BTG2 Serves as a Potential Prognostic Marker and Correlates with Immune Infiltration in Lung Adenocarcinoma.

Background: B-cell translocation gene 2 (BTG2) has been revealed to be involved in the occurrence and development of multiple cancers. However, the role of BTG2 in lung adenocarcinoma (LUAD) is still ambiguous. Thus, this study aims to investigate the prognostic value of BTG2 and its correlation with immune infiltration in LUAD.
Methods: The expression of BTG2 in LUAD was analyzed using the TIMER and UALCAN databases. The correlations between BTG2 expression and clinicopathological factors were investigated using the UALCAN databases. The Kaplan-Meier plotter, GEPIA, and TCGA databases were employed to assess the prognostic value of BTG2. The STRING database and Cytoscape software were used to construct an interaction network and mine co-expression genes. The TISIDB database was examined for a correlation between BTG2 and driver genes in LUAD. Enrichment analysis of co-expressed genes and BTG2 was performed using the LinkedOmics database. Finally, the correlations between BTG2 and immune infiltrates were investigated using the TIMER, GEO, and TISIDB database.
Results: BTG2 was significantly downregulated in LUAD. The decreased expression of BTG2 in LUAD was significantly correlated with higher cancer stages and shorter duration of overall survival. The expressions of BTG2-related co-expression genes were associated with the prognosis in LUAD. The expression of BTG2 was closely associated with the mutations of TP53 and ROS1. Enrichment analysis revealed that BTG2 was significantly correlated with immune-associated signaling pathways and function. In addition, the expression of BTG2 was found to be closely related to immune infiltration, multiple gene markers of immune cells, chemokines, and chemokine receptors.
Conclusion: Our findings have effectively demonstrated that BTG2 expression was downregulated in LUAD, indicating poor prognosis. Closely relating to immune cell infiltration, BTG2 may be a promising immune-related biomarker and molecular target for patients with LUAD.

Introduction

Lung cancer is one of the most common malignant tumors, ranking first in terms of cancer-related mortality worldwide.1–3 Non–small cell lung cancer (NSCLC) accounts for approximately 85% of lung cancer.4 Lung adenocarcinoma (LUAD) is the major histological subtype of NSCLC.4,5 Owing to the high proportion of patients diagnosed with advanced and metastatic disease, the 5-year survival rate of LUAD remains extremely low.6 The treatment of LUAD remains a major challenge for clinicians. With the introduction of immune checkpoint inhibitors into clinical practice, the treatment outcome of LUAD has been remarkably improved. PD-L1 and tumor mutation burden (TMB) are indicated as predictive biomarkers for immunotherapy. Numerous patients with high PD-L1 expressions or high TMB show limited benefit from immunotherapy, while negative PD-L1 expression or low TMB demonstrated response to PD-1/PD-L1 inhibitors. Cumulative evidence has indicated that immune infiltration plays a vital role in tumorigenesis, and affects the efficacy of immunotherapy, and prognosis of patients with cancers including LUAD.7,8 Therefore, it is crucial to clarify the potential regulating mechanisms of immune cell infiltration and further identify sensitive prognostic immune-related biomarkers and molecular targets to identify more effective therapeutic strategies for patients with LUAD.

B-cell translocation gene 2 (BTG2), a member of the anti-proliferative BTG/TOB gene family,9 maps within a chromosomal segment (1q32)10 and contains two regions named box A and box B.11 BTG2 is closely involved in a variety of biological processes, such as cell cycle,12 cell senescence,13 cell differentiation,14 hematopoiesis,15 oxidative damage,16 gene transcription,17 and DNA damage repair.18 Noteworthily, accumulating evidence has revealed that down-regulated BTG2 expression plays a vital role in tumor occurrence and development.19,20 Several studies reported that the downregulation of BTG2 was associated with the poor prognosis of cancers including NSCLC, breast cancer, hepatocellular carcinoma and bladder cancer.21–24 Zhang and colleagues reported that circular RNA circCRIM1 was able to promote BTG2 expression via down-regulating miR-125b-5p expression, thereby suppressing the migration, invasion, epithelial-mesenchymal transformation (EMT), and glycolysis in lung adenocarcinoma cell.25 These studies suggested that BTG2 was a potential prognostic marker for cancer. Recent experimental data demonstrated that BTG2 was closely involved in the regulation of immune response, including the regulation of B cell development, and T cell activation, and the induction of apoptosis in monocyte, which played key roles in anti-tumor immunity.26–30 However, the prognostic value of BTG2 and its correlation with immune infiltrates in LUAD remain unclarified.

Given the crucial role of BTG2 in tumorigenesis and progression, and the potential effect of BTG2 on immune cell infiltration, the aim of this study is to investigate the prognostic value of BTG2 and its correlation with immune infiltration in LUAD.

Materials and Methods

TIMER Database

TIMER () is an online analysis tool for immune infiltrating cells across diverse cancer types.31 The BTG2 expression for pan-cancer was analyzed using the TIMER database, and we also performed a detailed evaluation on the correlations between BTG2 expression and immune cell infiltrations as well as multiple immune markers. Further investigation was conducted to analyze the relationships between BTG2 expression and key driver genes (TP53 (p53), ROS1, ALK, KRAS, EGFR, BRAF, ERBB2 (HER-2), NRAS).

UALCAN Database

UALCAN () is a comprehensive web resource for analyzing cancer OMICS data.32 We compared the BTG2 expression between LUAD and normal samples using this database. Besides, we assessed the correlations between BTG2 expression and clinicopathological parameters in this database.

Kaplan–Meier Plotter Database

The Kaplan–Meier plotter () is capable of assessing the effect of 54k genes (mRNA, miRNA, protein) on survival in 21 cancer types.33 In Kaplan–Meier plotter database, the survplot function of the survival Bioconductor package in R software was used for univariate and multivariate Cox regression analysis.34 We used Kaplan–Meier Plotter database to determine the prognostic role of BTG2 in overall survival (OS). Based on the upper quartile of BTG2 expression, LUAD patients were divided into high and low groups. The prognostic value of BTG2-co-expressed genes for LUAD was also analyzed using this database.

GEPIA Database

GEPIA () is an open database for analyzing RNA sequencing expression of tumors and normal samples based on TCGA and GTEx.35 Survival analysis was performed using the python package lifelines.35 We chose quartiles as cut-off point for BTG2 grouping and obtained the correlations between OS and BTG2 expression using the GEPIA database.

TCGA Database

The Cancer Genome Atlas (TCGA) contained the clinical characteristic information available in public databases.36 UCSC Xenawebsite () is a handy site to help quickly download TCGA database, which was used to perform univariate and multivariate Cox regression analysis using “survminer” and “survival” packages in R (V4.1.1) software. Consistently, we chose the upper quarter as the cut-off value for the BTG2 grouping.

STRING Database

STRING () provides integrated and consolidated protein–protein interaction data.37 We constructed a protein–protein interaction (PPI) network information, and co-expressed genes were identified using the Cytoscape 3.8.2 version software.38

TISIDB Database

TISIDB () integrates multiple types of data resources in oncoimmunology, including 988 genes related to anti-tumor immunity and the association between any gene and immune features for 30 TCGA cancer types.39 The correlations between BTG2 expression and chemokines as well as chemokine receptors were comprehensively investigated through this platform.

Single-Cell RNA Sequencing Datasets Acquisition and Processing

The LUAD single cell validation dataset was downloaded from the Gene Expression Omnibus Database (GEO) database () with the accession number GSE189357. The LUAD dataset was processed through a set of process steps including data filtering, quality control, principal component analysis (PCA), reducing the dimension, and t-distributed stochastic neighbor embedding (tSNE) clustering. The clustering results were annotated using SingleR of R software and previous literature reports.

LinkedOmics Database

LinkedOmics () contains multiple sets of data from 32 types of cancer within the TCGA.40 The site has three modules: LinkFinder, LinkInterpreter and LinkCompare. The LinkInterpreter module was used to perform GO analysis CC (cellular component), BP (biological process), and MF (molecular function), as well as analysis of KEGG pathway, panther pathway, reactome pathway, and WikiPathway.

Results

The Expression Level of BTG2 in LUAD Tissues

We compared the BTG2 expression between malignant cancer and normal tissues using the TIMER database. Compared to normal tissue, the BTG2 expression significantly decreased in many malignancies, including LUAD (Figure 1A). These results were further confirmed based on the analysis using UALCAN database, with decreased BTG2 expression in LUAD (Figure 1B).

Figure 1

The expression levels of BTG2 in LUAD tissues. (A) The TIMER database showed that BTG2 expression levels were reduced in most of the different tumor types, especially in LUAD, compared with corresponding normal tissues. (B) BTG2 expression in LUAD was lower than normal tissues in the UALCAN database (***p < 0.001).

Relationship Between BTG2 Expression and Clinicopathological Factors in LUAD

Next, we evaluated the correlations between BTG2 expression and the clinicopathological factors (cancer stages, status of lymph node metastasis (LNM), age, gender, smoking habits, race). Results from the UALCAN database showed that the BTG2 expression was significantly associated with cancer stages and LNM in LUAD (Figure 2A and B), while the relationship with age, gender, patient’s smoking habits, and race was of no significance (Figure 2C–F).

Figure 2

The expressions of BTG2 in LUAD based on clinicopathological parameters in UALCAN database. (A–F) The correlation between BTG2 expression and cancer stages, nodal metastasis status, age, gender, smoking habits, and race (*p < 0.05, **p <0.01, ***p < 0.001).

Prognostic Analysis of BTG2 in LUAD

Subsequently, the Kaplan–Meier plotter database was used to assess the prognostic significance of BTG2 expression in LUAD patients. Kaplan–Meier curves showed that the down-regulated BTG2 expression in LUAD was significantly associated with short OS duration [HR: 0.44, 95% CI: 0.31–0.61, logrank P=7.6e-07] (Figure 3A). In addition, given the correlation between BTG2 expression and some clinical features, we also further discussed the effect of the interaction between BTG2 expression and clinical characteristics of LUAD patients on prognosis. Low BTG2 expression was associated with shorter OS duration in either females or males, in ever-smokers, and in patients with stage I, but the opposite was true in patients receiving chemotherapy (Table 1; P < 0.05). The results of Log rank test in GEPIA database also support the negative effect of low BTG2 expression on OS [HR: 0.4, logrank P=2.2e-05] (Figure 3B). To further verify the prognostic value of BTG2 in LUAD, TCGA database with 468 LUAD patients was used to perform univariate and multivariate Cox regression analysis. Our results confirmed that BTG2 was still an independent prognostic factor of OS after adjusting for potential confounders (Table 2). Baseline characteristics of TCGA LUAD patients were presented in .

Figure 3

Down-regulated BTG2 in LUAD was associated with poor prognosis of overall survival in Kaplan–Meier plotter database (A) and GEPIA database (B).

Table 1

The Kaplan–Meier Plotter Analysis of Predictive Effect of BTG2 Expression on Prognosis in LUAD Based on Clinicopathological Factors

Clinicopathological Factors	OS (n=719)
	N	HR (95%CI)	p
Gender
Female	317	0.53(0.31-0.9)	1.7e-02
Male	344	0.38(0.24-0.62)	5.7e-05
Smoking history
Ever	246	0.29(0.14-0.61)	4.6e-04
Never	143	0.61(0.21-1.8)	0.37
Stage
1	370	0.27 (0.13-0.53)	4.6e-05
2	136	0.73(0.42-1.29)	0.28
3	24	1.74 (0.58-5.16)	0.31
4	4	–	–
T stage
T1	123	0.66(0.31-1.39)	0.27
T2	105	0.66(0.33-1.32)	0.24
T3	4	–	–
T4	0	–	–
N stage
N0	184	0.61 (0.33-1.15)	0.12
N1	44	0.79 (0.33-1.88)	0.59
N2	3	–	–
M stage
M0	231	0.67 (0.41-1.1)	0.11
M1	1	–	–
Chemotherapy
No	21	1.3 (0.27-6.33)	0.74
Yes	36	10.16(1.94-53.37)	7.9e-04

Note: *: p<0.05, ***: p<0.001, ****: p<0.0001. Consistent P-value notations are used throughout the paper.

Abbreviations: OS, overall survival; HR, hazard ratio; CI, confidence interval; p, p values.

Table 2

Univariate and Multivariate COX Regression Analysis of OS for BGT2 from TCGA Cohort (n = 468)

	Characteristics	Univariate Analysis		Multivariate Analysis
		HR (95%CI)	P	HR (95%CI)	P
Gender	FEMALE VS. MALE	0.91 (0.67-1.2)	0.53	1 (0.69-1.5)	0.91
Age	≥ 65 VS. < 65	1.2 (0.9-1.7)	0.19	1.2 (0.83-1.8)	0.32
Smoking_history	YES VS. NO	0.97 (0.64-1.5)	0.9	1.1 (0.64-1.9)	0.75
T stage	T3–T4 VS. T1–T2	2.1 (1.3-3.2)	0.001	2.3 (1.3-4)	0.0027
N stage	N2–N3 VS. N0–N1	2.2 (1.5-3.2)	<0.001	2.1 (0.91-4.8)	0.082
M stage	M1 VS. M0	2.1 (1.3-3.6)	0.0049	2 (0.83-4.7)	0.12
Pathologic_stage	III–IV VS. I–II	2.4 (1.7-3.3)	<0.001	1 (0.44-2.4)	0.96
Targeted_molecular_therapy	YES VS. NO	1.1 (0.83-1.6)	0.42	0.63 (0.4-0.99)	0.046
History_of_neoadjuvant_treatment	YES VS. NO	14 (5.1-40)	<0.001	9.8 (3.2-31)	<0.001
Radiation_therapy	YES VS. NO	2.4 (1.7-3.5)	<0.001	2.4 (1.5-4)	<0.001
BTG2 expression	High VS. Low	0.63 (0.45-0.87)	0.005	0.66 (0.45-0.98)	0.041

Abbreviations: HR, hazard ratio; CI, confidence interval; p, p values.

Protein–Protein Interaction (PPI) and BTG2-Co-Expressed Genes Analysis

In order to elucidate the potential interactions between proteins, the protein–protein interaction (PPI) network was conducted based on the STRING database, and the top 10 BTG2-correlated genes were screened by the MCC algorithm in the cytoHubba plug-in of Cytoscape software. The interactions are plotted in Figure 4A, and the top 10 BTG2-correlated genes were TP53 (p53), CCND1, MDM2, HSP90AA1 (HSP90N), CDK2, CDK4, ATM, SIRT1, CDK6, and BCL2L1 (Figure 4B). Further analysis from the Kaplan–Meier plotter database using Log rank test showed that all but CCND1 [HR : 0.92, 95%=0.73–1.15, logrank p=0.45], HSP90AA1 [HR: 1.02, 95%=0.81–1.29, logrank p=0.85], and CDK6 [HR : 0.94, 95%=0.74–1.2, logrank p=0.62] were closely associated with the duration of OS in LUAD. (Figure 5A–J).

Figure 4

Protein–protein interaction network (PPI). (A) The PPI network of BTG2 was constructed in the STRING database. (B) The top ten genes with high degrees were obtained by the MCC algorithm in the cytoHubba plug-in of Cytoscape software.

Figure 5

Relationship between co-expression genes and prognosis. (A–J) Overall survival of the top ten co-expression genes was analyzed in the Kaplan–Meier plotter database.

Relationship Between BTG2 Expression and Key Driver Gene Mutations in LUAD

Key driver gene mutations were closely related to the treatment outcome and prognosis of LUAD.41–44 We therefore used the TISIDB database to analyze the correlations between BTG2 expression and key driver genes in LUAD, including TP53 (p53), ROS1, ALK, KRAS, EGFR, BRAF, ERBB2 (HER-2), NRAS (Figure 6A–H). Our results showed that the down-regulated BTG2 expression was significantly associated with TP53 (p<0.001) and ROS1 (p<0.05) mutations (Figure 6A and B).

Figure 6

The association between BTG2 expression and driver genes mutation status in TISIDB database. (A) TP53 status. (B) ROS1 status. (C) ALK status. (D) KRAS status. (E) EGFR status. (F) BRAF status. (G) ERBB2 status. (H) NRAS status.

Enrichment Analysis of BTG2 in LUAD

To explore the potential functions and signaling pathways of target genes, GO analysis of BP (biological process), MF (molecular function), and CC (cellular component), KEGG pathway (Kyoto Encyclopedia of Genes and Genomes), Reactome pathway, Panther pathway, and WikiPathway were assessed by using the LinkedOmics database. The biological processes were mainly enriched in transmembrane transport and cell differentiation (Figure 7A). For molecular function, the BTG2 was enriched primarily in cytokine binding and protein activity (Figure 7B). The cell component enrichment analysis indicated that the BTG2 was correlated with membranes and endosome lumen (Figure 7C). It is worth noting that the analysis of KEGG pathway, Reactome pathway, Panther pathway, WikiPathway, BP, and MF revealed that BTG2 was significantly correlated with immune responses and immune signaling pathways, including the production of molecular mediator involved in inflammatory response, interleukin-17 production, leukocyte differentiation, cytokine binding, Th17 cell differentiation, cytokine–cytokine receptor interaction, B cell activation, TGF-β signaling pathway, and B cell receptor signaling pathway (Figures 7A–D and 8A–C).

Figure 7

GO and KEGG enrichment analysis of BTG2 co-expressed genes in LinkedOmics database. (A) GO biological process. (B) GO molecular function. (C) GO cellular components. (D) KEGG pathway. Dark blue and orange indicate FDR ≤ 0.05, light blue and orange indicate FDR > 0.05.

Figure 8

Pathway enrichment analysis of BTG2 co-expressed genes in LinkedOmics database. (A) Reactome pathway. (B) Panther pathway. (C) WikiPathway. Dark blue and orange indicate FDR ≤0.05, light blue and orange indicate FDR >0.05.

Relationship of BTG2 Expression With the Immune Infiltration Levels, Markers of Different Subsets of Immune Cells, and Chemokines

Since BTG2 was significantly correlated with immune responses and immune signaling pathways, we further analyzed the correlation between BTG2 expression and the immune infiltration levels using the TIMER database. Our results showed that BTG2 expression significantly correlated with tumor purity, and the infiltration levels of B cells, CD8+ T cells, CD4+ T cells, macrophages, neutrophil cells, and dendritic cells in LUAD (Figure 9A). With these promising results, we next focused on the correlation between BTG2 expression and various immunocytochemical markers of different subsets of immune cells in LUAD, including CD8+T cells, T cells (general), B cells, monocytes, TAMs, M1 and M2 macrophages, neutrophils, natural killer cells, dendritic cells, functional T cells (Th1 cells, Th2 cells, Tfh cells, Th17 cells) and Tregs, as well as exhausted T cells. We found that BTG2 expression had a significant correlation with most markers of different subsets of immune cells, which mainly concentrated on T cells (general) (CD3D, CD3E and CD2), B cells (CD19 and CD79A), neutrophils (CD66b, CD11b and CCR7), natural killer cells (KIR2DL1 and KIR2DL4), dendritic cells (HLA-DPB1, HLA-DQB1, HLA-DRA, HLA-DPA1, CD1C and CD11c), Th1 cells (TBX21, STAT4, STAT1, and TNF), Th2 cells (STAT6, STAT5A and IL-13), Th17 cells (STAT3 and IL-17A), Tregs (FOXP3, CCR8, STAT5B and TGFB1) (Table 3). Strikingly, we found that immune escape markers such as CTLA4 and GZMB were significantly correlated with BTG2 expression in LUAD (Table 3). Additionally, we further analyzed the correlation between BTG2 expression and multiple chemokines and chemokine receptors in LUAD using the TISIDB database (Figure 9B–K). It is worth noting that the chemokines CCL14, CXCL17, CXCL2 and CCL7 (Figure 9C–F), as well as chemokines receptors CCR6, CCR4, CXCR2 and CCR2, were the most relevant adjustment factors to BTG2 expression in LUAD (Figure 9H–K).

Figure 9

The association between BTG2 and the immune infiltration levels and chemokines in LUAD. (A) Relationship of BTG2 with tumor purity and infiltrating levels of B cell, CD8+T cell, CD4+T cell, macrophage, neutrophil, and dendritic cell in TIMER database. (B–F) The correlation between BTG2 and chemokines in TISIDB database. (G–K) The correlation between BTG2 and chemokine receptors in TISIDB database.

Table 3

Correlation Analysis Between BTG2 and Markers of Immune Cells in Lung Adenocarcinoma from the TIMER Database

Description	Gene Markers	None		Purity
		Cor	p	Cor	p
CD8+ T cell	CD8A	0.094	*	0.068	0.133
	CD8B	0.09	*	0.067	0.138
T cell (general)	CD3D	0.142	**	0.111	*
	CD3E	0.232	****	0.222	****
	CD2	0.209	****	0.188	****
B cell	CD19	0.288	****	0.283	****
	CD79A	0.306	****	0.308	****
Monocyte	CD86	0.098	*	0.06	0.186
	CD115 (CSF1R)	0.16	***	0.13	**
TAM	CCL2	0.062	0.157	0.022	0.627
	CD68	0.068	0.121	0.04	0.375
	IL-10	0.179	****	0.15	***
M1 Macrophage	INOS (NOS2)	0.121	**	0.105	*
	IRF5	-0.017	0.704	-0.049	0.278
	COX2 (PTGS2)	0.053	0.233	0.069	0.125
M2 Macrophage	CD163	0.103	*	0.071	0.117
	VSIG4	0.06	0.175	0.027	0.553
	MS4A4A	0.118	**	0.08	0.0742
Neutrophils	CD66b (CEACAM8)	0.258	****	0.253	****
	CD11 b (ITGAM)	0.139	**	0.107	*
	CCR7	0.295	****	0.284	****
Natural killer cell	KIR2DL1	0.121	**	0.109	*
	KIR2DL3	0.041	0.352	0.018	0.687
	KIR2DL4	-0.117	**	-0.127	**
	KIR3DL1	0.079	0.073	0.061	0.179
	KIR3DL2	0.044	0.319	0.027	0.548
	KIR3DL3	-0.028	0.525	-0.033	0.466
	KIR2DS4	0.085	0.053	0.071	0.115
Dendritic cell	HLA-DPB1	0.301	****	0.287	****
	HLA-DQB1	0.188	****	0.152	***
	HLA-DRA	0.226	****	0.202	****
	HLA-DPA1	0.268	****	0.249	****
	BCDA-1 (CD1C)	0.323	****	0.299	****
	BDCA-4 (NRP1)	0.095	*	0.084	0.0632
	CD11c (ITGAX)	0.198	****	0.185	****
Th1	T-bet (TBX21)	0.176	****	0.157	***
	STAT4	0.17	***	0.144	**
	STAT1	-0.074	0.0934	-0.098	*
	IFN-γ (IFNG)	-0.036	0.418	-0.062	0.173
	TNF-α (TNF)	0.182	****	0.149	***
Th2	GATA3	0.053	0.228	0.009	0.844
	STAT6	0.311	****	0.318	****
	STAT5A	0.241	****	0.223	****
	IL-13	0.137	**	0.115	*
Tfh	BCL-6	0.225	****	0.238	****
	IL-21	0.07	0.11	0.055	0.226
Th17	STAT3	0.299	****	0.307	****
	IL-17A	0.108	*	0.104	*
Treg	FOXP3	0.15	***	0.118	**
	CCR8	0.212	****	0.183	****
	STAT5B	0.34	****	0.331	****
	TGF-β1 (TGFB1)	0.113	*	0.096	*
Exhausted T cells	PD-1 (PDCD1)	0.067	0.13	0.044	0.329
	CTLA4	0.192	****	0.172	***
	LAG3	0.051	0.25	0.033	0.467
	TIM-3 (HAVCR2)	0.062	0.159	0.017	0.706
	GZMB	-0.075	0.0882	-0.109	*

Notes: *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001.

Abbreviations: Cor, R value of Spearman correlation; p, p values; TAM, tumor-correlated macrophage; Th1, T helper 1 cell; Th2, T helper 2 cell; Tfh, follicular helper T cell; Th17, T helper type 17 cell; Treg, regulatory T cell.

Validation of Single Cell Sequencing

Considering the heterogeneity of different cells, we further validated the expression of BTG2 in LUAD tissue and different immune cells by single cell sequencing. As shown in Figure 10A, we filtered single cells of LUAD that expressed too many and too few genes. According to standard deviation, 15 PC with relatively gentle dimensions were selected for dimensionality reduction (Figure 10B). Based on PCA results, we mapped the expression of BTG2 gene in LUAD patients and immune cell infiltration, which showed BTG2 expression was highly expressed in NK cells, T cells, B cells, monocytes, macrophages, and neutrophils (Figure 10C and D).

Figure 10

Validation of the expression by single cell RNA-seq. (A) Quality control of LUAD data. (B) Standard deviation diagram of principal components. (C) Cell annotation of tSNE cluster graph (color-coded by clusters, cell subsets). (D) tSNE plot of BTG2 expression (red represents high gene expression).

Discussion

BTG2 maps within the chromosomal segment (1q32),10 and belongs to the BTG/TOB gene family that regulates cell cycle, cell proliferation, differentiation, and apoptosis.9,45,46 BTG2 is described as an immediate early gene47,48 and regulated by p53 activation.12 Previous studies have reported low expression of BTG2 in a variety of malignant tumors, including NSCLC,21 breast cancer,49 laryngeal carcinoma,50 prostate cancer,51 renal cell carcinoma,21,49–52 and closely associated with poor outcomes.21–23 Down-regulated BTG2 expression can promote tumorigenesis and progression through activating proliferation, migration, and invasion of tumor cells, while upregulated BTG2 expression generates the opposite effect.53–55 For example, MiR-25-3p was found to promote the proliferation of triple negative breast cancer56 by targeting BTG2. Li et al57 found that miR-365 promoted pancreatic ductal adenocarcinoma development by suppressing BTG2 expression. Lin et al58 found that miR-1278 inhibited metastasis of colorectal cancer via upregulation of BTG2. Tsui et al59 reported that BTG2 overexpression inhibited the growth of bladder cancer cells. Huang and colleagues found that down-regulated BTG2 expression promoted progression through inhibiting cancer stem cell-like features of side population cells in hepatocellular carcinoma, and low expression of BTG2 was associated with shorter OS and disease-free survival.60 Wang et al61 reported that BTG2 was abnormally decreased in esophageal squamous cell carcinoma (ESCC) patients, and a more pronounced decrease was found in radiotherapy-resistant patients as compared with radiotherapy-sensitive patients, suggesting that downregulation of BTG2 was able to predict the progression and radioresistance in ESCC. Recent investigations also reported that BTG2 overexpression could inhibit the proliferation and invasiveness of LUAD cells,62 while circular RNA circCRIM1 was able to promote BTG2 expression to suppress the migration, invasion, EMT, and glycolysis of LUAD cell.25 Wang et al further confirmed that hypomethylation and high expression of BTG2 significantly correlated with improved survival outcome in patients with LUAD.63 Additionally, some studies have found that BTG2 played an important role in tumor immune regulation.26–30 These studies suggested that BTG2 might be a potential prognostic immune-related biomarker for LUAD, but the role of BTG2 in LUAD has not been reported. In this study, we observed that BTG2 was significantly downregulated in LUAD. Decreased expression of BTG2 in LUAD was significantly correlated with advanced cancer stages and nodal metastasis. Further results showed that low BTG2 expression was associated with worse OS in LUAD. Besides, results based on a PPI network reveal that the top 10 BTG2-correlated genes were TP53 (p53), CCND1, MDM2, HSP90AA1 (HSP90N), CDK2, CDK4, ATM, SIRT1, CDK6, and BCL2L1. Previous studies demonstrated that these BTG2-related co-expressed genes were closely involved in tumorigenesis and progression.64–70 Consistently, our further study showed that all BTG2-related co-expressed genes in LUAD but CCND1, HSP90AA1, and CDK6 were closely associated with the duration of overall survival. Collectively, all the findings above suggested that BTG2 played a key role as a tumor suppressor and might become a promising prognostic biomarker as well as a molecular target for patients with LUAD.

It is well-known that classic driver gene mutations largely determine the therapeutic approaches and prognosis in patients with LUAD. Previous studies have reported that EGFR and ALK mutations were associated with prognostic markers of LUAD.71 EGFR has been shown to be associated with immune cell infiltration and overall survival in LUAD.72 We thus did correlation analysis between BTG2 expression and key driver gene mutations, including TP53(p53), EGFR, ROS1, ALK, KRAS, BRAF, ERBB2 (HER-2), and NRAS. The analysis showed that down-regulated BTG2 was significantly associated with TP53 and ROS1 mutations. It has been confirmed that TP53 mutation is one of the most frequently mutated genes in LUAD.73,74 TP53 mutation acquires oncogenic features, which may trigger chromosomal/genomic instability and further leads to a high tumor mutation burden, eventually resulting in more aggressive malignancy and worse clinical outcome in LUAD.74–78 ROS1 rearrangement, the major type of ROS1 gene alteration, has been confirmed as a vital biomarker for targeted therapy in advanced LUAD.79–81 Strikingly, recent studies highlighted that both TP53 mutation and ROS1 rearrangement but not EGFR mutation frequently overlapped with high PD‐L1 expression in LUAD. Moreover, a study has reported that TP53 mutation was associated with immune-related prognostic markers of LUAD.82 Ge et al83 found that TP53 mutation significantly varied with immune scores. In this study, the BTG2 was correlated to TP53 and ROS1 mutations, but not EGFR, indicating that BTG2 might be involved in the effect of TP53- or ROS1‐mutations on immune activity, and closely associated with immunity.44,77,78,84,85

We next conducted the enrichment analyses, and the results revealed that BTG2 was closely related to multiple immune-related processes and pathways including interleukin-17 production, leukocyte differentiation, cytokine binding, Th17 cell differentiation, cytokine–cytokine receptor interaction, B cell activation, TGF-β signaling pathway, B cell receptor signaling pathway. Based on these findings, we further analysed the associations between BTG2 expression and immune infiltrates in LUAD. We observed an obvious positive correlation between BTG2 expression and the infiltrated level of B cells, CD8+ T cells, CD4+ T cells, macrophages, neutrophils, and DCs in LUAD, which suggested that high BTG2 expression might up-regulate the abundance of immune-infiltrating cells. Consistently, relevant studies have shown that BTG2 was directly able to regulate the proliferation, development, and differentiation of immune infiltrating cells. For example, the BTG2-PRMT1 protein complex was found to antagonize the proliferation of pre-B cells to promote the development of B cells.27,28 In a MO5 melanoma transplantation model, BTG2 was found to transiently regulate anti-tumor immunity by affecting the frequencies of granzyme B+ CD8+ T-cells and CD107a+ CD8+ T-cells in vivo.30 Research from Passeri et al86 identified BTG2 as a regulator of differentiations of myelocytic leukemia cells and CD34+ hematopoietic precursor cells. Considering the heterogeneity of different cells, we further validated the expression of BTG2 in different immune cells by single cell sequencing, finding that BTG2 expression was highly expressed in NK cells, T cells, B cells, monocytes, macrophages, and neutrophils in the LUAD immune microenvironment. Similarly, Terra et al87 also found that BTG2 was highly expressed in immune organs such as thymus and spleen. These findings further enhanced the reliability of BTG2 regulation of immune cell infiltration in LUAD. Based on the close relationship between BTG2 and immune cell infiltration, we further demonstrated that the BTG2 expression had strong correlations with various marker sets, including T cells (general) (CD3D, CD3E and CD2), B cells (CD19 and CD79A), neutrophils (CD66b, CD11b and CCR7), dendritic cells (HLA-DPB1, HLA-DQB1, HLA-DRA, HLA-DPA1, CD1C and CD11c), Th1 cells (TBX21, STAT4, STAT1, and TNF), Th2 cells (STAT6, STAT5A and IL-13), Th17 cells (STAT3 and IL17A), Tregs (FOXP3, CCR8, STAT5B and TGFB1). Immune checkpoint inhibitors, including cytotoxic T lymphocyte-associated antigen 4(CTLA-4) and PD-1/PD-L1 inhibitors, have been shown to have dramatic anti-tumor activity in a variety of solid tumors, immune-related biomarkers therefore attract intense interest.88–91 In our study, significant correlations were found between BTG2 expression and T cell exhaustion markers including CTLA-4 and GZMB. Further exploration of the potential role of BTG2 in immune marker may benefit patients treated with immune checkpoint inhibitors. Altogether, these data reveal a critical role of BTG2 on immune infiltration in LUAD.

As we know, different immune infiltrating cells are recruited by different chemokines and chemokine receptors in the tumor microenvironment.92–95 Liu et al96 found that CC chemokine receptors might be immunotherapeutic targets and inflammation-related prognostic biomarkers in LUAD. Yu et al97 also reported that CXC chemokines were related to prognostic value. Based on these findings, we analyzed and demonstrated that the BTG2 expression was closely related to the levels of many chemokines (CCL14, CXCL17, CXCL2, and CCL7, etc) and receptors (CCR6, CCR4, CXCR2, and CCR2, etc). Evidently, CCL14 was reported to induce macrophage bone marrow homing, proliferation, and polarization in multiple myeloma.98 CXCL17 was able to recruit CD11b+Gr1 high F4/80-cells and accelerated tumor development.99 Xu et al100 found that ILC2s promoted hepatocellular carcinoma progression via CXCL2-neutrophil induced immunosuppression. Zhang et al101 reported that CCL7 recruited cDC1 to promote antitumor immunity in NSCLC. Lian and colleagues found that Eomes recruited Treg cells through the CCL20-CCR6 pathway, thereby promoting esophageal carcinoma progression.102 CCR4 has been shown to recruit Th2 cells which could induce the further expression of CCR4 ligands.103 Nywening et al104 revealed that myeloid recruitment in pancreatic ductal adenocarcinoma was disrupted by targeting both tumour-associated CXCR2+ neutrophils and CCR2+ macrophages. Based on our results and these previous researches, it is conceivable to hypothesize that BTG2 plays a vital role in the regulation of chemokine-mediated immune infiltration in LUAD.

Inevitably, there are some limitations in our studies. First, due to this research based on the analysis of multiple public databases, more prospective clinical trials should verify the prognostic value of BTG2. Second, further effective external experiments through LUAD cell lines and molecular biological methods are needed to clarify the mechanism of BTG2 in LUAD immunity. Thirdly, considering the limited predictive value of single gene, further studies on BTG2 single gene combined with multi-gene or multi-group analysis are needed.

Conclusion

In conclusion, our findings have effectively demonstrated that BTG2 expression is downregulated in LUAD, which is associated with higher cancer stage, and predicts poor prognosis. With close relationship with immune cell infiltration and tumor immune regulation, BTG2 may be a promising immune-related biomarker and molecular target for patients with LUAD.