Since the imbalance between normal cells and the immune system is considered to induce tumorigenesis, several studies have demonstrated significant roles for CXC chemokines in tumor progression(32–34), tumor microenvironment(35), and prognostic outcomes(36). Previous studies suggested that CXC chemokines, especially CXCL3 and CXCL13, presented as the prognosis markers in rectal adenocarcinoma (READ). For instance, lower expression of CXCL3 exhibited poor overall survival in READ(13). In the rectal cancer liver metastases, M2 macrophage polarization promoted tumor colonization by secreting CXCL13 and activated the CXCL13/CXCR5 axis in rectal cancer cells(37). These findings indicated that both CXCL3 and CXCL13 possibly had dual biological functions (tumor promoter/tumor suppressor) in READ, which consisted of the previous researches about chemokines (38). In detail, chemokines might promote metastasis by acting directly on the invasion and migration of tumor cells, while chemokines also recruited immune cells towards tumors (39, 40). In different diseases, each chemokine might have its unique function (38). However, the role of the immune microenvironment in READ remained unclear, and studies with large samples were needed. Therefore, we aimed to investigate whether CXC chemokines played crucial roles in the regulation of the immune microenvironment in READ.
We first characterized the CXC chemokine expression patterns in READ. The aberrant expression of three CXC chemokines (CXCL3, CXCL12, and CXCL13) was consistent in READ tissues compared with normal ones in both TCGA and ONCOMINE databases. Significantly increased CXCL3 level was observed in READ tissues compared with normal ones, while decreased CXCL12 and CXCL13 levels were observed in READ tissues compared with normal ones. Further survival analysis revealed that both lower expression of CXCL3 and CXCL13 predicted a poorer prognosis of READ. As CXC chemokines mediate immunocyte migration(4), we also found that the expression of CXCL3 and CXCL13 correlated with the infiltration of several immune cells in READ. The expression of CXCL3 was negatively and significantly correlated with macrophage infiltration, while positively correlated with neutrophil infiltration. On the other hand, CXCL13 level was negatively and significantly correlated with tumor purity, while positively correlated with the infiltration of B cell, CD8 + T cell, neutrophil, and dendritic cell. Furthermore, there was a significant positive correlation between CNV of CXCL13 and macrophage infiltration.
CXCL3 is reported as a chemoattractant of neutrophils(41, 42), which is consistent with our TIMER results. Multiple studies indicate that CXCL3 is an oncogene that promotes the progression of various malignancies, such as prostate cancer, breast cancer, and colorectal cancer(20, 43, 44). Our results show that CXCL3 is upregulated in READ, indicating that CXCL3 could also be oncogenic in READ. Surprisingly, upregulated CXCL3 is associated with a better prognosis. Similarly, CXCL3 is also upregulated in colon cancer, in which upregulated CXCL3 indicates better clinical outcomes(45). In esophageal squamous cell carcinoma(46), upregulated CXCL3 secreted from cancer cells unexpectedly increased the recruitment and the antitumor effect of neutrophils. Collectively, these findings indicate that in READ, CXCL3 may present dual character, which promotes tumor progression and upregulates neutrophil recruitment simultaneously.
On the other hand, CXCL13 was originally identified in the stromal cells from B cell follicles which could regulate homing of B cells and follicular T cells (Tfh)(47, 48). It is constitutively secreted by stromal cells in secondary lymphoid tissues, including the spleen and lymph nodes. This is consistent with our results as CXCL13 expression positively correlated with the infiltration level of B cells in READ. In several solid tumors, CXCL13 can also promote tumor cell proliferation and metastasis through CXCL13/CXCR5 axis, such as colorectal cancer(49), breast cancer(50, 51), and prostate cancer(52, 53). In colorectal cancer(54), lung cancer(55), and gastric cancer(56), CXCL13 is upregulated and serves as a prognostic biomarker as well as a therapeutic target, in which lower CXCL13 expression associates with the better prognostic outcome. Reversely, our results show that CXCL13 is downregulated in READ, and lower CXCL13 expression is linked to a worse prognosis. In terms of mechanism, downregulated CXCL13 may reduce B cells and Tfh cells infiltration, and thus lead to READ progression and poor prognosis, which is similar to breast cancer and colorectal cancer. These findings indicate the dual character of CXCL13, similar to CXCL3, in different tumors, and CXCL13 mainly performs anti-tumor effect on READ.
Long non-coding RNAs (lncRNAs) function effect as miRNAs sponges to interfere the functions of miRNAs targeting to mRNAs in various tumors[60, 61]. In our study, 12 miRNAs and 14 lncRNAs were identified as ceRNA networks of CXCL3, while eight miRNAs and 22 lncRNAs were identified targeting CXCL13. Among all these miRNAs, miR-425-5p, which downregulated CXCL3, was reported as an oncogene[20, 21] and inducing chemo-resistance in colorectal cancer (CRC). It was consistent with our identification that lower expression level of miR-425-5p and higher expression level of CXCL3 were presented in CRC. Additionally, it was reported that overexpression of miR-425-5p could induce the M2 polarization of macrophages, which enhanced tumor metastasis by secreting vascular endothelial growth factors (VEGF). It suggests that miR-425-5p may regulate M2 polarization of macrophages via CXCL3. According to prediction of miRNA-lncRNA from database DIANA, we found that miR-425-5p bound to lncRNA chr22-38_28785274–29006793.1. Regarding to lncRNA chr22-38_28785274–29006793.1, it was considered as the member of mRNA/miRNA/lncRNA ceRNA networks and involved in the infiltration of CD4 + and CD8 + T cell in colon cancer by bioinformatic analysis without biological validation. Thus, we assume that CXCL3/miR-425-5p/chr22-38_28785274–29006793.1 network may upregulate CXCL3 in READ and promote immune infiltration to exert the anti-tumor immunological effect.
Several limitations should be addressed. First, the bioinformatics analysis only focused on the transcriptional level of CXC chemokines, which could only reflect specific changes but not global changes of READ immune status. Second, the data were acquired from public resources, so the bias in the study cannot be completely eliminated. Last, ceRNA networks were predicted based on bioinformatics algorithms instead of experiments. To overcome these issues, further experiments in vitro and in vivo are warranted.