It is well known that almost all the secreted-type proteins produced in eukaryotic cells are N-glycosylated, suggesting that the N-glycosylation of proteins and the subsequent modification of the glycan moiety are ubiquitous and pivotal biological process in eukaryotes, especially the multicellular organisms (Lerouge et al., 1998; Kobata, 2007). In the protein quality-control mechanism working in the endoplasmic reticulum (ER), it is believed that these N-glycans are linked to the nascent glycoproteins in the ER and play a critical role as ligands for calnexin or calreticulin, molecular chaperons that assist the protein-folding system in the ER (Suzuki et al., 2007). In plant cells, for example, deletion of ER α-glucosidase I, an enzyme that hydrolyzes α-1,3-glucosidic linkages in the nascent and premature N-glycans and controls the interaction of the unfolded nascent glycoproteins and the molecular chaperons, results in death (Boisson et al., 2001; Gillmor et al., 2002). N-Glycans that are biosynthesized in eukaryotic cells (regardless of animal, plant, insect, fungi, and yeast) share a common trimannosyl core structure [Manα1-6(Manα1-3)Manβ1-4GlcNAcβ1-4GlcNAc], but the final structure of these N-glycans varies by species. In the case of animals, plants, and insects, the N-glycan structures are classified into three subgroups: high-mannose type, hybrid type, and complex type depending on the location of additional sugar residues transferred to the trimannosyl core. The high-mannose type N-glycans are common in all eukaryotic cells, but the structural features of complex-type N-glycans vary widely depending on the species. The structural features of plant N-glycans are as follows: (1) the occurrences of β1,2-xylosyl (Xyl) residue linked to β1,4-mannosyl (Man) residue and α1,3-fucosyl (Fuc) residue linked to the reducing end N-acetylglucosaminyl (GlcNAc) residue, (2) the occurrence of Lewis a (Le^a) epitope [Galβ1-3(Fucα1-4)GlcNAc] at the non-reducing end, and (3) the lack of N-acetylneuraminic acid residue and β1,4-linkage galactosyl (Gal) residue. The combination of α1,3-Fuc and β1,2-Xyl linked to the trimannosyl core is an outstanding characteristic of the plant complex type N-glycans. The most abundant complex type N-glycan linked to storage glycoproteins in seeds or vacuole-accumulated glycoproteins is Manα1-6(Manα1-3)(Xylβ1-2)Manβ1-4GlcNAcβ1-4(Fucα1-3)GlcNAc, which is sometimes described as a truncated-type, pauci-mannose-type, or vacuolar-type structure. Furthermore, the secreted plant glycoproteins sometimes carry large complex type N-glycans with Le^a epitope. The plant N-glycans harboring the Le^a epitope, which are sometimes referred to as secreted type structures, have been found in many foodstuffs and pollen allergens (Fichette-Lainè et al., 1997; Alisi et al., 2001; Wilson et al., 2001; Kimura et al., 2005; Maeda et al., 2005), although their physiological function in plants remains to be elucidated. The conversion of N-glycan structures from high-mannose type to complex type seems to play an important role in adaptation to environmental changes. It has been reported that the maturation of N-glycans to the complex type N-glycans in the Golgi apparatus is a prerequisite for the sufficient cell-wall formation under salt stress (Kang et al., 2008). Typical structures of N-glycans linked to plant glycoproteins are shown in Figure 1, and the possible processing pathway of plant N-glycans is outlined in Figure 2.

As the name implies, N-glycans are linked to a specific asparagine (Asn) residue in eukaryotic proteins, but free forms of these glycans, namely free N-glycans (FNGs), have been found in seedlings, stems, developing fruits, and culture cells (Priem et al., 1990b, 1993; Faugeron et al., 1997a,b; Kimura et al., 1997; Kimura and Matsuo, 2000; Kimura and Kitahara, 2000). Since similar FNGs have been found in animal cells, the occurrence of these FNGs is not just limited to plants. FNGs found in both plants and animals are classified into two types, GN1 type and GN2 type, based on the reducing terminal structure; GN1 type FNGs with one GlcNAc residue and GN2 type FNGs with GlcNAcβ1-4GlcNAc (N-acetylchitobiosyl unit). The structures of typical FNGs found in plants are summarized in Figure 3.

It is generally believed that FNGs are produced from the misfolded glycoproteins by cytosolic peptide:N-glycanase (cPNGase), which hydrolyzes the amide linkage in the glycosylated Asn residue, prior to degradation of the misfolded proteins by proteasomes in the cytosol. The disposal of misfolded/miss-associated proteins is known as ER-associated degradation (ERAD), and the resulting FNGs produced by cPNGase always belong to high-mannose type structure with the N-acetylchitobiosyl unit at their reducing ends (GN2 type FNGs). As for plant cPNGase, it was confirmed that the product of putative Arabidopsis PNGase gene when transformed in budding yeast facilitates the glycoprotein ERAD activity, suggesting that the gene product probably possesses PNGase activity and is likely involved in deglycosylation of misfolded glycoproteins (Diepold et al., 2007; Masahara-Negishi et al., 2012). In the cytosolic metabolism of FNGs released from misfolded proteins by cPNGase, GN2 type FNGs are rapidly modified by cytosolic endo-β-N-acetylglucosaminidase (cENGase), which hydrolyzes the glycosidic linkage in the N-acetylchitobiosyl unit to form the GN1 type FNGs (Suzuki and Funakoshi, 2006). Since almost all high-mannose type FNGs belong to the GN1 type, it is clear that ENGase is involved in the production of the high-mannose type FNGs found in plants, but there is no experimental evidence suggesting that such high-mannose type GN1-FNGs are exclusively generated from GN2 FNGs (cPNGase products) and not from the misfolded glycoproteins by the action of ENGase.

On the other hand, almost all plant complex type and truncated type FNGs belong to GN2 type, suggesting that only PNGase, but not ENGase, must be involved in the production. As we discuss later, the plant ENGase shows strong activity against high-mannose type N-glycans having Manα1–2Manα1–3Manβ1-unit, but no activity against plant complex and truncated structures having GlcNAcβ1-4(Fucα1-3)GlcNAc-unit at their reducing ends. It is, therefore, reasonable to conclude that almost all plant complex type FNGs have the GN2 type structure. Based on the biosynthesis or processing mechanism of N-glycans linked to plant glycoproteins, as shown in Figure 2, it can be considered that these complex type FNGs must be released from matured and secreted glycoproteins fully processed in the Golgi apparatus, but not from the misfolded glycoproteins that are fated to be degraded in the cytosol. Plant specific aPNGase involved in the production of plant complex type FNGs is genetically different from cPNGase localized in the cytosol and has the optimum activity in the acidic environment. The occurrence of two subtypes of GN2-FNGs, with the truncated structure and large size structure as shown in Figure 3, suggests that there may be two kinds of aPNGase in plants; one works in the vacuole and the other in the extracellular space such as cell wall or the apoplast.

As described above, the structural features of FNGs found in plants indicate their origins (misfolded proteins in ER or fully processed and secreted proteins) and the enzyme, PNGase or ENGase, that is involved in the final stage of the FNGs-production. For example, structural analysis of FNGs found in cultured rice cells and culture broth provided valuable information on the deglycosylation mechanism in plant cells (Maeda et al., 2010). In the soluble fraction (a mixture of cytosolic and vacuolar fractions) of cultured cells, only high-mannose type GN1-FNGs and truncated complex type GN2-FNGs were found, suggesting that the former FNGs were produced by ENGase and the latter by aPNGase. It is believed that the plant complex type N-glycans linked to vacuolar-targeting glycoproteins are modified to the truncated type structure in the vacuole (Vitale and Chrispeels, 1984; Lerouge et al., 1998), and it is reasonable to assume that truncated complex type FNGs must be produced from deteriorated glycoproteins (or glycopeptides generated by endopeptidases) by vacuolar PNGase. On the other hand, high-mannose type GN2-FNGs (M_9-5GN₂) and two longer complex type FNGs having N-acetylglucosamine residue(s) or Le^a unit(s) at the non-reducing end (GN1 and GN2 species) have been identified from the culture broth (Figure 3B; Maeda et al., 2010). The occurrence of GN2-FNGs (both high-mannose type and longer complex type structures) in the culture broth suggests that the secreted-type aPNGase in either the cell wall or apoplastic area plays a critical role in the release of N-glycans in the extracellular space. In the structural analysis of rice FNGs, some plant complex type FNGs with unique structural feature were found in the culture broth from rice cell (Maeda et al., 2010). As shown in Figure 3B, the unusual FNGs carry the complex type structure at their non-reducing end and only one GlcNAc residue at their reducing end, suggesting that these FNGs were fully processed in the Golgi apparatus and the ENGase were involved in the process. As described previously, ENGases that have been purified and characterized to date cannot hydrolyze the glycosidic linkage of the N-acetylchitobiosyl unit of plant complex type N-glycans, and no other ENGase that is active against plant complex type N-glycans has been found. In summary, the occurrence of the plant complex type GN1-FNGs cannot be explained by the established scheme of deglycosylation and processing mechanism of N-glycoproteins. On the other hand, in the case of animal cells, the complex type GN1-FNGs were recently found in the stomach cancer cultured cells (Ishizuka et al., 2008), suggesting that the occurrence of complex type GN1-FNGs is not limited to plants. However, the animal complex type N-glycans usually carry α1-6 fucosyl (but not α1-3 fucosyl) residue linked to the innermost GlcNAc residue, and this type of animal complex type glycans can be a modest substrate for some types of ENGase. At present, therefore, the possibility of involvement of ENGase in the production of complex type GN1-FNGs in animal cells cannot be excluded.

A putative processing scheme for the formation of the unusual FNGs, the plant complex type GN1-FNGs, has been proposed on the basis of their chemical structure (Figure 4), the substrate specificity of plant ENGase, and the N-glycan processing mechanism in the ER and Golgi apparatus (Maeda et al., 2010). Prior to discussion of the production mechanism and physiological functions of plant complex type FNGs, we would like to describe some properties of plant ENGase and PNGase, including localization, substrate specificity, and gene structure, in the following section.

In the 1990s, auxin-like activity of plant FNGs to stimulate elongation of stems in seedlings or maturation of tomato fruit was proposed (Priem et al., 1990a; Priem and Gross, 1992; Yunovitz and Gross, 1994). As part of the study to elucidate the physiological function of FNGs involved in fruit maturation process, Nakamura et al. analyzed changes in the amount of FNGs, ENGase activities, and ENGase gene expression during tomato fruit maturation (Nakamura et al., 2008, 2009). They found that the amount of high-mannose type GN1-FNGs increased significantly with fruit maturation (mature green, breaker, pink, and mature red), but the enzyme activity (as total activity) and gene expression were nearly unchanged. These results suggested that tomato ENGase is constantly expressed during tomato fruit maturation step to produce high-mannose type GN1-FNGs. However, the amount of substrates (GN2-FNGs or misfolded glycoproteins) may increase or α-mannosidase activity (Hossain et al., 2010a; Meli et al., 2010) responsible for degradation of the GN1-FNGs may decrease. To examine the putative auxin-like activity to stimulate plant development, the transgenic A. thaliana plants, in which two ENGase genes were knocked out, were constructed (Kimura et al., 2011; Fischl et al., 2011). The high-mannose type GN1-FNGs found predominately in the wild plants were completely converted to GN2-FNGs in the double knockout plants and the ENGase activity was completely lost in the mutant plants, clearly indicating that plant cPNGase, as well as the animal cPNGase, is involved in the production of FNGs prior to the action of ENGase. Two single knockout (At3g11040 or At5g05460) plants produced the high-mannose type GN1-FNGs, and the ENGase activities found in two single knockout plants and wild plants were comparable to each other, but the structural features of FNGs found in these single knockout mutants were slightly different between each other, suggesting that these two ENGases (At3g11040 or At5g05460) may reside in the cytosol of different tissues (roots, stems, seeds, or leaves). No apparent morphological changes were observed among the wild-type plants, two single knockout lines, and double knockout plants under the normal cultivation conditions (Fischl et al., 2011; Kimura et al., 2011). The construction of double knockout plants, in which both ENGase and cPNGase genes are completely suppressed, is the prerequisite to elucidate physiological function(s) of FNGs involved in plant differentiation, growth, and fruit maturation.

As described previously, the plant complex type FNGs bearing one GlcNAc residue (GN1-FNGs) at their reducing ends were first found as secreted FNGs in the rice-cell culture broth. Since these FNGs were not the truncated type and some of them carried the Le^a epitope, these FNGs must have been fully processed in the Golgi apparatus and secreted to extracellular space (but not sorted to the vacuole) through the trans Golgi network. If special ENGase that are active against plant complex type N-glycans would occur in the plant extracellular space, the occurrence of such kind of FNGs (plant complex type GN1-FNGs) could be reasonably explained. However, such special ENGase activity has not been found in plants to date. Considering the structural features of the unique GN1-FNGs, subcellular localization of ENGase and PNGase, and N-glycan processing pathway, a putative mechanism of the unique GN1-FNGs has been proposed (Figure 4; Maeda et al., 2010). In this scheme, it has been assumed that (1) the high-mannose type GN1-FNGs may be first formed by a combination of cPNGase and ENGase in the cytosol, (2) some part of the resulting GN1-FNGs may be transported from the cytosol to the ER through a putative translocator (similar to TAP, one of ABC transporters involved in the cellular immune system) and transported to the Golgi apparatus, and then (3) the high-mannose type N-glycans may be processed to the longer plant complex type structures and secreted to the extracellular space together with fully folded and processed proteins.

It is noteworthy to consider another interesting function of N-glycan involved in protein folding or reconstruction of oligomeric proteins. In a previous study (Kimura et al., 1999), it was reported that high-mannose type N-glycans (Man₉GlcNAc₂ or Glc1Man₉GlcNAc₂) linked to the jack bean α-mannosidase stimulated or induced the reconstruction from denatured and dissociated structure to functional oligomeric structure to recover the enzyme activity. Furthermore, Kimura et al. (1998) and Jitsuhara et al. (2002) reported that glycosylated Asn (Asn-glycan) functioned as a kind of molecular chaperon to reconstruct the functional conformation of some enzymes from denatured state, suggesting that at least the high-mannose type FNGs can play a critical role in the protein-refolding. Since it is well known that sugar molecules or polyethylene glycol stabilize protein structure and prevent protein aggregations, it is not a total surprise that the FNGs produced from glycoproteins show similar function. On the other hand, in the context of generation and degradation mechanisms of FNGs accepted to date, it is difficult to assume that unfolded or prefolded proteins in the ER encounter the high-mannose type FNGs that are formed in the cytosol. However, if there is a possibility that the high-mannose type FNGs produced in the cytosol can be transported to the ER, as postulated above, the unfolded or premature proteins, both glycoproteins and non-glycoproteins, in the ER may be able to receive the sweet benefit from the free sugar chains holding the latent chaperon-like function. To prove the hypothetical chaperon-like function of FNGs, it must be essential to confirm the occurrence of FNGs in the ER and identify the putative transporter responsible for the translocation of FNGs. Furthermore, on the platform of the biological activity of FNGs involved in the protein folding, it will be possible to develop a new glyco-technology in which FNGs are used to promote refolding of recombinant proteins in vitro.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.