Skip to main content
NCBI home page
Glycobiology logoLink to Glycobiology
. 2020 Dec 9;31(5):649–661. doi: 10.1093/glycob/cwaa112

Contrasting the conformational effects of α-O-GalNAc and α-O-Man glycan protein modifications and their impact on the mucin-like region of alpha-dystroglycan

Andrew Borgert 1, B Lachele Foley 2, David Live 3,✉
PMCID: PMC8176777  PMID: 33295623

Abstract

We have carried out a comparative study of the conformational impact of modifications to threonine residues of either α-O-Man or α-O-GalNAc in the context of a sequence from the mucin-like region of α-dystroglycan. Both such modifications can coexist in this domain of the glycoprotein. Solution NMR experiments and molecular dynamics calculations were employed. Comparing the results for an unmodified peptide Ac- PPTTTTKKP-NH2 sequence from α-dystroglycan, and glycoconjugates with either modification on the Ts, we find that the impact of the α-O-Man modification on the peptide scaffold is quite limited, while that of the α-O-GalNAc is more profound. The results for the α-O-GalNAc glycoconjugate are consistent with what has been seen earlier in other systems. Further examination of the NMR-based structure and the MD results suggest a more extensive network of hydrogen bond interactions within the α-O-GalNAc-threonine residue than has been previously appreciated, which influences the properties of the protein backbone. The conformational effects are relevant to the mechanical properties of α-dystroglycan.

Keywords: α-Dystroglycan, O-glycosylation, hydrogen bonding, mucin glycoprotein

Introduction

The addition of α-O-mannosyl protein modifications, and their associated glycans, to the repertoire of protein glycosylation found in animals, is relatively recent. Investigations of these have emphasized the characterization of the range of primary structures of the carbohydrates elaborated and their novel biosynthetic pathways. Less attention has been focused on the conformational and dynamic impact of these modifications on the proteins in question and the comparison with other modes of O-glycosylation. Most commonly, O-glycosylation has been associated with glycans initiated with α-O-GalNAc, in the class referred to as mucin glycoproteins, characterized by domains rich in serine and threonine residues, where a number of those residues may carry α-O-GalNAc glycans, often with clustered sites of glycosylation. The study of α-dystroglycan (α-DG) has revealed that multiple glycans initiated by both α-O-Man and with α-O-GalNAc can coexist in such domains (Chai et al. 1999). Our goal here is to compare the conformational contributions of these two modes of glycosylation in this context to help illuminate their biological functions.

α-DG was the first example of a glycoprotein from the animal kingdom where modification with α-O-Man glycans was mapped in detail (Harrison et al. 2012; Nilsson et al. 2010; Stalnaker et al. 2010). This glycoprotein is present, among other places, in muscle cells as the extracellular element of the dystrophin-glycoprotein complex that links the cytoskeleton to the extracellular matrix, stabilizing muscle tissue (Barresi and Campbell 2006). It is tightly linked to β-DG, the integral membrane component. The α-O-Man modifications are more abundant in the N-terminal region of its mucin-like domain, while the α-O-GalNAc are more prevalent in its C-terminal section. The discovery of α-O-Man glycans on α-DG arose from the finding that defects in the enzymes responsible for their assembly, rather than mutations in the underlying protein sequence, were responsible for a subset of congenital muscular dystrophies that were also associated with additional neurological consequences (Dobson et al. 2013; Live et al. 2013). Thus, the α-O-Man glycan structures found on α-DG were shown to be key to its biologically critical intermolecular interactions, particularly the structure that is extended with the matriglycan polymer facilitating interaction with the extracellular matrix (Yoshida-Moriguchi and Campbell, 2015).

In contrast with the multiple members of the family of Golgi enzymes that initiate protein α-O-GalNAc modifications, which have been known for some time (Raman et al. 2012), the stepwise biosynthetic pathway of α-O-Man glycan assembly on α-DG was shown to be initiated by the POMT1/POMT2 enzyme complex (Barresi and Campbell, 2006) in the endoplasmic reticulum. It has also been noted that α-O-Man glycans account for about 30% of the O-glycans in the mammalian brain, suggesting that other proteins were similarly modified (Stalnaker et al. 2011). Among others, the cadherin family has emerged as an example (Lommel et al. 2013; Vester-Christensen et al. 2013). Moreover, it has now been shown that there are one and possibly two novel additional biosynthetic processes restricted to particular sets of these other proteins that are associated with initiating this posttranslational modification (Larsen et al. 2019; Larsen et al. 2017a; Larsen et al. 2017b). With the identification of these additional proteins, the biological relevance of α-O-mannosylation as well as its contribution to the conformation of the scaffold has taken on greater significance.

The current insight into the organization of fully processed α-DG comes from low-resolution electron micrographs of the fully processed α-DG (Brancaccio et al. 1995; Kunz et al. 2004), where the N-terminal section appears to take on a globular form, followed by an extended section and then another globular region that is assumed to correspond to the post mucin C-terminal section. The latter region is not modified by O-glycosylation. Given this distribution pattern, the EM images suggest that the α-O-Man glycans, associated with the most N-terminal region (Harrison et al. 2012; Stalnaker et al. 2010), do not dramatically perturb the intrinsic secondary structure preferences of the protein for a globular segment. The organization of this N-terminal segment may be important to the proper elaboration and functionality of the α-O-Man glycans. This is where the specialized extracellular matrix binding structures are located (Hara et al. 2011; Yoshida-Moriguchi et al. 2010).

Based on earlier studies of conformational aspects of α-O-GalNAc modification, these and larger associated α-O-GalNAc glycans have been found to induce a region of extended conformation in mucins (Gerken et al. 1989; Rose et al. 1984; Shogren et al. 1989). This conformational property in the mucin-like region may contribute to the role of α-DG in mediating force transmission to the extracellular matrix and muscle tissue stabilization (Barresi and Campbell, 2006). The α-O-GalNAc glycans also enhance stability by conferring resistance to proteolysis on α-DG (Yu et al. 2018). Interestingly, the role of membrane-bound α-O-GalNAc glycosylated mucin-like proteins has been noted to impact the topology of the cell surface membrane (Shurer et al. 2019). Although this was tested in the context of the conventional α-O-GalNAc modification, α-O-Man would also increase the radius of gyration of the molecule and possibly contribute to molecular crowding and thus influence the local membrane curvature.

Thus, considering the central extended portion in α-DG, the hypothesis is that this would correspond to the region dominated by α-O-GalNAc modifications since these are known to induce such an organization, although there are some α-O-Man glycans interspersed (Stalnaker et al. 2010). While no defects in the α-O-GalNAc modifications on α-DG have been identified or linked to pathologies, their potential for inducing an extended arrangement are likely functionally important in the spatial orientation of the membrane distal region and to the mechanical properties and potential stiffness in transmitting forces between the cytoskeleton of individual muscle cells and the extracellular matrix, contributing to organization and stabilization of muscle tissue. Interestingly, there is an interplay between the presence of the initially installed α-O-Man modifications, and nearby loci where α-O-GalNAc modifications are subsequently initiated (Tran et al. 2012). This suggests that if the level of O-mannosylation is reduced, there could be ectopic α-O-GalNAc modification that could disrupt the conformation of the N-terminus of α-DG that might have a functional impact.

A particular goal here is to understand the relative conformational perturbations of α-O-GalNAc and α-O-Man modifications on the protein backbone in detail on α-DG, in particular, and the effects of O-mannosylation in general. We and others have explored the conformations of various short glycopeptide sequences with α-O-GalNAc modifications (for review see, Martinez-Saez et al. 2017) as well as isolated α-O-Man modifications (Hinou et al. 2019; Mo et al. 2011). Here, in the context of clustered sites of glycosylation, we examined the same sequence from α-DG when it was modified with either α-O-Man or α-O-GalNAc, as well as in the unmodified state. Such clusters are consistent with the glycan mapping of α-DG (Stalnaker et al. 2010).

Solution NMR studies are well suited for extracting conformational and particularly dynamic properties. Quantitative structural studies require homogenous material, however, and glycosylated proteins as derived from natural sources, particularly those with mucin-like domains, present major challenges since there are multiple sites for modification, and often a degree of heterogeneity in site occupancy among the multiple sites, compounded by high molecular weight. These factors can preclude structural analysis, in general, and NMR experiments in particular. In order to overcome these challenges, we relied on well-defined glycopeptide constructs that were chemically synthesized (Liu et al. 2008). In addition to deriving NMR-based structures, the conformational effects of these modes of glycosylation have been explored using unrestrained molecular dynamics simulations, carried out with explicit waters, starting from the NMR derived structures. The computational approach further provides insights into the interactions with associated water molecules and hydrogen bonding.

The comparison here of the properties of glycoconjugates modified with α-O-GalNAc and α-O-Man, respectively, provides further perspective on the effect of the N-acetyl group at the two positions in influencing conformation. Additionally, our efforts extend the evaluation of clustered sites that do occur frequently in mucin domains. The conformational characteristics of these are significant as antibodies can differentiate among clustered motifs as well as between clustered and isolated sites of α-O-GalNAc modifications (Borgert et al. 2012), with evidence that such specific recognition may be useful in therapeutic and diagnostic applications (Matsumoto et al. 2020).

Results

NMR

The sequence, P1PTTTTKKP9, residues 419–427 of α-DG, is examined here in the absence of glycosylation, and with the hydroxyl groups of the threonines modified with either α-O-Man or α-O-GalNAc. This sequence is from a region where sites modified with O-Man had been mapped (Stalnaker et al. 2010). (This is toward the N-terminal of the fully processed α-DG whose N-terminal residue is 317.) It bears similarity to another fragment RIRTTTSGVP (479–488) later in the α-DG sequence where both α-O-Man and α-O-GalNAc modifications coexist and also has some similarity to several other α-O-GalNAc modified cluster structures we have previously examined (Borgert et al. 2012). To better emulate the context in the larger glycoprotein, the synthetic constructs were capped at the N-terminus with an N-acetyl group, and with a carboxamide function at the C-terminus. As noted in the earlier report describing the synthesis of these constructs along with qualitative preliminary analysis (Liu et al. 2008), the dispersion of shifts in the 1D NMR spectrum for the amide protons (Supplementary Table SI), whether of the unmodified peptide or with the cluster of successive threonine sites modified with an α-O-Man residue, is rather modest relative to that of the signals in the presence of α-O-GalNAc modifications. The NMR dispersion of amide protons is a good qualitative indicator of peptide secondary structural characteristics. To develop a quantitative description of the conformations of these constructs, more extensive NMR data have been obtained and analyzed, and molecular dynamics simulations carried out for a detailed comparison of the three constructs.

Insight into the conformations can be extracted from the vicinal (3J) coupling constants accessible from the 1D 1H NMR spectra (Table I), which are sensitive to the bond angles (Wuthrich, 1986). In the region of the glycosylated residues, the backbone coupling between Hα and NH are larger when the modification is α-O-GalNAc than when the modification is α-O-Man, or is absent. This implies more limited mobility for the backbone of the α-O-GalNAc construct, and further, the values are consistent with a more extended arrangement (Borgert et al. 2012; Coltart et al. 2002; Hayakawa et al. 2020). Another feature noted with the α-O-GalNAc modification, and also seen in other cases of α-O-GalNAc modified threonines, is a small value for the coupling between the Hα and Hβ of these residues (Bermejo et al. 2018; Borgert et al. 2012; Coltart et al. 2002; Hayakawa et al. 2020). This is consistent with a limited range of torsion angles of the Hβ-Cβ-Cα-Hα bonds (χ1 angle) approaching 90o at which this value has a minimum. In contrast, the larger values for the α-O-Man modified residues indicate averaging among rotamers. A computational study of O-glycosylated serine and threonine residues indicated an angle ~ 60° for the side chain of α-O-GalNAc-T, but suggested as well that this was the case for α-O-Man-T (Mallajosyula and MacKerell, 2011). The NMR coupling data reported here do not support this conclusion for the latter case.

Table I.

Vicinal coupling constants for the peptide backbone and threonine residue side chainsa

  Peptide at 25 °C Mannose at 5 °C GalNAc at 15 °C
HN-Hα
 T3 7.59 8.34 8.69
 T4 7.78 8.42 9.61
 T5 7.6 8.33 9.52
 T6 7.5 8.06 9.25
 K7 7.05 6.69 6.69
 K8 6.96 6.14 4.57
Hα-Hβ
 T3 4.51 4.06 2.35
 T4 4.65 4.02 1.72
 T5 4.55 3.87 1.42b
 T6 5.09 4.26 1.57
Open in a new tab

aTemperatures chosen to reduce peak overlap. HN couplings in 9/1 H2O/D2O, Hβ couplings in D2O. Other experimental details are given in Methods section.

bCould not be completely resolved due to overlap with solvent peak.

Additional information supporting this analysis is available from the 2D 1H NOESY data on the distance-dependent NOE (nuclear Overhauser enhancement) (Wuthrich 1986) interactions between the threonine Hβ protons and the peptide backbone amide protons at i or i + 1 position. There are distinct differences among the NOE patterns for the cases examined, as seen in the example of the traces from the 2D NOESY spectrum at the position of the respective T6β protons (Figure 1). The significant NOE interactions between both the relevant i and i + 1 amide protons and the T6β proton when the threonine side chain is unmodified or modified with α-O-Man are consistent with the side chain accessing several rotamer states. Analysis by NMR of threonine residues in denatured proteins shows a mixture of rotamer populations (Hennig et al. 1999; Vajpai et al. 2010). Data from mannosylated threonine residues in an X-ray crystal structure of Aspergillus awamori indicate that mannosylated threonine residues can adopt several possible χ1 angles (Aleshin et al. 1994). In the α-O-GalNAc form, the NOE interaction is almost exclusively to the i + 1 amide. The substantially larger Hβ-NH(i + 1) NOE of the α-O-GalNAc modified residue, along with the more extended peptide backbone and the extremely low value of the 3JHα-Hβ coupling constant is consistent with restriction for the side chain angle.

Fig. 1.

Fig. 1

Open in a new tab

Variation in the pattern of the NOE interactions between the T6β protons of (A) the peptide (B), the tetra α-O-Man form, and (C) the tetra α-O-GalNAc and the amide protons on their own residue and on the next one, indicating differences in side-chain rotamer state. Traces are at the position of T6β from the respective NOESY spectra in 90% H2O/10% D2O.

An extensive set of the vicinal couplings and additional 1H NOEs from 2D NOESY spectra, with Figure 2 showing a region with NOEs between amino acid α and β protons and amide protons, was obtained and analyzed for each of the three constructs as constraints for comprehensive structure calculations (Table II) using Xplor-NIH (Schwieters et al. 2003). Overall, the number of NOEs that could be assigned was greater for the α-O-GalNAc modified glycopeptide than for the α-O-Man modified glycopeptide or the peptide itself. In the case of the α-O-Man form, overlap of the signals from the Man residues, even of the anomeric sites which tend to have greater dispersion, limits the number of NOE assignments that could be made. This is also noted for the Hα signals. The overlap and fewer NOE cross-peaks, as well as the 3J couplings, are consistent with a more fluxional state for both the α-O-Man glycopeptide and the peptide relative to the α-O-GalNAc modified version. The RMSD values for all heavy atoms within the acceptable structures from the NMR refinement were 3.22 Å for the peptide, 4.18 Å for the α-O-Man construct but 1.79 Å for the α-O-GalNAc glycopeptide. These values suggest conformationally more diffuse arrangements for the peptide and the α-O-Man form relative to the α-O-GalNAc. Coordinates of the well-defined family of structures obtained for the α-O-GalNAc form have been deposited in the PDB under entry 2MK7 along with a list of NMR constraints used. The backbone Φ/Ψ plots for the structures are in Supplementary Figure S4. The difference in the degree of organization of the structures between α-O-Man and α-O-GalNAc forms is visualized in the superposition of the glycosylated core of the 16 best structures of each of these (Figure 3). Subsequent to this analysis, residual dipolar couplings (RDCs) in weakly aligning media (Ottiger and Bax, 1999) were measured for both glycopeptides and were incorporated in refinements. These provide additional independent constraints. With their inclusion, for the construct with the α-O-GalNAc modification, the RMSD for all heavy atoms in acceptable structures decreased to 1.29 Å, while for the α-O-Man construct, the RMSD actually increased to 4.34 Å. The trends in results after incorporating the RDCs in the refinement are consistent with the interpretation of a well-defined structure for the α-O-GalNAc form, but a more flexible behavior for the α-O-Man form.

Fig. 2.

Fig. 2

Open in a new tab

Sections of the amide to aliphatic region of the NOESY spectra in 90% H2O/10% D2O of the O-GalNAc (A) and O-Man (B) glycopeptides showing interactions of peptide amide protons with sugar anomeric and amino acid α and β protons. Assignments of amide protons are given within the plot and those of the aliphatic protons on the left. Chemical shifts are relative to DSS.

Table II.

Constraints for structure calculations and characteristics of final structures

Distance Restraints Peptide O-Man glycopep-tide O-GalNAc glycopep-tide
Peptide–Peptide
 Intra-residue 46 52 52
 Inter-residue 16 22 26
Peptide-Sugar
 Proximal 21 51
 Non-Proximal 5 25
 Sugar-Sugar 4 26
Total 62 104 182
3 J coupling restraints 6 6 6
Torsion restraints 6 0 0
Restraint Violations
 NOE Violations (>.25 Å) 0 0 0
 J-Coup Violations (>.5 Hz) 0 0 0
 Torsion Violations (>5°) 0 0 0
Deviations From Ideal Geometry
 Bond (Å): 0.00246 +/− 0.00008 0.0020+/− 0.00003 0.002+/−0.0002
 Angle (°): 0.54 +/−0.02 0.44+/−0.003 0.45+/−0.02
 Improper (°) 0.39+/−0.02 0.38+/−0.007 0.39+/−0.02
Average RMSD 48 structures 36 structures 16 structures
 Backbone Heavy Atoms (Å) 2.75+/−0.44 4.02+/−0.67 1.19+/−0.31
 All Heavy Atoms (Å) 3.22+/−0.37 4.18+/−0.62 1.79+/−0.44
Open in a new tab

Fig. 3.

Fig. 3

Open in a new tab

Superposition of the core glycosylated residues of the 16 best-computed structures (PDB 2MK7) of the α-O-GalNAc in teal (A) and α-O-Man in red (B) glycopeptides illustrating the comparative degree of ordering of the two glycoconjugates.

In the case of the flexible peptide and α-O-Man glycopeptide, the restraints used in the Xplor structure analysis are necessarily time averages reflecting the properties of the different interactions. While the structure solutions for the flexible cases may not describe those of the actual individual conformers making up the ensemble, if the region of conformational space is not too large, these should capture their general features. Molecular dynamics simulations starting from these structures allow us to better assess their properties.

Molecular dynamics

To further explore the dynamic properties of these constructs, as well as provide additional validation that the relative degree of conformation flexibility concluded from the NMR analysis was not biased by the differences in the number of NMR constraints among the molecules studied, and that the structural characteristics deduced for the peptides and α-O-Man glycopeptide are reasonable, multiple unrestrained molecular dynamics simulations were undertaken starting from each of the accepted NMR structures of the respective constructs. For the MD, the structures in each instance were neutralized with chloride ions (Joung and Cheatham, 2008; Joung and Cheatham, 2009) and solvated with TIP3P (Jorgensen et al. 1983) waters in silico. Since the number of accepted structures for the α-O-GalNAc form was significantly lower than the others, two MD runs were carried out for each of these structures to insure comparable sampling of the MD trajectories among the structures. The length of the simulations in each instance was approximately 100 ns with the sum of the times for each construct being 3 μs or more.

The distribution of distances across the four contiguous threonine residues extracted from the frames in the MD provides a measure of how fluctuations of the peptide backbone conformation respond to the presence and nature of carbohydrates linked to the threonine residues. The distributions are plotted as a histogram of the peptide backbone N(T3)-N(K7) distances across the region (Figure 4). The construct modified with α-O-GalNAc residues shows a narrow distribution of distances consistent with a more extended stable structure. In contrast, the α-O-Man modified peptide and the unmodified peptide have considerably broader distributions with, on average, less extended structures, although the α-O-Man construct distribution is skewed toward the longer distances in its range. The relative stability of these conformations is reflected in the RMS deviations within each of the trajectories (Figure 5), which indicates the sampling of a greater conformational range for the peptide and α-O-Man glycopeptide than for the α-O-GalNAc glycopeptide, consistent with our interpretation of the NMR data. Since there are no NMR constraints from couplings directly reporting on the glycosidic torsion angles, but only more remote sugar to peptide NOE constraints, the MD simulations can provide an independent test of whether the glycosidic angles of the NMR derived structures are those energetically preferred, and the extent to which the differences in the dynamic properties of the peptide scaffold perturb the glycosidic linkages between the α-O-Man and α-O-GalNAc constructs.

Fig. 4.

Fig. 4

Open in a new tab

Histogram of the distributions of distances across the for threonine from N(T3)-N(K7) for the peptide (blue), the construct with α-O-Man modified threonines (green), and the construct with α-O-GalNAc modified threonines (red).

Fig. 5.

Fig. 5

Open in a new tab

Average RMSD for all the backbone atoms in the threonine core of Ac-PPTTTTKKP-NH2 across each MD simulation for the unmodified peptide (blue), the α-O-Man modified glycopeptide (green), and the α-O-GalNAc modified glycopeptide (red). Because of the lower number of starting structures for the latter, two runs were initiated from each initial structure.

The range of glycosidic angles sampled in the MD trajectories is similar for both types of sugar residues as seen in Figure 6 for residue T5 with those for all residues in Supplementary Figures S1 and S2, with a slight increase in flexibility observed for the α-O-Man moiety. The NMR results (red triangles in Figure 6) are consistent with those found in the MD. For both sugar residues, these bond angles fall in the normal range for glycosidic bonds. The presence of the N-acetyl on the GalNAc does not appear to be a dominant factor in the glycosidic angles. Time-lapse videos of the motions of an α-O-GalNAc-T and an α-O-Man-T at position T5 are shown in Figure 7.

Fig. 6.

Fig. 6

Open in a new tab

Glycosidic angle distribution from the MD simulations (gray heat map) and NMR structures (red triangles) for the carbohydrate residues on T5 of the α-O-GalNAc (A) and α-O-Man (B) glycoconjugate peptides. The darker the gray in each of the 1° square bins in the heat map, the greater the relative populations of the corresponding ϕ/ψ coordinates, taken across all the simulations. The MD data represent instantaneous values for the angles, while the NMR angles represent their time-averaged values.

As is seen in the NMR structures and the MD, for an α-linked O-GalNAc there is a parallel relationship of the bonds between the C1 to glycosidic oxygen and the N-acetyl NH bond, since the latter is effectively fixed trans to the C2H bond as indicated by the large HN to HC2 coupling (Coltart et al. 2002; Davis et al. 1994; Hayakawa et al. 2020), and is consistent with other MD results (Mallajosyula and MacKerell, 2011). This sets up a favorable interaction between this proton and the glycosidic oxygen. With the angles adopted by the glycosidic linkage and the very limited range for the χ1 angles found for the modified threonine, the orientation of the α-O-GalNAc relative to the peptide backbone is constrained so that the N-acetyl NH is further predisposed for a favorable interaction with its backbone carbonyl. Additionally, the Φ and Ψ angles in the peptide backbone are such that the peptide and N-acetyl amide protons are at the peaks of two triangles that share as their base the line between the glycosidic and carbonyl oxygens of the modified threonine residue, suggesting two bifurcated interactions of hydrogen bond character that share the acceptors (Figure 8) and act synergistically to stabilize the structure. This kind of configuration for the α-O-GalNAc construct is seen consistently over the course of the MD runs as the dominant configuration, being present in the NMR structure of the α-O-GalNAc construct, with an example from one residue in the time-lapse video (Figure 7).

Fig. 8.

Fig. 8

Open in a new tab

Relationships between the GalNAc NH, the glycosidic oxygen, the C=O for the threonine, and the NH of the threonine for T6 in the α-O-GalNAc modified peptide from the NMR structure closest to the average.

In light of earlier indications of water molecules associated with α-O-GalNAc threonine in MD simulations of the capped α-O-GalNAc threonine (Bermejo et al. 2018; Corzana et al. 2007; Mallajosyula and MacKerell, 2011), or as a residue at an isolated glycosylation site in a peptide (Madariaga et al. 2014), we examined this aspect in the example of a cluster of α-GalNAc threonines studied here. The frames for the MD of our α-GalNAc construct during the latter part of the time allotted for the equilibration phase were analyzed, after the equilibration was largely complete. The waters were saved in the frames of this period, but not for the frames in the production run (see Materials and Methods section). Unlike the persistent intramolecular interactions described above, the presence of water molecules within interacting range is a more transient feature with fractional occupancy, on average ~20% for the four residues, of a triangular arrangement where the oxygen of a water interacts with the amide protons of both the N-acetyl group and the backbone amide of the threonine residue. A fractional occupancy, of ~20%, was also noted in a computational study by others for α-O-GalNAc threonine (Mallajosyula and MacKerell, 2011).

Discussion

The finding that the same mucin-like protein domain can display both α-O-Man glycans and α-O-GalNAc modifications motivated our interest in further understanding the relative conformational impact of these modifications. The solution conformational properties, determined by NMR methods, of the peptide with a sequence from α-DG, PPTTTTKKP, as well as its glycoconjugate where all the four T residues were modified with α-O-Man, found the latter to be quite flexible, similar to the unmodified peptide, even though the presence of clustered α-O-Man sites should accentuate any effect of the α-O-Man modification.

The limited conformational influence of the α-O-Man modifications on the scaffold to which it is attached, as described here and noted in NMR studies by us and others in several contexts (Hinou et al. 2019; Mo et al. 2011), implies that the α-O-Man modification has a limited impact on the native secondary structure tendencies of the protein sequence. The pattern of NOE interactions between the Man and the threonine residue to which it is attached, that we observe, is similar to those reported by Hinou et al. (2019), but, in our case, some are obscured by the inability to resolve resonances from H5 and H3 protons among the four individual Man residues in the construct. In considering the shift of the amide backbone resonances as a qualitative indicator of conformational change, comparing the peptide and the α-O-Man glycopeptide we observe (Supplementary Table SI) shifts between −0.03 ppm and +0.13 ppm for the amino acids with amide protons, TTTTKK, which are larger than what Hinou et al. observed in the immediate vicinity of their modification. We attribute this to the magnified steric effects of having four sequential α-O-Man modifications. In data we reported earlier for an α-DG glycopeptide with a larger glycan attached (Mo et al. 2011), interrogating the same sequence region, GAIIQTPTLG, as Hinou et al. (2019), having in our case an O-mannosyl trisaccharide on the first rather than the second threonine in the TPT motif, the amides of the modified residue and the leucine, +3 to the modification, both shifted about +0.06 ppm, while the glutamine in the −1 position shifted −0.03 ppm, with rather little perturbation on the others. For the constructs, Hinou et al. studied the most notable shift for the trisaccharide construct was a shift of -0.25 ppm at the leucine, which is +1 to the modified threonine in their glycopeptide, suggesting some difference in accommodation of the backbone depending on the site of modifications. Consistent with our conclusions, all the constructs reported by Hinou et al. show substantial mobility for the O-mannosyl residue. Overall, the results of us and others suggest a range of secondary structural elements accessible to α-O-Man modified peptides.

In the apparent absence of evidence for the specific interaction between the α-O-Man and the peptide, the backbone carbonyl, and amide proton functionalities are available for more typical interactions with other elements of the protein sequence in the formation of protein secondary and tertiary structural elements. This is consistent with the general appearance of the N-terminus of the fully processed α-DG apparently displaying globally globular features (Brancaccio et al. 1995; Kunz et al. 2004). This is the region where the α-O-Man glycan modifications predominate (Stalnaker et al. 2010). A crystal structure of a fungal glucoamylase from Aspergillus awamori has 10 α-O-Man modifications spread over 21 residues in a strand of an irregular structure at the C-terminus of the construct used, with limited direct contacts between the α-O-Man and the protein backbone (Aleshin et al. 1994). Several structures of molecules in the cadherin family have been reported e.g. E-cadherin EC1–5 (Harrison et al. 2011) and protocadherin γA8 EC1–3 (Rubinstein et al. 2015)), where α-O-Man modifications are found in beta-sheet strands or in loops. Together this indicates that α-O-Man modified elements can be present in various structural motifs. As more α-O-Man modified proteins are being identified and structures are examined, a better appreciation for the range of conformational elements compatible with α-O-Man modifications will be clarified. In cases where the α-O-Man is extended with a β-1,2 GlcNAc substitution, the most common extension of α-O-Man glycans, steric effects have been invoked to explain conformational modulation in the peptide backbone (Hinou et al. 2019). The qualitative results we reported on a glycopeptide with the rarer M3 glycan core where, in contrast, the GlcNAc is linked in the less common β-1,4 to the α-O-Man, further from the glycosidic linkage of the latter, did not suggest significant backbone effects (Mo et al. 2011). This supports the conclusion that the steric interactions specific to the β-1,2 GlcNAc linkage are responsible for the perturbation of the backbone with that extension. It is possible that the conformational plasticity of the scaffold may play a role in whether the β-1,2 extension takes place, possibly explaining why the α-O-Man modifications on cadherins are not extended (Larsen et al. 2019).

The α-O-GalNAc, by contrast, has significant interactions with the protein scaffold between the proton of the N-acetyl group and the threonine C=O group, as has been noted here and in earlier work (Coltart et al. 2002). The disposition of this proton also makes it amenable to an interaction with the O atom of the glycosidic linkage. A hydrogen bond involving the glycosidic oxygen and the backbone amide proton has been proposed in computational studies, particularly for glycosylated threonine residues (Mallajosyula and MacKerell, 2011). A more extensive hydrogen-bonding network is implied by the orientations of the N-acetyl proton, and the backbone amide proton of the modified threonine residue, relative to the glycosidic, and threonine C=O oxygen atoms in the structure here. The configuration of these two hydrogen and two oxygen atoms as a group persists in our MD of the modified threonine residues as discussed above and illustrated in Figures 7 and 9. The significant downfield shifts of the peptide amide protons in the GalNAc construct relative to that in the O-Man construct or unmodified peptide (Supplementary Table SI) are consistent with these backbone amide protons participating in enhanced hydrogen-bonding interactions in the α-O-GalNAc glycopeptide. Thus, this network can be viewed as comprised of two bifurcated hydrogen bonds, with two hydrogen atom donors sharing the two common oxygen acceptors.

Fig. 7.

Fig. 7

Open in a new tab

Time-lapse video of MD snapshots (view in full in Supplementary Material) showing the local dynamics of the GalNAc-T (A) and Man-T (B) residues for position T5 in the respective glycopeptides. Each trajectory was aligned on the C, N, O, and CA atoms of the T.

Fig. 9.

Fig. 9

Open in a new tab

(A) Fraction for each GalNAc threonine where the backbone oxygen or the glycosidic oxygen was in a geometry consistent with a bifurcated hydrogen bond as observed in the MD runs and following the criteria for hydrogen bonding in McDonald and Thornton (1994). The glycosidic oxygen participates in a bifurcated hydrogen bond consistently throughout. (B) Distribution of time that the backbone oxygen participates in either a bifurcated hydrogen bond, a single hydrogen bond to only one of the protons, or not interacting. Distance distributions from which these were derived are in Supplementary Figure S5.

This network persistence is quantified in the bar graph in Figure 9, based on hydrogen bond criteria from McDonald and Thornton (1994), discussed below. The density plots from which these values are derived are in Supplementary Figure S5. The proportion of time, the bifurcated hydrogen bond associated with the glycosidic oxygen acceptor persists is almost 100%. For the backbone oxygen, the bifurcated hydrogen bond occurs between 50% and 80% of the time depending on the residue. In the latter case, it should be noted that even though both hydrogen bonds to the backbone oxygen are simultaneously present the majority of the time, during a significant portion of the remainder, the fluctuations in the atomic positions of these atoms in the MD still allow for the participation of one or the other of these partners in an individual H bond to the oxygen.

In considering this hypothesis, operational classification for the arrangement of donor and acceptor atoms that correspond to hydrogen bonds in proteins has been developed-based largely on analysis of protein structure data (McDonald and Thornton, 1994). Criteria for generally suggesting a single hydrogen bond have the distance between the acceptor (A) and the donor (D) to which the proton is attached of up to 3.9 Å, corresponding to H-A distance of about 2.5 Å, and the D-H-A angle is between 90o and 180o. Subpopulations with lower angle and greater length of H to A distance of 3.9 Å are also considered to be hydrogen bonding interactions that can satisfy donors and acceptors (McDonald and Thornton, 1994). The geometries that are present in the interactions of the GalNAc amide proton, the glycosidic oxygen, the threonine amide, and its carbonyl fall within the range that are consistent with hydrogen bonding-like geometries as found by McDonald and Thornton in analyzing Protein Data Bank structures (McDonald and Thornton, 1994). Representative angles for the interactions from the structure here closest to the average among all the α-O-GalNAc modified T residues are shown in Table III and are consistent with their broader definition of donor-hydrogen-acceptor angles that can approach 60o, and proton-acceptor distances reach 3.9 Å. A number of examples also exist where a donor may be shared between more than one acceptor. In these instances, it has been noted the geometries are less ideal and can show longer hydrogen to oxygen distances (McDonald and Thornton, 1994). The energetics are less than additive in such shared H-bonds. While some of the geometrical parameters of the network of two bifurcated hydrogen bonds indicated here are a bit beyond those found for single bonds, these distortions are consistent with expectations in such situations. This has been rationalized by noting that multiple hydrogen bonds to an O atom point to the lobes of the oxygen lone pair orbitals, whereas for an O accepting a single hydrogen bond it is directed between these lobes (Baker and Hubbard, 1984). Networks with generally similar geometry to what we propose here have been noted in protein secondary structural elements (Baker and Hubbard, 1984). The larger network we propose is consistent with these parameters and would serve to stabilize both the Φ and Ψ angles of the modified residue along with the χ1 angles and provide a robust network for inducing the extended geometry of the scaffold associated with α-O-GalNAc modifications. The disposition of the two hydrogens and two oxygens are comparable to other structures with this modification that we have solved (Borgert et al. 2012; Coltart et al. 2002). Earlier strictly computational analysis of α-O-GalNAc modified serine and threonine residues proposed an interaction between the backbone amide and carbonyl groups, but only in the case of modified α-O-GalNAc serine residues (Mallajosyula and MacKerell 2011). In contrast, here it is noted for threonine residues.

Table III.

Distances and angles for the relationships of the GalNAc NH, glycosidic oxygen, threonine NH, and threonine C=O atoms from the NMR structure closest to the average

  GalNAc NH to glycosidic O dist Å GalNAc NH to glycosidic O angle GalNAc NH to Thr O dist Å GalNAc NH to Thr O angle Thr NH to glycosidic O dist Å Thr NH to glycosidic O angle Thr NH to Thr O dist Å Thr NH to Thr O angle
T3 2.61 91.3 3.04 103.5 2.52 97.8 2.95 66.0
T4 2.61 91.5 2.73 101.8 2.57 93.9 2.83 92.8
T5 2.61 91.5 2.72 101.7 2.65 94.8 2.83 75.7
T6 2.61 91.3 2.73 95.6 2.56 96.6 2.88 74.8
Open in a new tab

Studies of an isolated amino and carboxyl capped α-O-GalNAc-threonine in vacuo, with computational modeling for the effects of the addition of hydration, demonstrated the impact of full solvation on the disposition of the GalNAc and the side-chain methyl group (Bermejo et al. 2018) in a conformational arrangement analogous to what we have found. The network of hydrogen bonds proposed here provides a more extensive picture of how the impact of the orientation of the GalNAc and its interactions propagate to the N−H and C=O of the peptide backbone. In addition, we and others have also pointed to how interactions between functionalities both on the GalNAc residues and moieties on neighboring residues can also influence the longer range organization of the peptide backbone scaffold (Coltart et al. 2002; Hayakawa et al. 2020). Together, these provide a better picture of the mechanism behind how the α-O-GalNAc modification stabilizes extended glycoprotein segments.

The features of the construct with α-O-GalNAc described here are similar to systems with clustered sites that we have studied earlier (Borgert et al. 2012; Coltart et al. 2002). When aligning the heavy atoms in a cluster of three glycosylated residues of a MUC2-like fragment where three successive Ts are modified with α-O-GalNAc (Borgert et al. 2012), PDB 2LHW, with either residues 3,4, and 5, or 4,5, and 6 of the GalNAc modified construct here, the fits to these α-O-GalNAc-Ts in question are 0.65 Å, and 0.71 Å, respectively, suggesting a more general structural motif for the clusters. The intramolecular interactions associated specifically with the α-O-GalNAc modifications, as further elucidated in this work, enhance the understanding of the structural impact of these, both at the level of the modified amino acid and on the neighboring protein scaffold.

Materials and methods

NMR

Syntheses of the peptide and glycopeptides were based on the sequence Ac-PPTTTTKKP-NH2 and were reported previously (Liu et al. 2008). The N and C capping better emulate the environment of the native glycoprotein. NMR spectra were obtained at 600 MHz or 800 MHz on Varian Inova spectrometers. Samples were in D2O or 90%H2O/D2O solutions at ~pH 4.5, and a concentration of 2–10 mM. In addition to one-dimensional 1H spectra, two-dimensional NOESY, COSY, TOCSY, and 1H-13C or 1H-15N HSQC experiments (van de Ven, 1995) were collected for assignment and structure determination. Water suppression was achieved using the WATERGATE method (Piotto et al. 1992) or gradient coherence selection methods in the HSQC experiment (Kay et al. 1992). RDCs were measured in didodecyl-phosphatidylcholine/dihexylphosphatidylcholine 3/1 molar ratio at 10% in 90% H2O/10% D2O (Ottiger and Bax, 1999) at in the range of 30–35 °C using 1H-13C or 1H-15N HSQC sequences without 1H decoupling pulses in the heteronuclear evolution period, and couplings determined from splittings in the heteronuclear dimension. Data were processed with the Varian instrument software or the nmrPipe (Delaglio et al. 1995). Chemical shifts are given relative to DSS (4,4-dimethyl-4-silapentane-1-sulfonic acid).

Starting from an extended structure, peptide backbone and glycosidic linkage torsion angles were randomized and the structure was then subjected to an initial minimization consisting of four iterations of 2000 steps each. The first iteration included only bond length and angle terms, the second iteration added torsion terms, the third added improper torsion terms, and the fourth iteration added a repulsive van der Waals term. After initial minimization, the van der Waals term was turned off, and NMR derived distance and torsion restraints were activated. The system was then warmed to 50,000 K over 2 ps, which was followed first by 30 ps of dynamic, and then 32 ps of dynamics with ramped repulsive van der Waals terms. This was followed by 16 ps of slow cooling to 0 K and 2000 steps of final Cartesian minimization. Electrostatic terms were turned off for the duration of the calculations. All simulated annealing dynamics stages utilized a time-step of 2 fs. An ensemble of 50 structures was generated for each variant, which were then evaluated for restraint violations and deviations from ideal geometry based on the following criteria: no distance restraint violations greater than 0.25 Å, no backbone torsion restraint violations greater than 5.0°, no bond length constraint violations greater than 0.05 Å and no angle or improper torsion constraint violations greater than 5.0°. All structure determination calculations utilized Torsion Angle Dynamics (Stein et al. 1997) along with the Internal Variable Module (Schwieters et al. 2003). Throughout the calculations, proline, mannose, and GalNAc rings were held rigid, with the mannose and GalNAc rings held in the chair conformation.

Molecular dynamics

Detailed information regarding the MD simulations can be found in the supplementary data. Only a brief overview is provided here. The atomic coordinates for the NMR structures were manually converted into a format amenable to preparation for simulation (see SD). The coordinates were loaded into the TLEAP utility in AmberTools 12 (Case et al. 2012). The GLYCAM06 (Kirschner et al. 2008) and FF12SB (Cheatham et al. 1999, Perez et al. 2007, Zgarbova et al. 2011) force fields were used to model the carbohydrate and protein, respectively. Chloride ions (Cl-) (Joung and Cheatham, 2008, Joung and Cheatham, 2009) were added to neutralize the charged portions of the peptide. TIP3P (Jorgensen et al. 1983) water molecules were added so that the system formed a cube with the minimum distance from the edge of the (glyco-)peptide moiety to the surface of the box being 10 Å.

Each system was prepared for simulation by minimizing, heating, and equilibrating before being run in production, using AMBER’s (Case et al. 2012) PMEMD module, for at least 50 million steps with a step size of 2 fs (at least 100 ns of simulated time per simulation). The lengths of bonds including hydrogens were restrained using the SHAKE algorithm (Miyamoto and Kollman, 1992, Ryckaert et al. 1977) to allow for use of the 2 fs time-step. Since the α-O-GalNAc glycopeptide was assigned only 16 initial NMR structures, each of its initial structures was run twice, with differing conformational spaces assured by using different random number seeds during heating. Heating (Berendsen et al. 1984) was performed at constant volume and temperature (NVT). Equilibration and production were performed at constant pressure and temperature (NPT). Heating, equilibration, and production were performed under periodic boundary conditions, treating electrostatics using an Ewald summation (Crowley et al. 1997; Darden et al. 1993; Essmann et al. 1995; Sagui and Darden, 1999). During equilibration, coordinates for all atoms were saved every 1000 simulated steps. During production, coordinates for only the (glyco-)peptide moiety, excluding waters and Cl- ions, were saved every 5000 simulated steps. For reasons described in the SD, some saved frames had to be discarded, and some runs exceeded 50 million steps, but the total analysis comprised at least 3 μs for each moiety. Table IV provides details.

Table IV.

Summary of molecular dynamics data collected

Threonine modification Number of simulations Equilibration framesa Production framesb
α-O-GalNAc 32 278213 (0.56 μs) 317739 (3.18 μs)
α-O-Man 36 c 355642 (3.56 μs)
Unmodified 48 c 520407 (5.20 μs)
Open in a new tab

Coordinates for the waters and counterions (a) were or (b) were not saved in each frame. (c) Equivalent data for these are available but were not analyzed except for the usual diagnostic purposes of equilibration.

The MD study was originally undertaken to support and validate the NMR-derived structures; so for efficiency, water molecules and counterions were not been saved from the main trajectories, since these were not needed for comparison with the structures generated by the NMR experiments. However, in view of the reports on interactions between surrounding water molecules and the α-O-GalNAc threonine identified in MD, we revisited the MD data saved from the equilibration runs where coordinates for all components of the system were saved from those trajectories.

The systems equilibrated early (Supplementary Figure S3), leaving a total of a little over 500 ns of trajectory (Table IV) available for analysis. The frames were checked for the existence of water’s oxygen in a hydrogen-bonding configuration with both the NAc hydrogen (Hg) and the backbone hydrogen (Hb). A hydrogen bond was considered to be present if the water’s oxygen (Ow) was less than 2.5 Å away from both Hg and Hb and if both angles were made between Ow, each hydrogen and the hydrogen’s nitrogen were at least 100 degrees. While not as comprehensive as the structural analysis, they provide an indication of the water interaction with the α-O-GalNAc-modified peptide. Further details are included in the SD.

Supplementary Material

SD_rev_cwaa112
Click here for additional data file. (1.3MB, pdf)
TGalNAc5_cwaa112
Click here for additional data file. (15MB, mp4)
TMan5_cwaa112
Click here for additional data file. (7.8MB, mp4)

Contributor Information

Andrew Borgert, Department of Medical Research, Gundersen Health System, 1900 South Ave., La Crosse, WI 54601, USA.

B Lachele Foley, Complex Carbohydrate Research Center, University of Georgia, Athens, GA 30602, USA.

David Live, Complex Carbohydrate Research Center, University of Georgia, Athens, GA 30602, USA.

Funding

This work was supported by National Institutes of Health grants RO1-GM111939, RO1-GM066148, and P41-GM103390.

Conflict of interest statement

The authors declare no conflict of interest.

Abbreviations

NMR nuclear magnetic resonance.

2D NOESY 2 dimensional nuclear Overhauser effect spectroscopy.

TOCSY total correlation spectroscopy.

HSQC heteronuclear single-quantum coherence.

COSY correlation spectroscopy.

PDB Protein Data Bank.

MD molecular dynamics.

References

  1. Aleshin  AE, Hoffman  C, Firsov  LM, Honzatko  RB. 1994. Refined crystal-structures of glucoamylase from aspergillus-awamori var x100. J Mol Biol. 238:575–591. [DOI] [PubMed] [Google Scholar]
  2. Baker  EN, Hubbard  RE. 1984. Hydrogen-bonding in globular-proteins. Prog Biophys Mol Biol. 44:97–179. [DOI] [PubMed] [Google Scholar]
  3. Barresi  R, Campbell  KP. 2006. Dystroglycan: From biosynthesis to pathogenesis of human disease. J Cell Sci. 119:199–207. [DOI] [PubMed] [Google Scholar]
  4. Berendsen  HJC, Postma  JPM, Vangunsteren  WF, Dinola  A, Haak  JR. 1984. Molecular-dynamics with coupling to an external bath. J Chem Phys. 81:3684–3690. [Google Scholar]
  5. Bermejo  IA, Usabiaga  I, Companon  I, Castro-Lopez  J, Insausti  A, Fernandez  JA, Avenoza  A, Busto  JH, Jimenez-Barbero  J, Asensio  JL  et al.  2018. Water sculpts the distinctive shapes and dynamics of the tumor-associated carbohydrate tn antigens: Implications for their molecular recognition. J Am Chem Soc. 140:9952–9960. [DOI] [PubMed] [Google Scholar]
  6. Borgert  A, Heimburg-Molinaro  J, Song  XZ, Lasanajak  Y, Ju  TZ, Liu  M, Thompson  P, Ragupathi  G, Barany  G, Smith  DF  et al.  2012. Deciphering structural elements of mucin glycoprotein recognition. ACS Chem Biol. 7:1031–1039. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Brancaccio  A, Schulthess  T, Gesemann  M, Engel  J. 1995. Electron-microscopic evidence for a mucin-like region in chick muscle alpha-dystroglycan. FEBS Lett. 368:139–142. [DOI] [PubMed] [Google Scholar]
  8. Case  DA, Darden  TA, Cheatham  ITE, Simmerling  CL, Wang  J, Duke  RE, Luo  R, Walker  RC, Zhang  W, Merz  KM  et al.  2012. Amber 12. San Francisco, CA: University of California. [Google Scholar]
  9. Chai  WG, Yuen  CT, Kogelberg  H, Carruthers  RA, Margolis  RU, Feizi  T, Lawson  AM. 1999. High prevalence of 2-mono- and 2,6-di-substituted manol-terminating sequences among o-glycans released from brain glycopeptides by reductive alkaline hydrolysis. Eur J Biochem. 263:879–888. [DOI] [PubMed] [Google Scholar]
  10. Cheatham  TE, Cieplak  P, Kollman  PA. 1999. A modified version of the Cornell et al. force field with improved sugar pucker phases and helical repeat. J Biomol Struct Dyn. 16:845–862. [DOI] [PubMed] [Google Scholar]
  11. Coltart  DM, Royyuru  AK, Williams  LJ, Glunz  PW, Sames  D, Kuduk  SD, Schwarz  JB, Chen  XT, Danishefsky  SJ, Live  DH. 2002. Principles of mucin architecture: Structural studies on synthetic glycopeptides bearing clustered mono-, di-, tri-, and hexasaccharide glycodomains. J Am Chem Soc. 124:9833–9844. [DOI] [PubMed] [Google Scholar]
  12. Corzana  F, Busto  JH, Jimenez-Oses  G, de  Luis  MG, Asensio  JL, Jimenez-Barbero  J, Peregrina  JM, Avenoza  A. 2007. Serine versus threonine glycosylation: The methyl group causes a drastic alteration on the carbohydrate orientation and on the surrounding water shell. J Am Chem Soc. 129:9458–9467. [DOI] [PubMed] [Google Scholar]
  13. Crowley  MF, Darden  TA, Cheatham  TE, Deerfield  DW. 1997. Adventures in improving the scaling and accuracy of a parallel molecular dynamics program. J Supercomput. 11:255–278. [Google Scholar]
  14. Darden  T, York  D, Pedersen  L. 1993. Particle mesh ewald - an n.log(n) method for ewald sums in large systems. J Chem Phys. 98:10089–10092. [Google Scholar]
  15. Davis  JT, Hirani  S, Bartlett  C, Reid  BR. 1994. H-1-nmr studies on an asn-linked glycopeptide - glcnac-1 c2-n2 bond is rigid in H2O. J Biol Chem. 269:3331–3338. [PubMed] [Google Scholar]
  16. Delaglio  F, Grzesiek  S, Vuister  GW, Zhu  G, Pfeifer  J, Bax  A. 1995. Nmrpipe - a multidimensional spectral processing system based on unix pipes. J Biomol NMR. 6:277–293. [DOI] [PubMed] [Google Scholar]
  17. Dobson  CM, Hempel  SJ, Stalnaker  SH, Stuart  R, Wells  L. 2013. O-mannosylation and human disease. Cell Mol Life Sci. 70:2849–2857. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Essmann  U, Perera  L, Berkowitz  ML, Darden  T, Lee  H, Pedersen  LG. 1995. A smooth particle mesh ewald method. J Chem Phys. 103:8577–8593. [Google Scholar]
  19. Gerken  TA, Butenhof  KJ, Shogren  R. 1989. Effects of glycosylation on the conformation and dynamics of o-linked glycoproteins - C-13 nmr-studies of ovine submaxillary mucin. Biochemistry. 28:5536–5543. [DOI] [PubMed] [Google Scholar]
  20. Hara  Y, Kanagawa  M, Kunz  S, Yoshida-Moriguchi  T, Satz  JS, Kobayashi  YM, Zhu  ZH, Burden  SJ, Oldstone  MBA, Campbell  KP. 2011. Like-acetylglucosaminyltransferase (large)-dependent modification of dystroglycan at thr-317/319 is required for laminin binding and arenavirus infection. Proc Natl Acad Sci USA. 108:17426–17431. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Harrison  OJ, Jin  XS, Hong  SJ, Bahna  F, Ahlsen  G, Brasch  J, Wu  YH, Vendome  J, Felsovalyi  K, Hampton  CM  et al.  2011. The extracellular architecture of adherens junctions revealed by crystal structures of type i cadherins. Structure. 19:244–256. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Harrison  R, Hitchen  PG, Panico  M, Morris  HR, Mekhaiel  D, Pleass  RJ, Dell  A, Hewitt  JE, Haslam  SM. 2012. Glycoproteomic characterization of recombinant mouse alpha-dystroglycan. Glycobiology. 22:662–675. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Hayakawa  S, Matsushita  T, Yokoi  Y, Wakui  H, Garcia-Martin  F, Hinou  H, Matsuoka  K, Nouso  K, Kamiyama  T, Taketomi  A  et al.  2020. Impaired o-glycosylation at consecutive threonine ttx motifs in mucins generates conformationally restricted cancer neoepitopes. Biochemistry. 59:1221–1241. [DOI] [PubMed] [Google Scholar]
  24. Hennig  M, Bermel  W, Spencer  A, Dobson  CM, Smith  LJ, Schwalbe  H. 1999. Side-chain conformations in an unfolded protein: Chi(1) distributions in denatured hen lysozyme determined by heteronuclear c-13, n-15 nmr spectroscopy. J Mol Biol. 288:705–723. [DOI] [PubMed] [Google Scholar]
  25. Hinou  H, Kikuchi  S, Ochi  R, Igarashi  K, Takada  W, Nishimura  SI. 2019. Synthetic glycopeptides reveal specific binding pattern and conformational change at o-mannosylated position of alpha-dystroglycan by pomgnt1 catalyzed glcnac modification. Bioorg Med Chem. 27:2822–2831. [DOI] [PubMed] [Google Scholar]
  26. Jorgensen  WL, Chandrasekhar  J, Madura  JD, Impey  RW, Klein  ML. 1983. Comparison of simple potential functions for simulating liquid water. J Chem Phys. 79:926–935. [Google Scholar]
  27. Joung  IS, Cheatham  TE. 2008. Determination of alkali and halide monovalent ion parameters for use in explicitly solvated biomolecular simulations. J Phys Chem B. 112:9020–9041. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Joung  IS, Cheatham  TE. 2009. Molecular dynamics simulations of the dynamic and energetic properties of alkali and halide ions using water-model-specific ion parameters. J Phys Chem B. 113:13279–13290. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Kay  LE, Keifer  P, Saarinen  T. 1992. Pure absorption gradient enhanced heteronuclear single quantum correlation spectroscopy with improved sensitivity. J Am Chem Soc. 114:10663–10665. [Google Scholar]
  30. Kirschner  KN, Yongye  AB, Tschampel  SM, Gonzalez-Outeirino  J, Daniels  CR, Foley  BL, Woods  RJ. 2008. Glycam06: A generalizable biomolecular force field. Carbohydrates. J Comput Chem. 29:622–655. [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Kunz  S, Calder  L, Oldstone  MBA. 2004. Electron microscopy of an alpha-dystroglycan fragment containing receptor sites for lymphocytic choriomeningitis virus and laminin, and use of the receptoid body as a reagent to neutralize virus. Virology. 325:207–215. [DOI] [PubMed] [Google Scholar]
  32. Larsen  ISB, Narimatsu  Y, Clausen  H, Joshi  HJ, Halim  A. 2019. Multiple distinct o-mannosylation pathways in eukaryotes. Curr Opin Struct Biol. 56:171–178. [DOI] [PubMed] [Google Scholar]
  33. Larsen  ISB, Narimatsu  Y, Joshi  HJ, Siukstaite  L, Harrison  OJ, Brasch  J, Goodman  KM, Hansen  L, Shapiro  L, Honig  B  et al.  2017a. Discovery of an o-mannosylation pathway selectively serving cadherins and protocadherins. Proc Natl Acad Sci USA. 114:11163–11168. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Larsen  ISB, Narimatsu  Y, Joshi  HJ, Yang  Z, Harrison  OJ, Brasch  J, Shapiro  L, Honig  B, Vakhrushev  SY, Clausen  H  et al.  2017b. Mammalian o-mannosylation of cadherins and plexins is independent of protein o-mannosyltransferases 1 and 2. J Biol Chem. 292:11586–11598. [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Liu  MA, Borgert  A, Barany  G, Live  D. 2008. Conformational consequences of protein glycosylation: Preparation of o-mannosyl serine and threonine building blocks, and their incorporation into glycopeptide sequences derived from alpha-dystroglycan. Biopolymers. 90:358–368. [DOI] [PubMed] [Google Scholar]
  36. Live  D, Wells  L, Boons  GJ. 2013. Dissecting the molecular basis of the role of the o-mannosylation pathway in disease: Alpha-dystroglycan and forms of muscular dystrophy. ChemBioChem. 14:2392–2402. [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Lommel  M, Winterhalter  PR, Willer  T, Dahlhoff  M, Schneider  MR, Bartels  MF, Renner-Muller  I, Ruppert  T, Wolf  E, Strahl  S. 2013. Protein o-mannosylation is crucial for e-cadherin-mediated cell adhesion. Proc Natl Acad Sci USA. 110:21024–21029. [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Madariaga  D, Martinez-Saez  N, Somovilla  VJ, Garcia-Garcia  L, Berbis  MA, Valero-Gonzalez  J, Martin-Santamaria  S, Hurtado-Guerrero  R, Asensio  JL, Jimenez-Barbero  J  et al.  2014. Serine versus threonine glycosylation with alpha-o-galnac: Unexpected selectivity in their molecular recognition with lectins. Chem A Eur J. 20:12616–12627. [DOI] [PubMed] [Google Scholar]
  39. Mallajosyula  SS, MacKerell  AD. 2011. Influence of solvent and intramolecular hydrogen bonding on the conformational properties of o-linked glycopeptides. J Phys Chem B. 115:11215–11229. [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Martinez-Saez  N, Peregrina  JM, Corzana  F. 2017. Principles of mucin structure: Implications for the rational design of cancer vaccines derived from muc1-glycopeptides. Chem Soc Rev. 46:7154–7175. [DOI] [PubMed] [Google Scholar]
  41. Matsumoto  Y, Kudelka  MR, Hanes  MS, Lehoux  S, Dutta  S, Jones  MB, Stackhouse  KA, Cervoni  GE, Heimburg-Molinaro  J, Smith  DF  et al.  2020. Identification of tn antigen o-galnac-expressing glycoproteins in human carcinomas using novel anti-tn recombinant antibodies. Glycobiology. 30:282–300. [DOI] [PMC free article] [PubMed] [Google Scholar]
  42. McDonald  IK, Thornton  JM. 1994. Satisfying hydrogen-bonding potential in proteins. J Mol Biol. 238:777–793. [DOI] [PubMed] [Google Scholar]
  43. Miyamoto  S, Kollman  PA. 1992. Settle - an analytical version of the shake and rattle algorithm for rigid water models. J Comput Chem. 13:952–962. [Google Scholar]
  44. Mo  KF, Fang  T, Stalnaker  SH, Kirby  PS, Liu  MA, Wells  L, Pierce  M, Live  DH, Boons  GJ. 2011. Synthetic, structural, and biosynthetic studies of an unusual phospho-glycopeptide derived from alpha-dystroglycan. J Am Chem Soc. 133:14418–14430. [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. Nilsson  J, Nilsson  J, Larson  G, Grahn  A. 2010. Characterization of site-specific o-glycan structures within the mucin-like domain of alpha-dystroglycan from human skeletal muscle. Glycobiology. 20:1160–1169. [DOI] [PubMed] [Google Scholar]
  46. Ottiger  M, Bax  A. 1999. Bicelle-based liquid crystals for nmr-measurement of dipolar couplings at acidic and basic ph values. J Biomol NMR. 13:187–191. [DOI] [PubMed] [Google Scholar]
  47. Perez  A, Marchan  I, Svozil  D, Sponer  J, Cheatham  TE, Laughton  CA, Orozco  M. 2007. Refinenement of the amber force field for nucleic acids: Improving the description of alpha/gamma conformers. Biophys J. 92:3817–3829. [DOI] [PMC free article] [PubMed] [Google Scholar]
  48. Piotto  M, Saudek  V, Sklenar  V. 1992. Gradient-tailored excitation for single-quantum nmr-spectroscopy of aqueous-solutions. J Biomol NMR. 2:661–665. [DOI] [PubMed] [Google Scholar]
  49. Raman  J, Guan  Y, Perrine  CL, Gerken  TA, Tabak  LA. 2012. Udp-n-acetyl-alpha-d-galactosamine:Polypeptide n-acetylgalactosaminyltransferases: Completion of the family tree. Glycobiology. 22:768–777. [DOI] [PMC free article] [PubMed] [Google Scholar]
  50. Rose  MC, Voter  WA, Sage  H, Brown  CF, Kaufman  B. 1984. Effects of deglycosylation on the architecture of ovine submaxillary mucin glycoprotein. Journal of Biological Chemistry. 259:3167–3172. [PubMed] [Google Scholar]
  51. Rubinstein  R, Thu  CA, Goodman  KM, Wolcott  HN, Bahna  F, Mannepalli  S, Ahlsen  G, Chevee  M, Halim  A, Clausen  H  et al.  2015. Molecular logic of neuronal self-recognition through protocadherin domain interactions. Cell. 163:629–642. [DOI] [PMC free article] [PubMed] [Google Scholar]
  52. Ryckaert  JP, Ciccotti  G, Berendsen  HJC. 1977. Numerical-integration of cartesian equations of motion of a system with constraints - molecular-dynamics of n-alkanes. J Comput Phys. 23:327–341. [Google Scholar]
  53. Sagui  C, Darden  TA. 1999. In: Pratt  LR, Hummer  G, editors. Simulation and Theory of Electrostatic Interactions in Solution. Melville, NY: American Institute of Physics. p. 104–113. [Google Scholar]
  54. Schwieters  CD, Kuszewski  JJ, Tjandra  N, Clore  GM. 2003. The xplor-nih nmr molecular structure determination package. J Magn Reson. 160:65–73. [DOI] [PubMed] [Google Scholar]
  55. Shogren  R, Gerken  TA, Jentoft  N. 1989. Role of glycosylation on the conformation and chain dimensions of o-linked glycoproteins - light-scattering-studies of ovine submaxillary mucin. Biochemistry. 28:5525–5536. [DOI] [PubMed] [Google Scholar]
  56. Shurer  CR, Kuo  JCH, Roberts  LM, Gandhi  JG, Colville  MJ, Enoki  TA, Pan  H, Su  J, Noble  JM, Hollander  MJ  et al.  2019. Physical principles of membrane shape regulation by the glycocalyx. Cell. 177:1757–1770. [DOI] [PMC free article] [PubMed] [Google Scholar]
  57. Stalnaker  SH, Aoki  K, Lim  JM, Porterfield  M, Liu  MA, Satz  JS, Buskirk  S, Xiong  YF, Zhang  P, Campbell  KP  et al.  2011. Glycomic analyses of mouse models of congenital muscular dystrophy. J Biol Chem. 286:21180–21190. [DOI] [PMC free article] [PubMed] [Google Scholar]
  58. Stalnaker  SH, Hashmi  S, Lim  JM, Aoki  K, Porterfield  M, Gutierrez-Sanchez  G, Wheeler  J, Ervasti  JM, Bergmann  C, Tiemeyer  M  et al.  2010. Site mapping and characterization of o-glycan structures on alpha-dystroglycan isolated from rabbit skeletal muscle. J Biol Chem. 285:24882–24891. [DOI] [PMC free article] [PubMed] [Google Scholar]
  59. Stein  EG, Rice  LM, Brunger  AT. 1997. Torsion-angle molecular dynamics as a new efficient tool for nmr structure calculation. J Magn Reson. 124:154–164. [DOI] [PubMed] [Google Scholar]
  60. Tran  DT, Lim  JM, Liu  M, Stalnaker  SH, Wells  L, Ten Hagen  KG, Live  D. 2012. Glycosylation of alpha-dystroglycan o-mannosylation influences the subsequent addition of galnac by udp-galnac polypeptide n-acetylgalactosaminyltransferases. J Biol Chem. 287:20967–20974. [DOI] [PMC free article] [PubMed] [Google Scholar]
  61. Vajpai  N, Gentner  M, Huang  JR, Blackledge  M, Grzesiek  S. 2010. Side-chain chi(1) conformations in urea-denatured ubiquitin and protein g from (3)j coupling constants and residual dipolar couplings. J Am Chem Soc. 132:3196–3203. [DOI] [PubMed] [Google Scholar]
  62. van de  Ven  FJM. 1995. Multidimensional NMR in Liquids. New York, NY: Wiley-VCH. [Google Scholar]
  63. Vester-Christensen  MB, Halim  A, Joshi  HJ, Steentoft  C, Bennett  EP, Levery  SB, Vakhrushev  SY, Clausen  H. 2013. Mining the o-mannose glycoproteome reveals cadherins as major o-mannosylated glycoproteins. Proc Natl Acad Sci USA. 110:21018–21023. [DOI] [PMC free article] [PubMed] [Google Scholar]
  64. Wuthrich  K. 1986. NMR of Proteins and Nucleic Acids. New York, NY: John Wiley and Sons. [Google Scholar]
  65. Yoshida-Moriguchi  T, Campbell  KP. 2015. Matriglycan: A novel polysaccharide that links dystroglycan to the basement membrane. Glycobiology. 25:702–713. [DOI] [PMC free article] [PubMed] [Google Scholar]
  66. Yoshida-Moriguchi  T, Yu  LP, Stalnaker  SH, Davis  S, Kunz  S, Madson  M, Oldstone  MBA, Schachter  H, Wells  L, Campbell  KP. 2010. O-mannosyl phosphorylation of alpha-dystroglycan is required for laminin binding. Science. 327:88–92. [DOI] [PMC free article] [PubMed] [Google Scholar]
  67. Yu  SH, Zhao  P, Prabhakar  PK, Sun  TT, Beedle  A, Boons  GJ, Moremen  KW, Wells  L, Steet  R. 2018. Defective mucin-type glycosylation on -dystroglycan in cog-deficient cells increases its susceptibility to bacterial proteases. Journal of Biological Chemistry. 293:14534–14544. [DOI] [PMC free article] [PubMed] [Google Scholar]
  68. Zgarbova  M, Otyepka  M, Sponer  J, Mladek  A, Banas  P, Cheatham  TE, Jurecka  P. 2011. Refinement of the Cornell et al. nucleic acids force field based on reference quantum chemical calculations of glycosidic torsion profiles. J Chem Theory Comput. 7:2886–2902. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

SD_rev_cwaa112
Click here for additional data file. (1.3MB, pdf)
TGalNAc5_cwaa112
Click here for additional data file. (15MB, mp4)
TMan5_cwaa112
Click here for additional data file. (7.8MB, mp4)

Articles from Glycobiology are provided here courtesy of Oxford University Press

RESOURCES