Abstract
In order to identify new targets and treatment modalities for breast cancer, we searched the literature for circular RNAs (circRNAs) with efficacy in preclinical breast cancer-related in vivo models. From our search, we identified 26 up-regulated and six down-regulated circRNAs which mediate efficacy in breast cancer-related preclinical in vivo models. We discuss reconstitution and inhibition of the identified circRNAs, as well as druggability and validation of the targets identified in the context of chemoresistance, inhibition of proliferation and metastasis. Pathways driven by suppressors of cytokines and high-mobility group proteins, nuclear factor 
B and Hippo signaling emerged as important drivers of tumor growth and metastasis. The role of trefoil factor-1 with respect to metastasis of estrogen receptor-positive breast cancer also merits further investigation. In addition, mucin 19 has emerged as an unexplored target for treatment of breast cancer.
Breast cancer (BC) is the most common malignancy in women worldwide and occurs as ductal and lobular carcinomas (1). According to the expression of hormone receptors (HR) such as estrogen-receptor (ER), progesterone receptor and human epidermal growth factor receptor 2 (HER2), the following subtypes have been defined: Luminal A (HR+, HER2−), luminal B (HR+, HER2+), triple-negative BC (HR−, HER2−) and HER2 enriched (2). The treatment of BC is dependent on the molecular subtype and includes surgery, radiotherapy, chemotherapy and targeted therapy with agents such as tamoxifen, aromatase inhibitors, trastuzumab, pertuzumab, Kadcyla (trastuzumab-emtansine), lapatinib, poly-(ADP)-ribose polymerase inhibitors for BC with BRCA DNA repair-associated mutations, cyclin-dependent kinase (CDK) 4/6 inhibitors, tumor spread-inhibiting bisphosphonates and recently also immunotherapy with pembrolizumab (Keytruda) (3-6). BC is curable in 70-80% of patients with early-stage, non-metastatic disease (3). BC metastasizes to the bones, lungs, regional lymph nodes, liver and brain and metastases only poorly respond to therapies (7, 8). Another problem is the development of resistance to chemotherapy and targeted therapies (9, 10). In order to identify new targets and treatment modalities for BC, we searched the literature for circular RNAs (circRNAs) with efficacy in preclinical BC-related in vivo models. We excluded triple-negative BC because relevant circRNAs in this subtype will be summarized in a separate review.
Circular RNA
CircRNAs are generated by back-splicing of polymerase II transcripts, thus creating new junctions (11). Their size ranges from a hundred to several thousands of nucleotides (11). In humans, common size of a circRNA is a few hundred nucleotides comprising two to three exons (12). In mammalian cells, at least 30,000 different circRNAs have been identified (13). They affect processes such as transcription, splicing, protein scaffolding, and can act as micro-RNA (miR) sponges and decoys; in rare cases they can even function as translational templates (11-13). In BC they can act as tumor suppressors, as well as oncogenes affecting processes such as tumor initiation, progression, proliferation, cell-cycle, cell death, migration, invasion, metastasis, angiogenesis, modulation of the tumor microenvironment and chemoresistance (14-16). CircRNAs can be detected in peripheral blood encapsulated in exosomes and their roles as biomarkers, predictors of survival, metastasis and drug resistance are under active investigation (17, 18). The physiological relevance of circRNAs in cancer has recently been supported by the fact that inhibition of individual circRNAs can inhibit tumor growth in patient-derived xenografts of lung adenocarcinoma and gastric cancer (19, 20).
CircRNAs Down-regulated in BC
circRNAs affecting chemoresistance
Circ-lysine-specific demethylase 4C (circKDM4C) targets phenazine biosynthesis-like domain-containing protein (PBLD). CircKDM4C (Figure 1) was down-regulated in BC and its down-regulation correlated with poor prognosis (21). CircKDM4C repressed BC cell proliferation, metastasis and adriamycin resistance in vitro and in vivo. Mechanistically, it sponged miR-548p, which led to up-regulation of PBLD. In hepatocellular carcinoma (HCC), reduced expression of PBLD correlated with poor prognosis and forced expression of PBLD inhibited HCC growth in vitro and in vivo by interfering with mitogen-activated protein kinase (MAPK), nuclear factor 
B (NF
B), epithelial–mesenchymal transition (EMT) and angiogenesis signaling pathways (22).
Circular (circ) RNAs down-regulated in breast cancer with efficacy in preclinical in vivo models. CircRNAs involved in adriamycin (ADM) and tamoxifen (TAM) resistance, as well as signaling-related circRNAs are shown. Up- and down-regulation are indicated by up and down arrows, respectively. ERK1/2: Extracellular signal-regulated kinases 1,2; FOXO3A: forkhead-box protein O3A; HIPK3: homeodomain-interacting protein kinase 3; JAK: Janus kinase; KDM4C: lysine-specific demethylase 4C; MEK1: mitogen-activated protein kinase 1; MET: metastasis; miR: microRNA; NOL10: nucleolar protein 10; PBLD: phenazine biosynthesis-like domain-containing protein; SOCS 2,3: suppressor of cytokine signaling 2,3; STAT: signal transducer and activator of transcription; WNK2: WNK lysine-deficient protein kinase 2.
Circ0025202 targets forkhead-box-protein O3A (FOXO3A). Low expression of circ0025202 (Figure 1) was found in BC tissues (23). Overexpression of circ0025202 reversed tamoxifen resistance of MCF-7/TR cells (23). In vivo, it suppressed tumor growth and increased tamoxifen sensitivity of MCF-7/TR cells in a xenograft model. Mechanistically, circ0025202 acted as a sponge for miR-182-5p, and resulting in up-regulation of FOXO3A. The latter is a transcription factor which is involved in apoptosis, proliferation, cell-cycle progression, DNA damage and tumorigenesis. FOXO3A is frequently inactivated by mutation in tumors or sequestrated in the cytoplasm and overexpression of FOXO3 inhibits proliferation, tumorigenic potential and invasiveness, EMT and metastasis (24-26). FOXO3A-driven miR signatures modulate vascular endothelial growth factor/neuropilin 1 signaling and BC metastasis (27).
Circ0025202 targets homeodomain-interacting protein kinase 3 (HIPK3). Circ0025202 (Figure 1) was reduced in BC tissues and tamoxifen-resistant BC cells (MCF-7 and T47D) (28). Knockdown of circ0025202 elevated the half-maximal inhibitory concentration for tamoxifen, promoted cell proliferation, invasion and migration, and mediated cell-cycle progression and inhibition of apoptosis in vitro (28). Up-regulation of circ0025202 hindered xenograft growth of MCF-7/TR and promoted tamoxifen sensitivity in nude mice. Circ0025202 targeted miR-197-3p and thus led to up-regulation of HIPK3 (28). The latter is a member of the ser/thr kinase family with three members (HIK1, −2 and −3) and interacts with homeobox proteins and other transcription factors as transcriptional co-activators or co-repressors (29). HIPK2 acts as a suppressor of development and metastasis of many types of tumors, down-regulates vimentin and inhibits BC cell invasion (30). HIPK3 overexpression can inhibit growth of non-small-cell lung carcinoma (31) and drives p53 activation to limit colorectal cancer growth (32). HIPK1 is involved in DNA repair through its interaction with p53 (33).
circRNAs affecting suppressor of cytokine 2,3 (SOCS2,3) signaling
Circ-nucleolar protein 10 (circNOL10) targets suppressor of cytokine signaling 2 (SOCS2). CircNOL10 (Figure 1) was down-regulated in BC tissues and cell lines (34). In BT-549 and MDA-MB-231 BC cells, circNOL10 suppressed proliferation, migration, invasion, EMT by sponging miR-767 which resulted in activation of SOCS2 and inhibition of Janus kinase (JAK)/signal transducer and activator of transcription (STAT) signaling. In nude mice, circNOL10 suppressed the growth of BT-549 BC xenografts.
Circ0001785 targets SOCS3. Circ0001785 (Figure 1) was reduced in T47D, MCF-7, MDA-MB-453, MDA-MB-231 and BT-549 BC cells in comparison to MCF-10A normal breast cells. It inhibited proliferation, invasion and migration of BC cells, as well as tumor growth in nude mice by sponging miR-942 through up-regulation of SOCS3 (35).
SOCS family members inhibit JAK, signal transducers and signal transducer and activator of transcription (STAT) are down-regulated in BC (36, 37). In BC, loss of SOCS2 is related to cell proliferation and tumor growth (38). In patients with BC with lymph node metastasis, reduced expression of SOCS3 was found (39).
Circ001666 targets WNK lysine-deficient protein kinase 2 (WNK2). Circ001666 (Figure 1) was down-regulated in BC tissues and cell lines (40). It inhibited proliferation, migration, invasion and promoted apoptosis of BC cells in vitro and tumor growth of BC xenografts in nude mice. Circ001666 sponged miR-620, resulting in up-regulation of WNK2 tumor suppressors. WNK2 is a ser/thr kinase, with four paralogs, which inhibits cell proliferation by modulating the activation of MAPK/extracellular signal regulated kinase 1/2 (ERK1/2) (MEK1/ERK1/2) and negatively regulates sodium transport (41). WNK2 is down-regulated in BC and inhibits BC cell proliferation (42). During pancreatic ductal adenocarcinoma development, WNK2 was also found to be down-regulated (43). However, its function seems to be context-dependent, because in HCC, WNK2 acts as a driver of carcinogenesis and is a risk factor of early recurrence (44).
CircRNAs Up-regulated in BC
circRNAs mediating chemoresistance
Circ-ATP-binding cassette subfamily B member 10 (circABCD10) targets dual-specificity phosphatase 7 (DUSP7). An abundance of circABCD10 (Figure 2A) was associated with reduced sensitivity to paclitaxel in BC tissues and cells (45). In paclitaxel-resistant BC cell lines MCF-7/PTX and MCF-MB-231/PTX, circABCD10 sponged let-7a-5p and led to up-regulation of DUSP7. In nude mice, knockdown of circABCD10 suppressed growth of MCF-7/PTX xenografts (45). The DUSP family exert their functions through dephosphorylation of MAPK (ERK1/2, JUN kinase p38) (46). DUSP7 acts as an oncogene in BC (47). DUSPs play an emerging role in human cancer (48). Paclitaxel is widely used for treatment of early and advanced BC (49).
Chemoresistance-related circular (circ) RNAs up-regulated in breast cancer with efficacy in preclinical in vivo models. A: CircRNAs conferring paclitaxel (PTX) resistance. B: CircRNAs conferring adriamycin (ADM) resistance. C: CircRNAs conferring tamoxifen (TAM) or lapatinib resistance (LR). Up- and down-regulation are indicated by up and down arrows, respectively. ABCD10: ATP-binding cassette subfamily B member 10; ANLN: anillin; CDK8: cyclin-dependent kinase 8; CHD4: chromodomain helicase DNA-binding protein; DUSP7: dual specificity phosphatase 7; E2F3: transcription factor E2F3; ERα: estrogen receptor α; HIPK3: homeodomain-interacting protein kinase 3; HK2: hexokinase 2; INTβ1: integrin β1; LR: lapatinib resistant; MMP11: matrix metalloprotease 11; NCOA3: nuclear receptor co-activator A3; RNF111: ring finger protein 111; UBE2D2: ubiquitin-conjugating enzyme 2 D2.
Circ-ring finger protein 111 (circRNF111) targets transcription factor E2F3. High expression of circRNF111 (Figure 2A) was observed in paclitaxel-resistant BC tissues and cell lines (50). CircRNF111 knockdown suppressed paclitaxel resistance, cell viability, colony formation, invasion and glycolysis in paclitaxel-resistant BC cells MCF-7/PTX and MDA-MB-231/PTX. In nude mice, knockdown of circRNF111 in MCF-7/PTX xenografts suppressed paclitaxel-resistance. CircRNF111 sponged miR-140-5p resulting in up-regulation of E2F3 (50). The latter is a member of the E2F family of transcription factors which are deregulated in BC (51). E2F3 is a driver of EMT, cell invasion and metastasis in BC (52, 53).
CircHIPK3 targets HK2. CircHIPK3 (Figure 2A) was upregulated in paclitaxel-resistant BC tissues and cell lines such as MCF-7/PTX and MDA-MB-231/PTX (54). Silencing of circHIPK3 enhanced drug sensitivity in vitro due to sponging of miR-1286 and up-regulation of HK2. In nude mice, silencing of circHIPK3 diminished tumor growth and promoted paclitaxel sensitivity of MDA-MB-231/PTX xenografts (54). HK2 is a tumor driver in BC and is involved in chemoresistance of BC (55, 56). It was shown that Pim2 proto-oncogene serine protein kinase (PIM2)-mediated phosphorylation of HIPK3 is critical for tumor growth and paclitaxel-resistance in BC (57).
Circ0006528 targets CDK8. Circ0006528 (Figure 2A) was up-regulated in BC tissues and paclitaxel-resistant BC cell lines BT549/PTX and ZR-75-30-PTX (58). In vitro silencing of circ0006528 repressed proliferation, migration and invasion, as well as autophagy, and induced apoptosis in vitro (58). Circ0006528 sponged miR-1299 and up-regulated CDK8. The latter promoted proliferation, migration and autophagy in paclitaxel-resistant BC cells (58). Circ0006528 increased xenograft growth of ZR-75-30/PTX in nude mice (58). CDK8 is associated with paclitaxel resistance and activation of the WNT/ß-catenin signaling pathway (59, 60). CDK8 acts both as an activator and repressor of transcription, invasiveness and EMT (61). Inhibition of CDK8 inhibits growth and proliferation of BC cells (62, 63).
Circ0006528 targets chromodomain helicase DNA-binding protein 4 (CHD4). Higher levels of circ0006528 (Figure 2B) have been found in adriamycin-resistant BC tissues and cells (64). Overexpression of circ0006528 mediated proliferation, migration, invasion and adriamycin resistance in BC cells. Circ0006528 sponged miR-1236-3p and led to up-regulation of CHD4 (64). It represents the main component of the nucleosome remodeling and deacetylase complex and plays an important role in epigenetic transcriptional regulation (65). CHD4 promotes BC progression as a coactivator of hypoxia-inducible factors (66). In addition, CHD4 regulates the HER2 signaling pathway and autophagy in HER2+ BC cells (67) and is an essential gene for BC growth (68, 69). Independently, it has been shown that CHD4 is involved in DNA damage response and chemotherapy resistance (70).
Circ0001667 targets nuclear receptor co-activator A3 (NCOA3). Knockdown of circ0001667 (Figure 2B) in MCF-7/ADM and MDA-MB-231/ADM inhibited proliferation, migration, invasion and adriamycin resistance in vitro (71). Knockdown of circ0001667 repressed tumor growth and adriamycin resistance of BC cells in immunodeficient mice (71). miR-4458 was sponged by circ0001667 and led to upregulation of NCOA3, also known as amplified in BC1 (AIB1) or steroid receptor co-activator 3 (SRC3). NCOA3 is a transcriptional co-activator with several nuclear receptor-interacting domains and intrinsic histone acetyltransferase activity, resulting in acetylation of histones and assisting nuclear receptors in up-regulation of genes (72). NCOA3 is amplified in BC and acts as an oncogene (73).
Circ0085484 targets integrin β1 (INTB1). Depletion of circ0085484 (Figure 2B) repressed adriamycin resistance, proliferation, and metastasis of adriamycin-resistant BC cells (74). The phenomenon is due to sponging of miR-873-5p and subsequent up-regulation of INTB1 (74). It was shown that INTβ1 can bind to collagen type 1 and activate adriamycin efflux transporters (75).
Circ-ubiquitin-conjugating enzyme 2D2 (circUBE2D2) targets ERα. Tamoxifen is a selective ER modulator used for prevention and treatment of BC. Treatment resistance to tamoxifen is frequently observed (76, 77). CircUBE2D2 (Figure 2C) was up-regulated in tamoxifen-resistant BC tissues and cell lines such as MCF-7/TMX and T47/TMX (78). Deletion of circUBE2D2 (Figure 2C) inhibited tamoxifen resistance in MCF-7/TMX and T47/TMX cells. UBE2D2 was also found in exosomes of these cell lines. Intercellular transfer of circUBE2D2 enhanced tamoxifen resistance in vitro and in vivo. circUBE2D2 sponged miR-200a-3p and increased viability, metastasis and the level of ERα (78).
Circ-matrix metalloprotease11 (circMMP11) targets anillin (ANLN). Lapatinib, an inhibitor of HER2 and epidermal growth factor receptor is used for the treatment of HER2-overexpressing advanced and metastatic BC (79). CircMMP11 (Figure 2C) was up-regulated in lapatinib-resistant BC tissues and cell lines (MDA-MB-231/LR and MCF-7/LR) (80). CircMMP11 was transported by exosomes and enhanced lapatinib resistance in BC cell lines. CircMMP11 knockdown impeded tumor growth of MDA-MB-231/LR in nude mice and increased lapatinib resistance (80). CircMMP11 sponged miR-153-3p, resulting in upregulation of ANLN. The latter is a cytoskeletal protein, containing an actin/myosin-binding domain and a pleckstrin-homology domain, which is critical for cell division and EMT, and is frequently overexpressed in cancer (81, 82). Knockdown of ANLN inhibits growth of BC cells (82). ANLN is also involved in poor prognosis in patients treated with anthracycline-based chemotherapy (83).
CircRNAs up-regulating high-mobility group (HMG) proteins Circ0069094, circ-F-box and leucine-rich repeat protein 5 (circFBXL5), circ-homeodomain-interacting protein kinase 3 (circHIPK3) and circ0003645 target HMGA1, -A2 and - B1. Circ0069094 (Figure 3A) was up-regulated in BC samples in comparison to adjacent normal tissues and non-cancerous MC7-10A cells (84). Knockdown of circ0069094 inhibited cell glycolysis, glucose uptake, lactate production, HK2, proliferation, migration, invasion and increased apoptosis in BC cells in vitro and tumor formation in vivo in nude mice (84). From a mechanistic point of view, circ0069094 sponged miR-661, resulting in up-regulation of HMGA1.
Circular (circ) RNAs up-regulated in breast cancer with efficacy in preclinical in vivo models affecting high-mobility group (HMG) proteins, secreted factors and transmembrane proteins. A: CircRNAs up-regulating HMG proteins. B: CircRNAs up-regulating secreted and transmembrane proteins. Up- and down-regulation are indicated by up and down arrows, respectively. CXCL10: C-X-C Chemokine ligand 10; FBXL5: F-box/LRR repeat protein 5; HIPK3: homeodomain-interacting protein kinase 3 exon 2 splicing; IGF1: insulin-like growth factor 1; miR: microRNA; MUC19: mucin 19; PI3K: phosphatidylinositol-4,5-bisphosphate 3-kinase; PLK1: POLO-like kinase 1; TFF1: trefoil factor 1.
CircFBXL5 (Figure 3A) was up-regulated in BC samples, promoted migration and invasion, and inhibited apoptosis in MBA-MB-231 and MDA-MB-453 BC cells in vitro and its silencing reduced 5-fluorouracil-resistant BC growth in vivo in nude mice (85). CircFBXL5 sponged miR-216b, resulting in up-regulation of HMGA2 (85).
CircHIPK3 up-regulates HMGB1 (Figure 3A), which was increased in BC tissues and high expression predicted poor prognosis. Small interfering RNA (siRNA)- mediated knockdown of circHIPK3 reduced viability, migration and invasion of MDA-MB-231 BC cells in vitro and inhibited tumor growth in vivo in nude mice (86). CircHIPK3 sponged miR-193a, resulting in expression of HMGB1 and activation of phosphatidylinositol-4,5-bisphosphate 3-kinase (PI3K)/AKT serine/threonine kinase 1 (AKT) signaling in MDA-MB-231 cells (86).
Circ0003645 (Figure 3A) also up-regulated HMGB1 by sponging miR-139-3p (87). Circ0003645 was overexpressed in BC tissues, promoted proliferation of MCF-7 and MDA-MB-231 BC cells in vitro and tumor growth of MDA-MB-231 BC cells in vivo in nude mice after subcutaneous implantation (87).
HMG proteins are chromosomal DNA-binding proteins which are involved in the regulation of DNA-dependent processes such as transcription, replication, recombination and DNA repair (88). HMGA contains an AT-hook domain, HMGB contains an HMG-box domain and HMGN exhibits a nucleosomal-binding domain (89). They can directly interact with transcription factors and alter chromatin structure (89). Expression of HMGA1 and HMGA2 is correlated with malignant status and prognosis of patients with BC (90. 91). Knockdown of HMGA1 inhibited BC growth and metastasis in immunodeficient mice and attenuated BC angiogenesis (91, 92). HMGA2 is involved in proliferation, migration, invasion, acquisition of stem cell features, EMT and telomere restoration (93, 94). HMGA1 also occurs extracellularly and is involved in an autocrine loop by binding to the receptor for advanced glycation end-products, resulting in migration, invasion and metastasis, and can be targeted with monoclonal antibodies (95). HMGA2 promotes BC metastasis by modulating Hippo-YAP signaling (96). The druggability of HMGs deserves further attention.
CircRNAs up-regulating secreted and transmembrane proteins Circ-trefoil factor 1 (circTFF1) targets trefoil factor 1. CircTFF1 (Figure 3B) was found to be highly expressed in BC and its depletion restrained cell migration, invasion and EMT in BT-549 and MBA-MB-231 BC cells (97). CircTFF1 sponged miR-326, resulting in up-regulation of TFF1. CircTFF1 accelerated growth of BT-549 cells in nude mice (97). Three TFFs are known to contain at least one copy of a 40 amino acid domain with three conserved sulfide bonds (98). TFF1 is regulated by ER and can act as a tumor suppressor in mice (99). On the other hand, TFF1 stimulated migration of BC cells (100). TFF1 represents a potential prognostic biomarker with functional significance in patients with BC (101). TFF1 is expressed at higher levels in blood from patients with metastatic BC than in those without metastatic disease (102). Independently, it has been shown that TFF1 is up-regulated in ER+ BC and is correlated with increased bone metastasis (103).
Circ0000515 targets C-X-C chemokine ligand 10 (CXCL10). Circ0000515 (Figure 3B) was up-regulated in BC and its expression level correlated with poor prognosis (104). It promoted cell-cycle progression, proliferation, invasion, inflammatory response and pro-angiogenic potential of BC cells (104). Circ0000515 was shown to bind to miR-296-5p, preventing it from repressing CXCL10. CXCL10 binds to CXC motif chemokine receptor 3 (CXCR3) together with CXCL9 and CXCL11 (105). Loss of CXCL10 reduced growth of MCF-7 BC cells in nude mice (104). CXCL10 mediates BC resistance to tamoxifen and promotes estrogen-dependent and - independent proliferation (106). In BC, RAS induced CXCL10 signals through serine-threonine kinase RAF and PI3K signaling pathways. However, CXCL10 can have dual effects on cell growth (107, 108). CXCL10 plays a role in migration of tumor cells (109). CXCL10 and its receptor CXCR3 play a key role in BC metastasis to bone and osteoclast activation through NF
B activation (110). However, a major function of CXCL10 is enhancement of T-cell-dependent anticancer immunity, a property which cannot be assessed in the immunodeficient in vivo model as described above (105). Therefore, the role of CXCL10 as a cancer-promoting agent has to be scrutinized in more detail.
Circ-POLO-like kinase 1 (circPLK1) targets insulin-like growth factor 1 (IGF1). CircPLK1 (Figure 3B) was upregulated in BC tissues and cells (111). In BT549 and HCC38 BC cells, circPLK1 regulated cell proliferation, cell-cycle transition from G1 to S phase, and migration and invasion in vitro. In vivo, interference with circPLK1 restrained tumor growth in nude mice (111). CircPLK1 binds to miR-4500, resulting in up-regulation of IGF1. The latter functions as a potent mitogen in the mammary gland via IGF receptor 1 signaling (112). This signaling system also includes six IGF-binding proteins, which modulate the bioavailability of IGF1 and triggers PI3K/AKT and RAF/MAPK pathways (112). In addition, IGF1 also binds to the insulin receptor (113). In preclinical models of BC, IGF1/2 antibodies inhibited bone metastasis without affecting the growth of the primary tumor (114, 115). However, clinical trials with IGF receptor 1 antibodies, IGF receptor 1 tyrosine kinase inhibitors and IGF1/2 antibodies in patients with cancer have generated negative results.
Circ0001982 targets mucin 19 (MUC19). Circ0001982 (Figure 3B) was increased and its knockdown inhibited cell glycolysis, viability of and migration and invasion by MDA-MB-231 and MDA-MB-468 BC cells (116). Circ0001082 sponged miR-1297-5p, resulting in up-regulation of MUC19 and mediated xenograft growth of MDA-MB-231 cells in nude mice (116). Mucins are large O-glycoproteins which occur as transmembrane or secreted molecules to activate signaling pathways such as NF
B, WNT, SRC, p53, MAPK, hypoxia-inducible factor and JAK-STAT, affecting stemness, metabolism and chemoresistance (117). MUC1, −4 and −16 are also involved in the pathogenesis of BC (118). MUC16, which was discovered as cancer antigen 125 and functions as a ligand for mesothelin, is aberrantly expressed in many types of cancer and numerous clinical studies with corresponding monoclonal antibodies and their conjugates are ongoing. Cleavage of its extracellular domain represents a major problem for therapy (119, 120). Other approaches focus on the use of MUC1 as a cancer vaccine (121). For MUC19, more preclinical validation studies are needed to rank its significance for treatment of BC.
CircRNAs involved in signaling
Circ-inhibitor of NF
B kinase subunit beta (circIKBKB) targets NF
B. Overexpression of circIKBKB (Figure 4A) in MCF-7 and MDA-MB-231 BC cells induced osteoclastogenesis, formation of a bone pre-metastatic niche and bone metastasis after intracardiac injection in nude mice (122). Overexpression of circIKBKB activated NF
B signaling due to I
B kinase β-mediated phosphorylation of NF
B inhibitor α (I
Bα) and its subsequent degradation. CircIKBKB recruited NF
B to the promoters of several bone-remodeling factors such as receptor activator of NF
B ligand, macrophage colony-stimulating factor and granulocyte-macrophage colony-stimulating factor. Eukaryotic translation initiation factor 4A3 (EIF4A3) directly bound to circIKBKB pre-mRNA and induced its cyclization. Blocking of EIF4A3 by RNA interference or a specific, highly selective non-competitive inhibitor blocked NF
B signaling, osteoclastogenesis and bone metastasis (122). EIF4A3 is a nuclear matrix protein and a core component of the exon junction complex (123, 124). Bone metastasis occurs in 70% of patients with metastatic BC (125). Bisphosphonates and receptor activator of NF
B ligand monoclonal antibody denosomab are approved for treatment of metastatic bone disease but they do not have an impact on survival and exhibit severe side-effects (126). It remains to be studied as to whether the inhibition of hyperactivated NF
B signaling will result in improved agents.
Circular (circ) RNAs up-regulated in breast cancer with efficacy in preclinical in vivo models affecting signaling and induction of additional targets. A: CircRNAs involved in signaling. B: CircRNAs up-regulating additional targets. Up-regulated and downregulated are indicated by up and down arrows, respectively. AKT: serine-threonine kinase AKT; CBX4: chromobox protein 4; CCND1: cyclin D1; CFL2: cofilin 2; COL1A1: collagen 1A1; DENN4C: DENN domain-containing 4C; EGFR: epidermal growth factor receptor; GRB2: growth factor receptor-bound protein 2; HK2: hexokinase 2; IKBKB: inhibitor of NF
B kinase; I
Bα-P: phosphorylated inhibitor I
Bα; MAPK1: mitogen-activated protein kinase 1; MET: metastasis; miR: microRNA; NF
B: nuclear factor 
B; PAK4: serine-threonine protein kinase PAK4; RPPH1: ribonuclease P component H1; TAZ: transcriptional co-activator with a PDZ-binding motif; UBR1: ubiquitin-protein ligase 1; WWC3: WW and C3 domain-containing; YAP1: YES-associated protein 1.
Circ-WW and C3 domain-containing (circWWC3) targets RAS pathway genes. Zinc finger E-box binding homeobox 1 (ZEB1) up-regulates circWWC3, which correlated with BC progression (127). CircWWC3 (Figure 4A) increased invasion and migration of MCF-7 and MDA-MB-231 BC cells. CircWWC3 functioned as a sponge for miR-26b-3p and -660-3p and up-regulated RAS pathway genes epidermal growth factor receptor, growth factor receptor-bound protein 2, serine-threonine kinase PAK4, MAPK1, and AKT. In nude mice, circWWC3 promoted metastasis of MDA-MB-231 cells to the lungs and liver. The RAS/RAF/MEK/ERK MAPK pathway plays a crucial role in growth, survival and differentiation of cancer cells (128). Targeting mutant RAS has recently led to approved drugs (129-131).
Circ-ribonuclease P component H1 (circRPPH1) targets YES-associated protein 1 (YAP1). CircRPPH1 (Figure 4A) was overexpressed in BC and mediated proliferation, migration and invasion in vitro of MCF-7 and MDA-MB-231 cells as well as endothelial tube formation with supernatants from circRPPH1-transfected BC cells (132). CircRPPH1 sponged miR-556-5p, up-regulated YAP1 and promoted tumor growth in nude mice bearing MDA-MB-231 BC cells (132).
Circ0005273 targets YAP1. Circ0005273 (Figure 4A) was highly expressed in BC and correlated with TNM stage, lymph node metastasis, tumor size and distant metastasis (133). It induced proliferation, migration and cell-cycle progression in MDA-MB-231, MCF-7 and SKBR3 BC cells in vitro and promoted tumor growth in MDA-MB-231 cells in nude mice (133). This was due to sponging of miR-200a-3p, which led to up-regulation of YAP1.
Circ0000511 targets transcriptional co-activator with PDZ-binding motif (TAZ). Circ0000511 (Figure 4A) was upregulated in BC and accelerated proliferation, migration, invasion, and impeded apoptosis in MCF-7 and MDA-MB-468 BC cells in vitro (134). Circ0000511 sponged miR-326 and up-regulated TAZ. In vivo, circ0000511 promoted tumor growth in nude mice (134).
YAP and TAZ are transcriptional co-activators of the Hippo pathway, which signals through components of the extracellular matrix, cell adhesion sites, cell shape and the actomyosin cytoskeleton and determines organ size in animals (135-137). YAP/TAZ regulates BC-related metastasis targets (138, 139). Therefore, targeting of the Hippo pathway in BC is a high priority issue (140). Verteporforin has been identified as a compound which inhibits YAP-TEA domain family interaction (140) and BAY1238097 interacts with YAP/TAZ and down-regulates their transcriptional activity by inhibiting bromodomain-containing protein 4 (141).
CircRNAs up-regulating additional targets
Circ-ubiquitin protein ligase 1 (circUBR1) targets cyclin D1 (CCND1). CircUBR1 (Figure 4B) was up-regulated in BC and its silencing inhibited proliferation and metastasis, promoted apoptosis in vitro and restrained tumor growth in vivo (142). CircUBR1 sponged miR-1299 and led to overexpression of CCND1. The latter is overexpressed in more than 50% of BCs, causes mammary cancer in transgenic mice, promotes cellular proliferation and cell-cycle progression, and has a role in BC stem cell expansion (143). CCND1 has been found to be amplified in BC (144). It has been found that a CCND1 G870A polymorphism increased BC risk (145).
Circ0008673 targets cofilin 2 (CFL2). Circ0008673 (Figure 4B) was up-regulated in BC, promoted proliferation, migration and invasion, and inhibited apoptosis of BC cells in vitro. It acted as a sponge for miR-153-3p and its silencing repressed BC xenograft growth in nude mice (146). This was due to up-regulation of CFL2, which mediated growth of BC cells in vitro (146, 147). CFL2 reversibly controls actin polymerization and depolymerization in a pH-dependent manner and regulation of the actin cytoskeleton in cancer cell migration and invasion (148). CFL2 promotes cancer cell invasion through O-Glc-Nacylation (149) and BC invasion and metastasis (150).
Circ002178 targets collagen 1A1 (COL1A1). Circ002178 (Figure 4B) was overexpressed in BC tissues and correlated with poor prognosis (151). Silencing of circ002178 impaired proliferation, energy metabolism and angiogenesis. Circ002178 bound to miR-328-3p and led to up-regulation of COL1A1. In vivo, circ002178 stimulated growth of MDA-MB-231 BC cells in nude mice. COL1A1 is an extracellular matrix protein that is correlated with advanced BC, poor prognosis, invasion and metastasis (152). Development of the extracellular matrix is correlated with BC cancer progression (153, 154). Taken together, these findings show COL1A1 is a potential target for treatment of BC.
Circ0008030 targets chromobox protein 4 (CBX4). Circ0008030 (Figure 4B) was up-regulated in BC tissues and cells and expedited proliferation, invasion and migration of BC cells (155). It sponged miR-515-5p, which led to upregulation of CBX4. Knockdown of circ0008030 suppressed tumor growth of BC xenografts in nude mice. CBX family proteins are canonical components of polycomb repressive complex 1 (PRC1) which bind to DNA and transcriptionally repress target genes via chromatin modification, acting as oncogenes or tumor suppressors (156). CBX4 recognizes H3K27me3, a transcriptionally suppressive epigenetic marker (157). In BC, CBX4 exhibits oncogenic activity via NOTCH1 signaling (158). In addition miR-129-5p has been shown to suppress BC proliferation by targeting CBX4 (159).
Circ-differentially expressed in normal cells neoplasia domain-containing 4C (circDENN4C) targets HK2. CircDENN4C (Figure 4B) was up-regulated in BC cells in response to hypoxia (160). Knockdown of DENN4C in MDA-MB-453 and SKBR3 BC cells inhibited glycolysis, lactate production and HK2, as well as migration and invasion under hypoxia. It was found that the observed phenomena were due to sponging of miR-200b and -c by circDENN4C. In addition, EMT and expression of MMP2 and −9 were increased under hypoxia by circDENN4C (160). However, the underlying mechanisms have to be resolved in more detail. Knockdown of circDENN4C in MDA-MB-435 cells inhibited xenograft growth in nude mice (160). Increased glycolysis is one of the hallmarks of cancer (161, 162) and a glycolysis-related expression signature predicts recurrence of BC (163).
Approaches and Challenges of Targeting CircRNAs
Down-regulated circRNAs can be reconstituted by transfection of expression vectors for the corresponding circRNA into recipient cells (164). For up-regulated circRNAs, the new junctions generated via back-splicing events may be specifically targeted (165). Options include antisense-oligonucleotides (ASO), siRNAs (siRNA), short hairpin RNA (shRNA) and the clustered regularly interspaced short palindromic repeats (CRISPR)-Cas (CRISPR-associated proteins) method for specific cleavage of the target RNA. ASOs are chemically modified single-stranded oligonucleotides (12-24-mers) that bind to their RNA targets and, depending on the design, lead to the RNaseH-mediated degradation or inhibit the function of target RNAs by steric interference (166, 167). Medicinal chemistry-based backbone and sugar modifications give rise to optimized ASOs with improved stability, high potency, and enabled intracellular delivery without the help of transfection agents (166, 167). siRNA and shRNA are RNA interference-based agents. siRNAs are 12–24-mers double-stranded RNAs with one of the strand targets circRNAs through complementary pairing and incorporate the circRNA into the RNA-induced silencing complex (168). shRNAs, on the other hand, contain a tight hairpin loop structure and target-complementary sequences which can be delivered as plasmids or viral vectors, allowing for stable integration of shRNA and long-term knockdown of the target (18, 169). The CRISPR-CAS13-based method has recently been used for screening for functional circRNAs (170). siRNA and shRNA can be delivered as lipid-based polymers or with exosomes (171). However, low stability in cells, poor intracellular delivery, and lack of cell specificity are critical issues for RNA interference therapeutics in vivo (172). Immunogenicity is another critical issue of the agents as described above (173, 174). Progress in the design of new formulations and tissue-specific targeting methods are issues of paramount importance (175-178).
Conclusion
We have identified six down-regulated and 26 up-regulated circRNAs which mediate efficacy in preclinical BC-related in vivo models. According to their corresponding targets, they can be categorized into groups which mediate chemoresistance or signaling, or are related to HMG, transmembrane or secreted proteins, as well as other targets. Two down-regulated and eight up-regulated circRNAs mediate adriamycin, tamoxifen, paclitaxel and lapatinib resistance. Interestingly, down-regulation of circ0025202 mediates tamoxifen resistance via miR-182-5p and miR-197-3p, with FOXO3A and HIPK3 as targets (Figure 1). Circ006528 can mediate paclitaxel as well as adriamycin resistance via miR-1299 and miR-1236-3p with CDK8 and CHD4 as targets (Figure 2). SOCS2 and SOCS3 (Figure 1) are frequently down-regulated in BC, resulting in constitutive activation of JAK- and STAT-related signaling. Other identified circRNAs target tumor- and metastasis-promoting pathways driven by HMG proteins (HMGA1, -A2 and -B1) (Figure 3). There are druggability issues with small molecules, but interference with antibody-related moieties emerges as an alternative, because HMGs are also present on the cell surface (179). Further identified targets are mediators of metastasis-related pathways. NF
B signaling induced by circIKBKB (Figure 4) is a mediator of bone metastasis. Hippo-related components YAP1 (circRPPH1, circ0052372) and TAZ (circ0000511) are also involved in metastasis of BC (Figure 4). It remains to be established whether a therapeutic window for corresponding inhibitors can be defined. CircTFF1 up-regulates TFF1, which is involved in metastasis of ER+ BC and should be investigated in further detail in this subgroup of patients (Figure 3). A further unexplored target identified by our search is MUC19, which merits further investigation by target validation experiments (Figure 3).
Footnotes
Conflicts of Interest
UHW was and UB is an employee of Roche. H-E H is a postdoctoral fellow of Roche. Roche is interested in targeted therapies and diagnostics
Authors’ Contributions
The Authors contributed equally to all aspects of the article.
- Received March 3, 2023.
- Revision received April 3, 2023.
- Accepted April 4, 2023.
- Copyright © 2023, International Institute of Anticancer Research (Dr. George J. Delinasios), All rights reserved
This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY-NC-ND) 4.0 international license (https://creativecommons.org/licenses/by-nc-nd/4.0).
