Glycobiology Advance Access originally published online on June 14, 2006
Glycobiology 2006 16(10):969-980; doi:10.1093/glycob/cwl015
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Structure and dynamics of the conserved protein GPI anchor core inserted into detergent micelles
2 Grupo de Carbohidratos, Instituto de Investigaciones Químicas, CSIC, Isla de la Cartuja, C/Américo Vespucio 49, 41092 Seville, Spain; 3 Centre de Recherches sur les Macromolécules Végétales (CERMAV), CNRS, BP 53F-38941 Grenoble Cedex 9, France; and 4 Centro de Investigaciones Biologicas, CSIC C/Ramiro de Maeztu 9, 28040 Madrid, Spain
1 To whom correspondence should be addressed; e-mail: pedro.nieto{at}iiq.csic.es
Received on April 5, 2006; revised on June 7, 2006; accepted on June 10, 2006
 |
Abstract |
|---|
 Top Abstract IntroductionResults and discussionConclusionsMaterials and methodsSupplementary materialConflict of interest statementAcknowledgmentsReferences |
|---|
A suitable approach which combines nuclear magnetic resonance (NMR) spectroscopy and molecular dynamics (MD) simulations have been used to study the structure and the dynamics of the glycosylphosphatidylinositol (GPI) anchor Man
l-2Man1-6Manl -4GlcN1-6myo-inositol-1-OPO3-sn-1,2-dimyristoylglycerol (1) incorporated into dodecylphosphatidylcholine (DPC) micelles. The results have been compared to those previously obtained for the products obtainable from (1) after phospholipase cleavage, in aqueous solution. Relaxation and diffusion NMR experiments were used to establish the formation of stable aggregates and the insertion of (1) into the micelles. MD calculations were performed including explicit water, sodium and chloride ions and using the Particle Mesh Ewald approach for the evaluation of the electrostatic energy term. The MD predicted three dimensional structure and dynamics were substantiated by nuclear overhauser effect (NOE) measurements and relaxation data. The pseudopentasaccharide structure, which was not affected by incorporation of (1) into the micelle, showed a complex dynamic behaviour with a faster relative motion at the terminal mannopyranose unit and decreased mobility close to the micelle. This motion may be better described as an oscillation relative to the membrane rather than a folding event. Key words: GPI anchor / micelle / molecular flexibility / molecular modelling / NMR spectroscopy
|
Introduction |
|---|
|
TopAbstractIntroductionResults and discussionConclusionsMaterials and methodsSupplementary materialConflict of interest statementAcknowledgmentsReferences |
|---|
A considerable number of studies on the biology and the chemistry of glycosylphosphatidylinositols (GPIs) have appeared in the literature (Ferguson and Williams, 1988
; Thomas et al., 1990; Mcconville and Ferguson, 1993; Ferguson et al., 1994; Stevens, 1995; Udenfriend and Kodukula, 1995; Guo and Bishop, 2004) since the first structural characterization of these molecules, that was reported almost twenty years ago (Ferguson et al., 1988; Homans et al., 1988). This interest in GPIs has been primarily due to their main biological function (they attach extracellular structures to the outer face of cell membranes through a covalent linkage) and to the particular features of their conserved chemical structures. In mammals and lower eukaryotes GPIs serve to attach proteins to cellular membranes. These protein GPI anchors present the core structure NH2EtOPO3-6Man1
2Man16Man14GlcNH216myo-Ino1-OPO3-lipid. In protozoal parasites they attach non protein bound extracellular glycoconjugates and share the basic core Man14GlcNH216myo-Ino1-OPO3-lipid (Ferguson et al., 1994). In addition, free GPIs, which do not seem to be derived from the hydrolysis of GPI-anchored proteins or glycoconjugates, have been reported although their chemical structures are less defined than those of the GPI anchors (Deeg et al., 1992).
As a consequence of their high lateral mobility, GPIs are involved in a variety of functions in addition to their anchoring performance. These include facilitating the selective release of molecules from cell surfaces and the exchange of membrane proteins. They also control the turnover of anchored molecules and the transduction of transmembrane signals through their association to membrane microdomains (Thomas et al., 1990; Robinson, 1991; Deeg et al., 1992). Besides, it has also been postulated that GPIs may operate in an intracellular signaling system that involves a phospholipase cleavage at the cell surface to generate inositolphosphoglycan-type second messengers (VarelaNieto et al., 1996; Wang et al., 1997; Jones and Varela-Nieto, 1998, 1999; Jones et al., 2005).
In natural systems GPIs exert their biological functions attached to biological membranes. In these biomembranes, glycolipids assemble in microdomains ("rafts") that sometimes cluster and self-stabilize in flask-shaped invaginations known as caveolae (Severs, 1988; Harder and Simons, 1997; Simons and Ikonen, 1997; Hooper, 1998). The biological functions of GPIs occur in these microdomains, where GPIs and GPI-linked proteins appear located on the outer leaflet (Parpal et al., 1995; Kubler et al., 1996; Wang et al., 1997; Jones and Varela-Nieto, 1999). Understanding the properties of membrane-bound molecules in their natural surrounding and their behavior in these membrane environments is an important and difficult matter. Structural and dynamics studies are being performed using micelles, bicelles, liposomes or even reconstituted membranes (Evans, 1995). Detergent based micelles have been particularly useful to study the structure, the dynamics and the interactions of membrane bound peptides and proteins (Baleja, 2001; Fernandez and Wuthrich, 2003; Opella and Marassi, 2004) and have also been used to study the behavior of glycolipids such as gangliosides (Acquotti and Sonnino, 2000).
We have recently investigated the phospholipase cleavage of free GPIs and GPI-like compounds inserted on liposomes (Bonilla et al., 2002) as a part of a program on the role of inositolphoshoglycans in insulin signaling (Dietrich et al., 1999; Martin-Lomas et al., 2000a; Martin-Lomas et al., 2000b; Martin-Lomas et al., 2000c; Martin-Lomas et al., 2000d; Bonilla et al., 2002; Reichardt and Martin-Lomas, 2003; Chevalier et al., 2005). We now have performed a structural and dynamics study of the free GPI anchor structure Manl-2Man1-6Man1-4GlcN1-6myo-inositol-1-OPO3-sn-1,2-dimyristoylglycerol (1) inserted into a micelle using nuclear magnetic resonance (NMR) spectroscopy and molecular dynamics (MD) simulations. As indicated above, detergent micelles have been previously used to study the behavior of glycolipids in membranes (Bonilla et al., 2002). However, to the best of our knowledge, a protocol based on NMR spectroscopy and MD calculations in explicit water of the glycolipid incorporated into the micelle has never been used before in these studies. The structure and dynamics of the pseudopentasaccharide core in this system have been compared with those of the previously reported phospholipase cleavage products, Manl-2Man1-6Manl -4GlcN1-6myo-inositol (2) and Manl-2Man1-6Man1-4GlcN1-6myo-inositol-1,2-cyclic-OPO3 (3) (see structures in Scheme S1, supplementary material; Dietrich et al., 1999; Martin-Lomas et al., 2000d).
|
Results and discussion |
|---|
|
TopAbstractIntroductionResults and discussionConclusionsMaterials and methodsSupplementary materialConflict of interest statementAcknowledgmentsReferences |
|---|
NMR
For the NMR study sodium dodecyl sulphate (SDS) and dodecylphosphatidylcholine (DPC) micelles were prepared but DPC micelles were finally chosen as they showed a better behaviour with regard to GPI insertion and provided an environment which was thought to better resemble the lipid bilayer of cellular membranes. Detergent concentrations larger than the critical micelle concentration (CMC), a temperature higher than the Krafft point, and 10 mM phosphate buffer were used in order to ensure micelle formation. These conditions have been employed in structural studies of other biomolecules inserted into detergent micelles (Vinogradova et al., 1998
; le Maire et al., 2000; Krueger-Koplin et al., 2004). However, to determine the formation of the expected stable aggregates and the insertion of 1 into these micelles was not straightforward from the NMR spectra since the overlapping of the signals for the detergent and for 1 prevented the unambiguous detection of ligand-micelle nuclear overhauser effect (NOE) peaks, which would be a proof of insertion (Losonczi et al., 2000; Glover et al., 2002). Thus, in order to unequivocally characterize the formation of the aggregates and the insertion of 1 relaxation and diffusion NMR experiments were used.
The analysis of the rotational correlation time, 
c, calculated from NMR relaxation parameters, is an adequate method to ensure micelle formation and to estimate micelle size as c can be related through the Stokes-Einstein relationship with the volume of the vesicle (Poppe et al., 1994; Wand et al., 1998). In the case of micelles of deuterated phospholipids, as DPC, the measurement of 31P relaxation rates at different magnetic fields is also a suitable method as chemical shift anisotropy is the only active relaxation mechanism thus simplifying the correlation time and relaxation times relationship (Smith and Ekiel, 1984). Thus the longitudinal relaxation times for 1 in presence of DPC were measured at 11.74, 9.39 and 7.05 T magnetic fields. The values obtained (0.89, 1.29 and 1.87 s–1, respectively) were used to determine the global correlation time, c, the internal correlation time, i, and the generalized order parameter S2 (5.06 ns, 0.19 ns and 0.25 ns, respectively). These values were in agreement with those described for similar DPC micelles (5.07 ns, 0.12 ns and 0.21) indicating the formation of micelles in the NMR samples (Poppe et al., 1994). On the other hand, diffusion ordered spectroscopy (DOSY) (Asensio et al., 1999; Groves, Palczewska, et al., 2004; Groves, Rasmussen, et al., 2004) also provides an accurate method to assess the formation of micelles and the insertion of 1 by determining the diffusion coefficients (Ddiff) of the individual species since these coefficients are related with the hydrodynamic volume of the aggregate (Groves, Palczewska, et al., 2004; Groves, Rasmussen, et al., 2004). Free 1 should behave as a small molecule with a high diffusion constant, while when inserted into the micelle 1 will present a small diffusion constant as expected for the micelle. We have compared Ddiff for 1 in the presence of the micelles with the Ddiff for 2 either alone or in presence of the same type of micelles. As expected, the DOSY spectra of 2 showed similar Ddiff values either in the absence or in the presence of DPC micelles (2.6 and 3.0 x 10–10 cm2 s–1, respectively, corresponding to species with approximate molecular weights of 700 Da), indicating that the diffusion rate of free 2 is almost unaffected by the presence of detergent in the medium. (Figure 1A and C). The DOSY experiment of 1 in presence of DPC micelles (Figure 1B) yielded a Ddiff value of 7.8.10–11 cm2 s–1 which was that expected for an aggregate with a molecular weight of 16,000 Da corresponding to a micelle size of 40–60 molecules (15,500–23,300 Da) which was the size expected according to the aggregation number (Glover et al., 2002). These results are conclusive about the insertion of the 1 into the DPC micelles. Similar experiments with 1 in presence of SDS (data not shown) gave results that were not conclusive of the insertion of all the GPI molecules into the homogeneous SDS micelles.
|
The description of the DPC micelles according to the diffusion and to the relaxation data have been compared taking into account that the global correlation time can be estimated from the expression 1/c = (6/R2) (Dt + Ddiff), where Ddiff is the translational diffusion obtained from DOSY and Dt (Dt = KT/8
R
), the Brownian diffusion calculated from the radius of the micelle (R) estimated from molecular modelling (see below). The calculated global correlation time value was 5.35 ns at 298 K, in good agreement with that calculated from the relaxation data (5.06 ns).
The 1H NMR spectrum of 1 incorporated into these DPC micelles presented enough quality as to allow the assignment of all the signals and to register NOE spectroscopy (NOESY) and off-resonance rotating overhauser effect spectroscopy (ROESY) experiments (Figure 2). The NOE pattern and the interglycosidic distances were compatible with an extended averaged conformation of the pseudopentasaccharide core. The experimental distances determined from NOESY data were similar to those previously measured for 2, 3 and the GPI anchor of the variant surface glycoprotein of Trypanosoma brucei (Homans et al., 1988; Martin-Lomas et al., 2000c; Martin-Lomas et al., 2000d). Some small variations detected could be attributed to the different temperature used in the experiments with 2 experiments in order to avoid the null NOE regime which in this case was close to room temperature. Therefore, it can be concluded that, within experimental error, the insertion of the lipid tail of 1 into the micelle does not modify the three dimensional structure of its carbohydrate moiety with respect to that of the free glycolipid. (Table I).
|
|
The dynamics of 2 and 3, have been previously studied by us. In both cases, MD simulations indicated an appreciable degree of internal flexibility of the glycosidic linkages. The overall outcome of the oscillations of all the glycosidic linkages along the molecule could be described as a "hinge" like motion of the core pseudopentasaccharide, folding over its centre. Also, in both cases the simultaneous detection of positive and negative NOE effects, as a consequence of the dispersion on effective correlation times close to the NOE null point, supported this description (Martin-Lomas et al., 2000c; Martin-Lomas et al., 2000d). A similar study has now been performed with 1 inserted into the DPC micelles. However, in this case all NOE contacts presented negative values as a result of the large size of the aggregates and NOESY spectra could not be used as a qualitative proof of the dynamic behaviour of the carbohydrate moiety. Therefore, a quantitative analysis of the effective correlation times was performed, and the rotational correlation times were calculated by off-resonance ROESY experiments from the 
NOE/ROE ratio for pairs of protons corresponding to fixed distances (Poveda et al., 1997a, Poveda et al., 1997b) for 2, free and in presence of DPC micelles, and 1 inserted into DPC micelles (Table II).
|
The results showed the global trends of those obtained from the previously mentioned 31P relaxation study: the correlation times for 1 incorporated into the DPC micelle increased with respect to 2 (Table II). Moreover, the value of the correlation time of the residue closer to the DPC polar head was 5.06 ns, as calculated from the 31P relaxation analysis. The magnitude of such reduction of the molecular tumbling could not be explained by changes in the medium viscosity caused by the micelles, as the correlation times for the corresponding vectors of pseudopentasaccharide 2 in presence of DPC varied from 3.79 to 3.32 ns. The dispersion of c values for each monosaccharide of 1 incorporated into the micelle revealed differences in the extent of the internal motion along the carbohydrate chain increasing from GlcNII towards the terminal ManV. These results are indicative of relative motions faster or with large amplitude at the external residue, and decreased mobility closer to the micelle.
Molecular dynamics
The above NMR study was further completed by modelling the system composed of 1 inserted onto the DPC micelle in explicit water using molecular dynamics simulations. The DPC micelles have been previously modelled (Tieleman et al., 2000) but simulations of this system incorporating inserted glycolipids are rare. The description of the micelle structure and properties was firstly verified and once the reliability of the simulations was established the structural and dynamics properties of the GPI moiety was evaluated and compared with the experimental data.
System composition and properties.
The initial system was constructed from a pre-equilibrated DPC micelle (Bruce et al., 2002b). For the sake of simplicity, the diacyl glycerol moiety in 1 was replaced by a dodecylphosphate chain. The properties of the pseudopentasaccharide moiety should not be much affected by this replacement and the system should adequately model the possible influence of the zwitterionic surface of the micelle on the structural properties of the carbohydrate chain of the inserted phospholipid. Thus, the GPI molecule was constructed by adding the relaxed structure of the pseudopentasaccharide to a dodecylphosphate chain. The conditions of the simulation were chosen in order to favour the stability of the micelle, sodium and chloride ions were explicitly included, and the theoretical DPC concentration of the solvated system was 300 mM, well above critical micelle concentration (1.3 mM) (Vinogradova et al., 1998).
The performance of the simulation was evaluated by monitoring the density, temperature and volume of the periodic box, which presented reasonable values along the trajectory. The visual inspection of the micelle was also satisfactory; the shape of the micelle was nearly spherical as described for DPC micelles of this size (Tieleman et al., 2000). The aliphatic chains of the DPC molecules were essentially extended while the choline polar heads were bent, causing the quaternary amine groups to lie on the micelle surface. This arrangement was detected as the sum of the distances between the centre of mass of the micelle and the phosphorus atom (17.6 Å) and between the phosphorus and the nitrogen atoms (4.6 Å) was larger than the average distance from the centre and the nitrogen atom (18.6 Å). The average area of the solvent accessible surface, calculated disregarding the outer part of 1, was 203 Å2, similar to that previously reported for a DPC micelle MD by Tieleman and others. The global micelle radius (calculated by the method described by Bruce et al., 2002b) was 21.7 Å. This value was comparable to the experimental values reported for similar DPC micelles as deduced from physicochemical properties like ultracentrifugation and light scattering techniques (Lauterwein et al., 1979).
Both the stability and physicochemical properties of lipid aggregates largely depend on the interactions of the lipid polar heads and the ions and solvent molecules. Therefore, the MD simulations have been performed including explicit water molecules and sodium and chloride ions and using the PME approach for the treatment the electrostatic energy term (Patra et al., 2003; Patra and Karttunen, 2004). In order to verify the performance of the model calculations, the interfacial distributions of ion and water molecules along the trajectories were explored by calculating the radial distribution function (RDF) of water, sodium and chloride around the phosphorus and nitrogen atoms of the DPC polar heads (Figures 3 and 4).
|
|
The calculated sodium radial distribution function (RDF) from the phosphorus atom (Figure 3A) showed two maxima at 3.6 Å and 5.8 Å. The observation of a coordination layer at a shorter distance than the separation between the phosphorus and the nitrogen atoms (4.6 Å) implies that sodium cations can penetrate into the lipid polar head. In addition, the distribution of Na+ around nitrogen showed the first maximum at 5.6 Å (Figure 3B) suggesting that Na+ cations were inside the micelle, probably coordinating phosphate groups from different DPC chains. This type of Na+ distribution around the phosphorus has been found for charged SDS micelles (Bruce et al., 2002a), and neutral DPPC bilayers (Pandit et al., 2003). Chloride anions have a different behaviour and are located outside the DPC polar heads as indicated by the coordination layer at 5.0 Å from nitrogen (Figure 3D) and the absence of apparent coordination with phosphorus (Figure 3C). Finally, the water radial distribution around phosphorus showed an average of five molecules of water at the first shell and more than seventeen within the first two shells. The ion population calculated from the RDFs indicates that the interaction with the micelle is weak and similar to that described for zwitterionic bilayers (Pandit et al., 2003). For Na+, most of the ions were located in the bulk solvent and only 10% were in the first coordination shell of the micelle, and 40% in the two first layers. For Cl– less than 10% of anions resided in the first shell of micelle while the 37% did lay within the first two and the average distance from the centre of mass of the micelle to the chloride anions was large (44 Å) indicating that most of these anions were in the bulk water. Both, the anions and water distribution around the polar head of the lipids were satisfactory indicating that the system used and the MD simulation conditions were adequate for the modelling of the micelle.
The radial distribution functions of water around the oxygen atoms and the nitrogen of the pseudopentasaccharide were also calculated as they can also play an important role in the structure of the carbohydrate moiety by forming hydrogen bonds and organizing hydration shells around the carbohydrate (Figure 4) (Pandit et al., 2003). As described for free carbohydrates, the first water shell around oxygen atoms were detected at less than 3.5 Å, and the acetal oxygens showed a lower degree of hydration than the hydroxyl ones (Perez et al., 1998). While the water distribution around the endocyclic oxygen atoms were analogous to that in sucrose, with hydration numbers close to 1.6, the glycosidic oxygen atoms were less solvated (0.55–0.94). An exception was the myo-inositol residue which was more extensively hydrated (glycosidic oxygen has a hydration number of 1.66). The calculated radial distribution function of water and the hydration number for the hydroxyl oxygens were the expected for this group (from 3.0 to 3.5 Å and from 1.9 to 3.5 water molecules). From the average hydration number it was estimated that the pseudopentasaccharide was solvated by nearly 60 water molecules. Although some of these water molecules could be involved in hydrogen bonds bridging or stabilizing the structure, there was no NMR evidence of long lived hydrogen bonds in 2 in our previous study (Chevalier et al., 2005).
Micelle and GPI-like molecule structure.
Once the suitability of the model system and the conditions for the simulation were assessed, the structural properties of the carbohydrate moiety of 1 incorporated in the micelle were studied. This analysis was based on the comparison of experimental and theoretical representative distances for the carbohydrate moiety of 1 in the micelle with those for 2 and 3 and on the evaluation of the conformational space accessible to the glycosidic linkages for 1 inserted in the micelle, 2, and 3. The significant interglycosidic linkage distances were calculated as the <r–6> average over the last 4.0 ns of the MD trajectory (Table III). The agreement with experimental distances was fairly good and both the calculated and the experimental distances were in accordance with those calculated for 2 and 3 (Martin-Lomas et al., 2000c; Martin-Lomas et al., 2000d). With regard to the conformational behaviour of the glycosidic linkages along the molecular dynamics, the trajectories covered relatively narrow and well defined regions of the torsional space, suggesting some degree of internal flexibility caused by discrete oscillations around the global minimum but not large conformational transitions were observed (Figure 5). The regions of the 
/
space visited were similar to the low energy areas described by molecular mechanics and molecular dynamics for 2 and 3. In any case, the most populated regions found were indicative that the structure of the carbohydrate moiety of 1 incorporated into the micelle was essentially linear, as those found for 2 and 3 .(Martin-Lomas et al., 2000c; Martin-Lomas et al., 2000d). Therefore, it can be deduced that the three-dimensional structure of the GPI carbohydrate backbone is not influenced by the proximity of the membrane, at least in this simplified model.
|
|
Concerning the dynamics of the system, MD simulations and NMR studies with 2and 3 (Martin-Lomas et al., 2000c; Chevalier et al., 2005) have previously concluded that the common pseudopentasaccharide core has a characteristic flexibility that is disclosed as a molecular motion that makes the molecule bend over its central residue, as a hinge. The NMR correlation times calculated for 1 inserted into the micelle also were indicative of a complex dynamic behaviour, but in this case the pattern was different than for 2 (Chevalier et al., 2005). Contradictorily, the global structure defined by the experimental and theoretical interglycosidic distances and conformations were equivalent in both cases. However, the MD simulations of 1 in the micelle can conceal the different flexibility pattern deduced from c with a common average three-dimensional structure. The superimposition of the structures of the aggregate along the trajectories reveals, together with the rotation around the lipid axis, a large flexibility of the carbohydrate moiety of 1 that displays a rocking motion (Figure 6). This type of motion would explain the c pattern along the chain as the amplitude of the motion increases towards the external end of the glycolipid, causing a decrease of c. Some less represented structures, analogous to the folded structures found for 2 and 3, were also identified, but no attempt was made to compare their stability due to the large difference on the methodology used. The MD reveals that the two types of flexibility found for the carbohydrate moiety of 1 when incorporated in the micelle and for 2 and 3, folding (bending around the central mannose in a hinge-like motion) and rocking (pivoting with respect to the lipid part), are originated by the same kind of local dynamics around the glycosidic linkages. In the case of 1, the larger mass of the micelle linked to one end dampens the absolute motion of this side while increase the amplitude of the oscillations of the external mannose, as it was detected by the influence on c.
|
|
Conclusions |
|---|
|
TopAbstractIntroductionResults and discussionConclusionsMaterials and methodsSupplementary materialConflict of interest statementAcknowledgmentsReferences |
|---|
In order to assess the possible influence of the membrane environment on the structure and the dynamics of the carbohydrate moiety of GPI anchors, the three-dimensional structure and dynamics of the pseudopentasaccharide motif of a synthetic GPI anchor (1) inserted into DPC micelles have been studied and compared with those for the free analogues 2 and 3 in aqueous solution (Martin-Lomas et al., 2000c
; Chevalier et al., 2005). The use of DPC micelles for these studies presents several advantages since their size is suitable for NMR and modelling and the probability to have several molecules of the compound studied inserted into the same micelle is negligible thus simplifying the analysis. The formation of the micelle, its average size and insertion of the glycolipid could be undoubtedly proved by NMR methods using 31P relaxation or diffusion spectroscopy. Although, detergent micelles have been often used in NMR studies of glycolipids (Acquotti and Sonnino, 2000), this is the first time, to our knowledge, that the NMR study is combined with MD simulations considering the glycolipid inserted into the micelle immersed in explicit water. This combination of NMR and MD improves the results, is mature and it can be applied to other glycolipids such as gangliosides or cerebrosides.
The three-dimensional structure of the carbohydrate component of 1 incorporated into the micelle, as deduced from the NMR NOE data and the MD trajectory, was essentially extended and comparable with the shape of similar chemical structures such as 2 and 3 (Martin-Lomas et al., 2000c; Chevalier et al., 2005). The available conformational space visited by the glycosidic linkages of 1 included into the micelle was also similar to that calculated for 2 and 3. The interglycosidic linkages displayed some degree of flexibility around the global minimum, but not large conformational changes. The pseudopentasaccharide backbone of glycolipid 1 incorporated into the micelle had, as previously found for 2 and 3 a complex dynamic behaviour that, although originated by a similar internal motion around the glycosidic linkages, is disclosed as a rocking oscillation instead of a folding event. This flexibility may have implications in GPI function as protein anchor to the cell membrane providing plasticity to the system.
We can conclude from the present study that the structural properties of the carbohydrate moiety present in GPI/inositol phosphoglycan (IPG) compounds seems not deviate from its intrinsic behaviour, as represented by the free models 2 and 3, by the vicinity of the lipid–water interface, as seen in 1 inserted into micelles. This moiety, shared among the entire GPI anchor family, presents an extended but flexible structure which can be related with its biological role connecting proteins to cell membranes. Thus, as the spatial relationship between anchored protein and membrane (relative orientation and distance) is controlled by the GPI moiety, the flexibility shown by the carbohydrate would allow change these properties.
|
Materials and methods |
|---|
|
TopAbstractIntroductionResults and discussionConclusionsMaterials and methodsSupplementary materialConflict of interest statementAcknowledgmentsReferences |
|---|
Compounds 1 and 2 were synthesised following described procedures and their synthesis will be fully described elsewhere (Lopez-Prados et al., 2005
).
NMR
NMR sample preparation.
NMR samples of free 1 and 2 were prepared in d3-MeOD and in D2O respectively. Samples of 1 and 2 in the presence of micelles were prepared by mixing methanolic solutions of the carbohydrate with the sodium salt of the corresponding detergent, SDS or DPC (purchased from Cambridge Isotope Laboratories, Inc., Andover, MA). The resulting solutions were evaporated and resuspended in 10.0 mM phosphate buffer D2O (99.97%, Euriso-top) solution. Two samples of 1 were prepared in 100 mM SDS (2.85 mM and pH* 7.5); and in 150 mM DPC (2.85 mM and pH* 7.2). Two samples of 2 were prepared, one in 150 mM DPC (7.25 mM and pH* 7.25) and the other without detergent in phosphate buffer 10 mM (7.25 mM at pH* 7.3).
NMR experiments were recorded in Bruker Avance instruments (500 MHz). dqf-COSY (Davis et al., 1991), total correlation spectroscopy (TOCSY) (Bax and Davis, 1985), NOESY (Parthasarathy et al., 2000), ROESY (Bothnerby et al., 1984), and 1H detected single quantum correlation (HSQC) (Schleucher et al., 1994) were performed using manufacturer standard sequences, implemented with z-pulsed field gradients when possible, and acquired using time proportional phase incrementation mode (Marion and Wuthrich, 1983). The size of the acquisition data matrix for homonuclear experiments was typically of 2k x 512 in F2 and F1 dimensions, respectively. Data were processed using manufacturer software or MestreC program (Cobas and Sardina, 2003), raw data were multiplied by shifted square sine window function prior to Fourier transform, and the baseline was corrected using polynomial fitting. Seven mixing times (150, 200, 250, 300, 400, 450 and 500 ms) were used for NOESY spectrum acquisition. The cross relaxation rates, NOE, were estimated as the initial NOE growth rate calculated from the build-up curves by extrapolation at the null mixing time (Maaheimo et al., 2000). Experimental proton-proton distances were calculated from NOE assuming the isolated spin pair approximation. Off-resonance ROESY experiments (Berthault et al., 1995) were recorded at several mixing times (100, 200, 300, 400 and 500 ms) using four different offsets of the spin locking carrier for the estimation of the longitudinal and transversal cross relaxation rates. Effective correlation times were calculated from: ROE/NOE = (5+22
02c2+804c4)/(5+02c2–402c4) (Poveda et al., 1997a).
DOSY experiments were recorded according to the manufacturer software at 298 K, using a stimulated echo scheme with longitudinal eddy current compensation, bipolar gradient pulses, and two purge gradients (Stilbs, 1981). Thirty-two 1H spectra were collected using a fix encoding gradient of 1.0 ms and an echo delay of 100 ms. The gradient strength (53.5 G cm–1) was linearly varied between 65 and 95% for 2 and between 2 and 95% for 1. The 1D spectra were processed and automatically baseline corrected. The diffusion dimension, zero-filled to 1k, was exponentially fitted according to preset windows for the diffusion dimension (–8.5 < logD < –10.0). The diffusion coefficient of the residual D2O signal was used to assure the performance of the experiment by comparison with the 2.10–9 cm2 s–1 (Stilbs, 1981).
31P spin-lattice relaxation time, T1, has been determined at three different magnetic fields: 7.05 T, 9.39 T and 11.75 T at 298 K; and at 7.05 T and 295 K. Inversion recovery experiments were recorded using 19 recovery delays between 0.5 and 15.0 s and a relaxation delay of 20.0 s. The correlation times, and order parameters were calculated assuming that the main source of 31P longitudinal relaxation of fully deuterated DPC is the CSA mechanism: 1/T1 = 3/10*C2*p2*J(p), and considering the model free approach spectral density function (Lipari and Szabo, 1982a,b). A value of 115.7 ppm for the chemical shift tensor was used (Dufourc et al., 1992).
Molecular dynamics
Molecular dynamic simulations were performed using the sander module of AMBER 6.0 program package (Case et al., 1999), the Parm98 force field (Kollman et al., 1997) and the GLYCAM-1993 extension for carbohydrates (Tumova et al., 2000). The micelle containing GPI-like molecule was hydrated in the Xleap module of AMBER by a periodic box of TIP3P water (Jorgensen et al., 1983). All simulations were run using periodic boundary conditions, a time step of 1.0 fs (0.001 ps), a temperature of 300 K by applying the Berendsen temperature coupling (Berendsen et al., 1984), SHAKE algorithm with a tolerance of 0.000001 (Ryckaert et al., 1977), a 9-Å cutoff was applied to the Lennard–Jones interactions, and at constant pressure nPT (NTB = 2) of 1.0 atm using isotropic position scaling (NTP = 1). The compressibility of system was adjusted to 44.6 10–11 m kg–1 s2. Scaling factors of 2.0 and 1.2 were used for 1–4 electrostatic and van der Waals interactions respectively. Electrostatic forces were calculated using the Particle Mesh Ewald technique (PME) (Essmann et al., 1995), as implemented in AMBER 6. Charges have been calculated with GAUSSIAN98 using HF/6-31G* level of calculation (Frisch et al., 1998).
The starting structure was constructed based on an equilibrated DPC micelle containing 40 phospholipid molecules (Tieleman et al., 2000) (downloaded from http://moose.bio.ucalgary.ca/downloads) by adding the carbohydrate moiety (-D-Mannopyranosyl–(12)--D-Mannopyrannosyl-(16)--D-Mannopyranosyl-(14)--2-deoxy-2-amino-glycosyl-(16)-D-myo-inositol) and adapting the DPC topology to AMBER. A periodic box of TIP3P water was extended by 6.5 Å in each direction from all atoms of the system to contain 7236 water molecules, 40 sodium ions and 40 chloride ions were added to neutralize DPC and GPI-like molecule. Equilibration process started by the minimization of the structure with 2250 steepest descent iterations followed by 2250 conjugated gradient steps. After this minimization, the system was heated at 300 K using a constant volume (nVT) short MD (15,000, 1.5 ps steps). Then the system was gradually equilibrated at 300 K by several short nPT restrained MD. During these MD the positions of the atoms of the micelle and the GPI were restrained, using constrains of decreasing strength: first run, 20,000 steps with 500 kcal/mol restraint; second run, 7000 steps with 25 kcal/mol; third run, 7000 steps with 20 kcal/mol; fourth run, 7000 steps with 20 kcal/mol; fifth run, 7000 steps with 15 kcal/mol; sixth run, 7000 steps with 10 kcal/mol; seventh run, 7000 steps with 5 kcal/mol; eight run, 25,000 steps with 1 kcal/mol; ninth run 500,000 (500 ps) steps with 0 kcal/mol. Productive run was performed after these equilibration procedure. Trajectory coordinates were recorded every 0.5 ps. Dimensions of the periodic box after equilibration steps were stable and ranged at x = 63.7, y = 60.7 and z = 62.7 Å. The final density of the system was 1.0077 g/mL.
Ion and water radial distribution.
Radial pair distribution functions g(r) were calculated using Carnal and Ptraj modules of AMBER by:
 |
where r is the interatomic distance between the solvent and the solute atom, 
w is the bulk water number density and N(r) is the average number of solvent (or ions) molecules (water oxygens considered) at a distance between r and d(r) from the solute. G(r) x d(r) is the probability of finding a water molecule (or ion) in the range r to r + d(r) from the solute atom. The density has been calculated upon the volume and the variations of the periodic box dimensions. The number of close water (or ion) molecules was deduced by integration of g(r) function. Smoothing of the curves was performed by using Adjacent averaging, considering four adjacent points.
Diffusion.
Diffusion coefficients of sodium and chloride counterions, micelle and self-diffusion of water molecule have been calculated from the MD trajectories using the equation:
where D is the diffusion coefficient and r(t) is the displacement between initial t0 and time t. The diffusion has been considered in the whole system, giving an average value for ions present in the inner and outer shells of hydration, and an average value for the micelle.
Accessible surface area.
This area has been determined by Sybyl 6.9 calculating the Connolly surfaces considering a probe of radius 1.4 Å. The analysis was done over structures that have been saved every thirty ps during the last four ns.
|
Supplementary material |
|---|
|
TopAbstractIntroductionResults and discussionConclusionsMaterials and methodsSupplementary materialConflict of interest statementAcknowledgmentsReferences |
|---|
Supplementary data are available at Glycobiology online (http://glycob.oxfordjournals.org/). Table S1: 1H-Chemical shift observed for 1 and 2 in the presence of SDS and DPC micelles at 298 K, 500 MHz. Figure S1: From top to bottom, dqf COSY of 1 in D2O, 2 into SDS micelles in D2O, 2 into DPC micelles in D2O, and 1 with DPC micelles in D2O. Scheme S1: Chemical structures of 1, 2, and 3.
|
Conflict of interest statement |
|---|
|
TopAbstractIntroductionResults and discussionConclusionsMaterials and methodsSupplementary materialConflict of interest statementAcknowledgmentsReferences |
|---|
None declared.
|
Acknowledgments |
|---|
|
TopAbstractIntroductionResults and discussionConclusionsMaterials and methodsSupplementary materialConflict of interest statementAcknowledgmentsReferences |
|---|
This research was supported by the Ministry of Science and Technology (Grant BQU2002-0374). We thank to Alain Rivet for his support in molecular modelling, and to D. Anne Milet for Gaussian access at the LEDSS (http://www-chimie.ujf-grenoble.fr/LEDSS/RESUM/resum.html) and the CECIC at the Institut de Chimie Moléculaire, Grenoble. We also thank to Prof J. Jiménez-Barbero for his help in diffusion experiments. We would like to thank the European Union for the predoctoral fellowship to F. C. (TMR programme: Grant HRN-CT-2000-00001/GLYCOTRAIN) and Spanish Ministry of Education for P. G. Ramon y Cajal Fellowship.
|
Abbreviations |
|---|
DOSY, diffusion ordered spectroscopy; DPC dodecylphosphatidylcholine; GPI, glycosylphosphatidylinositol; NMR, nuclear magnetic resonance; NOE, nuclear overhauser effect; NOESY, NOE spectroscopy; ROESY, rotating overhauser effect spectroscopy; SDS, sodium dodecyl sulphate; TOCSY, total correlation spectroscopy
|
References |
|---|
|
TopAbstractIntroductionResults and discussionConclusionsMaterials and methodsSupplementary materialConflict of interest statementAcknowledgmentsReferences |
|---|
Acquotti, D. and Sonnino, S. (2000) Use of nuclear magnetic resonance spectroscopy in evaluation of ganglioside structure, conformation, and dynamics. Part B: sphingolipid metabolism and cell signaling. Methods Enzymol., 312, 247–272.[ISI][Medline]
Asensio, J.L., Canada, F.J., Garcia-Herrero, A., Murillo, M.T., Fernandez-Mayoralas, A., Johns, B.A., Kozak, J., Zhu, Z.Z., Johnson, C.R., and Jimenez-Barbero, J. (1999) Conformational behavior of aza-C-glycosides: experimental demonstration of the relative role of the exo-anomeric effect and 1,3-type interactions in controlling the conformation of regular glycosides. J. Am. Chem. Soc., 121, 11318–11329.[CrossRef]
Baleja, J.D. (2001) Structure determination of membrane-associated proteins from nuclear magnetic resonance data. Anal. Biochem., 288, 1–15.[CrossRef][ISI][Medline]
Bax, A. and Davis, D.G. (1985) Mlev-17-based two-dimensional homonuclear magnetization transfer spectroscopy. J. Magn. Reson., 65, 355–360.[ISI]
Berendsen, H.J.C., Postma, J.P.M., Vangunsteren, W.F., Dinola, A., and Haak, J.R. (1984) Molecular-dynamics with coupling to an external bath. J. Chem. Phys., 81, 3684–3690.[CrossRef]
Berthault, P., Birlirakis, N., Goldman, M., and Piotto, M. (1995) Improved versions of off-resonance Roesy. J. Magn. Reson., Series A, 113, 47–52.[CrossRef]
Bonilla, J.B., Munoz-Ponce, J.L., Nieto, P.M., Cid, M.B., Khiar, N., and Martin-Lomas, M. (2002) Synthesis and structure of 1-D-6-O-(2-amino-2-deoxy-alpha- and -beta-D-gluco- and -galactopyranosyl)-3-O-methyl-D-chiro-inositol. Eur. J. Org. Chem., 889–898.
Bothnerby, A.A., Stephens, R.L., Lee, J.M., Warren, C.D., and Jeanloz, R.W. (1984) Structure determination of a tetrasaccharide – transient nuclear overhauser effects in the rotating frame. J. Am. Chem. Soc., 106, 811–813.[CrossRef]
Bruce, C.D., Berkowitz, M.L., Perera, L., and Forbes, M.D.E. (2002a) Molecular dynamics simulation of sodium dodecyl sulfate micelle in water: micellar structural characteristics and counterion distribution. J. Phys. Chem. Series B, 106, 3788–3793.
Bruce, C.D., Senapati, S., Berkowitz, M.L., Perera, L., and Forbes, M.D.E. (2002b) Molecular dynamics simulations of sodium dodecyl sulfate micelle in water: the behavior of water. J. Phys. Chem. B, 106, 10902–10907.[CrossRef]
Case D.A., Pearlman D.A., Caldwell J.W., Cheatham T.E., III, Ross W.S., Simmerling C.L., Darden T.A., Merz K.M., Stanton R.V., Cheng A.L., and others. (1999) AMBER 6.0; University of California San Francisco, San Francisco, CA.
Chevalier, F., Lopez-Prados, J., Perez, S., Martin-Lomas, M., and Nieto, P.M. (2005) Conformational study of GPI anchors: the common oligosaccharide GPI anchor backbone. Eur. J. Org. Chem., 3489–3498.
Cobas, J.C. and Sardina, F.J. (2003) Nuclear magnetic resonance data processing. MestRe-C: a software package for desktop computers. Concept. Magn. Reson. Part A, 19A, 80–96.
Davis, A.L., Laue, E.D., Keeler, J., Moskau, D., and Lohman, J. (1991) Absorption-mode 2-dimensional Nmr-spectra recorded using pulsed field gradients. J. Magn. Reson., 94, 637–644.
Deeg, M.A., Murray, N.R., and Rosenberry, T.L. (1992) Identification of glycoinositol phospholipids in rat-liver by reductive radiomethylation of amines but not in H4Iie hepatoma-cells or isolated hepatocytes by biosynthetic labeling with glucosamine. J. Biol. Chem., 267, 18581–18588.
Dietrich, H., Espinosa, J.F., Chiara, J.L., Jimenez-Barbero, J., Leon, Y., Varela-Nieto, I., Mato, J.M., Cano, F.H., Foces-Foces, C., and Martin-Lomas, M. (1999) Glycosyl inositol derivatives related to inositolphosphoglycan mediators: synthesis, structure, and biological activity. Chem. Eur. J., 5, 320–336.[CrossRef]
Dufourc, E.J., Mayer, C., Stohrer, J., Althoff, G., and Kothe, G. (1992) Dynamics of phosphate head groups in biomembranes – comprehensive analysis using P-31 nuclear-magnetic-resonance lineshape and relaxation-time measurements. Biophys. J., 61, 42–57.
Essmann, U., Perera, L., Berkowitz, M.L., Darden, T., Lee, H., and Pedersen, L.G. (1995) A smooth particle mesh Ewald method. J. Chem. Phys., 103, 8577–8593.[CrossRef]
Evans, J.N.S. (1995) Biomolecular NMR Spectroscopy. Oxford University Press, Oxford.
Ferguson, M.A.J., Brimacombe, J.S., Cottaz, S., Field, R.A., Guther, L.S., Homans, S.W., Mcconville, M.J., Mehlert, A., Milne, K.G., Ralton, J.E., and others. (1994) Glycosyl-phosphatidylinositol molecules of the parasite and the host. Parasitology, 108, S45–S54.
Ferguson, M.A.J., Homans, S.W., Dwek, R.A., and Rademacher, T.W. (1988) Glycosyl-phosphatidylinositol moiety that anchors Trypanosoma-brucei variant surface glycoprotein to the membrane. Science, 239, 753–759.
Ferguson, M.A.J. and Williams, A.F. (1988) Cell-surface anchoring of proteins via glycosyl-phosphatidylinositol structures. Annu. Rev. Biochem., 57, 285–320.[CrossRef][ISI][Medline]
Fernandez, C. and Wuthrich, K. (2003) NMR solution structure determination of membrane proteins reconstituted in detergent micelles. FEBS Lett., 555, 144–150.[CrossRef][ISI][Medline]
Frisch M.J., Trucks G.W., Schlegel H.B., Scuseria G.E., Robb M.A., Cheeseman J.R., Zakrzewski V.G., Montgomery J.A., Jr., Stratmann R.E., Burant J.C., and others. (1998) GAUSSIAN [computer program], Gaussian, Inc., Pittsburgh PA, USA.
Glover, K.J., Whiles, J.A., Vold, R.R., and Melacini, G. (2002) Position of residues in transmembrane peptides with respect to the lipid bilayer: a combined lipid NOEs and water chemical exchange approach in phospholipid bicelles. J. Biomol. NMR, 22, 57–64.[CrossRef][ISI][Medline]
Groves, P., Palczewska, M., Molero, M.D., Batta, G., and Canada, F.J. (2004) Protein molecular weight standards can compensate systematic errors in diffusion-ordered spectroscopy. Anal. Biochem., 331, 395–397.[CrossRef][ISI][Medline]
Groves, P., Rasmussen, M.O., Molero, M.D., Samain, E., Canada, F.J., Driguez, H., and Jimenez-Barbero, J. (2004) Diffusion ordered spectroscopy as a complement to size exclusion chromatography in oligosaccharide analysis. Glycobiology, 14, 451–456.
Guo, Z.W. and Bishop, L. (2004) Chemical synthesis of GPIs and GPI-anchored glycopeptides. Eur. J. Org. Chem., 3585–3596.
Harder, T. and Simons, K. (1997) Caveolae, DIGs, and the dynamics of sphingolipid-cholesterol microdomains. Curr. Opin. Cell Biol., 9, 534–542.[CrossRef][ISI][Medline]
Homans, S.W., Ferguson, M.A.J., Dwek, R.A., Rademacher, T.W., Anand, R., and Williams, A.F. (1988) Complete structure of the glycosyl phosphatidylinositol membrane anchor of rat-brain Thy-1 glycoprotein. Nature, 333, 269–272.