Dr Tamao Endo PhD, Molecular Glycobiology, Research Team for Mechanism of Aging, Tokyo Metropolitan Institute of Gerontology, 35-2 Sakaecho, Itabashi-ku, Tokyo 173-0015, Japan. Email: email@example.com
Glycosylation is the most common post-translational modification of proteins. Protein sequence data suggested that more than half of all proteins produced in mammalian cells are glycoproteins. Recent studies showed that glycans of secreted glycoproteins affect many protein properties, such as solubility, stability, protease sensitivity and polarity, whereas glycans on cell-surface glycoproteins are involved in various cellular functions, including cell–cell communication. Accordingly, the investigation of glycoprotein changes caused by aging is expected to help understand the aging process and to elucidate age-associated diseases. The present review will summarize our current knowledge of changes found in brain glycoproteins resulting from the aging process. Geriatr Gerontol Int 2010; 10 (Suppl. 1): S32–S40.
Recent advances in glycobiology have shown the importance of glycans as biosignals for multicellular organisms, including cell–cell communication. Unlike proteins and nucleic acids, which are linear molecules, glycans are branched. Another way in which glycans differ from nucleic acids and proteins is that they have a number of positional isomers and anomeric configurations, which allow them to form many possible structures.
The biosynthesis of glycans is not achieved by the intervention of a template. but is under the control of the expression of glycosyltransferases. There is growing evidence that these enzymes have a variety of roles in cellular differentiation and development, as well as in disease processes. Mouse knockout studies have shown that some glycosyltransferases are essential for neuronal development, and their defects lead to neurological abnormalities.1 The importance of glycans to the nervous system is further highlighted by congenital disorders of glycosylation (CDG, which are caused by defects in N-glycans) that result in hypotonia, psychomotor retardation and other neuropathological symptoms.2 Furthermore, recent findings that aberrant O-mannosylation is the primary cause of some forms of congenital muscular dystrophy and neuronal migration disorder3 indicate that glycans that are not part of the N-linked pathway in the nervous system are also important.
Glycoproteins are found at the cell surface and in extracellular compartments and play pivotal roles in nervous system development, the modification of synaptic activity and the regeneration of nerve connections after damage in adults.4 In the present review, we discuss the importance of protein glycosylation during aging.
Proteins produced by eukaryotic cells are frequently post-translationally modified by the addition of glycans (Fig. 1). The protein sequence data suggested that more than half of all proteins are glycoproteins.5 The glycan moieties of these glycoproteins play important roles not only in modulating properties, such as protein stability and conformation, but also as key elements in various molecular recognition processes, such as bacterial and viral infection, cell adhesion in inflammation and metastasis, differentiation, development, and many other events characterized by intercellular communication. Despite the importance of glycan recognition, the precise mechanisms underlying the many glycan-mediated recognition processes are not well understood.
The major glycans of glycoproteins can be classified into two groups according to their glycan-peptide linkage regions (Fig. 2). Those that are linked to asparagine (Asn) residues of polypeptides are termed N-glycans, whereas those that are linked to serine (Ser) or threonine (Thr) residues are called O-glycans. In N-glycans, the reducing terminal N-acetylglucosamine (GlcNAc) is linked to the amide group of Asn, through an aspartylglycosylamine linkage. In O-glycans, the reducing terminal N-acetylgalactosamine (GalNAc) is attached to the hydroxyl groups of Ser and Thr residues of polypeptides; however, in addition to the abundant O-GalNAc forms, several unique types of protein O-glycosylation have recently been reported, such as O-linked fucose,6–8O-linked glucose,9O-linked GlcNAc10 and O-linked mannose.3,11O-Glycans are formed by the stepwise addition of monosaccharides to the Ser and Thr residues of polypeptides from nucleotide sugars. In contrast, N-glycans are formed by a series of complex pathways, including lipid-linked intermediates. First, GlcNAc-1-P is transferred from uridine diphosphate (UDP)-GlcNAc to a polyisoprenol monophosphate, dolichyl phosphate (Dol-P). The GlcNAc residue of GlcNAc-PP-Dol is the starting point of N-glycans. To this GlcNAc residue, another GlcNAc and five mannose residues are transferred from UDP-GlcNAc and GDP-Man, respectively. Lipid-bound heptasaccharide is converted to Glc3Man9GlcNAc2-PP-Dol by the further addition of four mannose residues from Dol-P-Man and three glucose residues from Dol-P-Glc. The tetradecasaccharide of the lipid derivative is then transferred en bloc to the Asn residue of the polypeptide chain, which is translated in the rough endoplasmic reticulum by the catalytic action of a Dol-P-oligosaccharide, polypeptide oligosaccharyltransferase. Only the Asn residue in the sequence of Asn-X-Ser/Thr, where X can be any amino acid other than proline, is glycosylated. The completely translated protein with tetradecasaccharide is then transported to the Golgi apparatus. After the three glucose residues and up to six mannose residues are removed, a set of glycosyltransferases work to create multiple structures. The order of monosaccharide addition and deletion is prescribed, but not template driven and several enzymes can compete for the same structure, which yields a range of N-glycan structures.
There is growing evidence that glycosyltransferases are involved in cellular differentiation and development, and disease processes.1 The importance of glycans is further highlighted by CDG. CDG were initially identified in 1980; since then, 20 distinct CDG have been identified.2 Each is autosomal recessive and caused by mutations in different genes involved in N-glycosylation. Many more types of CDG will probably be found because approximately 50 genes are required for N-glycan synthesis. The CDG are a group of inherited multisystemic disorders, which are commonly associated with severe psychomotor disturbance and mental retardation. CDG type I is caused by defects of the assembly of lipid-linked oligosaccharides, whereas CDG type II is caused by the incomplete processing of protein-bound N-glycans. The molecular nature of 12 CDG-I types and eight CDG-II types has been identified.2 CDG studies show that correct N-glycosylation of proteins is essential for normal development, including the brain.
Glycans other than those of the N-linked pathway are also important. This is shown by the finding that aberrant O-mannosylation is the primary cause of some forms of congenital muscular dystrophy with abnormal neuronal migration. Mammalian O-mannosylation is present in a limited number of glycoproteins of the brain, nerves and skeletal muscle.11 We previously found that glycans of α-dystroglycan include a sialyl O-mannosyl glycan, Siaα2-3Galβ1-4GlcNAcβ1-2Man.12 We identified and characterized the enzymes involved in the biosynthesis of O-mannosyl glycans. We first showed that the enzyme that forms the GlcNAcβ1-2Man linkage is a glycosyltransferase, UDP-GlcNAc, protein O-mannose β1,2-N-acetylglucosaminyltransferase (POMGnT1).13,14 We then identified the human enzymes that initiate protein O-mannosylation, protein O-mannosyltransferase 1 and 2 (POMT1 and POMT2), and showed that they have protein O-mannosyltransferase activity only when they are co-expressed.15,16
Muscle–eye–brain disease (MEB; OMIM 253280) is an autosomal recessive disorder characterized by congenital muscular dystrophy, ocular abnormalities and brain malformation.17 Walker–Warburg syndrome (WWS; OMIM 236670) is another form of congenital muscular dystrophy that is characterized by severe brain malformation and eye anomalies.18 Both are caused by defects of protein O-mannosylation. The POMGnT1 gene is responsible for MEB14 and POMT1 and POMT2 genes are responsible for WWS.19–21 Additionally, selective deficiency of glycosylated α-dystroglycan in MEB and WWS patients was found. These findings suggest that defective protein O-mannosylation of α-dystroglycan is a common trait of muscle cell degeneration and abnormal brain structure.
After reporting that MEB and WWS are caused by O-mannosylation defects, some muscular dystrophies were suggested to be caused by abnormal glycosylation of α-dystroglycan;22,23 however, it is still unclear whether these muscular dystrophies are a result of O-mannosylation defects. Future studies might show that presently uncharacterized forms of muscular dystrophy are also caused by defects in glycosyltransferases.
Gene-knockout studies have shown the absolute requirement of protein N-glycosylation and O-mannosylation for neural tissue development in early embryogenesis.1,24–26
Glycan-binding proteins in the brain
Although animals have a large variety of glycan-binding proteins (lectins), only a few have been isolated from the brain and characterized with respect to their roles. They include β-galactose binding lectin (galectin), mannose-binding lectin and Siglec (sialic acid-binding immunoglobulin superfamily lectin).
Immunohistochemical studies have shown that the expressions and localizations of galectins are regulated developmentally in the brain.27 Previous findings have suggested that galectins play important roles during brain development and are involved in axonal tracing and neurite fasciculation.28–30 These observations have been supported by a more recent study using galectin-1 knockout mice. The phenotypes observed in galectin-1 knockout mice included a failure of neurite outgrowth and, thus, a failure of the neurite processes to target olfactory neurons,31 indicating that galectin-1 is functionally involved in promoting olfactory axon fasciculation, that is, in cross-linking adjunct axons and promoting axonal adhesion to the extracellular matrix. Recently, it was reported that galectin-1 is involved in the proliferation of adult neural stem cells.32
Two mannose-binding lectins with different physicochemical properties, CSL33 and R1,34 have been isolated from the cerebellum. The glycan-binding specificities of these two lectins are similar but not identical. CSL is a soluble protein with a preference for Man6-GlcNAc2-Asn. It is involved in adhesion processes, such as stabilization of the myelin structure, formation of contact between axons and myelinating cells, and contact guidance of neuron migration during development. It is also thought to have a role in multiple sclerosis.35 R1 is a membrane-bound protein with a preference for Man8-GlcNAc2-Asn and Man5-GlcNAc2-Asn, and is involved in neuronal recognition as a first step in synaptogenesis.
The Siglecs are single-pass type I transmembrane proteins.36 All contain an N-terminal V-set Ig domain with a sialic acid binding site. Eleven different Siglecs have been characterized in humans. Because all Siglecs, except Siglec-4, which is exclusively found in the nervous system, are expressed in hematopoietic cells, only Siglec-4 will be described here. Myelin-associated glycoprotein (MAG)/Siglec-4a is a glycoprotein expressed on oligodendrocytes in the central nervous system and on Schwann cells in the peripheral nervous system. Both cell types are involved in forming myelin sheaths around neural fibers. The minimal structural requirement for MAG recognition is Neu5Acα2-3Galβ1-3GalNAc, although MAG also recognizes Neu5Acα2-3Galβ1-4GlcNAc.37,38 MAG-deficient mice showed rather subtle phenotypes.39 The formation of morphologically intact myelin sheaths in the central nervous system is affected, and to a minor extent, the integrity and maintenance of myelin. In the peripheral nervous system, in comparison, only the maintenance of myelin is impaired. Thus, other molecules might compensate for MAG in MAG-knockout mice. Independent support for the role of MAG in myelin stability comes from studies of mice deficient in the synthesis of gangliosides, which are the best ligands for MAG.40 The pathological features of their nervous systems resemble those in MAG-deficient mice. Interestingly, such mice showed an age-dependent progressive decline in MAG expression, suggesting that maintenance of MAG protein levels depends on gangliosides.41 SMP/Siglec-4b is an avian Siglec expressed on Schwann cells and oligodendrocytes, and shows near-identical binding specificity towards gangliosides as MAG, suggesting that SMP/Siglec-4b is a functional equivalent of MAG in the avian nervous system.42
The neural cell adhesion molecule (NCAM) shows both homophilic and heterophilic interactions. NCAM binds to high-mannose type N-glycans of the cell adhesion molecule, L1.43,44 This appears to cause the phosphorylation of L1 and hippocampal long-term potentiation (LTP).45 In contrast, the interaction between L1 and CD24 is mediated by O-glycans on CD24, and is dependent on α2-3-linked sialic acid.46 This interaction might modulate the signal induced by L1. NCAM also binds to complex-type N-glycans of phosphacan/protein tyrosine phosphatase ζ/β47 to heparin/heparan sulfate48,49 and to chondroitin sulfate.50,51 Further studies are needed to examine the glycan-binding activity of NCAM.
NCAM has important roles in cellular communication and neurogenesis.52 NCAM-null mice showed a smaller olfactory bulb, reduced LTP in the hippocampal CA3 region and reduced spatial memory formation. Some diversity in NCAM is known to occur from post-translational modifications and differences in the glycosylation pattern of NCAM could generate distinct isoforms.53,54 Although the structure of NCAM glycans has not been analyzed in detail, two novel glycans (polysialic acid and HNK-1) of NCAM have been determined to be functionally important.
Polysialic acid (homopolymer of Siaα2-8 linkage) is a unique glycan moiety on NCAM and is known to modify the affinity of NCAM.55 It has been shown that polysialic acid chain lengths markedly decrease in the adult form of NCAM. Polysialic acid on NCAM is regarded as an important regulator that prevents strong binding between NCAM. Polysialic acid is formed by two sialyltransferases, ST8SiaII and ST8SiaIV.56 These two enzymes are expressed differently in the brain. Mice deficient in ST8SiaII and ST8SiaIV have shown the roles of polysialic acid in synaptic plasticity. Polysialic acid formed by ST8SiaIV is important in long-term depression and LTP in the hippocampal CA1 region but not in CA3 LTP.57 In contrast, the polysialic acid formed by ST8SiaII is not important in hippocampal synaptic plasticity, but its absence caused higher exploratory drive and reduced behavioral responses.58 These results suggest that the polysialic acid formed by ST8SiaII and ST8SiaIV plays different roles in the brain, but each partially compensated for the absence of the other. The absence of polysialic acid by simultaneous ablation of ST8SiaII and ST8SiaIV confers a lethal phenotype with specific defects of major brain tracts.59 In contrast, HNK-1 epitope (3-sulfated glucuronic acid) is also present on NCAM, and is involved in neural development and synaptic plasticity through the mediation of cellular interactions.4 HNK-1 is formed by two glucuronyltransferases (GlcAT-P and GlcAT-S) and one sulfotransferase (HNK-1ST). Studies have been carried out with GlcAT-P- and HNK-1ST-deficient mice, suggesting that the HNK-1 structure plays important roles in synaptic plasticity and spatial memory function.60,61
Different expression of glycoproteins in the brain during aging
As already discussed, the structures of glycans are much less rigidly defined than those of proteins and nucleic acids because the biosynthesis of glycans is not controlled by the intervention of a template. This means that glycans can be easily altered by the physiological conditions of cells. Accordingly, age-related alteration of the glycoproteins could be an important element in solving the various pathological problems found in elderly individuals.
Although many studies testify to the importance of the structural changes of glycans of glycoproteins during development, limited data are available concerning the alteration of glycans during aging. The first reliable report was obtained by the studies of glycans of human serum immunoglobulin G (IgG). The decrease of galactose content of human serum IgG with age was reported.62,63 Because nongalactosyl human IgG binds less effectively to C1q and Fc-receptors,64 this alteration will partly explain the phenomenon of immunodeficiency observed in aged individuals.
For the purpose of investigating the alteration of glycoproteins in the brain during aging, we comparatively analyzed glycoproteins in the soluble fraction and the membrane preparations of various portions of the brain, including spinal cords, obtained from 9-week-old rats and 29-month-old rats, by SDS-polyacrylamide gel electrophoresis and lectin staining.65 The glycoprotein patterns of each brain part showed marked differences with the age of donors (Fig. 3). The most prominent evidence in the soluble fractions of white matter, basal ganglia and spinal cord detected by wheat germ agglutinin (WGA) is that glycoproteins with an apparent molecular weight of 123K and 115K were increased in aged rats. In addition, the reactivity of 115K with concanavalin A (Con A) and peanut agglutinin (PNA) was also increased in aged rats. In contrast, reactivity of an apparent molecular weight of 115K with WGA was increased in the membrane fractions of white matter, basal ganglia, hippocampus, cerebellum and spinal cord from aged rats. In contrast, with Maackia amurensis lectin (MAL), which is specific for Siaα2-3Gal linkage, an apparent molecular weight of 115K was detected only in the membrane fraction of the cerebellum and was decreased in aged rats. Reactivity with an apparent molecular weight of 133K and 125K in the membrane fractions of white matter and basal ganglia with Lens culinaris agglutinin (LCA) was decreased in aged rats. In contrast, reactivity of the front band with LCA and Aleuria aurantia lectin (AAL) was increased and that of 130K with AAL was decreased in aged rats, respectively. Taken together, we conclude that the glycosylation state of protein in the brain changes during aging.
The most prominent alteration was detected in the membrane fraction of spinal cords through staining with LCA. A glycoprotein with an apparent molecular weight of 30 kDa (gp30) was detected in aged rats, but not in young adult rats. Based on amino acid sequence data around the glycosylation site, gp30 was identified as P0, which is a member of the immunoglobulin superfamily and a major structural component of peripheral nerve myelin. This is the first report showing that P0, which has been considered a peripheral nerve-specific glycoprotein, also occurs in the spinal cord of mammals.66 In addition, non-glycosyl P0 could be detected in the spinal cord of young adult rats by anti-P0 antibody. These results show that the glycosylation state of P0 in the spinal cord changes during aging. Furthermore, we determined, by immunohistochemical and immunocytochemical analyses, that neurons in the spinal cord expressed P0. Our data also showed that the number of neurons expressing P0 decreased and became smaller with age.67 Thus, the results emphasize the importance of neurons expressing P0 in the spinal cord in the formation and maintenance of the neural network. Previous studies showed that the glycan moiety of P0 plays a very important role in cell–cell adhesion by homophilic binding. It was also found that a non-glycosyl P0, obtained by site-directed mutagenesis, did not show homophilic adhesion;68 therefore, the biological meaning of the occurrence of non-glycosyl P0 in the spinal cord of young adult rats is an interesting target to be elucidated in the future.
Finally, we examined the distribution of sialoglycoconjugates in the brain of 9-week-old rats and 30-month-old rats using light microscopy and electron microscopy in combination with two lectins, MAL, for Siaα2-3Gal, and Sambucus sieboldiana agglutinin (SSA), for Siaα2-6Gal. The results showed that Siaα2-3Gal and Siaα2-6Gal were expressed in distinct regions of the brain and their expression patterns changed in the aged brain.69–71 At least two possibilities can be considered to explain these observations. First, the glycan structure of glycoproteins in young adult rats is different from that in aged rats. The second possibility is that the expression level of glycoprotein molecules themselves changes in aged rats. Furthermore, it is necessary to identity glycoproteins showing different expression patterns during aging in the future.
It is noteworthy that we observed weak but distinct staining of polysialic acid residues in the marginal zone between the granule cell layer and polymorph layer in 29-month-old rat hippocampal formations. Because a good correlation was found between the number of polysialic acid-positive cells and that of newly generated granule cells in the adult rat brain,72 the presence of polysialic acid residues of the brain of even 29-month-old rats suggests that the generation of granule cells occurs at least up to this age. Additionally, polysialylation of hippocampal cells was shown to be required for the induction of LTP,73 and also for learning in the adult rat.74 The presence of polysialic acid even in 29-month-old rats might be associated with memory and learning in aged animals.
Alzheimer's disease (AD) is the most common cause of dementia in the elderly population. Although the pathogenesis of the disease is not fully understood, AD is characterized by the presence of two histopathological hallmark brain lesions, senile plaques and neurofibrillary tangles. It was found that β-amyloid is the major component of senile plaques and hyperphosphorylated tau is the major component of neurofibrillary tangles. Generally, N-glycosylated proteins do not exist in the cytosol; however, in the case of AD, a cytosolic protein tau was N-glycosylated.70 Because AD was thought to be involved in pathological aging, it prompted us to examine the possibility that cytosolic glycoproteins accumulate during aging. We found that 14 N-glycosylated proteins were accumulated in the rat cerebral cortex cytosolic fraction in the aging process by a comparative study with two-dimensional gel electrophoresis and Con A staining (Fig. 4). All proteins had high mannose and/or hybrid-type N-glycans, as indicated by the fact that they were sensitive to endoglycosidase H digestion. Three of these cytosolic glycoproteins were identified as cathepsin D, a lysosomal protease, by tryptic digestion and mass spectrometry.75 The increase of cytosolic cathepsin D during aging was not a result of lysosomal membrane disruption, as shown by the fact that other lysosomal enzymes did not increase in the cytosolic fraction. Although the total amount of cathepsin D increased during aging, the amount of cathepsin D in the microsomal fraction did not change, indicating the selective increase of cytosolic cathepsin D. This phenomenon was also observed in the hippocampus, cerebellum, kidney, liver and spleen. Based on these results, we propose that cytosolic cathepsin D is a new biomarker of aging.
Because of the complexity of cellular interactions in the brain, much more work will be needed to gain a complete understanding of the role of glycan in brain functions. To understand the structure–function relationship of brain glycoproteins in detail, it is necessary to carry out glycoproteomics; for example, protein identification, location of glycosylation sites and structures of glycans attached to each glycosylated site.76 Such a study will help to better understand the function of glycan-mediated cell interactions in the brain. As already discussed, glycans can be altered by the physiological conditions of cells. Accordingly, age-related alteration of glycoproteins is likely to be an important target to solve the various pathological problems found in aged individuals and to be new markers of the aging process.
This work was supported by a Grant-in-Aid for Scientific Research-B (20390031) from the Japan Society for the Promotion of Science. I (T.E.) wish to dedicate this manuscript to the life and career of Yuji Sato, who died on June 21, 2006.