Data from 47 DLBCL tumors in The Cancer Genome Atlas (TCGA-DLBCL) and 444 normal control (NC) tissues were downloaded from the University of California Santa Cruz (UCSC) database (https://genome.ucsc.edu). The 4th edition of the World Health Organization (WHO) Classification of Tumors of Hematopoietic and Lymphoid Tissues in 2018 was taken as the basis (23). Paraffin-embedded tissue specimens were collected from 137 patients with pre-chemotherapy DLBCL and 20 patients with reactive lymphoid hyperplasia in the First Affiliated Hospital of Xinjiang Medical University from 2011 to 2020 ( Supplementary Table 1 ). Patients with pre-chemotherapy DLBCL diagnosed pathologically (n = 100) and 30 control samples from healthy peripheral blood samples at the Tumor Hospital affiliated with Xinjiang Medical University, collected from 2021 to 2022, were included ( Supplementary Table 2 ).

The Linear fitting methods, Bayesian analysis and T-test algorithms to identify differentially expressed genes (DEGs) between DLBCL and NC by “limma” package (24) in R software. A total of 2136 differential genes were identified, of which 1313 were up-regulated and 823 down-regulated according to |log2 foldchange|>1 and P < 0.05.

Gene Ontology (GO), including biological processes, cellular components (CCs), and molecular functions (MFs), and Kyoto Encyclopedia of Genes and Genomes (KEGG) analyses were applied to explore the function of DEGs using the “clusterProfiler” package (25) in R software.

Levels of LAG3, PDCD1, and other related factors in TCGA and GTEx database were analyzed for differences and the T-test used to assess the significance of differential expression between cancer and normal tissues.

CIBERSORT (26), an immune tumor biological computational tool, was used to calculate immune cell levels and evaluate the infiltration status of each sample.

Tissue microarrays were immersed in citrate buffer and left to cool to room temperature, then treated with 3% hydrogen peroxide for 10 min, followed by addition of a drop goat serum after washing with PBS, and incubating at 37°C for 20 min. Next, samples were incubated with anti-PD-1 (Working fluid, Catalog Number: UMAB199, ZSGB-BIO) or anti-LAG-3 (1:2000, Catalog Number: 16616-1-AP, Proteintech) and chromogenic immunoassays performed with DAB after washing in PBS, followed by hematoxylin re-staining. Human tonsils served as positive control specimens for both antibodies. Negative control specimens were DLBCL tissue treated with phosphate buffer in place of antibody. Final scores were obtained by averaging frequency scores from two replicate tests conducted for each DLBCL case.

PD-1 positivity was localized to the cell membrane or cytoplasm of TILs, with median = 4 (range, 0–54) PD-1⁺ TILs per high power field (HPF); ≤ 4; PD-1⁺ TILs/HPF was considered low expression and > 4 PD-1⁺ TILs/HPF as high expression. LAG-3 positivity was also localized to the cell membrane or cytoplasm of TILs (median = 11, range = 0–103 LAG-3⁺ TILs/HPF); ≤ 11 LAG-3⁺ TILs/HPF was considered low expression, with > 11 considered high expression.

The DLBCL cells (OCI-LY3 and SU-DHL-6) were cultured in RPMI-1640 (Thermo Fisher Scientific, USA) supplemented with 10% fetal bovine serum and 1% penicillin and streptomycin at 37°C and 5% CO₂. Anti-LAG-3 monoclonal antibody (0.5 nm/ml, MedChemExpress, USA) and/or anti-PD-1 monoclonal antibody (0.5 nm/ml, MedChemExpress, USA) were added into OCI-LY3 and SU-DHL-6 cells, respectively. Peripheral blood CD8⁺ T cells from patients with DLBCL were sorted using immunomagnetic beads (Becton, Dickinson and Company, USA) and co-cultured with OCI-LY3 or SU-DHL-6 cells.

The CD8⁺ T cells obtained by MACS were placed in RPMI-1640 complete medium containing 10% FBS and culture overnight at 37°C, 5% CO₂. At the same time, inoculate logarithmic growth SU-DHL6 and OCI-LY3 cells into a 24-well plate, with 6×10⁴ cells per well, and culture overnight. The next day, the already adhered SU-DHL6 and OCI-LY3 cells were incubated separately with anti-PD-1 or/and anti-LAG-3 monoclonal antibodies and isotype controls at room temperature for 2 h. After that, 4×10⁵ CD8⁺ T cells were added to each well and incubated at 37°C for 12 h. Finally, flow cytometry analysis was performed.

Freshly collected sodium heparin (100 μl) was added to each of two flow tubes. The two sets of fluorescently labeled antibodies were added to each of the two tubes: 1) CD4 and CD8 cells were labelled with 5 μl PD-1-PECy7, LAG3-PE, CD4-FITC, and CD8-PCP5.5; 2) CD4-FITC and CD8-PCP5.5 were used as negative controls. Cells were incubated for 15 min at room temperature in the dark, then 2 mL 1× erythrocyte lysate buffer was added per 100 μl of whole blood, and red blood cells lysed at room temperature for 8–12 min, until the cell suspension was transparent. Cells were then washed and centrifuged (3,000 g, 15 min), supernatants discarded, and 300 μl PBS added. CD4⁺ T cells were gated as T helper lymphocytes and CD8⁺ T cells as cytotoxic T lymphocytes. For detection of tumor cell apoptosis, after co-culture, CD8⁺ T cells were sorted and 5 μl Annexin V/PE (Becton, Dickinson and Company, USA) and 10 μl 7-AAD (Becton, Dickinson and Company, USA) added to the remaining cancer cells, mixed gently, and incubated. After incubation cells were analyzed by flow cytometry.

Data analysis and visualization were both conducted using Graphpad Prism 8.0 and Xiantao Academic database, which as a public information platform used for analyzing a large amount of sample data and visualizing the results (https://www.xiantaozi.com/). Pearson correlation analysis was applied to analyze the relationships of LAG-3 and PD-1 levels with immune cells. The Student’s T test was used for comparisons between two groups. P < 0.05 was considered a statistically significant difference.

Firstly, we screened 2136 DEGs, including 1313 up-regulated genes and 823 down-regulated genes, from 48 DLBCL-TCGA and 444 normal tissues from GTEx ( Figures 1A, B ). Most DEGs mainly involved in the “regulation of signaling receptor activity”, “system development”, “PI3K-Akt signaling pathway” and “Wnt signaling pathway” ( Figures 1C, D ). The GO enrichment analysis showed that in DLBCL patients, changes in gene expression related to system development, organ development, tissue development, epithelial cell differentiation, cellular keratinization, extracellular processes, and cellular adhesion occurred, thereby affecting the originally normal physiological environment in DLBCL patients ( Figure 1C ). In addition, KEGG pathway analysis shows significant changes in genes involved in the PI3K-Akt, cAMP, MAPK, Wnt, JAK-STAT, TLR, IL-17, TGF-β, and chemokine-related signaling pathways in DLBCL patients, suggesting significant alterations in multiple signaling transduction pathways in DLBCL patient cells. It is worth noting that there have been significant changes in the genes involved in the cytokines-cytokine receptor interaction, as well as in cell cytotoxicity, in DLBCL patients, which suggests that immune function is disrupted in DLBCL patients ( Figure 1D ). The mRNA level of both LAG-3 and PD-1 was noticeably elevated in DLBCL ( Figure 2A ) and levels of LAG-3 and PD-1 were positively correlated with one another ( Figure 2B ).

PD-1 and LAG-3 were widely expressed and co-expressed on DLBCL TILs (PD-1⁺ TILs, n = 68, 49.6%; LAG-3⁺ TILs, n = 65, 47.4%; and PD-1⁺LAG-3⁺ TILs, n = 45, 32.8%) ( Figure 2C ; Supplementary Tables 3 , 4 ). The relationships among PD-1 and LAG-3 expression and clinical features of DLBCL patients were also analyzed. PD-1 expression was significantly associated with Eastern Cooperative Oncology Group (ECOG) score and Ann-Arbor stage, while LAG-3 levels were significantly associated with primary tumor site and lactate dehydrogenase (LDH) level. Further, co-expression of PD-1 and LAG-3 was significantly associated with Ann-Arbor stage and PD-1 levels were positively correlated with those of LAG-3 on TILs in DLBCL ( Figure 2D ).

With the median expression levels of the PDCD1 and LAG-3 genes obtained from the database as the dividing line, the data of each group were divided into high expression group and low expression group in order to conduct subsequent analysis. The data of co-expression of PD-1 and LAG-3 was selected from the samples that overlap in expression between PD-1 and LAG-3 samples. There was no significant difference in overall survival (OS) between patients with DLBCL classified into two groups (high and low) according to PD-1 expression ( Figure 3A ), whereas OS was lower in patients with DLBCL in the high LAG-3 expression group than in those with low LAG-3 expression ( Figure 3B ). Furthermore, patients with elevated levels of both LAG-3 and PD-1 were characterized by shorter survival time and inferior prognosis ( Figure 3C ). Survival time of patients in National Comprehensive Cancer Network-International Prognostic Index (NCCN-IPI) group ≤ 3 was longer, and patients had better prognosis (P = 0.001) ( Figure 3D ), while the survival time of patients in the nodal, ECOG ≥ 2, and older age groups was shorter, indicating worse prognosis ( Figures 3E–G ). Non-germinal central B cell-like (GCB) phenotype, abnormal LDH or β2-microglobulin (β2-MG) levels, sex, B symptoms (fever, weight loss, and night sweats), Ann-Arbor stage, and site of invasion were not significantly associated with DLBCL disease prognosis ( Supplementary Figures 1A–G ). Multivariate Cox regression indicated that LAG-3 levels and age were independent risk factors associated with DLBCL ( Figure 3H ).

To better understand the differential immunological function of DLBCL and normal tissue, we used CIBERSORT and ssGSEA. The analysis revealed that DLBCL tissues had markedly elevated immune cell fractions ( Figure 4A , Supplementary Figure 2 ). Additionally, LAG-3 and PD-1 expression levels were closely associated with most TILs ( Figure 4B ). LAG-3 levels were strongly positively correlated with T cells, Th1 cells, dendritic cells (DCs), CD8 T cells, eosinophils, T helper cells, neutrophils, Th17 cells, and Th2 cells, and strongly negatively correlated with B cells ( Figure 4C ). No correlations were detected between PD-1 levels and Th17 cells, Tregs, DCs, B cells, macrophages, neutrophils, or natural killer cells ( Figure 4D ).

Levels of CD8⁺ T cells and CD4⁺CD8⁺ T cells in the peripheral blood of patients with DLBCL were higher than those in the healthy control (HC) group, while CD4⁺ T cell and CD4⁺/CD8⁺ T cell levels were lower than those in the HC group ( Supplementary Figures 3A–D ). LAG-3 expression on CD4⁺ T cells was higher in the peripheral blood of patients with DLBCL than those in HC samples ( Figure 5A ). There was no significant difference in PD-1⁺CD4⁺ T cells between the two groups ( Figure 5B ). In patients with DLBCL, levels of LAG-3⁺CD8⁺ T cells and PD-1⁺CD8⁺ T cells were significantly higher than those in HC ( Figures 5C, D ).

Additionally, patients with DLBCL had significantly lower levels of peripheral blood LAG-3^-PD-1^-CD8⁺ T cells ( Figure 6A ), while those of LAG-3⁺PD-1⁺CD8⁺ T cells were significantly higher in DLBCL than in HC ( Figure 6D ). There was no significant difference between the two groups in LAG-3⁺PD-1^-CD8⁺ T or LAG-3^-PD-1⁺CD8⁺ T cells ( Figures 6B, C ).

CD8⁺ T cells from patients with DLBCL were purified (purity > 90%) using magnetic beads ( Supplementary Figure 4 ). Then, we examined the expression of LAG-3 and PD-1 on CD8⁺ T cells after co-culture with SU-DHL6 or OCI-LY3 cells and treatment with anti-LAG-3 and (or) anti-PD-1 blocking agents. After administration of anti-LAG-3, LAG-3 expression was significantly decreased without affecting PD-1 expression. Similarly, anti-PD-1 treatment decreased PD-1 expression without affecting LAG-3 expression ( Figures 7A–D ).

Levels of perforin were higher in CD8⁺ T cells from patients with DLBCL cultured with SU-DHL6 or OCI-LY3 cells after treatment with anti-LAG-3, anti-PD-1, or a combination of the two, relative to the PBS control group, and levels were highest after combination therapy ( Figures 8A, C ). Similarly, levels of granzyme B were higher in the groups treated with blocking agents than in the untreated PBS group ( Figures 8B, D ).

CD8⁺ T cells from patients with DLBCL were co-cultured with SU-DHL6 or OCI-LY3 cells and treated with LAG-3 and/or PD-1 blocking agents, followed by examination of SU-DHL6 or OCI-LY3 cell apoptosis. Administration of anti-LAG-3 or anti-PD-1 inhibitors, or a combination of both blocking agents, all induced a significant increase in overall SU-DHL6 and OCI-LY3 cell apoptosis rates ( Figures 9A, B ). These results indicate that increased LAG-3 and PD-1 on the surface of CD8⁺ T cells in patients with DLBCL result in impaired CD8⁺ T cells functionality, leading to their inability to kill tumor cells.