Wnts belong to the glycoprotein family. This family has 19 members, which are typically secreted glycoproteins of 350–400 amino acids that exert pivotal functions in stem cell maintenance and development via the Wnt-β-catenin signaling pathway (Nusse, 2005). This canonical Wnt signaling pathway has been extensively investigated and is involved in many stages of invertebrate and vertebrate development and adult tissue homeostasis (Wodarz and Nusse, 1998; Logan and Nusse, 2004). Mechanistic studies have revealed that the transduction of Wnt-β-catenin signaling mediated by Wnts is activated through interplay with seven-pass transmembrane receptors of the Frizzled (Fz) family and LRP5/6 (Wodarz and Nusse, 1998). Upon binding of Wnts to the conservative cysteine-rich domain of Fz receptors and LRP5/6, a cytoplasmic protein called Dishevelled (Dvl) is stimulated. Dvl can block axis inhibition protein (Axin), preventing formation of the destruction complex consisting of adenomatous polyposis coli, Axin, glycogen synthase kinase-3β, and casein kinase-1α. This in turn prevents the phosphorylation and consequential degradation of β-catenin, a key tumor cell proliferation regulator within the Wnt-β-catenin signaling pathway (Figure 1A). Subsequently, β-catenin is stably translocated into the nucleus and binds to the N-terminus of lymphoid enhancer factor/T cell factor transcription factors, activating transcription of the downstream target genes and thus exerting the corresponding functions (Behrens et al., 1996; Niida et al., 2004; Chamorro et al., 2014). However, several natural extracellular antagonists, including the Fz-related protein sFRP, Wnt inhibitory factor 1, sclerostin, Wise, and DKK1, exert inhibitory effects through competitively binding LRP5/6 to effectively block associations between Wnts and LRP5/6, thereby attenuating the expression of downstream genes in the canonical Wnt signaling pathway (Piccolo et al., 1999; Itasaki et al., 2003). Among them, DKK1, a typical Wnt antagonist, antagonizes Wnt signaling by triggering the internalization of coreceptor LRP5/6 (Semënov et al., 2001). DKK1 has also been found to participate in the Wnt pathway by binding to Kremen, a transmembrane protein interacting with DKK1 with high affinity; however, Kremen might not be a necessary component in DKK1-mediated Wnt antagonism (Mao et al., 2002; Wang et al., 2008). Overall, a wide range of functions of the Wnt pathway influencing developmental decisions in humans or animals have been uncovered over the years. However, gaining further insight into several outstanding mechanisms of signaling transduction and developing interventions for the treatment of diseases remain formidable challenges (Logan and Nusse, 2004).

The JNKs, also called stress-activated protein kinases, form a well-identified subfamily of mitogen-activated protein kinases (Kyriakis and Avruch, 2001; Weston and Davis, 2002). They were initially found to catalyze the phosphorylation of the N-terminal transactivation domain of c-Jun (Ser-63 and Ser-73), leading to the inhibition of its ubiquitination, which was the origin of the name “JNK.” However, JNK can also phosphorylate other transcription factors, including JunB, JunD, c-Fos, and ATF, all of which have critical roles in controlling the levels of several stress-responsive genes (Yang et al., 1998). There are three main mammalian JNK genes, JNK1, JNK2, and JNK3, located on different chromosomes. Their expression and functions vary among individual cells or organs, owing to differences in substrate affinity between the multiple JNK isoforms resulting from splicing of JNK genes (Dérijard et al., 1994). The JNK pathway is involved in many neurological disorders (Savage et al., 2002), diabetes (Davis et al., 2000; Hirosumi et al., 2002), inflammatory bowel disease (Malamut et al., 2006; Craige et al., 2019), non-alcoholic fatty liver (Ferreira et al., 2011), heart diseases (Cook et al., 1999), and cancers (Hui et al., 2008; Chang et al., 2009; Barbarulo et al., 2013). Previous studies have also demonstrated that JNK functions are important in apoptosis in a variety of cell death types, including necrosis and autophagy, and that the loss of JNK signaling can cause suppressed tumor necrosis factor (TNF)-stimulated cell death (Ventura et al., 2004).

The expression of DKK1 has been reported to be dependent on the Ap-1 family member c-Jun. DKK1 levels and bone morphogenetic protein (Bmp)-triggered apoptosis were found to be considerably reduced in c-Jun^−/− mice, indicating a pivotal role of c-Jun in transcriptional activation of DKK1 and regulation of Bmp signaling (Grotewold and Rüther, 2002). In addition, the apoptosis induced by Bmp requires c-Jun to activate DKK1, potently inhibiting the propagation of Wnt signals. Hence, the Bmp signaling pathway is likewise entangled with DKK1. Some evidence suggests that JNK signaling occurs via the β-catenin-independent Wnt pathway. In prostate cancer cells, ovarian cancer cells, and osteosarcoma, JNK signaling was activated by DKK1; in these cases, β-catenin levels were not affected, and promotion of cell invasion or tumor growth was observed. One proposed hypothesis for the activation of JNK signaling is that competitive binding of DKK1 to LRP5/6 could shift the Fz receptors to the JNK pathway because of the higher availability of Wnts (Kagey and He, 2017). When Wnts bind to coreceptors ROR and RYK, the recruited Dvl contributes to the activation of Rac GTPase, which activates JNK (Figure 1B). However, more evidence is needed to elucidate the mechanism more precisely. Dvl-dependent RhoA signaling is also activated, and this cascade promotes actin cytoskeletal reorganization and cell movement (Weston and Davis, 2007). Colla et al. (2007) reported that the activation of JNK signaling induced DKK1 expression in multiple myeloma (MM; a neoplastic disorder of plasma cells in the bone marrow) plasma cells. In their study, DKK1 was found to be responsive to the JNK signaling cascade, and they assumed that the sensitivity of myeloma cells in the early stages to various stresses activated JNK signaling, thereby increasing DKK1 expression.

Cytoskeleton-associated protein 4 (CKAP4), also termed p63, ERGIC-63, or CLIMP-63, is a type-II transmembrane protein of 63 kDa mainly localized to the endoplasmic reticulum (Bhavanasi et al., 2016). DKK1 binds to receptor LRP5/6 through a cysteine-rich domain (CRD1; DKK1c); CKAP4 has recently been identified as another DKK1 receptor, where the interactive domain is CRD2 (the N domain). The extracellular domain of CKAP4 directly binds to DKK1, and the K_D value of DKK1 towards CKAP4 is 0.42 nM. The binding affinity of DKK1-LRP6 (K_D, about 0.34–0.5 nM) is similar to that of DKK1-CKAP4, suggesting a certain significance of CKAP4 in DKK1-related regulation (Figure 1C). Once DKK1 interacts with CKAP4, the subsequent formation of the CKAP4-PI3K (phosphatidylinositide 3-kinase) complex via the proline-rich domain of CKAP4 and the Src homology 3 domain of the p85 subunit of PI3K (composed of p85 and p110 subunits) accelerates the generation of phosphatidylinositol-3,4,5-triphosphate and thus activates AKT and stimulates cell proliferation (Kimura et al., 2016). Intriguingly, the presence of LRP6 does not compete with CKAP4 for binding with DKK1 owing to the different locations of the binding domain; instead, it enhances DKK1-CKAP4 signaling, thereby promoting cancer cell proliferation more than DKK1-CKAP4 interaction alone (Kikuchi et al., 2021). The specific mechanism of the stimulation of PI3K mediated by DKK1/CKAP4 signaling is still under investigation. Clinically, DKK1- and CKAP4-positive patients with pancreatic ductal adenocarcinoma have shorter 5-years survival and relapse-free survival than DKK1- or CKAP4-negative patients. Conversely, high expression of CKAP4 was found to be consistent with longer overall survival (OS) in patients with cholangiocarcinoma and hepatocellular carcinoma (HCC), indicating that the cellular context might influence the function of CKAP4 (Kimura et al., 2019). Furthermore, DKK1 and CKAP4 were found in tumor lesions in more than 70% of lung squamous cell carcinoma patients. However, CKAP4 expression was reduced to a nearly undetectable level in noncancerous regions. The pharmacological treatment with an anti-CKAP4 antibody suppressed tumor formation in mice, indicating the potential of CKAP4 as an anti-tumor target (Kimura et al., 2016). Surprisingly, DKK1 functions through the DKK1/CKAP4/PI3K pathway to augment the expression of plasmalemma vesicle-associated protein in cholangiocarcinoma cells, which is positively linked to angiogenesis in various tumors, hence also making this pathway a prospective anti-angiogenic target (Wang et al., 2021).

Wu et al. (2014a) virtually screened several compounds that could disrupt the crosstalk between LRP5 and DKK1 using the National Cancer Institute (NCI) database, including NCI106164, NCI39914, and NCI660224. This resource is freely searchable and contains more than 250,000 small chemical compounds. Owing to the common anthra-9, 10-quinone substructure found in NCI39914 and NCI660224, screening based on substructure helped to identify another two compounds, NCI366218 and NCI8642. Biological assays showed that NCI8642 was the best compound for disrupting the interplay of DKK1 with LRP5/6 and reversing the Wnt signaling mediated by DKK1 (Wu et al., 2014a; Klavdianou et al., 2017). NCI8642 could efficiently displace DKK1 from LRP5/6 and impede DKK1 inhibitory activity on canonical Wnt signaling, as demonstrated by binding and cellular assays, respectively. In the presence of exogenously added DKK1, NCI8642 could reverse DKK1-mediated inhibition; the effect was concentration-dependent and significantly distinct from that of dimethyl sulfoxide treatment at 25 and 50 μM. Although the micromolar potency of NCI8642 might be too low to enable its in vivo use, this compound could serve as a reference small molecule for further drug development aiming to inhibit therapeutically important protein-protein interactions (Iozzi et al., 2012). The above studies demonstrate that the DKK1-LRP5/6 interplay can be targeted by small molecules, which unlocks the possibility of new therapeutic tools for diseases associated with DKK1 dysregulation. However, the exact mechanisms by which these compounds restrain the function of DKK1 in the Wnt signaling pathway remain to be fully clarified (Glantschnig et al., 2010). It is found that epimedium-derived flavonoids with multiple benzene rings were also capable of downregulating DKK1 expression (Figure 2). This basic structure could thus have an important function in inhibiting DKK1 and is a potential candidate for accelerated virtual screening (Xu et al., 2011).

BHQ880 (Novartis, Cambridge, MA, United States) is a phage-derived DKK1 antibody that neutralizes both human DKK1 and murine DKK1 but does not show high affinity for DKK2 or DKK3. In the SCID-hu murine model, osteoblast number and function, serum human osteocalcin levels, and trabecular bone were significantly increased after administration of BHQ880, although there was no direct effect on MM cell growth. Nonetheless, BHQ880 inhibited the growth of MM cells in the presence of bone marrow stromal cells (BMSCs) in vitro, as demonstrated by significantly lower cell adhesion and IL-6 levels in the presence of BMSCs. In addition, β-catenin levels were upregulated under the regulation of BHQ880, with a simultaneous decrease in nuclear factor-κB activity in BMSCs (Fulciniti et al., 2009). Another in vivo study reported obvious anti-osteolytic effects of BHQ880 in the 5T2MM murine model of myeloma, with concurrent protection against 5T2MM-induced trabecular bone loss, but BHQ880 treatment to inhibit DKK1 did not reduce tumor burden (Heath et al., 2009). However, a study using a mouse model of subcutaneous osteosarcoma xenografts, aiming to determine the linkage between tumor growth velocity and DKK1 level, reported a statistically significant decrease in tumor growth velocity in mice treated with BHQ880 compared with control mice. In an orthotopic implantation/amputation model of osteosarcoma metastasis, BHQ880 was found to activate the Wnt pathway and attenuate metastasis, promoting differentiation and slowing tumor growth. Additionally, BHQ880 suppressed the growth and metastasis of orthotopically implanted patient-derived osteosarcoma xenografts, which was relevant to the increase of nuclear β-catenin staining and an elevated level of osteopontin (Goldstein et al., 2016). To evaluate the safety and efficacy of BHQ880, Padmanabhan et al. (2009). carried out a phase IB multicenter dose-determination study. The results showed no dose-limiting toxicity in the combination treatment with anti-myeloma therapy and zoledronic acid, but 96.4% of patients reported adverse events (AEs), although only one-third of these might have been related to BHQ880 or zoledronic acid (ZA). No grade 3–4 AEs or severe AEs related to the treatment were reported; the most common AEs were grade 1–2 treatment-emergent effects such as arthralgia, fatigue, and pyrexia (Munshi et al., 2012). Of three patients who received 24 cycles of treatment, one had stable disease and the other two had partial responses, and 75% of the patients had an active skeletal-related event at baseline. Overall, although studies in a larger patient population are required to confirm the efficacy of BHQ880 monotherapy, BHQ880 shows clinical potential for treating MM (Iyer et al., 2014). As well as its use in MM, BHQ880 was shown to inhibit bone metastases and tumor growth in prostate cancer, indicating that BHQ880 treatment could be an effective therapy for advanced prostate carcinoma (Sottnik et al., 2012).

DKN-01 is a humanized IgG4 monoclonal antibody that neutralizes human and murine-derived DKK1 and specifically recognizes the C-domain of DKK1. A study of kinetic parameters showed that the K _D towards DKK1 of clinical DKN-01 was 28 pmol/L, similar to that of mouse-derived mDKN-01 (21 pmol/L) (Haas et al., 2021). Given the high expression and close involvement of DKK1 in MM, DKN-01 has been evaluated in patients with MM in preclinical studies and phase I testing. The data showed that DKN-01 possessed indirect anti-MM effects secondary to inhibition of osteoclastogenesis (Pozzi et al., 2013). In concordance with this, mDKN-01 exhibited no direct antiproliferative effect on B16F0 tumors, but a lack of DKK1 expression was detected after treatment (Haas et al., 2021). In an ovarian cancer mouse model, DKN-01 effectively modulated antitumor immunity, increasing natural killer (NK) cell infiltration (p = 0.045) and NK PD-L1 expression (p = 0.001), although there was no evident improvement in the survival outcomes (Betella et al., 2019). A mechanistic study demonstrated that DKN-01 required NK cells to exert its role in a syngeneic tumor model, which also illustrated the importance of NK cells in the efficacy of DKN-01 (Haas et al., 2019). In a phase I study of patients with advanced NSCLC, the median OS was 6.6 months after treatment with DKN-01, and the median progression-free survival was 2.2 months. No dose-limiting toxicities or serious AEs were found (Edenfield et al., 2014). An ongoing phase II basket study on patients with recurrent gynecologic cancer also confirmed that DKN-01 showed increased activity in combination therapy, but other correlative data in terms of safety and efficacy are still pending (Arend et al., 2019a).

Novel fully human monoclonal antibodies including RH2-18, RH2-31, RH2-59, RH2-80, and RH2-18LC01 have been discovered through screening with a phage-displayed scFv library. The anti-DKK1 neutralizing antibody RH2-18 possesses excellent affinity for DKK1, with a mean K _D value reaching 249 pM. Epitope mapping experiments showed that these antibodies recognize conformational epitopes rather than specific DKK1 sequences. To be specific, they recognize only non-denatured native DKK1 (Glantschnig et al., 2010). To map the key DKK1 domains for functional interplay with these antibodies, deletion constructs and Ala substitution mutants were obtained. The results suggested that two continuous protected fragments, Val188-Lys217 and Cys239-Cys245 in the C-domain, were the interacting sites (Arend et al., 2019a). The function of RH2-18 in interfering with Wnt signaling by blocking DKK1 was validated in human embryonic kidney 293 cells and mouse pluripotent mesenchymal cells; RH2-18 could mediate the restoration of Wnt-induced activities (Glantschnig et al., 2011). Furthermore, although an obvious neutralizing effect appeared at an antibody concentration of 8 nM, the neutralization of the suppressive effect of recombinant human DKK1 on cell differentiation was dose-related. Generally, the half-life (t1/2) of RH2-18 was 12–19 days, and pharmacokinetic analyses in rhesus monkeys showed an extended half-life compared with that in lower species, demonstrating its applicability in non-human disease models. Modulation of DKK1 function through an experimental fully human anti-DKK1 antibody exhibited translational potential as a bone-anabolic agent for the treatment of human low bone mass disease, a characteristic of postmenopausal osteoporosis. When RH2-18 was injected subcutaneously into ovariectomized mice and rhesus macaques, the BMD was obviously increased, especially the trabecular bone BMD, and the bone microarchitecture in the mouse vertebra was improved. In addition to exerting function on the axial skeleton, integral volumetric BMD, along with trabecular thickness at the distal radius and tibia, was significantly ameliorated as well after the administration of RH2-18 (Glantschnig et al., 2010; Glantschnig et al., 2011).

Owing to the compensatory effects of DKK1 levels and sclerostin expression, and the therapeutic potential of antibodies that target sclerostin (Scl-Ab) and DKK1 (DKK1-Ab) in treating bone disorders, including for improving fracture healing and increasing bone formation, Florio et al. (2016) proposed that a synergistic effect was triggered by the simultaneous blockage of DKK1 and sclerostin. Through hybridizing heavy chains and light chains of human Scl-Abs and human DKK1-Abs, a human sclerostin and DKK1 bispecific antibody (hetero-DS) was created. It had a K_D of 42 and 702 pM towards sclerostin and DKK1, respectively, and neutralized DKK1 and sclerostin with a similar potency to parental antibodies. Both fracture repairing experiments in rats and bone-forming effects experiments in primates showed that hetero-DS possessed more robust therapeutic efficacy than the parental antibodies (Florio et al., 2016). Furthermore, blocking DKK1 not only rescued the deterioration in osteoblastogenesis but neutralized TNF-mediated sclerostin expression in entirely differentiated osteoblasts in vitro and in vivo, resulting in stronger activation of the Wnt pathway. This makes concomitant targeting DKK1 and sclerostin a more promising approach to treating bone diseases relative to either monotherapy (Heiland et al., 2010). To date, no other study has reported the development of a bispecific antibody binding DKK1 (Table 1). However, the method for constructing bispecific antibodies will aid the future study of antibodies. For example, the development of multispecific antibodies combining the advantages of several antibodies is a faster route to overcome their disadvantages, especially when the defective monospecific antibodies target the same proteins.

Park et al. (2017) synthesized a cyclized oligopeptide consisting of 10 amino acids, with the amino acid composition Cys-Asn-Ser-Asn-Ser-Ile-Arg-Phe-Cys-Gly. This oligopeptide contained the Asn-X-Ile (NXI) motif that appears to be conserved in DKK1 (Park et al., 2017). As reported previously, the NXI motif closely matches the E1 domain of LRP5/6, which is critical in the binding between DKK1 and LRP5/6 (Bourhis et al., 2011). The cyclized oligopeptide specifically inhibited the DKK1-LRP5/6 interaction, but it did not abrogate the inhibitory effect of sclerostin on Wnt signaling, irrespective of the existence of a similar NXI motif in sclerostin. In an in vitro experiment, this oligopeptide presented bone-anabolic effects. To evaluate its effects on cancer cells in vivo, the MOPC315. BM.Luc MM mouse model was adopted. Treatment with this oligopeptide significantly reduced the tumor burden, whereas MM cell growth was not influenced directly. Although its mechanism of action on osteoblasts and MM cells remains uncertain, this oligopeptide sequesters DKK1 from LRP5/6 and reduces tumor cell growth, providing an option for managing tumor burden in MM (Park et al., 2017).

In addition to anti-DKK1 antibodies, endothelin-1 (ET-1), an endothelium-derived vasoconstrictor that mediates osteoblastic bone metastases, was found to target DKK1 and elicit responses by osteoblasts (Vaiou et al., 2016). When mouse neonatal calvariae were treated with ET-1, DKK1 mRNA levels and protein secretion were significantly decreased, and a robust increase in new bone formation was observed (Clines et al., 2007). Nonetheless, ET-1 can trigger resistance to bortezomib in human MM cells, and it has been suggested that ET-1 might be a key factor involved in the escape mechanism in MM patients treated with bortezomib (Vaiou et al., 2016). Another type of inhibitor suppresses DKK1 by inhibiting DKK1 mRNA levels, via a mechanism similar to that of microRNAs (miRNAs). For instance, protein arginine methyltransferase 5 binds to the promoter region of DKK1 and DKK3, and symmetrically methylates the H3R8 and H4R3 histones, achieving epigenetic silencing of DKK1 (Shailesh et al., 2021).

Anti-sense oligonucleotides are small pieces of DNA or RNA that hybridize to target molecules of RNA, blocking the transcription of RNA into protein by preventing ribosome recruitment and thereby modulating specific protein expression (Muntoni and Wood, 2011; Rinaldi and Wood, 2018). As a novel therapeutic intervention, antisense oligonucleotides have actually been widely utilized to inhibit various disease-related gene expressions. They are easy to make or modify and have shown higher success rates in targeting the source of the pathogenesis than drugs targeting downstream pathways (Baker et al., 1997; Leech et al., 2000). End-capped phosphorothioate DKK1 antisense oligonucleotides (DKK1-AS) include DKK1-AS1 and DKK1-AS2, which are complementary to nucleotides 4–21 and 243–260 of DKK1 mRNA respectively. They were utilized to alleviate DKK1 expression in osteoarthritic knees. A resultant significant decrease in DKK1 expression was observed, and the decline in BMD was considerably reversed (Weng et al., 2010; Weng et al., 2012; Wijesinghe et al., 2021). Before this, DKK1-AS had been found to abrogate loss of bone mass and biomechanical properties in ovariectomized rats; the regulatory mechanism seems to be the suppression of the ovariectomy-promoted osteoclastogenesis-regulatory activities of bone cells (Wang et al., 2007). This suggests that controlling DKK1 signaling using antisense oligonucleotides might be an alternative strategy in treating postmenopausal osteoporosis. To exclude the implication of DKK1 in kainate-induced seizures, Busceti et al. (2007) injected DKK1-AS into adult male Sprague-Dawley rats to achieve the knockdown of DKK1. The antisense treatment not only prevented DKK1 expression but also neuronal death in the olfactory cortex and hippocampus at 3 days (Busceti et al., 2007). Consistently, treatment with DKK1-AS prevented ischemic injury to hippocampal neurons and N-methyl-d-aspartate toxicity from cultured cortical neurons, thus demonstrating the great importance of DKK1 in the development of ischemic and excitotoxic neuronal death (Cappuccio et al., 2005). Taken collectively, these data demonstrate that rescuing Wnt signaling by DKK1 anti-sense oligonucleotides has therapeutic potential at least for treating bone diseases.

RNA interference (RNAi) is a prominent biological process of post-transcriptional gene silencing mediated by specific double-stranded RNA sequences, which turns off the expression of related genes (Tijsterman and Plasterk, 2004). After the enzymatic digestion of the input double-stranded RNA by DICER, the resulting siRNAs destruct the expression of homologous mRNA and achieve the effect of RNAi (Xia et al., 2002). Feng et al. (2019). reported that the transfection of DKK1-siRNA into U266 cells significantly decreased DKK1 expression relative to that in the control group, simultaneously reducing the activity of U266 cells and IL-6 expression. More significantly, transfecting MDA-MB-231 cells with DKK1 siRNA completely reversed the NBAT1-induced upregulation of DKK1 (Zhang et al., 2019). In an in vivo mouse model experiment, treatment with siRNA was capable of increasing osteocalcin content and stimulating osteoblast differentiation (Feng et al., 2019). Silencing DKK1 expression by transfecting a siRNA targeting human DKK1 (Hs_DKK1_1) rescued the dexamethasone-induced inhibition of osteoblast differentiation in primary human osteoblasts (Butler et al., 2010). Some pieces of evidence showed that when positive regulatory domain I (PRDM1) was overexpressed, DKK1 expression was unexpectedly enhanced. To determine that PRDM1 is independent of DKK1 regulation in glioma cells, siRNA was introduced for the purpose of silencing DKK1 expression. There were no side-effects resulting from treatment, which highlighted the biosafety of siRNA (Wang et al., 2013). In addition, siRNA was found to substantially downregulate DKK1 expression in esophageal cancer cells, which contributed to a decrease in abnormal cell proliferation (Yi et al., 2013). Therefore, siRNA-mediated RNAi targeting DKK1 might be a promising therapy for treating esophageal carcinoma (Lu et al., 2017b). Indeed, studies so far have merely used DKK1 siRNA as a tool for gene silencing and applied it to diverse disease models; thus, the corresponding drug development has a long way to go.

As highly conserved non-coding RNAs of approximately 17–25 nucleotides, miRNAs regulate the expression of target genes by binding to the 3′-UTR (untranslated region) (Inui et al., 2010). MiRNAs have been extensively studied in the past few years, and DKK1 expression has been found to be regulated by a variety of miRNAs (Supplementary Table S1). In oral squamous cell carcinoma (OSCC) cells, miR-1-3p suppressed the progression of OSCC by attenuating DKK1 expression and thereby triggering apoptosis, which implied its potential for cancer treatment (Wang et al., 2018c). MiR-29a and miR-335-5b are also effective regulators of DKK1 in osteoblasts, and there might be some complementarities between them and DKK1 mRNA (Kapinas et al., 2010; Zhang et al., 2011). Similarly, miR-203 showed regulatory effects on osteoblasts, but it could also further reverse the osteogenic differentiation in mesenchymal stem cells and participate in the pathogenesis of osteoporosis by binding to DKK1 (Xia et al., 2018). In osteosarcoma and adjacent tissues, DKK1 expression was negatively correlated with miR-107 (p = 0.01). By introducing mutations at predicted miR-107-binding sites, it was proved that the transcription of DKK1 could be hindered by binding of miR-107 to the 3′-UTR of DKK1 (Zhang et al., 2017). In colon cancer cells and HCC tissues, miR-217 directly interacted with the 3′-UTR of DKK1 and diminished DKK1 expression, constitutively activating Wnt signaling and promoting the tumorigenesis and progression of HCC. In turn, DKK1 overexpression antagonized the stimulatory role of miR-217 in Wnt signaling, further indicating the importance of miR-217 in HCC cells (Jiang et al., 2017; Jia et al., 2019). In mouse embryonic stem cells, overexpression of miR-290 resulted in lower expression of DKK1 and promoted cluster pluripotency against differentiation (Zovoilis et al., 2009). Likewise, DKK1 was identified as a direct target of miR-372 and miR-373 in breast cancer and CRC cells, and knockdown of DKK1 promoted the growth and invasive activity of cancer cells (Zhou et al., 2012). Many other miRNAs, in addition to the ones mentioned above, have been identified as related to DKK1, including miR-9, miR-31, miR-33a-5p, miR-34a, miR-92b, miR-101, miR-103a-3p, miR-130b-3p, miR-152, miR-291a-3p, miR-302b, miR-373-3p, miR-410, miR-433-3p, miR-488, miR-493, miR-501-5p, miR-522, miR-523-3p, miR-543, miR-590-3p, miR-613, miR-3064-3p, and miR-BART10-3p (Liu et al., 2016b; Fan et al., 2016; Jia et al., 2016; Zhang et al., 2016; Feng et al., 2017; Lv et al., 2017; Tang et al., 2017; Weng et al., 2017; Wang et al., 2018a; Wang et al., 2018b; Chen et al., 2018; Zhao et al., 2018; Liu et al., 2019; Min and Lee, 2019; Wang et al., 2019; Li et al., 2020a; Li et al., 2020b; Liu et al., 2020; Xiang et al., 2020; Zeng et al., 2020; Zhou et al., 2020; Liao et al., 2021; Song et al., 2021; Wu et al., 2021).

DNA vaccination is a new approach to directly induce humoral and/or cellular immune responses without live vectors. For the generation of DNA vaccines, the antigen-encoding gene is inserted into a plasmid with an appropriate eukaryotic promoter, and the obtained recombinant plasmids are then administrated to animals for the production of antigens, which will elicit immune responses (Lu et al., 2008; Hasson et al., 2015). To determine whether the immunity elicited by a DKK1 vaccine could alleviate myeloma, a DNA (murine DKK1/defensin-2 fusion) vaccine was employed in the murine MOPC-21 myeloma model. Although only three of ten mice injected with the DNA vaccine survived in prophylactic studies, the percentage of mice with no tumor burden increased to 60% after treatment with the DNA vaccine plus adjuvant CpG. In the tumor inhibition experiment, the therapeutic efficacy was remarkable (Table 2). Concurrently, CD⁴⁺ and CD⁸⁺ T cells in vaccinated mice were activated, resulting in stronger tumor-specific responses compared with those of the mice without vaccination (Qian and Yi, 2012; Liu et al., 2018). Zhang et al. (2015a) constructed a multiepitope DNA vaccine against DKK1 by screening and optimizing four B cell epitopes; the results implied that vaccination could attenuate bone erosion relevant to DKK1 in collagen-induced arthritis (CIA) mice and lower the disease incidence of arthritis in CIA-susceptible mice. To further facilitate the efficiency of the DNA vaccine and circumvent the limitations of the conventional DNA-immunization approach, a novel hybrid phage particle was used to carry the DKK1 gene (Khalili et al., 2018). The phage-immunization results showed higher efficiency of the immunization vehicle and stronger humoral immune responses in contrast to classical DNA-immunization vectors, which meant that larger amounts of antibodies against DKK1 were produced. By means of combining epitopes originating from MM special antigen-1 (MMSA-1) and DKK1, Lu et al. (2017a) constructed a multi-epitope peptide vaccine that improved the immune response and attenuated myeloma growth. Indeed, peptide vaccines have presented impressive potential in provoking immune reactions and eliminating cancer cells (Perez et al., 2010; Walz et al., 2015). Wu et al. (2014b). explored the inhibitory effects of DKK1-derived peptides on cell differentiation and bone metabolism in MC3T3-E1 pre-osteoblastic cells and found increased bone mass, improved bone microstructure, reduced bone loss, and decreased mineralization in mice vaccinated with DDPs. Remarkably, the serum anti-DKK1 antibody titers were distinctly increased, with augmentation of more than 100,000 times compared with average titers 6 weeks after the first injection. Therefore, whether it is better to develop DNA vaccines or other traditional vaccines depends on the final clinical effects (Wu et al., 2014b).

Long noncoding RNA (lncRNA) LINC01133 was shown to methylate the DKK1 promoter, and its upregulation was associated with increased DKK1 methylation and decreased DKK1 expression in pancreatic cancer (Weng et al., 2019). Another lncRNA, HOTTIP, also mediated DKK1 downregulation in CRC cells (Rui et al., 2018). Two genes with longer sequences, muscle segment homeobox 1 (MSX1) and GATA6, suppressed DKK1 transcription by directly binding to the DKK1 promoter (Zhong et al., 2011; Menezes et al., 2012). Likewise, c-Myc, which is a frequently amplified gene in breast cancers, suppressed DKK1 and transformed human mammary epithelial cells (Cowling et al., 2007).