Discoidin domain receptors (DDRs) are novel receptor tyrosine kinases (RTKs) discovered by Johnson et al. in 1993 (1). As a RTK family members, DDRs also include two subfamily members: discoidin domain receptor 1 (DDR1) and discoidin domain receptor 2 (DDR2) (2). Similar to other RTK family members, the structure of DDRs includes an extracellular region, a transmembrane (TM) domain, and an intracellular kinase region. Unlike other members of the RTK family, the extracellular regions of DDRs contain a globular discoidin (DS) domain of 155 amino acids, a discoidin-like (DS-like) domain and a long extracellular juxtamembrane (JM) region between the DS domain and the TM domain (3, 4). The JM segment is responsible for transmitting extracellular signals to the intracellular tyrosine kinase region (5). Both of these molecular structure features indicate that DDRs can bind to their ligands through an unusual transmembrane mechanism and play a vital role in cell biology. Based on the alternative splicing of intracellular kinase mRNA, there are five subtypes of DDR1: a, b, c, d, and e. However, only one single DDR2 subtype exists. There is kinase activity in the DDR1a, DDR1b and DDR1c coding sequences but not in the DDR1d or DDR1e coding sequences due to the deletions of exon 11 and exon 12, respectively (Figure 1) (6, 7). Notably, in addition to their role in collagen synthesis and degradation, DDRs promote tumour cell adhesion, migration, and differentiation and the release of inflammatory factors (8, 9).

However, the underlying mechanisms of DDRs in premalignant and malignant liver disease remain obscure. In this review, the characteristics of the structure, activation and phosphotyrosine-based interactions of DDRs are briefly introduced, and the molecular mechanism of DDRs in the liver are systematically discussed, with the aim of promoting the development of DDR-targeted therapies and new cancer treatment methods.

Most RTKs are activated by phosphorylated tyrosine residues within seconds of ligand binding, followed by a rapid decline in activity due to dephosphorylation or internalization and degradation of the receptor/ligand (10). Collagen, a key constituent of the extracellular matrix, is the specific ligand of DDRs, which are not activated by soluble peptide-like growth factors. The DS domain of DDRs binds with collagen (11). Unlike the signal transduction of other typical RTKs with acute and rapid responses, the binding of collagen to DDRs is slow and continuous (4). However, only collagen with a natural triple helical structure, but not heat-denatured collagen, can bind to DDRs (11, 12). The DS domain of DDR1 and DDR2 recognizes the specific amino acid sequence motif GVMGFO (O, hydroxyproline) on fibrillar collagen types I, II and III (13, 14). Collagen of the basement membrane (type IV) only activates DDR1, whereas collagens of types V and X only interact with DDR2 (15–17).

Although their single DS domain contains sites for binding to collagen, DDRs require dimerization of the DS domain to bind collagen with high affinity (11). Strikingly, DDRs develop independent and stable dimers mediated by one leucine-based sequence motif in the transmembrane domains in the absence of ligand recognition, which distinguishes them from other RTKs (18). Furthermore, mutation of cysteine residues in the extracellular JM region in DDRs results in independent and covalent dimers, which occur during biosynthesis, according to the study authors (19). The collagen binding region of charged residues surrounded by the hydrophobic layer in the three surface-exposed loops of the DS domain is activated by forming lateral clusters after binding to collagen ( Figure 2 ) (11, 20). In other words, the activation of DDRs and the lateral clusters between DDR proteins are reciprocal causations. The lateral clusters reinforce the binding of DDRs to collagen and induce the activation of DDRs, while the lateral clusters leading to trans-phosphorylation between dimers are the result of collagen binding (18, 21). DDRs regulate adhesion and traction force on collagen by binding to myosin IIA, which condenses collagen fibrils into a denser alignment (22, 23). Activation of the DDR kinase domain enhances the association of the kinase domain with myosin IIA filaments, optimizing myosin-dependent contractile force delivery to collagen fibrils (21). The magnitude of DDR activation is proportional to collagen contractile forces (21). Upon binding of DDRs to collagen, autophosphorylation of the intracellular JM region and tyrosine residues of the intracellular kinase region in DDRs occurs (10). This phosphorylation recruits intracellular signalling protein complexes, such as Src Homology-23 (SH2/3) and the phosphotyrosine binding (PTB) domain, for assembling and transmitting receptor signals that play a cellular signal transducing function (11). Moreover, there is a non-canonical but functional crosstalk between insulin/insulin-like growth factor system (IIGFS) and DDRs, which was recently discovered apart from canonical collagen-dependent DDR activation. Interestingly, DDR1 functionally interacts with IIGFS better than DDR2 (24). As a complex network, IIGFS is constituted of transmembrane receptors, corresponding ligands, and binding proteins (24). Insulin-like growth factor 1 receptor (IGF1R) and insulin receptor (IR)-A, which are members of transmembrane receptors of IIGFS, and their common ligands consisting of insulin-like growth factors (IGF1 and IGF2) and insulin, are involved in the crosstalk with DDR1 (24). Upon stimulation by cognate ligands, IGF1R or IR-A physically interacts with DDR1 to induce rapid and sustained DDR1 phosphorylation, independent of the binding capacity of DDR1 to collagen (25). IIGFS not only stimulates DDR1 phosphorylation but also up-regulates DDR1 protein level to some extent via the activation of phosphoinositide 3-kinase (PI3K)/protein kinase B (Akt) cascade and further inhibition of downstream miR-199a-5p, a negative regulator of DDR1 (26).

Upon ligand activation, distinct tyrosine residues in the intracellular JM and kinase regions are phosphorylated and act as docking sites for SH2/3 or PTB, which can serve as signal adaptors to assemble intracellular signalling molecules (Figure 3) (10). DDR1a has 13 tyrosine residues in the cytosolic JM and kinase regions, while both DDR1b and DDR1c have 15, and DDR2 has 14. It is beyond the scope of this paper to go into their tyrosine residues, given that both DDR1d and DDR1e, with an incomplete JM domain or the absence of an entire kinase domain, are nonfunctional truncated proteins (27). Interactors of DDRs localized on the phosphotyrosine residues of the receptor have been identified. Lemeer S et al. (28) outlined the tyrosine residue sites and corresponding interactors in the intracellular JM and kinase regions based on phosphopeptide affinity purifications in human placenta. For example, Crk2 (Tyr520), ShcA (Tyr513), C-terminal Src kinase (Csk) (Tyr520), STATs (Tyr543 and Tyr 547), Nck1/2 (Tyr569), phosphoinositide 3-kinase (PI3K) (Tyr703) and other phosphotyrosine interactors containing SH2/3 and PTB domains were detected to point to the docking sites of corresponding phosphotyrosine residues in the DDR1 receptor to transduce early signals from the receptor to the downstream cascade (28). Given that DDR1a lacks Tyr513 and Tyr520 docking sites compared with DDR1b/c, adaptor molecules, such as ShcA and Csk, associated with specific docking sites cannot bind to DDR1a (12). Regarding DDR2, Iwai LK et al. (29) identified two new docking sites in the DDR2 kinase domain, namely, Tyr684 and Tyr813. Interestingly, the docking site of Tyr481 in the JM region was shown to be constitutively phosphorylated, and Tyr471 was found to be a docking point for the adaptor ShcA (29, 30). Moreover, novel adaptor molecules, such as Grb2 and EphA2, were confirmed using a DDR1b immunoprecipitation assay (28). In addition to known adaptors with well-defined functions, other new adapters appear to involve RasGAP and VAV2/3. However, whether the new adaptor molecules bind to collagen-dependent activated receptors and generate signalling effects following binding remains to be seen.

Several cell types express DDR1, mainly epithelial cells and smooth muscle cells, as well as other types of cells, such as the epithelium of the large intestine, lung and breast cells, adrenal cortical cells, pancreatic ducts, and macrophages (31, 32). DDR2 is mainly expressed in connective tissues of mesenchymal origin, including in cartilage and smooth muscle cells, fibroblasts and myofibroblasts, but also in skeletal muscle, kidney, myocardium, and lung cells (2, 23, 33, 34). Both DDR1 and DDR2 play important roles in the process of embryonic development. DDR1 is mainly involved in organ development and maturation, while DDR2 is involved in bone growth in humans (35). The physiological effects of DDRs do not persist into human adulthood, even though they play important physiological roles during embryonic development (27). Their physiological effects are not carefully elaborated here. The pathological functions of DDRs in premalignant and malignant liver diseases are described in the following sections ( Table 1 ).

The slow development of liver fibrosis associated with a variety of chronic diseases usually takes years (83). There is a slow transition from normal tissue repair to profibrotic tissue and tissue remodelling resulting from a prolonged lesion (84). Early diagnosis and early treatment of liver fibrosis need urgent improvements, and antifibrotic treatment remains stagnant. In addition, the emerging evidence described in this review suggests that DDRs play a crucial role in tumour progression and metastasis. Therefore, this section explains how DDR kinase inhibitors and multikinase inhibitors are used in the treatment of fibrosis and cancer. The effects of multikinase inhibitors may be mediated in part through inhibiting DDRs.

Most DDR inhibitors tend to be adenosine triphosphate (ATP)-competitive inhibitors, which can act on both DDR1 and DDR2 with poor kinase selectivity (85). The multikinase inhibitors dasatinib, imatinib, nilotinib, bafetinib, ponatinib, MGCD516 (sitravatinib) and LCB 03-0110 have been shown to inhibit both DDR1 and DDR2 kinase activity and are mainly applied in the treatment of chronic myeloid leukaemia (CML), lung adenocarcinoma, and colorectal cancer, among other cancer types (86–91). Among them, dasatinib can inhibit the proliferation of squamous cell carcinoma (SQCC) with DDR2 mutation, and ponatinib is prescribed to treat imatinib-resistant patients with CML (88, 92). Concerning DDR kinase inhibitors, the compound 7f is representative of a range of 2-amino-2,3-dihydro-1H-indene-5-carboxamide derivatives; compounds 2a, 4a, and 4b belong to a family of pyrazolourea-containing compounds; and DDR1-IN-1 and DDR1-IN-2 exhibit notable inhibitory effects on DDR1 and DDR2 kinase activity in cancer (93–95). WRG-28 has also been shown to inhibit tumour invasion and metastasis as a selective DDR2 small-molecule inhibitor (96). Moreover, Chen C et al. (97) reported that the novel DDR1 inhibitor AH-487/41940522 can inhibit collagen deposition in a rat model of idiopathic pulmonary fibrosis. Nilotinib has been shown to block the association of myosin IIA with the DDR1 kinase domain to reduce tractional collagen contraction in vitro (Figure 4) (21). Other potent lead compounds for inhibiting selectively DDRs include compounds 13-21 (98). Among them, compounds 13, 18, and 21 are representative of 1, 2, 3, 4-tetrahydroisoquinoline derivatives, pyrazole fused pyrimidine derivatives, and indazole derivatives against DDR1, respectively; compounds 14-16, compound 17, and compound 19 separately belong to di-amide derivatives, dasatinib analogs and indole analogs against DDR1 and DDR2; and the compound 20 derives from pyridine derivatives against DDR2 (98). Currently, several clinical trials have been conducted with drugs such as dasatinib, nilotinib, and sitravatinib for different cancers, including squamous cell lung carcinoma, advanced or refractory lymphoma, advanced solid malignancies, hematopoietic neoplasm, etc (99–104).

The mechanisms of drug resistance linked to DDRs are associated with DDR mutations. Part of DDR inhibitors loses their inhibitory activity on the background of DDR mutations. The aminopyrimidine head group of imatinib binds to the “gatekeeper” residue Thr701 in DDR1 via the hydrogen bond. In CML, the mutation of the “gatekeeper” Thr701 in DDR1 confers steric hindrance and causes resistance to imatinib (105). Ponatinib has been proven to circumvent the steric hindrance and may be selected for the treatment of imatinib-resistant CML patients (86). Considering that Ponatinib is much less selective than imatinib, DDR1-IN-1, therefore, is designed to have a similar pharmacophore model while conferring highly a selective pharmacological profile (94). Beauchamp et al. (106), found that the mutation of the “gatekeeper” T654I in DDR2 and loss of NF1 conferred acquired resistance to dasatinib in lung cancer cell lines via targeted exome sequencing. Although the phosphorylation of DDR2 in lung cancer cell lines was decreased by the administration of dasatinib, the acquired mutation in DDR2 T654I blocked this therapeutic effect. And, the loss of NF1 activated a bypass pathway that conferred the activation of the RAS signaling pathway. WRG-28 has been proven to inhibit the collagen-stimulated phosphorylation of DDR2 with T654I mutation via its allosteric action upon the extracellular domain (96). Interestingly, a case report by Pitini et al. (107), found a significant reduction in tumor size during administration of dasatinib in one patient with lung squamous SQCC accompanied by S768R mutation of DDR2 kinase gene. However, The high-frequency occurrence of dasatinib toxicity in clinical trials should be noted. Moreover, advanced melanoma with acquired BRAF (V600) mutations is resistant to initially effective BRAF/MEK inhibitors, which is the main reason restricting patient benefit. The key lies in the reactivation of alternate signaling networks, in which the DDR-mediated extracellular matrix is an important component of cancer cell adaptation and resistance to targeted therapy (108, 109). Thus, DDR-targeted therapy can enhance the efficacy of BRAF-targeted therapy to overcome drug resistance.

Here, we discussed the roles of DDR1 and DDR2 in premalignant liver diseases, primary HCC and metastatic liver cancer derived from several other cancer types. Briefly, the progression of liver fibrosis can be promoted when DDR1 is overexpressed in hepatocytes, and invasion, migration and liver metastasis can be stimulated when DDR1 is overexpressed in tumour cells. However, DDR2 signalling in HSCs can protect against chronic liver fibrosis progression but not early-stage liver injury (prefibrotic stage) and can inhibit liver metastasis of colorectal cancer. These perspectives are critically significant and are first described in detail in this review. Based on relevant original studies, DDR2 can also be expressed in HCC cells and plays a pathogenic role. Although the pathological functions of DDRs in premalignant and malignant liver diseases have been reported, to aid in the development of more targeted DDR inhibitors and facilitate their application in clinical practice, additional studies exploring the concrete molecular mechanisms of DDRs are needed.

HG: Writing-original draft. H-MX: Designing some contents of the manuscript. D-KZ: Writing, and revising the manuscript. All authors contributed to the article and approved the submitted version.