Introduction
Globally, breast cancer is the most common cancer among women [
1‐
4]. Despite progress in diagnosing and treating breast cancer, approximately 12% of patients ultimately developed tumor metastasis [
4]. Of all cancer-related fatalities, breast cancer contributed to 23%, posing a substantial threat to women’s well-being among malignant diseases [
5‐
6].
Now, the main treatment methods include chemotherapy, targeted therapy, immunotherapy, surgery, endocrine therapy, and radiotherapy [
7]. The therapeutic choice depends on clinical-pathological risks and molecular sub-type. However, genetic analysis advancement has offered more precise decision-making to cope with the diversity of breast cancer [
8].
Most tumors exhibited hypoxia, which meant an imbalance between oxygen consumption and supply [
9‐
10]. Hypoxia was related to stem cell characteristics, angiogenesis, extracellular matrix organization, protein ubiquitination, immune evasion, and cancer cell metastasis [
9‐
15]. New insights had been gained from recent research on intricate cellular and genomic regulation networks involved in the hypoxic response. This included the epigenetic regulation of transcriptional coregulators, histone, chromatin modifications by hypoxia-inducible factor (HIF), and the expression of various non-coding RNAs [
16]. Almost all solid tumors exhibited hypoxia as a typical tumor micro-environmental (TME) feature due to the uncontrolled and rapid proliferation of tumors. Besides, hypoxia, a sign of TME, was essential in drug resistance. Hypoxia-induced drug resistance was closely related to these signaling pathways, including autophagy, drug efflux, and mitochondrial activity [
17].
Current studies showed that hypoxia stimulation could initiate epithelial-mesenchymal transition (EMT), which played a key rold in cancer progression [
18]. On one hand, EMT was associated with the acquisition of stem cell characteristics from breast cancer cells, which led to drug resistance and poor prognosis [
19‐
20]. On the other hand, EMT was involved in the formation of immunosuppressive microenvironment, resulting the impairment of anti-tumor immunity. For example, EMT destroed the immune synapses of breast cancer cells, which changed the susceptibility of cancer cells to T cell-mediated immune surveillance. It ultimately brought about the weakening of cellular immune function and immune escape [
21]. Moreover, Lei Xiang et al. revealed that the expression of Hypoxia-Inducible Factor-2a (HIF-2a) was significantly correlated with higher histology-grade and Ki67 index of breast cancer [
22]. As a result, hypoxia might be a hidden prognostic factor for breast cancer.
Previous studies had delineated a prognostic signature in breast cancer using hypoxia-related genes [
23‐
24]. However, these studies had some limitations, including needing more validation and so on. In this research, we aimed to construct a robust signature to predict the outcome of breast cancer patients with hypoxia genes and to explore the cell line function of these genes.
Discussion
Our research successfully established a nine hypoxia-related gene signature through TCGA database. The robust of the signature was validated by external dataset GSE131769. Through in vitro experiment, we found PSME2 played an anti-tumor role in breast cancer.
The incidence of breast cancer has increased dramatically in the past decade [
26]. As a marker of TME, hypoxia was involved in various aspects of tumor progression [
9,
27‐
28]. Moreover, hypoxia was associated with drug resistance in the treatment of breast cancer [
29‐
30]. According to our findings, hypoxia-related genes might be useful as biomarkers in predicting the long-term prognosis of breast cancer patients. Some scientists have tried to create risk models to elucidate the outcome of breast cancer patients [
31‐
33]. Nonetheless, these models either lacked clinical validation or experimental validation. As a result, a robust signature was urgently needed.
This study constructed a hypoxia-related signature containing nine genes in the TCGA training set. This signature could distinguish the outcome of high-risk and low-risk patients. It revealed excellent performance in the training set. The robustness of the model was double-validated by the test and external independent validation set (GSE131769).
To explore the function of DEGs. GO enrichment analysis was utilized. As a result, the DEGs were mainly involved in the nuclear division, chromosomal region, and division organelle fission, providing further insight into the main underlying mechanisms. KEGG pathway analysis indicated that the most significant pathway was the cell cycle and cellular senescence. According to Druker’s research, hypoxia played an essential role on the cell cycle at the transcriptome level [
34]. Protein synthesis could also be modified in the hypoxia environment [
34]. Thus, hypoxia-induced changes in the cell cycle were influenced by several factors.
Among the nine prognostic hypoxia genes, most of them were
widely reported in previous studies, except KCNJ11 and PSME2. CD24, CHEK1, and HOTAIR were possibly oncogenic genes. Cluster of differentiation 24 (CD24) was a glycosyl-phosphatidyl-inositol-anchored glycoprotein [
35]. Barkal et al. demonstrated that a role for tumor-expressed CD24 in promoting immune evasion through its interaction with the inhibitory receptor sialic-acid-binding Ig-like lectin 10 [
36]. In a meta-analysis, Wang et al. revealed that the putative stem cell marker CD24 was significantly associated with worse survival based on 5697 BC cases [
37]. In conclusion, the negative role of CD24 in breast cancer was evident.
CHEK1 was a checkpoint kinase. Previous studies reported that CHEK1 played a critical role in maintaining genomic stability and preventing the accumulation of DNA damage during cell division [
38]. Xu et al. found that CHK1 inhibition would enhance adriamycin (ADR) chemosensitivity [
39]. In addition, lncRNA HOTAIR was engaged in cellular metastasis in various cancers, such as colorectal cancer, hepatocellular cancer, and non-small-cell lung cancer [
40‐
42]. Notably, HOTAIR could serve as a molecular sponge for miR-20a-5p, promoting breast cancer progression and tumorigenesis by activating the expression of the HMGA2 protein [
43]. As a result, CHEK1 and HOTAIR also served as oncogenic roles in breast cancer.
ALOX15B, CA9, FOXM1, KCNJ11, NEDD9, and PSME2 might be protective factors in the hypoxia signature. Evidence has shown that carbonic anhydrase 9 (CA9), a glycoprotein of the zinc-containing enzyme family, was an inducible expressed gene in response to hypoxia in cancers [
44‐
45]. The exterior cellular acidity of CA9 has a supportive effect on carcinoma cells [
46].
For FOXM1 (forkhead box protein M1), previous research articles showed it played a fundamental role in tumorigenesis, which was mainly related to the regulation of cell cycle progression [
47‐
48]. For NEDD9, Hu et al. discovered histone deacetylase inhibitors promoted breast cancer metastasis by elevating NEDD9 expression [
49]. Also, NEDD9 over-expression caused hyper-proliferation of luminal cells and cooperated with the HER2 oncogene in tumor initiation [
50]. So, both CA9 and FOXM1 were novel markers of poor prognosis for breast cancer patients.
Proteasome activator subunit 2 (PSME2) was a protein involved in the regulation of the proteasome [
51‐
52]. According to prior research, PSME2 served as an indicator of the metastasis of tumors [
53]. In breast cancer, PSME2 had the potential to identify immune hot tumors and predict the response to immunotherapy [
54]. But the function of PSME2 in breast cancer cell lines was still unclear. When searching in Pubmed, the research about KCNJ11 and breast cancer was rare. To further investigate the function of PSME2 and KCNJ11 in breast cancer, cell line experiment was performed. MDA-MB-231 and MCF7 breast cancer cell lines were introduced. Our study demonstrated that reducing PSME2 expression with siRNA could inhibited cell viability, weakened colony formation, and suppressed invasion in BC cells. The cell line experiment further validated the tumor suppression role of PSME2. Nonetheless, the decrease in KCNJ11 expression had no notable impact on the survival of the cells. Multiple factors influenced the result. Cell experiments could not fully reflect the situation of KCNJ11 in the human, as the human body was a sophisticated system. The formation and progression of cancer was a complicated procedure that involved numerous genes and environmental factors. Also, the alterations in cancer cells were not solely caused by a single gene variation.
By analyzing the prognostic-related hypoxia genes mentioned above, it was found that these genes could affect tumor behavior in terms of causing changes and regulating immune response in tumor microenvironment, which would lead to drug resistance of tumor cells. Thus, the expression of hypoxia-related genes could predict the prognosis of breast cancer patients. Similarly, Jianxin wang et al. also successfully constructed a diagnostic signature with hypoxia related genes in the TCGA and GEO databases [
55]. However, they did not confirm their results or elucidate the molecular mechanisms through cell experiments. Interestingly, they also found that the HIF-1 signaling pathway was involved in the activation of cancer stem cells during tumor occurrence and development. Currently, more and more research tried to target the cellular response to hypoxia in human cancers. It appeared that inhibiting HIF-1 alpha activation was the primary approach and might enhance chemotherapy response. Xia Yang et al [
56] developed a signature by combining hypoxa and immune genes to predict prognosis of breast cancer. But they just focused on triple-negative breast cancer subtype. Also this research lacked of wet experiment.
There were some limitations of this study. Firstly, some important clinical factors, such as chemotherapy regime and duration of endocrine therapy were missing, which would lower the accuracy of the signature. Chemotherapy and endocrine therapy regimens were important factors affecting the therapeutic efficacy of patients. Secondly, the signature was established based on retrospective data. The fundamental flaw of retrospective data was the existence of various biases. These bias may led to a decrease in the authority of the model. A prospective study will be needed to verify the model. Thirdly, further exploration of the function of hypoxia genes in vivo is needed. Animal experiment could better simulate the internal environment of human body. Nonetheless, our research successfully developed a robust prognostic signature. In clinical practice, oncologist would face a dilemma whether the patient should offer intensive treatment. With the help of this signature to predict the risk of recurrence, oncologist could offer personalized treatment strategies according to the risk score. What’s more, PSME2 would be an indicator of good prognosis for breast cancer patients. Through further research, PSME2 may develop a kit to identify breast cancer patients with good prognosis and avoid over-treatment in the future.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.