SOX factors as cell-state regulators in the mammary gland and breast cancer

Abstract

Emerging evidence has shown that several SOX family transcription factors are key regulators of stem/progenitor cell fates in the mammary gland. These cell-fate regulators are often upregulated in breast cancer and contribute to tumor initiation and progression. They induce lineage plasticity and the epithelial-mesenchymal transition, which promotes tumor invasion, metastasis, and therapeutic resistance. SOX factors act through modulating multiple oncogenic signaling pathways in breast cancer. In addition to the cell-autonomous functions, new evidence suggests they can shape the tumor immune microenvironment. Here, we will review the molecular and functional evidence linking SOX factors with mammary gland development and discuss how these cell-fate regulators are co-opted in breast cancer.

1. Introduction

Despite progress in screening, diagnosis, and treatment strategies, breast cancer remains a leading cause of cancer death among women, with more than 42,000 estimated deaths annually in the US alone [1]. Breast cancers are stratified based on the expression of markers such as estrogen receptor (ER), progesterone receptor (PR), and human epidermal growth factor receptor 2 (HER2). Transcriptomic and genomic profiling has uncovered substantial genetic and molecular heterogeneity of breast cancer. At least six intrinsic subtypes were identified based on gene expression profiling: Luminal A, Luminal B, HER2-enriched, Basal-like, Normal-like, and Claudin-low [2–4]. A combination of copy number variations and gene expression profiles has further expanded the classification to ten integrated subgroups [5]. These subtypes show distinct features in incidence, treatment responses, and clinical outcomes [4,6,7]. In particular, basal-like breast cancer (BLBC), which accounts for over 70% of the aggressive triple-negative breast cancer (TNBC), displays increased genomic instability, poor prognosis, and high risks of relapse [6,8,9]. Because the triple-negative feature excludes the benefits of endocrine therapy or anti-HER2 therapies, the patients with BLBC mainly rely on chemotherapy, which is accompanied by adverse side-effects and drug resistance [10]. To develop effective therapeutic strategies, there is a pressing need to understand the causes of the biological heterogeneity of breast cancer cells including their molecular characteristics and cellular states.

Delineation of the mammary epithelial differentiation hierarchy lays the groundwork for understanding breast cancer heterogeneity. The mammary epithelium is composed of two main lineages: luminal cells that line the central lumen and myoepithelial cells that are located in a basal position and contact the basement membrane. The luminal cells are further classified into ER-positive and ER-negative cells based on the expression of ESR1. Two main approaches have been applied to define stem/progenitor cells and their differentiation trajectories in the mammary gland in vivo. Initially, cleared mammary fat pad transplantation was used to measure the repopulating capacity of defined cell subpopulations, which led to the identification of multipotent stem cells that are capable of regenerating the entire mammary ductal tree [11–14]. Subsequently, in situ genetic lineage tracing was used to map the cell fate under unperturbed physiological conditions rather than measuring the regeneration potential of cells under a stress condition during transplantation. These studies reveal the existence of multipotent and unipotent stem/progenitor cells [15,16]. Both multipotent and unipotent stem/progenitor cells are involved in mammary gland development and homeostasis [15–17]. A population of embryonic mammary epithelial cells is multipotent, which commits to unipotent stem/progenitor cells during late embryogenesis [18–20]. The unipotent stem/progenitor cells play a predominant role in postnatal mammary epithelium growth and homeostasis. Lineage tracing and cell division kinetics experiments in mice show that the postnatal mammary basal cells, ER⁺ and ER⁻ luminal cells are three separate lineages maintained by distinct unipotent stem/progenitor cells [15,21–26]. However, the multipotent stem cells can still be detected in subpopulations of postnatal basal cells, which may serve as reserved stem cells [16,27]. In this review, we will refer to the multipotent stem cells as mammary stem cells (MaSCs) and the unipotent cells as progenitors for clearly distinguishing these cells although the unipotent progenitor cells do have long-term self-renewal abilities and contribute to cancer pathogenesis as discussed below.

The longevity and self-renewal properties of the multipotent stem and unipotent progenitor cells underline their potential to be the “cells-of-origin” of cancer. Interestingly, a comparison of gene expression profiles of different mammary cell types and breast tumors suggests that distinct breast cancer subtypes may originate from specific mammary cell lineages [28,29]. For instance, BLBC is considered to originate from ER-negative luminal progenitors on the basis of similarity in gene expression profiles between BLBC and luminal progenitors and the ability of mouse and human luminal progenitors to generate BLBC [29–31]. Interestingly, BLBC also shares high degrees of transcriptomic similarity with fetal multipotent mammary stem cells [20,32,33]. This supports the notion that BLBC tumorigenesis may involve the reactivation of embryonic multipotency in committed luminal progenitors [34–36]. Hence, elucidating the molecular basis regulating stem/progenitor cells will pave the way for understanding the etiology of breast cancer. As well-established stem/progenitor cell fate regulators, the SOX family transcription factors have been implicated in the development of mammary epithelium and breast cancer progression and metastasis. In this review, we will discuss the recent findings regarding the role of SOX factors in the context of mammary gland development and breast cancer.

2. Classification and characterization of the SOX family transcription factors

The Sex-determining Region Y-related HMG-box (SOX) family transcription factors play pivotal roles in cell fate decisions during development [37,38]. The Sry gene is the founding member of the Sox family, which is encoded in mammalian Y chromosomes and carries a highly conserved high-mobility group (HMG) domain that mediates DNA binding [39,40]. So far, 20 different Sox genes have been identified and are classified into eight groups based on their sequence similarities and domain structures (Table 1) [41].

Table 1.

Classification of the SOX family transcription factors

Groups	Genes
SOXA	Sry
SOXB1	Sox1, Sox2, Sox3
SOXB2	Sox14, Sox21
SOXC	Sox4, Sox11, Sox12
SOXD	Sox5, Sox6, Sox13
SOXE	Sox8, Sox9, Sox10
SOXF	Sox7, Sox17, Sox18
SOXG	Sox15
SOXH	Sox30

Each SOX subgroup has unique defining features. SOXB1 proteins are composed of a short N-terminal sequence followed by the HMG domain and a long C-terminal sequence that contains a transcriptional activation domain [42]. SOXB2 proteins share similar HGM domains and group-B homology sequences with SOXB1, however, SOXB2 proteins have a transcription repression domain in the C-terminal regions [43]. SOXC, SOXE, and SOXF proteins contain an HMG domain close to the N-terminus and an activation domain in the C-terminus, while the amino acid sequence outside the HMG domain varies among individual groups [37]. Some of the SOX factors can homo- or heterodimerize. For example, SOXD proteins are highly conserved in a group-specific coiled-coil domain that mediates dimerization with other SOXD proteins [44], while SOXE proteins encode a dimerization motif on the N-terminal upstream of the HMG domain which enables homo- or hetero-dimerization [45,46]. In spite of the conserved HMG domains, the diversity of the amino acid sequences outside the HMG domain allows SOX proteins to interact with different transcriptional cofactors and regulators to regulate distinct cellular activities and developmental processes [37,38]. Interestingly, many SOX proteins have pleiotropic functions and act in a context-dependent manner [37,38].

3. SOX factors in governing mammary stem and progenitor cell fates

Several SOX factors are expressed at significant levels in the mammary epithelium in lineage-specific or developmental stage-specific fashions. Two SOXE factors, SOX9 and SOX10 are expressed in both fetal mammary stem cells (fMaSCs) and postnatal unipotent progenitors [18,32,47–49]. SOX9 is expressed in fMaSCs, ER⁻ luminal progenitors, and a small population of basal cells [18,22,32], while SOX10 is expressed in postnatal basal cells and ER⁻ luminal progenitors and at a much higher level in fMaSCs [26,32,47,48]. Two other SOX factors are expressed in the mammary epithelium in interesting patterns. SOX11 is uniquely expressed in early embryonic mammary epithelial cells but not in the postnatal mammary glands [33,50,51]. In addition, SOX4 is specifically expressed in postnatal basal cells [32].

Both SOX9 and SOX10 are important determinants for mammary stem and progenitor cells. SOX9 is required for multipotent gland-reconstituting activity. Its knockdown in primary mouse epithelial cells with shRNAs inhibits organoid formation in vitro and the ability to regenerate mammary ductal trees upon transplantation in cleared mammary fat pads [49]. Furthermore, ectopic co-expression of SOX9 and the EMT transcription factor Slug in mature luminal cells is able to confer these cells with the organoid-forming ability and long-term mammary gland reconstituting activity [49]. A subsequent mouse genetic study showed that conditional Sox9 knockout using MMTV-Cre results in delayed postnatal mammary gland ductal elongation and side-branching [52].

The role of SOX9 in mammary stem/progenitor cells was also investigated by in vivo lineage tracing. Using Sox9-CreER^T2 transgenic mice, these studies showed that SOX9⁺ cells contribute to the formation of luminal cells and a small proportion of basal cells during pubertal ductal tree development and adult mammary gland homeostasis [22,25]. Interestingly, clonal lineage-tracing analyses revealed that distinct postnatal SOX9⁺ cells generate basal and luminal cells separately [22,25]. Within the luminal cells, SOX9 is expressed in the ER⁻ cells, and the postnatal SOX9⁺ luminal cells specifically contribute to the generation of ER⁻ luminal cells but not ER⁺ cells in mammary ductal trees [22]. Lineage-tracing during pregnancy further showed that SOX9⁺ cells also serve as progenitors for ER⁻ alveolar cells during alveologenesis [22]. Consistent with its expression in luminal progenitors, Sox9 conditional knockout inhibits the organoid-forming ability of ER⁻ luminal progenitors and impairs alveologenesis during early pregnancy [53]. Furthermore, CRISPR knockout of Sox9 in an embryonic mammary progenitor cell line inhibits its response to lactogenic stimuli [54]. SOX9 is similarly expressed in human ER^– luminal progenitors and required for their clonogenic potential in 3-dimension (3D) Matrigel culture in vitro [53,55]. These studies demonstrate a functional requirement of SOX9 in ER⁻ luminal progenitors.

SOX10 plays an essential role in embryonic mammary stem cells and the development of the mammary ductal tree. Interestingly, SOX10 is highly expressed in fetal multipotent mammary stem cells (fMaSCs) compared to postnatal mammary progenitor cells, suggesting a unique role of SOX10 in fMaSCs [26,32,47,48]. SOX10 expression is upregulated by the FGF10 signal, which is important for embryonic mammary rudiment development [26]. SOX10-high embryonic and adult mammary cells are enriched in stem/progenitor cell activities as revealed by the 3D organoid assay [26]. Importantly, Sox10 deletion in fMaSCs inhibits their ability to form organoids and to reconstitute mammary ductal trees upon transplantation [26]. Furthermore, chromatin accessibility analysis using ATAC-seq identified that fMaSCs exhibit a unique open chromatin landscape that supports multi-lineage differentiation potential [32]. Interestingly, SOX10-binding motifs are the most enriched transcription factor motifs in the unique accessible regions (UARs) of fMaSCs, supporting SOX10 acts as a master regulator of the fMaSC state [32]. Genetics studies using Sox10 conditional knockout mice further showed that SOX10 is essential for both the prenatal and pubertal branching of the mammary epithelium, supporting a critical role of SOX10 in fMaSCs and postnatal progenitors [56].

The exact function of SOX11 and SOX4 in the mammary gland development in vivo remains to be determined although in vitro evidence suggests their potential role in regulating stem/progenitor activity. Overexpression of SOX11 in human breast epithelial cell line MCF710A increases myoepithelial colony and mammosphere formation [50]. Furthermore, as discussed in detail below, both SOX4 and SOX11 can promote epithelial-mesenchymal transition, a process associated with mammary cell state plasticity [57]. Studies in other developmental settings, such as chondrocyte and neural crest cell development, have shown the requirement of cooperation among SOX factors [58,59]. Given the co-expression of SOX factors in mammary stem and progenitor cells, it will be interesting to dissect the potential crosstalk of SOX factors the mammary gland development.

4. SOX factors in breast cancer

Cancer cell plasticity and intra-tumor heterogeneity pose major challenges for cancer therapy [57]. Accumulating evidence suggests that cancer cells co-opt evolutionally conserved mechanisms of cellular plasticity required for normal tissue repair. Stem and progenitor cells maintain tissue homeostasis and restore tissue integrity during tissue repair. In response to tissue injury and other stress conditions, these cells can alter cell fates to broaden their regenerative potential and lineage outputs [60]. In addition, committed cells can dedifferentiate to become facultative stem cells to facilitate tissue repair [61]. In normal tissue, cell plasticity is tightly controlled and only transiently activated as injury responses to enable efficient regeneration and tissue repair. These evolutionally conserved mechanisms of cellular plasticity are aberrantly and persistently activated in cancer to generate intra-tumoral heterogeneity, promote adaptation to stresses encountered during metastatic cascade and confer therapeutic resistance [57,60,62,63]. As cell fate determinants in the normal mammary gland, SOX factors have recently been shown to play an important role in regulating cell state plasticity and promoting tumor progression and metastasis in breast cancer.

4.1. Promoting lineage plasticity in cancer

Mammary epithelial cells exhibit high levels of plasticity for epigenetic reprogramming to multipotency [57]. Culturing basal cells on feeder cells or in 3-dimensional organoids spontaneously gives rise to cells with multipotent repopulating activities [64,65]. In addition, lineage-restricted basal cells can reacquire multipotency upon transplantation in cleared mammary fat pads [15]. Genetic ablation of luminal cells in the mammary gland elicits basal cells to acquire a multipotent cell state reminiscent of fMaSCs [66]. Interestingly, adult basal cells have an open-chromatin landscape that is permissible for multi-lineage gene expression, which is likely to render basal cells with the increased ability to reacquire multipotency [32]. Increased cell plasticity also occurs during the malignant transformation of mammary epithelial cells. Dedifferentiation of committed mammary cells has been shown as a common event during transformation induced by different oncogenic mutations, including activating mutation of Pik3ca or deletion of Trp53 and Brca1 [34–36]. These findings suggest lineage plasticity is a common mechanism for generating transformation-competent intermediate cells during breast carcinogenesis.

Lineage plasticity events observed in health and tumorigenic mammary glands are summarized in Table 2.

Table 2.

Lineage plasticity in healthy and tumorigenic mammary glands and potential involvement of SOX factors

Models	Phenotype of mixed-lineage cells	Involvement of SOX factors	Citations
fMaSC-derived organoids	K8+/K14+	High SOX10 expression	[20,26]
Ablation of luminal cells using K8rtTA/TetO-DTA	CD29highEpCamhigh, co-expression of basal and luminal signatures	Not known	[66]
Activation of Pik3caH1047R in basal or luminal cells using K5 or K8 Cre drivers	K5+/K8+	Not known	[36]
Activation of Pik3caH1047R in basal or luminal cells using Lgr5 or K8 Cre drivers	K8/18+/K14+, acquisition of mammary-repopulating ability	Not known	[34]
Trp53 and Brca1 deletion in luminal cells with Ad-K8-Cre intraductal injection	K5+/K8+, EMT	Increased SOX9 and SOX4 expression	[35]
Early lesions in C3/Tag mice	K8+/K14+	Dependent on SOX9 upregulation	[53]
Pten and Trp53 deletion in luminal cells with Elf5 Cre driver	EMT	Not known	[67]

SOX factors have been shown to play a role in promoting lineage plasticity driven by cancer mutations. In the C3/Tag BLBC mouse tumor model, inactivation of p53 and Rb results in SOX9 upregulation in luminal cells. These SOX9^high luminal cells express both luminal and basal cytokeratins and acquired transcriptomic and open-chromatin signatures associated with basal cells [53]. Similar mixed-lineage cells have been observed in fetal mammary glands, BRCA1-deficient mammary tumors as well as PI3KCA^H1047R-induced mammary tumors [34–36,48]. Importantly, the conditional knockout of Sox9 in the C3/Tag mouse model inhibits the induction of such mixed-lineage cells, suggesting that SOX9 upregulation is required for lineage plasticity during BLBC formation [53]. This is accompanied by the inhibition of the progression from benign lesions to invasive tumors, suggesting this hybrid cell state is important for BLBC tumorigenesis [53].

Epigenetic and transcriptomic profiling of mammary cell subpopulations and tumor cells has also revealed that SOX10 acts as a key regulator for plasticity and multipotentiality in breast cancer cells [32]. Interestingly, in both PyMT and C3/Tag mammary tumors, cancer cells express heterogeneous levels of SOX10. SOX10^high tumor cells exhibit the transcriptomic and open-chromatin features associated with fMaSCs and luminal progenitors, whereas the SOX10^low tumor cells show features of mature luminal cells [32]. ChIP-seq further showed that SOX10 chromatin occupancy in tumor cells is significantly enriched in the fMaSC and luminal progenitor unique-accessible regions, highlighting the developmental origin of cancer cell plasticity and multipotency [32].

4.2. Enhancing EMT, tumor progression and metastasis

Epithelial-mesenchymal transition (EMT), a form of cellular plasticity, is a developmental program that enables polarized epithelial cells to acquire a highly invasive mesenchymal phenotype [68,69]. In cancer, EMT contributes to tumor progression, metastasis, intra-tumoral heterogeneity, and therapeutic resistance [70,71]. In the past years, several SOX family members have been shown to be involved in the EMT program to promote tumor invasion and metastasis.

SOXE factors promote tumor progression and metastasis by modulating EMT-related cell plasticity. Both SOX10 and SOX9 are important for neural crest induction, a developmental process generating highly mesenchymal and motile neural crest cells (NCC) [59]. Overexpression of these factors in tumor cells can activate this EMT/NCC-related developmental program. Interestingly, SOX10 overexpression in primary mammary organoids leads to the upregulation of genes associated with EMT and induces NCC-like invasive properties [26]. In mouse mammary tumors, SOX10^high tumor cells exhibit dedifferentiated and mesenchymal-like phenotypes with low expression levels of epithelial cytokeratins [32]. These tumor cells show enriched NCC gene expression and epigenetic signatures [32]. In human breast cancer, high SOX10 expression is correlated with NCC-related gene signatures [32]. SOX9 expression is also required for the progression of benign hyperplastic lesions to invasive tumors in the C3/Tag BLBC tumor models [53]. Furthermore, cooperating with the EMT-inducing transcription factor Slug, SOX9 overexpression greatly increases the lung metastasis-forming ability in poorly metastatic MCF7ras cells [49]. Both Sox9 and Sox10 are overexpressed in the BLBC or TNBC and are associated with shorter overall survival [26,32,53,55,72,73]. Given the overlapping role of SOX9 and SOX10 in mammary stem/progenitor and NCCs and their ability to heterodimerize [59,74], these two SOX factors may cooperate to drive tumor progression and metastasis.

SOX11 also plays an important role in tumor progression and metastasis. Normally only expressed in the embryonic mammary epithelium and silenced postnatally, SOX11 expression is reactivated in breast cancer, especially in BLBC [33,75]. High SOX11 expression correlates with poor survival in breast cancer patients [75]. SOX11 overexpression in the ductal carcinoma in situ cell model DCIS.com increases tumor cell invasion in 3D spheroid culture and promotes tumor growth and the formation of invasive tumors in vivo [50]. SOX11 knockdown in aggressive MDA-MB-231 and MDA-MB-468 cells inhibits tumor cell invasion [75]. Furthermore, inducible expression of SOX11 in DCIS.com cells leads to partial EMT to generate cells with hybrid epithelial/mesenchymal features and increases metastasis to bone and brain after tail-vein injection [76].

Another SOX factor that regulates EMT is SOX4. Overexpression of SOX4 induces mesenchymal traits along with increased cell migration and invasion in human MCF10A cells [77], while SOX4 silencing in TGF-β-treated mouse NMuMG cells and Py2T cells reverts these mesenchymal cells to an epithelial phenotype [78]. In addition, TGF-β-mediated upregulation of SOX4 is essential for the induction of the mesenchymal phenotype during EMT in human mammary epithelial cells HMLE [79], indicating that SOX4 is required for TGF-β-induced EMT in both mouse and human cells (Figure 1). Through activating the expression of the known EMT inducers, including Snail and ZEB1, SOX4 indirectly leads to the downregulation of E-cadherin, the loss of which is considered as a hallmark of EMT [70,71,77]. Interestingly, SOX4 also regulates EMT through epigenetic reprogramming mediated by histone methyltransferase EZH2 (Figure 1) [78], which has been reported to promote the expansion of breast cancer stem cells [80]. EZH2 increases H3K27me3 modification on key EMT genes and promotes primary tumor growth and metastasis to the lymph node, lung, and liver [78].

Figure 1. The action of SOX factors in breast cancer.

SOX factors can regulate multiple aspects of breast cancer initiation, progression, and metastasis. Through epigenetic reprogramming and activating oncogenic pathways such as Wnt/β-catenin signaling, TGF-β signaling, and NF-κB signaling pathways, SOX factors regulate cell proliferation, EMT, tumorigenesis, and metastasis in breast cancer. SOX factors play an important role in lineage plasticity, which is associated with mammary cell state reprogramming and tumor progression. Furthermore, SOX factors in breast cancer cells can regulate the infiltration of tumor-associated macrophages and enable latency competent breast cancer cells the ability to evade NK-cell mediated killing.

Lineage plasticity and EMT are closely related to the acquisition of stem-like properties and therapeutic resistance [71,81]. Anti-estrogen endocrine therapy is an effective treatment strategy for ER⁺ breast cancer. However, the development of endocrine therapy resistance is a major clinical issue, especially in metastatic diseases. Several SOX factors have been implicated in endocrine resistance. Tamoxifen-resistant MCF7TamR and T47DTamR breast cancer cells show upregulation of SOX9, which could be a consequence of aberrant ER chromatin binding [82] or regulated by HDAC5-mediated deacetylation and nuclear localization [83]. Furthermore, SOX9 overexpression is sufficient to increase endocrine resistance to estrogen deprivation or tamoxifen treatment in MCF7 cells [82,83]. In addition, the downregulation of SOX11 in MCF7TamR cells reverses EMT and reduces tamoxifen resistance, which can be partially restored by Slug [84]. SOX2 is also enriched in MCF7-TamR cells, and downregulation of SOX2 in tamoxifen-resistant BT-474 cells and T47DTamR cells elevated their sensitivity to tamoxifen [85]. Interestingly, SOX2 may act through upregulating SOX9 to promote tamoxifen resistance [55].

4.3. Regulating oncogenic signaling pathways

How SOX factors mechanistically promote cell fate plasticity in cancer is not completely understood. Their regulation of several key oncogenic signaling pathways may contribute to the increased plasticity and aggressive cancer traits. One prominent pathway controlled by SOX factors is Wnt signaling (Figure 1). The Wnt pathway regulates the development of the mammary gland and maintains stem/progenitor cells in the breast [86]. Aberrant activation of the Wnt/β-catenin pathway has been associated with metastasis and poor clinical outcomes in TNBCs [87,88]. Patients with dysregulated Wnt signaling have an increased risk for brain and lung metastases [87]. Co-immunoprecipitation experiments showed the interaction of SOX2 with β-catenin, one of the essential transducers of the Wnt signaling [89,90]. SOX2 and β-catenin can act synergistically to activate the transcription of Cyclin D1 in MCF7 cells [90]. This further facilitates the G₁/S transition, promotes anchorage-independent soft-agar colony formation and tumor growth in a xenograft model [90]. SOX9 also regulates the Wnt signaling in breast cancer. In TNBC cell lines MCF10A-DCIS and HCC1937, SOX9 silencing reduces the Wnt/β-catenin activity through downregulating the transcription of LRP6 and TCF4, two Wnt/β-catenin pathway components [91]. Furthermore, a Sox2-Sox9 axis has also been shown to activate the Wnt signaling and upregulate Wnt target genes AXIN2 and FZD4 in tamoxifen-resistant MCF7TamR and MDA-MB-231 cells [55].

SOX factors also regulate TGF-β signaling, which has a well-established role in breast cancer progression [92,93]. SOX4 can be induced by TGF-β in MCF10A cells and acts in a feedback loop to activate TGF-β signaling to induce EMT [77]. Another important signaling pathway regulated by SOX factors is NF-κB (Figure 1). Both canonical and non-canonical NF-κB signaling is required for BLBC tumorigenesis [94–96]. Interestingly, SOX9 enhances canonical and noncanonical NF-κB activity by upregulating NF-κB transcription factors in ER⁻ luminal cells [53]. This, together with elevated RANK receptor expression in ER⁻ luminal cells [94], is likely to predispose these cells to be hyper-responsive to NF-κB stimuli, therefore making them susceptible to transformation.

4.4. Shaping the tumor immune microenvironment

The tumor immune microenvironment is a critical determinant of tumor initiation and progression. Deciphering the complexity and dynamic interactions among cancer cells and immune cells within the tumor microenvironment will aid in exploiting new therapeutic targets [97]. SOX factors have an emerging role in regulating the interaction of tumor cells and tumor immune microenvironment. Via paracrine signals, immune cells can influence SOX factor expression in cancer cells and, conversely, SOX factors in cancer cells can regulate the infiltration of immune cells (Figure 1). For example, tumor-associated macrophages (TAMs)-derived CXCL1 activates SOX4 expression in MDA-MB-231 cells and MCF7 cells, which can in turn promote cell migration and invasion in vitro [98]. SOX2 has also been reported to promote the recruitment of TAMs via modulating the expression of chemokines including MIP-1α and ICAM-1 in 4T1 cells [99]. The inoculation of 4T1 cells with mouse peritoneal M2 TAMs reverts the inhibition of tumor growth and lung metastasis caused by SOX2 silencing in xenograft models [99].

Certain SOX factors can promote immune evasion of cancer cells. Recent studies showed that SOX2 and SOX9 confer the so-called latency competent cancer cells (cancer cells capable of establishing metastatic relapse after lying dormant) the ability to evade NK cell killing, hence promoting metastatic relapse (Figure 1) [100,101]. They have shown that the latency competent cells (LCCs) isolated from breast and lung cancer persist as dormant disseminated tumor cells (DTCs) in mouse models [100]. By entering a quiescent state, LCCs downregulate NK cell ligand, thus evading NK-cell-mediated killing. LCCs express stem cell-associated transcription factors including SOX2 and SOX9. Importantly, SOX2 and SOX9 are required for maintaining the LCC phenotype [100]. These findings provide additional mechanisms through which SOX factors promote tumor metastasis by regulating tumor-immune cell interaction.

5. Conclusion

The past few years have witnessed significant progress in understanding the functions of SOX factors in mammary gland development and breast cancer. SOX factors are critical stem/progenitor cell fate regulators in the normal tissue. Oncogenic signaling and stimuli from the tumor microenvironment can upregulate the expression of SOX factors to promote cancer cell plasticity and aggressive traits of cancer cells. SOX factors can, in turn, shape the tumor immune microenvironment. Some of the mechanistic actions of SOX factors in breast cancer are depicted in Figure 1. Further elucidating the regulatory network of SOX factors in breast cancer will reveal potentially useful therapeutic strategies. SOX factors are often overexpressed in cancer, which is likely to provide a therapeutic window for targeting SOX functions. Interestingly, several SOX factors, including SOX9, 10 and 11, are upregulated in BLBC, suggesting targeting these factors could be useful in this cancer subtype that lacks targeted therapy. Future studies are needed for identifying cancer-relevant functions/pathways of SOX factors for developing approaches to target cancers while sparing normal stem/progenitor cells.