Breast cancer originates from mammary epithelial cells, which is an aggressive tumour type with better prognosis [1–3]. Advanced breast cancer patients often develop distant metastasis in locations such as the bone. However, the underlying molecular mechanisms are still unknown. In the process of tumour occurrence and metastasis, key determining factors are the tumour’s molecular and cellular features, which are often used in the clinic to determine the prognosis. Bone metastasis are detrimental to the quality of life and life span of breast cancer patients and brings serious complications including pain, fracture, spinal cord compression, and malignant hypercalcemia [1–4]. Due to the frequent occurrence of bone metastasis in patients with advanced breast cancer, the pathogenesis and clinical management of bone metastasis are important and challenging topics in basic research and clinical practice. The ceRNAs network, including mRNA, miRNA, and lncRNA, and immune cells that infiltrate the tumour may be critical to further understand this phenomenon [12–15]. We observed that tumour samples with bone metastasis had significantly altered proportions of infiltrating cells compared with breast cancer without bone metastasis. We then developed two models to predict the prognosis of breast cancer patients with bone metastasis. The resulting AUC of the two nomograms demonstrated their value in a clinical setting.
Lncrna is a type of ncRNA with over 200 nucleotides, which has nothing to do with protein-coding [12–16]. New evidence suggests that lncRNA imbalance occurs frequently in many malignant tumours and are vital for carcinogenesis through post-transcriptional regulation and epigenetic modification [17, 18]. Here, we bioinformatically analysed the ceRNA network regulating bone metastasis of breast cancer with 20 protein encoded mRNAs, 2 lncRNAs and 18 miRNAs. We found a significant correlation between four protein-encoding genes (gjb3, camgv, ptprz1, fbn3) and their associated miRNAs and lncRNAs with the survival of breast cancer patients with bone metastasis. The AUC values of 1-, 3-, and 5-year survival were 0.746, 0.686, and 0.642 respectively. Using a hypergeometric test and correlation analysis, we found a significant correlation between gjb3 (dlx6-as1, hsa-mir-1-3p), fbn3 (dlx6-as1, hsa-mir-132-3p), camkv (dlx6-as1, hsa-mir-16-5p), ptprz1 (dlx6-as1, hsa-mir-181a-5p), and the total survival rate of breast cancer patients with bone metastasis. These results indicated that dlx6-as1 may occur and be a key player for bone metastasis in advanced breast cancer patients. We speculate that dlx6-as1 may regulate the occurrence and progression of metastasis by interacting with Wnt/β-catenin signalling.
Recently, an increasing number of studies show that aberrant lncrnas expression leads to the development of many kinds of malignant tumours, including breast cancer [12–18]. As a kind of lncrnas, dlx6-as1 is believed to be carcinogenic by regulating the progression of renal cell carcinoma, liver cell carcinoma, glioma, pancreatic cancer and lung adenocarcinoma [19–26]. Normal brain tissue has a high expression of DLX6-AS1, which is involved in the development regulation [24–26]. In recent years, it has been found that dlx6-as1 is abnormally expressed in a variety of tumour tissues and is closely related to a poor clinical outcome [19–26]. However, the molecular mechanisms of dlx6-as1 and how it contributes to the pathogenesis of breast cancer are still unclear. Zhao et al. demonstrated significant upregulation of dlx6-as1 in breast cancer tissues and cell lines [27]. Furthermore, high dlx6-as1 expression is linked to poor outcome regarding tumour size, lymph node metastasis, TNM stage, and survival of breast cancer patients [19–27]. SiRNA knockout of dlx6-as1 showed a reduction in proliferation, apoptosis, invasion, migration, and epithelial-mesenchymal transition (EMT) of breast cancer cells [25–27]. These findings suggest that the progression of breast cancer could be partly due to an overexpression of dlx6-as1. Studies have indicated that mRNA expression may be regulated by lncRNAs through competitive communication between cernas and miRNAs [28, 29]. For example, dlx6-as1 silencing inhibits cell proliferation, migration, and invasion in non-small cell lung cancer by interacting with mir-144. In addition, dlx6-as1 is important in the carcinogenesis of glioma by competing with mir-197-5p. In pancreatic cancer, knocking out the dlx6-as1 gene lead to inhibition of proliferation and metastasis of cancer cells through enhancement of mir-181b’s endogenous effects. In this study, dlx6-as1 may be important for the expression and regulation of miRNAs in breast cancer bone metastasis. Zhao et al. evaluated the expression of dlx6-as1 in breast cancer and analysis of the correlation between dlx6-as1 expression and clinicopathological parameters showed increased dlx6-as1 expression in tumour tissue compared with normal tissue, which was linked to poor prognosis of breast cancer patients [27]. Similar to pancreatic cancer, Dlx6-as1 gene knockout reduces the proliferation, invasion, and migration capacity of breast cancer cells and promoted apoptosis [27]. Furthermore, luciferase analysis confirmed that dlx6-as1 is an endogenous mediator of mir-505-3p and negatively regulates its expression. In addition, mir-505-3p inhibits runt related transcription factor 2 (Runx2) expression by binding directly to the 3′ untranslated region. Partial reversal of mir-505-3p’s carcinogenic effects can be achieved by overpressing Runx2. Wang et al. additionally showed that dlx6-as1 triggers its downstream effects on breast cancer cells by downregulating Fus [23]. These findings indicate that dlx6-as1 promotes breast cancer progression and is consistent with our results. Targeting Dlx6-as1 may be beneficial for the treatment of breast cancer.
The regulatory mechanism between lncrnas and miRNAs is extraordinarily complex. Among them, lncrna can be used as the combination of ceRNA and miRNA to compete and share mRNA, thus forming a complex lncRNA-miRNA-mRNA network [30]. Wnt/β-catenin signalling is critical in numerous cellular processes such as proliferation, invasion, and migration [31]. The involvement of Wnt/β-catenin signalling has been demonstrated in several malignant tumours including breast cancer and Wnt/β-catenin activation promotes tumour cell growth and metastasis [31, 32]. Previous studies have shown that Wnt activation inhibits memory T cells by reducing key transcription factors produced by these cells [33]. In our study, we found the activation of mast cells was correlated with Wnt6 expression. Zhang et al proved that Wnt signal is often activated under the action of dlx6-as1 [34]. Therefore, we speculate that Wnt pathway may be important for the impact of dlx6-as1 on the composition of immune cells. However, the molecular mechanisms of dlx6-as1-induced Wnt activation remains to be elucidated.
Guo et al. demonstrated dlx6-as1 upregulation in cell lines and tissues from bladder cancer patients, which augmented the proliferation, invasion, and migration of these cells by regulating EMT process and the activity of Wnt/β-catenin signalling [35]. dlx6-as1’s impact on the characteristics of cancer stem cells in osteosarcoma has also been investigated and was found to positively correlate with more advanced disease and poorer survival [36]. Similar findings have been demonstrated in prostate cancer patients. Recently, several studies have shown the participation of Wnt6, Barx1, ptprz1, and other genes in the regulation of Wnt/β-catenin signalling and their involvement in the development of malignant tumours [31–41]. These findings support our hypothesis that dlx6-as1 may affect the mRNA expression of Wnt6, Barx1, ptprz1 and other genes through ceRNA network to regulate Wnt/β-catenin signalling and participate in the distant bone metastasis of breast cancer.
In this study, we used six genes (camgv, alkal2, gabbr2, Barx1, fbn3, and Wnt6) to construct a multiple Cox risk regression model in the process of Lasso regression. Using these six genes to construct the nomogram based on Cox regression can predict the survival status of one year, three years and five years, and the prediction effect of the model is satisfactory.
In the study of immune infiltration of breast cancer bone metastasis, we found that plasma cell and follicular helper T cell were significantly different in two different samples. We also observed that the proportion of mast cells, gamma delta T cells, plasma cells, follicular helper T cells, and eosinophils varied depending on disease stage and progression. These results suggest that plasma cell, follicular helper T cell, mast cell and gamma delta T cell have potential biological prediction value for predicting bone metastasis, disease grading and staging of breast cancer, which is expected to be further verified and applied in clinical diagnosis and treatment. Subsequently, our correlation analysis showed that Wnt6 and gabbr2 positively correlated with mast cell activation. Therefore, our research focuses on the above six genes and their related pathways.
Our research inevitably has several limitations. The data included in our study are from Western countries and may not be directly extrapolated to patients in Asian countries. We were unable to comprehensively analyse the clinical and pathological parameters due to limited public information, which may impact our analyses. Therefore, we minimised bias by investigating the gene and protein expression of key biomarkers at the cell and tissue level.
Tumour microenvironment often affects the process of tumour invasion and migration [25, 26, 42]. Invasion of tumour cells is largely dependent on the composition of the extracellular matrix as well as growth factors that are secreted by the surrounding cells [25, 26]. Furthermore, metastasis of the tumour may have occurred in the early stage of the tumour development and is not directly related to proliferation. Therefore, it is necessary to determine the molecular mechanism leading to aggressive breast cancer with bone metastasis. We developed a ceRNA network based on breast cancer samples and the nomogram of tumour-infiltrating immune cells predicted the prognosis of breast cancer patients with and without bone metastasis with high accuracy. The predictive nomogram can provide more comprehensive clinical data for individual treatment of breast cancer bone metastasis. Further studies should be performed to show the interactions and communications between cancer cells and immune cells. In particular, the exosomes secreted by tumour cells contain ceRNAs and may also play a role mediating breast cancer metastasis.