GLS and GOT2 as prognostic biomarkers associated with dendritic cell and immunotherapy response in breast cancer

Abstract

Breast cancer is the females' most common cancer. Targeting the immune microenvironment is a new and promising treatment method for breast cancer. Nevertheless, only a small section of patients can profit by immunotherapy, and improving the ability to accurately predict the potential for immunotherapy response is still awaiting further exploration. In this study, we found that the key factors of glutamine metabolism, glutaminase 1 (GLS) and mitochondrial aspartate transaminase (GOT2), showed opposite expression patterns in breast cancer samples. Based on the expression level of GLS and GOT2, we divided the breast cancer samples into two clusters: Cluster 2 showed GLS expressed higher and GOT2 expressed lower, whereas Cluster 1 showed GOT2 expressed higher and GLS expressed lower. GSEA showed that the clusters were related to pathways of immunity. Further analysis showed that Cluster 2 was positively associated with immunity infiltration. Through WGCNA, we identified a module strongly correlated with glutamine metabolism and immunity and identified 11 dendritic cell-associated genes involved in dendritic cell development, maturation, activation and other functions. In addition, Cluster 2 also showed higher immune checkpoint gene expression, which suggest the Cluster 2 had even better response to immunotherapy. The validation dataset could also be clustered into two groups. Cluster 2 (GLS expressed higher and GOT2 expressed lower) of the validation dataset was also positively associated with dendritic cells and a better immunotherapy response. Thus, these data indicate that GLS and GOT2 are prognostic biomarkers which closely related to dendritic cells and better reacted to immunotherapy in breast cancer.

Keywords: Breast cancer; Dendritic cell; GLS1; GOT2; Immune checkpoint.

Fig. 1

The study design. (a) Workflow of data processing and bioinformatics analysis, comprising three main modules, including gene screening and consensus clustering, weighted gene coexpression network analysis and analysis of immune infiltration, and external validation.

Fig. 2

Consensus clustering of breast cancer samples on the basis of GLS and GOT2 expression. (a) ESTIMATE analysis of immune- and glutamine-associated regulators. The absolute value of four ESTIMATE indices (Stromal Score, Immune Score, ESTIMATE Score and Tumor Purity) > 0.250. Among them, red represented positive association, blue represented negative association, the darker the color, the stronger the association. (b) The expression of GLS and GOT2 in tumor and adjacent normal tissues. Red represents AdjN, blue represents tumor. ****p < 0.001. (c) The expression correlation analysis of GLS and GOT2. The expression of GLS was negative correlation with GOT2. R = 0.122, p = 3.693e-05. (d) Consistency matrix heat map was shown through consistency clustering (k = 2). Among them, light blue on behalf of Cluster 1, dark blue on behalf of Cluster 2. (e) The delta area helps to identification the most optimal number of clusters, which was k = 2. (f) Consistency score bar graph for subgroups with cluster counts between 2 and 9, and the optimal cluster counts was 2. Each colour represents a cluster, when the number of clusters is too large, and the number of patients in some clusters is too small, resulting in the disappearance of its colour block. (g) The GDC TCGA Breast Cancer (BRCA) dataset was divided into two clusters according to the expression of GLS and GOT2. The heatmap shows that GLS expressed lower and GOT2 expressed higher in Cluster 1 (n = 541), while GLS expressed higher and GOT2 expressed lower in Cluster 2 (n = 531).

Fig. 3

Comparing the characteristics of Immune Infiltration between the Cluster 1 and Cluster 2. (a) ESTIMATE analysis in Cluster 1 and Cluster 2. The Stromal Score, Immune Score and ESTIMATE Score of Cluster 2 were higher, and the tumor purity was lower comparing with Cluster 1. (b) CIBERSORT analysis revealed that the portion of dendritic cells in Cluster 2 was larger compared that with Cluster 1. (c) ssGSEA revealed that Cluster 2 showed a higher infiltration extent of dendritic cells compared with Cluster 1. Red represented Cluster 2, blue represented Cluster 1. ns, no significance, *p < 0.05, ***p < 0.005, and ****p < 0.001.

Fig. 4

Cluster 2 positively correlated with dendritic cell maturation-associated genes. (a) Volcano diagram showed the differential gene expression analysis between the Cluster 2 and Cluster 1. Red represented upregulated genes, blue represented downregulated genes, and black represented no change in genes comparing Cluster 2 with Cluster 1. (b) Network topology analysis of soft power (soft powers = 4). (c) Gene dendrogram and module colours. (d) Heatmap among module eigengenes, Cluster and ESTIMATE results. We identified a module (blue) correlated with glutamine metabolism (Cluster, R = 0.390, p = 8.000e-41) and immunity (Immune Score, R = 0.870, p = 0.000) (e) Scatter diagram of module eigengenes in the blue module (MM > 0.400 and GS > 0.300), and through it we found 71 hub genes. (f) Relevance between hub genes (11 genes which among hub genes were essential for dendritic cell development, maturation, activation and other functions.) and results of four ESTIMATE indices (Stromal Score, Immune Score, ESTIMATE Score and Tumor Purity). Red represented positive association, blue represented negative association, and the darker the color, the stronger the connection. (g) Association between the expression of 11 hub genes and GLS or GOT2. Red represented positive correlation, blue represented negative correlation. The darker the colour, the stronger the correlation. (h) Protein‒protein interaction network of 11 hub genes.

Fig. 5

Comparison of the expression of immune checkpoint gene between the Cluster 1 and Cluster 2. Comparing the expression of PD1-related genes (a), CTLA4-related genes (b), other immune checkpoint genes (c) and agonists of T-cell activation genes (d) between the Cluster 1 and Cluster 2. Red represented Cluster 2, blue represents Cluster 1. ****p < 0.001.

Fig. 6

Validation of clustering in the GDC TCGA Breast Cancer (BRCA) dataset. (a) Heatmap corresponding to consensus matrix using consensus clustering (k = 2). Among them, light blue represented Cluster 1, dark blue represented Cluster 2. (b) The delta area helps to make sure the most optimal number of clusters. (c) Consensus score bar graph for subgroups with cluster counts ranging from 2 to 9, and the optimal cluster counts was 2. Each colour represents a cluster, when the number of clusters is too large, and the number of patients in some clusters is too small, resulting in the disappearance of its colour block. (d) The validation dataset was divided into two clusters according to the expression of GLS and GOT2. The heatmap shows that GLS expressed low and GOT2 expressed high in Cluster 1 (n = 558), while GLS expressed high and GOT2 expressed low in Cluster 2 (n = 539).

Fig. 7

Validation of Characteristics of Immunity between the Cluster 1 and Cluster 2. (a) ESTIMATE analysis in Cluster 1 and Cluster 2. The Stromal Score, Immune Score and ESTIMATE Score of Cluster 2 were higher, and the tumor purity was lower compared with Cluster 1. (b) CIBERSORT analysis revealed that the portion of dendritic cells in Cluster 2 was larger compared with that in Cluster. (c) ssGSEA revealed that Cluster 2 showed a larger infiltration extent of dendritic cells compared with Cluster 1. Red represented Cluster 2, blue represented Cluster 1. ns, no significance, *p < 0.05, **p < 0.01, ***p < 0.005, and ****p < 0.001.

Fig. 8

Dendritic cell-associated gene expression analysis in the validation dataset. (a) Association between the expression of dendritic cell-associated genes and GLS or GOT2. (b) Association between hub genes and the ESTIMATE indice (Stromal Score, Immune Score, ESTIMATE Score or Tumor Purity). Red represented positive correlation, blue represented negative correlation. The darker the colour, the stronger the correlation.

Fig. 9

Comparison of the expression of immune checkpoint gene between the Cluster 1 and Cluster 2 in the validation dataset. Comparison of PD1-related genes (a), CTLA4-related genes (b), other immune checkpoint genes (c) and agonists of T-cell activation genes (d) between the Cluster 1 and Cluster 2. Red represented Cluster 2, blue represents Cluster 1. ****p < 0.001.