Statistical analysis of the protein environment of N-glycosylation sites: implications for occupancy, structure, and folding

We recently reported statistical analysis of structural data on glycosidic linkages. Here we extend this analysis to the glycan-protein linkage, and the peptide primary, secondary, and tertiary structures around N-glycosylation sites. We surveyed 506 glycoproteins in the Protein Data Bank crystallographic database, giving 2592 glycosylation sequons (1683 occupied) and generated a database of 626 nonredundant sequons with 386 occupied. Deviations in the expected amino acid composition were seen around occupied asparagines, particularly an increased occurrence of aromatic residues before the asparagine and threonine at position +2. Glycosylation alters the asparagine side chain torsion angle distribution and reduces its flexibility. There is an elevated probability of finding glycosylation sites in which secondary structure changes. An 11-class taxonomy was developed to describe protein surface geometry around glycosylation sites. Thirty-three percent of the occupied sites are on exposed convex surfaces, 10% in deep recesses and 20% on the edge of grooves with the glycan filling the cleft. A surprisingly large number of glycosylated asparagine residues have a low accessibility. The incidence of aromatic amino acids brought into close contact with the glycan by the folding process is higher than their normal levels on the surface or in the protein core. These data have significant implications for control of sequon occupancy and evolutionary selection of glycosylation sites and are discussed in relation to mechanisms of protein fold stabilization and regional quality control of protein folding. Hydrophobic protein-glycan interactions and the low accessibility of glycosylation sites in folded proteins are common features and may be critical in mediating these functions.     

Do N-glycoproteins have preference for specific sequons?

Protein N-glycosylation requires the presence of asparagine (N) in the consensus tri-peptide NXS/T (where X is any amino acid, S is serine and T is threonine). Several factors affect the glycosylation potential of NXS/T sequons and one such factor is the type of amino acid at position X. While proline was shown to negatively affect N-glycosylation, the nature of other amino acids at this position is not clear. Using Markov chain analysis of tri-peptide NXS/T from viral, archaeal and eukaryotic proteins as well as experimentally confirmed N-glycosylated sequons from eukaryotic proteins, we show here that the occurrence of most sequon types differ significantly from the expected probability. Sequon types with F, G, I, S, T and V amino acids are consistently preferred while those with P and charged amino acids are under-represented in all four groups. Further, proteins contained far fewer number of possible sequon types (maximum 20 types for NXS or NXT taken separately) for any given number of sequons, which may be explained based on random sampling. Consistent with the present finding, majority of the over-represented sequons found in two important viral envelope glycoproteins (hemagglutinin of influenza A H3N2 and glycoprotein120 of HIV-1) are indeed preferred sequon types, which may provide a selective advantage. Accordingly, although there seems to be some preference for sequons, this preference may not be unique to N-glycosylation.     

Cation-pi interactions studied in a model coiled-coil peptide

Cation-pi interactions between aromatic amino acids and the positively charged residues lysine and arginine have been proposed to play an important role in stabilizing protein structure. We have used a peptide that adopts a coiled coil structure as a model system to evaluate the energetic contribution of cation-pi interactions to protein folding. Peptides were designed in which phenylalanine, tyrosine, and tryptophan were placed at a solvent-exposed position of the helix, one turn removed from an arginine residue that could provide a favorable cation-pi interaction. Only the arginine-phenylalanine pairing provided significant stabilization of the peptide structure and it appears that hydrophobic packing, rather than the cation-pi effect, is more likely to be responsible for the stability of this peptide. We conclude that any stabilizing effect of cation-pi interactions in these peptides is much smaller than that predicted from computational studies.     

Increased N-Glycosylation Efficiency by Generation of an Aromatic Sequon on N135 of Antithrombin

The inefficient glycosylation of consensus sequence on N135 in antithrombin explains the two glycoforms of this key anticoagulant serpin found in plasma: α and β, with four and three N-glycans, respectively. The lack of this N-glycan increases the heparin affinity of the β-glycoform. Recent studies have demonstrated that an aromatic sequon (Phe-Y-Asn-X-Thr) in reverse β-turns enhances N-glycosylation efficiency and stability of different proteins. We evaluated the effect of the aromatic sequon in this defective glycosylation site of antithrombin, despite of being located in a loop between the helix D and the strand 2A. We analyzed the biochemical and functional features of variants generated in a recombinant cell system (HEK-EBNA). Cells transfected with wild-type plasmid (K133-Y-N135-X-S137) generated 50% of α and β-antithrombin. The S137T, as previously reported, K133F, and the double mutant (K133F/S137T) had improved glycosylation efficiency, leading to the secretion of α-antithrombin, as shown by electrophoretic and mass analysis. The presence of the aromatic sequon did not significantly affect the stability of this conformationally sensitive serpin, as revealed by thermal denaturation assay. Moreover, the aromatic sequon hindered the activation induced by heparin, in which is involved the helix D. Accordingly, K133F and particularly K133F/S137T mutants had a reduced anticoagulant activity. Our data support that aromatic sequons in a different structural context from reverse turns might also improve the efficiency of N-glycosylation.     

Subtle evolutionary changes in the distribution of N-glycosylation sequons in the HIV-1 envelope glycoprotein 120

Many viruses are known to undergo rapid evolutionary changes under selective pressures. The HIV-1 envelope glycoprotein 120 (gp120) shows extreme selection for NXS/T sequons, the potential sites of N-glycosylation. Although the average number of sequons in gp120 appears to be relatively stable in the recent past, even slight changes in the distribution of sequons may potentially play crucial roles in protein interaction and viral infection. This study tracked the prevalence and distribution of NXS/T sequons in gp120 over a period of 29 years (from 1981 to 2009). The gp120 showed location specific distribution of sequons with higher density in the outer domain of the molecule. The NXT sequon density decreased in the outer domain (despite the increase in the sequon specific amino acid threonine), but increased in the inner domain. By contrast, the NXS sequon density increased specifically in the outer domain. Related changes were also seen in the distribution probabilities of sequons in two domains. The results indicate that the gp120, chiefly in subtype B, is redistributing NXS/T sequons within the molecule with specific selection for NXS sequons. The subtle evolution of sequons in gp120 may have implications in viral resistance and infection.     

Role of aromatic stack pairing at the catalytic site of gelonin protein

Aromatic–aromatic interactions play an important role in the enzyme–substrate recognition mechanism and in stabilization of proteins. Gelonin – a ribosome inactivating protein (RIP) from the plant Gelonium multiflorum – belongs to type-I RIPs and shows N-glycosylation activity which has been used as a model to explain the role of aromatic–aromatic stack pairing in RIPs. RIPs have a different substrate binding site and catalytic site. Role of tyrosine residues at the binding site has already been known but the role of tyrosine residues at catalytic site is still unclear. In this study, the role of tyrosine–adenine–tyrosine aromatic stack pairing at the catalytic site was studied by in silico mutation studies using molecular dynamic simulations. Through this study we report that, despite the fact that aromatic stack pairing aids in recognition of adenine at binding site, both the tyrosine residues of stack pairing play a crucial role in the stabilization of adenine at catalytic site. In the absence of both the tyrosine residues, adenine was unstable at catalytic site that results in the inhibition of N-glycosylation activity of gelonin protein. Hence, this study highlights the importance of π–π stack pairing in the N-glycosidic activity of gelonin by determining its role in stabilizing adenine at catalytic site.     

Evidence for a strong selenium-aromatic interaction derived from crystallographic data and ab initio quantum chemical calculations

Increasing attention is being paid to the role of selenium, both as an essential component required for the activity of many enzymes and in the context of selenium-based pharmaceutical agents. A wide range of therapeutics that include selenium are on the market and under development, such as antihypertensive, anticancerogenic, antiviral, and immunosuppressive agents. Computer-aided drug design (CADD) has proven to be an important tool for the development of new drugs. Many CADD techniques, including docking, molecular dynamics simulation, and other receptor-based approaches, require an accurate understanding of the nature of the intermolecular forces that act to stabilize protein-ligand complexes; moreover, a quantitative assessment of these interactions furthers our efforts to rationalize the drug design process. In this paper, we consider one class of interaction involving selenium, that between Se and aromatic rings. Prior work has shown that interactions between divalent sulfur and aromatic rings are observed much more frequently than would be expected on the basis of chance, both in protein structures and the crystal structures of organic compounds that include these moieties. Recent studies on the optimization of inhibitor-protein binding also suggest that sulfur-aromatic interactions are important in stabilizing these complexes and may be crucial focal point for CADD. Given that selenium and sulfur have similar chemistry, and that selenium is significantly more polarizable, we propose that Se-aromatic interactions may also play an important stabilizing role in the structure of folded proteins and in drug-protein complexes. We have tested this hypothesis against data from the Cambridge Crystallographic Database and ab initio quantum chemical calculations. We have found evidence that selenium does interact strongly with aromatic rings and may play a role analogous to sulfur in stabilizing protein folds. In addition, selenium should be considered along with sulfur in rational drug design strategies that seek to improve binding to target protein sites that include aromatic rings.     

An analysis of amino acid sequences surrounding archaeal glycoprotein sequons

Despite having provided the first example of a prokaryal glycoprotein, little is known of the rules governing the N-glycosylation process in Archaea. As in Eukarya and Bacteria, archaeal N-glycosylation takes place at the Asn residues of Asn-X-Ser/Thr sequons. Since not all sequons are utilized, it is clear that other factors, including the context in which a sequon exists, affect glycosylation efficiency. As yet, the contribution to N-glycosylation made by sequon-bordering residues and other related factors in Archaea remains unaddressed. In the following, the surroundings of Asn residues confirmed by experiment as modified were analyzed in an attempt to define sequence rules and requirements for archaeal N-glycosylation.     

Proteome-wide analysis of single-nucleotide variations in the N-glycosylation sequon of human genes

N-linked glycosylation is one of the most frequent post-translational modifications of proteins with a profound impact on their biological function. Besides other functions, N-linked glycosylation assists in protein folding, determines protein orientation at the cell surface, or protects proteins from proteases. The N-linked glycans attach to asparagines in the sequence context Asn-X-Ser/Thr, where X is any amino acid except proline. Any variation (e.g. non-synonymous single nucleotide polymorphism or mutation) that abolishes the N-glycosylation sequence motif will lead to the loss of a glycosylation site. On the other hand, variations causing a substitution that creates a new N-glycosylation sequence motif can result in the gain of glycosylation. Although the general importance of glycosylation is well known and acknowledged, the effect of variation on the actual glycoproteome of an organism is still mostly unknown. In this study, we focus on a comprehensive analysis of non-synonymous single nucleotide variations (nsSNV) that lead to either loss or gain of the N-glycosylation motif. We find that 1091 proteins have modified N-glycosylation sequons due to nsSNVs in the genome. Based on analysis of proteins that have a solved 3D structure at the site of variation, we find that 48% of the variations that lead to changes in glycosylation sites occur at the loop and bend regions of the proteins. Pathway and function enrichment analysis show that a significant number of proteins that gained or lost the glycosylation motif are involved in kinase activity, immune response, and blood coagulation. A structure-function analysis of a blood coagulation protein, antithrombin III and a protease, cathepsin D, showcases how a comprehensive study followed by structural analysis can help better understand the functional impact of the nsSNVs.     

On the frequency of protein glycosylation, as deduced from analysis of the SWISS-PROT database

The SWISS-PROT protein sequence data bank contains at present nearly 75,000 entries, almost two thirds of which include the potential N-glycosylation consensus sequence, or sequon, NXS/T (where X can be any amino acid but proline) and thus may be glycoproteins. The number of proteins filed as glycoproteins is however considerably smaller, 7942, of which 749 have been characterized with respect to the total number of their carbohydrate units and sites of attachment of the latter to the protein, as well as the nature of the carbohydrate-peptide linking group. Of these well characterized glycoproteins, about 90% carry either N-linked carbohydrate units alone or both N- and O-linked ones, attached at 1297 N-glycosylation sites (1.9 per glycoprotein molecule) and the rest are O-glycosylated only. Since the total number of sequons in the well characterized glycoproteins is 1968, their rate of occupancy is 2/3. Assuming that the same number of N-linked units and rate of sequon occupancy occur in all sequon containing proteins and that the proportion of solely O-glycosylated proteins (ca. 10%) will also be the same as among the well characterized ones, we conclude that the majority of sequon containing proteins will be found to be glycosylated and that more than half of all proteins are glycoproteins.     

Biases and complex patterns in the residues flanking protein N-glycosylation sites

N-Glycosylation, the most common and most versatile protein modification reaction, occurs at the beta-amide of the aspargine of the Asn-Xaa-Ser/Thr sequon. For reasons that are unclear, not all such sequons are glycosylated. To find patterns that affect glycosylation, we examined the amino acid residues from the 20th preceding the sequon to the 20th residue following it, using bioinformatics tools. A clean data set of annotated, experimentally verified, glycosylated and nonglycosylated sequons derived from 617 well-defined nonredundant N- and N-,O-glycoproteins listed in SWISS-PROT (June 2002) was used. NXS and NXT sequons were analyzed separately. Although no overt patterns were found to explain sequon occupancy or nonoccupancy, trends for over- or underrepresentation of certain amino acids at particular positions were statistically significant and different in NXS and NXT sequons. In extension of earlier reports, none of the 80 Asn-Pro-Ser/Thr found were glycosylated, and a markedly low level of glycosylation was seen in sequons with Pro at the position following the Ser/Thr. In addition, a general observation was made that the considerable number of glycosylated sequons in the C-terminal 10 residues of glycoproteins suggests that N-glycosylation in these cases may be posttranslational and not cotranslational, as widely accepted.     

Structural characterization of recombinant soluble rat neuroligin 1: mapping of secondary structure and glycosylation by mass spectrometry

Neuroligins (NLs) are a family of transmembrane proteins that function in synapse formation and/or remodeling by interacting with beta-neurexins (beta-NXs) to form heterophilic cell adhesions. The large N-terminal extracellular domain of NLs, required for beta-NX interactions, has sequence homology to the alpha/beta hydrolase fold superfamily of proteins. By peptide mapping and mass spectrometric analysis of a soluble recombinant form of NL1, several structural features of the extracellular domain have been established. Of the nine cysteine residues in NL1, eight are shown to form intramolecular disulfide bonds. Disulfide pairings of Cys 117 to Cys 153 and Cys 342 to Cys 353 are consistent with disulfide linkages that are conserved among the family of alpha/beta hydrolase proteins. The disulfide bond between Cys 172 and Cys 181 occurs within a region of the protein encoded by an alternatively spliced exon. The disulfide pairing of Cys 512 and Cys 546 in NL1 yields a structural motif unique to the NLs, since these residues are highly conserved. The potential N-glycosylation sequons in NL1 at Asn 109, Asn 303, Asn 343, and Asn 547 are shown occupied by carbohydrate. An additional consensus sequence for N-glycosylation at Asn 662 is likely occupied. Analysis of N-linked oligosaccharide content by mass matching paradigms reveals significant microheterogeneous populations of complex glycosyl moieties. In addition, O-linked glycosylation is observed in the predicted stalk region of NL1, prior to the transmembrane spanning domain. From predictions based on sequence homology of NL1 to acetylcholinesterase and the molecular features of NL1 established from mass spectrometric analysis, a novel topology model for NL three-dimensional structure has been constructed.     

The role of aromatic residues in the hydrophobic core of the villin headpiece subdomain

Small autonomously folding proteins are of interest as model systems to study protein folding, as the same molecule can be used for both experimental and computational approaches. The question remains as to how well these minimized peptide model systems represent larger native proteins. For example, is the core of a minimized protein tolerant to mutation like larger proteins are? Also, do minimized proteins use special strategies for specifying and stabilizing their folded structure? Here we examine these questions in the 35‐residue autonomously folding villin headpiece subdomain (VHP subdomain). Specifically, we focus on a cluster of three conserved phenylalanine (F) residues F47, F51, and F58, that form most of the hydrophobic core. These three residues are oriented such that they may provide stabilizing aromatic–aromatic interactions that could be critical for specifying the fold. Circular dichroism and 1D‐NMR spectroscopy show that point mutations that individually replace any of these three residues with leucine were destabilized, but retained the native VHP subdomain fold. In pair‐wise replacements, the double mutant that retains F58 can adopt the native fold, while the two double mutants that lack F58 cannot. The folding of the double mutant that retains F58 demonstrates that aromatic–aromatic interactions within the aromatic cluster are not essential for specifying the VHP subdomain fold. The ability of the VHP subdomain to tolerate mutations within its hydrophobic core indicates that the information specifying the three dimensional structure is distributed throughout the sequence, as observed in larger proteins. Thus, the VHP subdomain is a legitimate model for larger, native proteins.     

N-linked glycosylation of rabies virus glycoprotein. Individual sequons differ in their glycosylation efficiencies and influence on cell surface expression.

Many eukaryotic proteins are modified by N-linked glycosylation, a process in which oligosaccharides are added to asparagine residues in the sequon Asn-X-Ser/Thr. However, not all such sequons are glycosylated. For example, rabies virus glycoprotein (RGP) contains three sequons, only two of which appear to be glycosylated in virions. To examine further the signals in proteins which regulate N-linked core glycosylation, the glycosylation efficiencies of each of the three sequons in the antigenic domain of RGP were compared. For these studies, mutants were generated in which one or more sequons were deleted by site-directed mutagenesis. Core glycosylation of these mutants was studied using two independent systems: 1) in vitro translation in rabbit reticulocyte lysate supplemented with dog pancreatic microsomes, and 2) transfection into glycosylation-deficient Chinese hamster ovary cells. Parallel results were obtained with both systems, demonstrating that the sequon at Asn37 is inefficiently glycosylated, the sequons at Asn247 and Asn319 are efficiently glycosylated, and the glycosylation efficiency of each sequon is not influenced by glycosylation at other sequons in this protein. High levels of cell surface expression of RGP in Chinese hamster ovary cells are seen with any mutant containing an intact sequon at Asn247 or Asn319, whereas low levels of cell surface expression are seen when the sequon at Asn37 is present alone; deletion of all three sequons completely blocks RGP cell surface expression. Thus, although core glycosylation at Asn37 is inefficient, it is still sufficient to support a biological function, cell surface expression. Future studies using mutagenesis of this model protein and its expression in these two well defined systems will aim to begin to unravel the rules governing core glycosylation of glycoproteins.     

Computation of non-covalent interactions in TNF proteins and interleukins

The roles played by the non-covalent interactions have been investigated for a set of six TNF proteins and nine Interleukins. The stabilizing residues have been identified by a consensus approach using the concepts of available surface area, medium and long-range interactions and conservation of amino acid residues. The cation-pi interactions have been computed based on a geometric approach such as distance and energy criteria. We identified an average of 1 energetically significant cation-pi interactions in every 94 residues in TNF proteins and 1 in every 62 residues in Interleukins. In TNF proteins, the cationic groups Lys preferred to be in helix while Arg preferred to be in strand regions while in Interleukins the Arg residues preferred to be in helix and Lys preferred to be in strand regions. From the available surface area calculations, we found that, almost all the cation and pi residues in TNF proteins and Interleukins were either in buried or partially buried regions and none of them in the exposed regions. Medium and long-range interactions were predominant in both TNF proteins and Interleukins. It was observed that the percentage of stabilizing centers were more in TNF proteins as compared to the Interleukins, while the percentage of conserved residues were more in Interleukins than in TNF proteins. In the stabilizing residues Lys was observed to be a stabilizing residue in both TNF proteins and Interleukins. Among the aromatic group, Phe was seen to be a stabilizing residue in both TNF and Interleukins. We suggest that this study on the computation of cation-pi interactions in TNF proteins and Interleukins would be very helpful in further understanding the structure, stability and functional similarity of these proteins.     

Contribution of cation-pi interactions to protein stability

Calculations predict that cation- interactions make an important contribution to protein stability. While there have been some attempts to experimentally measure strengths of cation-pi interactions using peptide model systems, much less experimental data are available for globular proteins. We have attempted to determine the magnitude of cation-pi interactions of Lys with aromatic amino acids in four different proteins (LIVBP, MBP, RBP, and Trx). In each case, Lys was replaced with Gln and Met. In a separate series of experiments, the aromatic amino acid in each cation-pi pair was replaced by Leu. Stabilities of wild-type (WT) and mutant proteins were characterized by both thermal and chemical denaturation. Gln and aromatic --> Leu mutants were consistently less stable than corresponding Met mutants, reflecting the nonisosteric nature of these substitutions. The strength of the cation-pi interaction was assessed by the value of the change in the free energy of unfolding [DeltaDeltaG(degrees) = DeltaG(degrees)(Met) - DeltaG(degrees)(WT)]. This ranged from +1.1 to -1.9 kcal/mol (average value -0.4 kcal/mol) at 298 K and +0.7 to -2.6 kcal/mol (average value -1.1 kcal/mol) at the Tm of each WT. It therefore appears that the strength of cation-pi interactions increases with temperature. In addition, the experimentally measured values are appreciably smaller in magnitude than calculated values with an average difference /DeltaG(degrees)expt - DeltaG(degrees)calc/av of 2.9 kcal/mol. At room temperature, the data indicate that cation-pi interactions are at best weakly stabilizing and in some cases are clearly destabilizing. However, at elevated temperatures, close to typical Tm's, cation-pi interactions are generally stabilizing.     

Glycosylation of the hepatitis C virus envelope protein E1 is dependent on the presence of a downstream sequence on the viral polyprotein

The addition of N-linked oligosaccharides to Asn-X-(Ser/Thr) sites is catalyzed by the oligosaccharyltransferase, an enzyme closely associated with the translocon and generally thought to have access only to nascent chains as they emerge from the ribosome. However, the presence of the sequon does not automatically ensure core glycosylation because many proteins contain sequons that remain either nonglycosylated or glycosylated to a variable extent. In this study, hepatitis C virus (HCV) envelope protein E1 was used as a model to study the efficiency of N-glycosylation. HCV envelope proteins, E1 and E2, were released from a polyprotein precursor after cleavage by host signal peptidase(s). When expressed alone, E1 was not efficiently glycosylated. However, E1 glycosylation was improved when expressed as a polyprotein including full-length or truncated forms of E2. These data indicate that glycosylation of E1 is dependent on the presence of polypeptide sequences located downstream of E1 on HCV polyprotein.     

Characterization of the functional role of the N-glycans in the AMPA receptor ligand-binding domain

The ligand-binding domains of AMPA receptor subunits carry two conserved N-glycosylation sites. In order to gain insight into the functional role of the corresponding N-glycans, we examined how the elimination of glycosylation at these sites (N407 and N414) affects the ligand-binding characteristics, structural stability, cell-surface expression, and channel properties of homomeric GluR-D (GluR4) receptor and its soluble ligand-binding domain (S1S2). GluR-D S1S2 protein expressed as a secreted protein in insect cells was found to be glycosylated at N407 and N414. No major differences in the ligand-binding properties were observed between the 'wild-type' S1S2 and non-glycosylated N407D/N414Q double mutant, or between S1S2 proteins expressed in the presence or absence of tunicamycin, an inhibitor of N-glycosylation. Purified glycosylated and non-glycosylated S1S2 proteins also showed similar thermostabilities as determined by CD spectroscopy. Full-length homomeric GluR-D receptor with N407D/N414Q mutation was expressed on the surface of HEK293 cells like the wild-type GluR-D. In outside-out patches, GluR-D and the N407D/N414Q mutant produced similar rapidly desensitizing current responses to glutamate and AMPA. We therefore report that the two conserved ligand-binding domain glycans do not play any major role in receptor-ligand interactions, do not impart a stabilizing effect on the ligand-binding domain, and are not critical for the formation and surface localization of homomeric GluR-D AMPA receptors in HEK293 cells.     

Role of aromatic localization in the gating process of a potassium channel

Position of the transmembrane aromatic residues of the KirBac1.1 potassium channel shifts from an even distribution in the closed state toward the membrane/solute interface in the open state model. This is the first example of an integral membrane protein making use of the observed preference for transmembrane aromatic residues to reside at the interfaces. The process of aromatic localization is proposed as a means of directing and stabilizing structural changes during conformational transitions within the transmembrane region of integral membrane proteins. All-atom molecular dynamics simulations of the open and closed conformers in a membrane environment have been carried out to take account of the interactions between the aromatic residues and the lipids, which may be involved in the conformational change, e.g., the gating of the channel.