Advances in Biomarkers and Endogenous Regulation of Breast Cancer Stem Cells

Breast cancer is one of the most common cancers. Even if breast cancer patients initially respond to treatment, developed resistance can lead to a poor prognosis. Cancer stem cells (CSCs) are a group of undifferentiated cells with self-renewal and multipotent differentiation characteristics. Existing evidence has shown that CSCs are one of the determinants that contribute to the heterogeneity of primary tumors. The emergence of CSCs causes tumor recurrence, metastasis, and therapeutic resistance. Previous studies indicated that different stemness-associated surface markers can identify other breast cancer stem cell (BCSC) subpopulations. Deciphering the critical signaling networks that are involved in the induction and maintenance of stemness is essential to develop novel BCSC-targeting strategies. In this review, we reviewed the biomarkers of BCSCs, critical regulators of BCSCs, and the signaling networks that regulate the stemness of BCSCs.


Introduction
Cancer has become one of the world's public health problems. According to the latest statistics, there are 19.3 million new cancer cases worldwide, and 2.26 million new cases of breast cancer among them, therefore breast cancer is ranked first [1]. In China, according to cancer statistics in 2018, it was found that the incidence of breast cancer is increasing, and the onset age is becoming younger [2]. Research suggests that breast cancer is closely related to genetic and environmental factors [3,4]. Breast cancer is a highly heterogeneous disease, divided into four subtypes, including Luminal A, Luminal B, HER2 positive, and triple-negative breast cancer (TNBC) [5].
Despite the rapid development of medical technology, breast cancer is still a recurrence risk. BCSCs are often associated with recurrence, metastasis, therapeutic resistance, and other biological behaviors [6]. Emerging evidence indicates that, although BCSCs account for only a small subset of tumor cells, they have tumorigenic potential, multipotent differentiation, and self-renewal capabilities [7]. BCSCs are derived from various sources, including normal stem-cell-gained mutations and de-differentiation of cancer cells induced by microenvironment components [8].

Epigenetic Factors
BCSCs have a unique gene expression signature, predominantly determined by an array of epigenetic and transcription factors. B lymphoma Mo-MLV insertion region 1 homolog (BMI1) is a polycomb-group (PcG) transcriptional repressor that is well documented in promoting BCSC self-renewal and tumorigenicity [22]. The molecular chaperone heat shock protein 90 alpha (HSP90α) can maintain the expression of BMI1, subsequently increasing the self-renewal ability of BCSCs [23]. In contrast, miR-494-3p inhibits BCSC self-renewal by targeting BMI1 [24]. Recently, the activated interleukin-1 receptor type 2 (IL1R2) was shown to recruit USP15 in order to stabilize BMI1 and promote the self-renewal and metastasis of BCSCs [25].
PRMT1 is an asymmetric arginine N-methyltransferase in mammalian cells that catalyzes the asymmetric demethylation of histone H4 arginine 3 (H4R3) and modifies the active chromatin [33]. In addition, PRMT1 modulates cell function through the methylation of forkhead box O1 (FOXO1) and ERα proteins [34]. PRMT1 can induce MCF10A cells to develop stem cell characteristics and self-renewal abilities [33]. PRMT1 may increase the attributes of TNBC stem cells by activating the signal transducer and activator of transcription 3 (STAT3). Moreover, the epidermal growth factor receptor (EGFR) signal mediated by PRMT1 may contribute to the upregulation of ZEB1 and the promotion of BCSC generation [34]. The same family member PRMT5 is a type II methyltransferase that controls the symmetrical demethylation of arginine residues on target proteins in the cytoplasm and nucleus. It modulates transcription through the methylation of transcription factors, such as nuclear factor kappa B (NF-κB), p53, and E2F Transcription Factor 1 (E2F1) [35]. PRMT5 promotes the expansion of stem cells through histone methylation and the expression of forkhead box P1 (FOXP1) and Kruppel-like Factor 4 (KLF4), thereby enabling the development of breast tumors and chemotherapy resistance [36]. Similarly, PRMT5 methylates KLF5 to prevent its phosphorylation, ubiquitination, degradation, and facilitates the transcription of downstream target genes, thereby promoting the maintenance and proliferation of BCSCs [37].
Epigenetic-factor histone deacetylases (histone deacetylases, HDACs) target lysine residues to facilitate chromatin condensation and to regulate biological processes such as mitosis, differentiation, autophagy, and apoptosis [38,39]. HDAC1, HDAC5, and HDAC7 are necessary to maintain BCSCs. HDAC7 overexpression is sufficient to increase the CSC phenotype, manifested in the increases in sphere formation and tumor-initiating cell frequency [39]. HDAC5 silencing inhibits the growth, migration, and invasion of BCSCs and increases apoptosis [38]. HDAC1 stabilizes KLF5 protein by preventing its ubiquitination and degradation [40]. The expression of HDAC1 and KLF5 is positively associated with breast cancer [40]. HDAC inhibitors inhibit the expression of KLF5 and the tumorigenesis of breast cancer in vivo [38,40]. At the same time, another study showed that

Transcription Factors and Co-Activators
In addition to epigenetic factors, BCSCs are also regulated by transcription factors (Table 3). Octamer-binding transcription factor-4 (OCT4) plays a vital role in stem cell self-renewal. The high expression of OCT4 in 4T1 breast cancer cells enhances the mammosphere formation of CSCs in vitro [77]. The knockdown of OCT4 in an MCF-7 tumor-model induced apoptosis and inhibited tumor growth [78]. Estrogen induced OCT4 expression in MCF-7 cells and promoted mammosphere formation [79]. Programmed death ligand (PD-L1) maintains CSC stemness by promoting OCT4 phosphorylation, but the histone demethylase jumonji domain-containing protein-3 (JMJD3) decreases breast cancer stem cell-like properties by downregulating OCT4 [80,81].
C-MYC and SOX2 are two stem cell transcription factors. The overexpression of c-MYC in MDA-MB-468 cells facilitates the expression of ALDH [91]. Caveolin-1 inhibits the c-MYC-mediated metabolic reprogramming function of BCSCs [92]. In addition, p62 enhances the stem-like properties of BCSCs by stabilizing c-MYC [93]. SOX2 knockdown reduces BCSC stemness [94]. The tumor suppressor transcriptional repressor GATA binding 1 (TRPS1) inhibits SOX2 expression and the tumorigenic ability of CSCs [95]. Similarly, FOXO3a inhibits SOX2 expression and BCSC tumorigenicity [96]. Additionally, knockdown of SOX9 significantly inhibits the tumorigenicity of MDA-MB-231 cells, with a 70-fold decrease in tumor initiation capacity and a 40-fold increase in the ability to inhibit lung metastasis compared with the control [97]. More importantly, SOX9 and SLUG have significant advantages in the synergistic reversal of differentiated luminal cells into mammary stem cells [97].
The activation of the Hedgehog signaling pathway also increases the number of BCSCs and the formation of mammospheres. TSPAN8 is upregulated in BCSCs, interacts with the SHH-PTCH1 complex, and promotes CSC stemness [14]. TSPAN8 was recently found to enter the nucleus by binding to 14-3-3θ, importin-β, and cholesterol [14]. Ubiquitin-specific peptidase 37 (USP37) activates the Hedgehog pathway by increasing Smo and GLI1 expression levels to enhance the characteristics of BCSCs [122]. The GLI1-derived tumors amplify a portion of CSCs featured by Keratin 6 and BMI1, supporting the role of Hedgehog signaling in breast cancer development [123].
TAZ, an effector of the Hippo pathway, is necessary to maintain the self-renewal ability of BCSCs [124]. Vascular endothelial growth factor (VEGF)/neuropilin-2 (NRP2) signaling participates in TAZ activation through a Rac1-dependent mechanism, enhancing the sphere-forming ability of BCSCs [125]. Silencing of the discs large homolog 5 (DLG5) might improve the activity of TAZ, thereby maintaining the self-renewal ability and stemness of BSCSs [126]. We previously reported that TAZ can stabilize KLF5 [127] and that tumor necrosis factor-alpha (TNFɑ) induces TAZ expression to increase the stemness of BCSCs [128]. Similarly, YAP was reported to promote the stemness of BCSCs [129]. Activating the competing endogenous RNA (ceRNA) network associated with star-related lipid transfer domain-containing 13 (STARD13) reduces YAP/YAZ activity, thereby inhibiting the formation of BCSCs [130]. YAP interacts with β-catenin, and TEA domain transcription factor 4 (TEAD4) cooperates in the nucleus to modulate CSCs in BLBC [131].
The activation of the Hedgehog signaling pathway also increases the number of BCSCs and the formation of mammospheres. TSPAN8 is upregulated in BCSCs, interacts with the SHH-PTCH1 complex, and promotes CSC stemness [14]. TSPAN8 was recently found to enter the nucleus by binding to 14-3-3θ, importin-β, and cholesterol [14]. Ubiquitinspecific peptidase 37 (USP37) activates the Hedgehog pathway by increasing Smo and GLI1 expression levels to enhance the characteristics of BCSCs [122]. The GLI1-derived tumors amplify a portion of CSCs featured by Keratin 6 and BMI1, supporting the role of Hedgehog signaling in breast cancer development [123].
TAZ, an effector of the Hippo pathway, is necessary to maintain the self-renewal ability of BCSCs [124]. Vascular endothelial growth factor (VEGF)/neuropilin-2 (NRP2) signaling participates in TAZ activation through a Rac1-dependent mechanism, enhancing the sphere-forming ability of BCSCs [125]. Silencing of the discs large homolog 5 (DLG5) might improve the activity of TAZ, thereby maintaining the self-renewal ability and stemness of BSCSs [126]. We previously reported that TAZ can stabilize KLF5 [127] and that tumor necrosis factor-alpha (TNFscripta) induces TAZ expression to increase the stemness of BCSCs [128]. Similarly, YAP was reported to promote the stemness of BCSCs [129]. Activating the competing endogenous RNA (ceRNA) network associated with star-related lipid transfer domain-containing 13 (STARD13) reduces YAP/YAZ activity, thereby inhibiting the formation of BCSCs [130]. YAP interacts with β-catenin, and TEA domain transcription factor 4 (TEAD4) cooperates in the nucleus to modulate CSCs in BLBC [131].
Receptor tyrosine kinases (RTKs) can activate the downstream PI3K/AKT/mTOR and MAPK signaling pathways. Type 1 insulin-like growth factor receptor (IGF-1R) maintains BCSCs by activating the downstream PI3K/Akt/mTOR pathway [132]. The downregulation of hypoxia-inducible factor 2 alpha (HIF-2α) expression can inhibit BCSCs by inhibiting the PI3K/AKT/mTOR signaling pathway [133]. B7-H3 activates MEK by binding to major vault protein (MVP), which regulates the MAPK kinase pathway and increases the proportion of BCSCs [134]. EGFR upregulates cyclooxygenase 2 (COX-2) to activate Nodal signaling and promote BCSC self-renewal [135]. SGCE stabilizes the level of EGFR by breaking the interaction between c-Cbl and EGFR, activating the PI3K/Akt pathway to foster breast cell stemness [16]. Wang et al. reported that the JAK/STAT3 pathway promotes BCSC differentiation by regulating lipid metabolism and that inhibition of this pathway blocks BCSC selfrenewal [136]. WW domain-containing oxidoreductase (WWOX) inhibits the proliferation and metastasis of breast cancer cells by inhibiting the phosphorylation of JAK2 to hinder STAT3 activation [137]. WWOX also inhibits the expression of KLF5 in breast cancer [138]. In contrast, in TNBC, EGFR activates the JAK/STAT3 pathway by promoting the phosphorylation of STAT3, thereby facilitating the proliferation and invasion of tumor cells [139].
Alec et al. found that transforming growth factor-beta (TGF-β) activated the cytokine receptor leukemia inhibitory factor-receptor (LIFR) to initiate the JAK-STAT signaling pathway, subsequently driving the formation of BCSCs [140]. Jun et al. showed that TGF-β enhances the self-renewal capacity of BCSCs by upregulating fibronectin and Smad3dependent COX-2 expression [141].

Conclusions
Although the 5-year survival rate of breast cancer has been dramatically improved, recurrence is still a big challenge for breast cancer treatment. BCSCs are closely associated with recurrence. Notwithstanding that biomarkers have demonstrated their value in identifying BCSCs, it is still difficult to distinguish between normal stem cells and BCSCs. Due to the heterogeneity of breast cancer, the expressions of BCSC markers are varied and show distinct clinical values in different breast cancer subtypes. Moreover, distinct BCSC subclones may co-exist with a heterogeneous tumor, and new BCSC subclones may be induced during tumor treatment. Therefore, it is necessary to combine the analysis of BCSC markers in certain phases during tumor progression to improve BCSC-based prognosis. The continuous updating of breast cancer-specific biomarkers has an important guiding significance for CSC identification and separation. Novel therapeutic strategies that target BCSC and BCSC niches and the rest of the cancer are required to target the entire cancer and prevent metastasis or recurrence. Currently, there are several treatments for BCSCs, including cytotherapy (NCT02915445), antibody-based biopharmaceuticals (NCT01954355), synthetic small molecule compounds (NCT00645333), natural compounds and their products (NCT01608867), and nucleic acid medicines. In the I-SPY2 (Investigation of serial studies to predict your therapeutic response with imaging and molecular analysis) clinical trial platform, 17 new agents were evaluated in combination with neoadjuvant chemotherapy for women with locally advanced breast cancer [148]. By conducting such clinical trials, promising drugs were approved more quickly, and the drug development process was shortened [149].
Cancer occurrence is a multi-factor, multi-stage, multi-gene mutation accumulation process. Various signaling pathways may be involved in cancer occurrence and development simultaneously. Most of the signaling pathways regulating BCSCs are evolutionarily conserved and shared with normal stem cells, which makes them inappropriate as therapeutic targets. With a deep understanding of BCSCs, we expect more specific regulation of BCSCs pathway factors to be found. The relationship between multiple pathways is closely supported, providing a more theoretical basis for developing new targeted therapies in order to overcome current breast cancer treatment limitations.
Author Contributions: W.C. and L.Z. wrote and edited the paper and generated the tables and figure. S.L. and C.C. revised the paper. All authors have read and agreed to the published version of the manuscript.