The FOS-family protein FRA-1 (Fos-Related-Antigen 1), named because of immunological cross-reactivity with c-FOS, is encoded by the FOSL1 (FOS-Like-1) gene on chr11q13. The 271-aa-long FRA-1 oncoprotein is one of the most frequently overexpressed FOS-family members in a large variety of solid tumors. The oncogenic roles and regulation of FOSL1/FRA-1 in tumorigenesis have been described in several recent exhaustive reviews [28,29,30,31].

Various transcriptional, posttranscriptional, and posttranslational mechanisms are implicated in FRA-1 accumulation in response to multiple oncogenic lesions. The RAF-MEK-ERK, IL6-Stat3, and Wnt-beta-catenin signaling pathways induce the FOSL1 transcription through multiple regulatory sites localized in both 5′ flanking and intronic enhancer elements. The mechanisms of c-MYC binding to the FOSL1 enhancer, triggering the chromatin changes associated with the sequential recruitment of multiple histone modifiers and readers [32,33], have been recently reviewed with other aspects of FOSL1 transcriptional and posttranscriptional regulation [28,29].

Recently, the c-MYC-dependent control of FOSL1 in breast cancer has been further characterized by studying the NRG-dependent control of c-MYC stability. The NRG-ERK1/2-FBXW7-c-MYC pathway is responsible for the ERK-induced recruitment of c-MYC to the FOSL1 promoter in TNBC. Being overexpressed in about 30% of HER2-negative breast cancers, the EGF family member Neuregulin/NRG1 is pathogenetically relevant in TNBC. NRG1 activates the MAPK pathway, and the ERK1/2-mediated phosphorylation of the ubiquitin ligase FBXW7 results in decreased polyubiquitylation and increased nuclear import of c-MYC, which induces FRA-1 accumulation and lung metastasis in vivo [34].

In addition to transcriptional induction, the cancer-associated FRA-1 overexpression depends on the downregulation of the oncosuppressor miRNAs, which target the FOSL1 transcript [28,29]. Remarkably, in breast cancer progression, the downregulation of miRNAs contributes to FRA-1 accumulation not only in neoplastic cells but also in tumor-associated cell types, such as TAMs (Tumor-Associated Macrophages). Relevant examples include miR-34 and miR-130a, which suppress breast cancer invasion and metastasis by targeting FRA-1 in cancer cells [35,36], while miR-19a-3p and miR-4516 affect cancer cell invasiveness by targeting FRA-1 in TAMs [37] and CAFs (Cancer-Associated Fibroblasts) [38].

Along with oncomiRs, the posttranscriptional control of FOSL1 involves RNA modifications and RNA-binding proteins. The cancer-associated epitranscriptome provides a new perspective in deciphering tumor progression. Dynamic RNA modifications involve the crosstalk among writers, erasers, and readers [39]. The most prevalent mRNA modification, N⁶-methyladenosine (m⁶A), affects the stability and translation of key oncogenic transcripts. The effect of the m⁶A modification on FOSL1 mRNA stability is suggested by various findings, including a recent analysis of the m⁶A methylome in response to triptolide in rheumatoid arthritis, in which FOSL1 mRNA methylation and abundance is strongly affected, along with IGF2BP3, encoding of the insulin-like growth factor 2 mRNA-binding protein 3 [40]. Interestingly, IGF2BP1/2/3 is a major m⁶A reader, mechanistically linking mRNA methylation and stability. These RBPs are critically involved in posttranscriptional control of stemness and neoplastic transformation by targeting thousands of m⁶A-modified mRNA transcripts [41]. A direct role in EMT has been proven for IGF2BP1 and IGF2BP3, implicated in the SNAI2/SLUG-dependent gain of mesenchymal features through the direct association of IGF2BP3 (IMP3) with the SNAI2/SLUG mRNA [42] or the LEF1-mediated transactivation of SNAI2/SLUG resulting by the IGF2BP1 association with the LEF1 mRNA [43].

Interestingly, one of the publicly available databases (RIP-chip GeneST from ENCODE, accessible through UCSC Genome Browser) shows that the FOSL1 mRNA is among the transcripts bound by IGF2BP.

Other RNA-binding proteins might cooperate with IGF2BP1/2/3 in posttranscriptional control of FOSL1. A likely candidate is represented by PTPB1 (RNA Polypyrimidine Tract-Binding Protein 1), recently implicated in the mechanism of FRA-1-mediated PD-L1 transcription, immune escape, and metastasis in response to the abnormal expression of HOXA11-AS1 lncRNA (long-noncoding RNA) in one subtype (hypopharyngeal carcinoma) of HNSCC (head and neck squamous cell carcinoma). In this system, the FOSL1 mRNA half-life is positively controlled by the direct interaction with PTPB1, which is enhanced by the HOXA11-AS1 lncRNA [44]. Interestingly, the paralogous factor PTPB3, which prevents the ZEB1 mRNA degradation by binding to its 3′UTRis, contributes to the EMT posttranscriptional regulation in breast cancer [45].

Multiple posttranslational modifications control the FRA-1 stability, DNA binding, and transactivating activity. The MEK/ERK/Rsk and PKC-theta pathways control the FRA-1 protein stability and DNA binding activity via a cascade of phosphorylation events on serine/threonine residues in response to numerous oncogenes, cytokines, and growth factors. As recently reviewed [29], in non-transformed cells, FRA-1 is an intrinsically unstable, short-lived protein, while in invasive cancer cells, the increased activity of the RAS/RAF/MEK/ERK pathway induces the phosphorylation of FRA-1 residues S252 and S265, which inhibit the C-terminal (DEST) domain implicated in FRA-1 proteasomal degradation [46]. In invasive breast cancer, PKC-theta contributes to both FRA-1 stabilization, via T223 and T230 phosphorylation, and transactivation activity, through the FRA-1 phosphoacceptor residues T217 and T227 [47,48]. Moreover, the RAS-independent Mixed-lineage-kinase 3 contribute to FRA-1/AP-1 accumulation by inducing the activity of both JNK and ERK, responsible for the phosphorylation-mediated stabilization of c-JUN and FRA-1, respectively [49].

FRA-1 overexpression has been implicated in cancer cell proliferation, survival, migration, invasion, and plasticity in a large variety of neoplastic diseases, including most solid tumors. FRA-1 oncogenic roles have been discussed in a number of relevant reviews, dealing with many aspects of FRA-1 in tumorigenesis and metastasis mechanisms, along with recently proposed strategies of FRA-1 therapeutic targeting [28,29,30,31,50,51,52,53,54,55].

Regarding EMT mechanisms, the FRA-1-mediated control of cell plasticity and cancer-associated embryonic signaling pathways have been addressed in a highly relevant review, dealing with the FRA-1 involvement in EMT regulatory networks [28].

Seminal evidence on the link between Fra-1 overexpression and the hallmarks of EMT was obtained in a mouse mammary cancer cell system. Comparison between the parental cell line and the metastatic derivative showed the mutually exclusive expression of Fra-1 and E-cadherin, which was supported by the analysis of multiple human adenocarcinoma cell lines. Moreover, FRA-1 ectopic expression could trigger cancer cell motility and invasion, along with the accumulation of various metastasis-associated proteins, such as HMGA1, S100A4, and components of the plasminogen activation system (uPA, uPAR, PAI-1) [56].

Complementarily, the pioneering expression profiling of a panel of (weakly vs. highly invasive) human breast cancer cell lines resulted in a 24-genes signature which highlighted the FRA-1 overexpression along with its tight association with vimentin upregulation in the phenotypically mesenchymal cell lines [57].

Subsequent functional studies were based on FRA-1 ectopic expression in the non-invasive ER-positive MCF7 cells vs. FRA-1 knockdown in the invasive MDA-MB-231 cells. Along with motility and invasion, FRA-1 drove the secretion of metalloproteases and VEGF, and FRA-1 overexpression/downregulation affected breast cancer cell proliferation and cyclin D1 expression [58].

First, in a study on the MAPK-dependent control of cell migration in multiple TNBC lines, the EMT-TF SLUG (SNAI2) was essential for in vitro invasion and lung metastasis and expressed in an AP-1-dependent manner, in response to the ERK-Fra-1/c-JUN axis [65].

In the NAMEC cell system, FRA-1 expression is essential for EMT, not only in spontaneously transitioning cell cultures but also in cell lines ectopically expressing the tamoxifen-inducible TWIST and SNAIL derivatives. FRA-1 accumulation results from direct transactivation by both EMT-TFs binding to two FOSL1 regions around the TSS and within the first intron. In addition, FRA-1 expression is required for both TWIST-induced expression of SNAIL and SNAIL-induced expression of TWIST. Therefore, FRA-1 is both a transcriptional target and a regulator of both TWIST and SNAIL. Importantly, FRA-1 was required for the TWIST- or SNAIL-mediated induction of ZEB1, ZEB2, and SLUG [19].

The mechanistic links between FRA-1 and ZEB1/2 have been further elucidated in subsequent analyses of the FRA-1 cistromes and transcriptomes in both human and mouse cell systems, in which the FRA-1-regulated transcriptional networks were dissected through ChIP-seq analyses combined with expression profiling of FRA-1 (and c-JUN) target genes [8,9,66].

The first direct evidence of Fra-1 as a transactivator of both Zeb1 and Zeb2 promoters was obtained in a mouse mammary epithelial cell (MMLE) system. In the fully polarized non-tumorigenic cell line EpH4, FRA-1 overexpression triggered the gain of mesenchymal features associated with in vivo tumorigenicity and formation of metastasis in orthotopic xenograft models. ChIP analyses and reporter assays showed the Fra-1 functional association with the Zeb1 first intron and the Zeb2 promoter region. Accordingly, the Zeb1 or Zeb2 knockdown restored the epithelial features in the Fra-1-overexpressing EpH4 cells [66].

According to the FRA-1 cistrome, investigated in the TNBC cell line BT549, most FRA-1 binding sites are shared with c-JUN, thus indicating a major role played by c-JUN/FRA-1 heterodimers in the FRA-1-dependent transcriptional control [8]. FRA-1 (along with c-JUN) accumulation is required for the induction of EMT in response to TNF-alpha through mechanisms involving the FRA-1-mediated transactivation and chromatin looping of the ZEB2 promoter region [68].

In addition to ZEB2 transcriptional induction, resulting in ZEB2 accumulation, the FRA-1/c-JUN dimers also control the ZEB2 activity, inhibiting the GATA-family transcriptional repressor TRPS1 via indirect mechanisms dependent on the FRA-1/AP-1-mediated induction of miR-221/222, as discussed below (Section 3.4).

In addition to the reciprocal transcriptional controls with the core EMT-TFs, FRA-1 participates in the breast cancer-associated EMT by further mechanisms, including several protein–protein interactions between FRA-1 and key TFs on specific subsets of regulatory targets.

The key interaction between FRA-1/AP-1 and YAP/TAZ complexes emerges from various lines of evidence. First, the frequent co-localization of the FRA-1 and c-JUN binding sites with the TEAD binding motif suggested the regulatory interaction between FRA-1/AP-1 and the Hippo pathway on target promoters [8]. The functional cooperation between AP-1 and the Hippo pathway was further highlighted by a study in which the TEAD coactivator YAP1 was identified among the genes able to rescue the effect of KRAS suppression in KRAS-transformed colorectal cancer cells

In TNBC cells, in-depth analyses of the Hippo downstream effectors showed that the YAP/TAZ-driven transcription program is critically involved in the control of cell proliferation. Genome-wide analyses of YAP, TAZ, and TEAD binding site distribution highlighted the co-occupancy with AP-1 on transcriptional enhancers. c-JUN was present on almost 80 percent of the YAP/TAZ/TEAD binding sites, and, in all the analyzed binding sites, FRA-1 was bound along with c-JUN, thus suggesting its role as a major c-JUN heterodimeric partner. In addition, proximity ligation and coimmunoprecipitation assays showed physical interactions between FRA-1, c-JUN, JUND, and TEAD1. Accordingly, YAP/TAZ/TEAD synergized with AP-1 in the control of oncogenic growth both in vitro and in orthotopic xenografts. Remarkably, the induction of mammospheres, reflecting the fraction of CSCs/TICs (Tumor-Initiating-Cells) in response to the ectopic expression of a constitutively active TAZ (TAZ-S89A) derivative in MCF10A cells, was antagonized by FRA-1 knockdown [71].

These findings indicate the functionally relevant cooperation between the chromatin-bound FRA-1/AP-1 and YAP/TAZ/TEAD nuclear factors through large numbers of composite regulatory elements in TNBC cells. Moreover, recent analyses of ZEB1 genomic binding sites in breast cancer show that ZEB1 binding sites extensively overlap with both AP-1 (33% ZEB1 overlap with c-JUN) and YAP binding sites. In addition, ZEB1 physically interacts with c-JUN, FRA-1, and YAP. The EMT transcriptional programs are epitomized by the transactivation of mesenchymal genes, associated with the repression of epithelial gene subsets, prototypically represented by CDH1 (encoding for E-cadherin). An important difference has emerged from the comparison of the genomic binding sites of the subsets of ZEB1-induced genes with respect to ZEB1-repressed genes. While the ZEB/YAP/AP-1 elements coincide with the positively regulated genomic elements, the ZEB1-only ChIP-seq peaks are associated with the negatively regulated genomic elements. Therefore, FRA-1, being the major c-JUN heterodimeric partner in TNBC cells, critically contributes to EMT in cooperation with YAP by selecting the target genes which are induced rather than repressed by ZEB1 [6].

In addition to the above-described core EMT-TFs, a variety of FRA-1-regulated genes, encoding for cytokines, extracellular proteases, and receptors, are involved in the EMT mechanisms in breast cancer.

EMT depends on multiple autocrine and paracrine interactions between cancer cells and non-neoplastic cell populations, such as CAFs, TAMs, mesenchymal stem cells (MSCs), and myeloid cells, recruited to the Tumor Microenvironment (TME). In this context, interleukin-6 (IL6) and TGF-beta likely represent the best-characterized mediators of EMT, implicated in the mechanisms triggering the mesenchymal transitions at the tumor–stroma interface [72].

FRA-1 was characterized as a transcriptional regulator of IL6 in a pioneering study on the mechanisms of AP-1-1 and NF-kB- mediated induction of the IL6 promoter in invasive BCCs [73]. Moreover, the PKC-theta-mediated FRA-1 phosphorylation, associated with the invasive front of mammary tumors, results in increased FRA-1 transcriptional activity leading to IL6 induction in BBCs [48].

Interestingly, FRA-1 controls the IL6 transcription not only in neoplastic cells but also in macrophages recruited by paracrine signals to the tumor microenvironment, in which the TAMs represent a major source of IL6. Conditioned media from BBCs (4T1) induce FRA-1 accumulation in macrophages (RAW264.7), in which FRA-1 transactivates the IL6 promoter [74,75]. In addition to its major role as an EMT inducer in breast adenocarcinoma and other epithelial tumors, IL-6 is critically involved in macrophage polarization. In breast tumors, the FRA-1-dependent secretion of IL-6 promotes the generation of M2d macrophages [74] through mechanisms negatively controlled by the miRNA-mediated downregulation of FRA-1 expression [37].

In the EpH4-Fra-1 murine cell system, TGF-beta secretion was increased in response to Fra-1-overexpression, and a TGFBR1 inhibitor partially restored the expression of the epithelial markers downregulated by the ectopic Fra-1 oncoprotein. Accordingly, Fra-1 directly binds and transactivates the Tgfb1 promoter [66].

In addition to the above-described EMT-TFs and cytokines, the FRA-1/AP-1 complexes drive the transcription of cellular components critically implicated in Extra-Cellular Matrix (ECM) degradation.

Multiple Matrix Metalloproteinases (MMPs), along with the serine protease urokinase-type Plasminogen Activator (uPA), encoded by PLAU, play major roles in the ECM degradation, responsible for the EMT-associated invasiveness of neoplastic cells. The urokinase-dependent extracellular proteolysis is finely controlled by interactions with other components of the plasminogen activation system (uPAR and PAI-1). Interestingly, the three cognate transcripts (encoded by PLAU, PLAUR, and SERPINE1) are upregulated in response to FRA-1 overexpression in a murine breast adenocarcinoma cell system [56].

The FRA-1-dependent transcriptional control of uPA expression in aggressive breast cancer is mediated by complex interactions between the PLAU enhancer and promoter regions [76]. Interestingly, the knockdown of PLAU along with MMP9, which is also FRA-1-regulated in breast cancer [58,77], restores the cell–cell adhesion and inhibits the expression of EMT-associated genes in BCCs [78].

Expression profiling of invasive breast tumors shows that PLAUR highly correlates with FOSL1. In a cohort (1093 patients) of the TCGA database [79], FOSL1 is among the top five genes coexpressed with PLAUR. While other EMT-related genes (e.g., SNAI1 and VIM) are highly coexpressed, ESR1 (Estrogen Receptor alpha) is one of the top anticorrelated genes, in agreement with the FOSL1-associated mesenchymal features of triple-negative (and basal-like) breast cancers.

FRA-1 controls the PLAUR promoter in response to the oncogenic signals [80]. Given the role of uPAR in EMT induction in the BCCs [81], PLAUR represents a functionally relevant FRA-1/AP-1 downstream effector. The functional cross-talk between FRA-1 and uPAR in driving cell polarization, motility, and invasiveness has been elucidated in human colon carcinoma cell lines [82]. In response to ERK-dependent signaling, FRA-1 inactivates β1-integrin and downregulates RHOA activity, thus enabling the activation of RAC by uPAR, necessary to form polarized lamellipodia extensions. The two ERK-dependent events do not act independently but rather in cooperation since the expression of PLAUR depends at least partly on the FRA-1 activity. Moreover, FRA-1 is posttranscriptionally regulated by the integrin-uPAR signaling pathway in BCCs. The vitronectin-induced activation of uPAR triggers the SRC-FAK-MEK-ERK2 pathway is responsible for FRA-1 phosphorylation and stabilization, thus contributing to breast cancer cell invasiveness [83].

Despite the apparent paradox concerning the role of PAI-1 as an inhibitor of the urokinase proteolytic activity, the expression of the cognate gene (SERPINE1) strongly correlates with breast cancer metastasis and, in combination with uPA, represents a validated prognostic biomarker [84]. SERPINE1 is a major TGF-beta-SMAD target gene in breast cancer, and the specific complexes formed by SMAD2/3 and FRA-1/AP-1 dimers are implicated in the TGF-beta-mediated induction of the SERPINE1 and other promoters, such as MMP10, in preneoplastic BCCs [85].

Therefore, the cooperation between FRA-1 and the TGF-beta pathway includes both the FRA-1/AP-1-mediated TGFB1 transcriptional induction [66] and the interaction between the SMAD2/3 transducers and FRA-1 on genomic target sites [85].

Along with CAFs, TAMs, immune cells, and mesenchymal stem cells, endothelial cells participate in the paracrine interactions in the context of the tumor microenvironment. Interestingly, the metastatic dissemination of TNBC cells depends on the crosstalk between cancer cells and endothelial cells. The tumor cells secrete the FRA-1-regulated PAI-1, which stimulates the expression of the chemokine CCL5 from endothelial cells. In their turn, CCL5 acts on TNBC cells through a paracrine mechanism to stimulate migration, invasion, and metastasis [86].

In addition to direct transactivation, FRA-1 induces the components of the plasminogen activation system components also by indirect pathways mediated by FRA-1-controlled factors. Interestingly, secretome analyses show that the extracellular accumulation of uPA, PAI-1, and uPAR tightly depends on the expression of the HMGA1 [87], which is transcriptionally controlled by the FRA-1/AP-1 [88].

Differing from the poorly druggable transcription factors, such as FRA-1, several members of the FRA-1/AP-1 transcriptome encode for cellular components, such as receptors, which can be pharmacologically targeted, to inhibit the FRA-1-driven EMT.

The tyrosine kinase receptor AXL is tightly coexpressed with FRA-1 in a variety of invasive cancer cell lines, including muscle-invasive bladder carcinoma cells, in which AXL was originally identified as an FRA-1 transcriptional target involved in the control of cancer cell motility [94].

AXL contributes to the invasiveness, metastatic dissemination stemness, and chemoresistance of TNBC cells, in which AXL is autocrinally activated by its own ligand (Gas6). Both the receptor and the ligand are often coexpressed in human breast cancer, in which AXL represents a strong negative prognostic factor and a downstream effector of EMT-TFs, as shown in preneoplastic cells ectopically expressing SLUG and SNAIL [95]. In the same cell system (MCF10A), AXL overexpression, in turn, activates a positive feedback loop mechanism by inducing both SLUG and SNAIL and regulating self-renewing of breast CSCs [96].

AXL and FOSL1 are connected by a therapeutically relevant positive feedback loop: in addition to being transcriptionally regulated by FRA-1, the AXL tyrosine kinase is an upstream regulator of FOSL1. AXL can be targeted by an antibody interfering with the natural ligand GAS6. In TNBC cell xenografts or patient-derived xenografts (PDXs), the therapeutic antibody inhibits the GAS6-mediated induction of cell migration and invasion and FRA-1 expression, along with the core EMT-TFs (ZEB1/2, TWIST, SNAIL and SLUG) and vimentin.

The downregulation of AXL using MP470 (Amuvatinib) reverts the EMT triggered by TGF-beta and TNF-alpha in preneoplastic cells along with decreasing the breast CSCs self-renewal and chemoresistance [96,97]. In addition to Amuvatinib, which also targets other RTKs (PDGFR, c-KIT, MET), more selective orally bioavailable AXL inhibitors have been recently characterized as promising therapeutic tools against metastatic breast cancer [98].

Other promising therapeutic targets have been revealed by functional studies on the FRA-1 transcriptome. High-throughput synthetic lethality screens, aimed at identifying drugs selectively killing the metastatic FRA-1–overexpressing, with but not the FRA-1-depleted MDA-MB-231 BCCs, resulted in the identification of an FRA-1 target gene (ADORA2B) encoding a pharmacologically tractable adenosine receptor. Among the adenosine receptor antagonists, the highly tolerable bronchodilator theophylline strongly synergized with docetaxel in inhibiting the metastatic activity of BCCs [7].

Although the ADORA2B knockdown recapitulated the inhibitory effect of theophylline on metastatic dissemination, other mechanisms should be considered. In addition to antagonizing the adenosine receptors, the theophylline-mediated inhibition of phosphodiesterase induces cAMP-PKA signaling. This pathway is known to induce MET (Mesenchymal-to-Epithelial Transition) in NAMEC and inhibit the TICs fraction in the Ras-transformed derivative [99].

Since theophylline strongly inhibited the tumor-initiating ability of NAMEC-Ras cells [97], and given the synergism between theophylline and docetaxel in MDA-MB-231 cells [7], it will be important to investigate the possible ADORA2B-independent mechanisms of restoration of epithelial features. For example, as shown for other PKA agonists, such as forskolin, which improves the sensitivity to doxorubicin by inhibiting the ERK activity in MDA-MB-231 cells [100], the reversion of EMT in the same cell system might be consequent to the theophylline-induced downregulation of ERK activity and FRA-1 expression.

Analysis of the FRA-1/Fra-1 transcriptomes in both human and mouse cell systems indicates the functional relevance of several genes negatively controlled by FRA-1/Fra-1. In the mouse Eph4 cells, the gene sets and pathways implicated in cell junction organization and tight junctions were downregulated in response to the ectopically expressed Fra-1 [66]. In a human cell system (BT549), twenty proliferation-repressive genes were upregulated in response to the c-JUN/FRA-1 combined knockdown. In addition to well-known EMT inhibitors, such as CDH1 (encoding E-cadherin), the downregulated subset also included genes with tumor suppressor activity and poorly characterized roles in EMT. Among the FRA-1-downregulated genes, CLCA2, encoding for the Chloride Channel Accessory 2 protein, is targeted by FRA-1 and c-JUN, interacting with the CLCA2 third intron. CLCA2 was previously characterized as a p53-induced inhibitor of cell proliferation [101] and, more recently, as a prognostically relevant inhibitor of EMT in TNBC cells. It was proposed that by inhibiting the chloride current, the CLCA2 downregulation contributes to the increased pHi and metabolic changes associated with cancer cell invasiveness [102]. According to more recent findings, CLCA2 is involved in the maintenance of the junctional anchoring and epithelial state by colocalizing at the cell-cell junctions and interacting with EVA1 and ZO-1. Moreover, in the membrane, CLCA2 colocalizes with E-cadherin and interacts with beta-catenin regulating homophilic cell–cell interactions while inhibiting the beta-catenin cytosolic signaling and downregulating the EMT-inducing beta-catenin target genes [103].

The FRA-1-downregulated genes also include indirect targets, posttranscriptionally controlled by negative regulators induced by FRA-1 the gene products. The miRNA-mediated regulatory circuits are highly relevant in EMT, as shown by the double-negative feedback loops formed by the EMT suppressor miRNAs (miR-200, miR-34, and miR-15/16 family members) and the core EMT-TFs. In BCCs, FRA-1 downregulates the transcriptional repressor TRPS1 through a miRNA-mediated mechanism [69]. TRPS1, which anticorrelates with FOSL1 in aggressive breast carcinoma (TCGA expression Atlas [79]), encodes for a GATA-type zinc-finger transcription repressor playing cell context-dependent roles. In mammary gland development, TRPS1 is essential for proliferation and lactogenic differentiation. In breast cancer, the loss of TRPS1 along with loss/inactivation of E-cadherin results in increased proliferation of mammary organoids and accelerated tumorigenesis in mouse models [104]. The FRA-1 effect on TRPS1 is mediated by the miR-221/222 oncomiRs, which are overexpressed in basal-like breast cancers. In response to the RTK/RAS/RAF/ERK pathway, FRA-1 binding to the miR-221/222 promoter region induces the expression of both miRNAs, which, in turn, target the TRPS1 transcript and downregulate the protein product. Among the downstream genes subjected to TRPS1-mediated repression, ZEB2 is strongly implicated in the effect of TRPS1 downregulation on the EMT induction [69]. Therefore, the FRA-1-miR-221/222-TRPS1-ZEB2 axis further reinforces the regulatory links between Fra-1 and ZEB2 in breast cancer.

Other oncomiRs are likely implicated in the FRA-1-mediated EMT mechanisms. The well-characterized onco-miRNA miR-21 is overexpressed in almost every cancer cell type. Differently from miR-221/222, miR-21 expression does not correlate with specific subtypes, such as the basal-like. However, the miR-21-mediated inhibition of the anti-metastatic gene (LZTFL1) is required for the in vitro invasiveness and in vivo metastasis along with the EMT markers in BCCs [105]. Since FRA-1 is a transcriptional regulator of miR-21, at least in some cell contexts, miR-21 might participate in the FRA-1-mediated EMT induction.

The characterization of FRA-1-binding genomic elements has highlighted several mechanisms of FRA1-dependent chromatin modifications and transactivation of target genes.

Dissection of the PLAU regulatory region shows that FRA-1 binding to the proximal (−1.9 kb) enhancer mediates the recruitment of the p300 HAT, which is associated with the transcription of multiple RNA species along with the bona fide PLAU mRNA precursor. These short unstable RNAs, transcribed bidirectionally across the enhancer regions upstream to the PLAU TSS, likely act as enhancer RNA molecules (eRNAs), regulating the epigenetic state of the chromatin [106]. In addition, a relatively stable transcript, extending from the −1.9 kb enhancer encompassing the PLAU mRNA from the TSS to the 3′UTR, is expressed in multiple basal-like cancer cell lines, although at low level with respect to the productive PLAU mRNA. Interestingly, FRA-1 positively controls the PLAU mRNA but inhibits the accumulation of the longer transcript initiating in the 5′ flanking region [76]. The functional significance of these findings awaits further investigations.

While the PLAU transcription mainly depends on the tracking by RNA polymerase of the region upstream to the PLAU locus, the ZEB2 transcriptional regulation points to the role of FRA-1 in the control of long-range chromatin interactions. The TNF-alpha-mediated induction of EMT in TNBC cells (BT549) is associated with the expression of two alternatively spliced ZEB2 mRNA isoforms transcribed from two distinct promoters located 2.4kb apart. FRA-1/c-JUN dimers are recruited on the distal promoter and mediate the looping of both promoters, which is essential for ZEB2 transcriptional induction.

The FOS family member FRA-2 is often co-expressed with FRA-1 in TNBC cells, and early findings suggested that FRA-2 could cooperate with FRA-1 in the control of breast cancer cell motility and invasiveness [107]. These findings raise the question of FRA-2’s contribution to the regulation of FRA-1 target genes. Interestingly, the AP-1-mediated control of HMGA1 transcription shows that, although both FRA-1 and FRA-2 bind to the last two introns of the gene, only FRA-1 is required for HMGA1 transcription. Chromatin conformation analyses show that the distal FRA-1 (and FRA-2) binding region interacts with the HMGA1 promoter. Surprisingly, however, at variance with ZEB2, FRA-1 is not required for the DNA looping. Moreover, as in the case of PLAU, FRA-1 expression is required for the recruitment of p300 to the enhancer region, but quite surprisingly, p300 (as well as CBP) is not required for HMGA1 transcription in MDA-MB-231 cells. In summary, the FRA-1-independent interaction between the enhancer and promoter regions allows the enhancer-bound FRA-1 to drive the recruitment of RNA polymerase (without affecting the Pol II CTD P-Ser5/P-Ser2 ratio) on the HMGA1 promoter [88]. Altogether, these studies suggest that FRA-1 controls its targets by gene-specific mechanisms.

A general picture has emerged from the in-depth analysis of FRA-1 and FRA-2 transcriptomes and genomic distribution, with the associated epigenomic modifications and long-range interactions. In agreement with its biological functions, FRA-1 regulates many more genes and exerts stronger transcriptional effects with respect to FRA-2. Genomic distribution, along with associated histone modifications, RNA Polymerase, and CBP/p300 recruitment and chromatin accessibility, shows that FRA-1 and FRA-2 prevalently bind to enhancers rather than promoter regions. FRA-1 controls the recruitment of CBP/p300 both positively and negatively, depending on the target. Moreover, as observed for HMGA1, FRA-1 exerts limited effects on the chromatin structure (DNA looping) of selected target genes, thus raising the question of the role of Fra-1 interactions with other transcription factors and coactivators within the context of 3D enhancer hubs [9].

Other lines of research stemming from the mass spectrometry-based identification of chromatin-bound FRA-1 interaction partners will contribute to shedding light on the mechanisms of FRA-1-mediated transactivation regarding EMT and therapeutic applications. The RNA helicase DDX (DEAD-box)5/p68, identified as the most enriched FRA-1-binding protein among 118 interactors in the chromatin of TNBC cells (BT549), shares with FRA-1 the majority (62%) of its genomic binding sites. DDX5 is essential for the transcriptional activation of FRA-1 target genes, and both positively and negatively regulated DDX5-dependent gene sets exhibit prognostic value in breast cancer patients. Accordingly, DDX5 is overexpressed in basal-like tumors, and DDX5 levels are predictive of worse outcomes in ER- but not ER+ breast cancers [108].

Recent findings, originating from seminal work in colorectal cancer cells [109], show that DDX5 promotes EMT by inducing the expression of the PDGF receptor in TNBC cells [110]. Therefore, given the role of the PDGF autocrine signaling in the maintenance of the FRA-1-driven EMT and cancer cell stemness [19], DDX5 might collaborate with FRA-1 by both direct mechanisms, mediated by the DDX5-FRA-1 interaction on chromatin [108], and indirect mechanisms, by sustaining the expression of PDGFR in TNBC cells [110]. Therefore, in addition to being posttranscriptionally regulated by RBPs (IGF2BP1) interacting with the FOSL1 transcript, FRA-1 also cooperates with other RBPs, such as the RNA helicase DDX5, in invasive breast cancer.

The same proteomics analysis [108] points to PARP1 as a therapeutically promising FRA-1 interactor. The interaction with PARP1 results in FRA-1 PARylation (Poly-ADP-Ribosylation), which can be inhibited by treatment with olaparib. The PARP inhibitor upregulates the FRA-1 protein and mRNA expression levels by mechanisms likely dependent on the AP-1-mediated FOSL1 transcriptional autoregulation. On the other hand, FRA-1 knockdown potentiates the proapoptotic effects of olaparib in TNBC cells. Transcriptomic analyses showed that a large fraction of the olaparib-regulated genes is induced in an AP-1-dependent manner. Remarkably, the EMT signaling pathway is one of the most enriched pathways in response to olaparib treatment, in agreement with the evidence that many of the olaparib-induced genes are direct FRA-1 targets [111].

The clinicopathological relevance of FRA-1 overexpression was originally suggested by immunohistochemical analyses of breast cancer cell lines and tumors, which showed the prognostically significant inverse correlation between the expression of FRA-1, accumulating in less differentiated cancers, and FOSB, expressed in normal mammary glands and well-differentiated tumors [113]. Subsequent analyses of breast cancer-associated AP-1 compositional changes showed the strong inverse correlation of FRA-1 vs. ER and PR expression, along with the FRA-1 accumulation in triple-negative compared to luminal carcinomas [114].

At the mRNA level, analyses of microarray data sets representing large cohorts of breast cancer patients showed that FRA-1 mRNA level inversely correlated with DMFS (Distant Metastasis-Free Survival), while c-FOS expression exhibited the opposite correlation [66].

The clinicopathological correlations were further reinforced by analyzing the FRA-1 transcriptome, representing a surrogate readout of FRA-1 activity in highly metastatic lines derived from the phenotypically mesenchymal MDA-MB-231 cells. A prognostically relevant subset of FRA-1 targets, further restricted to a 183-genes Fra-1 classifier, was defined by bioinformatic analysis. Remarkably, the FRA-1 classifier performed better than other available prognostic signatures in predicting the outcome (time to distant metastases or relapse) in TNBC patients [7]. The predictive value of the FRA-1 classifier was further delineated through a functional approach. A nine-genes subset was generated by investigating the effects on primary and/or metastatic growth of inhibition of 31 prognostically relevant FRA-1-regulated genes. Interestingly, within this subset, EZH2 (Enhancer of Zeste Homolog 2, the enzymatic component of the Polycomb Repressive Complex 2 PRC2 catalyzing the H3K27 trimethylation) represents a further functional axis linking FRA-1, lncRNAs, and EMT-TFs. An important link between EZH2 and EMT is represented by the SNAIL-mediated recruitment of EZH2 to specific genomic sites via interactions through the lncRNA HOTAIR (for HOX Transcript Antisense Intergenic RNA) during TGF-beta-induced EMT in human hepatocytes [115].

More recently, a four-gene prognostic signature predictive of poor overall survival has been characterized in basal B BCCs, in which constitutive FRA-1 phosphorylation is sustained by the FAK-SRC-MEK-ERK pathway in response to the integrin uPAR signaling. Interestingly, the signature includes the uPAR-coding gene, PLAUR, along with two uPAR ligands (PLAU and VTN) and FOSL1 itself [83].

Finally, powerful prognostic correlations originate from the recent analyses of cooperation mechanisms between FOSL1, ZEB1, and YAP in breast cancer. FOSL1 is highly coexpressed with ZEB1, and the prognostic value (relapse-free survival) of FOSL1+ ZEB1 expression is higher than that of ZEB1 alone. Accordingly, the high expression of a subset of eight common ZEB1/YAP/AP-1 target genes correlates with lower relapse-free and DMFS (Distant Metastasis-Free Survival) rates [6].