Role of glypican-1 in regulating multiple cellular signaling pathways

Abstract

Glypican-1 (GPC1) is one of the six glypican family members in humans. It is composed of a core protein with three heparan sulfate chains and attached to the cell membrane by a glycosyl-phosphatidylinositol anchor. GPC1 modulates various signaling pathways including fibroblast growth factors (FGF), vascular endothelial growth factor-A (VEGF-A), transforming growth factor-β (TGF-β), Wnt, Hedgehog (Hh), and bone morphogenic protein (BMP) through specific interactions with pathway ligands and receptors. The impact of these interactions on signaling pathways, activating or inhibitory, is dependent upon specific GPC1 domain interaction with pathway components, as well as cell surface context. In this review, we summarize the current understanding of the structure of GPC1, as well as its role in regulating multiple signaling pathways. We focus on the functions of GPC1 in cancer cells and how new insights into these signaling processes can inform its translational potential as a therapeutic target in cancer.

INTRODUCTION

Heparan sulfate proteoglycans (HSPGs) are glycoproteins consisting of a core protein covalently bound to several heparan sulfate (HS) glycosaminoglycan (GAG) chains. HSPGs are expressed ubiquitously on the cell surface and in the extracellular matrix where they interact with a wide range of ligands to mediate various cellular functions (1). HSPGs are categorized into three groups based on their location: membrane HSPGs (e.g., syndecans and glypicans), secreted extracellular matrix HSPGs (agrin, perlecan, type XVIII collagen), and secretory vesicle proteoglycans (serglycin) (2).

Glypicans, of the HSPGs family, are attached to the outer surface of the plasma membrane by a glycosyl-phosphatidylinositol (GPI) anchor (3, 4). Six human glypican family members (GPC1 to GPC6) have been identified and fall into two broad subfamilies, glypicans 1/2/4/6 and glypicans 3/5, both sharing ∼25% amino acid identity (3, 4). The three-dimensional structure of glypicans might be similar across the family, as the localization of 14 cysteine residues is conserved in all family members (3). All glypicans share attachment sites for HS chains, in particular, the two close to the cell membrane, and a hydrophobic sequence necessary for the GPI anchor in the C-terminal tail. Mounting evidence have demonstrated that glypicans are overexpressed in multiple human cancers (5). The current knowledge of the expression pattern of different glypicans in normal tissues and tumors has been summarized in Table 1.

Table 1.

Expression of glypicans in normal tissues and cancers

Glypicans	Normal Tissues (Antibodies)	Cancers (Overexpressed in Multiple Human Cancers) (5)
GPC1	Low to no expression (IHC, clone: 1–12) (6)	Pancreatic, breast, glioma, glioblastoma, prostate, uterine cervical cancers, and ESCC
GPC2	No expression except testis (IHC, clones CT3 and LH7) (7, 8)	Pediatric cancers, neuroblastoma, medulloblastoma, SCLC (9)
GPC3	No expression (IHC, clones YP7 and HN3) (10–13)	HCC, LSCC, melanoma, Merkel cell carcinoma, OCCC, thyroid cancer, urothelial carcinoma, salivary gland tumors, glioblastoma, hepatoblastoma, YST, choriocarcinoma, Wilms tumor, RMS, mesothelioma, and renal CCC
GPC4	N/A	Pancreatic, colorectal, and breast cancers and drug-resistant ovarian cancer
GPC5	N/A	Breast, prostate, and pancreatic cancers, glioma, RMS, and NSCLC
GPC6	N/A	Gastric cancer, ovarian cancer, melanoma, drug-resistant ovarian cancer, and retinoblastoma

CCC, clear cell carcinoma; ESCC, esophageal squamous cell carcinoma; GPC, glypican; HCC, hepatocellular carcinoma; IHC, immunohistochemistry; LSCC, lung squamous cell carcinoma; N/A, not available; NSCLC, non-small cell lung cancer; OCCC, ovarian clear cell carcinoma; RMS, rhabdomyosarcoma; SCLC, small-cell lung cancer; YST, yolk sac tumor.

Human GPC1 is expressed not only in the central nervous system (CNS) and skeletal system during development but also in other tissues in the adult (14). GPC1 is overexpressed in multiple types of cancers (5, 15–19), including breast cancer, esophageal squamous cell carcinoma (ESCC), glioma, and pancreatic cancer. Its high expression is corelated with poorer prognosis (18, 20, 21), making it a potential target for cancer therapy. In addition to its role in tumorigenesis, GPC1 is also involved in a variety of development pathways such as brain development (22), neurodegeneration (23), axonal guidance, and regeneration (24, 25), as well as prion conversion and scrapie infection (26–28).

GPC1 can act as a coreceptor for multiple signaling molecules known to regulate cell growth, motility, and differentiation through the HS chains and/or core protein. In this review, we discuss our current understandings on the structure of GPC1 and the roles of GPC1 in cellular signaling, mainly focusing on fibroblast growth factors (FGF), vascular endothelial growth factor-A (VEGF-A), transforming growth factor-β (TGF-β), Wnt, bone morphogenic protein (BMP), and Hedgehog (Hh). We also discuss the potential of GPC1 as a therapeutic target in cancer.

THE STRUCTURE OF GPC1

GPC1 is encoded by human GPC1 gene located at 2q37.3. As shown in Fig. 1A, GPC1 contains 558 amino acids and is composed of a secretory signal peptide (residues 1–23), an N-terminal core protein (residues 24–474), a C-terminal HS chain attachment region (residues 475–530), and a GPI anchor attached to cell membrane (residues 531–558) (29). GPC1 protein is substituted with a cluster of three HS chains at positions Ser-486, Ser-488, and Ser-490 and is also decorated with two N-linked glycans at Asn-79 and Asn-116 that affect the GPC1 expression level, as well as HS substitution (29, 30). Svensson et al. (31) determined the crystal structure of the N-glycosylated, C-terminally truncated GPC1 core protein (GPC1ΔC, residues 27–475) without HS chains at 2.6 Å (PDB entry 4ACR) and showed that it has an elongated cylindrical form and an all α-helical fold (α1–α14) with three major loops (L1, L2, and L3). The GPC1ΔC structure can be divided into three parts, a Cys-rich region near the N-terminal lobe, a central lobe stabilized by two evolutionarily conserved hydrophobic centers, and C-terminal lobe containing a protease site (31).

Figure 1.

GPC1 structure and the regulatory roles of GPC1 in multiple-signaling pathways. A: schematic of human GPC1 core protein linked to the cell membrane by its GPI anchor. The highly conserved Cys residues are indicated in yellow. The disulfide bonds formed by the Cys residues are indicated by purple lines. Two N-linked glycans indicated by black are at positions Asn-79 and Asn-116. Three HS chains are located at Ser-486, Ser-488, and Ser-490. B: overview of the GPC1ΔC crystal structure. Seven disulfide bonds are highlighted by red solid lines. The sequences before and after the predicted furin cleavage site are indicated by cyan and light pink, respectively. Cys residues before and after furin cleavage site are highlighted in blue and magenta, respectively. C: GPC1 modulates various signaling pathways in cancer progression and other physiological processes. GPC1 can interact with both heparin-binding growth factors (FGF and VEGF-A) and their receptors to stabilize their assembly and enhance MARK or PI3K/Akt signaling. GPC1 also binds TGF-β and its receptors to promote SMAD signaling. However, GPC1 plays a negative role in BMP cascades by competing with BMP receptors to interact with BMPs. GPC1 might serve as a Wnt coreceptor to activate Wnt signaling. BMP, bone morphogenic protein; FGF, fibroblast growth factor; FGFR, fibroblast growth factors receptor; GPC1, glypican-1; GPI, glycosyl-phosphatidylinositol; HS, heparan sulphate; TGF-β, transforming growth factor-β; VEGF-A, vascular endothelial growth factor-A.

Figure 1A shows the complete disulfide pattern of the 14 conserved Cys residues across the glypican family by sequencing analysis (31). The crystal structure of human GPC1 is shown in Fig. 1B (PDB entry 4AD7) and the disulfide bonds are highlighted by red sticks. Six disulfide bonds connect the structure in the Cys-rich lobe. Three disulfide bonds (Cys²⁶⁸-Cys⁴¹⁵, Cys²⁷²-Cys⁴⁰¹, Cys²⁴⁶-Cys²⁷⁹) are in α8 helix in the Cys-rich region, among which Cys²⁶⁸-Cys⁴¹⁵ and Cys²⁷²-Cys⁴⁰¹ connect the structure between the furin-like convertase site. Cys³²-Cys⁶⁸, Cys⁶²-Cys²⁵⁶, and Cys⁶⁹-Cys²⁵⁹ form three longitudinally placed disulfide bonds in the N-terminal part of the Cys-rich lobe. Lastly, Cys¹⁹¹-Cys³⁴³ forms the remaining disulfide bond and is located in the protease-site lobe.

To further elucidate the GPC1 C-terminal structure, Awad et al. (29) examined the N-glycans and C-terminus (Asp⁴⁷⁵–Thr⁵²⁹) of GPC1 using various purified 6xHisditine-tagged constructs of GPI anchorless GPC1. Various structure predictors [including the most recent AlphaFold2 (32, 33): https://alphafold.ebi.ac.uk/entry/P35052] indicated that GPC1 C-terminal region lacks significant secondary or tertiary structure. Crystal structure analysis of full-length GPC1 carrying two N-glycans but no HS chain (by mutagenesis of S486A, S488A, and S490A, termed as GPC1ΔHS) in conjunction with small angle X-ray scattering (SAXS), showed that the C-terminal domain is highly flexible and extends ∼35–40 Å from the core protein (29). Interestingly, the GPC1 C-terminus was predicted to orient the core protein transverse to the membrane, directing a surface evolutionarily conserved in GPC1 orthologs (L1 and L3 loops, α4, α5, and α14 helices) toward the membrane, where it may interact with signaling molecules and/or membrane receptors on the cell surface (29).

The structure of GPC1 with the HS chains is related to its biological function by modulating various signaling pathways through specific interactions with ligands and receptors in cell surface signaling complexes. The actual orientation and possible rotation of GPC1 (and other glypicans) relative to the cell surface is unknown. The impact of these surface signaling complexes, either activating or inhibitory, is dependent upon specific GPC1 domain interaction with signaling components under specific physiological conditions.

FGF SIGNALING

Fibroblast growth factors (FGFs) and their receptors (FGFRs) regulate a wide range of cellular processes, such as embryonic development and differentiation. Pathologically, FGFR signaling components are frequently altered in human cancers, suggesting their oncogenic potential (34).

FGFs are secreted glycoproteins that are readily sequestered by the extracellular matrix and the cell surface HPSGs, including GPC1. GPC1 has been proposed to act as a coreceptor for FGFs that enhances the binding of FGF to its receptor, subsequently promoting FGF-FGFR activation and signaling (35). FGFR signaling network has been thoroughly described in other reviews (36). The binding of FGF to its receptor leads to receptor dimerization and transphosphorylation of tyrosine kinase domains, resulting in subsequent activation of various signaling pathways, including Ras-MAPK, PI3K-AKT-mTOR, and DAG-PKC. All signaling cascades ultimately result in enhanced growth, survival, and angiogenesis.

FGF2 is one of the heparin-binding growth factors (HBGFs) that can drive tumor cell proliferation upon binding to its receptors. In addition to a cell proliferation stimulator, FGF2 also functions as an angiogenic growth factor in angiogenesis (37). Angiogenesis is an essential process involved in the blood vessel formation and a fundamental step in tumor transition from benign status to malignance. Mounting evidence indicates that changes in GPC1 expression have profound consequences on FGF2-induced cell proliferation or angiogenesis in tumors (Fig. 1C). For example, in breast cancer, GPC1 is upregulated and promotes tumor mitogenic signaling by modulating heparin-binding growth factors, including FGF2 (17). GPC1 has also been shown to cooperate with type V collagen to concentrate FGF2 at the extracellular cell matrix (ECM) interface, thereby affecting ECM stability and breast tumor cell proliferation (38).

In addition to breast cancer, GPC1 has also been reported to be associated with pancreatic cancer. An early study demonstrated that GPC1 was highly expressed in human pancreatic cancer. Downregulation of GPC1 expression abrogated mitogenic responses to FGF2 in pancreatic cancer cells (15). Whipple et al. (39) further studied GPC1 in pancreatic cancers using an oncogenic Kras-driven genetic mouse model of PDAC. These GPC1 null mice (GPC1^−/−) contained pancreas-specific Cre-mediated activation of oncogenic Kras (Kras^G12D) and deletion of a conditional INK4A/Arf allele (Pdx1-Cre;LSL-Kras^G12D;INK4A/Arf^lox/lox;GPC1^−/−). As compared with the comparable mouse model containing wild-type GPC1, pancreata isolated from GPC1^−/− mice were smaller and lighter. Of note, GPC1^−/− mice showed decreased pancreatic tumor growth and invasiveness. For example, at day 65, 100% of GPC1^+/+ transgenic mice harbored large and invasive pancreatic tumors that adhered to and invaded surrounding organs, whereas 80% of GPC1^−/− mice exhibited significantly smaller tumors that were not grossly invasive or adherent. Moderate attenuated cancer cell proliferation and angiogenesis were also observed in GPC1^−/− mice (39). Furthermore, GPC1^−/− pancreatic cancer cells isolated from GPC1 null mice had attenuated invasive properties in response to FGF2 compared with cancer cells isolated from wild-type mice. After implanting small (2 mm³) fragments prepared from 65-day old GPC1^+/+ and GPC1^−/− mice into the pancreata of athymic GPC1^+/+ and GPC1^−/− mice, GPC1^−/− mice also exhibited suppressed metastatic potential. Fourteen days after implantation, only 14% of GPC1^−/− mice developed metastases, whereas 60% of GPC1^+/+ mice developed numerous mesenteric metastases and 20% exhibited multiple renal metastases. All of these findings demonstrated that GPC1 enhanced PDAC tumor growth, angiogenesis, and invasion, indicating that GPC1 is a potential target in PDAC therapy (39).

In glioma tumor, GPC1 was frequently overexpressed in glioma vessel endothelial cells (ECs) and was undetectable in normal brain vessels. Mouse brain endothelial (MBE) cells express low levels of GPC1 and mirror normal brain vessel ECs in vivo. Qiao et al. (40) demonstrated that the MBE cells showed a robust proliferative response to FGF2. The proliferation of MBE cells was significantly inhibited when functional HS was abolished following treatment with sodium chlorate (a competitive inhibitor of glycosaminoglycan), whereas these responses were recovered when HS sulfation was restored, indicating the essential role of HS for FGF2 signaling. Furthermore, MBE cells with ectopic overexpression of GPC1 displayed a dramatically elevated base-line proliferation and increased sensitivity to FGF2-induced mitogenesis (40). In contrast, the overexpression of a different class of surface HSPG, syndecan-1, showed a slightly decreased growth rate and similar sensitivity response to FGF2 compared with mock-transfected controls, suggesting the specific role of GPC1 core protein in endothelial cell growth and FGF2-induced mitogenesis independent of HS chain composition (40). Su et al. (16) revealed significant overexpression of GPC1 in human astrocytoma and oligodendroglioma samples when compared with nonneoplastic gliosis. GPC1 may also contribute to the enhanced FGF2/FGFR1c signaling in glioma cells. This is likely due to highly elevated 2-O-sulfation- and 6-O-sulfation-containing disaccharides, indicating that GPC1 HS chains may also be important in FGF signaling pathway regulation (16).

GPC1 may also be involved in other physiological processes by mediating FGF signaling. GPC1 was identified as the major GPC expressed in human keratinocytes and whose expression was significantly decreased with aging. Downregulation of GPC1 results in attenuated proliferative effect of FGF2, confirming its role as a modulator of growth factor effects on keratinocytes (41). In another study, GPC1 loss significantly reduced brain size (∼30%) during early neurogenesis with very subtle effects on brain patterning, most likely resulting from a transient reduction of FGF17 (22).

Taken together, GPC1 plays an important role in multiple types of human cancers through FGF/FGFR signaling pathway regulation.

VEGF-A SIGNALING

Similar to FGF2, VEGF-A is also an angiogenic growth factor (37) and plays an essential role in angiogenesis. GPC1 acts as a coreceptor of VEGF-A (Fig. 1C). In 1999, GPC1 was first reported to specifically bind to VEGF₁₆₅ (VEGF-A) but not VEGF₁₂₁. This interaction is mediated by its HS chains, as HS removal by heparinase treatment abolished the ability of GPC1 to bind to VEGF₁₆₅. Cell binding assays using ECs also demonstrated that the addition of exogenous GPC1 could potentiate the VEGF-A/VEGFR binding, suggesting the potential role of GPC1 in the control of angiogenesis (42).

Since then, several mouse models by manipulating GPC1 expression levels have been established to study the role of GPC1 in tumor angiogenesis and migration through VEGF-A signaling. Aikawa et al. (43) revealed that both tumor cell and host-derived GPC1 are essential for full mitogenic, angiogenic, and metastatic potential of pancreatic cancer cells. Downregulation of GPC1 expression in PANC-1 cells significantly prolonged doubling times and attenuated anchorage-independent growth in vitro. When these cells were transplanted into athymic mice subcutaneously, the resulting tumors showed reduced growth when compared with tumors from sham-transfected cells. In addition, tumors derived from GPC1 downregulated pancreatic cancer cells showed an approximate 40% decrease in microvessel densities. These data demonstrated that downregulation of GPC1 resulted in attenuated tumor growth and suppressed angiogenesis in mice. Mechanistically, the levels of phospho-MAPK (p-ERK1 and p-ERK2) were significantly decreased in these tumors as compared with those derived from sham-transfected cells, indicating the correlation between GPC1 levels and the activation of the MAPK pathway in PANC-1 cells (43). In addition, quantitative real-time PCR analysis revealed that the mRNA level of VEGF-A was significantly decreased in tumors from GPC1 antisense PANC-1 clones by comparison with tumors from sham-transfected cells, suggesting that the VEGF-A signaling might be directly inhibited (43).

Parallel to this, PANC-1 or T3M4 cells were injected into the pancreas of both wild-type (WT) and GPC1-null nude mice. Derived tumors exhibited decreased angiogenesis and metastasis in athymic GPC1-null mice as compared with tumors in WT mice. The GPC1-null mice also developed fewer pulmonary metastases after intravenous inoculation with murine B16-F10 melanoma cells. Hepatic ECs isolated from mice lacking GPC1 exhibited a decreased mitogenic activity in response to VEGF-A (43).

Consistent results were reported by Whipple et al. (39) in a GPC1 knockout Kras-driven genetic mouse model of PDAC. These GPC1^−/− mice exhibited attenuated invasiveness and angiogenesis of implanted tumors. RNA expression of proangiogenic genes in GPC1^−/− mice, including VEGF-A, were downregulated. Hepatic ECs isolated from these GPC1 null mice also had decreased migration in response to VEGF-A treatment as indicated by a transwell migration assay (39). Taken together, findings based on GPC1 null mice models have demonstrated that GPC1 is essential in tumor growth, angiogenesis, and metastasis.

TGF-β SIGNALING

The transforming growth factor-β (TGF-β) belongs to a superfamily of cytokines that activate protein kinase receptors on the plasma membrane to regulate cell growth, death, differentiation, immune response, angiogenesis, and inflammation. Dysregulation of this pathway contributes to a broad variety of pathologies, including cancer (44). However, TGF-β signaling is considered a challenging target due to its dual functions and pleiotropic nature. In healthy cells and early-stage cancer cells, TGF-β acts as a tumor suppressor, whereas in the late stage of cancer, it drives tumor progression and metastasis (45). GPC1 has been shown to interact with TGF-β and its receptors to stabilize their assembly for enhanced Smad signaling (Fig. 1C).

Decreased GPC1 expression suppresses pancreatic cancer cell growth via modifying TGF-β signaling. Downregulation of GPC1 expression resulted in a slightly altered response toward TGF-β1, activin-A, and BMP2 in terms of growth, p21 induction, and Smad2 phosphorylation, ultimately leading to decreased anchorage-independent growth of T3M4 and PANC-1 cells (46).

Another study investigated the transcriptional cooperation between TGF-β and Wnt pathways in mammary and intestinal tumorigenesis. A novel gene expression profile in normal murine mammary gland (NMuMG) epithelial cells induced by the combination treatments of TGF-β and Wnt3a was revealed by using oligonucleotide microarray and quantitative PCR analysis; Gpc1 was one of the identified cooperative genes (47). In the same study, two transgenic mouse models were used to further investigate the correlation between TGF-β/Wnt and GPC1 in vivo. Min (APC^+/−) mice harbor a mutation in APC and develop multiple small-bowel adenomas and colon microadenomas (48). MMTV-Wnt1 mice are transgenic animals overexpressing Wnt1 oncogene in mammary epithelial cells, leading to mammary adenocarcinomas in 50% of females by the age of 6 mo (49). Both the Wnt/β-catenin and TGF-β/Smad pathways are fully activated in epithelial cells of Min intestinal adenomas or MMTV-Wnt1 mammary tumors. Accordingly, Gpc1 expression was significantly elevated in both intestinal and mammary tumors derived from these two transgenic mouse models, suggesting that activated Wnt and TGF-β signaling pathways might lead to increased Gpc1 expression (47). Moreover, double-transgenic MMTV/Wnt1/DNIIR mice were used to further study the involvement of GPC1 within TGF-β signaling. In these double-transgenic mice, TGF-β signaling was abrogated but Wnt1 overexpression remained by overexpressing dominant-negative TGF-β type II receptor (DNIIR) in MMTV-Wnt1 mice. TGF-β signaling interruption in these mice resulted in increased tumor latency, and enhanced tumor-free survival, and significantly reduced expression of Gpc1. This indicated the correlation between Gpc1 and TGF-β signaling in tumor progression (47).

Overall, all of these findings have emphasized the tumor-driven role of GPC1 via TGF-β mediation, further supporting the hypothesis that GPC1 targeting may be another effective strategy to treat human cancers. However, TGF-β cascades exert tumor-suppressive or -progression effects. The specific role of GPC1 in regulating TGF-β or whether GPC1 is correlated with the functional transition of TGF-β remain unclear.

BMP SIGNALING

Bone morphogenetic proteins (BMPs) play substantial roles in cell-cell communication during animal development and are potent growth factors promoting bone formation. Glypicans have been shown to regulate BMP activity (50, 51). For example, overexpression of GPC3 inhibited BMP-7 signaling through the Smad-6 pathway by luciferase reporter assay (50). GPC4 in another study attenuated BMP signaling pathways to promotes cardiac specification and differentiation during heart development (51). GPC1 protein is mainly expressed in the skeletal system in humans, and Gpc1 expression was also identified in the developing murine calvarium and skeletal structures (52). Thus, it may be inferred that GPC1 is also involved in the regulation of the BMP signaling pathway.

Dwivedi et al. (53) demonstrated that GPC1 and GPC3 function as negative modulators for BMP2 signaling to regulate osteogenesis in human suture mesenchymal cells. Craniosynostosis is a medical condition that occurs when premature bony fusion of one or more sutures results in a cessation of bone growth. Dwivedi et al. identified the co-expression of GPC1, GPC3, and the BMP type II receptors (BMPRII and ACTRIIB, the receptors for BMP2, 4 and 7) in human suture mesenchymal cells and further demonstrated that GPC1 and GPC3 were able to physically interact with BMP2 (53). ID1 is an immediate early BMP2 target gene in response to BMP2 (54). Increased GPC1 and GPC3 expression completely blocked BMP2 inductive activity at ID1. The addition of exogenous recombinant GPC1 and GPC3 could also dose-dependently inhibit BMP2 activity and effectively reduce BMP2-mediated mineralization in a human cranial osteogenesis model. Conversely, prior blockade of the endogenous GPC1 or GPC3 in human suture mesenchymal cells with specific antibodies significantly stimulated BMP2 responsiveness. Those findings highlighted the negative role of GPC1 and GPC3 in regulating BMP signaling and indicate that a GPC1/GPC3 fusion protein treatment might provide a therapeutic strategy to treat craniosynostosis (53). Accordingly, Bariana et al. designed and prepared nanoengineered Titania nanotubes (TNT) implants loaded with recombinant glypicans (GPC1 and GPC3) to treat craniosynostosis. BMP2 activity was significantly inhibited for up to 15 days by the glypicans released from polymer-coated implants, indicating their potential application in adjunctive craniosynostosis treatment (55).

Based on currently available evidence, GPC1 seems to act as an inhibitor in BMP signaling regulation in osteogenesis (Fig. 1C). GPC1 tethered at the cell surface or soluble GPC1 may compete with BMP receptors for interaction with BMP ligands, leading to reduced amounts of BMPs available for binding to their receptors and consequently attenuated Smad-dependent or -independent BMP signaling (53). However, it is unclear how the GPC1 core protein or HS chains function in the interaction with BMP2. Structural analysis is required to uncover the specific interactions between GPC1 and BMPs.

WNT SIGNALING

The Wnt signaling pathway plays significant roles in various biological and pathological processes, including embryonic development, differentiation, cell polarity, and tumorigenesis. Wnt signaling falls into two categories: the canonical pathway which involves the protein β-catenin and the noncanonical pathway which operates independently of it (56). The binding of Wnt ligands to Frizzled receptors triggers downstream Wnt signaling. In the last decade, increasing evidence has shown interactions between R-spondins (RSPOs), leucine-rich repeat-containing G-protein-coupled receptors (LGRs), and transmembrane E3 ubiquitin ligase (ZNRF3/RNF43) in Wnt signaling regulation. In the absence of RSPOs, ZNRF3/RNF43 limits Wnt signaling by reducing the cell-surface level of Frizzled receptors via direct ubiquitination (Fig. 2A). However, upon simultaneously binding to LGRs (LGR4/5/6) and to ZNRF3/RNF43, RSPOs induce internalization and lysosomal degradation of ZNRF3/RNF43, thereby stabilizing Frizzled receptors and amplifying Wnt cascades (57). Recent evidence showed that RSPO2 and RSPO3 can potentiate Wnt signaling in the absence of LGRs 4/5/6 through the interaction between the HS chains of glypicans and TSP/BR domains of RSPO3 (58, 59).

Figure 2.

The potential role of GPC1 in regulating Wnt signaling pathway. A: at baseline without R-spondins (RSPOs), transmembrane E3 ubiquitin ligases (ZNRF3 or RNF43) reduce cell surface levels of Wnt receptors Frizzled (FZD) by direct ubiquitination, leading to limited Wnt signaling. B: as GPC3, GPC1 might serve as a Wnt coreceptor by directly binding to Wnt via HS chains and/or its core protein to potentiate downstream signaling. C and D: in addition, GPC1 might act as a RSPOs coreceptor to present RSPOs via HS chains in the absence or presence of LGRs. With LGRs, RSPO3 can bind simultaneously to LGRs and GPC1, and together present the RSPO3 to ZNRF3/RNF43, leading to the internalization and degradation of ZNRF3/RNF43. Membrane clearance of ZNRF3/RNF43 results in increased number of FZD at the cell membrane with the consequent enhanced Wnt cascades. Without LGRs, RSPO3 is presented by GPC1 alone to sufficiently induce Wnt signaling potentiating through a similar mechanism. GPC1, glypican-1; HS, heparan sulphate; LGRs, leucine-rich repeat-containing G-protein coupled receptors.

In a more recent paper, Dubey et al. (60) demonstrated that glypicans act as coreceptors of RSPO3 to amplify Wnt signaling in the presence or absence of LGRs by binding to TSP/BR via HS chains. To study the ability of RSPO3 ligands to potentiate Wnt signaling, WT or LGR4/5/6^KO HAP1-7GTP cells were induced with Wnt3a conditional medium. The mutant RSPO3 protein that lost proper interaction with HS chains had significantly reduced potency in WT HAP1-7TGP cells and was nearly inactive in LGR4/5/6^KO cells. In contrast, an engineered RSPO3 (referred as RSPO3 ΔTSP/BR HS20) in which the entire TSP/BR domains were replaced with a single-chain Fv antibody (HS20) recognizing HS chains of GPC3 isolated by phage display (61–63), rescued the ability of RSPO3 to potentiate Wnt signals in both the presence and absence of LGRs. Interestingly, when mutations were introduced in the CDR3 region of the HS20 antibody critical for HS binding, the Wnt-potentiating activity was completely abolished. All these data suggested that the RSPO3-glypican interaction was important to potentiate Wnt singling in the presence of LGRs and becomes essential in the absence of LGRs.

Based on all of these findings, a RSPO-mediated working model, in the presence or absence of LGRs, was proposed (Fig. 2, C and D). With LGRs, RSPO3 can bind simultaneously to LGRs and glypicans, which in turn present the RSPO3 ligand to ZNRF3/RNF43, leading to the internalization and degradation of ZNRF3/RNF43 and enhanced Wnt cascades. Without LGRs, RSPO3 is presented by a glypican alone and is sufficient to induce Wnt signaling potentiation (60). However, there is no direct evidence showing the formation of a ternary RSPO3-LGR-glypican complex. Further structural studies will be required for further validation.

In addition to their potential role as coreceptors for RSPOs, numerous studies have demonstrated that glypicans also directly bind to Wnt ligands to trigger the activation of Wnt signaling (Fig. 2B). The role of GPC3 in Wnt signaling has been well demonstrated (3, 61, 64–67). GPC3 interacts with both Wnt and Frizzled to form a tripartite complex and triggers downstream cascades. Previous studies from our laboratory identified a Wnt-recognizing domain on the HS chains of GPC3 using a HS-specific antibody, HS20, that blocks GPC3 and Wnt3a interaction (61, 62). Moreover, we have demonstrated that the core protein of GPC3, in the absence of HS, is also involved in binding Wnt (65–67). Our studies have identified key amino acids in the Wnt binding domain on the core protein of GPC3. It was demonstrated that phenylalanine 41 (F41) within the hydrophobic groove located in the N-lobe of GPC3 is necessary for Wnt3a recognition in hepatocellular carcinoma (HCC) cell and mouse models (68). In addition, we also find that GPC2 is a modulator in Wnt signaling in neuroblastoma cells; genetic silencing of GPC2 suppresses Wnt/β-catenin signaling that leads to tumor cell growth inhibition and apoptosis (7).

To date, the role of GPC1 in the regulation of Wnt signaling in tumors remains elusive. However, there is evidence showing that Wnt/β-catenin signaling intertwines with TGF-β/Smad signaling and induces Gpc1 transcriptional expression that may contribute to tumor progression (47). GPC1 has also been shown to promote the aggressive proliferation of ESCC cells by regulating the PTEN/Akt/β-catenin pathway. GPC1 overexpression increased levels of p-Akt and β-catenin and reduced PTEN expression in ESCC cells, which ultimately induced enhanced epithelial mesenchymal transition, cell proliferation, and survival (68).

In contrast to the cancer studies described above, it has also been indicated that GPC1 works as a negative regulator of canonical Wnt signaling in trigeminal placodes development (69). Overexpression of full-length GPC1 in trigeminal placodes cells blocked proliferation and differentiation, resulting in loss of ganglia (69). Deletion of the GPI anchoring, HS attachment sites or both, failed to cause reduced ganglia as compared with the wild-type GPC1, suggesting the essential role of both GPI anchoring and HS GAG chains (69). In the same study, it was also demonstrated that GPC1 overexpression significantly reduced Wnt signaling activity, whereas the overexpression of a soluble truncated GPC1 construct (GPC1ΔGPI) enhanced the corresponding signal. In parallel, activation of Wnt signaling in the trigeminal placodal ectoderm by using a dominant-active form of β-catenin reversed the effects of GPC1 overexpression but phenocopied the effects of truncated, soluble GPC1. Taken together, all of these findings suggest a negative role of GPC1 in Wnt signaling regulation in the trigeminal placodes, while GPC1ΔGPI functioned as an antagonist presumably by competing with endogenous GPC1 for binding to Wnt factors (69).

Although the underlying mechanism by which GPC1 regulates Wnt signaling pathway in tumorigenesis remains unknown, the current findings are consistent with what has been found with GPC2 and GPC3. More investigations are needed to shed light on the GPC1/Wnt interaction in tumors. we hypothesize GPC1 acts as coreceptors for Wnt and RSPOs to activate and amplify Wnt cascades. However, the negative regulation in Wnt signaling mediated by GPC1 specifically in trigeminal placodes suggests that GPC1 has the opposite regulatory activity under different physiological or pathological conditions.

HEDGEHOG SIGNALING

The Hedgehog (Hh) signaling pathway is crucial in embryonic morphogenesis (70). Hyperactivation of this pathway has been shown to promote the progression of various cancer types, making the pathway an important drug target (71, 72). There are three Hhs in mammals: Sonic (Shh), Indian (Ihh), and Desert (Dhh) (73). Patched, a conserved 12-pass transmembrane protein, functions as the receptor of Hh proteins. In the absence of Hhs, Patched represses the activity of Smoothened (Smo), a 7-pass transmembrane protein from the Frizzled family of GPCRs that mediates downstream signaling, thereby inhibiting Hh signaling (Fig. 3A). With the binding of Hhs to Patched, the Smo-inhibitory function is abrogated. Activated Smo then triggers a signaling cascade through Gli family of transcription factors to regulate the expression of cell-type-specific genes that control cell proliferation, migration, and differentiation (74) (Fig. 3B).

Figure 3.

GPC1 seems to have dual function; GPC1 can augment or attenuate Hedgehog (Hh) signaling. A: in the absence of Hh ligands, Patched suppresses the activity of Smoothened (Smo), thereby inhibiting the downstream signaling. B: when Hh binds to Patched, the inhibition on Smo is released and the following Hh signaling cascade will be activated. C: in cholangiocytes, GPC1 is proposed to function as a negative modulator, probably by binding to Hh but does not interact with Patched, leading to reduced amount of Hh available for binding to Patched and ultimately decreased Hh signaling with attenuated expression level of target genes (gli2a, ptch1, foxl1, znf697, ccnd1). D: GPC1 acts as a positive regulator of Shh in commissural axons guidance. GPC1 interacts with both Shh and Patched in a way that requires the HS chains and core protein to facilitate Hh-Patched interaction, resulting in increased downstream Hh signaling. GPC1, glypican-1; Hhip, Hedgehog-interacting protein; HS, heparan sulphate; shh, sonic Hh.

Early work in Drosophila cultured cells first identified the regulatory activity of a glypican (dally-like protein) in Hh signaling (75). Differing effects of glypicans on Hh signaling regulation have been shown previously (4). For example, GPC3 has been reported to compete with Patched and acts as an inhibitor in Hh signaling (76, 77), whereas GPC5 on the cell membrane facilitates or stabilizes Hh-Patched coupling by interacting with both Hh and Patched and thereby increases the consequent signaling (78). GPC1 has also been shown to modulate Hh signaling pathway in multiple physiological processes or diseases, including biliary atresia (79) and axon guidance (80). Interestingly, GPC1 seems to have different functions in Hh regulation according to various cellular contexts (Fig. 3, C and D).

GPC1 has been reported to regulate Hh signaling in cholangiocytes in patients with biliary atresia (BA) in a negative manner (79). BA is a progressive fibroinflammatory disorder that affects the extrahepatic and intrahepatic biliary tree. Genome-wide association (GWA) studies previously revealed a region of interest at 2q37.3 that are potentially important in BA (81). Cui et al. (79) expanded the study and uncovered a statistically significant increase in deletions at 2q37.3 that led to deletion of one copy of GPC1 in patients with BA and indicated that GPC1 could be a BA susceptibility gene. Zebrafish GPC genes showed synteny with their human counterparts by genomic and in silico analysis. In situ hybridization studies also revealed the liver-specific expression of gpc1 in larval zebrafish 3 days postfertilization (dpf). Thus, zebrafish gpc1 is a likely ortholog of human GPC1, and zebrafish is a well-suited powerful animal model to study the role of gpc1 in biliary development. Knockdown of gpc1 in zebrafish larvae showed no gross defect at 5 dpf, but caused significant developmental biliary defects, such as less intense gallbladder, fewer and less complex-appearing ducts, as compared with the control group. A previous study has shown that Hh is hyperactive in patients with BA (82). In this study, Cui et al. examined the expression of Hh downstream genes and saw increased expression of the target genes gli2a, ptch1, foxl1, znf697, and ccnd1 in livers from gpc1-knockdown morphants. Moreover, exposure of those gpc1 morphants to the Hh antagonist, cyclopamine, partially rescued the gpc1-knockdown phenotype. These data indicated that elevated Hh activity is at least partially responsible for the biliary defects seen in gpc1 morphants (79). Interestingly, a decrease in GPC1 at the apical surface of cholangiocytes in patients with BA were also observed. Based on these findings, GPC1 was identified as a BA susceptibility gene and proposed to be a negative regulator of Hh signaling in the biliary tree (79), similar to previous report about GPC3 (73).

Wilson et al. (80) investigated the role of GPC1 in regulating Shh switch from attraction to repulsion in commissural axons. GPC1 knockdown in commissural neurons caused axon guidance defects at the neural tube midline (80). Hedgehog-interacting protein (Hhip) acted as a receptor of Shh mediating the repulsive guidance response to Shh (83). Downregulation of GPC1 displayed a specific loss of Hhip expression, indicating that the induction of Hhip expression in commissural neurons was dependent on GPC1 (80). In addition, a Shh-insensitive GPC1 mutant with 10 critical amino acids deletion that was unable to activate Shh signaling, failed to rescue Hhip expression and restore axon guidance following knockdown of endogenous GPC1. These findings demonstrated a functional link between the GPC1/Shh and the induction of Hhip expression and also provided evidence that GPC1 is a positive regulator in Shh-mediated signaling pathways in commissural axons guidance (80). In the same study, the authors also showed that the optimal activity of GPC1 requires both HS chains and GPC1 core protein, as forced expression of a mutated GPC1 that cannot be glycanated only partly rescued the axon guidance defects resulting from GPC1 silencing (80).

Taken together, GPC1 has different functions in Hh mediation according to various cellular contexts. In cholangiocytes cells, GPC1 acts as a negative modulator with attenuated expression levels of Hh target genes (gli2a, ptch1, foxl1, znf697, ccnd1). However, in commissural axons guidance, GPC1 facilitates Shh-Patched interaction and leads to increased downstream Hh signaling (e.g., Hhip).

GPC1 has been shown to modulate multiple signaling pathways in various physiological conditions and cell context including cancer cells. The mechanism of potential cross talk between Wnt, YAP, and other signaling pathways mediated by GPC1 remains undetermined. The evidence that GPC1 can regulate multiple cancer signaling pathways indicates a positive correlation between GPC1 expression and cancer development. Therefore, like GPC3 and GPC2, GPC1 has been examined as a potential biomarker and therapeutic target in cancer.

POTENTIAL ROLE OF GPC1 AS A BIOMARKER IN TUMOR DIAGNOSIS

GPC1 has been reported to be present in two forms, a membrane bound core protein (55–60 kDa) and secreted soluble forms (40 and 52 kDa), which can be detected in serum-free media harvested from prostate cancer cells DU-145 by immunoprecipitation (84), probably due to GPC1 shedding or proteolytic cleavage at GPI anchor that was also found in GPC3 (85). However, the specific cleavage mechanism or site has been unknown. Separately, Melo et al. isolated circulating exosomes (crExos) from serum of patients with breast cancer or pancreatic ductal adenocarcinoma (PDAC). They found that 75% (24/32) of patients with breast cancer and 100% (190/190) of patients with PDAC had remarkedly higher level of GPC1⁺ crExos than healthy individuals, indicating a strong correlation between GPC1⁺ crExos and cancers (86). Elevated GPC1 levels on the surface of exosomes isolated from patients with colorectal (87) and prostate cancer (88) suggest that GPC1 could be used as a noninvasive diagnostic biomarker (89). However, controversial results have been reported, showing that GPC1 may not be a robust biomarker to identify patients with pancreatic cancer (90). For example, GPC1-positive extracellular vesicles (EVs) were detected in patients with PDAC, but the levels were not significantly higher than in patients with benign pancreatic disease (BPD), indicating GPC1-positive EVs could not effectively discriminate between BPD and pancreatic cancer (90). Therefore, more investigations are needed to validate the role of GPC1 in pancreatic cancer diagnosis and its use as a serum biomarker.

GPC1 AS A THERAPEUTIC TARGET

In addition to its potential diagnostic value, GPC1 has been evaluated as a potential target for antibody-based strategies, including anti-GPC1 mAb (19, 91), antibody-drug conjugates (ADC) (92, 93), chimeric antigen receptor (CAR) T cell therapy (6, 91), radiotherapy (94), photoimmunotherapy (95), and bispecific T-cell engager (BiTE) (96) in preclinical settings (Table 2).

Table 2.

Antibody-based therapies specific for glypican-1 in preclinical settings

Format	Clone or Antibody Name	Epitope	Tumor Types	Animal Models	Mechanism	Reference
mAbs	1–12	GPC1-derived peptides 339–358, 388–404, and 405–421	ESCC	ESCC mouse xenograft model (TE14) and PDX model (ESCC-8)	Antibody-dependent cellular cytotoxicity-dependent and -independent manner; decreased angiogenesis	(19)
ADC	01a033	N/A	Uterine cervical cancer	Uterine cervical cancer mouse xenograft model (HeLa, ME180)	G2/M phase cell-cycle arrest mediated by MMAF	(92)
ADC	01a033	N/A	Pancreatic cancer	Pancreatic cancer xenograft mouse (BxPC-3) and PDX model	G2/M phase cell-cycle arrest mediated by MMAF	(93)
CAR-T	HM2	C-lobe of GPC1 close to the cell surface	Pancreatic cancer	Peritoneal dissemination pancreatic cancer xenograft mouse model	Tumor-specific cytotoxic T-cell response with greater amount of TNF-α, IFN-γ, and IL-2	(91)
CAR-T	1–12	GPC1-derived peptides 339–358, 388–404, and 405–421	Solid tumors	Xenogeneic mouse model (TE14); syngeneic mouse models (MC38-mGPC1 or MCA205-mGPC1)	Tumor-specific CD8+ T-cell responses; the murine CAR T cells enhanced endogenous T-cell responses against a non-GPC1 tumor antigen through the mechanism of antigen spreading	(6)
Radiotherapy	MIL-38 (177Lu-labeled Miltuximab)	N/A	Prostate cancer	Prostate cancer xenograft mouse model (DU-145)	Cancer cell apoptosis	(94)
Photoimmuno-therapy	MIL-38 (Miltuximab-IRDye700DX)	N/A	Solid tumors	N/A	Cytotoxicity; irreversible loss of membrane integrity and death of the tumor cells caused by photoactivation of the conjugate by near-infrared (NIR) light at 690 nm	(95)
BiTE	MIL-38 (Miltuximab)	N/A	Prostate cancer	N/A	Activation of peripheral blood T cells and the release of inflammatory cytokines TNF and IFN-γ	(96)

ADC, antibody-drug conjugates; BiTE, bispecific T-cell engager; CAR-T, chimeric antigen receptor T cells; ESCC, esophageal squamous cell carcinoma; GPC1, glypican-1; MMAF, monomethyl auristatin F; N/A, not available; PDX, patient-derived xenograft.

Our laboratory isolated the mouse monoclonal antibody HM2 that binds the C-lobe of GPC1 close to the cell surface (91). GPC1-overexpressing cancer cells were efficiently lysed by HM2 CAR T cells. GPC1-targeted CAR T cells demonstrated potent antitumor efficacy in a peritoneal dissemination xenograft mouse model, indicating that GPC1-targeted CAR T cell therapy could be a new approach for treating pancreatic cancer.

The anti-GPC1 antibody (clone 1–12) was isolated from chicken and recognized both human and mouse GPC1. It strongly inhibited tumor growth in both GPC1-positive ESCC xenograft models and patient-derived tumor xenograft (PDX) models via antibody-dependent cellular cytotoxicity-dependent and -independent manner (19). The same group then developed an antibody-drug conjugate (ADC) by conjugating another newly isolated anti-GPC-1 monoclonal antibody (clone 01a033) with the cytotoxic agent monomethyl auristatin F (MMAF). The GPC1/ADC showed significant and potent tumor growth inhibition in murine xenograft models bearing uterine cervical cancer or pancreatic cancer by G2/M-phase cell-cycle arrest (92, 93). Furthermore, GPC1-specific human and murine CAR T cells derived from anti-human/mouse GP1 antibody (clone 1–12) were generated and showed strong antitumor effects in xenogeneic and syngeneic solid tumor mouse models without any obvious adverse effects (6).

Chimeric antibody Miltuximab is a GPC1-specific human IgG1 derived from the parent antibody MIL-38 in clinical development for solid tumor therapy. The use of radio-labeled MIL-38-based therapy has shown promise in bladder cancer animal models (97). The ¹⁷⁷Lu-labeled Miltuximab was developed by conjugating with lutetium-177 and showed 47% reduction in the average tumor size as compared with the control in a prostate cancer mouse xenograft model (94). Polikarpov et al. then conjugated Miltuximab with a photosensitizer near-infrared dye IRDye700DX (IR700) and investigated its efficacy in solid tumors. Miltuximab-IR700 caused 67.3%–92.3% reduction in the viability of cells with medium-high GPC-1 expression due to the targeted binding of the conjugate with subsequent photoactivation (95). Moreover, a BiTE (MIL-38-CD3) was generated by combining the single-chain variable fragment (scFv) of Miltuximab and the CD3 binding sequence of blinatumomab. MIL-38-CD3 caused the activation of peripheral blood T cells, induced the release of inflammatory cytokines TNF and IFN-γ, and eventually redirected those activated T cells to lyse GPC1-positive prostate cancer cells without any killing to GPC1-negative Raji cells (96).

Taken together, GPC1 is being evaluated as a therapeutic target in multiple types of human cancers. Further preclinical and clinical studies will define the tumor specificity (e.g., on-target off-tumor effect) and efficacy of these candidates in various tumor microenvironments.

CONCLUSIONS

The structure of the GPC1 has been largely solved with the lack of the C-terminal flexible region that contains HS chains and GPC1 anchorage. The unknown orientation (and potential rotation) of GPC1 and other glypicans relative to the cell surface is a fundamental question related the physiological features of glypicans. GPC1 is clearly a crucial player in multiple signaling cascades that affect tumor progression or other diseases via its core protein and the HS chains. Important questions regarding the specific mechanisms by which GPC1 modulates the outlined signaling pathways in different cellular contexts, including tumor microenvironments, are poorly understood. It is also important to study the potential cross talk mechanisms between Wnt, YAP, and other signaling pathways modulated by GPC1. Additional investigations are required to provide a more complete picture of its precise role in cancers and other important physiological processes. It is intriguing that studies have demonstrated the potential role of GPC1 as a therapeutic target in cancer. Tumor specificity of these new therapies, including potential on-target off-tumor effects, should be carefully examined in preclinical studies. Various novel and emerging strategies, such as targeting tumor-specific isoforms/exons (8) and synNotch-controlled CAR T cells (98), have been proposed to improve tumor specificity and/or efficacy. Ongoing preclinical studies will help establish GPC1 as a new cancer target in major solid tumors such as pancreatic cancer, which have very limited therapeutic options.

GRANTS

This work was supported by the Intramural Research Program of NIH, Center for Cancer Research (CCR), National Cancer Institute (NCI) (Z01 BC010891 and ZIA BC010891 to M.H.).

DISCLOSURES

The authors are inventors on international patent applications no. PCT/US2020/013739, “High affinity monoclonal antibodies targeting glypican-1 and methods of use thereof.” The authors declare no other conflicts of interest. The content of this publication does not necessarily reflect the views or policies of the Department of Health and Human Services nor does mention of trade names, commercial products, or organizations imply endorsement by the US Government. The antibodies targeting glypican-1 (GPC1) are available for licensing, in a wide range of fields of use, from the National Cancer Institute, NIH. If you are interested in obtaining a license, please contact the principal investigator Dr. Mitchell Ho at homi@mail.nih.gov.

AUTHOR CONTRIBUTIONS

M.H. conceived and designed research; M.H. and J.P. interpreted results of experiments; J.P. prepared figures; J.P. and M.H. drafted manuscript; J.P. and M.H. edited and revised manuscript; M.H. approved final version of manuscript.