CD146, from a melanoma cell adhesion molecule to a signaling receptor

Laminins 411 and 421

Laminins are a family of large heterotrimeric αβγ proteins with over 15 different isoforms. Five laminin α chains (α1–α5), four laminin β chains (β1–β4), and three laminin γ chains (γ1–γ3) constitute αβγ heterotrimers. They are denominated according to their chain composition; for example, laminin α4β1γ1 is designated as Laminin 411.⁶¹ Laminins are predominantly found in basement membranes that compartmentalize different tissues and surround blood vessels, nerves, and adipocytes.^62,63 They play a crucial role in physiological and pathological remodeling of the ECM during angiogenesis, wound healing, embryogenesis, and tumor metastasis. Remodeling of the ECM during metastasis allows tumor cells to invade their surrounding ECM, spread via the vascular or lymphatic circulation, and extravasate into distant organs.

Laminin isoforms, particularly the laminin α chain, are expressed in a cell and tissue specific manner and are distinctly bound by almost ten different integrins and other cell-surface receptors.^62,63 The α4-laminins are mesenchymal laminins expressed by the cells of mesenchymal origin, such as vascular and lymphatic endothelial cells, pericytes, and leukocytes, and are required for normal development of the cardiovascular and neurological system in mice.^64–66 Under pathological conditions, α4-laminins are expressed and secreted by various tumor cells, such as melanoma and glioma,^67–72 or tumor stroma, lymphatic and vascular vessels, nervous system.^69,73,74

Laminin 411

Laminin 411 is expressed along the vascular endothelium.^68,73,75 This laminin isoform is recognized by various integrins, including α6β1, α3β1, α6β4, and αVβ3, which promote the migration of several cell types along vascular or nervous system tracks.^76–80

In 2012, Laminin 411 on the vascular endothelia was discovered as a specific ligand for CD146 on a subset of human CD4⁺ T helper (Th) cells.⁵¹ This subset of human T cells expresses CD146 and can enter tissues to promote pathogenic autoimmune responses. To determine the CAMs involved in the migratory capacity of Th17 cells into tissues, researchers used purified Laminin 411 to identify its receptor. In this study, the authors demonstrated that purified CD146-Fc binds to Laminin 411 with high affinity (27 nM) and that this binding disappeared when the endogenous Laminin 411 was specifically deleted. Correspondingly, blocking this binding by CD146 antibody in vivo also reduced Th17 lymphocyte infiltration into the CNS. Therefore, the authors concluded that Laminin 411 is a major tissue ligand for CD146⁺ lymphocyte.

However, the role of Th17 cells in the pathogenesis of malignant tumors is still remains controversial. Some studies revealed that increased percentage of Th17 lymphocytes among cells infiltrating ovarian cancer cells stimulate tumor progression;⁸¹ whereas other studies showed that Th17 lymphocytes have anticancer activity and can reduce tumor growth and metastasis.⁸² Therefore, the roles of CD146⁺ Th17 cells in cancer development may be worthy of further investigations.

Laminin 421

CD146 is a reliable biomarker of endothelia and is concentrated at the intercellular junctions of endothelial cells of vessel system.²¹ Most cancer cells, including melanoma, migrate along the abluminal sides of vascular and/or lymphatic vessels, as they disseminate throughout the body.⁸³ Laminin 421 is major laminins of along the tumor-dissemination tracks (blood and lymphatic vessels, nerves, and tumor stroma).^84–86

To determine the mechanism of CD146 roles in metastasis, researchers used melanoma cells to test what laminin isoforms, other than Laminin 411, can bind with the melanoma marker of CD146. Therefore, they used all laminin α chains to examine the binding affinity with human CD146 in a solid-phase ligand binding assay.⁸⁷ Finally, they found that only Laminin 421, of several laminin isoforms, readily bound to CD146, suggesting that Laminin 421 is a primary ligand for CD146 in melanoma. Accordingly, a function-blocking mAb to CD146 inhibited tumor cell migration on Laminin 421, but not on laminins 411 or 521. In addition, this investigation determined that the identity of NOF, previously identified as a ligand for chicken CD146 (gicerin), is actually Laminin 421.

In this study, the authors also determined that Laminin 411 and especially Laminin 421 are capable of stimulating migration of a broad panel of cancer cell lines through a filter. This investigation is consistent with the observation that the α4-laminins, including Laminins 411 and 421, expressed and secreted by various carcinoma cells, have already emerged as “onco-laminins.”^67–72 Melanoma CD146 binds with Laminin 421 but not 411, whereas lymphocyte CD146 only binds with Laminin 411; suggesting that the epitopes of CD146 on somatic cancer cells are different from those of CD146 on blood lymphocytes. Therefore, the infiltration of CD146⁺ invasive cancers into tumor-dissemination tracks is likely dependent on the interaction between CD146 and Laminin 421, and blocking their binding may affect the efficacy of cell–cell interactions and interfere metastasis.

Galectin-1 and -3

CD146 is a highly glycosylated junctional CAM involved in the control of vascular vessel integrity. Sequence analysis predicts the presence of eight putative N-glycosylation sites at residue positions 56, 418, 449, 467, 508, 518, 527, and 544.²¹ It has been estimated that 35% of the CD146 molecular mass is attributed to glycans.⁸⁸ The galactose residues in glycans can bind with galectins, and such binding can be inhibited by lactose.

Galectins are a family of soluble carbohydrate-binding lectins that modulate cell-to-cell and cell-to-ECM adhesions.⁸⁹ Up to now, 15 galectins have been identified in mammals and 11 are found in humans. Among them, Galectin-1, -3, and -9 are three best-investigated galectins and Galectin-1 and -3 promote tumor development, progression, and immune escape.⁹⁰ Galectin-1 and -3 can hamper antitumor responses and are considered multifunctional targets for cancer therapy.^91,92 The underlying mechanisms include interfering with drug efficacy/delivery or reducing the antitumor effect of immune cells. For instance, Galectin-1 confers drug resistance via inducing the expression of multidrug resistance protein 1, which in turn helps tumor cells to pump out cytotoxic drugs, facilitating cancer cells to combat anticancer drugs.⁹³ Regarding the immunosuppressive effects of Galectin-1 and -3 on T cells, in a mouse melanoma model, targeted inhibition of Galectin-1 enhanced T cell-mediated tumor clearance;⁹⁴ Galectin-3 can neutralize glycosylated IFN-γ in tumor matrices, ablating the immune response to tumors.⁹⁵ To increase overall responsiveness of tumors to chemo- or immune-therapy, inhibitors of Galectin-1^96,97 and -3^98,99 have been used in combination with anti-CTLA-4 or anti-PD-1 to treat cancer patients in clinical trials.

Because both CD146 and galectins are involved in the modulation of angiogenesis, researchers hypothesized that some galectins may be the ligands of CD146 and the interactions between them are required for functional CD146 in angiogenesis, as well as in cancer metastasis. To date, two galectins, 1 and 3, have been identified as the ligands of CD146.

Galectin-1

Galectin-1 prefers to bind with the branched N-glycans of cell-surface glycoproteins and mediates a glycosylation-dependent angiogenesis.^91,100–103 It has been reported that increased secretion of Galectin-1 in the ECM facilitates cancer cell proliferation and resistance to cancer therapy in prostate cancer¹⁰⁴ and Kaposi’s sarcoma.¹⁰⁵ Mechanistic investigation has revealed that Galectin-1 can bind to N-glycans on VEGFR2 to activate VEGF-like signaling in anti-VEGF-A refractory tumors, promoting tumor progression. Accordingly, disruption of the Galectin-1-N-glycan axis inhibits tumor growth by promoting vascular remodeling.¹⁰¹ This research highlights the importance of Galectin-1 in tumor angiogenesis and cancer metastasis. However, these studies cannot exclude the fact that other cell-surface proteins with branched N-glycans are also involved in this glycosylation-dependent pro-angiogenesis pathway.

Early in 2011, it was reported that the co-expression of Galectin-1 and CD146 is required for tumor vascularization in a human mesenchymal stem cell strain with significant angiogenic potential.¹⁰⁶ In 2013, Jouve et al. reported that Galectin-1 binds to CD146 on endothelial cells, facilitating cell survival.¹⁰⁷ In this report, they explained that CD146 glycosylation is mainly composed of branched N-glycans. They showed that the interaction of CD146 with Galectin-1 is carbohydrate-mediated using both an enzyme-linked immunosorbent assay and surface plasmon resonance assays. In addition, they demonstrated that the interaction between Galectin-1 and CD146 protects endothelial cells against apoptosis induced by Galectin-1. Thus, it is tempting to speculate that CD146 could be a decoy receptor for Galectin-1, preventing the Galectin-1 from binding to pro-apoptotic receptors.¹⁰⁷ However, whether this interaction affects tumor cell survival remains unknown. In 2015, Yazawa et al. thus further analyzed the functions of this interaction on melanoma and found that when Galectin-1 binds to CD146, it helps maintain intrinsic malignant features.¹⁰⁸ The authors examined the expression, identity, and function of Galectin-1 ligands in melanoma progression and demonstrated that CD146 is the major Galectin-1 ligand on melanoma cells.

These findings provide a perspective on the interactions between CD146 and its ligands, such as Galectin-1, as contributors to cancer malignancy. Indeed, various membrane glycoconjugates have been identified as binding partners of Galectin-1 such as β1 integrins, CD2, CD3, CD4, CD43, CD45, and GM1 ganglioside. In addition, Galectin-1 can bind to a number of ECM components in a dose-dependent and β-galactoside-dependent manner. For instance, laminin and fibronectin, which are highly N-glycosylated, interact with Galectin-1.¹⁰⁹ Because it has been reported that CD146 can interact with Laminin 411, Laminin 421, and β1 integrin, it is reasonable to speculate that CD146 may also interact with all of those Galectin-1 interactors within cancerous cells.

Since the tumor vasculature is an easily accessible target for cancer therapy, understanding how galectins influence cancer angiogenesis is important for the translational development of therapies intended to prevent tumor progression. Based on the fact that VEGF-targeted therapies often fail when tumors receive continued treatment,¹¹⁰ targeting glycosylation-dependent Galectin-1-receptor interactions, such as Galectin-1-CD146-VEGFR2 may increase the efficacy of anti-VEGF treatment.

Galectin-3

Like Galectin-1, Galectin-3 can also bind to various galactose-terminated glycans of cell-surface receptors and proteins of ECM and is involved in many physiological and pathological processes from cell adhesion and migration to cell activation.^111,112 In cancer cells, it modulates cell–cell and cell–microenvironment communications, contributing to cancer development, progression, and metastasis.^113–120 Patients with metastatic diseases tend to have higher concentrations of circulating Galectin-3 than those with only localized tumors.¹²¹ Increased circulating Galectin-3 promotes blood-borne metastasis due to the interaction of Galectin-3 with receptors on vascular endothelial cells, further causing endothelial secretion of several metastasis-promoting cytokines.

To identify the Galectin-3-binding molecules on the endothelial cell surface, using the Galectin-3 affinity purification method, Colomb et al. found that CD146 was the major cell-surface receptor to strongly bind and co-localize with Galectin-3, compared with other glycosylated receptors, CD31, CD144, and CD106. They also showed that Galectin-3 bound to N-linked glycans on CD146 and induced CD146 dimerization and subsequent activation of protein kinase B (AKT) signaling. Correspondingly, suppression of CD146 expression abolishes Galectin-3-induced secretion of metastasis-promoting cytokine from the endothelial cells. Thus, they concluded that CD146 is the functional Galectin-3-binding receptor on the endothelial cell surface responsible for Galectin-3-induced secretion of cytokines, and therefore influences cancer progression and metastasis.¹²²

Subsequently, the binding moieties of CD146 by Galectin-3 have been further identified. The authors demonstrated that Galectin-3 interacts with the highly glycosylated Domain 5 in the CD146 extracellular fragment regardless of the presence or absence of lactose. These findings provide a better understanding of how Galectin-3 interacts with cell-surface receptors to mediate endothelial cell migration and the secretion of cytokines.^123,124

The endothelial galectins are confined to four family members, i.e., Galectin-1, -3, -8, and -9, which contribute to tumor angiogenesis.⁹² Tumor-induced angiogenesis is a pathologic condition in which tumor cells secrete growth factors, such as VEGFs, to promote the growth of new blood vessels.^125,126 These growth factors activate quiescent endothelial cells in host tissue to facilitate them to invade into the tumor stroma for growth of new capillaries.¹²⁷ Endothelial galectins binding with glycoconjugates on tumors are involved in different processes during tumor-induced angiogenesis. Because Galectin-1 and -3 binding of glycoconjugates on tumor cells mediates many key processes in angiogenesis and elevated levels of Galectin-1 and -3 in the endothelium are correlated with tumor vascularization,^{105,128–131} the promotion of tumor vascular remodeling by tumor CD146 may be due to the interactions between CD146 with Galectin-1 and -3.

S100A8/A9

Matriptase

Matriptase is an epithelial-specific membrane-anchored serine protease that proteolytically degrades targets, such as ECM components and the pro-forms of growth factors.^183–186 Because most of solid tumors are originated from epithelia, matriptase is thus critically involved in cancer invasive growth through degradation events related to breaching the basement membrane, reorganization of the ECM, and activation of oncogenic signaling pathways.¹⁸⁷

During neurogenesis, matriptase, expressed on neural stem/progenitor (NS/P), plays a critical role in cell-contact signaling between NS/P and brain endothelial cells.¹⁸⁸ In 2017, the direct binding between brain endothelial CD146 and NS/P matriptase, was identified to be involved in the direct endothelia-NS/P contact.¹⁸⁹ Such binding can activate the downstream signaling cascades from CD146, including p38 and canonical Wnt/β-catenin pathways in endothelia, leading to secretion of cytokines and chemokines. These factors in turn act on NS/P cells to induce differentiation and migration for the adequate neurogenesis. Consistently, none of these signaling events occurred when either matriptase or CD146 is deleted. Thus, this study suggests that CD146, expressed on stromal cells in tumor microenvironment, plays essential roles in tumor invasion by interaction with matriptase expressed on invading cancer cells originated from epithelia.

Recent years, matriptase has received considerable attention in the field of cancer research, because it is upregulated in many cancers and is required for the degradation of the ECM and the maturation of a variety of oncogenic pro-growth factors.¹⁸⁴ Mice with reduced levels of matriptase display a significant delay in oncogene-induced mammary tumor formation and growth.^190,191 Therefore, interference of the interaction between CD146 on stromal cells and matriptase on cancers is a reasonable strategy to turn off the signaling pathway and invasive responses in cancers, especially epithelia-originated neoplasms.

CD146 as the co-receptor for VEGFR2 in tumor angiogenesis

There are five members of the human VEGF family: VEGF-A, -B, -C, -D and placental growth factor (PlGF).^196,197 VEGF-A is the best characterized family member and the most potent stimulator of angiogenic processes.¹⁹² VEGF-A has at least nine different splicing forms with different binding affinities within its receptors and ECM components.¹⁹⁸ VEGF-A is secreted by many cell types, such as endothelial cells,^199,200 fibroblasts,²⁰¹ smooth muscle cells,²⁰² platelets,²⁰³ neutrophils,²⁰⁴ macrophages, and ~60% of all tumors.²⁰⁵ VEGF-A secretion is induced by ischemia and inflammatory stimuli.²⁰⁶ Cellular responses to VEGF-A are mainly driven by their binding to its cognate receptors—the VEGFRs.

VEGFRs belong to the class IV receptor tyrosine kinase family and are expressed by endothelial cells, macrophages, hematopoietic cells and smooth muscle cells.^207–210 In humans, there are three VEGFR subtypes, which are encoded by separate genes: VEGFR1 (FLT1), VEGFR2 (KDR), and VEGFR3 (FLT4). Among them, VEGFR1 and VEGFR2 show high structural similarity.^198,211 In endothelial cells, VEGFR2, but not VEGFR1, mediates the full range of VEGF responses.^212,213

VEGFR2 activation leads to distinct activation patterns, including proliferation via mitogen activated protein kinase (MAPK),²¹⁴ cell migration via phosphoinositide-specific phospholipase C (PLC)-γ²¹⁵ and focal adhesion kinase (FAK),^216,217 and cell survival through phosphatidylinositol-4,5-bisphosphate 3-kinase (PI3K)/AKT.²¹⁸ In addition, VEGFR2 interaction with its co-receptors is essential for its functions. Till now, the well-established VEGFR2 co-receptors include Neuropilin-1, CD44, vascular endothelial cadherin, and β integrins.²¹⁹ Although the exact molecular mechanisms of how VEGFR2 induces diverging downstream signals have not been completely understood, ligand diversity and availability as well as interactions with co-receptors might explain most of these effects.

Similar to the functions of VEGFR2 in endothelia, CD146 also contributes to endothelial cell proliferation, migration, and angiogenesis.^220–222 Different from VEGFR2 with kinase activity, CD146 does not show any kinase activity. But similarly to VEGFR2, CD146 also can activate numerous signaling pathways, known as the downstream cascades of VEGFR2, such as focal adhesion formation mediated by FAK,¹⁴ cell migration mediated by PLC-γ,¹⁴ cell survival through PI3K/AKT,²²³ proliferation via MAPK and NF-κB.²²⁴

In 2012, we found that CD146 interacts with VEGFR2 on endothelia and acts as a co-receptor.⁵³ This interaction occurs in the extracellular protein domain of CD146 independently of VEGF-A. When CD146 was inhibited using the blocking antibody AA98 or CD146 siRNA, VEGF-A induced phosphorylation of VEGFR2 was suppressed in human umbilical vein endothelial cells (HUVECs), demonstrating that the interplay of CD146 with VEGFR2 is mandatory for functional VEGFR2 signaling.

In addition to the extracellular binding of CD146 to VEGFR2, intracellular CD146 signaling is also required for VEGF-A-induced signal transduction via VEGFR2. When associated with VEGFR2, the cytoplasmic tail of CD146 recruited ERM proteins and the actin cytoskeleton, to assemble a “signalosome,” which is required for signal transduction from VEGFR2 to AKT and P38 MAPK. Inhibition of CD146 by blocking antibody AA98 hinders the interaction between VEGFR2 and CD146, resulting in abrogation of the downstream cascade of p38 MAPK and AKT signaling. Therefore, our study revealed that the interaction of CD146 with VEGFR2 improves pro-angiogenesis signaling as part of such a signalosome in which CD146 binding to VEGFR2 enables the reorganization of the cytoskeletal network during angiogenesis.

Using a preclinical pancreas carcinoma xenograft model, we found that the efficacy of combined treatment with the anti-VEGF antibody Bevacizumab and the anti-CD146 antibody AA98 was significantly higher than treatment with either one of the two agents. The clinical benefit from VEGF-targeted therapies is always compromised during continued treatment with anti-VEGF antibodies, such as Bevacizumab in pancreatic carcinoma.^110,225,226 Thus, our findings indicate the possibility that the clinical efficacy of anti-VEGF therapy can be augmented by targeting CD146 treatment in the future.

Several mechanisms by which tumor angiogenesis may proceed in the presence of anti-VEGF have been identified. For example, pro-inflammatory cytokines and Galectin-1 produced from tumors activate VEGF-like pro-angiogenic pathways.^101,227 Based on our findings, we speculate that the interaction between CD146 and VEGFR2 might also underlie the mechanisms of anti-VEGF treatment failure. Thus, in the future, investigating the interaction network of Galectin-1, CD146, and VEGFR2 will be necessary for a better understanding of the mechanisms underlying the angiogenesis and metastasis of carcinoma.

CD146 as the co-receptor of PDGFR-β in vessel integrity

PDGF family members, including PDGF-A, -B, -C, and -D, play a crucial role in embryologic development based on the fact that all PDGF knockout mice are embryonic lethal. Similar to VEGFs, all PDGFs are dimers of disulfide-linked polypeptide chains.²²⁸ In the mouse embryo, PDGF-B is strongly expressed in the developing vascular endothelium,²²⁹ tip cells of angiogenic sprouts, and in the endothelia of growing arteries, where pericytes are actively recruited.^230,231 During angiogenesis, PDGF-BB released from endothelial cells promotes the recruitment of adjacent pericytes to the endothelial surface.^193,194

Both PDGFs and their receptors (PDGFR-α and -β) are implicated in tumor angiogenesis and lymphangiogenesis.^232–235 PDGF-AA and PDGF-CC mainly bind to PDGFR-α, whereas PDGF-BB binds to PDGFR-β.²³⁶ PDGFR-β is expressed in the mesenchyme, particularly in vascular smooth muscle cells and pericytes.^230,237 PDGF-BB/PDGFR-β signaling elicits several well-characterized signaling cascades, such as Ras-MAPK, PI3K, and PLC-γ. The PDGFR-β-Ras-MAPK cascade leads to the stimulation of cell proliferation, differentiation, and migration.^238,239 The PDGFR-β-PI3K branch promotes cell growth and cytoskeletal reorganizations.²⁴⁰ PDGFR-β-PLC-γ activates intracellular Ca²⁺ mobilization, stimulating Ca²⁺-dependent secondary signaling, such as tube formation and cell motility.^241,242 In addition, in adult mice, PDGF-BB/PDGFR-β signaling exerts a neuroprotective function by the recruitment of pericytes.²⁴³

In 2017, we revealed that CD146 controls pericyte recruitment and vessel stabilization through interactions with PDGFR-β to protect the CNS.⁵⁹ First, we extensively examined the expression patterns of CD146 in brain endothelial cells and pericytes throughout development and adulthood. In mouse brains, CD146 is first expressed in the cerebrovascular endothelial cells of premature blood vessels without pericyte coverage; with increased coverage of pericytes, CD146 could not be detected in cerebrovascular endothelial cells, it could only be found in pericytes. Furthermore, we studied the role of CD146 in endothelial-pericyte communication through its selective deletion in both cell types. We found that a knockdown in either cell type leads to the breakdown of the blood-brain barrier (BBB), which initiates the invasion of the endothelial cells toward the CNS and the recruitment of pericytes to the nascent vessels during embryogenesis. Furthermore, we demonstrated that CD146 functions as a co-receptor of PDGFR-β to mediate pericyte recruitment to cerebrovascular endothelial cells, indicating that pericyte CD146 was important for pericyte recruitment and vessel stabilization. The attached pericytes in turn downregulate endothelial CD146 by secreting TGF-β1 to promote further BBB maturation.

Because both CD146 and PDGFR-β are involved in regulation of growth and survival of different cell types, including cancer cells, further investigation of functions of CD146 and PDGFR-β interaction in cancers may help deeply understand the dysregulation of spatio-temporally controlled PDGF-BB induced signaling during tumor development and progression, especially tumor angiogenesis and lymphangiogenesis.

CD146 as the receptor of VEGF-C in lymphangiogenesis

VEGF-C belongs to the VEGF family.^197,244 Different from VEGF-A in angiogenesis, VEGF-C is the principal driver of lymphangiogenesis and controls the whole process of lymphangiogenesis.^245–247 Genetic studies in mice have revealed that loss of VEGF-C impairs the development of the lymphatic vasculature.^247,248 On the other hand, overexpression of VEGF-C in specific tissues induces in situ lymphangiogenesis, such as in the skin,^249,250 pancreas,²⁵¹ and lung.²⁵²

Full-length VEGF-C undergoes proteolytic processing and becomes a fully processed form, which increases its affinity for VEGFR2 and VEGFR3.^253,254 VEGF-C regulation of proliferation and migration of lymphatic endothelial cells is mainly through binding with VEGFR3.^255–257 VEGF-C/VEGFR3 activates the ERK-MAPK and PI3K/AKT pathways and enhances diverse physiological effects in lymphatic endothelial cells, such as growth, proliferation, mobility, and invasiveness. Therefore, the signaling axis of VEGF-C/VEGFR3 is under extremely sophisticated control.^258–260 However, the receptors that mediate VEGF-C signal transduction for lymphatic sprouting, the initiating step of lymphangiogenesis, remain elusive.

In 2017, we found that VEGF-C is also the ligand of CD146 and the binding site exists between D4 and D5 domains of the extracellular domain of CD146.⁵⁷ VEGF-C activates CD146 to mediate sprouting during lymphangiogenesis, although the exact amino acid residues in the intracellular domains of CD146 responsible for transmitting VEGF-C signals still need to be defined. Using a zebrafish model, we discovered why CD146 is required for the lymphatic sprouting during development. Knockdown of CD146 inhibited phosphorylation of p38 and ERK, while knockdown of VEGFR3 inhibited phosphorylation of AKT and ERK. Furthermore, we confirmed that inhibition of p38 mainly reduced sprouting of lymphatic endothelial cells during lymphangiogenesis.

Maladjustment of VEGF-C-elicited signaling results in tumor metastasis, especially via lymphatic vessels.^261–265 Lymphatic metastasis is a challenge for clinical treatment of tumors and is the cause of death for some cancers,^266,267 such as breast cancer,²⁶⁸ lung cancer,²⁶⁹ and melanoma.²⁷⁰ Soluble VEGFR3,^271,272 VEGF-C inhibitor,²⁷³ VEGFR3 antibodies,²⁷⁴ and VEGF-C siRNA²⁷⁵ have been used in the treatment of lymphatic metastasis. Our findings suggest that targeting CD146 may also be an effective therapeutic strategy to treat lymphatic metastasis.

CD146 as the receptor of Wnt5a in cell migration

The Wnt family includes several secreted glycoproteins that are involved in the regulation of a wide variety of normal and pathological processes, including embryogenesis, differentiation, and tumorigenesis.²⁷⁶ In human, there are 19 wnt genes that encode functionally distinct Wnt proteins, which can bind to their receptors, FZD/LRP heterodimers. Wnt signaling pathways include canonical and non-canonical cascades. The canonical pathway causes stabilization and nuclear translocation of β-catenin, which regulates transcription of Wnt target genes. The non-canonical pathway is β-catenin-independent and can be further divided into Wnt/planar cell polarity (PCP), Wnt/Ror2, and Wnt/Ca²⁺ signaling cascades.²⁷⁷ The PCP pathway is activated by c-Jun N-terminal kinases (JNKs).²⁷⁸

Wnt5a transmits signals through either canonical or non-canonical Wnt pathway.^279,280 In 2013, we found that CD146 is the receptor of Wnt5a and is required for the Wnt5a-controlled cell migration and convergent extension during zebrafish embryogenesis.⁵⁴ The biochemical experiments revealed that CD146 binds to Wnt5a with the high affinity, leading to activation of JNK-PCP pathway and downregulation of β-catenin expression.⁵⁴ Further analysis demonstrated that CD146 can interact with Dvl2, and this interaction is enhanced under Wnt5a treatment. Accordingly, knockout of CD146 results in dysregulation of the Wnt/PCP pathway. Thus, our findings provide the first direct evidence that CD146 turns on the non-canonical Wnt signaling branch as a functional Wnt5a receptor in cell migration during development.

Wnt5a is upregulated in various types of human cancers.^281,282 Meanwhile, Wnt5a activation of JNK is linked with cytoskeletal remodeling and cell motility in various cell systems.^283–285 For example, in melanoma, Wnt5a is thought to directly affect cell motility and metastasis.²⁸⁶ In this view, CD146 may represent the prime target to develop more effective and less toxic therapies toward Wnt5a/CD146/JNK activation for meeting the challenges from tumor metastasis.

CD146 as the receptor of Wnt1 in fibroblast activation

Wnt1, originally known as oncogene int-1, was initially discovered by analysis of host cell sequences adjacent to viral integration sites in tumors of mice infected with mouse mammary tumor virus.^287,288 Subsequent evidence suggests that the oncogenic functions of Wnt1 is mediated via upregulation of proliferative genes by canonical β-catenin pathway.

Similar with CD146, Wnt1 protein expression levels are high at developmental stage and low in adults, and ectopic expression of Wnt1 causes tumor development.²⁸⁹ In 2018, we found that CD146 can directly bind with Wnt1 in fibroblast, activating fibroblast via canonical Wnt/β-catenin pathway. Such interaction is essential for Wnt1-induced fibroblast proliferation and ECM production.⁵⁸

Because cancer-associated fibroblasts (CAFs), as the main cellular constituent of the cancer-associated stroma, can drive cancer cell invasion but can also impair the migration and activation of T cells. Herein, researchers should further examine whether CD146 is a marker of CAFs, and if so, inhibiting the interaction of CD146 with Wnt1 on CAFs may benefit cancer treatment.

CD146 as the receptor of Netrin-1 in angiogenesis

Netrin-1 belongs to a family of Laminin related secreted proteins that control axonal and cellular migration during the development of the nervous system.^290–293 Netrin-1 is a 640 amino acid protein and its carboxy-terminal sequence is enriched in basic amino acids.^294–296 This sequence can bind HSPGs with high affinity, contributing to retaining them in the ECM.²⁹⁷ Netrin-1 is not only expressed in the nervous system but also in other non-neuronal organs.^290,291,298 It regulates diverse processes, such as neuronal navigation, cellular adhesion, motility, proliferation, and differentiation during development.^299–302 Dysregulation of Netrin-1 is involved in diverse pathological processes, such as cancer, cardiovascular disease, and kidney disease, making it an attractive potential therapeutic target.^301,303,304

Netrin-1 acts through two main receptors, DCC (deleted in colorectal cancer) and UNC5B (uncoordinated-5 homolog), to alter the architecture of the cytoskeleton networks.^305,306 Binding to its receptors activates chemotropic responses and adhesive mechanisms, regulating inflammation, angiogenesis, and apoptosis.^299,301,302

Our laboratory clarified that netrin-1 binds CD146 with a higher affinity than the classical UNC5B receptor.⁵⁵ Netrin-1 preferentially binds CD146 at low concentrations (50–200 ng/mL) and binds UNC5B at high concentrations (1000–2000 ng/mL). In addition, our study demonstrated the dual action of Netrin-1 on angiogenesis: the pro-angiogenic roles through CD146 and the antiangiogenic functions through UNC5B. CD146 silencing or deletion inhibits Netrin-1-induced HUVEC proliferation, migration, and tube formation, as well as VEGFR2, p38, and extracellular regulated MAP kinase (ERK)1/2 activation. In contrast, Netrin-1 binding with UNC5B at high concentrations triggers signals that counteract the CD146-mediated pro-angiogenic pathway. CD146 can also interact with VEGFR2 as a co-receptor;⁵³ because of this, CD146-Netrin-1 binding may further improve VEGF-A signaling-mediated angiogenesis.

Thus, this finding highlights the importance of CD146 in Netrin-1-induced angiogenesis, although other factors are also likely contributors in the absence of CD146.³⁰³ Similarly to CD146, Netrin-1 also takes part in tumor growth, these properties of Netrin-1 in cancer offer reasonable therapeutic approaches. Therefore, preclinical and clinical studies should be planned to investigate the therapeutic potential of the Netrin-1/CD146 pathway in tumor angiogenesis.

CD146 as the receptor of FGF4 in apical-basal polarization

FGFs were initially recognized as fibroblast-specific growth factors, and it is now recognized that FGFs regulate multiple biological functions in diverse cell types beyond fibroblasts. FGFs constitute a large family with 23 members,³⁰⁷ which share sequences and structural similarities.³⁰⁸ These factors include paracrine or endocrine forms and participate in important pathophysiological processes such as cell proliferation, survival, migration, angiogenesis, wound healing, differentiation, and endocrine secretion regulation during development and adult life.^309–311

Among FGFs, only FGF4 and eight have been revealed to possess the chemotactic activity and FGF4 acts as a chemo-attractant during morphogenesis.³¹² FGF4 belongs to the paracrine FGFs that signal FGFRs by forming a tripartite complex with FGFRs (1–4) and HSPGs.³¹³ Although extensive investigations focus on the FGF signaling, there has been no indication that FGF ligands can bind with other receptors beyond FGFRs. Meanwhile, no direct evidence has indicated that any one of the FGFRs mediates the FGF4-elicited chemo-attracting activities.

The chief intracellular substrates of FGF signaling are FGF receptor substrate 2 (FRS2) and PLC-γ. Activated FRS2 activates the transcription factors of activator protein 1 (AP-1) and forkhead box protein O (FOXO) through RAS–ERK or the PI3K-pyruvate dehydrogenase kinase pathway, respectively, whereas PLCγ leads to the activation of Ca²⁺-dependent nuclear factor of activated T cells (NFAT).³¹³ There are several points where other pathways can cross-talk with FGF signaling, for instance, activated JNK is required for AP-1 activation and FOXO nuclear translocation.³¹⁴

In 2017, we found that CD146 is an independent receptor of FGF4.⁵⁶ The binding affinity between CD146 and FGF4 is higher than that between FGF4 with its receptors (FGFR1-4). In addition, the presence of HSPG is not required for the binding of FGF4 to CD146. Using zebrafish as the model system, we found that CD146, but not FGFR, is the responsive receptor for FGF4. It provided local spatial cues to organize apical-basal polarity in participating cells during morphogenesis. An in vitro lumen formation assay further confirmed the migration of CD146, but not FGFR1, toward FGF4 as the key activity during FGF4-induced establishment of the apical-basal polarity. By investigating the effects of CD146 on FGF signaling output, we found that the cooperative actions between CD146-dependent activation of JNK and NFAT together with FGF signaling-dependent activation of ERK ensure that CD146⁺ cells concomitantly upregulate the transcriptional activity of AP-1 and NFAT during organ morphogenesis.

Thus, our findings suggest the essential roles of CD146 in FGF4-elicited morphogenetic signaling. In light of the coincident spatiotemporal distribution of CD146/FGF4 on developmental embryos, CD146 may be a more preferable receptor in FGF4 biological responses than FGFRs. The cooperation of CD146/FGF4 in cell polarity establishment suggests that CD146 could be the genuine responsive receptor in all of FGF4-executed chemo-attracting actions. Because FGF4 possesses the potent pro-angiogenetic activity and fails to bind heparan sulfate in the heart and blood vessels,³¹⁵ it is possible that CD146-FGF4 is also a critical partner in angiogenesis. For this reason, in the therapeutic regimes of pathological angiogenesis, simultaneous targeting of both CD146 and FGF4 could have better efficacy than settings with that singular target either.