IR-spectral signatures of aromatic-sugar complexes: Probing carbohydrate-protein interactions

Article abstract of DOI:10.1002/anie.200605116

(Figure Presented) Key interactions: Sugar-arene complexes have been created in molecular-beam experiments and observed by IR ion-dip spectroscopy in the gas phase. These complexes are powerful models of the selective recognition seen in protein-sugar complexes, for example between the galactose-specific lectin from Artocarpus hirsute and MeGal (see picture).

Full text of DOI:10.1002/anie.200605116

Communications
DOI: 10.1002/anie.200605116
Sugar–Aromatic Complexes
IR-Spectral Signatures of Aromatic–Sugar Complexes:
Probing Carbohydrate–Protein Interactions**
James Screen, E. Cristina Stanca-Kaposta, David P. Gamblin, Bo Liu,
Neil A. Macleod, Lavina C. Snoek,* Benjamin G. Davis,* and John P. Simons*
Angewandte
Chemie
3644
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2007, 46, 3644 –3648

Angewandte
Chemie
Molecular interactions between carbohydrates and pro-
teins—lectins and enzymes—are central to a broad range of
biological processes.[1–3] Binding at carbohydrate-recognition
domains can be mediated by hydrogen-bonding,van der
[1–8]
Waals,and ionic interactions.
Despite the high hydro-
philicity of most sugars,their interactions with proteins can
include binding between aromatic amino acid residues and
the apolar faces or “patches” of pyranose rings (Figure 1).[9]
The planar aromatic platforms of tryptophan,tyrosine,and
less commonly,phenylalanine are often exploited in nature in
carbohydrate-binding modules as selective recognition points
in stacking or “sugar-tong” structural motifs.[9,10]
The near-IR vibrational spectra of individual carbohy-
drate conformers,isolated under molecular-beam conditions
in the gas phase,are extraordinarily sensitive to the local
hydrogen-bond environment of their OH groups. A large shift
of a band towards lower wavenumbers,or conversely,the
absence of a significant shift,provides a sensitive diagnostic
ꢀ
for the incidence or absence of specific OH X intramolecular
hydrogen-bonding interactions.[11] Molecular-beam experi-
ments also provide an ideal “laboratory” for studying
carbohydrate interactions within complexes,through a com-
bination of IR spectroscopy and computational modeling.[11]
A key first step along the computational path to under-
standing such interactions was taken recently by JimØnez-
Barbero and co-workers,who explored the nature of specific
galactose–lectin interactions through a series of density
functional theory (DFT) and ab initio calculations of b-l-
fucose and benzene.[4,7] Here,we describe the first results of
the experimental approach: a direct determination of the
spectroscopic signatures of sugar–arene complexes to identify
the presence or absence of specific hydrogen-bonding inter-
actions and any structural changes in the carbohydrate
promoted by complex formation. Toluene,chosen as the
model aromatic partner,provides a surrogate for the side
chains of phenylalanine and tyrosine—amino acids often
associated with stacking and sugar-tong binding.[9] Represen-
tative carbohydrates chosen for initial investigation included
Figure 1. Representative apolar “patches” in carbohydrates and the
monosaccharide derivatives chosen for this study.
the a-d-gluco- and -galactopyranosides MeGlc and MeGal,
the a-l-fucopyranosides MeFuc and PhFuc,and,in the light
of references [4,7], fucose (Fuc) itself (Figure 1). These were
used to probe: the effects of varying the configuration at C4
and hence the “shapes” of apolar faces,for example,the
B faces of MeGal and MeGlc; the structural influence of an
OH group (OH6) more remote from the apolar face,by
comparison of the toluene complexes of MeGal and MeFuc;
and the influence of the aglycon,by comparison of the
binding signatures in the complexes of Fuc,MeFuc,and
PhFuc (Figure 2).
The UV spectrum of the MeFuc·Tol complex revealed two
structures; their IR spectra (recorded by using the IR ion-dip
[*] J. Screen, Dr. E. C. Stanca-Kaposta, Dr. B. Liu, Dr. N. A. Macleod,
Dr. L. C. Snoek, Prof. J. P. Simons
(IRID) technique[12]) are shown in Figure 2,together with
Department ofChemistry, University ofOxof rd
Physical and Theoretical Chemistry Laboratory
South Parks Road, Oxford, OX13QZ (UK)
Fax: (+44)186-528-5002
E-mail: Lavina.Snoek@chem.ox.ac.uk
John.Simons@chem.ox.ac.uk
[11]
those of PhFuc·Tol,PhFuc,
and MeFuc. In the uncom-
plexed fucosides only the global-minimum conformation is
populated; its structure presents a counterclockwise (cc)
orientation of the hydrogen-bonded chain (OH4!OH3!
OH2!O1). None of the bands in the MeFuc·Tol complex is
displaced towards lower wavenumbers,implying bonding
Dr. D. P. Gamblin, Prof. B. G. Davis
Department ofChemistry, University ofOxof rd
Chemistry Research Laboratory
Mansfield Road, Oxford, OX13TA (UK)
Fax: (+44)186-527-5674
ꢀ
through CH p interactions alone; complexing with the
aromatic ring cannot significantly affect the structure of the
pyranose unit to which it is attached.
E-mail: Ben.Davis@chem.ox.ac.uk
The IRID spectrum of the complex between Fuc and
toluene is shown in Figure 3 together with the (computed)
spectra of the a and b anomers of uncomplexed Fuc. The
Fuc·Tol complex clearly displays a strongly shifted band
[**] Financial support was provided by the EPSRC (Grant GR/T26542),
the Leverhulme Trust (Grant F/08788G), and the Royal Society
(L.C.S., University Research Fellowship). We also acknowledge
support from the CLRC Laser Support Facility, the Physical and
Theoretical Chemistry Laboratory, and Corpus Christi College,
Oxford (L.C.S.).
ꢀ1
located at 3530 cmꢀ1,approximately 75 cm lower in energy
than the next band. This is in sharp contrast with the
MeFuc·Tol complex,where any OH1– p interaction is blocked
and the binding can be attributed solely to dispersive CH–p
Supporting information for this article is available on the WWW
under http://www.angewandte.org or from the author.
Angew. Chem. Int. Ed. 2007, 46, 3644 –3648ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
3645

Communications
and s4—to the weak features lying between 3600 cmꢀ1 and
3640 cmꢀ1.
This identification of a hydrogen-bonding OH–p inter-
action in the bFuc·Tol complex contrasts with the computed
CH–p interaction in the complex with benzene.[4,7] These
computations were initiated from starting geometries based
upon crystal structures of related galactoside–protein com-
plexes taken from the protein database. The computations
generated stable bound complexes,provided dispersive
interactions were taken into account.[7] NMR spectra of
methyl b-galactoside in aqueous solutions of benzene or
phenol were consistent with the retention of similar structures
in the condensed phase,and there was no evidence for specific
hydrogen-bonding interactions involving OH groups.[7] In the
context of this apparent conflict with our observations it
should be noted that restrictions in the calculations of
bFuc·benzene imposed a key starting topology and the
experimental NMR observations utilized the glycoside
MebGal,in which OH1 is unavailable. Nonetheless,the
prevalence of the fucoside rather than free fucose in nature
does highlight Fuc as an anomalous carbohydrate for study.
Unlike galactosides and fucosides,in a-methyl glucoside
(MeGlc) all of the secondary OH groups (OH2–OH4) are
equatorial,potentially allowing access to two parallel hydro-
phobic faces. In protein complexes in which hydrophobic
interactions between glucose and aromatic residues have
been identified,both of the hydrophobic (A and B) faces are
presented to aromatic protein side chains.[13] The IRID
spectrum of the MeGlc·Tol complex is presented in Figure 4
together with the computed spectra of each of the three
lowest-lying conformers of isolated aMeGlc. The isolated
glucoside presents a weakly hydrogen-bonded counterclock-
wise chain (OH4!OH3!OH2!O1) in its three lowest-
lying conformations. In the complex,one of the vibrational
bands is significantly displaced to a lower wavenumber
(approximately 3560 cmꢀ1),indicating a hydrogen-bonding
Figure 2. IRID spectra ofa) MeFuc·Tol, b) PhFuc·Tol, and c) PhFuc,
together with the calculated IR spectrum ofd) “rfee” MeFuc. The
dotted lines indicate the positions ofthe vibrational bands in the
(phenyl-tagged) uncomplexed saccharide, PhFuc, that are associated
with the three OH stretching modes s2, s3, and s4. The dip in the
central band at 3615 cmꢀ1 is due to the absorption ofatmospheric
water which reduces the intensity ofthe IR laser. The ball-and-stick
models next to (c) and (d) correspond to PhFuc and MeFuc,
respectively. These and the other structures shown in this Communica-
tion were generated by DFT calculations.
ꢀ
OH X intermolecular interaction,despite the blocking of the
anomeric hydroxy group,OH1. The bands observed at
3631 cmꢀ1 and 3642 cmꢀ1 lie close to those computed for the
OH modes s3 and s6. The absence of a band around
3600 cmꢀ1,associated in isolated,counterclockwise-oriented
a anomers with the vibration s2,[11] could suggest the
possibility of hydrogen bonding via OH2,in which case the
feature at 3613 cmꢀ1 would correspond to s4. Alternatively,if
the observed H-bonding involved OH4,these assignments
would need to be reversed. The band at highest wavenumber
(3642 cmꢀ1) is characteristic of a noninteracting hydroxy-
methyl group which excludes binding via OH6.
Figure 3. a,c) Computed IR spectra of aFuc (a) and bFuc (c), and
b) the experimental IRID spectrum ofthe complex Fuc·Tol. The
computed structures and conformations of aFuc and bFuc are shown
on the left.
Figure 5 compares the IR spectra associated with MeFuc·
Tol and MeGal·Tol with the (computed) IR spectra of the
lowest-lying conformers of MeGal·Tol and MeGal alone. Like
the fucoside (6-deoxy-l-galactoside),none of the vibrational
bands of MeGal were significantly shifted towards lower
wavenumbers when the sugar was bound to toluene,including
s6,so replacement of the exocyclic methyl (in MeFuc) by a
hydroxymethyl group (in MeGal) did not lead to a hydrogen-
bonding interaction with toluene through OH6. As with
MeFuc,the spectral fingerprint implies a complex bound
solely by dispersive CH–p interactions. The pattern displayed
interactions. The remaining vibrational structure at high
wavenumbers corresponds quite well with that calculated for
the uncomplexed b anomer,suggesting the occurrence of
selective binding,most probably through an OH1– p hydrogen
bond. The displaced band at 3530 cmꢀ1 can then be assigned
to the vibrational mode s1,and the remaining bands— s2, s3,
3646 www.angewandte.org
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2007, 46, 3644 –3648

Angewandte
Chemie
Figure 4. a) IRID spectrum ofthe complex between MeGlc and toluene
recorded experimentally. b–d) Computed IR spectra for the three
lowest-lying conformers of the isolated carbohydrate; computed struc-
tures of the conformers of MeGlc are shown on the left, and the
relative energies (in brackets) are given in kJmolꢀ1. The designation cc
refers to the counterclockwise orientation of the peripheral OH
groups; the designations G-g+, G-g-, and Tg+ indicate the orienta-
tion ofthe exocyclic hydroxymethyl and OH6 groups.
Figure 5. IRID spectra ofa) MeGal·Tol and b) MeFuc·Tol; c) the com-
puted IR spectra ofMeGal·Tol (uniflled bars) and uncomplexed MeGal
(filled bars); the computed conformation of uncomplexed MeGal is
shown as a ball-and-stick model.
by the remaining bands reflects retention of the weakly
hydrogen-bonded counterclockwise chain,with the band at
lowest wavenumber corresponding to the vibration of OH2
(s2),characteristic of an a anomer.[11] Since none of the bands
were significantly perturbed by complex formation,the
conformation adopted by the bound monosaccharide is
likely to be very similar to that preferred by isolated MeGal.
Initial computational analysis,including single-point MP2
counterpoise corrections,has allowed us to suggest structures
for some of these complexes. Analysis of the Tol·MeGal
complex reveals a lowest-lying complex (1 kJmolꢀ1 lower in
energy than the next) with a clear p–A face interaction
(Figure 6). Excitingly,this complex bears a striking similarity
to that observed directly in the 3D structure of MeGal bound
to the galactose-specific lectin from Artocarpus hirsute.[14]
This confirms,at least in part,the validity of the approach
described here for predicting and understanding biologically
relevant hydrophobic sugar–protein interactions.
These investigations were designed to explore and char-
acterize local interactions involved in the molecular recog-
nition of carbohydrates by proteins through binding to
aromatic residues. The possibility of forming stable complexes
involving dispersive “hydrophobic” CH–p interactions has
been established for a-methyl galactoside and its close
analogue, a-methyl fucoside. In fucose itself where the
anomeric hydroxyl group OH1 is no longer blocked and
which is rarely alone as a ligand in nature,the spectra imply a
Figure 6. The closely related A-face-binding mode of MeGal in a) the
galactose-specific lectin from Artocarpus hirsute and b) the DFT-
computed structure ofMeGal·Tol.
Angew. Chem. Int. Ed. 2007, 46, 3644 –3648ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
3647

Communications
provided relative energies. Zero-point corrections to the electronic
energy were calculated using the harmonic frequencies.
complex bound through an OH1–p interaction. A little
surprisingly,a hydrogen-bonding OH– p interaction also
operates in the complex with a-methyl glucoside,where
OH4 is oriented equatorially rather than axially; in this case
the hydrogen-bonding has been associated with OH2 or OH4.
Switching the orientation of OH4 from axial to equatorial
(MeGal!MeGlc) clearly tips the balance in favor of
intermolecular hydrogen bonding. The balance between
hydrogen-bonding and dispersive interactions is therefore a
delicate one,sensitive to the configuration of the carbohy-
drate. In the complexes bound exclusively by dispersive CH–
p interactions the global-minimum carbohydrate structures
are broadly retained. Initial modeling of an arene complex of
a-methyl galactoside suggests that this closely mimics a
known,observed binding mode in protein–sugar interactions.
This and our other results suggest that binding to aromatic
apolar platforms does not seem to necessitate large con-
formational penalties which may in part explain their wide-
spread use by nature,as carbohydrate-binding modules.
Received: December 19,2006
Published online: March 27,2007
Keywords: carbohydrates · density functional calculations ·
.
molecular recognition · vibrational spectroscopy
[1] W. I. Weiss,K. Drickamer, Annu. Rev. Biochem. 1996, 65,441 –
473.
[2] R. A. Dwek, Chem. Rev. 1996, 96,683 – 720.
[3] A. Varki, Glycobiology 1993, 3,97 – 130.
[4] J. JimØnez-Barbero,J. L. Asensio,F. J. Caæada,A. Poveda, Curr.
Opin. Struct. Biol. 1999, 9,549 – 555.
[5] H. Lis,N. Sharon, Chem. Rev. 1998, 98,637 – 674.
[6] R. Palma,M. E. Himmel,J. W. Brady, J. Phys. Chem. B 2000,
104,7228 – 7234.
[7] M. C. Fernµndez-Alonso,F. J. Caæada,J. JimØnez-Barbero,G.
Cuevas, J. Am. Chem. Soc. 2005, 127,7379 – 7386.
[8] M. I. Chµvez,C. Andreu,P. Vidal,N. Aboitiz,F. Freire,P.
Groves,J. L. Asensio,G. Asensio,M. Muraki,F. J. Caæada,J.
JimØnez-Barbero, Chem. Eur. J. 2005, 11,7060 – 7074.
[9] A. B. Boraston,D. N. Bolam,H. J. Gilbert,G. J. Davies,
Biochem. J. 2004, 382,769 – 781.
[10] X. Robert,R. Haser,T. E. Gottschalk,F. Ratajczak,H. Driguez,
B. Svensson,N. Aghajari, Structure 2003, 11,973 – 984.
[11] P. C¸ arꢀabal,T. Patsias,I. Hꢁnig,B. Liu,C. Kaposta,L. C. Snoek,
D. P. Gamblin,B. G. Davis,J. P. Simons, Phys. Chem. Chem.
Phys. 2006, 8,129 – 136.
[12] E. G. Robertson,J. P. Simons, Phys. Chem. Chem. Phys. 2001, 3,
1 – 18.
[13] A. B. Boraston,D. Nurizzo,V. Notenboom,V. Ducros,D. R.
Rose,D. G. Kilburn,G. J. Davies, J. Mol. Biol. 2002, 319,1143 –
1156.
Experimental Section
Methyl-a-l-fucopyranoside was obtained in
a near-quantitative
fashion following a standard Fischer glycosylation from fucose,
using a methanolic solution of HCl generated in situ from acetyl
chloride.
A detailed description of the molecular-beam experiment has
been published previously.[11]
Quantum chemical calculations were performed using the
Gaussian 03 package.[15] Geometry optimizations and calculations of
harmonic vibrational frequencies were conducted using density
functional theory (the B3LYP functional) with a 6-31 + G* basis set.
Single-point MP2 calculations with a larger basis set (6-311 + + G**)
[14] K. N. Rao,C. G. Suresh,U. V. Katre,S. M. Gaikwad,M. I. Khan,
Acta Crystallogr. Sect. D 2004, 60,1404 – 1412.
[15] M. J. Frisch,et al.,see the Supporting Information.
3648 www.angewandte.org
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2007, 46, 3644 –3648