CarbohydrateâAromatic Interactions: Vibrational Spectroscopy and

Article pubs.acs.org/JPCB

Carbohydrate−Aromatic Interactions: Vibrational Spectroscopy and Structural Assignment of Isolated Monosaccharide Complexes with p‑Hydroxy Toluene and N‑Acetyl L‑Tyrosine Methylamide E. Cristina Stanca-Kaposta,†,⊥ Pierre Ç arçabal,‡ Emilio J. Cocinero,§ Paola Hurtado,∥ and John P. Simons*,† †

Department of Chemistry, Physical and Theoretical Chemistry Laboratory, University of Oxford, South Parks Road, Oxford OX1 3QZ, U.K. ‡ Institut des Sciences Moléculaire d’Orsay-CNRS, Université Paris Sud, 91405 Orsay Cedex, France § Departamento de Química Física, Facultad de Ciencia y Tecnología, Universidad del País Vasco, (UPV-EHU), Apartado 644, E-48940 Bilbao, Spain ∥ Department of Physical, Chemical and Natural Systems, Universidad Pablo de Olavide, 41013 Seville, Spain ABSTRACT: The nature of carbohydrate binding first to p-hydroxy toluene and then the capped amino acid, N-acetyl L-tyrosine methyl amide (AcTyrNHMe), has been investigated in a solvent-free environment under molecular beam conditions. A combination of double resonance IR-UV spectroscopy and quantum chemical calculations has established the structures of complexes with the α and β anomers of methyl D-glucoand D-galacto- and L-fucopyranosides (α/βMeGlc, MeGal, MeFuc). The new results, when combined with dispersion-corrected DFT calculations, reveal gas phase structures which are dominated by hydrogen bonding but also with evidence of CH−π bonded interactions in complexes with α/βMeGal. These adopt stacked intermolecular structures in marked contrast to those with α/βMeGlc; p-OH → O bonds linking AcTyrNHMe, or p-hydroxy toluene, to the carbohydrate provide an anchor that facilitates further binding, both through OH → O and NH → O hydrogen bonds to the peptide backbone and through CH−π dispersion interactions with the aromatic side group. “Stacked” structures associated with dispersion interactions with the aromatic ring are not detected in the corresponding complexes of capped phenylalanine, despite their common occurrence in bound carbohydrate−protein structures.

■

INTRODUCTION Carbohydrate−protein interactions which result in selective molecular recognition are central to a vast range of biochemical processes.1−8 The interactions can involve a complex interplay of contributory factors:9−13 obvious ones include the nature, architecture (and flexibility) of the carbohydrate binding modules, and the conformational structure (and flexibility) of the ligands. They can be bound through hydrogen bonding between the carbohydrate and neighboring side groups on the protein, or its peptide backbone, and they can be mediated, in crystal structures, through “bridging” water molecules or through electrostatic interactions with neighboring ions, particularly Ca2+.4 The frequent occurrence of “stacking” between aromatic side groups and the apolar faces of bound carbohydrate ligands14−18 suggests an important contribution from CH−π dispersion interactions.14−23 In aqueous environments, stacked structures might also be driven by hydrophobic effects24a consequence of entropic as well as enthalpic contributionsalthough the role and the relative importance of desolvation, the exclusion of water from apolar regions of the carbohydrate binding sites, remains controversial.18,22 Excellent reviews have appeared recently by Jiménez-Barbero and Gabius.13,18,25 © 2013 American Chemical Society

Whatever its mechanism, the biological importance of carbohydrate binding at aromatic sites in proteins is a very well-recognized “fact of life”, and quantifying the interactions that might be involved has stimulated a variety of experimental and computational investigations of idealized model aromatic− carbohydrate systems.14,18,22,25−30 Because of the propensity for stacking, they have been focused principally, but not exclusively, on the incidence and relative importance of CH−π dispersion interactions (and also OH−π and OH−O hydrogen bonding). In solution, examples include thermochemical and NMR measurements of model systems19−22 and dispersion-corrected density functional theoretical (DFT-D)26−28,30 and ab initio29 calculations of their structure and interactions in the absence of solvent. In the gas phase, vibrational spectroscopy of isolated carbohydrate−aromatic complexes, necessarily free of solvent, coupled with DFT calculations has been used to assign their actual structures.31−35 These began with an exploration of the structure and bonding of a trial set of carbohydrate−toluene complexes,31−33 with toluene providing a “poor man’s” model Received: May 7, 2013 Revised: June 17, 2013 Published: June 17, 2013 8135

dx.doi.org/10.1021/jp404527s | J. Phys. Chem. B 2013, 117, 8135−8142

The Journal of Physical Chemistry B

Article

aromatic ring in the isolated complexes of AcPheNHMe (with α/β methyl D-gluco- and galactopyranoside), neither OHπ nor CHπ.34,35 Instead, the interactions were provided by OH → OC and NH → O hydrogen bonding, to and from the dipeptide backbone, with the bound carbohydrates adopting configurations which optimized the number and strength of hydrogen bonds linking them to the dipeptide chain, either in its folded or extended configuration (Figure 1b). In protein−carbohydrate complexes, stacked structures are most prevalent at tyrosine and tryptophan binding sites.14 Their relative incidence appears to correlate with the computed strengths of the CH−π contact interactions in isolated complexes (of fucose or glucose) with phenol, 3-methyl indole, and benzenepoor man’s models for tyrosine, tryptophan, and phenylalanine.26−30 However, the p-hydroxy substituent in tyrosine also introduces the possibility, or more likely the probability, of hydrogen bonded interactions with the carbohydrate ligand. If the gas phase experiments were repeated, using p-hydroxy toluene and then capped tyrosine (AcTyrNHMe) as a more realistic model, perhaps they might reveal structures which reflected contributions from hydrogen bonding to the p-hydroxy group (providing an “anchor”, a coarse control) and CH−π dispersion interactions (a fine control) to create stacked structures in both model systems. The present Article explores this hypothesis through a new series of IR ion-dip (IRID) spectroscopic experiments probing the complexes formed between p-hydroxy toluene and the trial set of carbohydrates (α/βMeGlc, α/βMeGal, α/βMeFuc) and between AcTyrNHMe and α/β-methyl D-galactopyranoside; see Figure 1. Comparisons between the experimental spectra and DFT calculations, using the M05-2X functional (which is sensitive to dispersion interactions), are used to assign their structures. Some calculations were also conducted using the B97 functional with (B97D) and without (B97) a dispersioncorrection, to expose the structural “fine-tuning” associated with medium-range dispersion interactions.

for phenylalaninea simplification subsequently avoided by replacing it with the capped peptide, N-acetyl phenylalanine methylamide (AcPheNHMe).34,35 The trial set of carbohydrates included the α and β anomers of methyl D-gluco-, Dgalacto-, and L-fucopyranoside (α/βMeGlc, α/βMeGal, α/ βMeFuc), systematically chosen to explore the results of anomeric change (α ↔ β), epimeric change at OH-4 (Glc ↔ Gal), and changing the exocyclic CH2OH group (in Gal and Glc) to CH3 (in Fuc); see Figure 1. The complexes were stabilized at low temperature in a molecular beam, free of an aqueous environment, and were probed through mass-selected IR ion-dip spectroscopy. Although, in some of the toluene complexes, there were spectroscopic indications of OHπ bonding (revealed through the displacement of one of the OH vibrational modes to lower wavenumbers), the majority of the complexes were bound, albeit weakly, through CHπ dispersion interactions.31−33 In contrast, there was no indication of any interaction with the

■

METHODS Spectroscopy. p-Hydroxy toluene, N-acetyl-L-tyrosine methylamide (AcTyrNHMe), and the α and β anomers of methyl D-gluco-, D-galacto-, and L-fucopyranoside were all obtained as commercial samples. The carbohydrate complexes were generated in the gas phase, as described earlier32−35 using seeded molecular beam and either oven-heating (for p-hydroxy toluene) or pulsed laser desorption (for AcTyrNHMe) procedures. Conformer-specific and mass-selected vibrational spectra were recorded in the OH and NH stretch regions through IR ion-dip (IRID) double-resonance spectroscopy. UV radiation was provided by a frequency doubled dye laser (Sirah) tuned onto selected regions of the resonant two-photon ionization (R2PI) spectra, and IR radiation was provided by the idler output of an OPO/OPA laser system (LaserVision) or in some experiments by a dye laser system (Continuum) equipped with a lithium niobate crystal for difference frequency generation. Computation. The structural calculations followed the well-established procedures described previously.32−35 In brief, they began with a series of unrestricted surveys using the molecular mechanics, Monte Carlo multiple-minimization procedure and the large-scale low-frequency-mode torsional sampling procedures36 implemented in the MacroModel software (MacroModel version 8.5.207, Schrödinger, LLC). The initial sets of structures were grouped into families

Figure 1. Molecular structures of (a) p-hydroxy toluene, (b) capped tyrosine in its extended and C7-folded conformations, and (c) representative members of the “trial set” of monosaccharides, methyl D-glucopyranoside, methyl D-galactopyranoside, and methyl L-fucopyranoside. The numbering scheme is indicated on the structure for βMeGlc. 8136



Article

Figure 2. Comparisons between experimental IRID spectra of the trial set of carbohydrate·p-hydroxy toluene complexes and corresponding vibrational spectra associated with their calculated (M05-2X/3-61+G(d)) low energy structures, shown alongside; zero point corrected relative energies (kJ mol−1) are shown in brackets. The symbols σ2, σ3, ..., σ6 indicate the stretching vibrational modes at OH-2, OH-3, ..., OH-6; see Figure 1. (The gap at ∼3500 cm−1 appearing in some of the IRID spectra is due to IR absorption in one of the lithium niobate frequency mixing crystals that were employed.)

vibrational spectra that could then be compared with experiment to allow their assignment. To aid the comparisons, the predicted harmonic wavenumbers of the NH and OH stretch modes were scaled by the factor 0.94. “Best fit” assignments were assessed by comparing the sets of scaled DFT and IRID spectra, with the correspondence between each set of vibrational wavenumbers and their spectral patterns providing the primary guide, while their calculated relative energies provided a secondary guide. In almost every case, the “best fit” corresponded to the calculated minimum energy structure. The degree to which dispersion interactions influenced the

distinguished by the number of strong hydrogen bonded interactions they presented and also, in the complexes with AcTyrNHMe, by their association with extended or folded peptide backbones. The structures of the 20 lower-energy conformers and a representative member of each family (typically ∼50 structures with relative energies ≤15 kJ mol−1) were optimized using dispersion-corrected density functional theory, employing the M05-2X functional37 and a 6-31+G(d) basis set, as implemented in the Gaussian 03 suite of programs.38 These led to new sets of relative energies (corrected for zero-point energy), molecular structures, and 8137



Article

Table 1. Summary of “Best Fit” Vibrational and Principal Structural Assignments for α/β-Methyl Monosaccharide·p-Hydroxy Toluene Complexes Based on Comparisons with M05-2X Calculations, and the Corresponding Assignments for Phenyl Monosaccharide·H2O Complexes41−44a p-hydroxy toluene αMeGlc βMeGlc αMeGal βMeGal αMeFuc βMeFuc

H2O αPhGlc βPhGlc αPhGal βPhGal αPhFuc βPhFuc

cG−g+.ins4,6; cTg+.ins4,3 cG−g+.ins4,6 ccG−g+ (p-OH → O-6) ccG−g+ (p-OH → O-6) cc.ins3,2; cc.ins4,3 cc.ins2,1; cc.ins3,2

cG−g+.ins4,6 cG−g+.ins4,6; ccG+g−.ins6,5 cG−g+.ins6,5 cG−g+.ins6,5 cc.ins3,2 cc.ins2,1

The notation, “c” and “cc”, indicates a clockwise (OH-6 ← OH-4 ← OH-3 ← OH-2) or counter-clockwise (O-1 ← OH-2 ← OH-3 ← OH-4) orientation of the peripheral OH groups on the pyranose ring; G+, G−, T and g+, g−, t indicate gauche or trans (anti) orientations of the exocyclic hydroxymethyl group and its terminal OH-6 group; and “ins x,y” indicates the p-OH, or water, binding (insertion) site with x being the hydrogen bond donor and y the acceptor.

a

Figure 3. (a) R2PI, (b) IR-UV hole-burn spectra of AcTyrNHMe (hole-burn wavenumbers are shown); (c, d) the corresponding IRID spectra and the calculated spectra and structures of its two lowest energy conformers, calculated using the M05-2X functional and a 6-31+G(d,p) basis set.

intermolecular structures of the p-hydroxy toluene complexes was subsequently assessed by comparing the predictions obtained using the B97 functional (employing a TZVPP basis set and the resolution of identity (RI) approximation) with those calculated using Grimme’s “d1” dispersion-correction39 (B97D) as implemented in Turbomole.40 Similar assessments were made for the capped tyrosine complexes.

spectra associated with their lowest energy structures calculated using the M05-2X functional. The complexes with α/βMeGlc and α/βMeFuc each display two broad, intense, and strongly displaced bands at ∼3400 and ∼3520 cm−1, indicating two strongly hydrogen bonded OH groups, and a group of sharper, less intense features between 3600 and 3650 cm−1, associated with more weakly bound groups. The band centered at ∼3400 cm−1 also appears in the corresponding spectrum of the p-hydroxy toluene dimer, suggesting its association with a p-OH → O hydrogen bond. Since their vibrational signatures are characteristic of “insertion” structures, OH-x → p-OH → O-y, similar to those presented in the corresponding microhydrated carbohydrates,42−45 the other broad band would then be associated with the interaction, OHx → p-OH. The IRID spectra associated with the α/βMeGal complexes display qualitatively different profiles; the strong

■

RESULTS p-Hydroxy Toluene Complexes. The R2PI spectra of the complexes were broad and unstructured, and their IRID spectra, recorded in the OH stretch region, were unaffected by variations in the choice of the UV probe wavenumber. The IRID spectra of the complexes of p-hydroxy toluene with the α and β anomers of MeGlc, MeGal, and MeFuc are shown in Figure 2, where they can be compared with the vibrational 8138



Article

Figure 4. Comparisons between the experimental IRID spectra of (a) the AcTyrNHMe·αMeGal and (b) the AcTyrNHMe·βMeGal complexes and the computed (M05-2X/3-61+G(d) vibrational spectra associated with their lowest energy structures, shown alongside; zero point corrected relative energies (kJ mol−1) are shown in brackets. The labels NHMe, NHAc, and p-OH indicate the CH3N−H, CH3CON−H, and p-O−H stretching modes of the capped amino acid; the symbols σ2, σ3, ..., σ6 indicate the stretching modes, O−H-2, O−H-3, ..., O−H-6 of the carbohydrate.

broad band centered at ∼3400 cm−1 is retained, but the second at ∼3520 cm−1 is much weaker or absent altogether, suggesting their association with a different kind of predominant structure. Comparisons between the experimental and computed vibrational spectra confirm these qualitative speculations; the “best-fit” assignments based on the M05-2X calculations are summarized in Table 1. The complexes with α/βMeGlc and α/ βMeFuc are all associated with low energy insertion structures which tend to resemble those adopted by the microhydrated complexes of the corresponding (phenyl) monosaccharides.41−44 OH-4/OH-6 provides the favored binding site in both sets of α/βMeGlc complexes, in each case promoted by a change in the OH-2,3,4 orientation from “counter-clockwise” to clockwise (cc to c) and the hydroxymethyl orientations from G +g− to G−g+. There is no change in OH orientation in α or βMeFuc, and OH-3/OH-2 or OH-2/OH-1 continue to be favored insertion sites. The preferred structures of the α/ βMeGal complexes, however, bound by a single p-OH → O-6 hydrogen bond, do not resemble those of the corresponding hydrated complexes. The most strongly displaced band in the complexes with α/ βMeGlc and α/βMeFuc, at ∼3400 cm−1, is indeed associated with the hydrogen bonded phenolic OH group, p-OH → O-y; since it is more acidic than OH in H2O, the corresponding hydrogen bonded OHW (→ O-y) band is located at somewhat higher wavenumbers, typically ∼3460 cm−1.41−44 The insertion complexes do not display stacking interactions; the α/βMeGal complexes, in contrast, do. They are bound by a single p-OH → O-6 hydrogen bond which forms part of an extended cooperative chain, p-OH → OH-6 → OH-4 → OH-3 → O-2, aided by the axial orientation of OH-4 in the galacto-pyranose ring. (In the corresponding hydrated complexes, where insertion structures are formed, stacking is not an option of course, since the aromatic ring is attached to the carbohydrate rather than the hydrogen bond donor.) What happens when the tyrosine analogue is replaced by the capped amino acid itself?

AcTyrNHMe. Since the resonant two-photon ionization (R2PI), IR-UV hole-burn, and IR ion depletion (IRID) spectra of the isolated capped amino acid, AcTyrNHMe, have not been reported previously, they are shown together in Figure 3. The R2PI spectrum, Figure 3a, presents two distinct components: a broad and densely structured band system which is centered around 35 475 cm−1 and a second, much weaker vibronic band system with its origin at 35 660 cm−1. Their corresponding IRID spectra, shown in Figure 3c,d, are in very good accord with those calculated for the two lowest energy conformers. They are associated, respectively, with an extended dipeptide chain (the global minimum) and a folded chain which incorporates a C7 turn linked by a hydrogen bond from CH3NH → OC; this displaces the broadened CH3NH vibrational band (labeled NHMe, in Figure 3c,d) to 3353 cm−1. IR-UV “hole-burn” spectra, Figure 3b, recorded with an IR laser tuned onto this band, or onto the CH3CONH band (labeled NHAc) at 3470 cm−1 in the other vibrational spectrum, deplete the minor and major components, respectively, of the R2PI spectrum. The vibrational band at 3660 cm−1 which appears in the IRID spectra of both conformers is associated with the phenolic OH (labeled p-OH). Otherwise, not surprisingly, the extended and C7-folded conformations are virtually identical to those of capped phenylalanine, AcPheNHMe,34,45 and also the tryptophan analogue, AcTrpNHMe,46 though the dense vibronic structure displayed in the UV spectrum of the extended conformer is not seen in the corresponding spectrum of capped phenylalanine. AcTyrNHMe Complexes. The R2PI spectra of the α/ βMeGal·AcTyrNHMe complexes were unstructured, displaying a single broad band, centered at ∼35 525 cm−1 (α) and 35 440 cm−1 (β), displaced a little from the corresponding band associated with the extended conformer of the free dipeptide. There were no indications of the second band system at higher wavenumber, associated with its folded conformer. The corresponding IRID spectra are shown in Figure 4, where they can be compared with the calculated vibrational spectra of the lowest energy structures. 8139



Article

The spectrum of the αMeGal complex, Figure 4a, although poorly resolved, accords well with the computed spectrum associated with the global minimum structurea complex incorporating an extended peptide chainbut not with the high energy structures. (The second most stable structure presents a similar computed spectrum but incorporates a folded peptide chain, which conflicts with the expectation based on the location of its R2PI spectrum.) The extended peptide is linked to the carbohydrate through a strong hydrogen bond, r[OH O] ∼1.91 Å, that links the p-OH group on the tyrosine side chain to O-6 to create a cooperative motif involving the axial OH-4 group; cf. the corresponding complex with p-hydroxy toluene. Two further hydrogen bonds, NH → O-2 and OH-3 → OC, link the carbohydrate to the peptide backbone. These support a stacked structure (see later discussion) in which the αMeGal structure is slightly distorted, adopting a ccG−g− conformation rather than the minimum energy conformation, ccG+g−, predicted for the isolated carbohydrate. The IRID spectrum recorded for the βMeGal complex is also in best accord with the predicted lowest energy structure. Once again, a strong hydrogen bond, r[OHO] ∼ 1.84 Å, links the p-OH group to OH-6, which is co-operatively bonded to the axial OH-4 group. Additional bonds between OH-2 and the (CH3NH)CO group, and between O-1 and the (CH3CO)NH group, link the carbohydrate to the peptide backbone, in its extended conformation, again to create a stacked structure but this time incorporating a ccG−g+ conformer of the (bare) carbohydrate.

Figure 5. The influence of dispersion. The “best fit” structures of phydroxy toluene complexes with (a) βMeGal and (b) βMeFuc, showing the results of calculations using RI-B97 (green), which does not include a correction for dispersion, and RI-B97D (blue), which does.

p-OH group to the carbohydrate providing a pivot about which the structural “fine-tuning” can occur. The aromatic ring tilts toward the carbohydrate, bringing the closest contact distance (calculated using the B97 functional) in the βMeFuc complex (CH-2−Carom) down to 2.63 Å, cf. 3.76 Å without dispersion. In the βMeGal complex, where the equatorial orientation at C-1 brings the methoxy group near the aromatic ring, the closest CHmethyl−Carom distance is reduced to 2.73 Å, cf. 4.33 Å without dispersion. These large changes are the result of the change in tilt angle between the two partners (Figure 5). At the same time, the p-OH−O hydrogen bond distance decreases by ∼0.1 and ∼0.05 Å, respectively. Interestingly, in both of these examples, the hydrogen bonding prevents what might have been anticipated as a more favorable CH−π bonded structure, where the apolar face would present its axial CH-1,3,5 bonds toward the aromatic ring. In the αMeGal complex, where the methoxy group is axial (and CH-1 is equatorial), both CH-2 and CH-1 are directed toward the aromatic ring and the closest CH− Carom contact distances, including dispersion, are 2.91 Å (CH1) and 2.75 Å (CH-2). These distances all lie comfortably within the range associated with CH−π dispersion interactions.22 Hydrogen-bonding from p-OH(Tyr) → OH-6 also plays a key role in providing a flexible pivot in the most stable complexes of α/βMeGal with the capped amino acid, AcTyrNHMe, but now some of their CH groups are brought into close contact with the aromatic ring primarily through the additional hydrogen bonds which link the carbohydrate to the peptide backbone. The inclusion of dispersion just brings the partners a little closer together. In the AcTyrNHMe·αMeGal complex, CH-1 and CH-2 both point toward the aromatic ring, and the shortest CH−Carom distances (calculated using the B97 functional) are reduced from 3.19 to 2.84 Å. CH-2 also points toward the aromatic ring in the βMeGal complex, and the closest CH−Carom contact distances are reduced from 3.21 to 2.79 Å. These CH−π contacts, which parallel those displayed in the corresponding p-hydroxy toluene complexes, reinforce the stacked carbohydrate−peptide structures created by the hydrogen bonded scaffold. In the absence of the p-hydroxy group, the picture is quite different. Stacked structures are not formed when α/βMeGal (or α/βMeGlc) are bound, in the gas phase, to capped phenylalanine, AcPheNHMe; hydrogen bonds link the carbohydrates to the peptide backbone, but their intermolecular structures provide no indication of any additional CH−π (or OH−π) interactions with the aromatic ring.34,35 These results led to the suggestion that the stacking at phenylalanine sites observed in protein−carbohydrate complexes crystallized from

■

DISCUSSION The intermolecular binding in complexes between p-hydroxy toluene and the trial set of carbohydrates is dominated by the hydrogen bonded interaction with the acidic p-hydroxy group. This behavior confirms theoretical predictions based on high level ab initio calculations29 (for phenol·fucose) and also force field calculations (for saccharide complexes with p-hydroxy toluene).31 The doubly hydrogen bonded insertion structures formed with α/βMeGlc and α/βMeFuc are similar to those of the corresponding microhydrates, but in α/βMeGal the axial orientation of OH-4 coupled with the exocyclic hydroxymethyl group favors the formation of “addition” structures, linked by a single H-bond, p-OH → O-6. The binding energies in the insertion complexes, involving two hydrogen bonds, are likely to be of the same order as those of the corresponding microhydrates,47 typically ∼30−40 kJ mol−1 and considerably larger than those associated with CH−π interactions, calculated to be ∼18 kJ mol−1 in the toluene·αMeFuc complex.32 This is unlikely to be true for the singly hydrogen bonded α/βMeGal complexes. On the other hand, the α/βMeGlc (cG−g+) conformers, bound into the insertion complexes, are calculated to lie well above their global minima (by ∼10 kJ mol−1), in contrast to the α/βMeGal (ccG−g+) conformers. The structural assignments have been based on calculations using the M05-2X functional, which was designed to accommodate medium range dispersion interactions.39 In order to assess their influence, calculations were also conducted using the B97 functional with, and without, an additional dispersion correction; Figure 5 presents illustrative results for the βMeFuc·p-hydroxy toluene and βMeGal·p-hydroxy toluene complexes. They both indicate a significant change toward more closely bound intermolecular structures when dispersion is taken into account with the hydrogen bonded link from the 8140



Article

i2BASQUE and the Oxford Supercomputing Centre (OSC) and also the computing facility cluster GMPCS of the LUMAT federation (FR LUMAT 2764). Dr Zheng Su contributed some of the early calculations.

an aqueous environment might be driven primarily by desolvation and the hydrophobic effect.34,35 Changes in carbohydrate−aromatic binding affinities in mixtures of aqueous and non-aqueous polar solvents have been interpreted by the Davis group in a similar way.21 On the other hand, Nishio22 has identified bonding through CH−π contacts as the key interaction in aqueous environments, since water will disrupt OH → O and NH → O hydrogen bonds but not dispersion interactions. In the gas phase, when the water is taken away, although hydrogen bonded interactions will dominate, dispersion can still play a role. In the present example, the p-OH → O-y bonds linking Tyr (or p-hydroxy toluene) to the carbohydrate provide an anchor that facilitates further binding, both through OH → O and NH → O hydrogen bonds to the peptide backbone and through CH−π dispersion interactions with the aromatic side group.

■

(1) Varki, A.; Cummings, R.; Esko, J.; Freeze, H.; Hart, G.; Marth, J. Essentials of Glycobiology; Cold Spring Harbor Laboratory Press: New York, 1999. (2) Avci, F. Y.; Kasper, D. L. How Bacterial Carbohydrates Influence the Adaptive Immune System. Annu. Rev. Immunol. 2010, 28, 107− 130. (3) Varki, A. Biological Roles of Oligosaccharides: All of the Theories Are Correct. Glycobiology 1993, 3, 97−130. (4) Weiss, W. I.; Drickamer, K. Structural Basis of LectinCarbohydrate Recognition. Annu. Rev. Biochem. 1996, 65, 441−473. (5) Dwek, R. A. Glycobiology: Toward Understanding the Function of Sugars. Chem. Rev. 1996, 96, 683−720. (6) Bertozzi, C. R.; Kiessling, L. L. Chemical Glycobiology. Science 2001, 291, 2357−2364. (7) Dwek, R. A.; Butters, T. D. Glycobiology - Understanding the Language and Meaning of Carbohydrates. Chem. Rev. 2002, 102, 283− 284. (8) Gabius, H. J.; Siebert, H. C.; Andre, S.; Jiménez-Barbero, J.; Rudiger, H. Chemical Biology of the Sugar Code. ChemBioChem 2004, 5, 740−764. (9) Lemieux, R. U. The Origin of the Specificity in the Recognition of Oligosaccharides by Proteins. Chem. Soc. Rev. 1989, 18, 347−374. (10) Quiocho, F. A. Protein-Carbohydrate Interactions: Basic Molecular Features. Pure Appl. Chem. 1989, 61, 1293−1306. (11) Lis, H.; Sharon, N. Lectins: Carbohydrate-Specific Proteins That Mediate Cellular Recognition. Chem. Rev. 1998, 98, 637−674. (12) Jiménez-Barbero, J.; Asensio, J. L.; Cañada, F. J.; Poveda, A. Free and Protein-Bound Carbohydrate Structures. Curr. Opin. Struct. Biol. 1999, 9, 549−555. (13) Gabius, H. J.; André, S.; Jiménez-Barbero, J.; Romero, A.; Solís, D. From Lectin Structure to Functional Glycomics: Principles of the Sugar Code. Trends Biochem. Sci. 2011, 36, 298−313. (14) Boraston, A. B.; Bolam, D. N.; Gilbert, H. J.; Davies, G. J. Carbohydrate-Binding Modules: Fine-Tuning Polysaccharide Recognition. Biochem. J. 2004, 382, 769−781. (15) Ambrosi, M.; Cameron, N. R.; Davis, B. G. Lectins: Tools for the Molecular Understanding of the Glycocode. Org. Biomol. Chem. 2005, 3, 1593−1608. (16) Hashimoto, H. Recent Structural Studies of CarbohydrateBinding Modules. Cell. Mol. Life Sci. 2006, 63, 2954−2967. (17) Mazik, M. Molecular Recognition of Carbohydrates by Acyclic Receptors Employing Noncovalent Interactions. Chem. Soc. Rev. 2009, 38, 935−956. (18) Asensio, J. L.; Arda, A.; Cañada, F. J.; Jiménez-Barbero, J. Carbohydrate-Aromatic Interactions. Acc. Chem. Res. 2013, 46, 946− 954. (19) Bernardi, A.; Arosio, D.; Potenza, D.; Sanchez-Medina, I.; Mari, S.; Cañada, F. J.; Jiménez-Barbero, J. Intramolecular CarbohydrateAromatic Interactions and Intermolecular van der Waals Interactions Enhance the Molecular Recognition Ability of GMI Glycomimetics for Cholera Toxin. Chem.Eur. J. 2004, 10, 4395−4406. (20) Laughrey, Z. R.; Kiehna, S. E.; Riemen, A. J.; Waters, M. L. Carbohydrate-π Interactions: What Are They Worth? J. Am. Chem. Soc. 2008, 130, 14625−14633. (21) Klein, E.; Ferrand, Y.; Barwell, N. P.; Davis, A. P. Solvent Effects in Carbohydrate Binding by Synthetic Receptors: Implications for the Role of Water in Natural Carbohydrate Recognition. Angew. Chem., Int. Ed. 2008, 47, 2693−2696. (22) Nishio, M. The CH/π Hydrogen Bond in Chemistry. Conformation, Supramolecules, Optical Resolution and Interactions Involving Carbohydrates. Phys. Chem. Chem. Phys. 2011, 13, 13873− 13900.

■

CONCLUSIONS Carbohydrate binding to p-hydroxy toluene and the capped aminoacid, N-acetyl L-tyrosine methyl amide (AcTyrNHMe), investigated under molecular beam conditions, has revealed structures which are dominated by hydrogen bonding but which may also be modulated by CH−π bonded interactions. Complexes of p-hydroxy toluene with α,β methyl D-gluco and Lfucopyranosides form open, doubly hydrogen bonded “insertion” structures, which resemble their microhydrates. With α,β methyl D-galactopyranoside, however, the axial orientation of OH-4 coupled with the exocyclic hydroxymethyl group favors the formation of “addition” structures, linked by a single hydrogen bond, p-OH → O-6, and they adopt stacked structures reinforced by further CH−π interactions. The hydrogen bond provides a flexible anchor, which facilitates the supplementary, longer range binding. The p-OH → O-6 anchor also plays a key role in the most stable complexes of the galactopyranosides with the capped amino acid, AcTyrNHMe, facilitating further binding both through OH → O and NH → O hydrogen bonds to the extended peptide backbone and through CH−π dispersion interactions with the aromatic side group. In the corresponding complexes with toluene or capped phenyl alanine, where there is no p-OH anchor, “stacked” structures do not form, despite their common occurrence in bound carbohydrate−protein structures.

■

REFERENCES

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Present Address ⊥

E.C.S.-K.: Institut für Experimentalphysik, Freie Universität Berlin, Arnimallee 14, 14195 Berlin, Germany. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS This work was supported by the CNRS international collaboration project PICS 5987. Financial support from the Spanish Ministry of Science and Innovation (MICINN, CTQ2011-22923) is gratefully acknowledged, and E.J.C. acknowledges also a “Ramón y Cajal” contract from the MICINN. J.P.S. thanks the Leverhulme Trust for the award of an Emeritus Fellowship. We are grateful for access to supercomputing resources provided by SGI/IZO-SGIker and 8141



Article

(23) Balaji, P. V. Contribution of CH−π Interactions to the Affinity and Specificity of Carbohydrate Binding Sites. Mini Rev. Org. Chem. 2011, 8, 222−228. (24) Lemieux, R. U. How Water Provides the Impetus for Molecular Recognition in Aqueous Solution. Acc. Chem. Res. 1996, 29, 373−380. (25) Roldos, V.; Cañada, F. J; Jiménez-Barbero, J. CarbohydrateProtein Interactions: A 3D View by NMR. ChemBioChem 2011, 12, 990−1005. (26) Raju, R. K.; Ramraj, A.; Hillier, I. H.; Vincent, M. A.; Burton, N. A. Carbohydrate-Aromatic π Interactions: a Test of Density Functionals and the DFT-D Method. Phys. Chem. Chem. Phys. 2009, 11, 3411−3416. (27) Sharma, R.; McNamara, J. P.; Raju, R. K.; Vincent, M. A.; Hillier, I. H.; Morgado, C. A. The Interaction of Carbohydrates and Amino Acids with Aromatic Systems Studied by Density Functional and SemiEmpirical Molecular Orbital Calculations with Dispersion Corrections. Phys. Chem. Chem. Phys. 2008, 10, 2767−2774. (28) Kozmon, S.; Matuška, R.; Spiwok, V.; Koča, J. ThreeDimensional Potential Energy Surface of Selected Carbohydrates’ CH/π Dispersion Interactions Calculated by High-Level Quantum Mechanical Methods. Chem.Eur. J. 2011, 17, 5680−5690. (29) Tsuzuki, S.; Uchimaru, T.; Mikami, M. Magnitude and Nature of Carbohydrate-Aromatic Interactions in Fucose-Phenol and FucoseIndole Complexes: CCSD(T) Level Interaction Energy Calculations. J. Phys. Chem. A 2011, 115, 11256−11262. (30) Kumari, M.; Raghavan, B. S.; Balaji, P. V. Exploration of CH−π Mediated Stacking Interactions in Saccharide: Aromatic Residue Complexes through Conformational Sampling. Carbohydr. Res. 2012, 361, 133−140. (31) Su, Z.; Stanca-Kaposta, E. C.; Cocinero, E. J.; Davis, B. G.; Simons, J. P. Carbohydrate-Aromatic Interactions: a Computational and IR Spectroscopic Investigation of the Complex, Methyl α-LFucopyranoside·Toluene, Isolated in the Gas Phase. Chem. Phys. Lett. 2009, 471, 17−21. (32) Screen, J.; Stanca-Kaposta, E. C.; Gamblin, D. P.; Liu, B.; Snoek, L. C.; Davis, B. G.; Simons, J. P. IR Spectral Signatures of AromaticSugar Complexes: Probing Carbohydrate-Protein Interactions. Angew. Chem., Int. Ed. 2007, 46, 3644−3648. (33) Stanca-Kaposta, E. C.; Gamblin, D. P.; Screen, J.; Liu, B.; Snoek, L. C.; Davis, B. G.; Simons, J. P. Carbohydrate Molecular Recognition: a Spectroscopic Investigation of Carbohydrate-Aromatic Interactions. Phys. Chem. Chem. Phys. 2007, 9, 4444−4451. (34) Cocinero, E. J.; Ç arçabal, P.; Vaden, T. D.; Davis, B. G.; Simons, J. P. Exploring Carbohydrate-Peptide Interactions in the Gas Phase: Structure and Selectivity in Complexes of Pyranosides with NAcetylphenylalanine Methylamide. J. Am. Chem. Soc. 2011, 133, 4548− 4557. (35) Cocinero, E. J.; Ç arçabal, P.; Vaden, T. D.; Simons, J. P.; Davis, B. G. Sensing the Anomeric Effect in a Solvent-Free Environment. Nature 2011, 469, 76−79. (36) Mohamadi, F.; Richards, N. G. J.; Guida, W. C.; Liskamp, R.; Lipton, M.; Caufield, C.; Chang, G.; Hendrikson, T.; Still, W. C. Macromodel: an Integrated Software System for Modelling Organic and Bioorganic Molecules Using Molecular Mechanics. J. Comput. Chem. 1990, 11, 440−467. (37) Zhao, Y.; Truhlar, D. G. Density Functionals for Noncovalent Interaction Energies of Biological Importance. J. Chem. Theory Comput. 2007, 3, 289−300. (38) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M. A.; Cheeseman, J. R.; Montgomery, J. A., Jr.; Vreven, T.; Kudin, K. N.; Burant, J. C.; et al. Gaussian 03, revision C.02; Gaussian, Inc.: Pittsburgh, PA, 2003. (39) Grimme, S. Semi-Empirical GGA-Type Density Functional Constructed with a Long-Range Dispersion Correction. J. Comput. Chem. 2006, 27, 1787−1799. (40) TURBOMOLE, V6.2 2010, a Development of University of Karlsruhe and Forschungszentrum Karlsruhe GmbH, 1989−2007. (41) Ç arçabal, P.; Jockusch, R. A.; Hünig, I.; Snoek, L. C.; Kroemer, R. T.; Davis, B. G.; Gamblin, D. P.; Compagnon, I.; Oomens, J.;

Simons, J. P. Hydrogen Bonding and Cooperativity in Isolated and Hydrated Sugars: Mannose, Galactose, Glucose, and Lactose. J. Am. Chem. Soc. 2005, 127, 11414−11425. (42) Ç arçabal, P.; Patsias, Th.; Hünig, I.; Liu, B.; Kaposta, E. C.; Snoek, L. C.; Gamblin, D. P.; Davis, B. G.; Simons, J. P. Spectral Signatures and Structural Motifs in Isolated and Hydrated Monosaccharides: Phenyl α- and β-L-Fucopyranoside. Phys. Chem. Chem. Phys. 2006, 8, 129−136. (43) Cocinero, E. J.; Stanca-Kaposta, E. C.; Scanlan, E. M.; Gamblin, D. P.; Davis, B. G.; Simons, J. P. Conformational Choice and Selectivity in Singly and Multiply Hydrated Monosaccharides in the Gas Phase. Chem.Eur. J. 2008, 14, 8947−8955. (44) Mayorkas, N.; Rudić, S.; Cocinero, E. J.; Davis, B. G.; Simons, J. P. Carbohydrate Hydration: Heavy Water Complexes of α and β Anomers of Glucose, Galactose, Fucose and Xylose. Phys. Chem. Chem. Phys. 2011, 13, 18671−18678. (45) Gerhards, M.; Unterberg, C.; Gerlach, A.; Jansen, A. β-Sheet Model Systems in the Gas Phase: Structures and Vibrations of Ac-PheNHMe and Its Dimer (Ac-Phe-NHMe)2. Phys. Chem. Chem. Phys. 2004, 6, 2682−2690. (46) Dian, B. C.; Longarte, A.; Mercier, S.; Evans, D. A.; Wales, D. J.; Zwier, T. S. The Infrared and Ultraviolet Spectra of Single Conformations of Methyl-Capped Dipeptides: N-Acetyl Tryptophan Amide and N-Acetyl Tryptophan Methyl Amide. J. Chem. Phys. 2002, 117, 10688−10702. (47) Ç arçabal, P.; Cocinero, E. J.; Simons, J. P. Binding Energies of Micro-Hydrated Carbohydrates: Measurements and Interpretation. Chem. Sci. 2013, 4, 1830−1836.

8142


CarbohydrateâAromatic Interactions: Vibrational Spectroscopy and

Recommend Documents

CarbohydrateâAromatic Interactions: Vibrational Spectroscopy and