL2, CCL3, CCL4, CCL5, CCL8, CCL18, CCL19, CCL21, CXCL9, CXCL10, CXCL11, and CXCL13 (figure 4E).37 38 Taken together, these results highlighted a reduction of B lymphocyte infiltration in the corresponding primary lesion of patients with LN-positive lung cancer, and this may lead to the inhibition of GC formation and TLS maturation.

    Tumor invasion of TDLN inhibits memory B cell formation

    Then we zeroed in on the transcriptomic features of the B cell population in TDLNs. As is shown in figure 5A–B and subclusters of B lymphocytes (C1–C5) were identifiable. Cluster 1 B cells show a preponderant expression of markers associated with naïve B cells, clusters 2–4 B cells exhibited a predominant expression of markers linked to memory B cells, and cluster 5 B cells demonstrated a relatively high expression of GC markers. Pseudotime trajectory analysis of B cell clusters also substantiated the state distribution of these five B-cell clusters (figure 5C,D), categorizing cluster 1 cells as naïve B cells, clusters 2–4 cells as memory B cells, and cluster 5 cells as GC B cells.

    Figure 5
    Figure 5

    Change of B cell differentiation in tumor-invaded TDLNs. (A) Naïve, memory and GC B cell scores as represented in the tSNE plot. (B) Heatmap of functional gene sets in B cell clusters. (C–D) Unsupervised transcriptional trajectory of B lymphocytes from Monocle, colored by pseudotime and cell clusters. (E–F) Percentages of the memory and naïve B cells across tumor-free and tumor-invaded TDLNs (E), and TDLNs from different patients (F). (G) Representative images of IF staining for TDLNs from GC− patients and GC+ patients (n=5 for each group). For figure 5E,F, error bars indicate the 95% CI for the calculated relative frequencies, and *p<0.05 using χ2 test of independence. For figure 5G, error bars indicate the IQRs, and *p<0.05 using Wilcoxon rank-sum test. GC, germinal center; IF, immunofluorescence; TDLN, tumor-draining lymph node; TLS, tertiary lymphoid structure; tSNE, t-distributed stochastic neighbor embedding.

    Then we compared the estimated percentages of B cell clusters in TDLNs, and observed that while the presence of naïve B cells were augmented, the percentage of memory B cells was diminished in cases of tumor-invaded TDLNs (figure 5E). This indicates a major reduction of memory B cells in positive LNs, suggesting that B lymphocyte reduction in tumor-invaded TDLNs is mainly due to the inhibition of memory B cell formation. To further explore the relationship between this reduction of memory B cells in TDLNs and inhibited GC formation at primary tumor sites, we first compared the memory B cell abundance in TDLNs from the four corresponding patients. H&E staining showed that patient A and B are GC−, while patient C and D are GC+ (online supplemental figure S4). The subpopulation analyses of scRNA data revealed increased memory B cells and decreased naïve B cells in TDLNs of patients C and D (figure 5F). Furthermore, a comparison of IF staining of TDLNs from GC− and GC+ patients showed an increase in the percentage of CD19+CD27+ memory B cells in TDLNs from GC+ patients (figure 5G). These results are consistent with our findings in a small cohort of patients with NSCLC, where the presence of CD19+CD27+ memory B cells in primary tumors was well-correlated with their presence in TDLNs (online supplemental figure S5A,B). Indeed, infiltration of memory B cells was also found to be diminished in LN-positive primary tumors (online supplemental figure 5C). Taken together, these results indicated a reduction of memory B cells in tumor-invaded TDLNs of NSCLS, which is associated with decreased GC formation in the primary tumor lesions.

    Supplemental material

    Supplemental material

    Cell-to-cell interaction and gene expression enrichment analyses of B lymphocytes in TDLN

    To further dissect how the formation of memory B cells was inhibited when tumor cells invaded TDLNs, we employed the method of CellChat, which is based on the examination of ligand-receptor network, to analyze intercellular interactions. Incoming communication is defined as an interaction where a cell expresses a receptor, while outgoing communication is defined as an interaction where a cell expresses a ligand. We first compared the overall information flow for each signaling pathway by summing up the communication probabilities between all pairs of cell groups with B cells as the incoming end in the inferred network. Significant signaling pathways were ranked based on differences in the overall information flow, with the top signaling pathways enriched in the tumor-free TDLNs highlighted in red, and those enriched in the tumor-invaded TDLNs highlighted in green. As is shown in figure 6A, major intercellular ligand-receptor pairs with B lymphocytes as signal receivers in tumor-free TDLNs include CCL, ITGB2, IFN, CXCL, and pleiotrophin pathways, while signals such as the osteopontin (SPP1), Midkine, CD22, CD45, semaphorin-4, and GALECTIN pathways were among major pathways in tumor-invaded TDLNs.

    Figure 6
    Figure 6

    Cell-to-cell interaction and gene expression enrichment analyses. (A) Comparison of the overall information flow of each signaling pathway between tumor-free and tumor-invaded TDLNs with B cells as signal receiver, and significant signaling pathways were ranked based on differences in the overall information flow. (B) Dot plots showing comparison of the communication probabilities mediated by ligand-receptor pairs between tumor-free and tumor-invaded TDLNs with B cells as signal receiver. (C) Bar plots showing altered hallmark pathways scored by GSVA in B lymphocytes from tumor-invaded TDLNs. (D) Bar plots showing enrichment of genes differentially expressed by B lymphocytes from tumor-invaded TDLNs through Gene Ontology analysis of biological process. (E) Violin plots showing the gene expression distribution of IFN-γ among major cell types from TDLNs and between GC− patients and GC+ patients. (F) Western blot analysis of TDLNs from footpad xenograft of NSCLC between control group and NK cell-depleted group. (G) Representative images of IF staining for NCR1 and IFN-γ in human TDLNs. Ctrl, control; GC, germinal center; GSVA, gene set variation analysis; IF, immunofluorescence; IFN, interferon; NK, natural killer; NSCLC, non-small cell lung cancer; TDLN, tumor-draining lymph node.

    We then compared the communication probabilities mediated by each ligand-receptor pair from different cell groups to B cells. The increased signaling pathways are defined as those with higher communication probabilities in tumor-invaded TDLNs and the decreased signaling pathways as those with lower communication probabilities. As is shown in figure 6B, in positive TDLNs, inhibiting interactions, such as SPP1-CD44/(ITGA4+ITGB1), and LGALS9-CD45/CD44 were increased, while stimulation-provoking interactions, such as CCL19/CCL21-CCR7, CXCL12-CXCR4, IFNG-(IFNGR1+IFNGR2), and ITGB2-ICAM1 were decreased. Comparison of B cells in TDLNs also showed that expression of ICAM1, on which the ITGB2 signal relies for boosting lymphocyte activation, was decreased in B cells from positive LNs (online supplemental figure S6A). This trend was particularly pronounced in the memory lineage of B lymphocytes (online supplemental figure S6B). On the outcoming ends of these interactions, as is shown in online supplemental figure S6C, the expression of IFNG was decreased in NK cells, the expression of CCL19, CCL21, and CXCL12 was decreased in fibroblasts. Conversely, expression of LGALS9 was increased in endothelial, mast, and myeloid cells, and SPP1 was increased in myeloid and epithelial cells. Collectively, the cell-cell communicating analysis revealed an overall reduction of stimulation-provoking and induction of inhibitory signals on B lymphocytes in tumor-invaded TDLNs.

    Supplemental material

    To further illustrate the signaling pathway change, GSVA analysis comparing B cells in tumor-invaded and tumor-free TDLN was conducted, which revealed a reduction of genes regulated by MYC and genes in response to IFN-γ, indicting a decrease in B cell proliferation and differentiation to memory cells (figure 6C). GO enrichment analysis was also performed (figure 6D, online supplemental figure S6D), and it showed that the altered genes were primarily enriched in biological process including leukocyte differentiation, B cell activation, and response to IFN-γ. As the above analyses pointed to a central role of IFN-γ signaling, we then compared its expression from major cell types in TDLNs. The results also showed a reduction of IFN-γ expressed by NK cells, being the major source of IFN-γ production, from TDLNs of GC− patients (figure 6E). To further confirm the role of NK cells in the IFN-γ signaling activation, we employed the murine footpad xenograft model of lung cancer and injected NK1.1 antibody to deplete NK cells in C57/BL6J mice. After harvesting the ipsilateral inguinal LNs, western blot analysis showed decreased expression of IFN-γ in the NK-depleted cases (figure 6F). We also did IF staining of human TDLNs, and the results showed remarkable overlaps of IFN-γ expression and the NK cell marker NCR1 (figure 6G). Moreover, publicly accessible scRNA-seq data of primary NSCLC tumors with LN-negative disease was analyzed to explore the role of NK cells in primary lesions (online supplemental figure S7A).39 While NK cells are major sources of effector molecules including IFN-γ, granzyme A, granzyme B, and perforin (online supplemental figure S7B,C), signals for stimulating memory B cell differentiation was not seen from NK cells (online supplemental figure S7C). Indeed, no difference was found in either total NK cells (NCR1+) or IFN-γ-expressing NK cells (IFN-γ+NCR1+) between GC− and GC+ primary tumors with LN-negative disease, as determined by IF staining (online supplemental figure S7D). These thus together suggested an important role of IFN-γ from NK cells in regulating memory B cell differentiation in TDLNs for the maturation of TLS in primary tumor lesions.

    Supplemental material

    Figure 7
    Figure 7

    Regulation of memory B cells by IFN-γ signaling in TDLNs. (A) Gating strategy of single-cell suspensions from TDLNs of footpad xenograft model of NSCLC. (B) Flow cytometry analysis of single-cell suspensions of TDLNs and percentages of memory and naïve B cells in TDLNs in the corresponding groups (n=5 for each group). (C) Representative images of H&E staining of orthotopic xenograft of NSCLC from the corresponding groups (n=5 for each group). (D) Schematic illustration of the cross-talk between TDLNs and primary tumor lesions through B lymphocytes. Student’s t-test was used to compare the means between groups and results were presented as mean±SD, *p<0.05, **p<0.01, ***p<0.001. Ctrl, control; IFN, interferon; NK, natural killer; NSCLC, non-small cell lung cancer; TDLN, tumor-draining lymph node; TLS, tertiary lymphoid structure.

    IFN-γ signaling regulates memory B cell differentiation in TDLNs

    To provide further insight into the impact of IFN-γ signaling on memory B cell differentiation in TDLNs and GC formation in primary tumor lesions, we then first used the footpad xenograft model which has been described above in the method section of murine model of lung cancer. The experimental groups were treated with anti-IFN-γ neutralizing antibody, or injection of tumor cells into the ipsilateral inguinal LNs with and without recombinant murine IFN-γ. The gating method for flow cytometry is shown in figure 7A, and analysis of the flow cytometry data revealed decreased memory B cells and increased naïve B cells in the anti-IFN-γ group and the intranodal tumor injection group, which can be partly rescued by IFN-γ treatment (figure 7B). Additionally, using the murine orthotopic xenograft model of NSCLC, we evaluated the effect of IFN-γ treatment on TLS maturation, and the results showed that IFN-γ can significantly increase GC formation in primary tumor lesions, and this effect can be partially reversed by FTY720 treatment, which is an effective blocker of immune cell outflow from LNs (figure 7C). Taken together, the above results indicated that IFN-γ plays a crucial role in regulating memory B cell differentiation in TDLNs, which can then impact TLS maturation in the primary tumor lesions of NSCLC.

    Discussion

    In this research, through the morphological examination of tissue slides from surgically resected NSCLC tumors, we showed that maturation of intratumoral TLS is associated with better prognosis after surgery. Further analyses based on clinical features revealed that invasion of TDLN influenced the predicting relevance of intratumoral TLS, which is due to reduced GC formation. scRNA-seq data revealed a reduction of memory B cell formation in tumor-invaded TDLNs, which may result from an increase of inhibitory signals together with a decrease of stimulatory ones. IHC and MIF further revealed a reduction of B lymphocyte infiltration in the corresponding primary tumor sites as a result of tumor invasion of TDLNs, which was further confirmed through analysis of the TCGA database. Murine xenograft models of NSCLC provided direct evidence that tumor invasion of TDLNs can inhibit GC formation in primary tumor lesions, and this may be due to a decrease of IFN-γ signaling from NK cells and subsequent reduced memory B cell differentiation in TDLNs. Together, our research revealed that tumor-invaded TDLNs in lung cancer can inhibit the maturation of intratumoral TLSs, which may be a result of reduced memory B cells and IFN-γ signaling from NK cells (figure 7D).

    The tumor microenvironment is composed of both tumor cells and non-tumor cells of various lineages. With the successes of immunotherapy in NSCLC,40 41 the role of the immune microenvironment in tumor development and progression has been extensively investigated. TLS is considered to be a hub for both cellular and humoral immune responses against tumor cells and is associated with favorable clinical outcomes in various types of solid tumors.13 42 43 In this regard, we hypothesized that intratumoral TLS can play a role in effective antitumor immune responses, and showed that it indeed correlates well with better survival after surgical resection. Briefly, intratumoral TLS can be classified as immature ones, manifesting as aggregates of vague and ill-defined clusters of lymphocytes, and mature ones, within which GC formation can be found.44 As the degree of TLS maturation may be a better prognostic factor,45 we compared the prognostic relevance of GC− TLS and GC+ TLS. Our results, consistent with previous studies,15 45 46 revealed that GC+ TLS is associated with better survival after surgery, while GC− TLS is not. Our research emphasized the existence of GC response and maturation of TLS in effective antitumor immunity within primary tumor lesions.

    While the induction of antitumor immune response in TLS has been documented, little is known about its relationship with the draining LNs. Considering the association between TLS and better efficacy of immunotherapy,14 we further showed that predicting capability of TLS is diminished in LN-positive patients. Despite unchanged rates of intratumoral TLS, GC formation, as a hallmark of TLS maturation, was significantly reduced. In this regard, we showed that draining LNs play a crucial role in TLS maturation process. The standard practice of radical surgery of solid tumors involves the systematic resection of TDLNs to prevent local and distant recurrence, however, this has been questioned in recent large-scale clinical trials.47 48 Additionally, as a platform to facilitate antigen presentation and intercellular communication, TDLNs play important roles in shaping immune cell differentiation and antitumoral immunity. Studies on ICBs as neoadjuvant therapies showed that response rate is reduced in patients with positive TDLNs19–21 (online supplemental figure S2). Although current studies of TME mainly focused on the analysis of primary tumor, our results highlight a crucial role of TDLNs in the maturation process of intratumoral TLS, and the need for further investigation into the communication between TDLNs and primary tumors.

    Recent evidence pointed towards a central role of B cells in the formation and maturation of TLS.23 36 Morphologically, TLSs show similar features to TDLNs, forming a T cell zone with mature DCs, a GC zone with follicular DCs and proliferating B cells, and high endothelial venules.49 Although the detailed underlying mechanism is still not clear, the presence of B lymphocytes in TLS is associated with protective immunity and a better response to immunotherapy.22 50 Indeed, our results showed a remarkable decrease of B cells in primary tumor sites of LN-positive patients, which is also validated through the TCGA NSCLC cohort, suggesting the involvement of TDLNs in the activation of TIL-B cells. Activated B lymphocytes in TME can result from de novo differentiation of naïve B cells, or a rapid expansion of memory B cells that have encountered antigens before, such as those derived from GC response in SLOs. It is thus possible that B cells can bridge the communication between immune response in TDLNs and intratumoral TLSs.25

    With the advent of scRNA-seq analysis, we can now better decipher the intricacies of tumor microenvironment. Through scRNA-seq analysis of TDLNs, we found a dramatic decrease of B lymphocytes, primary attributed to the suppression of memory B cells. This observation is consistent with the findings of mass cytometry studies, which showed an enrichment of memory B cells in primary tumors of responders to ICBs.22 It is thus possible that formation of memory B cells accounts for the role that TDLNs play in the maturation process of intratumoral TLS, and that in tumor-invaded TDLNs, differentiation of naïve B cells into memory B cells is suppressed. In structurally intact TDLNs, antigen-presenting cells (APCs) loaded with tumor-antigens from primary tumor sites can trigger potent antitumor immunity. The memory B cells formed may then travel back to the primary tumor, leading to a rapid expansion and differentiation into effector B cells.51 Indeed, percentage of memory B cell in TDLNs correlated well with its infiltration into the primary tumors (online supplemental figure S5A,B). While molecular and cellular similarities between GCs in LNs and TLS in primary tumor lesions suggest possible communication between them, the exact mechanisms for this remains unclear. Based on our analyses, the TDLNs may act as a source of memory B cells with strong antitumor potential, thereby promoting the formation of mature intratumoral TLS.

    Memory B cell is an important player in antitumor immunity. After being loaded with tumor antigens in the primary tumor sites, APCs, such as macrophages and DCs, migrate into TDLNs, where they may activate B cells and induce the formation of GC. While complex differentiation and selection processes give rise to the generation of plasma cells, some B cells escape the GC reaction and are stored as memory B cells. They can survive for long term and rapidly differentiate into effector B cells on antigen re-encounter.52 When they re-enter GCs during the recall response, further somatic hypermutation can occur leading to stronger protective immunity.51 Thus, the induction of memory B lymphocytes is likely to efficiently augment local antitumor immunity.

    While a complex intercellular communication network may exist, we identify a general decrease of stimulatory and increase of inhibitory signals in tumor-invaded TDLNs. Among these interactions, the IFN-γ signal is noteworthy, as both cell-cell interaction analysis and gene enrichment analysis pointed to a significant decrease of its production from NK cells and a suppression of the downstream pathways in B cells. Indeed, IFN-γ, acting on IFN-γR, can upregulate Th1 cell-defining transcription factor (T-bet) of B-cell in synergy with the BCR, toll-like receptor, and CD40 signals,53 which is required for the generation of IgG1+ or IgG3+ memory B cells.54 However, in tumor-invaded TDLNs, the production of IFN-γ from NK cells is significantly reduced. In this sense, the interconnection between NK and B cells through the IFNG-IFNGR pathway should explain the decreased formation of memory B cells in TDLNs. Furthermore, murine experimental models provided direct evidence that IFN-γ signaling is involved in TLS maturation through regulating memory B cell differentiation in TDLNs. These together thus pointed to a central role of IFN-γ from NK cells in the relationship between TDLN and TLS maturation, although roles of other signaling pathways on memory B cell formation may also warrant investigation.

    With the success of immunotherapy, dissection of the intricate interactions in TME is a burgeoning area of lung cancer research. However, most studies focused on primary tumor sites. Our results called for more attention to the role of TDLNs as an essential part of TME, where antitumor immune response can occur and interaction with primary tumor may ensue. The strength of this study is that multiple aspects of data including patient follow-up, scRNA-seq and murine NSCLC models, together with experimental methods of IHC, multiplex IF, and flow cytometry were integrated to shed light on the currently less-noticed interaction between TDLN and primary tumor sites. Limitations of this study may include a lack of the dissection into the phenotypic change of NK cells in TDLNs and differential process that memory B cells undergo during TLS maturation in primary tumors. Thus, future studies are needed to provide more detailed evidence for addressing these unsolved problems and delve deeper into the logic of this communication between TDLN and primary tumor.

    In conclusion, through the above analysis, we show that maturation of intratumoral TLS is associated with a lower risk of relapse in surgically resected patients with NSCLC, and this can be influenced by tumor invasion of TDLNs. B lymphocytes, specifically the memory B cell population, are reduced in tumor-invaded TDLNs, leading to decreased B cell infiltration in primary tumor sites, and this is likely to be the result of changes of regulatory signals, specifically the IFN-γ response of B cells, in the signaling network of TDLNs. The study highlights the impact of TDLN on intratumoral TLS maturation and provides basis for further research into the role of B lymphocytes in the communication between TDLNs and immune microenvironment of the primary tumor.

    Data availability statement

    Data are available on reasonable request. The raw sequence data reported in this paper has been deposited into CNGB Sequence Archive (CNSA) of China National GeneBank DataBase (CNGBdb) with accession number CNP0002869 and is publicly accessible at: https://db.cngb.org/cnsa/. Detailed information will be available from the corresponding authors on reasonable request.

    Ethics statements

    Patient consent for publication

    Ethics approval

    This patient information and tissue samples were collected according to the ethical guidelines set by the Internal Review Board of the First Affiliated Hospital of Guangzhou Medical University, #2020105. This study was conducted following the Declaration of Helsinki. Animal procedures were approved by the Animal Ethics Committee of Guangzhou Medical University under protocol #2021118. Written informed consent was obtained from all patients for permitting researchers to analyze clinical features and tissue samples.

    Acknowledgments

    We thank Juanjuan Long for kindly helping with the graphics.

    References

    1. ↵
      1. Siegel RL,
      2. Miller KD,
      3. Jemal A
      . Cancer statistics, 2019. CA Cancer J Clin 2019;69:7–34. doi:10.3322/caac.21551
      OpenUrlCrossRefPubMed
    2. ↵
      1. Asamura H,
      2. Chansky K,
      3. Crowley J, et al
      . The International association for the study of lung cancer lung cancer staging project: proposals for the revision of the N descriptors in the forthcoming 8th edition of the TNM classification for lung cancer. J Thorac Oncol 2015;10:1675–84. doi:10.1097/JTO.0000000000000678
      OpenUrlCrossRef
    3. ↵
      1. Hotta K,
      2. Matsuo K,
      3. Ueoka H, et al
      . Role of adjuvant chemotherapy in patients with resected non-small-cell lung cancer: reappraisal with a meta-analysis of randomized controlled trials. J Clin Oncol 2004;22:3860–7. doi:10.1200/JCO.2004.01.153
      OpenUrlAbstract/FREE Full Text
    4. ↵
      1. Yue D,
      2. Xu S,
      3. Wang Q, et al
      . Erlotinib versus vinorelbine plus cisplatin as adjuvant therapy in chinese patients with stage IIIA EGFR mutation-positive non-small-cell lung cancer (evan): a randomised, open-label, phase 2 trial. Lancet Respir Med 2018;6:863–73. doi:10.1016/S2213-2600(18)30277-7
      OpenUrl
    5. ↵
      1. Mitsudomi T,
      2. Soh J
      . Adjuvant therapy of operable nonsmall cell lung cancer: an update. Curr Opin Oncol 2021;33:47–54. doi:10.1097/CCO.0000000000000702
      OpenUrl
    6. ↵
      1. Somasundaram A,
      2. Burns TF
      . The next generation of immunotherapy: keeping lung cancer in check. J Hematol Oncol 2017;10:87. doi:10.1186/s13045-017-0456-5
    7. ↵
      1. Fridman WH,
      2. Zitvogel L,
      3. Sautès-Fridman C, et al
      . The immune contexture in cancer prognosis and treatment. Nat Rev Clin Oncol 2017;14:717–34. doi:10.1038/nrclinonc.2017.101
      OpenUrlCrossRefPubMed
    8. ↵
      1. Trüb M,
      2. Zippelius A
      . Tertiary lymphoid structures as a predictive biomarker of response to cancer immunotherapies. Front Immunol 2021;12:674565. doi:10.3389/fimmu.2021.674565
    9. ↵
      1. Sautès-Fridman C,
      2. Petitprez F,
      3. Calderaro J, et al
      . Tertiary lymphoid structures in the era of cancer immunotherapy. Nat Rev Cancer 2019;19:307–25. doi:10.1038/s41568-019-0144-6
      OpenUrlCrossRefPubMed
    10. ↵
      1. Sautès-Fridman C,
      2. Verneau J,
      3. Sun C-M, et al
      . Tertiary lymphoid structures and B cells: clinical impact and therapeutic modulation in cancer. Semin Immunol 2020;48:101406. doi:10.1016/j.smim.2020.101406
    11. ↵
      1. Goc J,
      2. Germain C,
      3. Vo-Bourgais TKD, et al
      . Dendritic cells in tumor-associated tertiary lymphoid structures signal a Th1 cytotoxic immune contexture and license the positive prognostic value of infiltrating CD8+ T cells. Cancer Res 2014;74:705–15. doi:10.1158/0008-5472.CAN-13-1342
      OpenUrlAbstract/FREE Full Text
    12. ↵
      1. Dieu-Nosjean M-C,
      2. Antoine M,
      3. Danel C, et al
      . Long-Term survival for patients with non-small-cell lung cancer with intratumoral lymphoid structures. J Clin Oncol 2008;26:4410–7. doi:10.1200/JCO.2007.15.0284
      OpenUrlAbstract/FREE Full Text
    13. ↵
      1. Munoz-Erazo L,
      2. Rhodes JL,
      3. Marion VC, et al
      . Tertiary lymphoid structures in cancer-considerations for patient prognosis. Cell Mol Immunol 2020;17:570–5. doi:10.1038/s41423-020-0457-0
      OpenUrlCrossRefPubMed
    14. ↵
      1. Cabrita R,
      2. Lauss M,
      3. Sanna A, et al
      . Tertiary lymphoid structures improve immunotherapy and survival in melanoma. Nature 2020;577:561–5. doi:10.1038/s41586-019-1914-8
      OpenUrlPubMed
    15. ↵
      1. Siliņa K,
      2. Soltermann A,
      3. Attar FM, et al
      . Germinal centers determine the prognostic relevance of tertiary lymphoid structures and are impaired by corticosteroids in lung squamous cell carcinoma. Cancer Res 2018;78:1308–20. doi:10.1158/0008-5472.CAN-17-1987
      OpenUrlAbstract/FREE Full Text
    16. ↵
      1. Calderaro J,
      2. Petitprez F,
      3. Becht E, et al
      . Intra-tumoral tertiary lymphoid structures are associated with a low risk of early recurrence of hepatocellular carcinoma. J Hepatol 2019;70:58–65. doi:10.1016/j.jhep.2018.09.003
      OpenUrlPubMed
    17. ↵
      1. Tang J,
      2. Ramis-Cabrer D,
      3. Curull V, et al
      . B cells and tertiary lymphoid structures influence survival in lung cancer patients with resectable tumors. Cancers (Basel) 2020;12:2644. doi:10.3390/cancers12092644
    18. ↵
      1. Núñez NG,
      2. Tosello Boari J,
      3. Ramos RN, et al
      . Tumor invasion in draining lymph nodes is associated with Treg accumulation in breast cancer patients. Nat Commun 2020;11:3272. doi:10.1038/s41467-020-17046-2
    19. ↵
      1. Provencio M,
      2. Nadal E,
      3. Insa A, et al
      . Neoadjuvant chemotherapy and nivolumab in resectable non-small-cell lung cancer (NADIM): an open-label, multicentre, single-arm, phase 2 trial. Lancet Oncol 2020;21:1413–22. doi:10.1016/S1470-2045(20)30453-8
      OpenUrlCrossRefPubMed
    20. ↵
      1. Forde PM,
      2. Chaft JE,
      3. Smith KN, et al
      . Neoadjuvant PD-1 blockade in resectable lung cancer. N Engl J Med 2018;378:1976–86. doi:10.1056/NEJMoa1716078
      OpenUrlCrossRefPubMed
    21. ↵
      1. Gao S,
      2. Li N,
      3. Gao S, et al
      . Neoadjuvant PD-1 inhibitor (sintilimab) in NSCLC. J Thorac Oncol 2020;15:816–26. doi:10.1016/j.jtho.2020.01.017
      OpenUrlCrossRefPubMed
    22. ↵
      1. Helmink BA,
      2. Reddy SM,
      3. Gao J, et al
      . B cells and tertiary lymphoid structures promote immunotherapy response. Nature 2020;577:549–55. doi:10.1038/s41586-019-1922-8
      OpenUrlCrossRefPubMed
    23. ↵
      1. Ruffin AT,
      2. Cillo AR,
      3. Tabib T, et al
      . B cell signatures and tertiary lymphoid structures contribute to outcome in head and neck squamous cell carcinoma. Nat Commun 2021;12:3349. doi:10.1038/s41467-021-23355-x
    24. ↵
      1. Nakagawa R,
      2. Toboso-Navasa A,
      3. Schips M, et al
      . Permissive selection followed by affinity-based proliferation of GC light zone B cells dictates cell fate and ensures clonal breadth. Proc Natl Acad Sci U S A 2021;118:e2016425118. doi:10.1073/pnas.2016425118
    25. ↵
      1. Alberts E,
      2. Wall I,
      3. Calado DP, et al
      . Immune crosstalk between lymph nodes and breast carcinomas, with a focus on B cells. Front Mol Biosci 2021;8:673051. doi:10.3389/fmolb.2021.673051
    26. ↵
      1. Hao Y,
      2. Hao S,
      3. Andersen-Nissen E, et al
      . Integrated analysis of multimodal single-cell data. Cell 2021;184:3573–87. doi:10.1016/j.cell.2021.04.048
      OpenUrlCrossRefPubMed
    27. ↵
      1. Qiu X,
      2. Mao Q,
      3. Tang Y, et al
      . Reversed graph embedding resolves complex single-cell trajectories. Nat Methods 2017;14:979–82. doi:10.1038/nmeth.4402
      OpenUrlCrossRefPubMed
    28. ↵
      1. Jin S,
      2. Guerrero-Juarez CF,
      3. Zhang L, et al
      . Inference and analysis of cell-cell communication using cellchat. Nat Commun 2021;12:1088. doi:10.1038/s41467-021-21246-9
    29. ↵
      1. Becht E,
      2. Giraldo NA,
      3. Lacroix L, et al
      . Estimating the population abundance of tissue-infiltrating immune and stromal cell populations using gene expression. Genome Biol 2016;17:218. doi:10.1186/s13059-016-1070-5
    30. ↵
      1. Subramanian A,
      2. Tamayo P,
      3. Mootha VK, et al
      . Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles. Proc Natl Acad Sci U S A 2005;102:15545–50. doi:10.1073/pnas.0506580102
      OpenUrlAbstract/FREE Full Text
    31. ↵
      1. Hänzelmann S,
      2. Castelo R,
      3. Guinney J
      . GSVA: gene set variation analysis for microarray and RNA-Seq data. BMC Bioinformatics 2013;14:7. doi:10.1186/1471-2105-14-7
    32. ↵
      1. Ritchie ME,
      2. Phipson B,
      3. Wu D, et al
      . Limma powers differential expression analyses for RNA-sequencing and microarray studies. Nucleic Acids Res 2015;43:e47. doi:10.1093/nar/gkv007
    33. ↵
      1. Wu T,
      2. Hu E,
      3. Xu S, et al
      . ClusterProfiler 4.0: a universal enrichment tool for interpreting omics data. Innovation (Camb) 2021;2:100141. doi:10.1016/j.xinn.2021.100141
    34. ↵
      1. He M,
      2. Yu W,
      3. Chang C, et al
      . Estrogen receptor α promotes lung cancer cell invasion via increase of and cross-talk with infiltrated macrophages through the CCL2/CCR2/MMP9 and Cxcl12/Cxcr4 signaling pathways. Mol Oncol 2020;14:1779–99. doi:10.1002/1878-0261.12701
      OpenUrl
    35. ↵
      1. Lin Y,
      2. Xue J,
      3. Liao S
      . Blocking lymph flow by suturing afferent lymphatic vessels in mice. J Vis Exp 2020;2020:159. doi:10.3791/61178
      OpenUrl
    36. ↵
      1. Patil NS,
      2. Nabet BY,
      3. Müller S, et al
      . Intratumoral plasma cells predict outcomes to PD-L1 blockade in non-small cell lung cancer. Cancer Cell 2022;40:289–300. doi:10.1016/j.ccell.2022.02.002
      OpenUrlCrossRefPubMed
    37. ↵
      1. Coppola D,
      2. Nebozhyn M,
      3. Khalil F, et al
      . Unique ectopic lymph node-like structures present in human primary colorectal carcinoma are identified by immune gene array profiling. Am J Pathol 2011;179:37–45. doi:10.1016/j.ajpath.2011.03.007
      OpenUrlCrossRefPubMedWeb of Science
    38. ↵
      1. Prabhakaran S,
      2. Rizk VT,
      3. Ma Z, et al
      . Evaluation of invasive breast cancer samples using a 12-chemokine gene expression score: correlation with clinical outcomes. Breast Cancer Res 2017;19:71. doi:10.1186/s13058-017-0864-z
    39. ↵
      1. Kim N,
      2. Kim HK,
      3. Lee K, et al
      . Single-Cell RNA sequencing demonstrates the molecular and cellular reprogramming of metastatic lung adenocarcinoma. Nat Commun 2020;11. doi:10.1038/s41467-020-16164-1
    40. ↵
      1. Shu CA,
      2. Gainor JF,
      3. Awad MM, et al
      . Neoadjuvant atezolizumab and chemotherapy in patients with resectable non-small-cell lung cancer: an open-label, multicentre, single-arm, phase 2 trial. Lancet Oncol 2020;21:786–95. doi:10.1016/S1470-2045(20)30140-6
      OpenUrlCrossRefPubMed
    41. ↵
      1. Borghaei H,
      2. Paz-Ares L,
      3. Horn L, et al
      . Nivolumab versus docetaxel in advanced nonsquamous non-small-cell lung cancer. N Engl J Med 2015;373:1627–39. doi:10.1056/NEJMoa1507643
      OpenUrlCrossRefPubMed
    42. ↵
      1. Di Caro G,
      2. Bergomas F,
      3. Grizzi F, et al
      . Occurrence of tertiary lymphoid tissue is associated with T-cell infiltration and predicts better prognosis in early-stage colorectal cancers. Clin Cancer Res 2014;20:2147–58. doi:10.1158/1078-0432.CCR-13-2590
      OpenUrlAbstract/FREE Full Text
    43. ↵
      1. Kroeger DR,
      2. Milne K,
      3. Nelson BH
      . Tumor-Infiltrating plasma cells are associated with tertiary lymphoid structures, cytolytic T-cell responses, and superior prognosis in ovarian cancer. Clin Cancer Res 2016;22:3005–15. doi:10.1158/1078-0432.CCR-15-2762
      OpenUrlAbstract/FREE Full Text
    44. ↵
      1. Denton AE,
      2. Innocentin S,
      3. Carr EJ, et al
      . Type I interferon induces CXCL13 to support ectopic germinal center formation. J Exp Med 2019;216:621–37. doi:10.1084/jem.20181216
      OpenUrlAbstract/FREE Full Text
    45. ↵
      1. Posch F,
      2. Silina K,
      3. Leibl S, et al
      . Maturation of tertiary lymphoid structures and recurrence of stage II and III colorectal cancer. Oncoimmunology 2018;7:e1378844. doi:10.1080/2162402X.2017.1378844
    46. ↵
      1. J Gunderson A,
      2. Rajamanickam V,
      3. Bui C, et al
      . Germinal center reactions in tertiary lymphoid structures associate with neoantigen burden, humoral immunity and long-term survivorship in pancreatic cancer. Oncoimmunology 2021;10:1900635. doi:10.1080/2162402X.2021.1900635
    47. ↵
      1. Faries MB,
      2. Thompson JF,
      3. Cochran AJ, et al
      . Completion dissection or observation for sentinel-node metastasis in melanoma. N Engl J Med 2017;376:2211–22. doi:10.1056/NEJMoa1613210
      OpenUrlCrossRefPubMed
    48. ↵
      1. Giuliano AE,
      2. Ballman KV,
      3. McCall L, et al
      . Effect of axillary dissection vs no axillary dissection on 10-year overall survival among women with invasive breast cancer and sentinel node metastasis: the ACOSOG Z0011 (Alliance) randomized clinical trial. JAMA 2017;318:918–26. doi:10.1001/jama.2017.11470
      OpenUrlCrossRefPubMed
    49. ↵
      1. Dieu-Nosjean M-C,
      2. Goc J,
      3. Giraldo NA, et al
      . Tertiary lymphoid structures in cancer and beyond. Trends Immunol 2014;35:571–80. doi:10.1016/j.it.2014.09.006
      OpenUrlCrossRefPubMed
    50. ↵
      1. Germain C,
      2. Gnjatic S,
      3. Tamzalit F, et al
      . Presence of B cells in tertiary lymphoid structures is associated with a protective immunity in patients with lung cancer. Am J Respir Crit Care Med 2014;189:832–44. doi:10.1164/rccm.201309-1611OC
      OpenUrlCrossRefPubMed
    51. ↵
      1. Laidlaw BJ,
      2. Cyster JG
      . Transcriptional regulation of memory B cell differentiation. Nat Rev Immunol 2021;21:209–20. doi:10.1038/s41577-020-00446-2
      OpenUrlCrossRef
    52. ↵
      1. Ripperger TJ,
      2. Bhattacharya D
      . Transcriptional and metabolic control of memory B cells and plasma cells. Annu Rev Immunol 2021;39:345–68. doi:10.1146/annurev-immunol-093019-125603
      OpenUrl
    53. ↵
      1. Knox JJ,
      2. Myles A,
      3. Cancro MP
      . T-bet(+) memory B cells: generation, function, and fate. Immunol Rev 2019;288:149–60.
      OpenUrlCrossRef
    54. ↵
      1. Wang NS,
      2. McHeyzer-Williams LJ,
      3. Okitsu SL, et al
      . Divergent transcriptional programming of class-specific B cell memory by T-bet and RORα. Nat Immunol 2012;13:604–11. doi:10.1038/ni.2294
      OpenUrlCrossRefPubMed

    Supplementary materials

    Footnotes

    • MH, QH, XC and JL contributed equally.

    • Contributors MH, QH, XC and JLiu designed the research study, performed the experiments, analyzed the data, and wrote the manuscript. FL, RZ and YL performed the experiments. CL, JLi and HD discussed the data. HP and XW reviewed the data and the manuscript. ZC and SL reviewed the manuscript. WL and JH conceived the concept, oversaw the interpretation and presentation of the data, and wrote the manuscript. WL is responsible for the overall content as the guarantor. All authors read and approved the final manuscript.

    • Funding China National Science Foundation (Grant No. 82022048, 81871893,82203090), Key Project of Guangzhou Scientific Research Project (Grant No. 201804020030),Natural Science Foundation of Guangdong Province (Grant No.2023A1515012837), High-level University Construction Project of Guangzhou Medical University (Grant No. 20182737, 201721007, 201715907, 2017160107); National Key R&D Program (Grant No. 2017YFC0907903 and 2017YFC0112704); and the Guangdong High-level Hospital Construction ‘Reaching Peak’ Plan.

    • Competing interests No, there are no competing interests.

    • Provenance and peer review Not commissioned; externally peer reviewed.

    • Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.