Glycogenin-1
Created by Michael Tanes
Glycogenin-1 (PDB ID = 1LL2) from Oryctolagus cuniculus initiates glycogen synthesis. Glycogen synthase and a branching enzyme need a glucose oligosaccharide (maltosaccharide) to construct glycogen. Glycogenin-1 is responsible for creating the maltosaccharide primer by taking the glucose from uridine 5’-diphosphoglucose (UDPG) and linking the sugars through α-1,4-glycosidic linkages. Non-primate mammals express the single isoform of glycogenin-1 in liver and muscle cells while humans actually have two isoforms of glycogenin: glycogenin-1 and glycogenin-2. Skeletal muscles contain glycogenin-1 and the liver contains glycogenin-2 while heart muscle contains both isoforms (1,2). Glycogen stores energy for cells, and glycogenin-1 is essential to its synthesis. Diseases in humans that affect glycogenin-1 are characterized by an absence of glycogen in muscle, which leads to fatigue, exertional dyspnea, and heart complications (2).
Glycogenin-1 is part of glycosyltransferase family 8, a set of proteins that are known for having a highly conserved motif to facilitate the removal of uridine 5’-diphosphate (UDP) from a sugar group. However, proteins within the group have dissimilar sequences. The rest of the structure depends on the sugar group, the type of catalysis, and the acceptor of the sugar group. Glycogenin-1 is a retaining glycosyltransferase because the stereochemistry of the anomeric carbon is conserved between reactant and product (1,3). It is a globular protein that has a single domain consisting of an α+β folding pattern arranged into 4 layers. The two interior layers are composed of a six-stranded β-sheet and a three-stranded β-sheet, which are surrounded by α-helices in the two outer layers (1). Glycogenin-1 is a 333-residue protein that has a molecular weight of 37,397.02 Da and its isoelectric point (pI) is 5.07.
A channel forms along the C-terminal ends of the six-stranded β-sheet in the tertiary structure to form the active site of glycogenin-1. The six-stranded β-sheet is a mixed parallel-antiparallel sheet where four strands run parallel to each other. The fifth strand runs antiparallel to the first four, and the sixth runs antiparallel to the fifth strand (parallel to the first four). The first three strands run parallel because the sequence forms a pattern of β-strand, turn, α-helix, turn, β-strand, turn, α-helix, turn, β-strand. This β-α-β folding pattern is known as a Rossman-like fold (a complete Rossman fold uses a parallel, six-stranded β-sheet instead of the mixed sheet in glycogenin-1) and is involved in the binding of nucleotides. Many residues in the active site confer specificity for the uracil ring. Tyr-15 and Val-82 sandwich the uracil ring between them through hydrophobic interactions, while Thr-11 forms hydrogen bonds with the edge of the ring. However, these interactions do not exclude cytidine from the active site. Asn-12, which hydrogen-bonds with the carbonyl group on the uracil ring, gives the active site specificity for uracil because its carboxamide is restricted in motion by hydrogen bonds formed with its α-helix so it cannot form a hydrogen bond with the amine group on the cytosine ring. The ring is held in place through hydrogen bonds formed between hydroxyl groups (on the uridine ring) and the peptide carbonyl group of Leu-9 and the peptide nitrogen atoms of Ala-103 and Asp-104 (1).
Glycosyltransferases have a highly conserved DXD motif to bind a divalent cation in the active site. This motif is composed of Asp-102–X-103–Asp-104 (the middle residue does not matter because the DXD motif is part of a turn and it faces away from the active site) in glycogenin-1 and is essential to its function because any mutation of the aspartate residues results in a non-functional enzyme. The DXD motif plus His-212 coordinate with a manganese (II) ion (Mn+2), which stabilizes the diphosphate group of UDPG and facilitates the removal of UDP from glucose. The distances between Mn+2, Asp-102, Asp-104, His-212, and the diphosphate group are such that, if Mn+2 were substituted with any other divalent cation, enzyme activity would decrease considerably. Tyr-15, Lys-218, and the peptide nitrogen atom of Gly-215 form hydrogen bonds with the diphosphate group, as well (1).
Asp-125, Asn-133, and Gln-164 form hydrogen bonds with the 4’-hydroxyl group that is present on glucose, thereby giving glycogenin-1 specificity for glucose instead of galactose. Ser-134 forms a hydrogen bond with the 6’-hydroxyl group on glucose (1). All interactions at the active site can be visualized in Figure 1.
The mechanism glycogenin-1 uses to transfer the glucose from UDPG to the acceptor molecule, which is either Tyr-195 or the growing oligosaccharide, is still a bit of a mystery. The controversy is whether the glucose is transferred by a double SN2 reaction, where two inversion reactions take place as the glucose is transferred from an intermediate acceptor to the final acceptor molecule, or by a SNi mechanism, which lacks an intermediate acceptor. In either mechanism, the roles of two aspartate residues in the active site remain the same. Asp-163 stabilizes the oxocarbenium intermediate, which facilitates the removal of the UDP group from glucose, and Asp-160 guides the acceptor molecule to the active site (3). In this crystallization, Asp-163 is too far from the glucose in the active site, which suggests a necessary conformational change in glycogenin-1 to move Asp-163 towards the oxocarbenium intermediate. Non-proline cis peptide bonds are rare in proteins, and when they do occur, are important to the function of the protein. Glycogenin-1 contains a cis peptide bond between Glu-119 and Leu-120. The bond would be unstable, but hydrogen bonds formed between Glu-119 and Trp-90, the peptide nitrogen atom of Gln-140, and Lys-181 plus hydrophobic interactions between Leu-120 and Trp-90, Phe-132, Phe-137, Leu-167, and Phe-171 stabilize the cis peptide bond. If Glu-119 or Leu-120 rotate to a trans conformation, glycogenin-1 could change its conformation and move Asp-163 close enough to the oxocarbenium intermediate (1).
Glycogenin-1 forms dimers with two identical subunits. Gibbons et al. also crystallized glycogenin-1 as a decamer arranged as a pentamer of dimers. However, they discarded this quaternary structure was discarded as solely a product of crystallization because the active sites of the inner ring of glycogenin-1 subunits would face the inside of the ring. This arrangement would restrict the access of glycogen synthase to the maltosaccharide primer preventing glycogen synthesis, which makes the decamer form of glycogenin-1 physiologically irrelevant. The function of glycogenin-1 is linked to its dimer form. The attachment of the first 1-4 glucose units is achieved through an intermolecular process between two dimers. When the chain grows longer, additional glucose units are added to the growing oligosaccharide via an intramolecular process between the two subunits. One subunit adds glucose molecules to the oligosaccharide anchored at the Tyr-195 of the adjacent subunit because the growing chain elongates towards the active site of the adjacent subunit (1,4). Romero et al. found that the maltosaccharide anchored at Tyr-195 adopts a left-handed helical structure such that after 13 glucose units, the end of the maltosaccharide is too far away from the glycogenin-1 active site for any additional units to be added (5). Glycogenin-1 cannot transfer glucose groups to free oligosaccharides. Tyr-195 is essential to the function of glycogenin-1 because it anchors and positions the growing oligosaccharide near the active site of the adjacent subunit. If Tyr-195 is replaced with a phenylalanine residue, glycogenin-1 ceases functioning (6).
Four cysteine groups (Cys-89, Cys-98, Cys-131, and Cys-271) on each subunit facilitate the association of the two glycogenin-1 monomers into a dimer (4). The dimer interface consists of residues 125-132, 160-165, 176-193, 194, 197, 198, 200-202, 204, and 205 (1). Cys-89 and Cys-98 are not part of the interface but are located towards the interior of the glycogenin-1 monomer and face each other which indicates a disulfide bond. Cys-131 is located in the dimerization domain and the whereabouts of Cys-271 are unknown because residues 232-240 and ~260-333 were disordered in the crystallization (1). All four cysteine groups contribute to dimerization in some way because, in an experiment by Dweck et al. when all permutations of two to four of the cysteine groups were replaced with serine groups, glycogenin-1 lost the ability to dimerize. However, glycogenin-1 monomers would still dimerize when a serine group replaced only one of the four cysteine groups (4). This suggests that Cys-89 and Cys-98, which are located in the interior of the protein, form a disulfide bond that stabilizes the tertiary structure of glycogenin-1 in such a way that allows the Cys-131 and Cys-271 residues of one monomer to form disulfide linkages with the Cys-131 and Cys-271 residues of another monomer. If one of the interior cysteine groups is replaced with serine, the monomer loses the stability in tertiary structure, but the two dimerization groups of one monomer can still match up with another monomer and dimerize. If one of the dimerization groups is replaced with serine, the stability in tertiary structure keeps the remaining dimerization group exposed so that it can still bond with the other dimerization group on another monomer. When two or more cysteine groups are replaced with serine, the protein either loses both dimerization groups or the loss of tertiary structure stability coupled with loss of one or more dimerization groups, which causes glycogenin-1 to lose the ability to dimerize.
Dimerization of glycogenin-1 is concentration dependent. Bazan et al. found that monomeric glycogenin-1 existed at concentrations less than 0.6 μM in solution and that it was able to autoglucosylate via an intramolecular process. However, the rate of autoglucosylation for the dimer was twice as fast as the rate for the monomer. The relative abundance of glycogenin-1 monomer to dimer could be an important mechanism for regulating glycogen synthesis (7).
Unfortunately, the carboxy-terminal residues beyond residue 260 were disordered and, therefore, their structure could not be analyzed. However, residues 301-333 have been shown to bind to glycogen synthase. The W-E-X2-4-D-Y-L/M motif is conserved in the glycogenin of many species (8).
As the initiator, glycogenin-1 is poised to be a regulator of glycogen synthesis. However, glycogenin-1 contains no known regulatory domains. Protein-protein interactions are essential to glycogenin’s function: dimerization facilitates growth of the priming oligosaccharide, association with glycogen synthase, and co-localization with actin mediated by the binding of the carboxyl-terminal of glycogenin with actin. Skurat et al. found four glycogenin interacting proteins (GNIP): GNIP1, GNIP2, GNIP3, and TRIM7. Their effects on glycogenin-1 still need to be studied, but binding of GNIP2 to glycogenin-1 activated auto-glucosylation.GNIP2 might alter the conformation of the glycogenin-1 dimer to achieve its observed effect (9).
PSI-BLAST and a Dali server search identified proteins related to glycogenin-1. PSI-BLAST compares the primary structures of two proteins and calculates an E-value based on their similarity. An E-value of 0.05 or less means the sequences are significantly similar and an E-value of 0.0 indicates the two proteins have the same sequence. The Dali server compares the tertiary structures of two proteins and calculates a Z-score based on a sum-of-pairs method, which compares intermolecular distances (10). A Z-score above 2 indicates the two proteins have similar folds. PSI-BLAST and the Dali server search found two proteins of interest: glycogenin-1 from Homo sapiens (PDB ID = 3QVB) and α-1,4-galactosyltransferase (LgtC) from Neisseria meningitides (PDB ID = 1G9R).
Glycogenin-1 from Homo sapiens had an E-value = 3e-147 and Z-score = 41.7 (r.m.s.d.=0.8 angstroms), which indicate that sequence and tertiary structure are highly conserved in glycogenin-1 from different species. This makes sense since glycogenin-1 serves the same function in rabbit and human muscle.
LgtC had an E-value = 0.030 and a Z-score = 21.1 (r.m.s.d.= 2.6 angstroms). These results indicate that LgtC and glycogenin-1 have similar sequences and folding patterns. In fact, LgtC and glycogenin-1 share some similarities in structure and function. Both are part of glycosyltransferase family 8, are retaining glycosyltransferases, and bind Mn+2. The conserved DXD motif is responsible for binding Mn+2, and a Rossman-like fold is present for nucleotide binding. Unlike glycogenin-1, the substrate for LgtC is UDP-galactose (same nucleotide but different sugar group) and it contains two 310-helices. Glycogenin-1 is a primer for glycogen synthase so its C-terminal residues function as a binding site for glycogen synthase. But LgtC is associated with the production of lipooligosaccharides (LOSs) that are present on the surface of bacterial cells to help them evade the host’s immune system. Since LgtC is associated with LOSs and is not a primer for another protein, its C-terminal residues are radically different from those of glycogenin-1, which are used to bind to glycogen synthase. The C-terminal residues consist of many basic and hydrophobic and aromatic residues so that LgtC can associate with the cell membrane through electrostatic interactions with the heads of phospholipids and hydrophobic interactions with the tails. LgtC only adds one α-galactose to the lactose-terminal of a LOS so it does not need to covalently bond with the product like glycogenin-1 does with its growing oligosaccharide. This makes sense as to why LgtC exists as a monomer instead of a dimer since there is no need for autoglycosylation (11).
Due to the importance of glycogen as quick-energy storage, mutations that affect glycogenin-1 activity have serious, but non-fatal implications. For example, a mutation in the glycogenin-1 gene of a 27-year-old patient caused glycogenin-1 to lose the ability to autoglucosylate. Loss of glycogenin-1 function meant no primer could be formed and so there was no glycogen in the patient’s skeletal muscles. This led to shortness of breath and weakness in the chest and arm muscles. It also led to an episode of ventricular fibrillation after exercise. In order to make up for the lack of glycogen, the patient’s muscle compensated by increasing mitochondria counts and decreasing the prevalence of fast-twitch muscle fibers, which rely on glycolytic sources to generate ATP, in favor of slow-twitch muscle fibers, which rely on aerobic respiration to generate ATP (12).