The implications of (2S,4S)-hydroxyproline 4-O-glycosylation for prolyl amide isomerization

Article abstract of DOI:10.1002/chem.200900844

The conformations of peptides and proteins are often influenced by glycans O-linked to serine (Ser) or threonine (Thr). (25,4R)-4-Hydroxyproline (Hyp), together with L-proline (Pro), are interesting targets for O-glycosylation because they have a unique influence on peptide and protein conformation. In previous work we found that glycosylation of Hyp does not affect the N-terminal amide transieis ratios (K_trans/cis) or the rates of amide isomerization in model amides. The stereoisomer of Hyp-(25,4S)-4-hydroxyproline (hyp)-is rarely found in nature, and has a different influence both on the conformation of the pyrrolidine ring and on K_trans/cis Glycans attached to hyp would be expected to be projected from the opposite face of the prolyl side chain relative to Hyp; the impact this would have on K _trans/cis was unknown. Measurements of ³J coupling constants indicate that the glycan has little impact on the C^y-endo conformation produced by hyp. As a result, it was found that the D-galactose residue extending from a C^y-endo pucker affects both K _trans/cis and the rate of isomerization, which is not found to occur when it is projected from a C^y-exo pucker; this reflects the different environments delineated by the proline side chain. The enthalpic contributions to the stabilization of the trans amide isomer may be due to disruption of intramolecular interactions present in hyp; the change in enthalpy is balanced by a decrease in entropy incurred upon glycosylation. Because the different stereoisomers-Hyp and hyp-project the O-linked carbohydrates in opposite spatial orientations, these glycosylated amino acids may be useful for understanding of how the projection of a glycan from the peptide or protein backbone exerts its influence.

Full text of DOI:10.1002/chem.200900844

FULL PAPER
DOI: 10.1002/chem.200900844
The Implications of (2S,4S)-Hydroxyproline 4-O-Glycosylation for Prolyl
Amide Isomerization
Neil W. Owens, Adrian Lee, Kirk Marat, and Frank Schweizer*[a]
Abstract: The conformations of pep-
tides and proteins are often influenced
by glycans O-linked to serine (Ser) or
threonine (Thr). (2S,4R)-4-Hydroxy-
proline (Hyp), together with l-proline
(Pro), are interesting targets for O-gly-
cosylation because they have a unique
influence on peptide and protein con-
formation. In previous work we found
that glycosylation of Hyp does not
affect the N-terminal amide trans/cis
ratios (Ktrans/cis) or the rates of amide
isomerization in model amides. The
stereoisomer of Hyp—(2S,4S)-4-hy-
droxyproline (hyp)—is rarely found in
nature, and has a different influence
both on the conformation of the pyrro-
lidine ring and on Ktrans/cis. Glycans at-
tached to hyp would be expected to be
projected from the opposite face of the
prolyl side chain relative to Hyp; the
impact this would have on Ktrans/cis was
pucker; this reflects the different envi-
ronments delineated by the proline
side chain. The enthalpic contributions
to the stabilization of the trans amide
isomer may be due to disruption of in-
tramolecular interactions present in
hyp; the change in enthalpy is balanced
by a decrease in entropy incurred upon
glycosylation. Because the different
stereoisomers—Hyp and hyp—project
the O-linked carbohydrates in opposite
spatial orientations, these glycosylated
amino acids may be useful for under-
standing of how the projection of a
glycan from the peptide or protein
backbone exerts its influence.
3
unknown. Measurements of J coupling
constants indicate that the glycan has
little impact on the Cg-endo conforma-
tion produced by hyp. As a result, it
was found that the d-galactose residue
extending from a Cg-endo pucker af-
fects both Ktrans/cis and the rate of iso-
merization, which is not found to occur
when it is projected from a Cg-exo
Keywords: carbohydrates
· glyco-
conjugates · glycopeptides · peptide
glycosylation
Introduction
and glycopeptide mimics have been prepared over the
years.[7] These studies have concluded that the nature of the
glycosidic linkage not only influences the presentation of
the carbohydrate moiety, but also influences peptide back-
bone conformation.[2d,8] Typically, carbohydrates are O-
linked to serine (Ser) and threonine (Thr) or N-linked to as-
paragine (Asn). O-Glycosylation of (2S,4R)-4-hydroxypro-
line (Hyp) is widespread in the plant kingdom and occurs in
hydroxyproline-rich glycoproteins (HRGPs) that are associ-
ated with the cell walls of algae and flowering plants.[9] The
stereoisomer of Hyp—(2S,4S)-4-hydroxyproline (hyp)—is
rarely found in nature, but has been isolated from extracts
of the sandalwood tree Santalum album, several species of
fungi and the cyanobacteria Lyngbya majuscula.[10]
Glycosylation is a common post-translational modification
of proteins in eukaryotes, as well as in bacteria and arch-
aea.[1] Aside from direct involvement in various biological
processes involving protein–carbohydrate recognition,[2] gly-
cosylation also affects protein conformation and folding,[2a,b]
receptor binding and signalling,[2a,b] enhancement of the
thermal stabilities of proteins,[3] protection against proteolyt-
ic degradation,[4] hydration and hydrophilicity,[5] and may
also facilitate membrane penetration.[6]
In order to explore the effects of glycosylation on peptide
backbone conformation, many small-model glycopeptides
Proline (Pro), Hyp and hyp exhibit properties unique
among proteinogenic amino acids. Firstly, these amino acids
are characterized by limited rotation of their f dihedral
angles, because their side chains are fused to the peptide
backbones. As a result, there is a reduction in the energy
difference between the prolyl amide cis (w=08) and trans
(w=1808) isomers, making them nearly isoenergetic
[a] N. W. Owens, A. Lee, Dr. K. Marat, Dr. F. Schweizer
Department of Chemistry, University of Manitoba, Winnipeg, Mani-
toba, R3T 2N2 (Canada)
Fax : (+1)204-474-7608
E-mail: schweize@ms.umanitoba.ca
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/chem.200900844.
Chem. Eur. J. 2009, 15, 10649 – 10657
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
10649

(Figure 1);[11] this leads to higher cis N-terminal amide
isomer content than in the case of the other amino acids.
Secondly, prolyl cis/trans isomerization is often the rate-de-
constraint into glycopeptides. From these results, we became
interested in studying how galactosylation of hyp, the stereo-
isomer of Hyp, would influence the thermodynamics and ki-
netics of prolyl amide cis/trans isomerization. We anticipated
that O-glycosylation of hyp might have a different impact
on N-terminal amide isomerization, because 4S-hydroxyl-
ation would cause the hydroxy group to be projected from
the opposite face of the prolyl side chain relative to its ste-
reoisomer (Figure 2). Whereas Hyp has been found to adopt
Figure 1. Proline cis amide isomers showing relevant backbone torsion
angles as well as the trans amide isomer with the n ! p* interaction be-
tween the N-terminal amide carbonyl oxygen and the C-terminal carbon-
yl carbon.
termining step in the folding pathways of many peptides and
proteins,[11a,12] and thirdly, Pro and Hyp induce b turns and
extended helical structures (polyproline helixes), crucial in
protein–protein and protein–peptide interactions, in pep-
tides.[13,14] Moreover, both Pro and Hyp play important roles
in the stabilities of structural proteins such as collagen[15]
and have also been implicated in contributing to the stabili-
ties of HRGPs such as the extensins.[16] As a result, there is
growing interest in understanding and controlling prolyl N-
terminal amide isomerization in biological processes.
Figure 2. The Cg-exo pucker of Hyp would place the sugar below the
plane of the proline ring, whereas the Cg-endo pucker of hyp would place
the sugar above the plane of the proline side chain.
The prolyl cis/trans isomerization equilibrium is mostly
governed by an n ! p* interaction between the oxygen
lone pair from the prolyl N-terminal amide C=O and the an-
tibonding orbital of the C-terminal C=O (Figure 1).[11,17–20]
Recent studies have shown that 4-hydroxylation and attach-
ment of other electron-withdrawing groups have a further
influence on cis/trans isomerization through inductive and
stereoelectronic effects,[17,21–25] which affect the n ! p* inter-
action through changes to the conformation of the pyrroli-
a Cg-exo conformation, hyp is associated with a Cg-endo con-
formation;[31–35] the closely related Cb-exo conformation has
also been suggested.[36,37] This conformational switch has
been attributed to a stereoelectronic effect, through which
the 4-hydroxy group prefers to adopt a gauche orientation
with the prolyl nitrogen atom.[17,21–25] This orientation is fur-
b
g
ꢀ
ꢀ
ther stabilized by hyperconjugative s
N
ACHTUNGTRENNUNG(C O)
d
g
ꢀ
ꢀ
and s
N
ACHTUNGTRENNUNG
vours the trans amide conformation relative to Pro because
the Cg-exo pucker forces a y angle of 1508, which is ideal
for a favourable n ! p* interaction; this interaction be-
tween the N-terminal amide carbonyl oxygen and the C-ter-
minal carbonyl carbon has been shown to stabilize the trans
amide (Figure 1).[17,21–25] In contrast, the Cg-endo conforma-
tion associated with hyp has been shown to favour the cis
amide conformation because of an unfavourable y dihedral
angle for the same n ! p* interaction, which has been at-
tributed to several factors: experimental and computational
methods indicate that in hyp a hydrogen bond probably
exists between the 4-hydroxy group and the C-terminal car-
bonyl oxygen atom, as well as electrostatic repulsion be-
tween the oxygen atom in the 4-position and the C-terminal
carbonyl oxygen atom.[17,35] Both of these factors are likely
to force the prolyl y angle into a poor orientation for the n
! p* interaction, resulting in hyp favouring the cis amide
conformation more than Pro and Hyp.
dine ring[26,27] and the prolyl backbone
y
dihedral
angle.[17,21–25] It has also been shown that carbohydrates can
affect cis/trans isomerization, because Ser O-glycosylated
either with a-linked N-acetylgalactosamine or with b-linked
N-acetylglucosamine N-terminal to Pro stabilizes the trans
amide conformation.[28] Recently, our group has developed
several unnaturally C-glycosylated Pro analogues, which
have demonstrated strong abilities to vary the prolyl N-ter-
minal amide equilibrium (Ktrans/cis).[29]
In terms of clarifying the effects of 4-O-glycosylation of
Hyp, as found in HRGPs, we demonstrated in earlier work
that neither a- or b-glycosylation of Hyp in model amides
had any apparent effect either on the isomer equilibrium
constants or on the rates of amide isomerization.[30] Howev-
er, our results demonstrated that galactosylation of Hyp pro-
vides an inductive electron-withdrawing effect on the prolyl
ring. It is known that (4R)-electronegative substituents stabi-
lize the Cg-exo pucker of proline.[17,21–25] Moreover, NOE ex-
periments indicated that glycosylation of Hyp resulted in
distant contacts between the proline and galactose rings,
which indicates that glycosylation induces conformational
Because glycosylation of hyp effectively places the carbo-
hydrate moiety on a different face of the proline side chain
than in the case of Hyp, it may have a different impact on
proline isomerization. We explored the effects of glycosyla-
10650
www.chemeurj.org
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2009, 15, 10649 – 10657

Prolyl Amide Isomerization
FULL PAPER
tion of (2S,4S)-4-hydroxyproline on prolyl conformation,
and the kinetics and thermodynamics of prolyl amide iso-
merization, by using the monosaccharide d-galactose at-
tached in both a- and b-anomeric configurations. Galactose
was selected because of the occurrence of Hyp-Gal linkages
in HRGPs.[9,16]
tion with benzyl 1-thio-2,3,4,6-tetra-O-benzyl-b-d-galacto-
pyranoside[40] in the presence of N-iodosuccinimide and cat-
alytic amounts of silver triflate to yield both the a-linked
(8a) and the b-linked (8b) products in 35% yield over two
steps. A 1.9:1 ratio of the a- and b-anomers was obtained
under these conditions. Installation of the N-acetyl group
was carried out by N-Fmoc removal (piperidine in dichloro-
methane) followed by N-acylation (acetic anhydride in pyri-
dine) to give 9a and 9b in 78% yield. Removal of the
benzyl ether protecting groups by catalytic hydrogenolysis
in methanol gave the model amides 2 and 3 in 27% overall
yield from 7. Assignment of the a- and b-anomers was car-
Results and Discussion
Synthesis: Model peptides of the N-acetyl-Pro methyl ester
form are well established for the study of subtle effects of
proline side chain modification on N-terminal amide isomer-
ization.[17,21–25,26,38] This prevents the intramolecular hydro-
gen bonding and g turn formation observed in N-acetylpro-
line N’-methylamide model peptides.[39] Therefore, the
1
3
ried out by measuring the H NMR JH ,H coupling constant,
1
2
which for 2 was 1.9 Hz, indicative of the gauche dihedral
3
angle expected for the a-anomer, whereas JH ,H for 3 was
1
2
8.0 Hz, indicative of the trans diaxial relationship expected
for the b-anomer.
model amides Ac-Hyp-OMe (1), Ac-[Hyp
(2), Ac-[Hyp(b-d-Gal)]-OMe (3), Ac-hyp-OMe (4), Ac-
[hyp(a-d-Gal)]-OMe (5) and Ac-[hyp(b-d-Gal)]-OMe (6)
ACHTUNGTREN(NGNU a-d-Gal)]-OMe
T
The synthesis of the (4S) model amides was carried out in
a similar fashion beginning from 10 (Scheme 2). N-Boc re-
moval (trifluoroacetic acid in dichloromethane at 08C) gave
4 in 94% yield after N-acylation under standard conditions.
Glycosylation of 10 was carried out under the same condi-
tions as for 7. It was found that separation of the a- and b-
glycosides required replacement of the N-Boc group with
the N-Fmoc protecting group, which was carried out under
standard conditions to give 11a and 11b in 32% yield over
three steps, with a 1.5:1 ratio of the a- and b-anomers. In-
stallation of the N-acetyl group to give 12a and 12b, fol-
lowed by removal of the benzyl
A
ACHTUNGTRENNUNG
were synthesized by different strategies, depending on the
commercial availability of each proline derivative
(Schemes 1 and 2, below). We found that installation of the
N-Fmoc group allowed optimal separation of the a- and b-
anomers 8a/8b (Scheme 1) and 11a/11b (Scheme 2, below).
Model amides containing both a- and b-anomeric linkages
were made for comparison in order to explore how the
nature of the glycosidic linkage affects prolyl amide cis/trans
isomerization.
ether protecting groups, was
carried out under the same con-
ditions as for 8a and 8b, to give
the model peptides 5 and 6 in
25% overall yield over three
steps.
IR spectroscopic study: The fre-
quency of the amide I vibra-
tional mode (vamide), which is
primarily
a function of the
amide C=O stretching vibra-
tion, has been correlated with
changes in the bond order of
the amide C=O group.[41] In
D2O, the vamide values for the
model compounds 1, 2 and 3
were nearly identical, with
maxima at 1612, 1611.5 and
1611.5 cmꢀ1, respectively. Simi-
larly, nearly identical maxima
of 1612, 1611 and 1613 cmꢀ1
Scheme 1. Synthesis of 1, 2, and 3. a) Ac2O, Et3N, MeOH, 258C, 2 h, 95%; b) i) Fmoc-OPfp, NaHCO3, ace-
tone/H2O 3:1, 258C, 3 h, ii) benzyl 1-thio-2,3,4,6-tetra-O-benzyl-b-d-galactopyranoside, AgOTf, NIS, CH3CN, 0
to 258C, 2 h, 35% overall yield; c) i) CH2Cl2/piperidine 1:1, 258C, 3 h, ii) Ac2O, Py, 258C, 4 h, 78% overall
yield; d) H2/Pd(OH)2/C, MeOH/ethyl acetate 4:1, 258C, 4 h, quantitative.
The synthesis of 1 was carried out by selective N-acylation
of 7 with acetic anhydride and triethylamine in methanol in
95% yield (Scheme 1). The glycosylated (4R)-hydroxypro-
line model peptides 2 and 3 were obtained through N-Fmoc
protection of 7 (9-fluorenylmethyl pentafluorophenyl car-
bonate) under mild basic conditions, followed by glycosyla-
were exhibited by 4, 5 and 6, respectively. Therefore, for
compounds 1–6, no change in the amide carbonyl I vibra-
tional mode was found to occur with a- or b-glycosylation
or inversion of stereochemistry at the 4-position. A slightly
more pronounced effect was observed in a study of 4-fluoro-
proline (Flp) in Ac-Flp-OMe model compounds, which
Chem. Eur. J. 2009, 15, 10649 – 10657
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
10651

F. Schweizer et al.
changes in the shift of the Cg-
carbon (~9 ppm) were found to
occur upon glycosylation both
for the (4R) model compounds
(78.9, 77.6 ppm, in relation to
69.9 ppm for 2, 3 and 1, respec-
tively) and for the (4S) model
compounds (80.3, 80.6 ppm, in
relation to 69.9 ppm for 5, 6
and 4, respectively). Glycosyla-
tion therefore appears to cause
a
local electron-withdrawing
effect. This is similar in magni-
tude to the 13Cg chemical shifts
of ~8 ppm observed on attach-
ment of trifluoroacetate groups
to similar model peptides, but
more than that due to a simple
acyl group, which caused only a
shift of ~3 ppm.[33]
Scheme 2. Synthesis of 4, 5, and 6. a) i) TFA/CH2Cl2 1:1, 08C, 1.5 h. ii) Ac2O, Et3N, MeOH, 258C, 2 h, 94%
overall yield; b) i) benzyl 1-thio-2,3,4,6-tetra-O-benzyl-b-d-galactopyranoside, AgOTf, NIS, CH3CN, 0 to 258C,
2.5 h, ii) TFA/CH2Cl2 1:1, 08C, 1.5 h, iii) Fmoc-OPfp, NaHCO3, acetone/H2O 3:1, 258C, 2 h, 32% overall yield;
c) i) CH2Cl2/piperidine 1:1, 258C, 3 h, ii) Ac2O, Py, 258C, 4 h, 78% overall yield; d) H2/Pd(OH)2/C, MeOH/
ethyl acetate 4:1, 258C, 4 h, quantitative.
showed a maximum vamide shift—relative to (4R)-hydroxy-
NOE experiments: To determine the extent of interaction
between the galactose and prolyl rings, selective NOE trans-
fer experiments were performed on the galactosylated
model peptides 2, 3, 5 and 6 in D2O (Figure 3). For com-
proline in similar model compounds—of 3 cmꢀ1 in D2O.[21]
1
NMR spectroscopic studies: Full assignment of the H NMR
spectra of 1–6 was carried out with the aid of COSY and
HSQC experiments. Assignment of the major isomers of 1–6
as the trans amide isomers in each case was established by
selective one-dimensional GOESY[42] experiments, which
showed interproton effects from the N-acetyl methyl group
to the prolyl d-protons (1.1–2.4% NOE relative to the N-
acetate singlet signal). By comparison, this interaction was
not observed in the minor isomer.
pound 2, it was found that selective inversion of Hypb re-
1
sulted in 0.9% and 1.2% resonance transfers to the peaks at
d=3.96 ppm and d=3.99 ppm, which correspond to the gal-
actose H3 and H5 protons, respectively. This suggests that a-
galactosylation of Hyp results in close contacts between dis-
tant positions in the galactose and prolyl rings (Figure 3).
The overlap of the hydrophobic a-face of d-galactose with
the pyrrolidine ring of proline in 2 resembles other hydro-
phobic galactose–protein interactions found in several crys-
Prolyl side-chain conformation: The prolyl ring puckers of
the major isomers of 1–6 were established by comparison of
1H NMR coupling constants with literature values (see the
tal structures.[48,49] In contrast, selective inversion of Hypb in
1
3 only showed resonance transfer to H1 of galactose (1.5%),
and no other sugar protons. However, there was greater
Supporting Information).[43] The prolyl puckers for 1–3 were
overlap of the galactose protons in the H NMR spectrum of
1
3
each assigned as Cg-exo on the basis of Ja,b values of 8.2–
3, which made assignment of NOE contacts with certainty
1
3
8.3 Hz and Ja,b values of 8.5–8.7 Hz. The corresponding
more difficult. Selective inversion of the Hypb /Hypb signal
2
1
2
coupling constants for the Cg-exo pucker are expected to be
7–10 and 7–11 Hz, respectively. In contrast, the prolyl puck-
ers for 4–6 were each assigned as Cg-endo on the basis of
in 5 and 6 only showed an NOE contact to H1 of galactose
(0.7 and 0.4%, respectively), but not to other sugar protons.
Therefore, the galactose rings in the glycosylated (4S)-hyp
model peptides 5 and 6 are likely to be located distally to
the proline rings (Figure 3).
3
3Ja,b values of 7.8–9.8 Hz and Ja,b values of 2.4–4.3 Hz. The
1
2
coupling constants for the Cg-endo pucker are expected to
be 6–10 and 2–3 Hz, respectively. These results are as ex-
pected in the context of other experimental[17,21–25,31–33,43]
measurements and computational[34,44–46] predictions. The
coupling constants for 4 and 6 were nearly identical at 258C,
whereas those of 5 deviated slightly.
Measurement of Ktrans/cis: The ratios of trans/cis isomers
(Ktrans/cis) were established by integrating as many well-re-
solved peaks as possible for each isomer, and taking the
average for all peaks for each isomer (Table 1).[20] As previ-
ously found, glycosylation of (4R)-Hyp does not affect Ktrans/cis
in D2O (1–3 had Ktrans/cis values of 5.9–6.2).[30] However, it
was found that the glycosylation of (4S)-hyp does affect
Ktrans/cis. The a- and b-linked sugars in 5 and 6, respectively,
were found to stabilize the prolyl N-terminal trans amide
conformation to an equal extent (Ktrans/cis value of 2.9 in D2O
at 258C) relative to 4 (Ktrans/cis value of 2.4). This is in con-
Cg inductive effect: Changes in 13C chemical shifts have been
used to estimate the electron-withdrawing effects of sub-
stituents on the prolyl side-chain.[47] Therefore, measurement
of the Cg-carbon chemical shifts in the model compounds 1–
6 was used to assess the relative changes in electron-with-
drawing ability incurred upon glycosylation. Significant
10652
www.chemeurj.org
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2009, 15, 10649 – 10657

Prolyl Amide Isomerization
FULL PAPER
Figure 3. Relevant NOE interactions between the prolyl side chain and the carbohydrate ring for 1–6. These experiments indicate that close contacts be-
tween the rings exist when d-galactose is a-linked to Hyp, but not when it is a-linked to hyp. Furthermore, both a- and b-linked d-galactose seem to be
oriented away from the proline side chain when O-linked to hyp. Thus, Hyp and hyp have different impacts on the orientations of the glycans.
Table 1. Thermodynamic and kinetic parameters for 1–6.
[a]
[b]
[b]
[c]
[d]
[e]
[f]
[g]
Compound Ktrans/cis
DH8cis!trans [kcalmolꢀ1
]
DS8cis!trans [calmolꢀ1 Kꢀ1
]
DG8300 K [kcalmolꢀ1
]
kcis!trans
ktrans!cis
vamide [cmꢀ1
]
dACHTUNGTRENNUNG
(13Cg)trans
1
2
3
4
5
6
6.2ꢁ0.1 ꢀ1.43ꢁ0.04
6.0ꢁ0.1 ꢀ1.42ꢁ0.12
5.9ꢁ0.2 ꢀ1.45ꢁ0.13
2.4ꢁ0.1 ꢀ0.29ꢁ0.08
2.9ꢁ0.3 ꢀ0.94ꢁ0.13
2.9ꢁ0.1 ꢀ0.91ꢁ0.05
ꢀ1.17ꢁ0.11
ꢀ1.14ꢁ0.37
ꢀ1.38ꢁ0.41
0.79ꢁ0.25
ꢀ1.08ꢁ0.07
ꢀ1.08ꢁ0.23
ꢀ1.04ꢁ0.25
ꢀ0.53ꢁ0.16
ꢀ0.63ꢁ0.26
ꢀ0.62ꢁ0.10
0.81ꢁ0.01 0.18ꢁ0.01 1612
0.85ꢁ0.01 0.19ꢁ0.01 1611.5
0.77ꢁ0.02 0.18ꢁ0.02 1611.5
0.44ꢁ0.04 0.20ꢁ0.01 1612
0.59ꢁ0.06 0.25ꢁ0.03 1611
0.71ꢁ0.04 0.30ꢁ0.02 1613
69.9
78.9
77.6
69.9
80.3
80.6
ꢀ1.03ꢁ0.42
ꢀ0.96ꢁ0.16
[a] Carried out in D2O at 24.88C; ꢁSE determined by integration of two or more sets of trans/cis isomers. [b] Error limits obtained by linear least-
squares fitting of the van’t Hoff plots to the equation lnKct =(ꢀDH8/R) (1/T)+DS8/R. [c] Calculated from DG8=DH8ꢀTDS8. [d] Carried out in phos-
phate buffer (pH 7.2, 0.1m) at 67.38C; the temperature was calibrated by use of an ethylene glycol standard, and error values were obtained from the
linear least-squares fit of the data in Figures S7–S12 in the Supporting Information. [e] Calculated from kct and amide isomer equilibrium (Kt/c) at 67.38C.
[f] Determined in D2O at 258C. [g] Determined by 75 MHz NMR in D2O at 258C.
trast with previous studies of 4-O-modification of hyp,
which saw a slight stabilization of the cis amide isomer.
Taylor et al. found that O-methylation of hyp in Ac-Phe-(4S)-
hyp-NHMe model peptides caused a decrease in Ktrans/cis
from 1.9 to 1.4 in D2O at 258C.[36] However, this result is
complicated by the prolyl N-terminal l-phenylalanine resi-
due, which is known to stabilize the cis amide isomer
through interaction with the prolyl side chain.[14b] Similarly,
Jenkins et al. found that O-acylation of hyp in Ac-(4S)-hyp-
OMe model compounds caused a decrease in Ktrans/cis from
1.8ꢁ0.4 to 1.6ꢁ0.2 in D2O at 258C, although these values
were within experimental error.[33]
energy differences between the cis and trans amide isomers
are independent of temperature; the linear van’t Hoff plots
indicate that this assumption is probably valid.[21,33,38,50] We
found in each case that the Ktrans/cis value decreased with in-
creasing temperature. Compound 4 was slightly anomalous,
exhibiting less temperature dependence on Ktrans/cis than the
other model compounds. Accordingly, DH8 and DS8 could
be calculated from the least-squares fits of the van’t Hoff
plots. In each case, DH8 was <0, which correlates well with
other studies of proline model peptides (Table 1).[21,38,51,52]
The differences in Ktrans/cis values were reflected in DH8 and
DS8, with no significant differences in DH8 or DS8 between
compounds 1, 2 and 3 (DH8 values of ꢀ1.43ꢁ0.04, ꢀ1.42ꢁ
0.12 and ꢀ1.45ꢁ0.13 kcalmolꢀ1, respectively; DS8 values of
ꢀ1.17ꢁ0.11, ꢀ1.14ꢁ0.37 and ꢀ1.38ꢁ0.41 calmolꢀ1 Kꢀ1, re-
spectively). In contrast, there was an increase of
0.6 kcalmolꢀ1 [relative to 4 (DH8 of ꢀ0.29ꢁ0.08 kcalmolꢀ1)]
Thermodynamics: The effect of temperature on the Ktrans/cis
values for compounds 1–6 was measured by NMR spectros-
copy and the resulting van’t Hoff plots are shown in
Figure 4. This assumes that the enthalpic and entropic
Chem. Eur. J. 2009, 15, 10649 – 10657
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
10653

F. Schweizer et al.
to 4 cannot be explained by such a conformational change
in this case.[26,27]
The local electron-withdrawing effect caused by glycosyla-
tion may diminish the intramolecular electrostatic repulsion
between the 4-hydroxy groups and the C-terminal carbonyl
oxygen atoms in 5 and 6.[35] This would allow the prolyl y
angle to relax from 1808 closer to the optimal angle of 1508
for a favourable n ! p* interaction, which has been esti-
mated to contribute 0.7 kcalmolꢀ1 to the stability of the
trans amide isomer.[22,55] Because this is specific for the Cg-
endo conformation, it would explain the lack of effect in 1–
3, which each have a Cg-exo pucker.
The effect of the glycosylation on the intramolecular hy-
drogen bond in hyp should also be considered, because it
has been estimated to contribute 1.5 kcalmolꢀ1 to stabiliza-
tion of the Cg-endo conformation.[35] The presence of the
glycosidic linkage would be expected to eliminate this intra-
molecular interaction, and because NOE experiments indi-
cated that the sugar is not proximal to the prolyl side chain,
it is unlikely to be restored through the sugar hydroxy
groups. The removal of the intramolecular hydrogen bond
should destabilize the Cg-endo conformation in 5 and 6 and
thereby destabilize the cis amide isomer. This would also ex-
plain why there is no impact from glycosylation on the Ktrans/cis
values for 1–3, because there are no equivalent intramolecu-
lar hydrogen bonds. Taylor et al., however, have down-
played the effects of the intramolecular hydrogen bond be-
cause O-methylation of hyp had little effect on Ktrans/cis
values.[36]
Figure 4. van’t Hoff plots for 1–6 in D2O.
in the enthalpic energy differences in 5 and 6 (DH8 values
of ꢀ0.94ꢁ0.13 and ꢀ0.91ꢁ0.05 kcalmolꢀ1, respectively).
Also, there seemed to be different entropic contributions to
the ground state energy difference (DG8) between the cis
and trans amide isomers, because glycosylation seemed to
cause the trans amide to become more ordered in 5 and 6
(DS8 values of ꢀ1.03ꢁ0.42 and ꢀ0.96ꢁ0.16 calmolꢀ1 Kꢀ1,
respectively), whereas the trans amide isomer was more dis-
ordered in 4 (DS8 of +0.79ꢁ0.25 calmolꢀ1 Kꢀ1).
The proposed reductions in the intramolecular electrostat-
ic repulsion and the loss of the intramolecular hydrogen
bonds, which have an influence on the enthalpic contribu-
tions (DH8) to the ground state energy differences (DG8)
between the trans and cis amide isomers, seem to be coun-
teracted by decreases in entropy (DS8) incurred upon glyco-
sylation in compounds 4–6 (Table 1). In previous work, it
has been shown that peptide and protein glycosylation
causes more ordered hydration spheres.[56] Here, glycosyla-
tion apparently had a further impact on the entropic differ-
ences established between the prolyl trans and cis amide iso-
mers in 4–6, which have themselves has been attributed to
differences in solvation.[52] Interestingly, this effect is less ap-
parent for 1–3, so the influence of solvation differences
seem to be specific to the face of the proline side chain.
NMR magnetization inversion transfer experiments indi-
cated that the hyp model compounds 1–3 have faster amide
isomerization rates overall than 4–6. Improta et al. have cal-
culated that the prolyl nitrogen is more pyramidalized in the
Cg-exo pucker than in the Cg-endo pucker,[35] which should
facilitate isomerization for 1–3, which each have a Cg-exo
pucker, with respect to 4–6, each with a Cg-endo pucker.
This is in contrast with the findings of Beausoleil et al., who
found the reverse effect: hyp had a faster rate than Hyp in
Ac-Xaa-NHMe model amides in D2O at 608C (2.05ꢁ0.50
and 1.46ꢁ0.13 sꢀ1, respectively).[31] This was attributed to
the intramolecular hydrogen bond in hyp reducing coulom-
bic repulsion between the C-terminal carbonyl oxygen atom
Kinetics: The rates of cis/trans amide isomerization were cal-
culated for 1–6 with the aid of H NMR magnetization in-
1
version transfer experiments in phosphate buffer (pH 7.2,
0.1m).[21,30,31,53] These experiments were performed at elevat-
ed temperature, because at physiological temperature the ki-
netic rates are too slow to be determined by this assay. It
was found that the glycosylated hyp model compounds 5
and 6 exhibited increases in their rates of isomerization
(kcis!trans values of 0.59ꢁ0.06 and 0.71ꢁ0.04 sꢀ1, respective-
ly) relative to the unglycosylated model peptide 4 (kcis!trans
of 0.44ꢁ0.04 sꢀ1) at 678C (Table 1). This is in contrast with
the glycosylated Hyp compounds 2 and 3, which showed
almost no differences in their rates of isomerization (kcis!trans
values of 0.85ꢁ0.02 and 0.77ꢁ0.01 sꢀ1, respectively) relative
to the unglycosylated model compound 1 (kcis!trans of 0.81ꢁ
0.01 sꢀ1) at 678C, which is consistent with previous work.[30]
The changes of ~10 ppm in the 13Cg chemical shifts of 5
and 6 relative to 4 (and similarly for 2 and 3 relative to 1)
are indicative that glycosylation causes a local electron-with-
drawing effect. An increased electron-withdrawing ability
has been attributed to stabilization of a given pucker
through a stereoelectronic effect.[17,21–25,54] This probably ex-
plains why glycosylation does not significantly change the
prolyl puckers for 1–3 or 4–6 as determined by the measure-
ment of 3J coupling constants. However, since changes in
Ktrans/cis have been correlated with changes in prolyl puckers,
the stabilization of the trans amide isomer in 5 and 6 relative
10654
www.chemeurj.org
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2009, 15, 10649 – 10657

Prolyl Amide Isomerization
FULL PAPER
and the prolyl nitrogen, although the values were within ex-
perimental error.
tension of the glycopeptide model amides to larger peptides
or to oligohydroxyprolines containing multiple glycosylation
sites might result in larger and additive effects, as seen with
other oligoproline-based models.[58,59]
The different stereoisomers of 4-hydroxyproline provide
an opportunity to understand how each face of the prolyl
ring has an influence on Ktrans/cis. Furthermore, Hyp and hyp
can be used to project a glycan rigidly in opposite spatial
orientations; these building blocks might therefore be useful
for studying carbohydrate binding interactions and the influ-
ences of glycans on peptide and protein structures in which
the orientations of the glycans are important.[60]
Glycosylation caused increases in the rates of isomeriza-
tion of 5 and 6 (0.59ꢁ0.04 and 0.71ꢁ0.06 sꢀ1, respectively)
relative to 4 (0.44ꢁ0.04 sꢀ1), whereas no increases in iso-
merization rates were observed in 1–3. In contrast, the acy-
lation of hyp by Jenkins et al. caused no observable increase
in the rate of amide isomerization.[33] Changes in cis/trans
amide isomerization have been correlated to inductive elec-
tron-withdrawing effects from prolyl g-substituents,[21,23] in
which the g-position group withdraws electron density from
the peptide bond, thereby increasing N-pyramidaliza-
[57]
tion[17,35] and reducing the C N bond order; this effective-
ꢀ
ly weakens the amide bond and enables isomerization to
occur.[36] Whereas glycosylation does appear to cause a local
electron-withdrawing effect, it does so both for (4S) and for
(4R) stereoisomers, and does not therefore explain the rela-
tive increases in isomerization rates for 5 and 6 relative to 4,
with no changes with 1–3.
Experimental Section
General procedures: Reagent grade solvents were used without further
purification. Thin-layer chromatography was performed on precoated
silica gel plates (Si250F, 250 mm). Column chromatography was per-
formed on SilicaFlash silica gel (P60, 40–63 mm). NMR spectra were as-
signed with the aid of 2D COSY and 2D HSQC experiments. For
1H NMR, minor isomers are listed between square brackets. For
13C NMR, when assigned, carbon peaks for the minor isomer are listed in
brackets.
An increase in C=O bond order is indicative of a lower
ꢀ
C N bond order, and has been used to explain the increased
rate of isomerization of 4-fluoroproline.[23] Here, perhaps be-
cause of weaker electron-withdrawing effects, glycosylation
either of the (4R)- or of the (4S)-hydroxyproline model
compounds did not affect the amide I vibrational mode
maxima, and so also cannot be used to explain the changes
in the rates of isomerization of 5 and 6 relative to 4. There-
fore, the basis for the increases in the rates of amide isomer-
ization in 5 and 6 relative to 4, while in 1–3 there is no
effect, remains unclear.
General preparation of N-acetyl-amino acid methyl esters: The amino
acid (1.0 equiv) was dissolved in CH2Cl2/piperidine 3:1, stirred for 2 h at
ambient temperature and then co-distilled with toluene (3ꢁ10 mL). The
crude product was then dissolved in acetic anhydride/pyridine 1:1, and
the mixture was stirred at ambient temperature for 4 h. The solution was
concentrated under reduced pressure and was then co-distilled with tolu-
ene (3ꢁ10 mL) before purification by flash chromatography.
Preparation of 4-O-galactopyranosyl-N-acetyl-amino acid methyl esters:
The amino acid (1.0 equiv) was dissolved in CH3OH. The catalyst (20%
palladium hydroxide on carbon, approx. 0.5 equiv) was added, and the
flask was flushed with N2 for 5 min. The reaction mixture was then
stirred under hydrogen (10 psi) for 6 h, after which it was flushed with ni-
trogen and filtered. The product was then concentrated under reduced
pressure.
Whereas the glycosylation of (4R)-hydroxyproline was
not found to affect amide isomerization, the glycosylation of
(4S)-hydroxyproline affects both the prolyl N-terminal
amide equilibrium and the rate of amide isomerization.
Here, we found that both a- and b-anomeric linkages to
(4S)-hydroxyproline in 5 and 6 stabilize the trans amide con-
formation relative to the unglycosylated model compound 4.
Glycosylation does not significantly alter the Cg-endo con-
formation of 4 induced by the (4S) hydroxylation, but does
cause a local electron-withdrawing effect; this probably re-
duces repulsion between the 4-oxygen atom and the C-ter-
minal carbonyl group and allows the y angle to relax closer
to 1508 to restore the n ! p* interaction that stabilizes the
trans amide isomer. The changes in Ktrans/cis for 5 and 6 rela-
tive to 4 might also be due to the loss of an intramolecular
hydrogen bond that is specific to the (4S) stereoisomer. This
interaction is probably not restored in 5 and 6 because the
sugar was not found to form close contacts to the prolyl side
chain. Regardless of their origin, the enthalpic contributions
to DG8 seem to be offset by entropic changes, which create
a more ordered environment for the trans amide isomer in 5
and 6 relative to 4, resulting in only small net changes in
Ktrans/cis. The glycosylation of (4S)-hydroxyproline does seem
to cause a small increase in rate of amide isomerization, but
it is not reflected in the amide I vibrational mode and thus
the bond order of the N-terminal amide group. Therefore,
the cause of the increases in the rates remains unclear. Ex-
trans-N-Acetyl-4-hydroxy-l-proline methyl ester (1): Compound
7
(0.250 g, 1.38 mmol) was dissolved in methanol (2 mL), followed by the
addition of triethylamine (0.77 mL, 5.52 mmol, 4.0 equiv) and acetic an-
hydride (0.78 mL, 8.28 mmol, 6.0 equiv). The reaction mixture was stirred
for 2 h at ambient temperature before being concentrated under reduced
pressure. The product was purified by flash chromatography with ethyl
acetate/methanol 9:1 to yield 1 as a white solid (0.244 g, 1.31 mmol,
95.0%). [a]D25 =ꢀ83.38 (c=0.7, CH3OH); m.p. 75–788C; 1H NMR
(500 MHz, D2O, 298 K): d=[4.83, dd, Ja,b =8.0 Hz, Ja,b =7.6 Hz, 0.14H;
1
2
Procais], 4.61 (dddd, Jb ,g =2.1 Hz, Jb ,g =4.4 Hz, Jg,d =4.4 Hz, Jg,d =1.9 Hz,
1
2
1
2
0.86H; Progtrans), 4.54 (dd, Ja,b =8.4 Hz, Ja,b =8.6 Hz, 0.86H; Protarans),
1
2
[4.51–4.57, m, 0.14H; Procgis], 3.85 (dd, Jd ,d =13.7 Hz, 0.86H; Protdrans),
1
2
1
cis
[3.83, s, 0.4H; COCH3 ], 3.79 (s, 2.6H; COCH3trans), [3.71, dd, Jg,d
=
ꢀ
ꢀ
1
2.2 Hz, Jd ,d =12.6 Hz, 0.14H; Procdis], 3.66 (dd, 0.86H; Protdrans), [3.56, dd,
1
2
Jg,d =4.5 Hz, 0.14H; Procdis], [2.510, ddd,
J
b ,g =1.7 Hz, Jb ,b =13.7 Hz,
2
2
1
1
2
2
0.14H; Procis], 2.39 (ddd, Jb ,b =13.7 Hz, 0.86H; Protrans), [2.34–2.40, m,
b1
b1
1
2
0.14H; Probcis], 2.18 (ddd, 0.86H; Probtrans), 2.14 (s, 2.6H; NCOCH3trans),
ꢀ
2
2
[2.03 ppm, s, 0.4H; NCOCH3cis]; 13C NMR (75 MHz, D2O, 298 K): d=
172.9, (172.7), (170.8), 170.2, 69.6, (68.0), (58.7), 57.5, 55.9, (54.3), (52.7),
52.2, (39.6), 37.8, 22.1, (21.5 ppm); MS (ES): m/z: calcd for C8H13NNaO4:
210.07 [M+Na]+; found: 209.73 [M+Na]+.
ꢀ
trans-4-O-(a-d-Galactopyranosyl)-N-acetyl- 4-hydroxy-l-proline methyl
ester (2): The general preparation method was followed for O-debenzyla-
tion of 9a (0.153 g, 0.215 mmol) to yield 2 as a clear oil (0.075 g,
0.215 mmol, quant.). [a]D25 =+105.98 (c=0.3, CH3OH); 1H NMR
(500 MHz, D2O, 338 K): d=5.07 (d, J=1.9 Hz, 0.8H; H1), [5.03, brd,
0.2H; H1], [4.83, dd, J=7.1, 7.6 Hz, 0.2H; Procais], 4.54–4.61 (m, 1.6H;
Chem. Eur. J. 2009, 15, 10649 – 10657
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
10655

F. Schweizer et al.
Protgrans, Proatrans), [4.51, dddd, J=1.4, 5.0, 5.0, 7.5 Hz, 0.2H; Procgis], 3.98–
4.01 (m, 1H; H5), 3.93–3.98 (m, J=6.0, 6.2 Hz, 1H; H3), 3.71–3.89 (m,
cis-4-O-(b-d-Galactopyranosyl)-N-acetyl-4-hydroxy-l-proline-methyl
ester (6): The general preparation method was followed for O-debenzyla-
tion of 12b (0.047 g, 0.066 mmol) to yield 6 as a clear oil (0.023 g,
0.066 mmol, quant.). [a]D25 =ꢀ40.68 (c=0.7, CH3OH); 1H NMR
(500 MHz, D2O, 338 K): d=[5.09, dd, J=1.3, 9.2 Hz, 0.3H; Procais], 4.86–
4.98 (m, 1.7H; Prog, Proatrans), 4.73 (d, J=8.2 Hz, 0.7H; H1trans), [4.71, d,
J=8.2 Hz, 0.3H; H1cis], 4.20–4.25 (m, 1H; H4), 4.16 (dd, J=4.8 Hz. J=
ꢀ
8.8H; H2, H4, H6a, H6b, Prod , Protrans
,
CO2CH3), [3.59, dd, J=5.0,
ꢀ
d2
1
12.9 Hz, 0.2H; Procdis], [2.67, ddd, J=7.5, 7.6, 13.4 Hz, 0.2H; Procbis], 2.58
2
1
(ddd, J=5.3, 7.9, 13.3 Hz, 0.8H; Protbrans), [2.43, ddd, J=5.0, 7.1, 13.4 Hz,
1
0.2H; Procbis], 2.23 (ddd, J=5.0, 8.3, 13.3 Hz, 0.8H; Protbrans), 2.13 (s, 2.4H;
2
2
NCOCH3trans), [2.01 ppm, s, 0.6H; NCOCH3cis]; 13C NMR (75 MHz,
D2O, 298 K): d=179.6, (179.2), (178.8), 178.2, 102.9, (102.8), 81.3, (79.4),
76.6, (76.5), 74.3, (74.2), 73.0, (72.9), 66.3, (64.1), 62.9, (58.4), 58.3, 58.0,
(56.2), 53.9, (41.8), 40.4, 26.3, (25.7 ppm); MS (ES): m/z: calcd for
C14H23NNaO9: 372.13 [M+Na]+; found: 372.03 [M+Na]+.
ꢀ
ꢀ
11.8 Hz, 0.7H; Protdrans), 4.02–4.13 (m, 5.7H; H5, H6a, Protrans
, CO2CH3),
d2
3.90–3.99 (m, 2.6H;1H3, H6b, Procis, Procdis), 3.75 (dd, J=7.5, 9.6 Hz, 0.7H;
d1
1
H2trans), [3.71, dd, J=8.0, 10.0 Hz, 0.3H; H2cis], [2.90–2.96, m, 0.3H;
Probcis], 2.79–2.90 (m, 1H; Probtrans, Probcis), 2.67–2.74 (m, 0.7H; Probtrans),
1
2
2
2.431 (s, 2.1H;
NCOCH3trans), [2.34 ppm, s, 0.9H;
NCOCH3cis];
ꢀ
ꢀ
trans-4-O-(b-d-Galactopyranosyl)-N-acetyl-4-hydroxy-l-proline methyl
ester (3): The general preparation method was followed for O-debenzyla-
tion of 9b (0.100 g, 0.141 mmol) to yield 3 as a clear oil (0.049 g,
0.141 mmol, quant.). [a]D25 =ꢀ33.68 (c=0.5, CH3OH); 1H NMR
(500 MHz, D2O, 338 K): d=[4.81, dd, J=7.3, 7.8 Hz, 0.2H; Procais], 4.66–
4.73 (m, 0.8H; Protgrans), [4.59–4.65, m, 0.2H; Procgis], 4.55 (dd, J=8.2,
8.2 Hz, 0.8H; Proatrans), 4.49 (d, J=8.0 Hz, 0.8H; H1trans), [4.47, d, J=
13C NMR (75 MHz, D2O, 298 K): d=(175.0), 174.7, (174.3), 173.8, 102.3,
(102.0), 78.3, (77.0), 75.6, (75.5), 72.9, (71.2), 71.1, 68.9, 61.3, (59.9), 58.0,
54.2, (53.7), 53.5, (52.9), (37.6), 35.9, 21.7, (21.5 ppm); MS (ES): m/z:
calcd for C14H23NNaO9: 372.13 [M+Na]+; found: 372.03 [M+Na]+.
8.3 Hz, 0.2H; H1cis], 3.92–3.99 (m, 1.2H; H4, Procis), 3.86–3.92 (m, 1.6H;
d1
Protdrans, Prodtrans), [3.85, s, 0.6H; CO2CH3cis], 3.74–3.83 (m, 4.4H; H6a, H6b
ꢀ
,
ꢀ
1
2
CO2CH3trans), 3.64–3.74 (m, 2H; H3, H5), [3.61, dd, J=4.7, 12.9 Hz,
Acknowledgements
0.2H; Procdis], 3.52 (dd, J=8.6, 9.1 Hz, 0.8H; H2trans), [3.82–3.53, m, 0.2H;
2
H2cis], [2.60–2.69, m, 0.2H; Procbis], 2.55 (ddd, J=2.4, 8.2, 13.4 Hz, 0.8H;
The authors thank the Natural Sciences and Engineering Research Coun-
cil of Canada (NSERC) and the University of Manitoba for financial sup-
port.
1
Protbrans), [2.42, ddd, J=5.6, 6.6, 13.4 Hz, 0.2H; Procbis], 2.21 (ddd, J=8.2,
1
2
10.0, 13.4 Hz, 0.8H; Protbrans), 2.14 (s, 2.4H; NCOCH3trans), [2.02 ppm, s,
ꢀ
2
0.6H; NCOCH3cis]; 13C NMR (75 MHz, D2O, 298 K): d=179.5, (179.1),
ꢀ
(178.8), 178.2, (106.8), 106.6, 82.6, (81.3), (80.3), 80.2, (77.65), 77.61,
(75.54), 75.48, 73.5, 65.9, (63.7), 62.6, 59.3, (58.4), 58.0, (57.3), (53.9),
(41.2), 39.6, 26.3, (25.7 ppm); MS (ES): m/z: calcd for C14H23NNaO9:
372.13 [M+Na]+; found: 372.03 [M+Na]+.
[1] a) E. Weerepana, B. Imperiali, Glycobiology 2006, 16, 91R–101R;
b) P. Messner, J. Bacteriol. 2004, 186, 2517–2519; c) J. Eichler,
M. W. Adams, Microbiol. Mol. Biol. Rev. 2005, 69, 393–425.
[2] a) A. Varki, Glycobiology 1993, 3, 97–130; b) R. A. Dwek, Chem.
Rev. 1996, 96, 683–720; c) G. Reuter, H. J. Gabius, Cell. Mol. Life
Sci. 1999, 55, 368–422; d) A. J. Petrescu, A. L. Milac, S. M. Petrescu,
R. A. Dwek, M. R. Wormald, Glycobiology 2004, 14, 103–114.
[3] a) G. Mer, H. Hietter, J.-F. Lefꢂvre, Nat. Struc. Biol. 1996, 3, 45–53;
b) B. Imperiali, K. W. Rickert, Proc. Natl. Acad. Sci. USA 1995, 92,
97–101.
cis-N-Acetyl-4-hydroxy-l-proline methyl ester (4): Compound 10
(0.050 g, 0.204 mmol, 1.0 equiv) was dissolved in CH2Cl2 (5 mL) and
cooled to 08C, before addition of TFA (5 mL). After stirring for 1.5 h,
the solution was co-distilled with toluene (3ꢁ6 mL) and concentrated
under reduced pressure. The residue was then re-dissolved in methanol
(2 mL), followed by the addition of triethylamine (0.112 mL, 0.816 mmol,
4.0 equiv) and acetic anhydride (0.113 mL, 1.224 mmol, 6.0 equiv). The
reaction mixture was stirred for 2 h before concentration and purification
[4] a) J. F. Fisher, A. W. Harrison, G. L. Bundy, K. F. Wilkinson, B. D.
Rush, M. J. Ruwart, J. Med. Chem. 1991, 34, 3140–3143; b) S.
Mehta, M. Meldal, J. O. Duus, K. Bock, J. Chem. Soc. Perkin Trans.
1 1999, 1445–1451.
with ethyl acetate/methanol 9:1 to yield
4 as a clear oil (0.036 g,
0.192 mmol, 94.7%). [a]D25 =+89.98 (c=0.7, CH3OH); 1H NMR
(500 MHz, D2O, 298 K): d=[4.82, dd, Ja,b =3.2 Hz, Ja,b =6.7 Hz, 0.3H;
2
1
[5] B. Imperiali, Acc. Chem. Res. 1997, 30, 452–459.
Procais], 4.65 (dd, Ja,b =2.4 Hz, Ja,b =9.6 Hz, 0.7H; Protarans), 4.57 (dddd,
2
1
[6] E. J. Bilsky, R. D. Egleton, S. A. Mitchell, M. A. Palian, P. Davis,
J. D. Huber, H. Jones, H. I. Yamamura, J. Janders, T. P. Davis, F. Por-
reca, V. J. Hruby, R. Polt, J. Med. Chem. 2000, 43, 2586–2590.
[7] a) B. Meyer, H. Moller, Top. Curr. Chem. 2007, 267, 187–251; b) D.
Specker, V. Wittman, Top. Curr. Chem. 2007, 267, 65–107.
[8] a) C. J. Bosques, S. M. Tschampel, R. J. Woods, B. Imperiali, J. Am.
Chem. Soc. 2004, 126, 8421–8425; b) S. E. OꢃConnor, B. Imperiali,
Chem. Biol. 1998, 5, 427–437; c) Y.-L. Pao, M. R. Wormarld, R. A.
Dwek, A. C. Lellouch, Biochem. Biophys. Res. Commun. 1996, 219,
157–162; d) R. Liang, A. H. Andreotti, D. Kahne, J. Am. Chem.
Soc. 1995, 117, 10395–10396.
[9] a) D. T. A. Lamport, Nature 1967, 216, 1322–1324; b) D. T. A.
Lamport, Recent Adv. Phytochem. 1977, 11, 79–115; c) R. B. Knox,
A. Clarke, S. Harrison, P. Smith, J. J. Marchalonis, Proc. Natl. Acad.
Sci. USA 1976, 73, 2788–2792; d) A. M. Showalter, Plant Cell 1993,
5, 9–23; e) D. J. Bowles, Annu. Rev. Biochem. 1990, 59, 873–907;
f) M. Josꢂ-Estanyol, P. Puigdomꢂnech, Plant Physiol. Biochem. 2000,
38, 97–108; g) J. Sommer-Knudsn, A. Bacic, A. E. Clarke, Phyto-
chemistry 1998, 47, 483–497; h) M. J. Kieliszewski, Phytochemistry
2001, 57, 319–323; i) Z. S. Khashimova, Chem. Br. Chem. Nat.
Comput. 2003, 39, 229–236.
Jg,d =1.0 Hz,
J
b ,g =2.5 Hz, Jg,d =4.6 Hz,
J
b ,g =4.6 Hz, 0.7H; Protgrans),
2
2
1
1
[4.51–4.54, m, 0.3H; Progcis], 3.84 (dd, Jd ,d =11.6 Hz, 0.7H; Protdrans), [3.80,
1
2
1
s, 0.9H; CO2CH3cis], 3.77 (s, 2.1H; CO2CH3trans), [3.64, dd, Jg,d
=
=
=
ꢀ
ꢀ
1
2
2
4.4 Hz, Jd ,d =13.4 Hz, 0.3H; Procdis], 3.61 (dd, 0.7H; Prodtrans), [3.47, Jg,d
1
2
1
2
1.0 Hz, 0.3H; Prodcis], 2.41–2.52 (m, 1.3H; Prob , Procis), 2.22 (ddd, Jb ,b
b2
1
1
2
14.2 Hz, 0.7H; Probtrans), 2.14 (s, 2.1H; NCOCH3trans), [2.08 ppm, s, 0.9H;
ꢀ
2
NCOCH3cis]; 13C NMR (75 MHz, D2O, 298 K): d=(174.7), 174.5,
(174.4), 173.8, 69.9, (68.8), (59.8), 57.9, 56.1, (54.9), (53.5), 53.3, (38.8),
37.3, 21.7, (21.5 ppm); MS (ES): m/z: calcd for C8H13NNaO4 [M+Na]+:
210.07; found: [M+Na]+: 209.73.
ꢀ
cis-4-O-(a-d-Galactopyranosyl)-N-acetyl-4-hydroxy-l-proline
methyl
ester (5): The general preparation method was followed for O-debenzyla-
tion of 12a (0.087 g, 0.123 mmol) to yield 5 as a clear oil (0.043 g,
0.123 mmol, quant.). [a]D25 =+93.68 (c=0.3, CH3OH); 1H NMR
(500 MHz, D2O, 298 K): d=4.90 (d, J=3.8 Hz, 0.75H; H1trans), [4.88, d,
J=3.8 Hz, 0.25H; H1cis], [4.69, dd, J=1.1, 9.2 Hz, 0.25H; Procais], 4.53 (d,
J=4.3, 7.8 Hz, 0.75H; Proatrans), 4.34–4.38 (m, 0.75H; Protgrans), [4.30–4.33,
m, 0.25H; Progcis], 3.78–3.87 (m, 2.5H; H3trans, H5, Protrans), [3.71–3.75, m,
d1
0.25H; H3cis], 3.45–3.71 (m, 8.25H; H2, H4, H6a
,
H6b
,
Procdis
,
Prod ,
2
1
CO2CH3), [2.48–2.55, m, 0.25H; Procis], 2.26–2.37 (m, 1.75H; Protbrans
,
ꢀ
b1
1
Prob ), 1.98 (s, 2.25H; NCOCH3 ), [1.92 ppm, s, 0.75H; NCOCH3cis];
trans
ꢀ
ꢀ
[10] A. B. Mauger, J. Nat. Prod. 1996, 59, 1205–1211.
2
13C NMR (75 MHz, D2O, 298 K): d=(174.39), (174.34), 174.2, 173.8, 99.1,
(98.7), 77.9, (76.3), 72.1, (71.6), 69.75, 69.69, (69.5), 68.4, (68.3), 61.7,
(61.3), (59.8), 57.9, 54.5, (53.6), 53.4, (36.4), 35.1, 21.64, (21.59 ppm); MS
(ES): m/z: calcd for C14H23NNaO9 [M+Na]+: 372.13; found: [M+Na]+:
372.03.
[11] a) J. F. Brandts, H. R. Halvorson, M. Brennan, Biochemistry 1975,
14, 4953–4963; b) M. W. MacArthur, J. M. Thornton, J. Mol. Biol.
1991, 218, 397–412; c) R. L. Stein, Adv. Protein Chem. 1993, 46, 1–
24.
[12] G. Fischer, F. X. Schmid, Biochemistry 1990, 29, 2205–2212.
10656
www.chemeurj.org
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2009, 15, 10649 – 10657

Prolyl Amide Isomerization
FULL PAPER
[13] a) A. M. Sapse, L. Mallah-Levy, S. B. Daniels, B. W. Erickson, J.
Am. Chem. Soc. 1987, 109, 3526–3529; b) S. Tanaka, H. A. Schera-
ga, Macromolecules 1974, 7, 698–705; c) G. D. Rose, L. M. Gierasch,
J. A. Smith, Adv. Protein Chem. 1985, 38, 1–109.
[14] a) L. Halab, F. Gosselin, W. D. Lubell, Biopolymers 2000, 55, 101–
122; b) L. Halab, W. D. Lubell, J. Am. Chem. Soc. 2002, 124, 2474–
2484; c) S. Kakinoki, Y. Hirano, M. Oka, Polym. Bull. 2005, 53,
109–118.
[37] N. Shamala, T. N. G. Row, K. Venkatesan, Acta Crystallogr. B 1976,
32, 3267–3270.
[38] a) E. S. Eberhardt, S. N. Loh, A. P. Hinck, R. T. Raines, J. Am.
Chem. Soc. 1992, 114, 5437–5439; b) E. S. Eberhardt, S. N. Loh,
R. T. Raines, Tetrahedron Lett. 1993, 34, 3055–3056.
[39] a) T. Matsuzaki, Y. Iitaka, Acta Crystallogr. B 1971, 27, 507–516;
b) G.-B. Liang, C. J. Rito, S. H. Gellman, Biopolymers 1992, 32,
293–301.
[15] a) R. A. Berg, D. J. Prockop, Biochem. Biophys. Res. Commun.
1973, 52, 115–120; b) A. V. Persikov, J. A. M. Ramshaw, A. Kirkpa-
trick, B. Brodsky, Biochemistry 2000, 39, 14960–14967.
[16] a) G. J. Van Holst, J. E. Varner, Plant Physiol. 1984, 74, 247–251;
b) E. Shpak, E. Barbar, J. F. Leykam, M. J. Kieliszewski, J. Biol.
Chem. 2001, 276, 11272–11278; c) P. J. Ferris, S. Waffenschmidt,
J. G. Umen, K. Ishida, T. Kubo, J. Lau, U. W. Goodenough, Plant
Cell 2005, 17, 597–615.
[40] F. K. Griffin, D. E. Paterson, P. V. Murphy, R. J. K. Taylor, Eur. J.
Org. Chem. 2002, 1305–1322.
[41] a) S. Krimm, J. Bandekar, in Advances in Protein Chemistry, Vol. 38
(Eds.: C. B. Anfinsen, J. T. Edsall, F. M. Richards), Academic Press,
New York, 1986, pp. 181–364; b) M. Jackson, H. H. Mantsch, Crit.
Rev. Biochem. Mol. Biol. 1995, 30, 95–120.
[42] GOESY (Gradient enhanced nuclear Overhauser effect spectrosco-
py) is a selective 1D NOESY experiment, see: J. Stonehouse, P.
Adell, J. Keeler, A. J. Shaka, J. Am. Chem. Soc. 1994, 116, 6037–
6038.
[17] L. E. Bretscher, C. L. Jenkins, K. M. Taylor, M. L. DeRider, R. T.
Raines, J. Am. Chem. Soc. 2001, 123, 777–778.
[18] J. A. Hodges, R. T. Raines, Org. Lett. 2006, 8, 4695–4697.
[19] a) S. S. Zimmerman, H. A. Scheraga, Macromolecules 1976, 9, 408–
416; b) S. S. Zimmerman, H. A. Scheraga, Biopolymers 1977, 16,
811–843.
[20] C. M. Taylor, R. Hardrꢄ, P. J. B. Edwards, J. H. Park, Org. Lett. 2003,
5, 4413–4416.
[43] a) J. T. Gerig, R. S. McLeod, J. Am. Chem. Soc. 1973, 95, 5725–
5729; b) M. Cai, Y. Huang, J. Liu, R. Krishnamoorthi, J. Biomol.
NMR 1995, 6, 123–128; c) C. A. G. Haasnoot, F. A. A. M. De
Leeuw, H. P. M. De Leeuw, C. Altona, Biopolymers 1981, 20, 1211–
1245.
[44] I. K. Song, Y. K. Kang J. Phys. Chem. B 2006, 110, 1915–1927.
[45] C. Benzi, R. Improta, G. Scalmani, V. Barone, J. Comput. Chem.
2002, 23, 341–350.
[21] E. S. Eberhardt, N. Panasik, R. T. Raines, J. Am. Chem. Soc. 1996,
118, 12261–12266.
[22] M. L. DeRider, S. J. Wilkens, M. J. Waddell, L. E. Bretscher, F.
Weinhold, R. T. Raines, J. L. Markley, J. Am. Chem. Soc. 2002, 124,
2497–2505.
[23] C. Renner, S. Alefelder, J. H. Bae, N. Budisa, R. Huber, L. Moroder,
Angew. Chem. 2001, 113, 949–951; Angew. Chem. Int. Ed. 2001, 40,
923–925.
[24] S. A. Cadamuro, R. Reichold, U. Kusebauch, H.-J. Musiol, C.
Renner, P. Tavan, L. Moroder, Angew. Chem. 2008, 120, 2174–2177;
Angew. Chem. Int. Ed. 2008, 47, 2143–2146.
[46] J. S. W. Lam, J. C. P. Koo, I. Hudꢅky, A. Varro, J. G. Papp, B. Penke,
I. G. Csizmadia, J. Mol. Struct. 2003, 666, 285–289.
[47] H. Friebolin, in Basic One- and Two-Dimensional NMR Spectrosco-
py, 3rd ed., Wiley-VCH, New York, 1998.
[48] F. A. Quiocho, Annu. Rev. Biochem. 1986, 55, 287–315.
[49] N. K. Vyas, M. N. Vyas, F. A. Quiocho, Science 1988, 242, 1290–
1295.
[50] D. A. Stauffer, R. E. Barrans, Jr., D. A. Dougherty, J. Org. Chem.
1990, 55, 2762–2767.
[25] L.-S. Sonntag, S. Schweizer, C. Ochsenfeld, H. Wennemers, J. Am.
Chem. Soc. 2006, 128, 14697–14703.
[51] A. Radzicka, L. Pedersen, R. Wolfenden, Biochemistry 1988, 27,
4538–4541.
[26] C. L. Jenkins, G. Lin, J. Duo, D. Rapolu, I. A. Guzei, R. T. Raines,
G. R. Krow, J. Org. Chem. 2004, 69, 8565–8573.
[27] M. D. Shoulders, J. A. Hodges, R. T. Raines, J. Am. Chem. Soc.
2006, 128, 8112–8113.
[52] A. Troganis, I. P. Gerothanassis, Z. Athanassiou, T. Mavromousta-
kos, G. E. Hawkes, C. Sakarellos, Biopolymers 2000, 53, 72–83.
[53] E. Beausoleil, W. D. Lubell, J. Am. Chem. Soc. 1996, 118, 12902–
12908.
[28] Y.-L. Pao, M. R. Wormarld, R. A. Dwek, A. C. Lellouch, Biochem.
Biophys. Res. Commun. 1996, 219, 157–162.
[54] However, there may be a slight destabilization of the pucker, be-
cause Alabugin et al. have proposed that there is an inverse rela-
ꢀ
[29] a) N. W. Owens, C. Braun, F. Schweizer, J. Org. Chem. 2007, 72,
4635–4643; b) K. Zhang, F. Schweizer, Synlett 2005, 20, 3111–3115;
c) K. Zhang, F. Schweizer, Carbohydr. Res. 2009, 344, 576–585;
d) K. Zhang, R. B. Teklebrhan, G. Schreckenbach, S. Wetmore, F.
Schweizer, J. Org. Chem. 2009, 74, 3735–3743.
[30] N. W. Owens, C. Braun, J. D. OꢃNeil, K. Marat, F. Schweizer, J. Am.
Chem. Soc. 2007, 129, 11670–11671.
[31] E. Beausoleil, R. Sharma, S. W. Michnick, W. D. Lubell, J. Org.
Chem. 1998, 63, 6572–6578.
[32] N. Panasik, Jr., E. S. Eberhardt, A. S. Edison, D. R. Powell, R. T.
Raines, Int. J. Pept. Protein Res. 1994, 44, 262–269.
[33] C. L. Jenkins, A. I. McCloskey, I. A. Guzei, E. S. Eberhardt, R. T.
Raines, Biopolymers 2005, 80, 1–8.
[34] L. Vitagliano, R. Berisio, A. Mastrangelo, L. Mazzarella, A. Zagari,
Protein Sci. 2001, 10, 2627–2632.
tionship between the hyperconjugative acceptor ability of the C X
bond and the electronegativity of X: I. V. Alabugin, T. A. Zeidan, J.
Am. Chem. Soc. 2002, 124, 3175–3185.
[55] M. P. Hinderaker, R. T. Raines, Protein Sci. 2003, 12, 1188–1194.
[56] H. L. Bagger, C. C. Fuglsang, P. Westh, Eur. Biophys. J. 2006, 35,
367–371.
[57] L. Pauling, in The Nature of the Chemical Bond, 3rd ed., Cornell
University Press, Ithaca, NY, 1960, pp. 281–282.
[58] M. Kumin, L.-S. Sonntag, H. Wennemers, J. Am. Chem. Soc. 2007,
129, 466–467.
[59] J.-C. Horng, R. T. Raines, Protein Sci. 2006, 15, 74–83.
[60] a) M. R. Wormald, R. A. Dwek, Structure 1999, 7, R155–R160;
b) K. D. White, R. D. Cummings, F. J. Waxman, J. Immunol. 1997,
158, 426–435.
[35] R. Improta, C. Benzi, V. Barone, J. Am. Chem. Soc. 2001, 123,
12568–12577.
[36] C. M. Taylor, R. Hardrꢄ, P. J. B. Edwards, J. Org. Chem. 2005, 70,
1306–1315.
Received: April 1, 2009
Revised: June 29, 2009
Published online: September 8, 2009
Chem. Eur. J. 2009, 15, 10649 – 10657
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
10657