Trends Glycosci.Glycotechnol.8:163-194
*Gannon, Mary C.; and *Nuttall, Frank Q.
Metabolic Research Laboratory and the Section of Endocrinology, Metabolism and Nutrition, Minneapolis Veterans Affairs Medical Center and the Departments of Medicine* and Food Science & Nutrition1 ,University of Minnesota Minneapolis, MN 55417 USA, FAX: 1-612-725-2273, E-mail: ganno004@maroon.tc.umn.edu
Key Words: glycogen, glycogenin, liver, proteoglycan
It generally is agreed that initiation of a glycogen molecule occurs through the addition of a linear array of glucose molecules to a protein referred to as glycogenin. The addition of glucosyl units is catalyzed by glycogenin itself. The product is then branched and expanded by other enzymes. This process has been studied best in skeletal muscle. In skeletal muscle, all of the glycogenin is incorporated into glycogen molecules (1) The glycogen molecules also are all presumed to be proteoglycans.
In liver, glycogenin is free, i.e., not incorporated into a proteoglycogen molecule in the fasted state and it becomes incorporated into a glycogen molecule only later in the feeding phase. Furthermore, preliminary data suggest that most of the glycogen molecules are not proteoglycans but rather are independent of glycogenin. Thus, the character and biosynthesis of glycogen in liver appears to be much more complex than in skeletal muscle. Regulation of the various forms of glycogen also is likely to be complex.
Glycogen is the storage form of glucose in animals. It is a highly branched structure composed of thousands of glucose molecules attached to each other in an alpha-1,4 glucosidic linkage with branch points consisting of alpha-1,6 linkages. The external branches on the surface of the glycogen molecule are estimated to consist of a linear array of 12-14 glucose molecules. The internal branches are ~ 4-6 glycosyl units in length.
The molecular weight (MW) of glycogen is polydisperse. Particles with an estimated MW up to 1 billion have been reported on gradient density analysis and by electron microscopy (2). However, these very large particles most likely represent aggregates or assemblies of individual glycogen molecules. The largest individual glycogen molecule appears to have a MW of ~ 8 million and to contain 50,000 glucose molecules arranged in 12 tiers of branches (2, 3). Why there is a limit in glycogen molecular weight is uncertain. It has been attributed to steric hindrance (4).
In spite of the very large size of some of these molecules, each begins with a single glucose molecule which is referred to as the reducing end because unless it is derivatized to protein or another substance, it has a free aldehyde group which is chemically reactive. Thus, theoretically a glycogen molecule could consist of anywhere from less than 100 glycosyl units to several thousand.
As currently defined, glycogen consists of those structures which after potassium hydroxide (KOH) digestion of tissue are precipitable by a 66% ethanol (EtOH) solution. This is the method against which other extraction methods are compared (5). This procedure, however, is known to degrade glycogen particles. Primarily it affects the association of the glycogen molecules, probably through formation of isosaccharinic acid groups, although some degradation of individual molecules also occurs (6). In order to isolate intact glycogen other methods must be used. A number of these have been described (2, 3, 7-9). Whether ethanol also will precipitate small nascent or immature small glycogen molecules, particularly if they are attached to protein is uncertain, although ethanol has been reported to precipitate even oligosaccharides (10). To our knowledge, this issue has not been rigorously evaluated. Nevertheless, ethanol precipitation currently is considered the best method for isolating and quantifying the total amount of glycogen present in a tissue.
Small glycogen molecules or even malto-oligosaccharides are substrate for glycogen synthase which adds glucosyl units to the non-reducing ends of the branches of glycogen or the non-reducing end of oligosaccharides in an alpha-1,4 linkage. UDP-glucose is the glucosyl donor. Once the chain is elongated to 12-14 units, branching enzyme removes a linear chain of 4-6 glucose molecules and inserts it in an alpha-1,6 linkage forming a branch point. This chain also then may be elongated by the action of glycogen synthase.
We have purified the branching enzyme from rat liver and have found the activity to be 150-fold greater than that of the synthase (unpublished), confirming that it is the synthase which is rate limiting in expansion of the nascent glycogen molecule to a mature glycogen molecule.
Our laboratory also has shown that specific isozymes of synthase are present in liver and skeletal muscle (11-14). Both isozymes are regulated by a complex dephosphorylation-phosphorylation mechanism which also interacts in a complex fashion with allosteric effectors (15, 16). However, the liver enzyme is more highly phosphorylated, and its physiological regulation is different than that of skeletal muscle (15). Recently we have cloned and sequenced the gene for the human liver enzyme (11), have localized the gene to chromosome 12 (17) and have begun site-directed mutagenesis studies in order to probe the structural basis for its regulation. The regulation of synthase has been a major interest in our laboratory for many years.
Although synthase and the branching enzyme activities can explain the expansion of small or nascent glycogen molecules, they are not likely to explain the de novo synthesis of glycogen molecules. Nevertheless, this is a theoretical possibility. The minimal length of a malto-oligosaccharide chain that will serve as a substrate for glycogen synthase was reported to be 1 to 4 glucose units (maltotetraose) (18-20). However, the affinity was much less than when acting on the branches of glycogen and contamination of the synthase preparation with glycogenin cannot be ruled out.
In 1934 WillstŠtter and Rohdewald (9)extracted tissues with hot water. A soluble fraction was obtained, referred to as lyo-glycogen. From the insoluble material a second smaller fraction was obtained after proteolytic enzyme treatment or alkaline digestion which they referred to as desmo-glycogen. They believed the latter form of glycogen to be less soluble due to combination with protein. They also reported that it was lyo-glycogen that changed with feeding. Others referred to the desmo-glycogen fraction as "residual" glycogen (5).
Whether there were two forms of glycogen, one trichloroacetic acid (TCA) precipitable and one not, and whether their metabolism was regulated differently, resulted in numerous publications [reviewed in reference (5)] and became a contentious issue that has never been completely resolved. In a review, Stetten and Stetten (5), as well as Roe et al. (21)suggested that there were many forms of glycogen present in cells and that the classification into an easily extractable form and a more difficult to extract form was somewhat arbitrary. Until recently there has been little interest in pursuing the possibility of different forms being present or the possibility that they may be metabolically important and regulated differently.
Stetten and Stetten (5) also suggested that the protein associated with glycogen as fractionated was due both to a non-specific binding of proteins and to non-covalent attachment of the enzymes concerned with glycogen metabolism. Later, Fischer and Stetten (22) reported that protein made up ~ 40-50% of the glycogen as isolated from skeletal muscle. They and subsequently Cohen (23) attributed the protein to non-covalently bound enzymes important in glycogen metabolism. These included glycogen phosphorylase, phosphorylase kinase, glycogen synthase, synthase kinase 2, and cAMP kinase. Small amounts of other enzymes and sarcoplasmic reticulum also were present. That the protein associated with glycogen represented non-covalently bound proteins, primarily glycogen associated enzymes, was the accepted explanation for the small amounts of protein present in highly purified glycogen preparations from liver, skeletal muscle and other organs.
However, data began to accumulate that the synthesis of an alpha-1-4 glucan attached to a protein was present in liver (24, 25), heart (26), and retina (27) of mammals, Escherichia coli (28, 29), Neurospora crassa (30), potato tubers (31) and carrots (32). UDP-glucose at a micromolar concentration functioned as the glycosyl donor. The product formed was generally but not exclusively (33, 34) found in the TCA precipitable fractions of cell extracts (33- 35). Data were obtained in potato tubers indicating that the product produced was a linear chain of glucose molecules in alpha-1,4 linkage attached to a protein (33). The nature of the protein or the protein-glucan bond was not known. It was shown not to be a serine or threonine-O-glycosyl bond.
It also was speculated that the protein was a foundation or primer upon which new glycogen or starch molecules could be formed (25, 33). That is, it became incorporated into the glycogen or starch molecule. Also, in potato tubers enzymes were present which could cleave it from the ethanol precipitable product (starch).
Earlier investigators also had suggested that formation of a glucoproteic intermediate was a general mechanism for the synthesis of alpha- and betaeta-glucans(36). Thus, the general concept was developed that a UDP-glucose-protein transglucosylase was present in plants and animals which produced a product (either a single glucose or most likely a linear array of glucose molecules attached to a protein) which then was acted upon by other enzymes to produce a mature starch or glycogen molecule.
Subsequently, the protein in potato tubers and carrots (32) and mammals to which glucosyl units were being attached was shown to be a 37-40 kDa protein (37, 38). In carrots it was shown to be self-glycosylating (32). This protein has been purified from rabbit skeletal muscle, the amino acid sequence determined, and a specific tyrosine to which the glucose molecules are attached identified (39). The gene also has been cloned and sequenced (40). It is a manganese dependent self-glycosylating enzymic protein which may add up to 8-12 glucosyl units to itself in a linear array (41). This includes the addition of the first glucose to a specific tyrosine in the protein. Thus, it is a very unique enzyme. Not only is it a self-glucosylating enzyme, it gets incorporated in to the glycogen molecule. It has been given the name, glycogenin (42).
Subsequently, we (unpublished data) and others have obtained data indicating that not only does it glycosylate itself, it also will glycosylate malto-oligosaccharides and artificial alkylglucosides or maltosides (43-45). Indeed, dodecyl-betaeta-D-maltoside is reported to be an excellent substrate for the kidney enzyme (44). The rate of reaction is even faster than for self glycosylation. These observations then raise the possibility that not only does glycogenin self glycosylate and become incorporated into the growing glycogen molecule as the backbone or primer, it also may catalyze the glycosylation of endogenous maltosaccharides or other endogenous substrates such as alkyl derivatives of maltose or malto-oligosaccharides, which in turn may function as primers. That is, glycogenin may have a dual function.
The presence of malto-oligosaccharides in tissue has been reported by several investigators. Beloff-Chain et al. (46) identified a series of malto-oligosaccharides in diaphragm muscle the largest of which was at least malto-heptaose (7 glycosyl units in length). The production of these was accelerated by addition of insulin to an incubated diaphragm preparation. Fishman and Sie (10) reported the isolation of the same family of oligosaccharides up to malto- decaose (10 glycosyl units in length) and even higher from rat liver. They speculated that they were derived from the endogenous degradation of glycogen. However, in a subsequent publication using 14C labeling experiments (47), they concluded that it was likely that the oligo-saccharides were precursor substrates for glycogen synthesis. Mordoh et al. (3) homogenized rat liver in HgCl2 in order to inhibit enzymic activity and reported the presence of oligosaccharides but the concentration (1.3 µmol/g liver using maltose as a standard) was lower. The concentration also did not change with oral sucrose administration. This was in contrast to the increase with feeding reported by Sie and Fishman (47). The higher concentration reported by Sie and Fishman they attributed to endogenous-alpha-amylase not being inhibited during preparation of the liver for assay. They did not consider the amount present to be significant.
Giri et al. (48) described an enzyme preparation from rat liver and brain which generated from maltose a series of alpha-1,4 oligo-saccharides, liberating glucose in the process. A similar glucosylating activity was reported in rabbit liver (49) but these activities have never been confirmed. A D- or dispropor-tionating enzyme has been described in potatoes. This enzyme transferred maltosyl or higher malto-oligosaccharyl residues from oligo- and polysaccharides to various receptors. The glucose concentration in the incubation mixture appeared to regulate the average chain lengths of the maltodextrin produced (50).
The importance, if any, of these transglucosylase activities or of alpha-amylase or alpha-glucosidase activities in glycogen metabolism remains unknown. However, they could provide a non-protein bound primer for synthesis of a glycogen molecule by glycogen synthase and branching enzyme. It should be noted that glycogen synthase may use oligosaccharides as short as maltotetraose as a substrate but the rate was very slow (19).
Thus, that oligosaccharides or small branched or unbranched maltodextrins either produced by cleavage of these from a glycogenin initiated product or from partial degradation of protein-free glycogen molecules are substrates for a formation of new glycogen molecules is a distinct possibility.
A lysosomal, acid, alpha-glucosidase is present which hydrolyzes both alpha-1,4 and alpha-1,6 glycosyl bonds and thus can completely hydrolyze glycogen to glucose or maltose. In its absence glycogen of normal structure accumulates in lysosomes resulting in death or disability (glycogen storage disease, Type II or Pompe's disease) (51). Thus, autophagy of glycogen has been considered to be important in glycogen degradation. The mechanism by which glycogen molecules are selected for and taken up by lysosomes remains unknown, as does its possible regulation. Also, whether the glycogen taken up by the lysosomes is derivatized by glycogenin is not known, nor are the actual products released from microsomes in vivo.
An additional alpha-glucosidase active at neutral pH has been described which releases glucose from glycogen. It is referred to as alpha-glucosidase C. It has been purified, and the gene cloned, sequenced, and assigned to chromosome 15 (52, 53). However, its role, if any, in glycogen metabolism is unknown. In this regard, some investigators have attributed the inability to identify large proteoglycogen species in skeletal muscle extracts, after labeling with 14C using 14C-UDP-glucose as substrate, to the presence of an alpha-glucosidase (54).
Whether there is a cytosolic alpha-amylase active at a neutral pH in addition to the lysosomal acid alpha-amylase remains uncertain but has not been identified in liver.
In animal skeletal muscle, glycogen preparations have invariably contained protein with amounts up to 3% by weight(8, 55) . The minimal amount present generally is ~ 0.3-0.4% (54). Based on the latter, it has been calculated that there is at least one glycogenin molecule present for each glycogen molecule(61, 62). Also, it has been reported that free reducing groups are not present in skeletal muscle glycogen (54). This has led to the concept that in skeletal muscle, all of the glycogen molecules have attached to them a glycogenin molecule. That is, they are all proteoglycans even though the glycan may be extremely large relative to the size of the protein.
However, highly purified preparations of liver glycogen obtained under conditions which minimize degradation have yielded preparations in which the protein content varied between 0.04 to less than 0.01% or was unmeasurable (2, 3, 54) . In liver, if one assumes a molecular weight of 8,000 kDa for the fundamental, mature, glycogen particle, then the molar ratio of glycogenin to glycogen would be 1:12 at a protein content of 0.04% by weight for ethanol precipitable glycogen. That is, only 1 glycogenin protein for each 12 glycogen molecules. The ratio is likely to be greater if even smaller amounts of protein are present as has generally been reported. This is less than the proposed presence of one glycogenin for each alpha particle (63).
Many years ago the presence of reducing groups in rabbit liver glycogen samples with molecular weights of 0.1 to 34 million (6) was reported. The glycogen samples were subjected to anaerobic strong alkali treatment. The latter results in production of isosaccharinic acid at the reducing end of the molecule. Skeletal muscle or other tissues were not studied. In these studies care was not taken to rapidly remove and process the liver samples and all samples were subjected to acid or alkali extraction. Thus, they likely had been modified.
Overall, the best evidence obtained many years ago is that the majority of glycogen molecules in liver in contrast to skeletal muscle (and probably other tissues), do not have a glycogenin protein attached. Whether the majority have a free reducing end present has been less certain.
We have obtained preliminary data which indicate that indeed free reducing ends are present (Table I), and that glycogen molecules with free reducing ends may represent the majority of molecules present.
Table I. Stoichiometry of Incorporation of Tritium into Glycogen. Glycogen pellets (200,00 x g) were prepared from fed rat livers using the least manipulation possible, and compared with a commercial rabbit liver glycogen preparation. They were incubated for 2 hours with an excess of 3H NaBH4. Replicate experiments were performed on different days. The number of moles of aldehyde reduced in a population of glycogen molecules was calculated based on a molecular weight of glycogen of 8000 kDaltons, and the appropriate cpm/pmole of 3H NaBH4.
If a large number of free reducing groups are present and if glycogenin is assumed to initiate the formation of all new glycogen molecules, then a mechanism for cleaving glycogenin from the glycogen molecules must be present. This could be an enzymic process and could be important in regulating the rate of formation of new individual glycogen molecules. That is, it could regulate the number of glycogen molecules present in different physiological states. Alternatively, malto-oligosaccharides may be produced and function as primers for new glycogen synthesis. These are possibilities that we currently are pursuing.
Using bovine retinal membrane preparations Lacoste et al. (56)have provided evidence that a small proteoglycogen containing glycogenin is found primarily in the microsomal fraction on ultracentrifugation and is attached to a membranous structure. This proteoglycan is TCA precipitable. Incubation with a micromolar concentration of UDP-glucose resulted in labeling of the proteoglycogen. Incubation with a higher concentration of UDP-14C glucose resulted in release from the membrane structure and the product now became soluble in TCA and precipitable by ethanol. This suggested expansion of the glycan to a large polysaccharide structure. A pulse-chase experiment indicated the conversion of the proteoglycan, in which the glycogen moiety was small, to one in which it was considerably larger. The expansion of the glycogen structure was attributed to the activity of glycogen synthase and branching enzyme, although the actual structure of the product was not determined. Addition of a detergent (SDS) also resulted in dissociation from the membranes. In this regard it has been reported that the glycogen in liver of fasted rats is difficult to solubilize and that the glycogen in a 12,000 ¥ g pellet was only completely extracted when chloroform was added (57). Thus, these data suggest that a fraction of nascent glycogen may be membrane bound.
Interestingly, in the retinal membrane preparations (56), addition of rabbit liver glycogen in excess during the incubation with 10 mM UDP-glucose did not result in a competitive inhibition of the expansion of the proteoglycogen species. This suggests the presence of a mechanism for addition of glucosyl units to a proteoglycogen which is independent from that of "free" glycogen. Whether glycogen synthase and branching enzyme are responsible for expansion of both forms remains to be determined. We also have provided data suggesting that the two processes are regulated independently in vivo (58).
In liver extracts from 24 hour fasted rats, essentially all glycogenin was present in a free form, i.e. it was not incorporated into a glycogen molecule (58). Glucose could be transferred from UDP-glucose to the glycogenin. Following oral glucose administration to fasted rats, the glycogen concentration increased for one hour before glycogenin was incorporated into a large proteoglycan product. Thus, there appears to be a "free" glycogen synthetic pathway, and a proteoglycogen synthetic pathway which are regulated independently in liver as well as in retinal membrane preparation.
In the TCA precipitable fractions of the retinal membranes, the glycogen to which glycogenin was attached had an average Mr of 470 kDa as determined on a sizing column. The TCA soluble fraction had an Mr of ~ 700 kDa, i.e., both were considerably smaller than a mature glycogen molecule (35). Lomako and associates (59)subsequently reported that in skeletal muscle preparations the TCA precipitable form of glycogen also had an Mr of approximately 400 kDa, i.e., considerably smaller than the Mr of the TCA soluble glycogen (~ 8000-10,000 kDa). They have referred to the former as "proglycogen". They also have commented that species smaller than 400 kDa are uncommonly reported. This suggested to the authors that a glycogenin initiated nascent glycogen molecule is rapidly expanded to this 400 kDa species. Using primary cultures of astrocytes from 1-2 day old rats they also provided evidence for an independent regulation of the TCA insoluble and soluble fractions and for the TCA precipitable species being a precursor of the TCA soluble species (60). Thus, in retina, skeletal muscle, astrocytes and liver(58)there is evidence for different forms of glycogen being present and for independent regulation of each. In skeletal muscle, glycosylation of the 400 kDa species and those with a Mr between 42 and 400 kDa occurred with a micromolar concentration of 14C-UDP-glucose(59)a concentration considerably below the Km for UDP-glucose of glycogen synthase. Whelan and associates (60, 62)suggest a specific proglycogen synthase in addition to classical glycogen synthase is present and this accounts for the activity observed or that glycogen synthase having different phosphorylation states has affinities that are different for "proglycogen" and glycogen. To our knowledge, conversion of proteoglycogen species to a mature (8,000 kDa) species has not been reported in vitro.
In liver little is known about the distribution of glycogen between TCA precipitable and soluble fractions, the size of the proteoglycogen in these fractions or how they may vary with feeding or fasting or with glucose administration, etc.
In summary, evidence is developing that indicates the presence of two or more forms of proteoglycogen. In liver a protein-free species appears to be present as well. The interrelationships among these species, their regulation, and their role in fuel metabolism remains to be determined.
Current data also indicate that the initiation of glycogen molecules is a much more complex process than originally believed. It now is clear that a self-glycosylating protein containing a glucose molecule attached by its reducing end to a tyrosyl residue in the protein may initiate the synthesis of a glycogen molecule and that the protein may become permanently attached to it. This enzyme which glycosylates itself has been best studied in skeletal muscle. However, even in this organ regulation of the amount of enzyme or of its catalytic activity is poorly understood. Its role in the overall regulation of the amount of glycogen that may be stored in different physiological or pathological states also remains uncertain.
In liver the initiation of glycogen molecules is likely to be even more complex. A similar enzyme is present but whether it is identical to skeletal muscle glycogenin is not known but it is likely to be the case (40, 62-64) . In addition, in liver the majority of glycogen molecules apparently do not have attached to them a glycogenin molecule. How these glycogenin-free glycogen molecules are generated is completely unknown. Whether they are derived from the hydrolysis of the bond between glycogenin and glycogen (Fig. 1a) or hydrolysis of a glycogenin (Fig. 1b) synthesized malto-oligosaccharide attached to glycogenin or are generated by fragmentation of glycogen itself by amylase or another enzyme producing either linear or branched products (Fig. 1c) also is completely unknown. It also is possible that the glycogenin-bound and glycogenin-free glycogen molecules are synthesized independently and are regulated independently and have a different cellular location. In any regard, how glycogenin-free glycogen molecules are generated is likely to play an important role in determining the number of glycogen molecules in a liver cell in different physiological and pathophysiological states. It also is likely to regulate the total amount of glycogen that can be stored, i.e. the cellular capacity to store glycogen.
Supported by Merit Review Research Funds from the Department of Veterans Affairs, and grant #DK43018 from the National Institutes of Health. The authors would like to thank Claudia Durand for her expert secretarial assistance.
Received on February 9, 1996, accepted on April 15, 1996
ReferencesBack to contents
Up to top