The study design is shown as a flowchart for the analysis of the BTN3A2 gene (Figure 1A). To detect BTN3A2 expression in cancer and non-malignant tissues, the BTN3A2 mRNA contents in different cancer type tissues were analyzed using the Sangerbox and TIMER data resources. The results illustrated that the BTN3A2 expression was lower in LUAD, lung squamous cell carcinoma, as well as kidney renal clear cell carcinoma than that in the non-malignant tissues (Figure 1B). Moreover, data from TCGA demonstrated that the BTN3A2 content was lower in the LUAD group than that in the healthy group (Figures 1C,D). Interestingly, the protein expression of BTN3A2 in LUAD was still much lower than in that in non-malignant tissues from CPTAC samples (UALCAN) (Figure 1E). To verify the histological level of BTN3A2, we used the Human Protein Atlas database. The results showed that BTN3A2 is upregulated in normal tissue and downregulated in LUAD tissues (Figure 7F). The result was found in immunofluorescence staining (Supplementary Figure S1). Otherwise, BTN3A2 is also highly expressed in cells with high CD3 expression.

Further analysis of multiple clinic features of 515 LUAD samples in TCGA exhibited the low BTN3A2 mRNA content. The BTN3A2 mRNA content was lower in LUAD patients than that in healthy individuals in subgroup analysis on the basis of gender, race, nodal metastasis, smoking, stages, as well as tumor grade (Figure 2 and Table 1). Therefore, BTN3A2 levels may have a promising LUAD diagnostic value.

To explore whether BTN3A2 expression was related to the prognosis of LUAD patients, we evaluated the influence of BTN3A2 contents on OS and FP through the Kaplan–Meier plotter data resource. The data demonstrated that the high BTN3A2 contents were linked to the good prognosis in LUAD from 204820_s_at probe (OS HR = 0.71, p = 8.5e-08; FP HR = 0.71, p = 5.5e-04) (Figures 3A,B). The results were consistent with those in 209846_s_at probe and 212613_s_at probe from the Kaplan–Meier plotter (Figures 3C–F). To further verify the prognostic value of BTN3A2 in LUAD, the PrognoScan database was used and the results showed that the low BTN3A2 expression group was linked to good lung cancer prognosis (OS jacob-00182-CANDF HR = −0.72, p = 0.00027; RFS GSE31210 HR = −0.28, p = 0.033) (Figures 3G,H). Thus, our data suggested that the low BTN3A2 content is a protective factor and results in a good prognosis in LUAD.

To identify the different signal pathways activated in LUAD, we used GSEA between high and low BTN3A2 expression datasets. A total of six hallmark gene-sets (KEGG_T_CELL_RECEPTOR_SIGNALING_CASCADE, KEGG_B_CELL_RECEPTOR_SIGNALING_CASCADE, KEGG_NATURAL_KILLER_CELL_MEDIATED_CYTOTOXICITY, GO_IMMUNE_RECEPTOR_ACTIVITY, GO_IMMUNOLOGICAL_SYNAPSE, and GO_T_CELL_ACTIVATION) were chosen for analysis (Figures 4A–F). The results showed that T-cell receptor signaling cascade, B-cell receptor signaling cascade, natural killer cell–mediated cytotoxicity, immune receptor activity, immunological synapse, and T-cell activation are differentially enriched in the BTN3A2 high expression phenotype of LUAD. Moreover, the bar graph shows the top 10 messages of biological processes, cell components, molecular functions, and KEGG, respectively. The GO term annotation indicated that these genes primarily participated in cell proliferation, multicellular organismal process, cell communication, protein binding, and structural molecule activity (Supplementary Figure S2A). The KEGG pathway analysis indicated that these genes chiefly participated in peroxisome and cell adhesion molecules (Supplementary Figure S2B).

To assess the potential relationship of immune invasion with BTN3A2 expression in LUAD, we used TIMER to conduct the following analysis. First, the “Gene” module analysis revealed that BTN3A2 expression is remarkably positively correlated with invading levels of tumor purity, B cells, dendritic cells, CD8+ T cells, macrophages, neutrophils, and CD4+ T cells in LUAD (Figure 5A). Then, the “SCNA” module analysis revealed immune cell infiltration may relevant to altered BTN3A2 gene copy numbers, including B cells, dendritic cells, CD4+ T cells, macrophages, CD8+ T cells, and neutrophils in LUAD (Figure 5B). In addition, the “SURVIVAL” module analysis demonstrated that high B-cell and dendritic cell levels were linked to a good prognosis of LUAD (Figure 5C, p < 0.05, respectively).

To study the relationship between immune cell markers and BTN3A2 expression, we analyzed the markers of B cells, macrophages, T cells, dendritic cells, and neutrophils by the TIMER data resource. Interestingly, we discovered that the expression levels of CD19 along with CD79A in B cells, CD3E, CD2, CD8A, and CD8B in T cells, CEACAM8, ITGAM, and CCR7 in neutrophils, T helper cells (Th1 and Th2), macrophages, and dendritic cell have markedly positively related to the BTN3A2 expression in LUAD (Table 2).

The data imply that high BTN3A2 expression in LUAD may influence the prognosis because of immune invasion.

To further elucidate the miRNA targets of BTN3A2 in LUAD, we used miRarbase, starbase, and miRDB databases to predict miRNA-target genes of BTN3A2 shown in the Venn plot (Figure 6A). The result showed that it had six miRNA targets (hsa-miR-17-5p, hsa-miR-20a-5p, hsa-miR-93-5p, hsa-miR-106b-5p, hsa-miR-20b-5p, and hsa-miR-519d-3p) of BTN3A2 and which was visualized in Figure 6B. Then, we analyzed the prognostic potential of 6 miRNA-target in lung cancer using the Kaplan–Meier Plotter database. Interestingly, we found that only hsa-miR-17 had differential expression (p = 1.62E-12) and prognostic value (OS, HR = 1.4, p = 0.035) (Figures 6C,D). Moreover, the target site in the BTN3A2 3'UTR was predicted to pair with hsa-miR-17-5p by miRDB and RNA22Sites (Figure 6E).

We have confirmed that BTN3A2 expression was related to the immune infiltration in LUAD, and the height expression of BTN3A2 was also correlated with the better prognosis of LUAD. Therefore, we hypothesized that hsa-miR-17 expression may be the miRNA targets of BTN3A2. The prognosis analysis on the basis of the expression of hsa-miR-17 of LUAD in linked immune cell subtype was performed using the Kaplan–Meier plotter, and data illustrated that the high hsa-miR-17 content of LUAD in abundant B cells (HR = 0.03), abundant CD4+ memory T cells (HR = 0.042), abundant macrophages (HR = 0.04), abundant regulatory T cells (HR = 0.031), and type 2 T helper cells (HR = 0.0077) had a poor prognosis (Figures 7A–E). Thus, the results may provide a potential mechanism that hsa-miR-17-5p may be miRNA targets of BTN3A2.

We analyzed the protein expression along with mRNA expression levels of SKA3 in the UALCAN data resource (http://ualcan.path.uab.edu/index.html) (Chandrashekar et al., 2017).

BTN3A2 expression in cancers was determined by the TIMER data resource (https://cistrome.shinyapps.io/timer/) (Li et al., 2017). We identified the expression of BTN3A2 in different tumors by the TIMER data resource. After that, the association of BTN3A2 with immune invasion in LUAD was estimated by the TIMER algorithm. Furthermore, the relationship of ACE2 with the type biomarkers of tumor purity, B cells, CD8+ T cells, CD4+ T cells, macrophages, neutrophils, and dendritic cells in LUAD was verified.

We used the Human Protein Atlas (https://www.proteinatlas.org/) to detect BTN3A2 protein expression in immunohistochemistry.

We employed the Kaplan–Meier plotter data resource (http://kmplot.com/) (Györffy et al., 2010) to explore the value of BTN3A2 expression in LUAD prognosis. This threshold was logrank p value < 0.05 in the Kaplan–Meier plotter data resource.

We analyzed the prognosis in LUAD using the PrognoScan database (http://dna00.bio.kyutech.ac.jp/PrognoScan/index.html) (Mizuno et al., 2009). Cox p-value < 0.05 was considered as the significant difference.

GSEA is the most commonly used statistical method (Subramanian et al., 2005). GSEA was performed to clarify the molecular mechanisms of the prognostic gene signature. GSEA was performed using GSEA v. 3.0 and was searched to determine the enriched biological processes, cellular components, molecular functions, and KEGG pathway associated with survival of the high-risk group. FDR < 0.05 and |NES| > 1 were considered statistically significant. Then, we used the STRING website to analyze co-expression genes of BTN3A2. Also, we analyzed GO and KEGG pathway using WebGestalt.

To further explore the targets of BTN3A2 in LUAD, we used miRarbase, starbase, and miRDB databases to predict miRNA-target genes of BTN3A2. Moreover, the target site in the BTN3A2 3'UTR was predicted to pair with hsa-miR-17-5p by miRDB (http://www.mirdb.org/) (Chen and Wang, 2020) and RNA22Sites (https://cm.jefferson.edu/rna22v2.0/) (Miranda et al., 2006).

The human primary LUAD tissue and corresponding normal tissue were collected from LUAD patients in the Suzhou Science and Technology Town Hospital, Gusu School, Nanjing Medical University. The embedded LUAD and normal clinical sample tissue were sliced and routinely deparafnized to water as described before. The tissue sections were blocked with goat serum after antigen retrieval. These were incubated overnight with anti-BTN3A2 and anti-CD3. The tissue was treated with rhodamine-labeled anti-mouse IgG and fluorescein-labeled anti-rabbit IgG (Qi et al., 2021). DAPI was used for nuclear staining. The images were captured using a confocal microscope (DM6000 CFS; Leica) and processed using LAS AF software.

The results of the survival analysis were acquired from a log-rank test. We used Spearman’s correlation to evaluate BTN3A2 with immune invasion and type biomarkers of immune cells. Comparing two independent samples using Student’s t test. p < 0.05 signified statistical significance.