Serine versus threonine glycosylation: The methyl group causes a drastic alteration on the carbohydrate orientation and on the surrounding water shell

Article abstract of DOI:10.1021/ja072181b

Different behavior has been observed for the ψ torsion angle of the glycosidic linkages of D-GalNAc-Ser and D-GalNAc-Thr motifs, allowing the carbohydrate moiety to adopt a completely different orientation. In addition, the fact that the water pockets fou

Full text of DOI:10.1021/ja072181b

Published on Web 07/07/2007
Serine versus Threonine Glycosylation: The Methyl Group
Causes a Drastic Alteration on the Carbohydrate Orientation
and on the Surrounding Water Shell
Francisco Corzana,*,† Jesu´s H. Busto,† Gonzalo Jime´nez-Ose´s,†
Marisa Garc´ıa de Luis,† Juan L. Asensio,‡ Jesu´s Jime´nez-Barbero,§
Jesu´s M. Peregrina,† and Alberto Avenoza*,†
Contribution from Departamento de Qu´ımica, UniVersidad de La Rioja, UA-CSIC, Madre de
Dios 51, E-26006 Logron˜o, Spain, Instituto de Qu´ımica Orga´nica (CSIC), Juan de la CierVa 3,
E-28006 Madrid, Spain, and Centro de InVestigaciones Biolo´gicas (CSIC), Ramiro Maeztu 9,
E-28040 Madrid, Spain
Received March 28, 2007; E-mail: alberto.avenoza@unirioja.es
Abstract: Different behavior has been observed for the ψ torsion angle of the glycosidic linkages
of D-GalNAc-Ser and D-GalNAc-Thr motifs, allowing the carbohydrate moiety to adopt a completely different
orientation. In addition, the fact that the water pockets found in R-D-GalNAc-Thr differ from those obtained
for its serine analogue could be related to the different capability that the two model glycopeptides have
to structure the surrounding water. This fact could have important biological inferences (i.e., antifreeze
activity).
Introduction
It is well-known that the R-O-glycosylation of Ser and Thr
has a profound effect on the underlying peptide backbone,
The most abundant form of O-linked glycoproteins in higher
eukaryotes, termed “mucin-type”, is characterized by R-D-N-
acetylgalactosamine (R-D-GalNAc) attached to the hydroxyl
groups of serine/threonine (Ser/Thr) side chains.1 These gly-
coproteins are involved in fundamental biological processes2
and are attracting a great deal of interest in therapeutic
approaches,3 particularly for the development of vaccines for
cancer treatment.4
forcing it into an extended conformation.5
On the other hand, recent studies have indicated that the
rotation around the glycosidic linkage in R-D-GalNAc-Thr is
restricted, a feature not observed for the Ser-glycosylated
analogues,6 and it has been assumed that this mainly affects
the lateral chain of the amino acid residue. However, to the
best of our knowledge, the key differences concerning the
geometry of the glycosidic linkage and, in particular, the Ψs
torsion angle have not been reported to date. The way in which
these factors could affect the biological behavior of these
molecules has also not been studied.
We have observed in the literature5a,7 that diverse glycopep-
tides present remarkable differences between the 3D-orientation
of the sugar attached to Ser or Thr. Indeed, when Ser is present,
the torsion angle Ψs of the glycosidic linkage takes a value of
around 180°, providing an anti arrangement for the bulky
GalNAc residue. However, in Thr glycopeptides, Ψs frequently
adopts a value close to 120°, resulting in the Hâ-Câ and O1s-
C1s bonds in an eclipsed conformation (see Figures 1 and 2).
Bearing in mind that this torsional angle (Ψs) has a primary
influence on the orientation of the carbohydrate antenna, a
detailed study of the factors that govern the different Ser versus
† Universidad de La Rioja.
‡ Instituto de Qu´ımica Orga´nica (CSIC).
§ Centro de Investigaciones Biolo´gicas (CSIC).
(1) (a) Strous, G. J.; Dekker, J. Crit. ReV. Biochem. Mol. Biol. 1992, 27, 57-
92. (b) Hang, H. C.; Bertozzi, C. R. Bioorg. Med. Chem. 2005, 13, 5021-
5034. (c) Fumoto, M.; Hinou, H.; Ohta, T.; Ito, T.; Yamada, K.; Takimoto,
A.; Kondo, H.; Shimizu, H.; Inazu, T.; Nakahara, Y.; Nishimura, S.-I. J.
Am. Chem. Soc. 2005, 127, 11804-11818.
(2) (a) Dwek, R. A. Chem. ReV. 1996, 96, 683-720. (b) Sears, P.; Wong,
C.-H. Cell Mol. Life Sci. 1998, 54, 223-252. (c) Baldus, S. E.; Engelmann,
K.; Hanisch, F.-G. Crit. ReV. Clin. Lab. Sci. 2004, 41, 189-231. (d)
Hanisch, F.-G.; Mu¨ller, S. Glycobiology 2000, 10, 439-449.
(3) (a) Davis, B. G. Chem. ReV. 2002, 102, 579-601. (b) Watt, G. M.; Lund,
J.; Levens, M.; Kolli, V. S. K.; Jefferis, R.; Boons, G.-J. Chem. Biol. 2003,
10, 807-814. (c) Lui, H.; Wang, L.; Brock, A.; Wong, C.-H.; Schultz, P.
G. J. Am. Chem. Soc. 2003, 125, 1702-1703. (d) Gamblin, D. P.; Garnier,
P.; van Kasteren, S.; Oldham, N. J.; Fairbanks, A. J.; Davis, B. G. Angew.
Chem., Int. Ed. 2004, 116, 827-833. (e) Davis, B. G. Science 2004, 303,
480-482. (f) Dube, D. H.; Bertozzi, C. R. Nat. ReV. Drug DiscoVery 2005,
4, 477-488.
(4) (a) Dube, D. H.; Prescher, J. A.; Quang, C. N.; Bertozzi, C. R. Proc. Natl.
Acad. Sci. U.S.A. 2006, 103, 4819-4824. (b) Dziadek, S.; Kunz, H. The
Chem. Rec. 2004, 3, 308-321. (c) Jaracz, S.; Chen, J.; Kuznetsova, L. V.;
Ojima, I. Bioorg. Med. Chem. 2005, 13, 5043-5054. (d) Danishefsky, S.
J.; Allen, J. R. Angew. Chem., Int. Ed. 2000, 39, 836-863. (e) Slovin, S.
F.; Ragupathi, G.; Musselli, C.; Olkiewicz, K.; Verbel, D.; Kuduk, S. D.;
Schwarz, J. B.; Sames, D.; Danishefsky, S. J.; Livingston, P. O.; Scher, H.
I. J. Clin. Oncol. 2003, 21, 4292-4298. (f) Chen, X.; Lee, G. S.; Zettl, A.;
Bertozzi, C. R. Angew. Chem., Int. Ed. 2004, 43, 6112-6116.
(5) (a) Coltart, D. M.; Royyuru, A. K.; Williams, L. J.; Glunz, P. W.; Sames,
D.; Kuduk, S.; Schwarz, J. B.; Chen, X.-T.; Danishefsky, S. J.; Live, D.
H. J. Am. Chem. Soc. 2002, 124, 9833-9844. (b) Pratt, M. R.; Bertozzi,
C. R. Chem. Soc. ReV. 2005, 34, 58-68.
(6) (a) Mimura, Y.; Inoue, Y.; Maeji, N. J.; Chuˆjoˆ, R. Int. J. Pept. Protein
Res. 1989, 34, 363-368. (b) Naganagowda, G. A.; Gururaja, T. L.;
Satyanarayana, J.; Levine, M. J. J. Pept. Res. 1999, 54, 290-310. (c)
Kindahl, L.; Sandstro¨m, C.; Norberg, T.; Kenne, L. Carbohydr. Res. 2001,
336, 319-323. (d) Kogelberg, H.; Solis, D.; Jime´nez-Barbero, J. Curr. Opin.
Struct. Biol. 2003, 13, 646-653.
(7) (a) Tachibana, Y.; Fletcher, G. L.; Fujitani, N.; Tsuda, S.; Monde, K.;
Nishimura, S.-I. Angew. Chem., Int. Ed. 2004, 43, 856-862. (b) Yeh, Y.;
Feeney, R. E. Chem. ReV. 1996, 96, 601-617. (c) Ben, R. N. ChemBioChem
2001, 2, 161-166.
9
9458
J. AM. CHEM. SOC. 2007, 129, 9458-9467
10.1021/ja072181b CCC: $37.00 © 2007 American Chemical Society

Ser versus Thr Glycosylation
A R T I C L E S
compound 2 was transformed into the R-anomer (compound
6R) with a 58% yield. Transformation of the different groups,
using a similar methodology to that described above, led to
compound 2R in a good yield.
Finally, we used the thioglycoside strategy developed by
Schmidt and co-workers11 to obtain 2â. In this case, we used
compound 3 as the starting material, and the reaction of
compound 3 with the thioglycoside shown in Scheme 1 gave
compound 6â in a moderate yield. Then, the removal of the
Boc group with TFA, and further transformation of the nitro
group into the corresponding amine, gave, after N-acetylation,
derivative 8â. Finally, this compound was transformed into the
desired model glycopeptide 2â in a moderate yield.
Figure 1. Model glycopeptides.
NMR Experiments. The conformational analysis of all the
compounds was carried out using NMR spectroscopy. NOE and
homonuclear J couplings were interpreted with the assistance
of molecular dynamics (MD) simulations. The resulting torsional
angles and the numbering used in this work for the model
compounds are shown in Figure 2.
In a first step, full assignment of the protons in all of the
compounds was carried out using COSY and HSQC experi-
ments. Selective 1D-NOESY experiments in D2O (20 °C, pH
) 5.2) and 2D-NOESY experiments in H2O/D2O (9/1) (20 °C,
pH ) 5.2) were then carried out for the model glycopeptides
(see Supporting Information and Figure 3). Distances involving
NH protons were semiquantitatively determined by integrating
3
Figure 2. Molecular structure of compound 2R, showing definitions of
the torsional angles and the numbering of the atoms. The same definitions
were used for the other model glycopeptides.
the volume of the corresponding cross-peaks. In addition, J
coupling constants were measured from the splitting of the
resonance signals in the 1D spectra and are gathered in Tables
1 and 2. These experimental data were used as restraints in MD
simulations.
The thorough study of the 2D NOESY spectra of 2R, 1â,
and 2â reveals that they present similar NOE patterns. Moreover,
this pattern was similar to that previously reported for the model
peptides (serine and threonine diamides)10 and also for the model
glycopeptide 1R.8 The strong NOE observed between HR and
NH1, along with a medium NOE between NH1 and NH2,
suggests the existence of extended conformations in the peptide
backbone of all the model glycopeptides (Figure 3).
MD Simulations. An experimentally derived ensemble of
2R, 1â, and 2â was obtained by carrying out 80-ns MD-tar12
(MD with time-averaged restraints) simulations including the
coupling constants and distances shown in Tables 1 and 2 as
time-averaged restraints. The simulations were performed
without explicit solvent, but by using a bulk dielectric constant
of 80 to reproduce the water environment. In addition, 4 ns
MD-tar simulations in explicit water [MD-tar (H2O) in Table
1], as well as 20 ns unrestrained MD simulations in explicit
water (see Supporting Information), were also carried out for
compound 2R to compare its hydration shell with that previously
reported10 for the model glycopeptide 1R (see Supporting
Information). As previously observed for 1R,8 the unrestrained
MD simulations failed to accurately reproduce the conformation
of the peptide backbone of 2R, suggesting a folded conformation
(distance NH1-NH2 < distance HR-NH1). This result is
characteristic of the AMBER force field, used in the simulations,
which favors helical structures for small peptides as is described
in the literature.13
Thr conformational behavior of glycopeptides is of paramount
importance. Indeed, these differences could play a significant
role in their distinct biological functions. For example, it is
known that when Ser replaces Thr in a sequence of an antifreeze
protein, the resulting structure cannot act as a freezing inhibitor.7
We report here the synthesis and the conformational study
in aqueous solution of the simplest model glycopeptides derived
from Thr and Ser, compounds 1R, 2R, 1â, and 2â (Figure 1).
Compound 1R has recently been reported by our group,8 while
compounds 1â and 2â were used as models to ascertain whether
the eclipsed conformation mentioned above depends on the
anomeric center.
Results
Synthesis. The synthesis of the target glycosides 2R, 1â, and
2â (Scheme 1) was carried out following the Schmidt procedure9
that involves a Michael-type addition of the amino acid
derivatives 1, 2, or 3 (Scheme 1), which were previously
prepared by our group,10 to tri-O-benzyl-2-nitro-D-galactal.
Therefore, the treatment of protected serine 1 with this nitro-
galactal in the presence of Et3N as a base gave 4â as the major
product. Then, reduction of the nitro group with Raney nickel
T4/H2 and N-acetylation gave derivative 5â, after column
chromatography. Further removal of the benzyl groups afforded
glycoside 1â.
Compound 2R was synthesized in a similar way but by using
tBuOK as a base in the Michael-type addition. Therefore,
(8) Corzana, F.; Busto, J. H.; Jime´nez-Ose´s, G.; Asensio, J. L.; Jime´nez-
Barbero, J.; Peregrina, J. M.; Avenoza, A. J. Am. Chem. Soc. 2006, 128,
14640-14648.
(9) Barroca, N.; Schmidt, R. R. Eur. J. Org. Chem. 1999, 1167-1171.
(10) Corzana, F.; Busto, J. H.; Engelsen, S. B.; Jime´nez-Barbero, J.; Asensio,
J. L.; Peregrina, J. M.; Avenoza, A. Chem. Eur. J. 2006, 12, 7864-7871.
(11) Barroca, N.; Schmidt, R. R. Org. Lett. 2004, 6, 1551-1554.
(12) Pearlman, D. A. J. Biomol. NMR 1994, 4, 1-16.
(13) Gnanakaran, S.; Garcia, A. E. J. Phys. Chem. B 2003, 107, 12555-12557.
9
J. AM. CHEM. SOC. VOL. 129, NO. 30, 2007 9459

A R T I C L E S
Corzana et al.
Scheme 1. Synthetic Routes to 1â, 2R, and 2â.
The calculated 3J coupling constant values were obtained from
the simulations by applying the appropriate Karplus equation14
to the corresponding torsion angles. As can be seen in Tables
1 and 2, both the distances and calculated JH,H values from
the simulations are in very good agreement with the experi-
mental ones.
distribution of the lateral chain (ø1 torsional angle) obtained
for the R and â anomers from the MD-tar simulations is shown
in Figure 5. Interestingly, the most stable rotamer presents a
value for ø1 close to 60° in all of the compounds. This result is
consistent with the medium observed NOE between the aceta-
mide methyl (Me3) of the carbohydrate residue and the methyl
amide (Me1) of the peptide moiety in compounds 1R8 and 2R
(see Supporting Information). As shown in Figure 5, the extra
rigidity of the lateral chain is especially remarkable in compound
2R, in which only the gauche(+) conformer was observed.
3
The distribution for the peptide backbone (Φp/Ψp) of the
model glycopeptides, obtained from the MD-tar simulations, is
shown in Figure 4. It can be seen that, according to the NOE
experiments mentioned above, the Φp/Ψp dihedral values
(backbone) of these models are similar to typical values for
extended conformations, such as PPII and â-sheet, and only a
small number of conformers showed Φp/Ψp dihedral values
corresponding to an R-helical conformation. These results
suggest that the O-glycosylation with GalNAc (via R- or â-) of
Ser and Thr diamides does not significantly affect the conforma-
tion of the peptide backbone of the underlying amino acid.
As far as the lateral chain (ø1 torsional angle) is concerned,
the model glycopeptides derived from Thr, 2R and 2â, show
3JHR,Hâ values that are smaller than those observed for the Ser
analogues; such values have previously been observed for larger
glycopeptides.6
The distribution of the glycosidic linkages (Φs/Ψs) obtained
for all the model glycopeptides from the MD-tar simulations is
shown in Figure 6. In this case, both torsional angles are, in
some way, restricted. With regard to Φs, this angle has a value
of around 60° for the R anomers and close to -60° for the â
derivatives, which is in good agreement with the exo-anomeric
effect. However, Ser and Thr derivatives showed markedly
different behavior in terms of the Ψs dihedral angle. Remark-
ably, in Thr-compounds 2R and 2â, Ψs was rather rigid with
values mainly close to 120°, resulting in the Hâ-Câ and O1s-
C1s bonds in an eclipsed conformation. In contrast, in Ser-
glycopeptides, Ψs was more flexible, providing in most of the
cases an anti arrangement for the GalNAc residue regarding
the peptide moiety. This result was independent of the config-
uration at the anomeric center. Moreover, this result appears to
This fact suggests that rotation around ø1 in the Thr
derivatives is to some extent restricted. On this basis, the
(14) (a) Marco, A.; Llinas, M.; Wuthrich, K. Biopolymers 1978, 17, 617-636.
(b) Vuister, G. W.; Bax, A. J. Am. Chem. Soc. 1993, 115, 7772-7777.
9
9460 J. AM. CHEM. SOC. VOL. 129, NO. 30, 2007

Ser versus Thr Glycosylation
A R T I C L E S
Figure 3. Section of the 800 ms 2D NOESY spectrum (500 MHz) in H2O/D2O (9:1) at 25 °C of model glycopeptide 2R, showing the amide cross-peaks
NH1, NH2, and NH3. Diagonal peaks and exchange cross-peaks connecting NH protons and water are negative (blue color). The NOE contacts are represented
as positive cross-peaks (red color).
Table 1. Comparison of the Experimental and MD Simulation
Derived Distances and 3J Coupling Constants for 2R
Table 2. Comparison of the Experimental and MD Simulation
Derived Distances and 3J Coupling Constants for 1â and 2â
exptla
MD-tar (ꢀ ) 80)
MD-tar (H2O)
compound 1
â
compound 2â
dNH1,NH2
dNH2,NH3
dHR,NH1
dHR,NH2
dHâ,NH1
dHâ,NH2
3JΗR,Hâ
3JHR,NH2
3JH2s,NH3
2.8b
3.3b
2.4b
2.9b
2.8b
3.5b
2.5
2.8
3.1
2.5
2.8
2.6
3.5
2.8c
8.1d
9.2d
2.8
3.2
2.5
2.9
2.6
3.7
3.2c
7.9d
9.4d
exptla
MD-tar (ꢀ ) 80)
exptla
MD-tar (ꢀ ) 80)
dNH1,NH2
dHR,NH1
dHR,NH2
2.9b
2.2b
2.6b
2.6b
2.8b
2.9
2.3
2.8
2.7
3.0b
2.3b
2.8b
2.9
2.5
2.9
dHâproR,NH2
dHâproS,NH2
3JHR,HâproR
3JHR,HâproS
3JHR,NH2
e
2.7
2.5
6.8
6.4c
5.6c
6.5d
9.3d
8.8
9.5
e
3.5
7.4
9.9
3.6c
7.1d
9.3d
6.6
9.6
3JH2s,NH3
a Distances are given in Å and 3J coupling constants in Hz. b Distances
involving NH protons were semiquantitatively determined by integrating
the volume of the corresponding cross-peaks. c Estimated using the Karplus
equation given in ref 14a. d Estimated using the Karplus equation given in
ref 14b.
a Distances are given in Å and 3J values in Hz. b Distances involving
NH protons were semiquantitatively determined by integrating the volume
of the corresponding cross-peaks. c Estimated by using the Karplus equation
given in ref 14a. d Estimated by using the Karplus equation given in ref
14b. e Not determined due to overcrowded NMR spectrum.
be independent of the carbohydrate moiety. A similar result was
obtained for model â-O-glucopeptides previously reported by
our group.10
Solvent Influence on the Conformational Behavior of
Compound 2R and DFT Calculations. Interestingly, the
orientation of the carbohydrate in 2R allows weak hydrogen
bonding between the amide proton of the GalNAc (dubbed NH3)
and the carbonyl oxygen of the O-linked Thr residue (O1). Thus,
while this hydrogen bonding was present for about 8% of the
total trajectory time in 2R, it was never detected for 1R during
the course of the MD simulations.8 This finding was also
reported by Gururaja and co-workers,6b who stated that the
intramolecular hydrogen bonding in Thr-glycopeptides is the
key structure stabilizing element. However, we think that the
As a consequence of the different Ψs values in Thr and Ser
derivatives, the carbohydrate moiety adopts a completely
different orientation. Thus, while in compound 2R, the carbo-
hydrate moiety is almost perpendicular to the peptide backbone;
in 1R the GalNAc adopts a parallel disposition (Figure 7). As
a consequence, the N-acetyl group of the carbohydrate in 2R is
in close proximity to the peptide backbone (see structure of 2R
in Figure 6).
9
J. AM. CHEM. SOC. VOL. 129, NO. 30, 2007 9461

A R T I C L E S
Corzana et al.
Figure 4. Distributions for the simulated peptide backbone (Φp/Ψp) of the model glycopeptides obtained from the MD-tar simulations.
very weak and, as a consequence, it should not be responsible
for the defined conformation of the model glycopeptide. On
the other hand, and due to the proximity of the carbohydrate
and the peptide moieties, it is not possible to accommodate any
water pockets/bridges in the neighborhood of NH3 and O1,
unlike the situation reported for 1R.8 In contrast, an interesting
water pocket was found between NH3 and NH2 (Figure 8a),
which was in good agreement with the NOE observed between
these two protons (Figure 3). A strikingly similar NOE has been
observed in MUC1 derivatives recently synthesized by Kunz
and co-workers.15 The density of this shared water site was 6.2
times the bulk density, having maximum and average residence
times of 9.5 and 1.0 ps. Interestingly, the fact that the water
pockets found in Thr derivative 2R differ from those previously
obtained for its Ser analogue 1R could be related to the different
capabilities that R-D-GalNAc-Ser and R-D-GalNAc-Thr motifs
have to structure the surrounding water. Therefore, this finding
could have important biological implications (i.e., antifreeze
activity).7a In this sense, although the mechanism of action of
the antifreeze proteins at molecular level still remains to be
elucidated, the current hypothesis indicates that the antifreeze
activity could be related with the irreversible binding of the
antifreeze molecules to the ice surface through a hydrogen-
bonding network.7b Presumably, and attending to the fact that
the water pockets found in 2R are more persistent than those
observed in 1R, this hydrogen-bonding network should be more
efficient in 2R, which could explain to some extent why the
Thr residues are necessary to maintain the antifreeze activity.
Figure 5. Distributions for the lateral chain (ø1) of the model glycopeptides
1R and 2R (a) and 1â and 2â (b) obtained from the MD-tar simulations.
absence of NOEs between the N-acetyl group of the GalNAc
and NH1 and/or Me1 indicates that the hydrogen bond must be
(15) Dziadek, S.; Griesinger, C.; Kunz, H.; Reinscheid, U. M. Chem. Eur. J.
2006, 12, 4981-4993.
9
9462 J. AM. CHEM. SOC. VOL. 129, NO. 30, 2007

Ser versus Thr Glycosylation
A R T I C L E S
Figure 6. Distributions of the glycosidic linkage (Φs/Ψs) of the model glycopeptides obtained from the MD-tar simulations. Newman projections of the
Câ-O1s bond are included into the diagrams.
mentioned above, the inclusion of water molecules from the
first solvation shell was required to obtain a structure compatible
with the experimental data. As can be seen in Figure 8b, the
DFT optimized structure of model glycopeptide 2R shows an
extended conformation for the backbone and an eclipsed
conformation around ψs (the alternate one could not be located
through this methodology). Notably, this result corroborates the
conformational preferences obtained from the MD simulations.
Furthermore, DFT calculations demonstrated that the existence
of bridging water molecules is not only possible but necessary
to stabilize the experimentally observed geometries.
Although it seems rather unusual that a rotamer containing
eclipsed atoms is found to be the most stable one, a survey of
Figure 7. Calculated ensembles for compounds 1R (a) and 2R (b) obtained
from the MD-tar simulations, showing the RMSDs for heavy atom
superimposition.
the Cambridge Structural Data Base revealed several structures
(CSD refcodes: DMGALP, RONHEH, and ZOSSEF) that
contain eclipsed H-C-O-C torsional angles.16 Moreover, some
carbohydrates containing eclipsed structures have been recently
reported, from a theoretical point of view.17 To gain some
insights into the behavior of the glycosidic linkage in Ser/Thr
model glycopeptides, we carried out a thorough theoretical study
of their intrinsic conformational preferences using the reduced
models shown in Figure 9.
Relaxed PES scans at the B3LYP/6-31G(d,p) level along the
Ψs dihedral angle were performed for these models of 1R and
2R. As can be seen in Figure 9b, the double-minimum potential
calculated for the R-Glyco-Ser model agrees quite well with
the experimental data previously reported for R-D-GalNAc-Ser,
On the other hand, our experience8 has demonstrated that the
combination of several methodologies is required to reach
reasonable conclusions consistent with the experimental data.
The combination of MD-tar simulations in explicit water
followed by DFT optimization of frames of the MD trajectory
has shown to be the most expedient and practical procedure to
obtain accurate geometries of the model glycopeptides together
with the first solvation shell. This protocol allows us to evaluate
both the influence of the surrounding water in the solute
geometry, and also the influence of the solute on the organization
of the surrounding water. Therefore, a reliable structure for
compound 2R was obtained, according to our previously
established protocol,8 by optimizing, through DFT methods, a
frame of the MD simulations in explicit water, in which the
model glycopeptide shows an extended conformation. As
(16) Allen, F. H.; Davis, J. E.; Galloy, J. J.; Johnson, O.; Kennard, O.; Macrae,
C. F.; Mitchell, E. M.; Mitchell, G. F.; Smith, J. M.; Watson, D. G. J.
Chem. Inf. Comput. Sci. 1991, 31, 187-204.
(17) Csonka, G. I.; Schubert, G. A.; Perczel, A.; Sosa, C. P.; Csizmadia, I. G.
Chem. Eur. J. 2002, 8, 4718-4733.
9
J. AM. CHEM. SOC. VOL. 129, NO. 30, 2007 9463

A R T I C L E S
Corzana et al.
Figure 8. (a) Two-dimensional radial pair distribution function for N2 and
N3 found in the 4 ns MDH2O-tar simulation of 2R. (b) Calculated B3LYP/
6-31G(d) geometry of compound 2R, including the surrounding water
molecules. Distances are given in ångstroms and torsion angles in degrees.
Figure 9. (a) Three-dimensional structure of the reduced model for 1R
and 2R. (b) PES of R-Glyco-Ser and R-Glyco-Thr reduced models. (c) Long-
range effects in R-Glyco-Ser and R-Glyco-Thr reduced models. All quantities
were calculated at the B3LYP/6-31G(d,p) level.
with the alternate conformation (ψs ) 190°) being the most
stable one. In the case of the R-Glyco-Thr model, only one
minimum with an eclipsed conformation (ψs ) 105°) was found.
In this case, the alternate conformations were clearly inacces-
sible.
the NBO partition scheme.18 These interactions were calculated
for both the rotation along Ψs (torsional effects) and selected
areas of the ring and amino acid moieties (long-range effects),
as depicted in the Supporting Information. Hence, whereas the
smallest torsional effects were located at alternate conformations
around Ψs, long-range interactions were minima at the eclipsed
ones in both structures. The driving force for the extra
stabilization of the R-Glyco-Thr model at the eclipsed confor-
mations is, therefore, the greater long-range effects arising at
values of Ψs close to 180°, which arise due to the growing steric
repulsions between the ring moiety (principally the lone pairs
of the endocyclic oxygen O5s) and the â-methyl group (Figures
9c and 10). In contrast to the R-Glyco-Ser model, these long-
range interactions clearly overcome the torsional preferences
Origin of the Conformational Preferences. We would like
to point out that the influence of the methyl group on the
torsional energy profiles around the Ψ angles is not at the origin
of the different orientation detected for the carbohydrate moiety
in compounds 1R and 2R. Thus, the minima of the torsional
energy of Ψs, according to AMBER/GLYCAM force field, lay
on alternate conformations in both derivatives (see Supporting
Information). Therefore, the conformational preferences must
be related to other additional stereoelectronic factors other than
the simple torsional energy. Since these effects are not easy to
evaluate from the MD simulations, the DFT study was carried
out. An effort to locate the source of these subtle conformational
differences was made by evaluating the steric and delocalizing
(hyperconjugative) interactions in both substrates, by means of
(18) Glendening, E. D.; Badenhoop, J. K.; Reed, A. E.; Carpenter, J. E.;
Bohmann, J. A.; Morales, C. M.; Weinhold, F. NBO 5.0; Theoretical
Chemistry Institute, University of Wisconsin: Madison, WI, 2001.
9
9464 J. AM. CHEM. SOC. VOL. 129, NO. 30, 2007

Ser versus Thr Glycosylation
A R T I C L E S
dissolved in THF (5 mL) under argon, and freshly activated molecular
sieve (3 Å, 0.6 g) was then added. After the reaction mixture stirred at
25 °C for 30 min, Et3N (1 mL, 6.9 mmol) was added, and stirring was
continued for 24 h. Acetic acid (0.5 mL) was used to acidify the reaction
mixture, the molecular sieve was filtered off, and all solvents were
removed by evaporation. The residue was purified by silica gel column
chromatography (ethyl acetate/methanol, 98:2) to give a mixture of
1
4R and 4â in a 3:7 ratio (110 mg), as an oil, in 38% yield. H NMR
(400 MHz, CDCl3) data for this mixture follow: δ 1.99, 1.98 (s, 3H),
2.76, 2.73 (d, J ) 4.8 Hz, 3H), 3.48-3.61 (m, 1H), 3.63-3.71 (m,
1H), 3.89-3.95 (m, 1H), 3.99-4.00 (m, 1H), 4.32-4.62 (m, 6H), 4.66-
4.74 (m, 2H), 4.79-4.87 (m, 2H), 5.02 (dd, J ) 12.0 Hz, J ) 4.0 Hz,
1H), 5.38 (d, J ) 4.0 Hz, 1H), 6.17-6.31 (m, 1H), 6.38-6.51 (m,
1H), 7.20-7.38 (m, 15H). Anal. Calcd for C33H39N3O9: C, 63.76; H,
6.32; N, 6.76. Found: C, 63.85; H, 6.30; N, 6.81.
Figure 10. Three-dimensional plots of the σC-H (â-methyl group) and n1
(endocyclic oxygen) NBOs of the R-Glyco-Thr reduced model, calculated
at the B3LYP/6-31G(d,p) level, corresponding to representative (a) alternate
(Ψs ) 180°) and (b) eclipsed (Ψs ) 120°) conformations.
O
Synthesis of Compound 5â. Platinized Raney-nickel (T4) catalyst
was freshly prepared as described in the literature.19 The catalyst
obtained by using 2 g of Raney nickel/aluminum alloy was suspended
in ethanol (10 mL) and pre-hydrogenated for 10 min before the addition
of the mixture of glycosides 4R and 4â (110 mg, 0.18 mmol) in ethanol
(7 mL). The reaction mixture was shaken under H2 (1 atm) for 3 h at
25 °C. The catalyst was filtered off and the solvent evaporated. The
residue was dissolved in pyridine/acetic anhydride (2:1, 6 mL) and
stirred at 25 °C for 3 h. Removal of the volatiles followed by a silica
gel column chromatographic purification (ethyl acetate/methanol, 95:
5) gave an oily residue corresponding to â-glycoside 5â (63 mg) in
55% yield. 1H NMR (400 MHz, CDCl3): δ 1.91 (s, 3H), 1.98 (s, 3H),
2.69 (d, J ) 4.8 Hz, 3H), 3.59-3.67 (m, 4H), 3.80-3.87 (m, 2H),
3.95 (s, 1H), 4.04 (dd, J ) 10.5 Hz, J ) 5.2 Hz, 1H), 4.41-4.54 (m,
4 H), 4.57 (d, J ) 11.4 Hz, 1H), 4.67 (d, J ) 11.9 Hz, 1H), 4.73 (d,
J ) 7.7 Hz, 1H), 4.88 (d, J ) 11.4 Hz, 1H), 5.54 (d, J ) 7.1 Hz, 1H),
6.64 (d, J ) 3.5 Hz, 1H), 6.72 (d, J ) 6.4 Hz, 1H), 7.27-7.39 (m,
15H). 13C NMR (100 MHz, CD3OD): δ 23.1, 23.7, 26.3, 52.6, 53.4,
68.4, 68.5, 71.8, 71.9, 73.3, 73.5, 74.5, 78.0, 101.2, 127.9, 128.1, 128.1,
128.3, 128.5, 128.6, 137.6, 137.8, 138.3, 170.2, 170.4, 171.0. Anal.
Calcd for C35H43N3O8: C, 66.33; H, 6.84; N, 6.63. Found: C, 66.41;
H, 6.91; N, 6.58.
in the R-Glyco-Thr one and consequently force it to adopt an
eclipsed conformation close to 120°.
Conclusions. We have synthesized and carried out a thorough
conformational analysis of the simplest model glycopeptides.
Strikingly different behavior was observed for the ψ torsion
angle of the glycosidic linkages of D-GalNAc-Ser (alternate
conformations) or D-GalNAc-Thr (eclipsed conformations)
motifs present in natural glycopeptides. Moreover, we have
demonstrated that this exceptional behavior of the glycosidic
linkage is independent of the anomeric center (R or â). On the
other hand, although in the Thr derivative 2R the carbohydrate
and backbone moieties are closer than in the Ser derivative 1R,
the hydrogen bond between these moieties is very weak, as
inferred from NMR data and MD simulations. Consequently,
hydrogen bonding cannot be responsible for the extended
conformation of the backbone. In addition, the rigidity of 2R
is, as for 1R, well explained by the presence of water pockets/
bridges between the carbohydrate and the peptide moieties.
Interestingly, the water pockets found in 2R differ from those
previously deduced for its Ser analogue. This finding suggests
that the R-D-GalNAc-Ser and R-D-GalNAc-Thr motifs structure
the surrounding water in rather different ways, which could
explain the loss of activity of an antifreeze protein (R-D-GalNAc-
Ser/Thr-Ala-Ala) when the Thr is replaced by the Ser.7 Finally,
a DFT study allowed us to identify the stereoelectronic origin
of the conformational preferences for 2R by evaluating both
the torsional and the long-range effects on the Ψs values. The
obtained results indicate that the different behavior observed
for the glycosidic linkage in the Thr versus Ser derivatives can
be explained in terms of steric repulsions between the carbo-
hydrate moiety (endocyclic oxygen) and the â-methyl group in
the former, forcing them to be located far away from each other
in the molecule, which causes an eclipsed conformation of the
Ψs torsion angle.
Synthesis of Compound 1â. To a solution of glycoside 5â (25 mg,
0.04 mmol) in ethyl acetate/methanol (1:1) (3 mL) was added 10%
palladium-carbon (12 mg) as a catalyst. The reaction mixture was
shaken under H2 (1 atm) for 12 h at 25 °C. Removal of the catalyst
and a further purification of the residue with C18 reverse-phase sep-
pak cartridge gave 1â (12 mg), as a colorless oil, in 80% yield. [R]25
) -0.5 (c ) 1.29, MeOH). H NMR (400 MHz, D2O): δ 1.99 (s,
D
1
3H), 2.02 (s, 3H), 2.70 (s, 3H), 3.61-3.70 (m, 2H), 3.71-3.79 (m,
1
2H), 3.80-3.93 (m, 3H), 3.96-4.04 (m, 1H), 4.36-4.46 (m, 2H). H
NMR (400 MHz, H2O/D2O, 9:1) for the region of amides: δ 7.96-
8.02 (m, 1H), 8.12 (d, J ) 9.6 Hz, 1H), 8.24 (d, J ) 6.6 Hz, 1H). 13
C
NMR (100 MHz, D2O): δ 24.1, 24.6, 28.3, 54.6, 56.3, 63.4, 70.1, 70.4,
73.2, 77.5, 103.7, 174.1, 176.8, 177.1. Anal. Calcd for C14H25N3O8:
C, 46.28; H, 6.93; N, 11.56. Found: C, 46.30; H, 6.99; N, 11.52.
Synthesis of Compound 6R. Compound 2 (212 mg, 1.22 mmol)
and tri-O-benzyl-2-nitrogalactal (468 mg, 1.01 mmol) were dissolved
in THF (15 mL) under argon, and freshly activated molecular sieve (3
Å, 0.3 g) was then added. After the reaction mixture was stirred at 25
Experimental Section
t
°C for 30 min, 1 M BuOK solution in THF (102 µL, 0.1 mmol) was
General Procedures. Solvents were purified according to standard
procedures. Analytical TLC was performed using Polychrom SI F254
plates. Column chromatography was performed using silica gel 60
(230-400 mesh). Melting points were determined on a Bu¨chi B-545
melting point apparatus and are uncorrected. Optical rotations were
measured on a Perkin-Elmer 341 polarimeter. Microanalyses were
carried out on a CE Instruments EA-1110 analyzer and are in good
agreement with the calculated values. 1H and 13C NMR data were
obtained on a Bruker Avance 400 MHz spectrometer.
added, and stirring was continued for 24 h. Acetic acid (0.5 mL) was
used to acidify the reaction mixture, the molecular sieve was filtered
off, and all solvents were removed by evaporation. The residue was
purified by silica gel column chromatography (ethyl acetate/methanol,
98:2) to give only the R-anomer 6R (370 mg), as an oil, in 58% yield.
1
[R]25 ) +112.9 (c ) 1.45, CHCl3). H NMR (400 MHz, CDCl3): δ
D
1.16 (d, J ) 8.0 Hz, 3H), 2.05 (s, 3H), 2.80 (d, J ) 4.0 Hz, 3H),
3.48-3.62 (m, 2H), 3.98-4.07 (m, 2H), 4.30 (dd, J ) 8.0 Hz, J ) 4.0
Synthesis of Compounds 4R and 4â. Compound 1 (100 mg, 0.62
mmol) and tri-O-benzyl-2-nitrogalactal (214 mg, 0.46 mmol) were
(19) Nishimura, S. Bull. Chem. Soc. Jpn. 1959, 32, 61-64.
9
J. AM. CHEM. SOC. VOL. 129, NO. 30, 2007 9465

A R T I C L E S
Corzana et al.
Hz, 1H), 4.32-4.51 (m, 5H), 4.68 (dd, J ) 20.0 Hz, J ) 12.0 Hz,
2H), 4.82 (d, J ) 12.0 Hz, 1H), 5.03 (dd, J ) 8.0 Hz, J ) 4.0 Hz,
1H), 5.56 (d, J ) 4.0 Hz, 1H), 6.21-6.26 (m, 1H), 6.38 (d, J ) 8.0
Hz, 1H), 7.19-7.37 (m, 15H). 13C NMR (100 MHz, CDCl3): δ 17.5,
23.3, 26.4, 56.6, 68.3, 70.2, 72.8, 72.8, 73.7, 74.1, 75.1, 84.9, 96.6,
127.8, 128.0, 128.1, 128.2, 128.4, 128.5, 128.6, 137.1, 137.7, 137.9,
169.2, 170.5. Anal. Calcd for C34H41N3O9: C, 64.24; H, 6.50; N, 6.61.
Found: C, 64.30; H, 6.47; N, 6.58.
Calcd for C37H47N3O10: C, 64.05; H, 6.83; N, 6.06. Found: C, 64.11;
H, 6.80; N 6.02.
Synthesis of Compound 7â. Derivative 6â (150 mg, 0.22 mmol)
was dissolved in CH2Cl2/TFA (1:1, 6 mL), and the solution was stirred
at 25 °C for 12 h. The solvent was evaporated, and the residue was
dissolved in pyridine/acetic anhydride (2:1, 6 mL) and stirred at 25 °C
for 4 h. Removal of the volatiles and a further silica gel column
chromatographic purification (CH2Cl2/methanol, 15:1) gave a white
solid corresponding to 7â (110 mg) in 80% yield. Mp 159-161 °C;
Synthesis of Compound 7R. Platinized Raney-nickel (T4) catalyst
was freshly prepared as described in the literature.19 The catalyst
obtained using 2 g of Raney nickel/aluminum alloy was suspended in
ethanol (10 mL) and pre-hydrogenated for 10 min before the addition
of 6R (190 mg, 0.30 mmol) in ethanol (8 mL). The reaction mixture
was shaken under H2 (1 atm) for 3 h at 25 °C. The catalyst was filtered
off and the solvent evaporated. The residue was dissolved in pyridine/
acetic anhydride (2:1, 6 mL) and stirred at 25 °C for 4 h. Removal of
the volatiles and a further silica gel column chromatographic purification
1
[R]25 ) +38.0 (c ) 1.05, MeOH). H NMR (400 MHz, CDCl3): δ
D
1.01 (d, J ) 6.4 Hz, 3H), 1.98 (s, 3H), 2.72 (d, J ) 4.8 Hz, 3H),
3.58-3.63 (m, 2H), 3.77 (t, J ) 6.6 Hz, 1H), 3.97-4.01 (m, 1H), 4.08
(dd, J ) 10.6 Hz, J ) 2.7 Hz, 1H), 4.14-4.23 (m, 1H), 4.43-4.55
(m, 4H), 4.57 (dd, J ) 6.2 Hz, J ) 3.7 Hz, 1H), 4.61 (d, J ) 11.6 Hz,
1H), 4.81-4.89 (m, 2H), 4.98 (d, J ) 8.1 Hz, 1H), 6.38-6.43 (m,
1H), 6.59 (d, J ) 6.4 Hz, 1H), 7.22-7.39 (m, 15H). 13C NMR (100
MHz, CDCl3): δ 15.2, 23.1, 26.2, 55.6, 67.6, 71.6, 72.4, 73.5, 73.8,
75.0, 76.3, 79.1, 87.4, 100.0, 127.7, 127.8, 128.0, 128.2, 128.3, 128.4,
128.4, 128.5, 128.6, 136.4, 137.3, 137.7, 168.7, 170.0. Anal. Calcd for
C34H41N3O9: C, 64.24; H, 6.50; N, 6.61. Found: C, 64.29; H, 6.55; N,
6.57.
(ethyl acetate/methanol, 95:5) gave an oil corresponding to 7R (139
1
mg) in 72% yield. [R]25 ) +9.6 (c ) 1.08, CHCl3/MeOH, 1:1). H
D
NMR (400 MHz, CDCl3): δ 1.20 (d, J ) 6.4 Hz, 3H), 2.03 (s, 3H),
2.04 (s, 3H), 2.76 (d, J ) 4.6 Hz, 3H), 3.53-3.58 (m, 2H), 3.67 (dd,
J ) 10.9 Hz, J ) 1.9 Hz, 1H), 3.90-4.00 (m, 2H), 4.13 (dd, J ) 6.0
Hz, J ) 2.4 Hz, 1H), 4.38-4.44 (m, 2H), 4.45-4.59 (m, 3H), 4.60-
4.65 (m, 1H), 4.73 (d, J ) 12.0 Hz, 1H), 4.92-5.00 (m, 2H), 6.18 (d,
J ) 9.2 Hz, 1H), 6.30-6.35 (m, 1H), 6.42 (d, J ) 8.3 Hz, 1H), 7.24-
7.38 (m, 15H). 13C NMR (100 MHz, CDCl3): δ 17.9, 23.2, 23.4, 26.3,
49.6, 56.7, 69.3, 70.5, 71.5, 72.7, 73.7, 74.4, 76.4, 77.2, 99.7, 127.6,
127.6, 127.7, 127.8, 127.9, 128.2, 128.3, 128.4, 128.5, 137.7, 138.2,
138.4, 170.6, 170.7. Anal. Calcd for C36H45N3O8: C, 66.75; H, 7.00;
N, 6.49. Found: C, 66.64; H, 7.06; N, 6.44.
Synthesis of Compound 8â. Platinized Raney-nickel (T4) catalyst
was freshly prepared as described in the literature.19 The catalyst
obtained using 1 g of Raney nickel/aluminum alloy was suspended in
ethanol (5 mL) and pre-hydrogenated for 10 min before the addition
of 7â (100 mg, 0.16 mmol) in ethanol (5 mL). The reaction mixture
was shaken under H2 (1 atm) for 3 h at 25 °C. The catalyst was filtered
off and the solvent evaporated. The residue was dissolved in pyridine/
acetic anhydride (2:1, 6 mL) and stirred at 25 °C for 4 h. Removal of
the volatiles and a further silica gel column chromatographic purification
(CH2Cl2/methanol, 9:1) gave an oil corresponding to 8â (46 mg) in
Synthesis of Compound 2R. To a solution of glycoside 7R (87 mg,
0.13 mmol) in ethyl acetate/methanol (1:1) (10 mL) was added 10%
palladium-carbon (40 mg), as a catalyst. The reaction mixture was
shaken under H2 (1 atm) for 12 h at 25 °C. Removal of the catalyst
and further purification of the residue with C18 reverse-phase sep-pak
1
45% yield. [R]25 ) +24.2 (c ) 0.58, CHCl3). H NMR (400 MHz,
D
CDCl3): δ 1.02 (d, J ) 6.4 Hz, 3H), 1.89 (s, 3H), 1.97 (s, 3H), 2.71
(d, J ) 4.8 Hz, 3H), 3.59-3.64 (m, 2H), 3.66-3.71 (m, 1H), 3.79-
3.99 (m, 3H), 4.13-4.20 (m, 1H), 4.40-4.48 (m, 3H), 4.51-4.55 (m,
1H), 4.58 (d, J ) 11.4 Hz, 1H), 4.69 (d, J ) 11.9 Hz, 1H), 4.76 (d, J
) 8.1 Hz, 1H), 4.91 (d, J ) 11.3 Hz, 1H), 5.25 (d, J ) 7.7 Hz, 1H),
6.72 (d, J ) 6.1 Hz, 1H), 6.74-6.80 (m, 1H), 7.27-7.38 (m, 15H).
13C NMR (100 MHz, CDCl3): δ 16.0, 23.3, 23.7, 26.4, 53.7, 56.2,
68.6, 71.9, 72.3, 73.5, 73.7, 74.9, 75.1, 78.2, 101.1, 127.8, 128.0, 128.3,
128.4, 128.6, 128.8, 137.8, 138.0, 138.6, 169.4, 170.2, 170.6. Anal.
Calcd for C36H45N3O8: C, 66.75; H, 7.00; N, 6.49. Found: C, 66.83;
H, 7.04; N, 6.47.
cartridge gave 2R (45 mg), as a colorless oil, in 88% yield. [R]25
)
D
+29.5 (c ) 1.46, MeOH). 1H NMR (400 MHz, D2O): δ 1.22 (d, J )
4.0 Hz, 3H) 2.02 (s, 3H), 2.12 (s, 3H), 2.70 (s, 3H), 3.68-3.74 (m,
2H), 3.83 (dd, J ) 11.0 Hz, J ) 3.1 Hz, 1H), 3.92-4.01 (m, 2H), 4.07
(dd, J ) 11.1 Hz, J ) 3.8 Hz, 1H), 4.29-4.43 (m, 2H), 4.86 (d, J )
1
3.8 Hz, 1H). H NMR (400 MHz, H2O/D2O, 9:1) for the region of
amides: δ 8.09-8.11 (m, 1H), 8.32 (d, J ) 9.5 Hz, 1H), 8.59 (d, J )
8.8 Hz, 1H). 13C NMR (100 MHz, CD3OD): δ 18.9, 22.8, 23.1, 27.0,
50.8, 58.9, 62.3, 68.6, 69.5, 72.3, 75.4, 99.5, 173.0, 175.1, 175.9. Anal.
Calcd for C15H27N3O8: C, 47.74; H, 7.21; N, 11.13. Found: C, 47.70;
H, 7.20; N, 11.11.
Synthesis of Compound 2â. To a solution of glycoside 8â (27 mg,
0.04 mmol) in ethyl acetate/methanol (4:1) (5 mL) was added 10%
palladium-carbon (20 mg), as a catalyst. The reaction mixture was
shaken under H2 (1 atm) for 7 h at 25 °C. Removal of the catalyst and
further purification of the residue with C18 reverse-phase sep-pak
Synthesis of Compound 6â. A mixture of 3 (182 mg, 0.78 mmol),
NIS (186 mg, 0.83 mmol), 4 Å molecular sieve, and the thiodonor
shown in Scheme 1 (302 mg, 0.53 mmol) in CH2Cl2 (12 mL) was stirred
for 1 h sheltered from light at 25 °C under dry Ar, then cooled to 0 °C.
TMSOTf (23 µL, 0.14 mmol) was added, and the mixture was stirred
for 20 min at 0 °C and then 12 h at 25 °C. Et3N (50 µL) was added,
and the mixture was diluted with CH2Cl2, filtered through a pad of
Celite, washed with 5% aqueous Na2S2O3, dried (MgSO4), and
concentrated. The solvent was removed by evaporation, and the products
were separated by flash chromatography on silica gel (ethyl acetate/
hexane, 2:3) to give 6â (75 mg), as a yellow solid, in 20% yield. Mp
162-164 °C; [R]25D ) +33.8 (c ) 1.00, MeOH). 1H NMR (400 MHz,
CDCl3): δ 1.04 (d, J ) 6.4 Hz, 3H), 1.43 (s, 9H), 2.67 (d, J ) 4.8 Hz,
3H), 3.58 (d, J ) 6.2 Hz, 2H), 3.68-3.74 (m, 1H), 3.94-3.98 (m,
1H), 4.02-4.08 (m, 1H), 4.20-4.30 (m, 2H), 4.44-4.50 (m, 3H), 4.53
(d, J ) 11.2 Hz, 1H), 4.61 (d, J ) 11.6 Hz, 1H), 4.79-4.94 (m, 3H),
5.53 (d, J ) 6.8 Hz, 1H), 6.34-6.39 (m, 1H), 7.25-7.36 (m, 15H).
13C NMR (100 MHz, CDCl3): δ 15.7, 26.2, 28.3, 56.9, 67.9, 71.7,
72.5, 73.6, 73.8, 75.0, 77.2, 79.2, 87.6, 100.0, 127.9, 128.0, 128.1, 128.2,
128.3, 128.4, 128.5, 128.5, 128.6, 136.4, 137.4, 137.7, 169.2. Anal.
cartridge gave 2â (8 mg), as a colorless oil, in 53% yield. [R]25
)
D
-0.2 (c ) 0.55, H2O). 1H NMR (400 MHz, D2O): δ 1.03 (d, J ) 6.2
Hz, 3H), 1.94 (s, 3H), 1.98 (s, 3H), 2.64 (s, 3H), 3.48-3.53 (m, 1H),
3.59 (dd, J ) 10.9 Hz, J ) 3.3 Hz, 1H), 3.62-3.68 (m, 2H), 3.72 (dd,
J ) 10.8 Hz, J ) 8.5 Hz, 1H), 3.81 (d, J ) 3.1 Hz, 1H), 4.16-4.23
1
(m, 2H), 4.31 (d, J ) 8.4 Hz, 1H). H NMR (400 MHz, H2O/D2O,
9:1) for the region of amides: δ 7.79-7.82 (m, 1H), 8.02 (d, J ) 7.4
Hz, 1H), 8.14 (d, J ) 9.5 Hz, 1H). 13C NMR (100 MHz, D2O): δ
18.8, 24.3, 24.7, 28.4, 55.0, 60.6, 63.5, 70.1, 73.3, 77.3, 77.5, 103.0,
174.6, 177.3, 177.5. Anal. Calcd for C15H27N3O8: C, 47.74; H, 7.21;
N, 11.13. Found: C, 47.68; H, 7.18; N, 11.18.
NMR Experiments. All the NMR experiments were recorded on a
Bruker Avance 400 spectrometer at 293 K, except for the 1D and 2D
NOESY spectra of compound 2R, which were recorded at 278 K on a
Varian Unity 500 spectrometer. 1H and 13C NMR spectra were recorded
in CDCl3 and CD3OD with TMS as the internal standard and in D2O
(chemical shifts are reported in ppm on the δ scale). Magnitude-mode
ge-2D COSY spectra were recorded with gradients and using the
9
9466 J. AM. CHEM. SOC. VOL. 129, NO. 30, 2007

Ser versus Thr Glycosylation
A R T I C L E S
cosygpqf pulse program with 90° pulse width. Phase-sensitive ge-2D
HSQC spectra were recorded using z-filter and selection before t1
removing the decoupling during acquisition by use of invigpndph pulse
program with CNST2 (JHC) ) 145. 2D NOESY experiments were
made using phase-sensitive ge-2D NOESY for CDCl3 spectra and
phase-sensitive ge-2D NOESY with WATERGATE for H2O/D2O (9:
1) spectra. Selective ge-1D NOESY experiments were carried out using
the 1D-DPFGE NOE pulse sequence. NOE intensities were normalized
with respect to the diagonal peak at zero mixing time. Experimental
NOEs were fitted to a double exponential function, f(t) ) p0(e-p1t)(1
- e-p2t) with p0, p1, and p2 being adjustable parameters.12 The initial
slope was determined from the first derivative at time t ) 0, f′(0) )
p0p2. From the initial slopes, interproton distances were obtained by
employing the isolated spin pair approximation.
carried out at the same level used in the geometry optimizations, and
the nature of the stationary points was determined in each case according
to the appropriate number of negative eigenvalues of the Hessian matrix.
Scaled frequencies were not considered since significant errors in the
calculated thermodynamical properties are not found at this theoretical
level.25 Electronic energies (∆E’s) were used for the discussion on the
relative stabilities of the considered structures. Steric interactions were
evaluated by means of the pairwise steric exchange energies for disjoint
interactions.26 Delocalizing interactions (hyperconjugation) were sti-
mated by calculating the second-order perturbation energies.27 These
quantities were calculated through a natural bond orbital/natural
localized molecular orbital (NBO/NLMO) analysis using the NBO 5.G
program18 and upgraded Gaussian 03 as interface. These attractive and
repulsive interactions were estimated by means of the localized σ, π,
n, σ*, and π* valence orbitals.
Molecular Dynamics Simulations. MD-tar simulations were per-
formed with AMBER20 6.0 (parm94),21 which was implemented with
GLYCAM 04 parameters22 to accurately simulate the conformational
behavior of the sugar moiety. NOE-derived distances were included
as time-averaged distance constraints, and scalar coupling constants J
Acknowledgment. We thank the Ministerio de Educacio´n
y Ciencia and FEDER (Project CTQ2006-05825 and Ramo´n
y Cajal contracts of F.C. and J.H.B.), the Universidad de
La Rioja (Project API-05/B01 and grants of G.J.-O. and
M.G.L.), and the Gobierno de La Rioja (ANGI-2004/03 and
ANGI-2005/01 projects). We also thank CESGA for computer
support.
Supporting Information Available: 1H and 13C NMR spectra,
as well as COSY, HSQC, and 2D-NOESY correlations, of
compounds 1â, 2R, and 2â; 1D NOE selective spectra for
compound 2R; TOCSY and 2D J-resolved correlations for
compound 2â; MD simulations in explicit water for derivative
2R; torsional barrier for ψs in 1R and 2R derived from AMBER/
GLYCAM force field; NBO analysis of reduced model of 1R
and 2R; B3LYP/6-31G(d) distances, dihedral angles, energy,
enthalpy, free energy, entropy, and coordinates of the optimized
structure of 2R together with its first hydration shell; complete
refs 20b and 24. This material is available free of charge via
the Internet at http://pubs.acs.org.
as time-averaged coupling constraints. A <r-6 -1/6 average was used
>
for the distances, and a linear average was used for the coupling
constants. Final trajectories were run using an exponential decay
constant of 8000 ps and a simulation length of 80 ns for the MD-tar
simulation with ꢀ ) 80, and using an exponential decay constant of
400 ps and a simulation length of 4 ns for the MD-tar simulations in
explicit water.
DFT and NBO Calculations. All calculations were carried out using
the B3LYP hybrid functional.23 The 6-31G(d) basis set was used in
the full optimization of 2R together with its first hydration shell, and
the 6-31G(d,p) basis set was used for the scans. In the case of reduced
models of 1R and 2R (see Supporting Information), full geometry
optimizations and relaxed potential energy surface (PES) scans with
step sizes of 5° were carried out using the Gaussian 03 package.24 BSSE
corrections were not considered in this work. Frequency analyses were
(20) (a) Pearlman, D. A.; Case, D. A.; Caldwell, J. W.; Ross, W. R.; Cheatham,
T. E., III; DeBolt, S.; Ferguson, D.; Seibel, G.; Kollman, P. Comput. Phys.
Commun. 1995, 91, 1-41. (b) Kollman, P. A., et al. AMBER 6; University
of California: San Francisco, 1999.
JA072181B
(21) Cornell, W. D.; Cieplack, P. C.; Bayly, I.; Gould, I. R.; Merz, K.; Ferguson,
D. M.; Spellmeyer, D. C.; Fox, T.; Caldwell, J. W.; Kollman, P. A. J. Am.
Chem. Soc. 1995, 117, 5179-5197.
(25) Bauschlicher, C. W., Jr. Chem. Phys. Lett. 1995, 246, 40-44.
(26) (a) Badenhoop, J. K.; Weinhold, F. J. Chem. Phys. 1997, 107, 5406-5421.
(b) Badenhoop, J. K.; Weinhold, F. J. Chem. Phys. 1997, 107, 5422-
5432. (c) Badenhoop, J. K.; Weinhold, F. Int. J. Quantum Chem. 1999,
72, 269-280.
(27) (a) Reed, A. E.; Curtiss, L. A.; Weinhold, F. Chem. ReV. 1988, 88, 899-
926. (b)Weinhold, F. In Encyclopedia of Computational Chemistry, Vol.
3: Natural Bond Orbital Methods; Schleyer, P. v. R., Ed.; 1998; pp 1792-
1811.
(22) Woods, R. J.; Dwek, R. A.; Edge, C. J.; Fraser-Reid, B. J. Phys. Chem. B
1995, 99, 3832-3846.
(23) (a) Lee, C.; Yang, W.; Parr, R. Phys. ReV. B 1988, 37, 785-789. (b) Becke,
A. D. J. Chem. Phys. 1993, 98, 5648-5652.
(24) Pople, J. A., et al. Gaussian 03, revision B.05; Gaussian, Inc.: Pittsburgh,
PA, 2003.
9
J. AM. CHEM. SOC. VOL. 129, NO. 30, 2007 9467