1. Analysis Procedure
Firstly, we downloaded the transcriptome profiling and corresponding clinical materials of 1109 breast tumor samples and 113 normal breast samples from the TCGA database. Secondly, we evaluated the proportion of immune/stromal component in TME using ESTIMATE algorithm and the relationship between Immune/Stromal Score and survival rates or clinic–pathological characteristics of BC patients. Thirdly, a total of 223 DEGs were obtained based on the immune score and stromal score. GO and KEGG enrichment analysis and PPI network were performed to learn the specific biological functions of these DEGs. Univariate COX regression was used to assess the effects of these DEGs on BC patients’ survival. Finally, CCL19 came to our eyes. And we focused on the association between the expression of CCL19 and the survival or clinic–pathological characteristics of BC patients. In addition, GSEA was carried out to find the biological functions of CCL19. CIBERSORT algorithm was applied to evaluate the TIC profile and correlation analysis was used to evaluate the relationship between the expression of CCL19 and the proportion of TICs in TIME.
2. Immune component in TME was associated the survival of BC patients
In order to explore the underlying relationship between the ratio of immune or stromal component in TME and the survival rate of BC patients, we first established the immune/stromal component evaluation system. The higher score reflected the larger amount of the immune or stromal component in TME. And then survival analysis was carried out by Kaplan–Meier plot and log-rank tests. The results showed that Immune Score was significantly associated with the survival rate of BC patients (Fig. 2b; p = 0.008), which meant that the higher Immune Score predicted the better survival. Yet, neither Stromal nor ESTIMATE Score was significantly related with the survival rate of BC patients (Fig. 2a, c; p = 0.687, 0.339). The above results implied that the immune component in TME might be more suitable for indicating the prognosis of BC patients.
3. Both immune and stromal component were significantly associated with the specific clinic-pathological features of BC patients
To further determine the associations between scores and clinic-pathological features of BC patients, we analyzed the corresponding clinical materials (Supplement Table 1) of BC patients, including the age, stages and TNM classification. First, we found that Stromal, Immune and ESTIMATE Score were significantly associated with the age of BC patients, which might indicate that the component of TME changed with age (Fig. 2d, i, n; p = 0.014, 0.03, 0.0068). For other clinic-pathological features, we found that there was significant difference in the Stormal Score in stage I compared with stage II (Fig. 2e, p = 0.017). There was also significant difference in the Stormal Score in stage T1 compared with stage T2 or T3, in stage N0 compared with stage N2 (Fig. 2f-g, p = 0.00023, 0.022, 0.015). However, for Immune Score, we were unable to find any difference in stages or TNM classifications (Fig. 2j-m). In addition, for ESTIMATE Score, there was also significant difference between T1 and T2 or T4 (Fig. 2p, p = 0.016, 0.029). Nonethless, none of Immune, Stromal or ESTIMATE Score was related with the distant metastasis of BC (Fig. 2h, m, f; p > 0.05). From above, we concluded that the proportion of immune and stromal component might be associated with the early progression of BC, instead of the distant metastasis.
4. DEGs Were Significantly Related With Immune-related Activities
In order to obtain the DEGs profile in TME, we performed the difference analysis between the high Stromal/Immune Score group and the low Stromal/Immune Score group. Heat map was ploted to show the gene profile in TME (Fig. 3a-b). In detail, there were a total of 773 DEGs obtained from the Stromal Score group (high Stromal Score vs. low Stromal Score), among which 634 DEGs were up-regulated and 139 DEGs were down-regulated (Fig. 3a, Supplement Table 2). Additionally, there were a total of 832 DEGs obtained from the Immune Score group (high Immue Score vs. low Immue Score), among which 729 DEGs were up-regulated and 103 DEGs were down-regulated (Fig. 3b, Supplement Table 3). Next, we conducted the intersection analysis to acquire the commonly up-regulated or down-regulated DEGs in Immune and Stromal Score groups. The venn diagram showed that 193 DEGs were up-regulated and 30 DEGs were down-regulated in both Immune Score group and Stromal Score group (Fig. 3c). Thus, these 223 DEGs (Supplement Table 4) might take a great part in modulating the status of TME. As a result, GO and KEGG enrichment analyses were further performed to assess the detailed functions of these 223 DEGs. Go enrichment analysis indicated that 223 DEGs greatly participated in lymphocyte activation, T cell activation and T cell differentiation (Fig. 3d). Similarly, KEGG enrichment analysis revealed that 223 DEGs were significantly associated with specific immune-related activities, such as cell adhesion molecules, cytokine-cytokine receptor interaction and chemokine signaling pathway (Fig. 3e-g). From above, these 223 DEGs seemed to be closely related with immune-related activities and might be a primary feature of TME in BC.
5. PPI network and univariate COX regression analysis of DEGs
To learn more about the potential interaction mechanisms of these 223 DEGs, PPI network was first constructed with the help of STRING database and Cytoscape software. The specific interaction network of DEGs was showed in Fig. 4a. The top 10 DEGs, which shared the leading nodes, were ranked in Fig. 4b. We next conducted the univariate COX regression analysis to acquire the DEGs, which were significantly associated with BC patients’ survival. As a result, 32 DEGs were found to be closely related with the hazard ratio (HR) of BC patients’ survival (Fig. 4c, p < 0.05; Supplement Table 5). Finally, the intersection analysis was used to find out the DEGs, which shared the leading nodes in PPI and were significantly associated with patients’ survival (Fig. 4d). Luckily, C-C motif chemokine 19 (CCL19) emerged.
6. CCL19 was significantly associated with the survival and specific clinic-pathological features of BC patients
CCL19, a ligand of the chemokine receptor C–C chemokine receptor type 7 (CCR7), is often expressed in the T-cell zones, including lymph nodes and thymus [12]. Recently, researches revealed that the overexpression of CCL19 in serum was associated with the progression of melanoma [13, 14]. In addition, CCL19 functioned as an immune-modulator for colon cancer therapy by closely communicating with DCs, T cells and B cells, thus regulating the adaptive immune responses [15]. However, there are few researches focusing on the role of CCL19 in BC. In the present study, the survival analysis suggested that the expression of CCL19 was positively related with the survival of BC patients (Fig. 5a, p = 0.009). As for the clinic-pathological features, the expression of CCL19 was significantly associated with ages, stages, T classifications and N classifications of BC patients (Fig. 5b-e, p < 0.05). However, it was not connected with M classification (Fig. 5f, p > 0.05), which was in accordance with the difference analysis between Score and clinic-pathological features (Fig. 2h, m, f; p > 0.05). From above, we might conclude that CCL19 was likely to be a prognostic factor and participate in the early progression of BC.
7. CCL19 Greatly Participated In Immune-related Activities
Given that CCL19 was significantly associated with the survival and specific clinic-pathological features of BC patients, we further carried out GSEA to explore the underlying mechanisms of CCL19. First, BC samples were divided into two groups according to the expression of CCL19, including CCL19 up-regulated group and CCL19 down-regulated group, compared with the median. For C2 collection, the KEGG gene sets database, genes in CCL19 up-regulated group were obviously enriched in immune-related activities, including B cell receptor signaling pathway, cell adhesion molecules cams, chemokine signaling pathway and so on (Fig. 5g). For hallmark gene sets, genes in CCL19 up-regulated group were primarily enriched in allograft rejection, complement, IL2 stat5 signaling and so on (Fig. 5h). For C7 collection defined by MSigDB, the immunologic gene sets, a variety of immune functional gene sets were enriched in the high CCL19 expression group (Fig. 5i). At last, for C5 collection, the gene ontology sets, the genes in CCL19 up-regulated group were also enriched in immune-related activities, such as adaptive immune response, cytokine binding, cytokine receptor activity and so on (Fig. 5j). Thus, we concluded that CCL19 greatly participated in the immune-related activities in BC patients and might be a modulator of TME.
8. CCL19 Was Associated With Tics In Time
Since we had found that CCL19 greatly participated in immune-related activities and might play an important role in the regulation of TME, we speculated that there were underlying connections between CCL19 and TIME. CIBERSORT algorithm was first applied to evaluate the proportion of TICs in BC samples (Fig. 6a). The associations between TICs were displayed in Fig. 6b. The difference analysis was performed to verify the associations between the expression of CCL19 and 22 TICs. The violin plot suggested that 15 TICs were significantly associated with the expression of CCL19 in BC samples (Fig. 6c, Supplement Table 6). Next, the correlation test showed that 10 TICs were positively related with the expression of CCL19 (Fig. 7a-j, Supplement Table 6), and 6 TICs were negatively related with the expression of CCL19 (Fig. 7k-p, Supplement Table 6). Finally, the intersection analysis between the difference test and correlation test revealed that 15 TICs were significantly associated with the expression level of CCL19 (Fig. 6d, Supplement Table 6). As a consequence, CCL19 was significantly associated with multiple TICs in TIME and might play an important role in modulating the TIME of BC through communicating with various TICs.