| Glycobiology | Pages |
Phosphoglycosylation: a new structural class of glycosylation?
Protein glycosylation
N- and O-linked glycosylation: carbohydrate-peptide linkages
Phosphoglycosylation
Outlook for the future
Acknowledgments
References
Phosphoglycosylation: a new structural class of glycosylation?
There are a number of different glycoproteins that have been identified relatively recently which contain oligosaccharides linked to serine or threonine in a peptide backbone via phosphodiesters. It is possible that these glycoproteins may form an alternative structural class of glycosylation. This modification has been referred to as phosphoglycosylation (Mehta et al., 1996; J. Biol. Chem., 271, 10897-10903), and has been reported in slime molds and several unicellular parasites. In this review, examples of phosphoglycosylation from different biological sources are discussed. Those which are well characterized have been found to be highly variable with respect to the glycan moiety, while sharing some common features. An experimental approach detailing how to determine whether a protein is phosphoglycosylated is also presented.
Protein glycosylation
Glycosylation is the most common covalent modification of newly synthesized proteins, and also the most diverse. This diversity is reflected in the range of functions which have been assigned to glycoproteins. A small sample of these includes roles in adhesion, cellular trafficking, decoys against bacterial and parasitic invasion, wound healing, inflammation, cancer metastasis and molecular mimicry (Springer, 1990; Lasky, 1992; Ratner, 1992). Much of the evidence for the participation of glycoconjugates in physiological function has been based on indirect methods that rely on glycosylation inhibitors, deglycosylation enzymes, and site-directed mutagenesis of proteins. Advances in analytical methodology in recent years, however, particularly in the field of mass spectrometry, have allowed the detailed structural analysis of oligosaccharide structures at ever-increasing levels of sensitivity.
These improvements in methodology, along with the increased awareness of the biological significance of many glycoproteins, have led to a significant increase in the number of known oligosaccharide structures, sugar constituents, and carbohydrate-peptide linkages of glycoproteins. A myriad of novel glycoprotein structures have been characterized, including those from newly purified molecules and also many from previously characterized glycoproteins where increased sensitivity has led to new findings (Cumming et al., 1989; Varki, 1996). Many of these novel glycan structures have thus far only been identified on a single molecule from a single organism. In the last 10 years two new classes of glycoproteins have been discovered, both of which have subsequently been found to be widely distributed in nature. These glycoproteins are formed by the addition of O-GlcNAc to nuclear and cytoplasmic proteins (Hart et al., 1989) and the addition of glycosylphosphatidylinositol (GPI) anchors, containing a conserved trimannosyl core, to the carboxyl terminus of membrane proteins (Ferguson and Williams, 1988).
The subject of this review is another group of related glycoprotein structures which have recently been identified from several different organisms. The glycoproteins in question are those where oligosaccharides are attached directly to the polypeptide chain via a phosphodiester linkage, a modification that has been referred to by Freeze and co-workers as phosphoglycosylation (Mehta et al., 1996). While there is no evidence as yet to suggest that they are widespread or abundant in nature, it seems likely that many more examples exist than the small number that have been identified so far.
N- and O-linked glycosylation: carbohydrate-peptide linkages
The vast majority of glycoproteins containing N-linked oligosaccharides conform to the classical motif of a reducing terminal GlcNAc linked to asparagine as part of the consensus sequence Asn-Xaa-Ser/Thr (where Xaa is any amino acid except proline; Marshall, 1974). The only reported exception to this consensus sequence is nephritogenoside, where the amino terminal sequence Asn-Pro-Leu is modified by a trisaccharide consisting of three glucose residues, with [alpha]-glucose as the linkage unit (Shibata et al., 1988). The carbohydrate moiety is more variable, as, for example, several asparagine-linked monosaccharides have been reported in bacterial glycoproteins, including [alpha]- and [beta]- glucose, [beta]-N-acetylgalactosamine, and rhamnose (Lis and Sharon, 1993).
O-linked glycosylation is far more promiscuous in terms of the carbohydrate-peptide linkage. The most abundant is the mucin-type linkage between N-acetyl-galactosamine and serine or threonine (Carraway and Hull, 1991). Other amino acids which have been reported to be modified by O-linked glycosylation include hydroxyproline linked to galactose in some collagens, and arabinose in plant cell walls (Kieliszewski et al., 1995), hydroxylysine linked to arabinose in plant extensins, and tyrosine linked to [alpha]-glucose in glycogenin (Smythe et al., 1988) and [beta]-glucose in the S layer glycoprotein of Clostridium thermohydrosulfuricum (Christian et al., 1993). Examples of other saccharides found linked to serine or threonine include [alpha]-galactose in potato lectin and some collagens (Lis and Sharon, 1993), mannose in yeast mannoproteins, glucose in bovine and human blood coagulation factors (Nishimura et al., 1989), xylose in maize root-cap cells (Green and Northcote, 1978), and fucose in tissue plasminogen activator (Harris et al., 1991).
A separate class of O-linked glycosylation, which appears to be functionally distinct, is the O-GlcNAc modification found as monosaccharides attached to either serine or threonine. This modification is unusual when compared to most O-glycosylation, since it is a simple monomeric structure and is also highly dynamic and responsive to cellular stimuli in a manner somewhat analogous to phosphorylation (Haltiwanger et al., 1992). A series of O-linked N-acetylglucosamine bound oligosaccharides have also been isolated from Trypanosoma cruzi (Previato et al., 1994), but this seems to be a single example which differs from the majority. Since it was first reported (Torres and Hart, 1984), O-GlcNAc modification has been found on proteins that reside almost exclusively in the nucleus and cytoplasm of cells of eukaryotes ranging from trypanosomes to humans (Hart et al., 1995). It seems certain that many more examples of this modification will be characterized in the future as analytical methods become increasingly sensitive and functional studies further elucidate its physiological role.
Another form of glycosylation which has been found to be ubiquitous in eukaryotes since it was first discovered (Ferguson et al., 1985) is the addition of GPI anchors at the carboxyl terminus of proteins (Ferguson, 1992). The inositol phospholipid portion of the GPI moiety is variable in structure, but the glycan core is conserved in all known examples, consisting of Man([alpha]1-2)Man([alpha]1-6)Man([alpha]1-4)GlcN. The glycan core may be substituted with either additional ethanolamine phosphate residues, additional mono- or oligosaccharides, or both. A number of functions have been described for GPI anchors of specific proteins in different systems, including transmembrane signaling, targeting to the apical membrane in polarized epithelial cells, and folate uptake (Ferguson, 1991). There is little evidence to suggest any more general physiological function, although a study of the major cell surface proteins of Trypanosoma brucei revealed the intriguing fact that the structure of the GPI group was modified in a developmentally regulated fashion during the life cycle (Field et al., 1992).
The previous two types of glycosylation discussed have undergone intensive study since their initial discovery. It is clear now that they are both far more widespread than was originally envisaged, and it is possible that the same may be true of phosphoglycosylation in the future.
Phosphoglycosylation
Dictyostelium discoideum cysteine proteinases
The first reported example of a protein modified by phosphoglycosylation was an endopeptidase known as Proteinase I, isolated from the cellular slime mold Dictyostelium discoideum. This was shown to contain GlcNAc-1-PO4 linked to serine (Figure 1; Gustafson and Milner, 1980). The linkage was demonstrated by the recovery of both GlcNAc-1-PO4 after alkaline hydrolysis and O-phosphorylserine after acid hydrolysis. The same enzyme was examined further in a more recent study, and the original findings regarding the structure of the protein modification were confirmed (Mehta et al., 1996). These authors also analyzed the products released by alkaline [beta]-elimination using mass spectrometry in order to show that only single GlcNAc residues, and no larger groups, were linked to the protein via phosphodiesters.
Figure
Two other cysteine proteinases from D.discoideum, known as cprD and cprE, have also been shown to carry GlcNAc-1-PO4 phosphoglycosylation (Souza et al., 1995). These findings were based on cross-reactivity with a monoclonal antibody, raised against Proteinase I, that was shown by competitive inhibition to be highly specific for the GlcNAc-1-PO4 group. Recent work has shown that these proteinases are members of a family of such enzymes present in D.discoideum, all of which appear to carry the same modification, which has now been extended to include cprF and cprG (Ord et al., 1996).
Phosphoglycoproteins of Leishmania species
The insect and mammalian life cycle stages of several species of Leishmania are known to secrete a number of phosphoglycan containing components (reviewed in Ilg et al., 1994b). These include lipophosphoglycan (LPG), extracellular hydrophilic phosphoglycan (Greis et al., 1992), and a filamentous polymeric protein complex. In Leishmania mexicana promastigotes, for example, this complex consists of a 100 kDa phosphoglycoprotein known as secreted acid phosphatase (sAP) which is associated with one or more polydisperse high molecular weight proteophosphoglycans (Ilg et al., 1994a).
Phosphoglycosylation attached to serine was identified and characterized in the sAP of L.mexicana (Ilg et al., 1994a). The linkage unit is Man[alpha]-1-PO4-serine, and the carbohydrate attached consists of monomeric mannose and a series of neutral and phosphorylated glycans similar to those found in the Leishmania LPGs (McConville et al., 1990). The neutral species consisted of a linear series of oligosaccharides with the structure (Man([alpha]1-2))0-5-Man. The phosphorylated oligosaccharides consisted of (PO4-6-Gal([beta]1-4)Man) and (PO4-6[Glc([beta]1-3)] Gal([beta]1-4)Man), assembled in short phosphoglycan chains with an average of two repeat units per chain, and capped with neutral mannose oligomers (Figure 1). The Man[alpha]-1-PO4-serine linkage unit was indicated by detection of phosphoserine residues in amino acid analysis, and the finding that the phosphate was not susceptible to removal by alkaline phosphatase in the absence of prior treatment with either jack bean [alpha]-mannosidase (to release neutral mannose glycans) or mild acid treatment (to release all glycans).
Despite the fact that the high molecular weight proteophosphoglycan was not completely resolved from the 100 kDa phosphoglycoprotein in this study (Ilg et al., 1994a), it was also shown to contain the same modifications, although probably enriched in the neutral mannose oligosaccharides. Analysis of the homologous molecule from a related species, the mucin-like proteophosphoglycan secreted by L.major promastigotes, revealed that it too contained phosphoglycosylation attached to serine (Ilg et al., 1996). The carbohydrates attached were similar to those previously described in the L.mexicana sAP, including a Man[alpha]-1-PO4-serine linkage unit, but contained additional substituents including arabinose. The glycan structures were reported as: Gal([beta]1-4)Man, Man([alpha]1-2)Man, Gal([beta]1-3)Gal([beta]1-4)Man, PO4-6(Gal([beta]1-3))0-2Gal([beta]1-4)Man, and PO4-6(Ara([beta]1-2)Gal ([beta]1-3))Gal([beta]1-4)Man. These were arranged in phosphoglycan chains with an average of three repeat units per chain, although the structure of individual glycan chains was not able to be determined. The linkage unit was determined by both phosphoamino acid analysis and 31P nuclear magnetic resonance spectroscopy, which demonstrated that phosphate was present exclusively as diesters.
A large panel of monoclonal antibodies of varying specificities (Ilg et al., 1993) has been used to demonstrate the presence of similar phosphoglycosylation in a number of related molecules, such as the proteophosphoglycan of L.mexicana amastigotes(Ilg et al., 1995), and the secreted acid phosphatases of L.donovani, L.amazonensis, L.aethiopica, and L.tropica (Ilg et al., 1994b). It is clear that phosphoglycosylation is a significant form of posttranslational modification in Leishmania, and it is fortunate that sufficient quantities of these proteins are synthesized to permit their detailed analysis. It seems certain that the structural characterization of more phosphoglycoproteins from Leishmania species will be reported in the future.
Trypanosoma cruzi phosphoglycoproteins
Glycoproteins from the kinetoplastid parasite T.cruzi, which react with the carbohydrate-specific monoclonal antibody WIC29.26 (Ferguson et al., 1983), are also phosphoglycosylated proteins. The WIC29.26-reactive glycopeptides purified from epimastigotes contained xylose, rhamnose, fucose, and galactofuranose (Haynes et al., 1996). These were found to be attached to protein via a mild acid labile linkage, suggesting the presence of phosphoglycosylation. This was confirmed by the identification of phosphothreonine, and a low level of phosphoserine, in the glycopeptides by 32P labeling of cells and thin layer electrophoresis of phosphoamino acids. The linkage unit is Xyl-1-PO4-threonine or serine.
Several of the phosphoglycans liberated by mild acid hydrolysis were characterized by electrospray mass spectrometry in conjunction with different derivatization techniques. The structures of two principal species were elucidated, and included an unusual trisubstituted xylose and a galactofuranose containing a cyclic-phosphate group (Figure 1). The presence of a cyclic-phosphate group was most likely an artifact of the mild acid hydrolysis procedure, rather than an intrinsic part of the native glycan as has been found in a polysaccharide of Vibrio cholerae (Knirel et al., 1995). The oligosaccharides also appeared to be arranged in phosphoglycan chains, with an average chain length of five repeat units per chain, capped with either xylose and galactofuranose monosaccharides or nonphosphorylated versions of the same oligosaccharides. The glycan structures determined were cyclic-PO4-Galf-(Galf)((±Galf),Rha,Fuc)-Xyl-Xyl.
A homolog of the major glycoprotein that reacts with the WIC29.26 antibody, GP72, was identified in a related parasite, T.brucei. (Nozaki et al., 1996). This glycoprotein, designated Fla1, was shown to have a high degree of similarity to GP72, including oligosaccharides that were sensitive to mild acid hydrolysis and thus likely to be phosphodiester linked. Fla1 did not, however, cross-react with the T.cruzi specific WIC29.26 antibody, indicating that the carbohydrate structures may be significantly different between the two proteins.
Biosynthetic enzymes involved in phosphoglycosylation
In order to elucidate the physiological function of the phosphoglycoproteins previously discussed, it is necessary to isolate and characterize the biosynthetic enzymes involved in phosphoglycosylation pathways, particularly the initial steps. A UDP-GlcNAc:Ser-protein N-acetylglucosamine-1-phosphotransferase activity was first identified in both Acanthomeba castellani and D.discoideum (Lang et al., 1986), and subsequently purified from cellular membranes of D.discoideum (Merello et al., 1995). The enzymic activity was found to be distinct from UDP-GlcNAc:glycoprotein N-acetylglucosamine-1-phosphotransferase, which phosphorylates protein-linked high mannose type oligosaccharides. Ongoing work has made substantial progress in the isolation of a glycan phosphotransferase from L.mexicana involved in the assembly of Man[alpha]-1-PO4-serine. This enzymic activity is apparently distinct from the glycan phosphotransferase involved in the assembly of the polymeric phosphodiester linked glycans of the L.mexicana lipophosphoglycan (M. A. J. Ferguson, personal communication).
The differences in intracellular location between the phosphoglycosylated proteins that have been characterized thus far suggest that their biosynthesis may also differ significantly. The cysteine proteinases of D.discoideum are found in lysosomes, the WIC29.26 reactive glycoproteins of T.cruzi are localized at the cell surface, and the acid phosphatases and proteophosphoglycans of Leishmania spp. are secreted. Since many of the phosphoglycan containing products in Leishmania are secreted (Ilg et al., 1994b), it is tempting to speculate that phosphoglycosylation, at least in Leishmania, may act as a secretory signaling modification. However, it is clear that this is not the case in other systems.
Outlook for the future
Most of the phosphoglycosylated proteins characterized thus far have been identified by monoclonal antibodies that recognize all or part of the oligosaccharides. Since these glycan chains vary widely in composition and structure, there is still no generally applicable technique that can be used to screen cellular preparations for the presence of phosphoglycoproteins. The basic requirements to demonstrate the presence of a phosphodiester linkage between carbohydrate and protein are threefold, as shown in Figure 2. Firstly, oligosaccharides must not be liberated from the glycoprotein by treatments appropriate for N-linked or mucin-type O-linked structures, such as peptide N-glycosidase F digestion (Plummer et al., 1984) or mild alkaline [beta]-elimination and reduction (Carlson, 1968). Phosphodiester linked monosaccharides have been released from glycoproteins under alkaline conditions (Gustafson and Milner, 1980; Mehta et al., 1996), while, in contrast, phosphodiester linked oligosaccharides have been found to be resistant to liberation under similar conditions (Haynes et al., 1996). Secondly, oligosaccharides should be liberated from the glycoprotein by mild acid hydrolysis, which preferentially cleaves hexose-1-phosphate bonds (McConville et al., 1990). Thirdly, the residual phosphate groups which remain associated with the protein following mild acid hydrolysis should be found in phosphoamino acids. The latter procedure is necessary to distinguish phosphoglycoproteins from glycoproteins that contain an N- or O-linked oligosaccharide which includes a phosphodiester linkage within the glycan chain (Varki and Kornfeld, 1980; Hayes and Varki, 1993).
Figure
It is only when all three of these criteria are satisfied that a modified protein can be considered an example of phosphoglycosylation as described in this review. There are a number of modified proteins in literature for which there is some evidence that they may contain phosphoglycosylation, but insufficient data to be conclusive. An example of this is the partially characterized O-glycosylation of D.discoideum prespore specific antigen (Haynes et al., 1993). This glycoprotein was found to be extensively modified by glycosylation that was insensitive to peptide N-glycosidase F or mild base catalyzed [beta]-elimination, while 3H labeled glycans were produced by aqueous hydrofluoric acid treatment followed by reduction in tritiated sodium borohydride (Haynes, 1992). These data suggest the presence of a phosphodiester group, but give no information concerning the attachment of glycans to amino acids. There is as yet no single function assigned specifically to phosphoglycosylation modifications. This is to be expected, since the carbohydrate groups involved vary widely in size and structure. The purpose of this review is to present phosphoglycosylation as a distinct structural, rather than functional, classification. One feature common to all such phosphoglycoproteins characterized is a high level of immunogenicity. This is particularly evident for both the WIC29.26 glycans, where the phosphoglycosylation appears to be part of an immunodominant carbohydrate epitope (Cooper et al., 1993), and the phosphoglycans found on a variety of substrates in Leishmania, which have been used to generate a panel of monoclonal antibodies of exquisite specificity (Ilg et al., 1993). The severity of the immune response indicated by these data suggest that phosphoglycosylation is unlikely to occur in mammalian cells. It is hoped that bringing this class of glycoproteins to the attention of a wider audience will stimulate interest in the field and increase awareness of an alternative to the more common oligosaccharide-protein linkages. This, in turn, may help in solving the primary structure of some modified proteins known to contain unspecified yet unusual carbohydrate modifications. Any researcher confronted by a glycoprotein, especially one isolated from a lower eukaryote, that is intractable to carbohydrate analysis using conventional methodology should certainly consider the possibility of phosphoglycosylation.
Acknowledgments
I wish to acknowledge funding support from the National Science Foundation as part of the Science and Technology Center for Molecular Biotechnology. I thank Ruedi Aebersold for providing such a stimulating laboratory to work in, Mike Ferguson for encouragement and critical discussions, George Cross for his attention to detail, and Julian Watts for editorial assistance.
References
This page is run by Oxford University Press, Great Clarendon Street, Oxford OX2 6DP, as part of the OUP Journals
Comments and feedback: www-admin{at}oup.co.uk
Last modification: 17 Jan 1998
Copyright© Oxford University Press, 1998.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  T. Buskas, S. Ingale, and G.-J. Boons Glycopeptides as versatile tools for glycobiology Glycobiology, August 1, 2006; 16(8): 113R - 136R. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 L. M. Thomson, D. J. Lamont, A. Mehlert, J. D. Barry, and M. A. J. Ferguson Partial Structure of Glutamic Acid and Alanine-rich Protein, a Major Surface Glycoprotein of the Insect Stages of Trypanosoma congolense J. Biol. Chem., December 6, 2002; 277(50): 48899 - 48904. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. J. Harrison, I. M. Wathugala, M. Tenkanen, N. H. Packer, and K.M. H. Nevalainen Glycosylation of acetylxylan esterase from Trichoderma reesei Glycobiology, April 1, 2002; 12(4): 291 - 298. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
R. G. Spiro Protein glycosylation: nature, distribution, enzymatic formation, and disease implications of glycopeptide bonds Glycobiology, April 1, 2002; 12(4): 43R - 56R. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. J. McConville, K. A. Mullin, S. C. Ilgoutz, and R. D. Teasdale Secretory Pathway of Trypanosomatid Parasites Microbiol. Mol. Biol. Rev., March 1, 2002; 66(1): 122 - 154. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||