Cancer is the second leading cause of death worldwide (Wang et al., 2016). Nevertheless, the basis of poor prognosis for cancer patients and the obstacle to a positive clinical outcome is not the primary tumor itself, but cancer cell plasticity, which enables local invasion, dissemination, and distant metastases. Evidence has accumulated that plasticity is driven by the process called epithelial-mesenchymal transition (EMT). Activation of EMT is widely believed to contribute to invasion, metastasis, tumor relapse, and therapy resistance (Zheng et al., 2015). In many epithelial malignancies, EMT is associated with a change in phenotypic features such as loss of cell-cell adhesion and polarity, change from a cobblestone-like shape to an elongated one, and development of a generally more aggressive mesenchymal-like phenotype (Kalluri and Weinberg, 2009). EMT originally observed during embryogenesis (Hay, 1995), is a reversible, evolutionary conserved process that is tightly regulated through the interplay between environmental signals from Wnt, TGF, FGF family members, interleukins, and various EMT-transcription factors (EMT-TFs), including Zinc-finger E-box binding protein 1 (ZEB1), ZEB2, Snail, Slug, and Twist (Kalluri and Weinberg, 2009). All of these processes are fine-tuned by oncogenic and tumor-suppressive microRNAs (miRNA). ZEB1 is a core EMT-TF of the ZEB family and is implicated in cellular plasticity, dissemination, and a dormant-to-proliferative phenotypic switch at the distant site as well as being a determinant of worse clinical prognosis in most human cancers (Zhang et al., 2015; Katsura et al., 2017; Krebs et al., 2017). Furthermore, activation and stabilization of ZEB1 via a miRNA- or ATM-dependent axis contribute to the resistance to various anticancer therapies (Burk et al., 2008; Zhang et al., 2014a). With regards to clinical relevance, ZEB1 expression increases progressively through the different stages of cancer progression, e.g., ZEB1 expression dramatically increases in advanced castration-resistant prostate cancer (CRPC) and PCa metastasis compared to clinically localized prostate cancer (Figiel et al., 2017). Moreover, patients with ZEB1-expressing metastases have shorter overall survival compared to patients with ZEB1-negative tumors (Figiel et al., 2017). In this review, we outline recent studies on the molecular function of ZEB1 in cellular plasticity and metastasis and elucidate its role in DDR and therapy resistance in an EMT-dependent or EMT-independent manner.

ZEB1 protein (also known as TCF8), encoded by the ZEB1 gene in humans, is a transcription factor characterized by the presence of two C2H2-type flanking zinc finger clusters, which are responsible for interaction with paired CACCT(G) E-box-like promoter elements on DNA, and a centrally located POU-like homeodomain, not binding DNA (Vandewalle et al., 2009). Additionally, ZEB1 contains Smad- (SID), CtBP- (CID), and p300-P/CAF (CBD) interaction domains that are instrumental in the control of its transcriptional activity (Vandewalle et al., 2009; Zhang Y. et al., 2019). ZEB1 can either downregulate or upregulate the expression of its target genes by epigenetic mechanisms, including DNA methylation or histone modifications and recruitment of different co-suppressors or co-activators through SID, CID, or CBD (Postigo et al., 2003). For instance, ZEB1 can activate transcription of TGF-beta responsive genes through its interaction with co-activators such as Smad, p300, and P/CAF (Postigo et al., 2003; Caramel et al., 2018). Conversely, recruitment of CtBP transcriptional co-repressors (histone deacetylases HDAC1/2) following direct ZEB1 binding onto the CDH1 gene promoter leads to repression of CDH1 transcription, resulting in downregulation of E-cadherin protein expression and induction of EMT (Zhang et al., 2015). This dual activity, which fosters the expression of genes encoding components for tight cell junctions, desmosomes or intermediate filaments, is unique for ZEB1/2 transcription factors and crucial for the EMT program (Caramel et al., 2018).

Regulation of ZEB1 expression can be accomplished on different levels by transcriptional or post-transcriptional mechanisms. First, the feedback loop between ZEB1 and the miRNA-200 family is a well-described mechanism of the regulation of cellular plasticity, (de)differentiation, and EMT machinery (Tian et al., 2014; Zhang Y. et al., 2019). Second, ubiquitination by E3 ligase complex Skp1-Pam-Fbxo (Xu et al., 2015) or, conversely, deubiquitination by USP51 enzyme has also been shown to regulate ZEB1 and EMT (Zhou Z. et al., 2017). Expression of ZEB1 is under the control of different positive (TGF-beta, Wnt/beta-catenin, NF-κB, PI3K/Akt, Ras/Erk) as well as negative regulators, including miRNA signaling (Chua et al., 2007; Bullock et al., 2012; Horiguchi et al., 2012; Kahlert et al., 2012; Zhang and Ma, 2012; Zhang Y. et al., 2019). For instance, ZEB1 represents the direct downstream target of Wnt-activated beta-catenin in bone metastasis of lung cancer, resulting in decreased levels of E-cadherin and EMT (Yang et al., 2015). In parallel, TGF-beta induces the mesenchymal phenotype in glioblastoma cells via pSmad2- and ZEB1-dependent signaling, leading to tumor invasion (Joseph et al., 2014). Finally, Han et al. have reported that hepatocyte growth factor increases the invasive potential of prostate cancer cells via the ERK/MAPK-ZEB1 axis (Han et al., 2016). Besides well-known transcription factors, Grainyhead-like 2 (GRHL2) has been described as a potential key player associated with the epithelial phenotype and an important regulator of ZEB1 and EMT. Studies have shown that GRHL2 modulates the expression of E-cadherin and Claudin 4, which are crucial for differentiation and maintenance of cell junctions (Werth et al., 2010). In breast cancer, GRHL2 acts as an EMT suppressor by forming a double-negative feedback loop with the EMT driver ZEB1 via the miR-200 family (Cieply et al., 2012). Similarly, GRHL2 regulates epithelial plasticity along with stemness in pancreatic cancer progression by forming a mutual inhibitory loop with ZEB1 (Nishino et al., 2017). Whereas combined (over)expression of GRHL2 and miR-200s increases E-cadherin levels, inhibits ZEB1 expression and induces MET (Somarelli et al., 2016), GRHL2 knockdown is associated with downregulation of epithelial genes, upregulation of ZEB1 or vimentin, and the onset of EMT (Chung et al., 2019). Hence, the reciprocal repressive relationship between GRHL2 and ZEB1 is considered to be a significant regulator of EMT cell plasticity and chemoresistance (Chung et al., 2019). These regulatory mechanisms make ZEB1 the core downstream target of broad spectra of signaling pathways implicated in various cellular processes, including differentiation, proliferation, plasticity, and survival.

Enhanced plasticity of cancer cells is considered an important driving force of tumor progression, allowing continuous adaptations to the demanding conditions in the ever-changing tumor microenvironment. Cellular plasticity is exerted by a reciprocal feedback loop between the EMT driver ZEB1 and the miR-200 family as an inducer of epithelial differentiation (Burk et al., 2008; Gregory et al., 2008; Brabletz and Brabletz, 2010). Within this feedback loop, ZEB1 promotes EMT, plasticity, dissemination, and drug resistance via inhibition of the transcription of miR-200 family members, while miR-200 family members promote MET, differentiation, and drug sensitivity by inhibition of ZEB1 translation (Brabletz, 2012). Thus, this regulatory mechanism was proposed as a molecular “engine” of cellular plasticity and a driving force toward cancer metastasis (Brabletz and Brabletz, 2010). Mathematical modeling of this feedback loop suggests that cells need not necessarily attain just epithelial or mesenchymal states; rather, they can stably acquire a hybrid epithelial/mesenchymal phenotype (Lu et al., 2013). The coupled system of ZEB1, GRHL2, and miR-200 drives the cellular dynamics of epithelial-hybrid-mesenchymal transition (Jolly et al., 2016; Chung et al., 2019). ZEB1 forms two additional indirect feedback loops including epithelial splicing regulatory factor ESRP1 and an enzyme that produces hyaluronic acid, HAS2 (Preca et al., 2015, 2017; Jolly et al., 2018) Thus, ZEB1-mediated feedback loops function as a hub of cellular plasticity during metastasis.

From another point of view, it is increasingly evident that genomic regions that do not encode proteins and are often transcribed into long non-coding RNAs (lncRNAs) represent important regulators of cancer development, dissemination, and aggressiveness (Huarte, 2015). Moreover, lncRNAs can directly interact with proteins and thereby regulate their stability (Huarte, 2015). A recent study has reported a novel lncRNA, namely RP11-138 J23.1 (RP11), as a positive regulator of migration, invasion, and EMT in colorectal carcinoma cells in vitro and enhanced liver metastasis in vivo (Wu et al., 2019b). Mechanistically, epigenetic upregulation of RP11 (m6A modification) accelerates the degradation of two E3 ligases and thus attenuates proteasomal degradation of ZEB1, resulting in dissemination of CRC cells (Wu et al., 2019b).

Genome-wide screening of ZEB1 targets using TNBC cell line Hs578T revealed more than 2,000 genes that are positively or negatively regulated by this transcription factor. In the context of plasticity, ZEB1 contributed to the regulation of cell polarity via DLG2 and FAT3 proteins, cell-to-cell adhesion via transmembrane protein TENM2 or anchorage-independent growth through interaction with metalloproteinase inhibitor TIMP3 (Maturi et al., 2018). Moreover, strong evidence indicates that ZEB1, but not Snail or Slug, is the master regulator of phenotypic as well as metabolic plasticity of pancreatic cells, affecting cancer cell dissemination and metastasis (Krebs et al., 2017). Moreover, metastasis remains one of the main obstacles in cancer therapy. Hence, effective anti-ZEB1 immunotherapy might serve as a promising tool for the reduction of cancer cell dissemination and metastasis and thus could help to eradicate various types of cancer. Notably, ZEB1 depletion significantly reduces stemness and colonization capacity and locks the cells in the homogeneous epithelial state, limiting cell heterogeneity and plasticity (Krebs et al., 2017). However, the similarities and overlaps in molecular networks mediated by ZEB1 vs. other EMT-inducing transcription factors remain to be identified. Thus, ZEB1 may play context-specific roles in repressing epithelial genes and/or activating genes involved with a mesenchymal phenotype (Watanabe et al., 2019), given its ability to function both as a repressor and as an activator depending on available co-factors (Lehmann et al., 2016).

The detachment of tumor cells from the main tumor bulk and invasion through surrounding stroma is an important step for the development of distant metastasis. A growing body of evidence proves that the stroma plays a major role in the budding of quiescent tumor cells, resulting in dissemination. ZEB1 has been shown to be strongly associated with this complex process, wherein EMT-like stromal cells possessing high ZEB1 levels trigger the tumor-budding phenotype by tumor-stroma crosstalk (Galvan et al., 2015). Strikingly, ZEB1 also governs the inflammatory phenotype in breast cancer cells by regulating the secretion of pro-inflammatory cytokines IL-6 and IL-8 and induction of fibroblasts and growth of myeloid-derived suppressor cells, indicating its key role in the tumor microenvironment and formation of the pre-metastatic niche (Katsura et al., 2017; Carpenter et al., 2018). At the same time, it has been shown that in the post-dissemination events, inflammation orchestrates ZEB1-dependent escape of disseminated tumor cells from dormant to active phenotype and induces EMT-associated metastatic outgrowth, highlighting the importance of ZEB1 in the regulation of cell plasticity (De Cock et al., 2016). Functional studies revealed that ZEB1 overexpression drives melanoma phenotypic plasticity and is sufficient to drive resistance to BRAF and/or MEK inhibitors, whereas ZEB1 inhibition sensitizes naive melanoma cells to BRAF inhibitors, prevents the emergence of resistance, and decreases the viability of resistant cells (Richard et al., 2016). Finally, ZEB1-driven phenotypic plasticity of epithelial pancreatic cancer cells was also observed in vivo, where differentiated primary tumor cells underwent dedifferentiation associated with an upregulation of ZEB1 at the invasive front, resulting in liver metastasis (Krebs et al., 2017). Taken together, these findings reveal the crucial role of ZEB1 in the phenotypic plasticity important for the dissemination of cancer cells and the establishment of metastasis in distant sites.

Importantly, ZEB1-mediated mechanisms and feedback loops can also drive an irreversible EMT (Jia et al., 2019). However, the different impacts of reversible vs. irreversible EMT on associated traits such as therapy resistance, immune evasion, and tumor-initiating potential remain to be investigated. Breaking the ZEB1/miR-200 feedback loop has been shown to alter the dynamics of phenotypic plasticity in a cell population and curb metastasis in vivo, but the mechanisms involved here are still elusive (Celia-Terrassa et al., 2018).

Nowadays, a growing body of evidence implicates intratumoral heterogeneity, EMT, and increased ZEB1 levels as among the main drivers of therapy resistance, exemplified by EMT-induced docetaxel resistance in prostate cancer (Hanrahan et al., 2017), gemcitabine resistance in pancreatic cancer (Wang et al., 2017), and multiple types of resistance within various malignancies (Shibue and Weinberg, 2017; Cui et al., 2018; Zhang et al., 2018; Orellana-Serradell et al., 2019). Besides being a key contributor to the regulation of cancer cell differentiation and metastasis, the potential role of ZEB1 in the modulation of tumor chemoresistance is not yet fully understood.

A growing body of publications has considered ZEB1 in normal and cancer cells to be a crucial regulator of fundamental intracellular processes as well a major denominator of plasticity, driving drug adaptation and phenotypic resistance to various types of anticancer therapy. Given the core downstream target of highly conserved pathways implicated in response to DNA damage and repair, proliferation, plasticity, and cell differentiation, ZEB1 plays a pivotal role in the determination of cell fate (Figure 1). Considering its phosphorylation and stabilization by ATM kinase, leading to a limited response to different types of anticancer therapy, combined targeting of ZEB1 with the ATR-CHK1 axis might represent an effective way to overcome these obstacles. Promising results were also obtained with MDM2 inhibitors, which could reactivate p53 tumor suppressor along with downregulating ZEB1 and decreasing stemness features and cancer aggressiveness. An additional possibility for reducing the expression of ZEB1 is inhibition of non-coding circular RNA (circRNA) hsa_circ_0057481, as shown in laryngeal cancer (Fu et al., 2019). Further studies are necessary in order to test clinical applicability. One could also consider inhibition of ZEB1 and EMT by down-regulation of valproic acid, which regulates the ZEB1 promoter (Zhang S. et al., 2019). Multiple studies have also focused on BET inhibitors in cancer. In this context, it was observed that the DNA endonuclease Mus81, which regulates ZEB1, may be targeted by BET4 inhibitors (Yin et al., 2019). In general, miRNA-based therapeutic options targeting ZEB1 might represent promising tools for targeting ZEB1 but need to be further developed, and delivery methods and therapeutic agent stability should also be investigated with priority.

Despite the association of high ZEB1 expression, EMT, and chemoresistance described in many studies, the role of EMT by itself in therapy resistance is rather controversial. It is not necessarily the epithelial or mesenchymal state that dictates cancer stem-like properties such as drug resistance; instead, they depend on the functions and mechanisms of action of EMT regulators, including ZEB1. Moreover, underlying mechanisms should be investigated for individual context, as the roles of ZEB1 and other transcriptional factors are highly treatment- and cancer type-dependent. The mechanistic links between ZEB1 expression, plasticity, the emergence of CSCs and therapy resistance represent important areas for future investigation. A novel, more specific inhibitors and a better understanding of ZEB1-driven plasticity, inflammation, and vascularization within the tumor and/or pre-metastatic niche microenvironments are inevitably needed to more effectively control resistance to various types of therapies.

SD wrote a complete draft and first version of the manuscript. All authors contributed to the manuscript revision, read, and approved final version and contributed to the principle layout of the article.