Numerous signal-transduction-related molecules are secreted proteins or membrane proteins, and the mechanism by which these molecules are regulated by glycan chains is a very important issue for developing an understanding of the cellular events that transpire. This review covers the functional regulation of epidermal growth factor receptor (EGFR), ErbB3 and the transforming growth factor β (TGF-β) receptor by N-glycans. This review shows that the N-glycans play important roles in regulating protein conformation and interactions with carbohydrate recognition molecules. These results point to the possibility of a novel strategy for controlling cell signalling and developing novel glycan-based therapeutics.
Cell signalling is ubiquitous in multicellular animals, and cell surface receptors mediate a variety of cellular responses. Since most cell surface receptors are glycosylated, an analysis of the glycan chains of receptors is therefore crucial for our understanding of cell signalling and subsequent physiological and pathological cellular events including cell growth, differentiation, tumour metastasis and immune reaction [1–8].
There are limited strategies for examining the functional aspects of protein glycosylation. To examine the function of a particular glycan, it is frequently necessary to prepare mutant proteins that lack glycans. The difference in the target protein with or without the particular glycan at glycosylation sites serves as an indicator of the function of the glycan. In this case, the target proteins used are frequently recombinant, and information concerning the function of the glycan structure is not available. To examine the function of a specific glycan structure, it is necessary to manipulate the glycan structure by introducing or knocking down/knocking out (a) glycosyltransferase gene(s). In this case, it is difficult to determine the target protein(s), and interpretating the results must be done with care [9,10].
N-glycans of many receptors have been studied by using these strategies. In this review, the functional regulation of epidermal growth factor receptor (EGFR), ErbB3 and the transforming growth factor β (TGF-β) receptor by N-glycans are summarized and discussed. The results suggest that the functional regulation of cell surface receptors by N-glycans have some common characteristics.
FUNCTIONAL REGULATION OF ErbB RECEPTORS BY N-GLYCANS
General aspects of ErbB receptors: EGFR, ErbB2, ErbB3 and ErbB4
The ErbB family consists of four members, EGFR (ErbB1/HER1), ErbB2/HER2, ErbB3/HER3 and ErbB4/HER4. They are involved in a variety of cellular events, including proliferation, differentiation and migration, and their aberrant signalling has been implicated in the initiation and maintenance of various types of cancers. More than 30% of breast cancers, 40% of glioblastomas, 60% of non-small cell lung cancers and 70% of metastatic colorectal cancers are associated with the dysregulation of ErbB receptors. Therefore, ErbB receptors are considered to be targets of cancer therapy [11–18].
The ErbB receptors are type I transmembrane receptor tyrosine kinases, and consist of an N-terminal extracellular domain, a transmembrane domain, an intracellular tyrosine kinase domain and a C-terminal tail (Figure 1). The extracellular domain is divided into four subdomains, i.e., I–IV. Domains I and III have β-helical folds, and domains II and IV contain several cysteine-rich elements. In the unliganded state, ErbB receptors exit in a ‘tethered form (=inactive form)’. In this state, molecules are folded in a manner that prevents them from undergoing dimerization and domains I and III are oriented such that the ligand cannot be in contact with both (Figure 2) [19–22]. In this conformation, the dimerization arm in domain II interacts with the disulfide-bonded loop in domain IV, termed the ‘tether loop’, to form an auto-inhibitory tether. The binding of ligands to the extracellular domain alter the conformation from a ‘tethered form’ to an ‘extended form (=active form) ’ in which the dimerization arm in domain II mediates homo- or hetero-dimers. The dimerization process is considered to activate tyrosine kinase, with the subsequent phosphorylation of the tyrosine residues in the C-terminal tail. These phosphotyrosine residues recruit downstream signalling molecules that contain the SH2 and/or PTB domains, and initiate downstream signalling cascades. Among the four members of the ErbB receptor family, ErbB2 lacks ligand-binding activity and is constitutively in an ‘extended form’ suitable for dimerization, and ErbB3 lacks tyrosine kinase activity. Consequently, these two receptors form heterodimers with other ErbB receptors.
The structure and N-glycosylation sites of ErbB receptors (adapted from Takahashi : Takahashi, M. (2014) Functional regulation of ErbB receptors by N-glycans. In Glycosciences: Biology and Medicine (Taniguchi, N., Endo, T., Hart, G.W., Seeberger, P.H. and Wang, C.-H., eds), pp. 983–989, Springer Japan, Tokyo)
Scheme showing the signal transduction of ErbB receptors
The extracellular domains of ErbB receptors are highly glycosylated. EGFR, ErbB2, ErbB3 and ErbB4 contain 12, 8, 10 and 11 N-glycosylation sites in their extracellular domains respectively. The scheme shown in Figure 1 indicates the N-glycosylation sites of EGFR and ErbB3, and the alignment of the N-glycosylation sites of all ErbB family members. It shows that some glycosylation sites are common among the ErbB receptors. (We use a numbering system that does not count the signal peptides of the N-terminal 24 amino acids of EGFR, the 22 amino acids of ErbB2, the 19 amino acids of ErbB3 or the 25 amino acids of ErbB4.)
N-glycans of EGFR
EGFR is a 170 kDa protein that consists of 1186 amino acid residues. The EGFR ligand family consists of seven growth factor members, including epidermal growth factor (EGF), transforming growth factor α, amphiregulin, epigen, β-cellulin, heparin-binding EGF-like growth factor and epiregulin. On ligand binding, EGFR forms a homodimer or a heterodimer, and activates downstream signalling pathways, including Ras/Erk, phosphoinositide 3-kinase (PI3K)/Akt, Src kinases and STAT transcription factors. EGFR signalling is involved in a wide variety of cellular events, and the aberrant expression or dysregulation of EGFR is implicated in cell transformation and tumour development.
EGFR has 11 typical (N-X-S/T, where X is any amino acid except proline) and 4 atypical (N-X-C) N-glycosylation consensus sequences, and the N-glycans of EGFR make up approximately 40 kDa of the total molecular mass [23–25]. The structure of the N-glycan on each site is different depending on the cell type. Figure 3 indicates the status and structure of N-glycans on each glycosylation site of endogenous EGFR in A431 human epidermoid carcinoma cells , recombinant EGFR, expressed in CL-1 human lung cancer cells  and recombinant soluble EGFR (sEGFR, the extracellular domain of EGFR), expressed by CHO-K1 cells . For example, in sEGFR in CHO-K1 cells, 11 out of 11 typical N-glycosylation consensus sequences (N104, N151, N172, N328, N337, N389, N420, N504, N544, N579, N599) are either fully or partially glycosylated, and one of the four atypical N-glycosylation consensus sequences (N32) is fully glycosylated . It is noteworthy that N337 is glycosylated by oligomannose type N-glycans in all cell types examined, a point that will be discussed in detail later in this section.
Site-specific N-glycans of EGFR
The roles of N-glycans in the functions of EGFR have been investigated. For glycosylation in general, it has been found that an initial N-glycosylation of EGFR is required for the proper sorting to the cell surface and for ligand binding [24,29–31]. Once EGFR is glycosylated, it acquires ligand-binding ability, and the processing of oligosaccharides from oligomannose type to complex type has no effect on this ability .
More recent studies provide information concerning the function of particular N-glycans of EGFR. It has been demonstrated, by using N420Q single N-glycan deletion mutants of EGFR, that the N-glycan on N420 of EGFR is involved in dimerization . Among the four single N-glycan deletion mutants of EGFR that each lack the glycosylation sites in domain III, the EGFR N420Q mutant was observed to undergo ligand-independent oligomerization, resulting in the phosphorylation of EGFR. Whitson et al.  reported that the deletion of the N-glycan on N579 by the N579Q mutation, which is in the domain IV auto-inhibitory tether loop, weakens the interaction between domains II and IV, thereby increasing the ratio of the high affinity ligand-binding population and also increasing ligand-independent dimerization. They assumed that the N-glycan attached to N579 contributes to tether stability and thus the deletion of the N-glycan results in increasing conformational flexibility. These studies suggest that specific N-glycans are involved in maintaining the conformation of EGFR. Those glycans may play important roles in stabilizing EGFR in the tethered form, and in preventing ligand-independent dimerization. Recent atomistic molecular dynamics simulations have also demonstrated that N-glycans are determinants of protein conformation [34,35]. Kaszuba et al.  suggested that N-glycans determine the conformation of EGFR, specifically the orientation of the EGFR extracellular domain to the membrane.
It is also important to note that some of the glycan structures of EGFR are involved in the regulation of its function. Based on the following, we have concluded that the interaction of specific N-glycans with carbohydrate recognition molecules such as lectins is another crucial mechanism of functional regulation of EGFR by N-glycans. The following are descriptions of such cases.
EGFR contains both oligomannose-type and complex-type N-glycans, although the glycoform depends on the cell type expressing the EGFR. As shown in Figure 3, N337 is modified by a oligomannose type of N-glycan in all three cell types examined. A recent study has suggested that pulmonary surfactant protein D (SP-D) binds to the oligomannose type N-glycan of EGFR, and down-regulates EGF signalling in lung adenocarcinoma cells and human EGFR stable-expressing CHO-K1 cells . SP-D is an apoprotein of a pulmonary surfactant and belongs to the collectin subgroup of the C-type lectin superfamily. It thus recognizes oligomannose type N-glycans in a Ca2+-dependent manner [36–38]. It has been reported that collectins interact with endogenous membrane proteins such as toll-like receptors [39–42], calreticulin, or SIRPα , Uroplakin Ia  and pathogenic microbes [45–49]. Hasegawa et al.  reported that SP-D directly binds to the extracellular domain of EGFR via interaction of the carbohydrate recognition domain of SP-D with N-glycans of EGFR, and down-regulates the binding of EGF to EGFR in A549 human lung adenocarcinoma cells. A structural analysis of N-glycans indicates that the EGFR in A549 cells and the sEGFR expressed in CHO-K1 cells contain oligomannose type N-glycans. For example, N328 and N337 of the sEGFR in CHO-K1 cells are modified by glycans with Man5GlcNAc2 to Man9GlcNAc2. These data suggest that SP-D binds to the oligomannose-type N-glycans of EGFR, and down-regulates EGF binding and downstream signalling. It is speculated that the binding of SP-D to EGFR directly interferes with EGF binding, or that SP-D affects the conformation of EGFR and alters the binding characteristics of EGF to EGFR.
It has been reported that EGFR with poly-N-acetyllactosamine-containing N-glycans, which is produced by the activity of N-acetylglucosaminyltransferase V (GnT-V), avoids constitutive endocytosis . It has been proposed that galectin-3 binds to poly-N-acetyllactosamine on the EGFR to form a molecular lattice, and sustains the cell surface residency of EGFR. Similarly, it has been observed that the endocytosis of EGFR is enhanced in N-acetylglucosaminyltransferase III (GnT-III) transfected HeLaS3 cells . GnT-III catalyses the introduction of GlcNAc to the mannose residue at the base of the trimannosyl core of the N-glycans, to produce ‘bisecting GlcNAc’ (Figure 4) [52,53]. The bisecting GlcNAc prevents further N-glycan processing, since other glycosyltransferases are unable to act on the resulting triantennary glycans , and therefore, the overexpression of GnT-III reduces the formation of poly-N-acetyllactosamine on N-glycans of EGFR. It is possible that the binding of N-glycans of EGFR to galectin-3 is decreased in GnT-III transfectants, resulting in the up-regulation of EGFR endocytosis. Thus, specific terminal structures of N-glycans may regulate EGFR endocytosis through interactions with carbohydrate recognition molecules, such as galectin-3.
The activities of GnT-III, GnT-V and Fut8
EGFR is also regulated by glycolipids. GM3 (NeuAcα3Galβ4Glcβ1Cer) has been shown to interact with EGFR and down-regulate its activation [55–62]. The binding activity of GM3 is much higher than that for other gangliosides, such as GM2, GD3, GM4, GM1, GD1a and GT1b . It has been revealed that GM3 interacts with complex-type N-glycans with multivalent (more than 3) GlcNAc termini through carbohydrate-to-carbohydrate interactions (CCI). GM3 binds to the N-glycans of EGFR and inhibits the activation of tyrosine kinase and subsequent downstream signalling without affecting the binding of EGF to EGFR [63–65]. Thus, CCI is involved in the functional regulation of EGFR by N-glycans, although CCI are usually weaker than protein-to-carbohydrate interactions. It has also been demonstrated that membrane-associated sialidase NEU3, whose selective substrates are GM3 and GD1a , activates EGFR , and the involvement of GM3 in cancer has been indicated [68–71].
The α1,6-fucosylation of N-glycans via the activity of α1,6 fucosyltransferase (Fut8) has been shown to affect EGFR function . Fut8 catalyses the addition of a fucose to the innermost GlcNAc residue of N-glycans, to produce α1,6-fucosylation, or ‘core fucose’ (Figure 4). The activity was first identified by Schachter's group , and it was purified to homogeneity and its cDNA was cloned by our group [74,75]. Core fucose has a variety of pathophysiological functions [9,10,76–79]. It has been reported that the loss of α1,6-fucosylation of N-glycans of EGFR causes a reduction in the binding of EGF to EGFR, and subsequently, a reduction in downstream signalling . It has also been suggested that the increased sialylation and α1,3-fucosylation of the N-glycans of EGFR suppress EGF-induced EGFR dimerization [27,80]. Sialylation is catalysed by sialyltransferases, and α1,3-fucosylation is catalysed by α1,3 fucosyltransferases (Fut4 or Fut6). These results of the functional regulation of EGFR by specific glycan structure are summarized in Figure 5.
Functional regulation of EGFR by specific N-glycans
N-glycans of ErbB3
ErbB3 is a 185 kDa protein that consists of 1323 amino acid residues. The ligands of ErbB3 are neuregulin1 (heregulin), neuregulin 2 and neuregulin 6. As discussed in section ‘General aspects of ErbB receptors: EGFR, ErbB2, ErbB3 and ErbB4’, the unique character of ErbB3 is its lack of tyrosine kinase activity [81–83]. Therefore, ErbB3 forms a heterodimer with other ErbB receptors and exerts downstream signalling, such as the PI3K/Akt pathway or Ras/Erk pathway. The ErbB2/ErbB3 heterodimer is considered to be the most active ErbB signalling dimer [84–86]. Most cancer cells require PI3K/Akt signalling for their survival, and PI3K activation has been implicated in the ErbB3-dependent progression of various types of cancer [87–89] due to the fact that ErbB3 contains seven binding sites for PI3K in contrast with one in ErbB4 and none in EGFR and ErbB2 [90–93]. The induction of ErbB3 expression or signalling is an important factor in drug resistance in several cancer models. For example, ErbB3 expression or signalling is associated with ErbB2 tyrosine kinase inhibitor resistance in ErbB2-amplified breast cancers , anti-ErbB2 monoclonal antibody pertuzumab resistance in ovarian cancers and breast cancers [95,96], and EGFR tyrosine kinase inhibitor resistance in lung cancers and head and neck cancers [97,98]. ErbB3 is also associated with resistance to anti-oestrogen therapy in oestrogen receptor positive breast cancers [99–101], hormone resistance in prostate cancers  and insulin-like growth factor-1 receptor inhibitor resistance in hepatomas . Thus, ErbB3 can be considered to be a novel target for cancer therapy [104–107].
ErbB3 contains 10 N-glycosylation consensus sequences in the extracellular domain. The N-glycan on N418 of ErbB3, which corresponds to N420 of EGFR based on sequence alignment (Figure 1), is involved in dimer formation. Among the 10 single N-glycan deletion mutants of ErbB3 that lack each of the glycosylation sites, only the ErbB3 N418Q mutant forms a heterodimer with ErbB2 and exerts downstream signalling without ligand stimulation, and promotes tumour formation in athymic mice . It is assumed that this phenomenon corresponds to that observed in the EGFR N420Q mutant discussed in section ‘N-glycans of EGFR’. Whether the conformation of the ligand-independent dimerization of the EGFR N420Q mutant or the ErbB3 N418Q mutant is similar to that of ligand-induced dimerization remains an issue yet to be resolved. Nevertheless, these findings suggest that the N-glycans play a role in maintaining the inactive form of ErbB receptors in the absence of a ligand. We assume that the conformational changes from the inactive to the active form occur with less energy in these N-glycan deletion mutants .
The extracellular domain of ErbB3 (=soluble ErbB3, sErbB3) has suppressive effects on heregulin signalling, and the effects are enhanced when the N-glycan on N418 is deleted [110,111]. Since it has been observed that sErbB3 acts on cell surface receptors but has no effect on the ligand, it is possible that the sErbB3 N418Q mutant binds to ErbB2 or other ErbB receptors on the cell surface at a higher frequency than the wild type. Similar results have been observed in sErbB3 with both oligomannose-type and complex-type N-glycans . It therefore appears that the effect is not influenced by the structural differences in the N-glycan termini.
FUNCTIONAL REGULATION OF TGF-β RECEPTORS BY N-GLYCANS
TGF-β plays important roles in cell proliferation, differentiation, apoptosis and development [112–115]. TGF-β signalling is mediated by type I transmembrane receptors, the TGF-β receptor type I (TβRI) and the type II (TβRII), which are the sole cell surface receptor serine/threonine kinases found in humans. Both TβRI and TβRII comprise approximately 500 amino acids, and consist of an N-terminal extracellular domain, a transmembrane domain and a C-terminal serine/threonine kinase domain. The binding of TGF-β to TβRII dimers recruits two TβRI molecules to form a ligand–receptor complex involving a heterotetrameric receptor and the ligand dimer. Ligand binding changes the orientation of TβRII so that TβRII phosphorylates TβRI in the short juxtamembrane GS domain. The phosphorylated GS domain of TβRI then interacts with the MH2 domains of Smad2 and Smad3, and the TβRI kinase phosphorylates the Smad on two serines within its conserved C-terminal SXS motif. The activated Smad proteins form trimeric complexes with Smad4 and are translocated to the nucleus, where they regulate the transcription of target genes.
Aberrant TGF-β signalling is involved in many diseases, including skeletal defects, cancer and fibrosis. For instance, TβRII has been found to be mutated and inactivated in most gastrointestinal cancers and Smad4 is mutated in more than half of all pancreatic cancers . TGF-β signalling is also involved in the homoeostasis in the respiratory systems of mammals and plays a key role in various diseases including pulmonary hypertension, asthma, pulmonary fibrosis and emphysema [117–119].
TGF-β receptors are regulated by post-translational modifications, such as ubiquitylation and sumoylation [120–122]. Among these modifications, N-glycosylation is thought to be important for the function of TβRII [50,123–125]. TβRII contains two or three potential N-glycosylation sites in its extracellular domain. (Isoform A contains three and isoform B contains two.) Kim et al.  reported that the N-glycans of TβRII determine the sensitivity toward the ligand. They indicated that at least one out of two N-glycans in the mouse TβRII isoform B is required for its transport to the cell surface. These two asparagine residues, N70 and N94, are conserved in humans, mice, rats and even zebrafish, implying that they have an important role in the function of TβRII.
It has been reported that TβRII with poly-N-acetyllactosamine-containing N-glycans produced by the activity of GnT-V, binds to galectin-3 at the cell surface, and avoids constitutive endocytosis in a manner similar to that described for EGFR in the previous section . The Golgi pathway is sensitive to hexosamine flux for the production of N-glycan branching, which binds to galectin to oppose endocytosis. As a glycoprotein with few N-glycans, TβRII exhibits enhanced cell surface expression with switch-like responses to increasing hexosamine concentration, whereas EGFR, being a glycoprotein with high numbers of N-glycans, exhibits hyperbolic responses, in which the lag phase of the sigmoidal responses is suppressed and the slopes are steeper at low hexosamine flux than at high flux . This suggests that both the degree of N-glycan branching and the number of N-glycans per receptor are important determinants for regulating cell surface expression levels in response to a hexosamine flux.
As stated in the previous section, core fucosylation has a variety of pathophysiological functions. It has been reported that Fut8-null mice exhibit emphysema changes in the lung with significant overexpression of matrix metalloproteinase (MMP), and cigarette smoking makes these changes more obvious [124,125]. TGF-β signalling is down-regulated in Fut8-null mice, and the binding of TGF-β is decreased significantly in Fut8-null cells, but is rescued by the reintroduction of Fut8. The impaired TGF-β signalling results in the down-regulation of Smad phosphorylation, which usually suppresses MMP gene expression, and in the overexpression of MMP-9, -12 and -13 in the lungs in Fut8-null mice (Figure 6). Although the mechanisms by which core fucosylation affects TGF-β binding have not yet been fully elucidated, it is assumed that some carbohydrate recognition molecules that specifically bind to the core fucose may be involved in the phenomenon.
Model scheme for the onset of emphysema in Fut8-null mice
The functions of cell signalling molecules are often controlled by N-glycans. The fundamental function of glycan chains is to modify the physicochemical properties of target molecules. In the case of glycoproteins, functional regulation by glycan chains can be mainly categorized into two patterns: regulation of a protein conformation, which depends on the glycan core, and the regulation of interactions with carbohydrate recognition molecules, which depends on the glycan termini. In the case of ErbB receptors, it is thought that specific N-glycans regulate dimerization, probably by affecting the conformation of receptors. N-glycans with specific terminal structures interact with specific carbohydrate recognition molecules such as SP-D, galectin-3 and GM3, and regulate ligand binding, autophosphorylation and the endocytosis of the receptors. These mechanisms may be applied to many other cell surface molecules. In particular, in most cases of the functional regulation of receptors by glycosyltransferases, specific carbohydrate recognition molecules may play key roles.
To understand cell surface receptor signalling, elucidating the role of glycosylation is imperative. Integrated analyses of N-glycans will give further insights into molecular mechanisms that govern cell signalling.
This work was supported by the Japan Society for the Promotion of Science [grant number 26440058 (to M.T.), 16K19461 (to Y.H.), 23591118 (to C.G.), 25461194 (to Y.K.), 15H04700, 15K14481 (to N.T.)]; the Takeda Science Foundation [grant number 1400343 (to M.T.)]; and the Suhara Foundation [grant number 1400918 (to M.T.)].
epidermal growth factor
epidermal growth factor receptor
surfactant protein D
transforming growth factor β
transforming growth factor β receptor type I
transforming growth factor β receptor type II