2r. .3.
. 5,... z 7 {
a .

 

 

LA...“

. 2.4.
.fl
2.9

1
.
.

 

 

 

r.

L 3....“

 

u. .h. ans,
«Eflmfi .. a.
, :9... 4
wwmri

»

. fixing

. J“
, V c
3% .
. xfimum”
”Fa“ V

1.“:
ml
.9313; .u
QFICMK
1 .El.
Sta, 3

d. P
.9... \.1. . I
& n. 4... L y

This is to certify that the

dissertation entitled

STATISTICAL, COMPUTATIONAL, AND HER
SPECTROSCOPIC METHODS FOR PREDICTING
CARBOHYDRATE CONFORHATIOHS
presented by

HUSSEN MOHAMED

has been accepted towards fulfillment
of the requirements for

PhD degree in WHY

 

 

Date

 

MS U i: an Aflirmativc Action/Equal Opportunity Institution 0- 12771

 

LEIRARY
Michigan State
University K

 

 

 

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

q
1.1
E
3

 

 

 

 

 

 

 

 

11m mus-p.14

 

STATISTICAL, COMPUTATIONAL, AND NMR SPECTROSCOPIC METHODS
FOR PREDICTING CARBOHYDRATE CONFORMATIONS

By

Hussen Mohammed

A DISSERTATION

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY

Department of Chemistry

2001

ABSTRACT

STATISTICAL, COMPUTATIONAL, AND NMR SPECTROSCOPIC METHODS
FOR PREDICTING CARBOHYDRATE CONFORMATIONS

By

Hussen Mohammed

In the first part of this work, we report a simple rule based method for
predicting the conformations of oligo— and polysaccharides. The method is based
on an exhaustive analysis of the entries in the Cambridge Structural Database
(CSD) of crystal structures. We analyzed the values of the critical interglycosidic
dihedral angles, 05'-Cl’-OX-CX (phi) and C1'-OX-CX-CX+1 (psi) from all of the
entries which ranged from di-, tri-, tetra- and hexa- to octasaccharides. These
results were used to construct a second database of dihedral angles. An almost
exclusive preference for specific values of phi and psi for a specific type of
linkage independent of the context was clearly evident. There was symmetry of
placement of the points on the phi-psi plots that was determined by the absolute
configuration of the anomeric carbon and the carbon to which the linkage is
made. These corresponded to the RR, SS, SR, and RS configurations. Three
simple and accurate rules are proposed to help predict the conformations of
oligo- and polysaccharides. The results support, expand and generalize earlier
reports from other groups.

One of the important methods for determining the three-dimensional

structures of carbohydrates is using the proton-carbon hetero-nuclear coupling

constant to calculate the critical glycosidic dihedral angle values. In this work we
report the synthesis of seven compounds that have rigid glycosidic linkages.
These compounds are important to parameterize a new Karplus-type equation.
We used these seven compounds and three commercially available molecules to
parameterize new Karplus-Type equations relating the vicinal proton-carbon
coupling constants, 3JHC, to the glycosidic dihedral angles ¢ (H1'-C1'-OX-CX) and
w (HX-CX-OX-C1'). The three dimensional structures of the synthetic model
compounds 1-7 and the commercially available compounds 8-10 have been
refined by Molecular Mechanics and AW and PM3 Semi-empirical calculations.
The vicinal proton-carbon coupling constants, 3JHC, have been measured in
solution by two-dimensional excitation-sculptured indirect-detection experiment
(EXSIDE) NMR spectroscopy. From the least squares fitting of the measured
coupling constants and the glycosidic dihedral angles of the rigid molecules, new
Karplus-type equations of the form: 3JHc; (9) = a Cos2 (e) +b Cos (9) + c, were

obtained where 6 stands for the dihedral angles defined by the four atoms H-C-

O-C (o and w for carbohydrates).

Finally, we applied the methods developed in this work in combination with
existing methods using multi-nuclear, multidimensional NMR spectroscopy,
molecular mechanics calculations and molecular dynamics simulations to study
the solution conformations and dynamics of the disaccharide maltose and the
tetrasaccharide reduced-maltotetraose, and the pseudo-tetrasaccharide

acarbose, which is a potent glycohydrolase inhibitor.

Cepyright by
Hussen Mohammed
2001

To my son, Alula
my wife, Elizabeth

and my mother, Nun'a

ACKNOWLEDGMENTS

First I would like to thank my thesis advisor, Dr. Rawle Hollingsworth, for
his guidance and patience. I thank Rawle very much for teaching me how to
approach problems with a positive attitude. Without his guidance, the completion
of my dissertation would have been impossible.

I also want to thank my committee members, Drs. B. Borhan, J. F.
Harrison, T. J. Pinnavaia and D. L. Ward and my former committee member Dr.
C. K. Chang for their encouragement and for providing me helpful critiques of my
dissertation. I deeply thank Drs. Rawle l. Hollingsworth, James F. Harrison,
Eugene LeGoff, and Ardeshir Azadnia for writing reference letters.

I am indebted to the Max T. Rogers NMR Facility (Mr. Kermit Johnson and
Dr. Long Lee), the Mass Spectrometry Facility and the Computational Chemistry
Facilities of the Chemistry and Biochemistry Departments. Especially, I want to
thank Kermit for helping me with the multidimensional NMR experiments.

I also want to thank the Hollingsworth group members especially for the
fruitful discussions and for making the lab a lively place to work.

I would like to show gratitude to my mother Nuria and my late father
Mohammed, to my sisters, Radia, Halima, SOphia and Abrehet and to my
brothers, Jemal, Ali, Suleyman, Abdu, Kedir, and Hafiz for their encouragement,
love and support. I want to thank my mother for teaching me that education is an

important part of life.

vi

Finally, my deepest thanks goes to my wife, Elizabeth Yohannes, and my
son, Alula Hussen, for their support, encouragement, sacrifices, and

unconditional love.

Thank you all

Hussen

vii

TABLE OF CONTENTS

List of Tables ........................................................................................ x

List of Figures ...................................................................................... xiii

List of Schemes .................................................................................. xviii

List of Abbreviations .............................................................................. xix

CHAPTER 1

Introduction to Conformational Analysis of Carbohydrates .............................. 1
Introduction .................................................................................. 2
The Anomeric and Exo-anomeric Effects ............................................ 9
Crystal Structures from X-ray Crystallography .................................... 13
Solution Conformations from NMR spectroscopy ................................ 14
Molecular Mechanics Calculations and Molecular
Dynamics Simulations of Carbohydrates ........................................... 21
References ................................................................................ 26

CHAPTER 2

A Simple and Accurate Rule-based Method for Predicting the

Conformations of Oligo- and Polysaccharides ............................................. 28
Abstract ..................................................................................... 29
Introduction ................................................................................ 30
Results and Discussion ................................................................. 33
Conclusions ................................................................................ 48
Acknowledgements ...................................................................... 48
References ................................................................................. 49

CHAPTER 3

Design and Synthesis of Molecules with Rigid Glycosidic

Linkages .............................................................................................. 58
Abstract ..................................................................................... 59
Introduction ................................................................................ 60
Design and Synthesis .................................................................... 60
Conclusions ................................................................................ 88
Experimental .............................................................................. 89
Acknowledgements ...................................................................... 99
References ............................................................................... 1 00

viii

CHAPTER 4

New Karplus-Type Equations Relating the Glycosidic Dihedral

Angles and Vicinal Proton-Carbon Coupling Constants ............................... 101
Abstract ................................................................................... 102
Introduction .............................................................................. 1 03
Materials and Methods ................................................................. 107
Results and Discussion ............................................................... 113
Conclusions .............................................................................. 1 33
Acknowledgements .................................................................... 1 34
References ............................................................................... 1 35

CHAPTER 5

Studies on the Solution Conformation and Dynamics of
Maltose, Reduced Maltotetraose and the Potent Glycohydrolase

Inhibitor Acarbose .............................................................................. 137
Abstract ..................................................................... ‘ ............. 138
lntrod uction .............................................................................. 1 39
Materials and Methods ................................................................ 142
Results and Discussion ............................................................... 149
Conclusions .............................................................................. 1 73
Acknowledgements .................................................................... 1 74
References ............................................................................... 1 75

Table 1.1.

Table 1.2.

Table 1.3.

Table 1.4.
Table 2.1 .

Table 2.2.

Table 4.1 .

Table 4.2.

LIST OF TABLES

The solution conformations of the tetrasaccharide a-L-Fuc-(1—>2)-
B-D-Gal-(l —>3)-B-D-GalNAc-(1 —>3)-a-D-GaI-1-0Pr and its
component di- and trisaccharides ............................................ 18
The solution conformations of the tetrasaccharide B-D-GIcNAc-
(1 ->2)-a-D-Man-(1 —>3)-[B-D- GlcNAc-(1—>4)-]-B-D-Man and its
component di- and trisaccharides ........................................... 19
The solution conformations of the branched tetrasaccharide
B-D-Xyl-( 1 —>2) )[a-D-Man-(1—->3)-][oL-D-Man-( 1-—>6)-]-B-D-Man-OM6 and
its component di- and trisaccharides ....................................... 20
The solution conformations of a family of trisaccharides ............. 21
The percentages of the total phi-psi space that is populated
by oligosaccharide crystals from Cambridge Structural
Database .......................................................................... 35
Results of the statistical analysis of the dihedral angles
(«1) and w) of some of the oligosaccharides from the Cambridge
Structural Database ............................................................. 40
1H and 130 chemical shift assignments and proton-proton coupling
constants of 1,6-anhydroglucopyranose (8) ............................ 110
1H and “‘0 chemical shift assignments and proton-proton coupling

constants of 1,6-anhydrogalactopyranose (9) .......................... 111

Table 4.3.

Table 4.4.

Table 4.5.

Table 4.6.

Table 4.7.

Table 4.8.

Table 4.9.

1H and 13C chemical shift assignments and proton-proton coupling
constants of 1,6-anhydromannopyranose (10) ....................... 112
Heteronuclear three-bond proton-carbon coupling constants

of compounds 1-5 measured from 20 EXSIDE NMR experiments
and the corresponding dihedral angles (torsions). .................. 115
Heteronuclear three-bond proton-carbon coupling constants of
compounds 6-10 measured from 2D EXSIDE NMR experiments
and the corresponding dihedral angles (torsions). . . . . . . . . ..1 16
Structures of 1,6-anhydroglucose (8) from Cambridge

Structural Database and from MM, AM1 and PM3 geometry
optimizations ................................................................... 1 19
Structures of 1,6-anhydrogalactose (9) from Cambridge

Structural Database and from MM, AM1 and PM3 geometry
optimizations .................................................................. 120
Structures of 1,6-anhydromannose (10) from Cambridge

Structural Database and from MM, AM1 and PM3 geometry
optimizations .................................................................. 121
The values of the dihedral angle C2-C3-C4-CS as a measure of the
degree of flattening of the chair conformations of the three anhydro
sugars, 1,6-anhydroglucose (8), 1,6-anhydrogalactose (9), and 1,6-
anhydromannose (1 0) from Cambridge Structural Database and

from MM, AM1 and PM3 geometry optimizations .................... 122

xi

Table 4.10.

Table 5.1 .

Table 5.2.

Table 5.3.
Table 5.4.

Table 5.5.

Table 5.6.

Table 5.7.

The values of the parameters a, b, c and the value of R2 from a
least squares fitting of the heteronuclear coupling constants and the
dihedral angles given in Tables 4.4 and 4.5 ........................... 131
1H and 130 chemical shift assignments of maltose .................. 144
1H and 13C chemical shift assignments of reduced

maltotetraose ................................................................. 1 45
1H and 13C chemical shift assignments of acarbose ............... 146
Distances calculated from NOESY experiment for reduced
maltotetraose .................................................................. 1 47
Distances calculated from NOESY experiment for

acarbose ....................................................................... 148
Glycosidic dihedral angles from constrained minimization and

grid search studies and from crystal structures ...................... 171
Comparison of the predicted vicinal proton-carbon coupling
constants using the new Karplus-type equations with that of the
three previous equations of Mulloy et. al., Tvaroska et. al., and
Cloran et. al. with the experimentally measured values. The
results are averages of the 2000 data points from Molecular

Dynamics simulations ...................................................... 172

xii

Flgure 1.1.

Flgure 1.2.

Flgure 1.3.

Flgure 1.4.

Flgure 1.5.

Flgure 1.6.

Flgure 2.1.

LIST OF FIGURES

The aldotrioses D- and L-glyceraldehyde ................................. 4
The structure and stereochemical relationships of D-aldoses
having three to six carbon skeletons. The highest numbered
asymmetric carbon determines the configuration of the

aldose in each case ............................................................. 5
Cyclization of D-glucose in aqueous solution resulting

in the formation of two pyranoses, which show distinct

physical, chemical and biological properties .............................. 8
The 4C1 conformation of B-D-glucose and the 1C.

conformation of B—L-fucose ..................................................... 8
The Twist and Envelope conformations of furanoses, with
the twist conformation having three of the ring atoms in

plane and the envelope conformation with four of the

ring atoms in plane ............................................................... 9
The anomeric effect that relates to the preference for
the axial orientation of the aglycon of glycopyranosides

and the exo-anomeric effect that relates to the preference
for the aglyconic carbon to be in near syn-clinal orientation
to both the ring oxygen and the anomeric hydrogen ................... 10
Structure of 1-O-methyl cellobioside and the glycosidic
dihedral angles, phi and psi, which determine its conformation.

Phi = 05'-C1'-O4-C4 and Psi = C1 ‘-O4-C4-C5 ........................ 31

xiii

Figure 2.2.

Figure 2.3.

Figure 2.5.

Flgure 2.6.

Flgure 3.1.

Figure 3.2.

Flgure 3.3.

Figure 3.4.

Flgure 3.5.

Biograf rigid residue phi-psi maps of (a) cellobiose (B—D-GIc-(1 -
>4)BD-Glc) and (b) melliobiose (oc-D-GaI-(1->6)-[3-D-Glc) ........... 36
(D)? map of 1-2, 1-3, 1-4 and 1-6 linked oligosaccharides from the
Cambridge Structural Database (CSD). Each point on the map
corresponds to one glycosidic linkage from the CSD .................. 41
Line representation of the structure of dermatan 4-sulfate predicted
using the rules in this paper and from X-ray fiber diffraction
study.104 ........................................................................... 44
Line representation of the structure of hyaluronan (sodium
hyaluronate Ill) predicted using the rules in this paper and from X-
ray fiber diffraction study.104 ................................................. 46
The ten molecules that have rigid glycosidic dihedral angles. While
molecules 8-10 are commercially available, molecules 1-7 are
synthesized in this work ....................................................... 62
Proton and carbon nuclear magnetic resonance spectra of 120-
[2-methyl-(S)-1 ,2-ethanediyl]-a-L-fucopyranoside (1) ................. 66
Proton and carbon nuclear magnetic resonance spectra of 4,60-
benzyledene-1 ,2-0-[2-methyl-(R)-1,2-ethanediyl1-a -D-
glucopyranoside (2) ............................................................ 69
Proton and carbon nuclear magnetic resonance spectra of 1,20-
[2-methyI-(R)-1 ,2-ethanediyl]-B-L-fucopyranoside (3). . . . . . . . . .73
Proton and carbon nuclear magnetic resonance spectra of 3-0-

methyl-1 ,2-0-[2-methyI-(S)-1 ,2-ethanedin]-B-D-

xiv

Figure 3.6.

Figure 3.7.

Flgure 3.8.

Flgure 4.1.

Flgure 4.2.

Figure 4.3.

Figure 4.4.

Flgure 4.5.

glucopyranoside (4) ............................................................ 76
Proton and carbon nuclear magnetic resonance spectra of 4,60-
benzyledene-1 ,2-0-[2-methyl-(S)-1 ,2-ethanedin]-B-D-
glucopyranoside (5) ............................................................. 80
Proton and carbon nuclear magnetic resonance spectra of 4.6-0-
benzyledene-1 ,2-0-[2-methyl-(S)-1 ,2-ethanediyl]~B-D-
galactopyranoside (6) .......................................................... 83
Proton and carbon nuclear magnetic resonance spectra of 3-0-
benzyl-1 ,2-0-[2-methyI-(R)-1 ,2-ethanediyl]-B-D-mannopyranoside
(7) ................................................................................... 86
The ten molecules that have rigid glycosidic dihedral angles. While
molecules 8-10 are commercially available, molecules 1-7 are
synthesized in this work (Chapter 3) .................................... 105
The new data points plotted against the Karplus-type curves of the
previous equations .......................................................... 124
The new‘AM1 data points plotted against the Karplus-type curves
of the previous equations and the new equations. ............... 126
The new PM3 data points plotted against the Karplus-type curves
of the previous equations and the new equations ................ 128
The difference in the predicted values of the vicinal coupling
constants between the new Karplus-type equations (AM1* and
PM3*) and the two previous equations (Tvaroska et. al and Mulloy

et. al.) .......................................................................... 130

XV

Flgure 5.1.

Figure 5.2.

Flgure 5.3.

Figure 5.4.

Flgure 5.5.

Figure 5.6.

Figure 5.7.

Schematic diagram of maltose, reduced maltotetraose, and
acarbose. ....................................................................... 140
Rigid residue potential energy (PE) surfaces of maltose (top)

and acarviosine (the a-b unit of acarbose) (bottom). lsoenergy
contours are drawn with interpolation of 1 kcal/mol above the
minimum. ...................................................................... 151
The molecular dynamics population map of maltose, shown above,
is consistent with the PE surface. The dihedral angles are defined
as (05'-C1'-O4-C4) and (C1'-O4—C4-C5) ........................... 153
Solution conformations of maltose, reduced maltotetraose and
acarbose as determined from constrained minimization using
distance constraints from 2D NOESY experiments ................. 156
Molecular dynamics population maps of reduced

maltotetraose from 200 ps simulation starting from its constraint-
minimized structure. The plots are for the glycosidic linkages

a-b, b-c, and c-d ............................................................. 157
Molecular dynamics trajectories of the phi dihedral angle of reduced
maltotetraose from 200 ps simulation starting from its constraint-
minimized structure. The plots are for the glycosidic linkages a-b
(top), b-c (middle) and c-d (bottom) ...................................... 159
Molecular dynamics trajectories of the psi dihedral angle of reduced

maltotetraose from 200 ps simulation starting from its constraint-

Flgure 5.8.

Figure 5.9.

Figure 5.10.

minimized structure. The plots are for the glycosidic linkages a-b, b-
c, and c-d ...................................................................... 160
Molecular dynamics population maps of acarbose from 200 ps
simulation starting from its constraint-minimized structure. The plots
are for the glycosidic linkages a-b, b-c, and c-d ...................... 164
Molecular dynamics trajectories of the phi dihedral angle of
acarbose from 200 ps simulation starting from its constraint-
minimized structure. The plots are for the glycosidic linkages a-b,
b-c, and c-d .................................................................... 166
Molecular dynamics trajectories of the psi dihedral angle of
acarbose from 200 ps simulation starting from its constraint-
minimized structure. The plots are for the glycosidic linkages a-b,

b-c, and c-d ...................................................................... 168

LIST OF SCHEMES

Scheme 3.1 Synthesis of 1,2-0-[2-methyl-(S)-1,2-ethanedin1-a-L-
fucopyranoside (1) .............................................................. 65
Scheme 3.2 Synthesis of 4,6-O-benzyledene-1,2-0-[2-methyl-(R)-1,2-
ethanediylj-oc -D-glucopyranoside (2) .................................... 68
Scheme 3.3 Synthesis of 1,2-0—[2-methyI-(R)-1,2-ethanediylj-B-L-
fucopyranoside (3) .............................................................. 72
Scheme 3.4 Synthesis of 3-O-methyl-1,2-0-[2-methyI-(S)-1,2-ethanediyl]-B-D-
glucopyranoside (4) ............................................................ 75
Scheme 3.5 Synthesis of 4,6-O-benzyledene-1,2-0-[2-methyl-(S)-1,2-
ethanediyl]-B-D-glucopyranoside (5) ........................................ 79
Scheme 3.6 Synthesis of 4,6-O-benzyledene-1,2-0-[2-methyl-(S)-1,2-
ethanediyl]-B-D-galactopyranoside (6) .................................... 82
Scheme 3.7 Synthesis of 3-O-benzyl-1,2-0-[2-methyl-(R)-1,2-ethanediyll-B-D-

mannopyranoside (7) .......................................................... 85

xviii

Ac

A020

dd

DMF
2-D NMR
COSY
eq.
EXSIDE
FAB
HMQC
HRMS
Hz

LAH
LRMS

MeOH
MM
MD

m.p.

LIST OF ABBREVIATIONS

Acetyl

Acetic anhydride

Doublet

Doublet of doublet

Chemical Shift

N,N-Dimethyl formamide

Two-dimensional Nuclear Magnetic Resonance
Correlated spectroscopy

Equivalent

Excitation-Sculptured Indirect-Detection Experiment
Fast Atom Bombardment

Heteronuclear Multi Quantum Coherence spectroscopy
High Resolution Mass Spectrometry

Hertz

Lithium aluminum hydride

Low Resolution Mass Spectrometry

Multiplet

Methanol

Molecular Mechanics

Molecular Dynamics

Melting point

xix

NMR
n.O.e.

NOESY

r.t.

TFA
THF
TLC
TOCSY
pTSA

Nuclear Magnetic Resonance
Nuclear Overhauser enhancement
Nuclear Overhauser enhancement spectroscopy
Pyridine

Quartet

Room temperature

Singlet

Trifluoroacetic acid
Tetrahydrofuran

Thin Layer Chromatography

Total Correlated Spectroscopy

p-Toluene sulfonic acid

XX

Chapter 1

Introduction to Conformational Analysis of Carbohydrates

Introduction

Carbohydrates, lipids, DNA, and proteins are four major classes of
macromolecules in biology. While proteins are linked by amide bonds and
nucleic acids by 3'-5' phosphodiester bonds, carbohydrates can be connected by
many different linkage types. In addition, while proteins and nucleic acids are
almost exclusively linear, carbohydrates can be highly branched. This
complexity allows carbohydrates to provide almost unlimited variations in their
structure.

Carbohydrates are commonly classified as monosaccharides,
disaccharides, trisaccharides, oligosaccharides and polysaccharides. A
monosaccharide is that unit which cannot be further hydrolyzed into smaller
carbohydrates. Di-, tri-, tetra-, and pentasaccharides can be hydrolyzed to two,
three, four, and five monosaccharide units, respectively. Glycoconjugates are
carbohydrates that are covalently attached to other biomolecules like proteins
(glycoproteins) and lipids (glycolipids and Iipopolysaccharides).

Monosaccharides can be represented by the formula anzno. or C..(H20),,.
The latter representation led to the name “carbohydrates” or hydrates of carbon.
Monosaccharides are further classified as triose, tetrose, pentose, hexose,
heptose, and octose depending on the number of atoms in the carbon skeleton.
Those monosaccharides with an aldehyde functional group are called aldoses
and those with a ketone functional group are called ketoses. The number of
carbon atoms and the position of the carbonyl group are both included in the

general terms used for identifying the monosaccharides. Aldohexose refers, for

example, to six-carbon sugars with an aldehyde functional group like glucose or
galactose. And ketoses are named by adding the suffix “ulose” to the base
name, which indicates the number of carbon atoms. Laevulose, commonly
known as fructose, is a hexulose containing six carbon atoms and a carbonyl
group at position 2. The simplest monosaccharides are water soluble, and most
taste sweet.

Glyceraldehyde, an aldotriose, is the smallest monosaccharide. When the
central carbon of glyceraldehyde has the R configuration, it is referred as D-
glyceraldehyde and when it has the 3 configuration it is referred as L-
glyceraldehyde (Figure 1.1). The higher aldoses derived from D-glyceraldehyde
belong to the D-series and those derived from L-glyceraldehyde belong to the L-
series. D-Sugars are the most common form of sugars in nature, in contrast to
the amino acids that commonly exist as L-amino acids in nature. L-Sugars such
as L-rhamnose, L-arabinose, and L-fucose also exist in nature. The higher
aldoses and ketoses are derived from their respective aldotrioses and
ketotetroses by inserting one or more hydroxymethylene (HCOH) groups
between the first chiral center and the carbonyl group of the corresponding
isomer (see Figure 1.2 for D-aldoses). A new chiral center is created by each
insertion of hydroxymethylene group, thus creating two isomers. In a given
monosaccharide with n chiral centers, there are 2" isomers or 2"'1 enantiomeric
pairs. An aldohexose, for example, has 2‘ = 16 isomers considering both D and

L forms.

The mnemonic, familiarized by Louis and Mary Fieser of Harvard
University, “all altruists gladly make gum in gallon tanks” is a very convenient
way to remember the eight aldohexoses. Each word in the mnemonic
symbolizes one enantiomer of the hexoses. And the mnemonic “Right AXLe” is a
convenient way to remember the aldopentoses, where the four capitalized letters

denote the first letter of the four aldopentoses (Figure 1.2).

CHO CHO
H—C—OH HO—c—H
CH20H CHZOH
D-glyceraldehyde L-glyceraldehyde

Figure1.1 The aldotrioses D- and L-glyceraldehyde.

The dominant forms of monosaccharides in solution are the cyclic
hemiacetal forms. Sugar hemiacetals are formed by addition of an alcohol to an
aldehyde or a ketone group resulting in anomerization to give two isomers a and
B, where the hydroxyl group at the anomeric carbon is axial or equatorial,
respectively (Figure 1.3). Aldoses with four or more carbon atoms and ketoses
with five or more carbon atoms can form five membered rings, furanose
structures. And aldoses with five or more carbon atoms and ketoses with six or

more carbon atoms can form six membered rings, pyranose structures. These

Figure 1.2. The structures and stereochemical relationships of D-aldoses that
have three to six carbon skeletons. The highest numbered asymmetric carbon

determines the configuration of the aldose in each case.

cab 083.0 5.9 880100 60.0 850 130082200 .cmzvmmocéz0 0.0088200 5008.20 6.3 822.0

10.10 10.10 10.10 10.10 10m10 1005 10.10 1010 0
101M_T1 101m11 101011 101011 101011 101011 101011 1010.11 0
1&101 11%101110101 11.0101 101011 101011 101011 101011 e
11m101 10101101011 101011 110101 110101101011 10101 0
..1... 5-... I III ..1... III II I
010 010 010
90?: 00903.0 . 93c 0003xé 3.3 00050020 3?: 00003-0
1010.011 101w11 101w11 10.1«11 v
IIOIQ._ IIOIQ1 10101: 101011 m
110101 101011 110101 101011 m
0.6 Q10 Q10 Q10 F
mwowfké mmoEthé
10.10 101m a.
101w11 101011 0
E 101011 N
010 025208200 010 F
:03 m
101n_T1 m

0.6 F

tautomers show distinct physical, chemical and biological properties. The
composition of sugars in solution, which can be estimated by NMR spectroscopy,
depends on the nature of the solvent, the temperature, and the nature of the
substituents at the different carbon atoms. Optical rotation measurement can
also be used to differentiate between the two anomers. In the D-series, the more
dextrorotatory anomer of each sugar is designated 01 and the less dextrorotatory
is designated B.‘ A freshly prepared solution of a—D-glucose, for example, has an
optical activity, [011020, of +112.2° while that of B-D-glucose is +18.7°. The
formation of cyclic sugars creates an additional chiral center at the anomeric
carbon, Cl in aldoses and C2 in ketoses. Although Haworth projection formulas
of sugars are still used by some, they are somewhat misleading because they
suggest that the five and six membered rings are planar which is not the case.
For aldohexoses a more clear representation is the chair conformation. There
are two possible chair conformations designated 4C1 and ‘C4 (Flgure 1.4) where
C stands for chair. In the notation 401, for example, the superscript 4 stands for
the atom puckered above the plane of the chair while the subscript 1 stands for
the atom puckered below the plane of the chair.

The furanose form is another form, which is found in nature for some of
the monosaccharides such as the aldopentoses ribose and deoxyribose in
ribonucleic acids (RNA) and deoxyribonucleic acids (DNA). The furanose rings
can have two types of conformations, the envelope (E) and the twist (T). In the

twist conformation, three of the ring atoms are in plane and one ring atom is

0” 0H 0H 01-1
@H V— &’m 11$;
0'1 +
O OH OH OH
Pyranose form
Figure 1.3 Cyclization of D-glucose in aqueous solution resulting in the formation

of two pyranoses, which show distinct physical, chemical and biological

properties.

OH 6 5 1
4 6 5 20 OH CH3 0 OH
HO1-10 0H1 4 2 OH
3 OH OH
4C1 104

Figure 1.4 The “c. conformation of B-D-glucose (left) and the ‘c. conformation

of B-L-fucose.

above and one below the plane of the ring. While in the envelope conformation,

only one of the ring atoms is out of plane (Figure 1.5).

 

A‘ /\;_—
v v

Twist form Envelope form

Flgure1.5 The Twist and Envelope conformations of furanoses, with the twist
conformation having three of the ring atoms in plane and the envelope

conformation with four of the ring atoms in plane.

The Anomeric and Exo-anomeric Effects and Carbohydrate

Conformational Analysis

The anomeric and exo-anomeric effects are phenomenon that specially
relate the conformations of sugars and deserve some special comment. The
anomeric effect, which refers to a special bonding effect found in acetals and
related structures was outlined by Lemieux and Koto.2 In anomeric and exo-
anomeric effects (Figure 1.6), extra bonding is provided to the carbon-oxygen
bonds of acetals due to the participation of the lone pairs of electrons when the
lone pair of electrons is in a specific orientation relative to the neighboring bonds

with the C-0 bond being the strongest when the lone pairs of electrons are anti-

gig OH 9:

HO % ‘CH3
HO

HO OH HO OH

CH3

Axial OMe at the anomeric position prefered over
equatorial OMe due to the anomeric effect

OH OH
9;) 90
110% 92C H3 H0 %'CH3
HO OH HO OH

Syn-clinal methyl at the anomeric oxygen
due to the exo-anomeric effect

Flgure 1.6 The anomeric effect that relates to the preference for the axial
orientation of the aglycon of glycopyranosides and the exo-anomeric effect that
relates to the preference for the aglyconic carbon to be in near syn-clinal

orientation to both the ring oxygen and the anomeric hydrogen.

10

periplanar to the neighboring C-O bond. Physically there is no difference
between the anomeric and exo-anomeric effects.3 The two different terms were
used for the simple practical reason that the anomeric effect relates to the
preference for the axial orientation of the aglycon of glycopyranosides while the
exo-anomeric effect relates to the preference for the aglyconic carbon to be in
near syn-clinal orientation to both the ring oxygen and the anomeric hydrogen
(Flgure 1.6). In the exo-anomeric effect the lone pairs of the exo-oxygen
stabilize the ring CO bond and hence favor syn-clinal methyl at the anomeric
oxygen.

As can be seen from the discussions and figures in the preceding
sections, carbohydrates can provide almost unlimited variations in their structure
and generate a wide variety of complex structures because they can be attached
to each other in many different linkage types and they can be highly branched.
For example, if we consider a disaccharide of say D-glucose and D-galactose, it

can have 32 different structures (each monomer can be 01 or B, so the

disaccharide can be 0101, 043,801, or [58 and the two monomers can be linked in
four different ways: 1->2, 1->3, 1->4, and 1->6 and the reducing and can be the
glucose or the galactose residue). So, carbohydrates provide diverse three-
dimensional (30) structural possibilities and have several applications. For
example, recent studies have demonstrated that oligosaccharides are involved in
a number of recognition events such as cell adhesion, metastasis, fertilization
and embryonic development, amongst others.“'5 In view of their diverse function

in a variety of biological systems, the primary structures and SD structures of

11

carbohydrates are of great interest. Due to their diversity and the inability to
crystallize such conjugates, the largest pure carbohydrates crystallized so far are

° nuclear magnetic resonance (NMR) spectroscopy in

tetrasaccharides,“
combination with molecular mechanics calculations, is the method of choice for
their identification and conformational analysis.

A major problem when using NMR spectroscopy to study carbohydrates is
that, except for the anomeric protons, all the protons on oxygenated atoms (e.g.,
those on carbons 2-6 in aldohexoses) appear in a range of ~3.2 — ~5.3 ppm.
Hence, as the number of monosaccharide units increases, the assignment of
peaks for each proton (and carbon) becomes increasingly difficult. This means
extracting coupling constants and NOESY volumes--which are useful for
determining dihedral angles and distances, respectively--becomes very difficult.
Doing pure molecular mechanics and grid search studies of higher oligomers and
polymers requires a lot of computational space and time. For example, consider
a hexasaccharide that has ten glycosidic bonds and thus ten dihedral angles
across the glycosidic linkages. To do a simple molecular mechanics grid search
study of the hexasaccharide by 10° rotations of each dihedral angle
independently involves 36 possible values for each dihedral angle. This gives,
for the ten dihedral angles, 3610 z 3.7 x 1015 starting conformations to be
evaluated by molecular mechanics. In addition, these calculations when used

alone cannot give reliable or meaningful information on structure. So, the

question is how best to predict the three dimensional structure of carbohydrates.

12

The focus of this work is therefore on developing methods that can be
used to predict carbohydrate conformations. In the next sections of this chapter
we will survey the three major current methods available for the conformational
analysis of carbohydrates. These are X-ray crystallography, nuclear magnetic

resonance spectroscopy, and computational methods.

Crystal Structure from X-ray Crystallography

The most direct and powerful experimental tool to determine molecular
structures and intermolecular interactions at atomic resolution is X-Ray diffraction
from single crystals. Under crystallizing conditions helical structures invariably
grow much faster along the helix axis than in the other directions. This means, it
is seldom possible to achieve the isotropic growth rate necessary to produce
single crystals in the case of helix forming polymers such as polysaccharides.

This is a major experimental setback with non-crystalline materials.
However, it is often possible to prepare polycrystalline and/or oriented specimens
in the form of fibers or films that are suitable for X-ray diffraction analysis.
Because of this X-ray structures of several polysaccharides composed of simple
to complex repeating units have been determined. These include the structural
polysaccharides; cellulose, mannan, chitin, and xylan; the storage
polysaccharide amylose and some of its derivatives: hyaluronan, chondroitin,
keratan, and other polysaccharides in the glycosaminoglycan family; agarose,
alginate, carrageenans, curdlan, gellan, and pectates, which are gel-formers; and

branched polysaccharides such as galactomannan, welan, and xanthan.11

13

A library of the solid-state conformations of carbohydrates from X-ray
crystallography studies will be built and statistically analyzed to come out with a
simple and accurate rule-based method to predict carbohydrate conformations

(Chapter two).

Solution Conformation from NMR Spectroscopy

Nuclear magnetic resonance spectroscopy (NMR) in combination with
molecular mechanics and molecular dynamics studies is one of the best known
and widely used experimental methods for the conformational analysis of
carbohydrates in solution. The first task here is to perform combinations of
homonuclear— and heteronuclear-(one- and) two-dimensional NMR experiments
in order to allow the assignments of the signals for the oligosaccharide. This
include: double quantum filtered J-correlated spectroscopy (DQF—COSY)
spectra,12 total correlated spectroscopy (TOCSY) spectra,13 heteronuclear multi-
quantum coherence (HMQC) experiment,14 heteronuclear multi-bond coherence
(HMBC) experiment,15 and nuclear Overhauser effect (n.O.e.) experiment .‘6

Analysis of the spin systems for the glycoside ring systems with 1-D traces
from DQF-COSY and TOCSY spectra gives the connectivities. Once the
assignments are made, n.O.e. experiments are done to determine the various
inter-nuclear distances. Spectra of the n.O.e. experiments are acquired at
different mixing times to address the potential problem of spin diffusion.

Assuming isotropic motion then, the cross-relaxation rates and thus inter-nuclear

distances between nuclei i and j, r“, for each pair wise interaction will be

14

extracted from the volume of the NOESY peak, V", measured at different mixing

times using the relation:

I'ij = rref (Vref/ViII1I6 (1)

where, r = distance, V = volume, ref = reference pair of nuclei the distance
between which is fixed and known. This is followed by constrained minimization
using the distances obtained at the previous step to determine the average
conformation about glycosidic linkages in solution.

The second important structural information comes from the calculation of
dihedral angles from coupling constants measured from NMR experiments. To
calculate the dihedral angles between the glycoside residues from the vicinal

17-19

proton-carbon coupling constants, 3JHC, we use a ‘Karplus type’ equation of

the form:

3J0. = a cos2 (111) + b Cos (¢) + c (2)

where 3J0" is the three bond proton-carbon hetero-nuclear coupling constant and
(1 is dihedral angle. The Haasnoot - De Leeuw - Altona relationship20 can be
used to calculate the co dihedral angle defined as H5-C5-06-H6 between the
glycoside residues from the coupling constants 3JHH for 1—16 glycosidic linkages.
The final step is to use the distances and the angles obtained from the

different NMR spectroscopy experiments as constraints in the force fields of

15

molecular mechanics calculations are performed using, for example, the Drieding
force fields21 implemented in the BIOGRAF program. The Drieding force field
includes a harmonic term for the bond distortions, a simple harmonic function for
the valence angle term, an improper torsion term to evaluate inversions about
atomic centers, a cosine expansion torsional term, a Coulombic potential for
evaluating electrostatic energy, a Leonard-Jones 12-6 potential for van der
Waals contributions and a Leonard-Jones 12-10 potential for evaluating
hydrogen bonding contributions.

NMR spectroscopy data represent an average conformational space when
used for the determination of SD structure. Often it is not possible to deduce the
magnitude of internal motions from NMR data alone. To further complicate the
matters, the average is not a linear average in the case of n.O.e. measurements,
but weighed by the r'6 dependence of the n.O.e. up on the inter-nuclear
distance. 22 This means that a single n.O.e. measurement is not a very precise
tool. The measurement of an n.O.e. of magnitude n "/0 could be interpreted in
terms of a semi-rigid conformer where the n.O.e. corresponds to exactly to an
inter-nuclear distance r, or equally, it could correspond to a situation where the
two nuclei are in proximity for 10 % of the time, and a considerable distance
apart for 90 % of the time, thus leading to a completely erroneous conclusions
regarding the dominant conformer if interpreted as a semi-rigid body. In
oligosaccharides, the number of constraints is small, and long-range constraints
are almost never observed. Therefore, n.O.e. data should be used in

combination with theoretical calculations. The combination of n.O.e. data and

16

theoretical predictions achieves several desirable purposes. First, the energy
minimized structure obtained using the n.O.e. constraints is more likely to
represent the global minimum energy structure than the use of purely theoretical
energies, since the available conformational space is restricted by the additional

2425 at ambient

constraints.23 Second, molecular dynamics simulations
temperature (300K) can be run on the energy minimized structure in the
presence and absence of n.O.e. constraints. The former allows one to determine
whether an alternative, lower energy conformer exists by searching over the
conformational space near the obtained minimum. The latter gives some
measure of the extent of molecular motion in the sense that the extent of
torsional oscillations about 111 and w for each linkage can be measured. This is
important because the energy minimized structures representing the
conformation at 0 K are not biologically important, and we are more interested in
the ‘conformation’ at normal temperatures. Since this is a dynamic structure, the
‘solution conformation’ is defined as an average in Cartesian space together with
root mean square deviations per atom with respect to the center of mass to gain
some measure of mobility.

Another approach in combining n.O.e. data with theoretical studies
involves calculation of the populations of conformers on the potential energy
surface followed by computation of theoretical n.O.e.s from these conformers,
and comparison with those generated experimentally.‘°"5'27

In the next few paragraphs we will discuss some examples of the solution

conformations of oligosaccharides from the literature. These examples

17

demonstrate that the solution conformation of disaccharides seems to be
preserved in structures containing three or more monosaccharide units.

The preferred solution conformations of 01-L-Fuc-(1—-)2)-j3-D-Gal-(1->3)-B-
D-GalNAc-(1->3)-0t-D-Gal-1-OPr and its component di- and trisaccharides were
determined by combination of n.O.e. and MM and MD studies”29 The results
are summarized in Table 1.1 where a, b, c and d denote the first, second, third
and fourth monosaccharide units, respectively. The consistency of the values of
the dihedral angles in the different oligosaccharides show that the conformations
of the disaccharides a-b, b-c, and c-d is preserved in the trisaccharides a-b-c and

b-c-d, and the tetrasaccharide a-b-c-d.

Table 1.1. The solution conformations of the tetrasaccharide 01-L-Fuc-(1—92)-B-

D-Gal-(1—>3)-B-D-GalNAc-(1——>3)-0t-D-Gal-1-OPr and its component di- and

 

 

 

 

 

 

 

 

 

 

 

trisaccharides.

Molecule (Ii/vi of (ti/1y of My of
a-b b-c c-d

01-L-Fuc-(1 —>2)-B-D-Gal 36/23

B-D-aGal-(1 —>3)-B-D-(:aINAc 39/20

B-D-bGalNAc-(1 —->3)-01-D-Gal 28/33

a—L-Fui-(1—12)-B-D-Gacl,-(1 —>3)-B-D-GalNAc 36/27 40/27

B-D-(gal-U —>3)-B-D-GalI)NAc-1 —>3)-0t-D-Galc 38/23 32/34

a—L—Futc-(1——>2)-B-D-Gal-(‘1:—>3)-B-D-GaIN‘Ac - 34/26 40/23 25/38

(1 —)3a)—01-D-C:al b c

 

18

 

A similar study was done on B-D-GlcNAc-(1—>2)-0t-D-Man-(1->3)-[B-D-
GlcNAc-(1—>4)-]-B-D-Man and related di- and trisaccharides.30 The results are
summarized in Table 1.2 where a, b, c and 0 denote the first, second, third and
fourth monosaccharide units, respectively. Here again, the conformation of the
individual disaccharides is consistent the same disaccharide in the trisaccharide
and tetrasaccharide units. Note that, while the previous example contains only
linear carbohydrate chains, the second example contains branched

oligosaccharides (a-b-[c-]-d).

Table 1.2. The solution conformations of the tetrasaccharide B-D-GlcNAc-

(1—>2)-01-D-Man-(1->3)-[B-D-GlcNAc-(1—>4)-]-B-D-Man and its component di- and

 

 

 

 

 

trisaccharides.
Molecule 111/11; 111/11; (ti/111
of a-b of b-d of c-d
a-D-Man-(1 —->3) )-B-D-Man -50/-10
b cl
B-D-GlcNAc-(1—->2)-01-D-Man-(1 —>3)-B-D-Man 40/30 -50/-20
a b d
B-D-GIcNAc-(1-—>2)-01-D-Man-(1—->3)-[B-D-GlcNAc- 40/30 -50/-20 60/10
a b c
(1 —94)-]-B--D-Man
d

 

 

 

 

 

In the third example the solution conformation of three oligosaccharides

have been carried out using rotating frame n.O.e. experiments in combination

19

with Hard Sphere Exo-Anomeric effect calculations31 and the results are shown
in Table 1.3 where 0’ and 111’ are the dihedral angles defined by 05’-C1’-OX-CX
and C1’-OX-CX-CX+1, respectively (where X: 2, 3 or 6). Except for the 1—>6
linkage, which is more flexible than other linkage types due to its three bonds
across the glycosidic linkage, the geometry of the other linkages has only one

value and do not change much from one compound to another.

Table 1.3. The solution conformations of the branched tetrasaccharide B-D-Xyl-

(1-92) )[a-D-Man-(1—>3)-][ot-D-Man-( 1—>6)-]-8-D-Man-0Me and its component and

 

 

 

 

 

 

 

 

 

 

trisaccharides.

Molecule , 0711/ of a- ¢’/\y’ of 0711/ of Pmso:
c b—c d-c Pm,=1 8°

B-D-Xyl-(1—>2)[01-D-Man-(1—>6)-]-B-D-Man-OMe -70/120 1 7/180 4:6

B-D-le-U—>2)[01-D-l\l:lan-(1—>3)-]-B-D-M:n-0Me -80/125 eel-140

B-D-:(yI-(1—->2) )[a-D-MZn-U —>3)-][01-D-clI/|an-( -80/1 25 80/1 70 80/-140 4:6

a1—>6)-]—B-D MES-OMe b

 

In the last example, on the basis of the n.O.e. data and theoretical
calculations, it was found out that only one predominant conformer was found to
be selected for each of the compounds shown in Table 1.4.32 Here also, even
though the oligosaccharides are branched, the change in the dihedral angles for

the same type of linkage from compound to compound is very small.

20

Table 1.4. The solution conformations of a family of trisaccharides.

 

 

 

 

 

 

Molecule 41/11: of 11/11; of My of (1)/1|! of
, a-c b-c d-c e-c

01-D-Glc-(1-+4)[01-D-Man-(1—93)-]-B-D-GaI-OMe -68.3/ -65.5/

a b c -40.0 -51.9
B-D-GIc-(1—>4)[0t-D-Man-(1 —>3)-]-B-D-Gal-OMe -68.4/ 57.7/

d b c -55.2 15.31
a—D-Glc—(l—>4)[0t-L-Rha-(1—>3)-]-B-D-GaI-OMe -71 .8/ 51 .7/

a e c -39.7 -10.0
B-D-Glc—(1—>4)[01-L-Rha-(1—>3)-]-[3-D-Gal-OMe 50.7/ 51 .7/

d e c 17.5 -23.4

 

 

 

 

 

 

From these and other examples we have Ieamed that the solution
conformation of disaccharides is reasonably consistent in the disaccharides and
when those same disaccharide units are in the middle of a bigger oligo- or

polysaccharide. This is further discussed in more detail in Chapter Two.

Molecular Mechanics Calculations and Molecular Dynamics

Simulations of Carbohydrates

Molecular mechanics grid search studies of the conformations about the
glycosidic linkages and molecular dynamics simulations will be done to
determine the possible conformational states with a weighting factor which tells
the probability of being at a certain conformational state. Due to the smaller
number of available constraints especially for long-range connectivities in
oligosaccharides, it is necessary to rely upon potential energy surfaces as an aid

to the interpretation of NMR data. In principle, the best estimates of the potential

21

surface for isolated molecules are obtained from ab initio calculations since a
minimal number of approximations are inherent in the method. However, it is
impractical to compute internal energies with full optimization for other than small
organic molecules, due to the huge number of calculations required. In fact, the
computational time increases approximately to the fifth power of the number of
degrees of freedom of the system.23 Semi-empirical calculations are also
quantum mechanical in nature, but in order to reduce the computational
complexity, a large number of integrals in the calculation are replaced by
empirical data.33 But, this is still dependent upon the cube of the number of
degrees of freedom of the system, and is therefore still impractical for
oligosaccharides.23 Molecular mechanics (MM) methods use an entirely classical
treatment. In molecular mechanics, a molecule is viewed as a collection of points
(atoms) connected by springs (bonds) with different elasticities (force constants).
The forces holding the atoms together are described by potential energy
functions of structural features like bond lengths, bond angles, non-bonded
interactions, and so on. In comparison with quantum mechanical methods,
molecular mechanics methods are very much faster, and it is possible to obtain a
minimized geometry for a large macromolecule in a useable time. The precise
form of the force fields used in this calculations varies according to its source, but
takes the general form of the equation below,23 where r refers to bond lengths, 9
to valence angles, 0 to dihedral angles, with equilibrium values denoted by
subscript e. The torsional term contains Fourier terms of the order n and phase 7

to simulate various torsional functions. The fourth term groups the non-bonded

22

(van der Waals) interactions, together with a coulombic term containing the
partial atomic charges q and Q], and the fifth term represents the hydrogen-
bonding potential. The accuracy of the calculations depends upon the choice of
K,, K9, Kn, Aij, Bi], qt, Qj, C11 and Di]. These parameters are derived from a
combination of ab initio calculations on small molecules and from experimental

data.

15,, = 2 mm.)2 + 2 1010-11..)2 + 2 Kn(1+cos(n¢-y)) + 2(A1/R1j‘2-B‘I/Rif) +

bond angle torsion angle van der Waals
2 (qiq/eRij) + X (011/ Rij12'Dij/ 91110) + Z Kn(n'ne)2 (3)
electrostatic hydrogen bond n.O.e.

Now in Chapter Two, we have collected and statistically analyzed crystal
structures of carbohydrates and came out with a simple rule-based method for
predicting the conformations of oligo- and polysaccharides. In Chapter Three,
we describe the design and synthesis of seven molecules with rigid glycosidic
linkages. These molecules are important to parameterize a new Karplus-type
equation that will be useful to calculate the glycosidic dihedral angles of
carbohydrates from vicinal proton-carbon coupling constants and hence help
predict carbohydrate conformations.

In Chapter tour, we optimized the geometries, of the seven molecules
whose synthesis is described in Chapter Three and three other commercially

available molecules, using molecular mechanics and AM1 and PM3 semi-

23

empirical calculations. We then measured the vicinal proton-carbon
heteronuclear coupling constants using a two-dimensional NMR spectroscopy,
excitation-sculptured indirect-detection experiment (EXSIDE). From the least
squares fitting of the coupling constants and the dihedral angles, we came out
with a new Karplus-type equation that will be used to predict the glycosidic
dihedral angles from NMR coupling constants or vice versa.

Finally, in Chapter Five we describe our study of the solution conformation
and dynamics of the disaccharide maltose, the tetrasaccharide reduced
maltotetraose, and the pseudotetrasaccharide acarbose using existing methods
and the new methods that we have developed and described in chapters two to

four.

24

References

1.

2.

10.

11.

12.

13.

14.

Hudson, C. S. J. Am. Chem. Soc. 1909, 31, 66-80.

Lemieux, R.; Koto, S. Tetrahedron, 1974, 30, 1933.

Lemieux, R.; Koto, S.; Voisin, D. The origin and Consequences of the
Anomeric Effect, 1979, p2.

Rademacher, T. W.; Parekh, R. B. & Dweck, R. A. Ann. Rev. Biochem.
1988, 57, 785.

Phillips, M. L.; Nudelman, E.; Gaeta, F. C. A.; Perez, M.; Shingel, A. K.;
Hakamori, S. ; Paulson, J. C. Science 1990, 250, 1130.

Gilardi, R.; Flippen-Andersen, J. L. Acta Crystallogr. 1986 C43, 806.
Jeffrey, G. A.; Huang, D. B. Carbohydr. Res. 1993, 247, 37.

Okuyama, K.; Noguci, K.; Saitoh, M.; Ohno, S.; FuJii, S.; Tsukada, M.;
Takeda, H.; Hidano, T. Bull. Chem. Soc. Jap. 1993, 66, 374.

Gessler, K.; Krauss, N.; Steiner, T.; Betzel, C.; Sandmann, C.; Saenger,
W. Science 1994, 266, 1027.

Raymond, S.; Heyraud, A.; Tran Qui, D.; Kuick, A.; Chanzy, H.
Macromolecules 1995, 28, 2096.

Chandrasekaran, R. Adv. Carbohydr. Chem. Biochem. 1997, 52, 311.
Ernst, R., Bodenhansen, G., & Wokahun, A. Principles of Nuclear
Magnetic Resonance in One and Two Dimension, 1987 pp 431 -440,
Oxford University Press, Oxford.

Bax, A.; Davis, D. G. J. Magn. Reson. 1985, 65, 355.

Bax, A.; Subramanian, S. J. Magn, Reson. 1986, 67, 565.

25

15.

16.

17.
18.
19.

20.

21.

22.

23.

24.

25.
26.
27.

28.

Titman, J.; Neuhaus, D.; Keeler, J. J. Magn. Reson. 1989, 85, 111.
Sangers, J. K.; Hunter, B. K. Modem NMR Spectroscopy, 1993, pp 193-
197, Oxford University Press, Oxford.

Tvaroska, I; Hricovini, M; Petrakova, E. Carbohydr. Res. 1989, 189, 359.
Mulloy, B.; Frenkiel, T. A.; Davies, D. B. Carbohydr. Res. 1988, 184, 39.

Cloran, F.; Carmichael, l.; Serianni, A. S. J. Am. Chem. Soc. 1999, 121,
9843.

Haasnoot, C. A.; De Leeuw, F. A.; Altona, C. Tetrahedron 1980, 36, 2783.
Mayo, S. L.; Olafson, B. D.; Goddard, W. A. J. Phys. Chem. 1990, 94,
8897.

Noggle, J. H.; Schirmer, R. E. The Nuclear Overhauser Effect-Chemical
Applications, Academic Press, London, New York, 1971.

Homans, S. W. Progress in NMR Spectroscopy 1990, 22, 55; Pearlman,
D.A.; Case, D.A.; Caldwell, J.W.; Ross, W.S.; Cheatham, T.C.; Ferguson,
D.M.; Seibel, G.L.; Singh, U.S.; Weiner, P.K.; Kollman, PA. AMBER 4.1,
University of California, San Francisco, 1995.

Kaptein, R.; Zuiderweg, E. R. 13.; Scheek, n. M.; Boelens, R.; van
Gunstem, W. F. J. Mol. Biol. 1985, 182, 179.

Karplus, M.; McCammon Ann. Rev. Biochemistry 1983, 52, 263.
Cumming, D. A.; Carver, J. P. Biochemistry 1987, 26, 6664.

Cumming, D. A.; Carver, J. P. Biochemistry 1987, 26, 6676.

Toma, L.; Ciuffreda, P.; Colombo, D.; Roncheti, F.; Lay, L.; Panza, L.

Helv. Chim. Acta 1994, 77, 668.

26

29.

30.

31.

32.

33.

Toma, L.; Colombo, D.; Roncheti, F.; Panza, L.; Russo, G. Helv. Chim.
Acta1995, 78, 636.

Brisson, J. R.; Carver, J. P. Biochemistry 1983, 22, 3671.

Leeflang, B. R.; Bouwstra, J. B.; Kerékgyarto, J.; Kamerling, J. P.;
Vliegenthant, J. F. B. Carbohydr. Res. 1990, 208, 117.

Lipkind, G. M.; Shashkov, A. S.; Nechaev, O. A.; Torger, V. I.; Shibaev, N.
N.; Kochetkov, N. K. Carbohydr. Res. 1989 195, 27.

Dewar, M. J. S.; Thiel, W. J. Am. Chem. Soc. 1977, 99, 4899.

27

Chapter 2

A Simple and Accurate Rule-based Method for Predicting the

Conformation of Oligo— and Polysaccharides

(Adapted from the paper entitled ‘A Simple and Accurate Rule-based Method for
Predicting the Conformation of Oligo— and Polysaccharides’, Hussen Mohammed

and Rawle I. Hollingsworth, J. Org. Chem. 2001, accepted for publication.)

28

Abstract

In this chapter, we report a simple rule based method for predicting the
conformations of oligo- and polysaccharides. The method is based on an
exhaustive analysis of the entries in the Cambridge Structural Database (CSD) of
crystal structures. We analyzed the values of the critical interglycosidic dihedral
angles, 05'-C1’-OX-CX (phi) and C1'-OX-CX-CX+1 (psi) from all of the entries
which ranged from di-, tri-, tetra- and hexa- to octasaccharides. These results
were used to construct a second database of dihedral angles. The maximum
percentage of the total possible phi-psi space that is populated by a given linkage
was found to be 4.0 °/o and that was for 01-1->2 linkages. An almost exclusive
preference for specific values of phi and psi for a specific type of linkage
independently of the context was clearly evident. These values coincided with
those indicated in a molecular mechanics potential energy search. There was
symmetry of placement of the points on the phi-psi plots that was determined by
the absolute configuration of the anomeric carbon and the carbon to which the
linkage is made. These corresponded to the RR, SS, SR, and RS configurations.
Three simple and accurate rules are proposed to help predict the conformations
of oligo- and polysaccharides where the glycosyl donor is a pyranose residue.

The results support, expand and generalize earlier reports from other groups.

29

Introduction

Carbohydrates are among the most abundant naturally occurring organic
substances on earth. They are found in large storage polymers such as starch, in
cell wall components such as cellulose and hemi-cellulose, in insect
exoskeletons as chitin, as bacterial cell surface antigens, on viruses as the
immunodominant species that govern infection, as the predominant substances
on blood cell surfaces that determine blood type and in a large variety of other
roles and functions. Nucleic acids contain a large proportion of carbohydrates in
the forms of ribose and deoxy-ribose. Despite their importance, methods for
determining the 3-dimensional structure and properties of even mildly complex
carbohydrates have lagged behind those of other biomolecules. There are
several reasons for this. Firstly, complex carbohydrates are generally difficult to
crystallize and there are therefore only a handful of X-ray crystal structures and
these are limited to fiber analyses. The nuclear magnetic resonance (NMR)
spectra of complex carbohydrates are very complicated and characterized by a
high degree of overlap and second-order coupling effects. Because of this, it is
usually not possible to get anywhere near a complete assignment. This often
precludes solution 3-dimensional structure determination by this method.1 The
third reason is that carbohydrates, for example those found in glycoproteins, are
often very heterogeneous in nature. Their biological function is based on this
heterogeneity. Current structural and analytical tools, especially those based on
diffraction methods, are extremely limited in utility because they require

homogeneity and uniformity of structure.

30

 

Figure 2.1. Structure of 1-O-methyl cellobioside and the glycosidic dihedral
angles, phi and psi, which determine its conformation. Phi = 05'-C1'-O4-C4 and

Psi = C1 '-O4-C4-C5.

There is, however, a fairly significant body of information on the 3-
dimensional structure of very small carbohydrate molecules such as
monosaccharides and di- and trisaccharides. It might be possible to make
projections from these structures to more complex ones. There are three
important questions to be asked. Firstly, how dependent is the conformation
about the glycosidic linkage on the nature of the residues attached at that
linkage? Secondly, how do substitutions in the glycosyl residue affect the
conformation of the glycosidic linkage of that same residue? Thirdly, how
sensitive is the conformation of a particular type of linkage (e.g. 01-1->4) between
two residues to the context in which that linkage appears? Is this conformation

the same in different molecules? It favorable and reliable answers to these

31

questions could be obtained, a way of predicting the conformation of complex
carbohydrate molecules could be developed.

There is more than just an indication that the conformation of the bonds
involved in a specific type of inter-glycosidic linkage between two carbohydrate
rings might be independent of the context in which the linkage appears. This is
. implied in the Hudson rules of iso-rotation.2 These rules allow the estimation of
the net optical rotation of carbohydrate molecules containing more than one
residue by summing fractional contributions from each residue. The anomeric
carbon makes the largest contribution because the ring oxygen, the anomeric
carbon and the anomeric oxygen atom form the system with the most intense
U.V. chromophore in typical carbohydrates without rt-systems. Because the
intensity of this chromophore (and therefore the optical rotatory dispersion) is a
function of the geometry of the interglycosidic bond, it is reasonable to conclude
that significant changes in geometry of a given type of linkage do not occur in
different molecular contexts. In an earlier study with a limited number of
glycosides, it was observed that the dihedral angles of a particular type of
linkages generally fall between the limits of 130°, irrespective of the
monosaccharides involved.3 The idea that the geometry about interglycosidic
linkages is relatively independent of the molecular context in which they occur is
exploited in the molecular builder computer program POLYS by Engelsen et al.
This program produces three-dimensional structures of carbohydrates from
preoptimized monosaccharide structures and from computer analyses describing

the linkage properties of disaccharide fragments.4

32

Results and Discussion

A case for supporting the conjecture that linkages should be relatively
insensitive to context can also be made by analyzing the potential energy surface
obtained by systematically varying the dihedral (11>) about the anomeric linkage
relative to the one about the linkage that connects the glycosyl residue to the
aglycon or second carbohydrate residue (1v) (Figure 2.1) and evaluating the
energy at each step using molecular mechanics methods. Figure 2.2 shows 0,111
maps of (a) cellobiose (B-D-Glc-(1->4)-D-Glc) and (b) melliobiose (01-D-Gal-(1-
>6)-D-Glc). Dih-1 stands for 0, which is defined as H1'-C1'-OX-CX and Dih-2
stands for 1y, which is defined as C1'-OX-CX-HX. For cellobiose X is 4 and for
melliobiose it is 6. The contour lines are drawn with a gap of 1 kcal/mol for each
line. Such potential energy surfaces in hexopyranosides are characterized by a
single deep potential energy well (Flgure 2.2a) except for 1-6 linkages where two
relatively broad but connected minima are obtained (Flgure 2.2b). The depth of
the potential energy well is typically 6 kcal/mol or higher. This is much higher
than kT (the characteristic energy) at room temperature. Because of this,
residues should be essentially locked in this conformation unless the penalty can
be offset by some strong interaction such as an electrostatic one if charged
groups are present.

A definitive proof of the conjecture was sought by examining the entire
Cambridge database of crystal structures and analyzing the phi and psi values
for all of the carbohydrate entries. These include simple glycosides,“

disaccharides,M2 trisaccharides,”97 tetrasaccharidesf’8'100 hexasaccharides,101

33

and octasaccharides‘oz. The database contained a total of 164 phi and 153 psi
angles. No database entries were ignored. The results of the search were
completely supportive of our conjecture. An almost exclusive preference for a
specific value of phi and psi for a specific type of linkage independently of context
was clearly evident. These values coincided with those indicated in the molecular
mechanics potential energy surface search. Table 2.1 shows the total area of the
Mr space populated by the different linkages of oligosaccharides where both the
donor and acceptor residues have the D configuration. Only a very small area of
this space is populated. Of the available 360 by 360 degrees of the 0,111 space,
the maximum area populated by a given linkage of oligosaccharides is 4.0% and
that is for the 01-1->2 linkage (Table 2.1). The percentages in Table 2.1 were
obtained by using the relation (0 max - ¢ min)(\v max - 111 min) / (360 x 360). The
actual area populated by the oligosaccharides within the rectangle is much
smaller. This indicates that the actual percentage of luv-space that is populated
is very limited. The value corresponding to the maximum probability that the
angle of a specific linkage lay within a given percentage of the total
configurational space of 360 degrees and the mean value of the dihedral angle is
given for some of the oligosaccharides in Table 2.2. Also included in Table 2.2
are the standard deviation and the percentage of (I) or w space populated by a
given linkage. From the linkages where both the donor and acceptor residues
are pyranose rings, the highest percentage of 0 space populated is 14 % and that
is for the 01-Glc-1->4-Glc linkage. The corresponding value for the \v space is

obtained for B-Glc-1->3-Glc linkage and it is 16 %.

Table 2.1. The percentages of the total phi-psi space that is populated by
oligosaccharide crystals from Cambridge Structural Database.

 

 

 

 

 

Linkage % Area Linkage % Area
[3 -1->4 1.66 01—1->4 3.35
B—l—>2 1.87 01 -1—>2 3.98
B—1->6 2.55 01—1->6 2.68
B—l—>3 2.99 01 —1->3 1.75

 

 

 

 

 

 

Because of geometric arguments, there is a symmetry of placement of the
points on the potential energy surface that is determined by the absolute
configuration of the anomeric carbon and the carbon to which the linkage is
made if that carbon is a carbohydrate ring carbon (Figure 2.3). These
correspond to the configurations RR, 88, SR and RS. The first character in each
pair corresponds to the configuration of the anomeric carbon. In the case of
alpha D-glucosides and the entire alpha D-aldohexopyranosides then this first
character is R. If the residue is attached to a primary carbon in the aglycon (e.g.
the 6-position of a hexose) the second character is not relevant. For the
corresponding beta glycosides the configuration at Ci is S. The configurations
for the acceptor carbons of a D-glucose are R, S, and R for the acceptor carbons
2, 3, and 4, respectively. The reverse holds for the L-sugars. As noted earlier,
there are two minima on the potential energy surface If the linkage is to the

primary carbon at the 6-position of an aldohexopyranoside.

35

Figure 2.2 Biograf rigid residue 0,111 maps of (a) cellobiose (B—D-Glc-(1->4)BD-
Glc) and (b) melliobiose (a—D-GaI-(1->6)-B-D-Glc). Din-1 stands for 0, which is
H1'-C1'-OX-CX where X is 4 for cellobiose and 6 for melliobiose and Dih-2
stands for 111, which is C1'-O4-C4-H4 for cellobiose and C1'-06-C6-C5 for
melliobiose. The contour lines are drawn with a gap of 1 kcanol for each

line. This figure is a color presentation for clarity.

36

 

 

 

 

 

0.1”?

360.0

DIHI

37

Structures corresponding to the occupation of one or the other of these
minima were observed. This study is limited to the anomeric linkage of
pyranosides and the results of the statistical analysis of the built database of
carbohydrate dihedral angles from CSD can be summarized in the following

rules:

(1) For a disaccharide linkage between D-pyranose sugars in which the bond
from the bridging oxygen to both residues is equatorial (e.g. a residue that is B-
linked to an equatorial site on the other residue) the average value :1: standard
deviation of 0 and iv (and a) for 1-6 linkages) are given below in degrees as 0,111
pairs:

(a) -88 :l: 11, -128 :l: 15 for 1-4 linkages,

(b) -79 :1: 10, 109 :1: 34 for 1-3 linkages,

(c) -75 a: 11, -129 :1: 23 for 1-2 linkages, and

(d) For 1-6 linkages o is -71 :1: 10 with (111,03) of (-170 :1: 3, 83 :l: 7) when the
acceptor is galactose and (164 :1: 4, 71 1 7) with a 30% probability and (-166 :1:

10, -63 :1: 1) with a 70% probability when the acceptor is glucose.

(2) When the oxygen is axial on the donor carbon the absolute value of 41 is
similar but the sign is reversed. The absolute value of w is similar but its sign is
reversed if the bridging oxygen is axial to the acceptor carbon (e.g. the 2-position
of mannose or the 4-position of galactose). The sign of o is reversed if the donor

residue is of the L configuration and the sign of \v is reversed when the acceptor

38

residue is of the L configuration. Thus for a D residue that is 01-linked to an
equatorial site of another D residue, the average value :1: standard deviation of (I)
and w (and (0101' 1-6 linkages) are given below in degrees as 111,111 pairs.

(a) 98 :l: 19, -128 :I: 22 for 1-4 linkages

(b) 87 :1: 14, 121 124 for 1-3 linkages,

(c) 99 a: 11, -146 :1: 28 for 1-2 linkages, and

(d) For 1-6 linkages <l> is 72 :t 8 with the same (111,10) values as in Rule 1(d).

(3) Rules 1 and 2 also apply for acyl glycosides or for glycosides bearing simple

substituents at other sites in the ring.

It is noteworthy that there is only a very small difference in the average
values of 1’ and 11; between the 1->4 and 1->2 linkages as shown in Rules 1 (a)
and (c) and 2 (a) and (c). The reason for this is that both 1->4 and 1->2 linkages
belong to the same region of 0,111 space, which is the SR region for the [3-1->4
and B-1->2 linkages and the RR region for the 01-1->4 and 01-1->2 linkages
(Figure 2.3). In the case of B-1->3 linkages (Rule 1(b)), the average value of 0
(magnitude and sign) is practically identical to the corresponding values of the B-
1->4 and [3-1->2 linkages (Rule 1(a) and (c)) but the average value of \II is of
similar magnitude and opposite in sign to the corresponding values of the [3-1->4
and B-1->2 linkages. Here again the reason is that while [3-1->3 linkages belong
to the 88 region of the 0,111 space, the B-1->4 and [3-1->2 linkages belong to the

SR region (Figure 2.3). Similarly, the 01-1 ->3 linkage belongs to the RS region

39

 

 

 

 

 

 

 

 

 

. 000%

2 e 2 2 2 E 1. .1.
530300 «r

.m : 2 cm 3 mm 0.898%

9;. 0:- . 5- _: E- 02- :82
02000.—

2 v 2 o 2 mm CO .02
000mm

0 a 2 w a E e .x.
soamm>0fl 9

S E m _ 8 a om Eng—Sm

2: SW. w? ow- mm- 00 .802
09000.—

3 v 2 o 2 mm mo .02

Ea 82 20 20 20 20

$40068 -AX-_0-§2-0 0.40-100 $401200 #401060 $40.208 800:5

 

 

 

 

 

 

 

 

 

 

0009800 Efiosbm

0315800 05 80h Detacogmowzo 05 no 0500 .«0 3 0:0 3 00—98 36050 05 m0 max—0:0 3030330 05 no 3300M .N.N 030,—.

40

 

 

 

 

 

 

 

1

180 1 l

1 SS ‘ #15111“ _ 1
120 ~— —:—~ ~1~ 139.141-- ~ 1 I
i A 1* 0’ I I
1 I; (O A ‘ I
1 e 60 + 1
' x i
1 g 0 , I
l 0 :
1 1'3 ,
13W 1
1 . ~ ° ' '
1 -12. -0..- .7: ,1
1 1
I SR 3 Q‘HI I
1 -180 1
, -180 -120 -60 0 60 120 180 I
1 Phi (05'01 'oxcx1 L
1. l
1. 1

Figure 2.3 4),? map of 1-2, 1-3, 1-4 and 1-6 linked oligosaccharides from the
Cambridge Structural Database (CSD). Each point on the map corresponds to
one glycosidic linkage from the CSD. Phi = 05'-C1'-OX-CX and Psi = C1'-OX-
CX-CX+1, where X = 2, 3, 4, and 6 for 1-2, 1-3, 1-4, and 1-6 linkages,

respectively.

41

while the 01-1->4 and 01-1->2 linkages residue in the RR region. Comparing rules
1(a-d) with the corresponding rules 2(a-d), we can see that while the average
values of w have the same sign and similar magnitudes, the average values of 0
have similar magnitude but opposite sign. This is because the anomeric carbon
of the donor residues in the two rules have opposite configurations, S in rules
1(a-d) and R in rules 2(a-d), while the configuration of the acceptor carbons stay
the same in the two rules.

It should be mentioned that care should be taken when using the average
values of 41 and w in Rules 1(0) and 2(d) for 1->6 linkages because of the
relatively greater flexibility of 1->6 linkages as compared to 1->4, 1->3 and 1->2
linkages. The higher flexibility of 1->6 linkages arise from the lower barrier of a
methylene group to rotation. We have observed that when a glucose residue is
attached at the anomeric carbon (C1) of the acceptor glucose of 1->6 linkages,
the signs of both 111 and a) are positive but when C1 has a free hydroxyl or a
fructose residue attached to it, the signs of both 111 and 0) are negative (Rule 1(d)
and 2(0)). It seems that the sign of 0) dictates the sign of w for 1->6 linkages
where the acceptor is glucose. A larger database of 1->6 linkages is needed to
confirm this. The (111,00) average values in Rules 1(d) and 2(d) are the same
because of the achiral nature of C6. Even though there are no 1->6 linkages with
the acceptor residue having the L configuration in the Cambridge Structural
Database, we expect that the sign of co and hence the sign of iv might be

opposite to the case when the acceptor residue has the D configuration. In this

42

work, we observed that the average values of 0 shift from 71° for 1->6 linked
oligosaccharides to 80° for B-1->X (X = 2, 3, 4) to 94° degrees for 01-1->X (X = 2,
3, 4) linked oligosaccharides as the steric demand increases.

To test the prediction power of the rules proposed in this paper we plotted
the structure of two polysaccharides, dermatan 4-sulfate (Figure 2.5) and
hyaluronan (sodium hyaluronate lll) (Figure 2.6), using the dihedral angles from
X-ray fiber diffraction studies104 with the BIOGRAF molecular mechanics
programl°° and compared them with those predicted using the rules. Dermatan
4-sulfate is a polysaccharide with the repeating unit [->3)-B-D-GalNAc4SOa'-(1-
>4)-B-L-ldoA-(1->] and interacts with collagen, elastin, and some glycoproteins to
maintain structural integrity of the tissues involved. Hyaluronan is an unsulfated
glucosaminoglycan found in mammalian connective tissues, where it forms the
central core of the proteoglycan aggregate.104 It is a polymer composed of
alternating 1->4 and 1->3 linkages [->3)-B-D-GlcNAc-(1->4)-B-D-GIcA-(1->]. As
shown in Figure 2.5 and Figure 2.6, the predicted structures are comparable to
the structures from X-ray fiber diffraction studies. These rules will be useful for
predicting the three dimensional structures of pure carbohydrates and the
carbohydrate parts of glycoproteins and glycolipids. These structures ' will be
useful in understanding molecular recognition in biological processes for

example.

43

Figure 2.5 Line representation of the structure of dermatan 4-sulfate predicted
using the rules in this paper (bottom) and from X-ray fiber diffraction study
(top).104 Dermatan 4-sulfate is a polysaccharide with the repeating unit [->3)-
B-D-GalNAc4SOa'-(1->4)-B-L-IdoA-(1->] and interacts with collagen, elastin,
and some glycoproteins to maintain structural integrity of the tissues involved.

This figure is a color presentation for clarity.

44

 

Figure 2.6 Line representation of the structure of hyaluronan (sodium
hyaluronate Ill) predicted using the rules in this paper (bottom) and from X-ray
fiber diffraction study (top).104 Hyaluronan is an unsulfated glucosaminoglycan
found in mammalian connective tissues, where it forms the central core of the
proteoglycan aggregate. It is a polymer composed of alternating 1->4 and 1->3
linkages [->3)-B-D-GlcNAc-(1->4)-B-D-GIcA-(1->]. This figure is a color

presentation for clarity.

46

 

47

Conclusions

We describe herein a reliable method for predicting oligo- and polysaccharide
conformations. The consistency of prediction especially when mixed linkages are
present is high. This indicates that further substitution adjacent to the site of the
linkage does not have a dominant impact at this level of refinement. The exo-
anomeric effect1 is a phenomenon that specially relates to the conformation
about glycosidic linkages and deserves some special comment. It relates to the
preference for a gauche conformation about the O-R bond of the aglycones of
sugars. The effect should lead to values of +60 and -65 degrees for the ID angles
of 01- and B-glycosides, respectively.1a Values of (I) that range from 61 - 74° for
01- and from 68 - 87 for B-methyl glycosides with corresponding ranges of 75 -
121° and 71 — 105°, respectively for 01- and B-linked oligosaccharides were
reported in an earlier work.‘°° These and the values reported here are within a
range consistent with an exo-anomeric effect. It is still not clear, however, to what
extent this effect plays a role especially since a term defining it was not implicitly
assigned in the force fields but results that agree with the databases were
obtained anyway. There is yet another level of refinement to the predictions that
one can make. This would include an assessment of steric, electronic and
volume effects of substituents proximal to the linkage sites. The biasing due to

the anomeric effect will doubtless play a key role in this next level of prediction.
Acknowledgements

This work was supported by the Michigan State University Research

Excellence Fund.

48

References

1.

For some of the early works on the prediction of the conformations of
oligosaccharides and modeling of their solution conformations by a
combination of NMR and calculation, see (a) Lemieux, R. U.; Koto, S.
Tetrahedron 1974, 30, 1933. (b) Lemieux, R. U.; Koto, 8.; Voisin, D. Origin
and Consequences of the anomeric effect, ACS Symposium series, No.
87, Chapter 21979, p17. (c) Lemieux, R. U.; Bock, K. Jap. J. Antibiot.
1979, 32, 163. (d) Lemieux, R. U.; Bock, K.; Delbaere, L. T. J.; Koto, S.;
Rao, V. S. Can. J. chem. 1980, 58, 631. (e) Thogersen, H.; Lemieux, R.
U.; Bock, K.; Meyer, B. Can. J. chem. 1982, 60, 44. (f) Bock, K.; Lemieux,
R. U. Carbohydr. Res. 1982, 100, 63. (g) Lemieux, R. U.; Bock, K. Arch.
Biochem. Biophys. 1983, 221, 125. (h) Sabesan, S.; Bock, K.; Lemieux, R.
U. Can. J. Chem. 1984, 62, 1034. (i) Vandonselaar, M.; Delbaere, L. T. J.;
Spohr, U.; Lemieux, R. U. J. Biol. Chem. 1987, 262, 10848. (j) Delbaere,
L. T. J.; Vandonselaar, M.; Prasad, L.; Quail, J. W.; Nikrad, P. V.;
Pearlstone, J. R.; Carpenter, L. 8.; smillie, L. B.; Spohr, U.; Lemieux, R. U.
Trans. Amer. Crystal/ogr. Assoc. 1989, 25, 65. (k) Otter, A.; Lemieux, R.
U.; Ball, R. G.; Venot, A. P.; Hidsgaul, 0.; Bundle, D. R. Eur. J. Biochem.
1999, 259, 295. (I) Bock, K.; Amap, J.; Loenngren, J. Eur. J. Biochem.
1982, 129, 171. (m) Bock, K. Pure & Appl. Chem. 1983, 55, 605. (n)
Cumming, D. A.; Carver, J. P. Biochemistry 1987, 26, 6664. (0) Meyer, B.
Topics in current chemistry: Carbohydrate chemistry, ed. Thiem, J., Berlin,

1990, 154, 141.

49

10.

11.

12.

13.

14.

Hudson, C. S. J. Amer. Chem. Soc. 1930, 52, 1680; Hudson, C. S. J.
Amer. Chem. Soc. 1930, 52, 1707.

Jeffrey, G. A. Acta Cryst. Section B1990, 46, 89.

Engelsen, S. 8.; Cross, 8.; Mackie, W.; Perez, S. Biopolymers, 1996, 39,
417.

Jeffrey, G. A.; Takagi, 8. Acta Crystallogr., Section B 1977, 33, 738.
Gatehouse, B. M.; Poppleton, B. J. Acta Crystallogr., Section B 1970, 26,
1761.

Jeffrey, G. A.; McMullen, R. K.; Takagi, S. Acta Crystallogr., Section B
1977, 33, 728.

Shalaby, M. A.; Fronczek, F. R.; Younathan, E. S. Carbohydr. Res. 1994,
258, 267. '
Taga, T.; Sumiya, S.; Osaki, K.; Utamura, T.; Koizumi, K. Acta
Crystallogr., Section B 1981, 37, 963.

Mo, H.; Jensen, L. H. Acta Crystallogr., Section B 1978, 34, 1562.
Longchambon, F.; Ohanessian, J.; Gillier—Pandraud, H.; Duchet, D.;
Jacquinet, J. C.; Sinay, P. Acta Crystallogr., Section B 1981, 37, 601.
Foces-Foces, C.; Cano, F. H.; Garcia-Blanco, S. Acta Crystallogr., Section
B 1980, 36,377.

Ollis, J.; James, V. J.; Angual, S. J.; Pojer, P. M. Carbohydr. Res. 1978,
60, 219.

Smith, J. L.; Sundaralingam, M. Acta Crystallogr., Section B 1981, 37,
1095.

50

15.

16.

17.

18.

19.

20.

21 .
22.
23.
24.

25.

26.

27.

28.
29.

Rihs, G.; Traxler, P. Helv. Chim. Acta1981, 64, 1533.

Neuman, A.; Becquart, J.; Avenel, D.; Gillier-Pandraud, H.; Sinay, P.
Carbohydr. Res. 1985, 139,23.

Hirotsu, K.; Simada, A. Bull. Chem. Soc. Jpn. 1974, 47, 1872.

Lugar, P.; Vangehr, K.; Bock, K.; Pauleen, H. Carbohydr. Res. 1983, 117,
23.

Nassimbeni, L. R.; Nlrer, M. L.; Cragg, G. M.; Pettit, G. R. Acta
Crystal/ogr. Section C (Cr. Str. Comm.) 1985, 41, 728.

Jacobson, R. A.; Wunderlich, J. A.; Lipscomb, W. N. Acta Crystallogr.
1961, 14,598.

Brown, C. J. J. Chem. Soc 1966, A 927.

Chu, S. S. C.; Jeffrey, G. A. Acta Crystallogr., Section B 1968, 24, 830.
Senma, M.; Taga, T.; Osak, K. Chem. Lett. 1974, 1415.

Perez, S.; Vergelati, C. Acta Crystallogr., Section B 1984, 40, 294.
Svensson, G.; Albertsson, J.; Svensson, C.; Mangusson, G.; Dahmen, J.
Carbohydr. Res. 1986, 146, 29.

Perez, 8.; Vergelati, C.; Tran, V. H. Acta Crystallogr., Section B 1985, 41,
262.

Jeffrey, G. A.; Nanni, R. Carbohydr. Res. 1985, 137,21.

Wong, R. Y.; Horowitz, R. M. J. Chem. Soc., Perkin Trans. 1, 1986, 843.
Brown, J. M.; Cook, S. J.; Jones, R. H.; Khan, R. Tetrahedron 1986, 42,

5089.

51

30.

31.

32.

33.

35.

36.
37.

38.

39.

40.

41.

Srikrishnan, T.; Chowdhary, M. S.; Matta, K.L. Carbohydr. Res. 1989, 186,
167.

Nikitin, A. V.; Andranov, V. I.; Myasnikova, R. M.; Firgang, S. I.; Usov, A.
l.; Sopin, V. F.; Pertsin, A. l. Kristal/ografiya1986, 31, 676.

Lamba, D.; Burden, C.; Mackie, W.; Sheldrick, B. Carbohydr. Res. 1986,
153,205.

Lamba, D.; Burden, C.; Mackie, W.; Sheldrick, B. Carbohydr. Res. 1986,
155, 11.

Ezure, Y.; Yoshikuni, Y.; Ojima, N.; Sugiyama, M.; Hirotsu, K.; Higuchi, T.
Acta Crystallogr. Section C (Cr. Str. Comm.) 1987, 43, 1809.

Nikitin, A. V.; Shibanov, T. A.; Firgang, S. l.; Usov, A. .l; Sopin, V. F.;
Myasnikova, R. M. Kristal/ografiya1987, 32, 896.

Takagi, 8.; Jeffrey, G. A. Acta Crystallogr., Section B 1977, 33,2377.
Arene, F.; Neuman, A.; Longchambon, F. C. R. Acad. Sci. 0 Chim. 1979,
288, 331.

Rohrer, D. C.; Sarko, A.; Bluhm, T. L.; Lee, Y. C. Acta Crystallogr.,
Section B 1980, 36, 650.

Ilin, S. G.; Lindeman, S. V.; Struchkov, Yo. T; Levina, E. V.; Storik, V. A.;
Yelyakov, G. B. Dokl. Akad. SSSR 1992, 323, 290.

Taga, T.; lnagaki, E.; Fujmori, Y.; Fujita, K.; Hara, K. Carbohydr. Res.
1993, 241, 63.

Duee, E.; Grand, A.; Tran, V. H. Acta Crystallogr., Section B 1981, 37,

850.

52

42.

43.

44.

45.

46.

47.

48.

49.

50.

51.

52.

53.

54.

55.

Kanters, J. A.; Schouten, A.; van Bommel, M. Acta Crystallogr., Section C
(Cr. Str. Comm.) 1990, 46, 2408.

Yamada, H.; Nishizawa, M.; Katayama, C. Tetrahedron Lett. 1992, 33,
4009.

Kanters, J. A.; Schrrenberg, R. L.; Leeflang, B. R.; Kroon, J.; Mathlouthi,
M. Carbohydr. Res. 1988, 180, 175.

Jin, G.; Yamagata, Y.; Tomita, K. Chem. Phami. Bull. 1990, 38, 297.
Bugg, C. E. J. Amer. Chem. Soc. 1973, 95, 908.

Cook, W. J.; Bugg, C. E. Acta Crystallogr., Section B 1973, 29, 907.
Beerers, C. A.; Hansen, H. N. Acta Crystallogr., Section B 1971, 27,
1323.

Noordik, J. H.; Beurskens, P. T.; Bennema, P.; Visser, R. A.; Gould, R. O.
2. Kristal/ogr. 1984, 168, 59.

Fries, D. C.; Rao, S. T.; Sundaralingam, M. Acta Crystallogr., Section B
1971, 27, 994.

Ohanessian, J.; Longchambon, F.; Arene, F. Acta Crystallogr., Section B
1978, 34, 3666.

Oliver, J. D.; Strickland, L. C. Acta Crystallogr., Section C (Cr. Str. Comm.)
1984, 40, 820.

Hanson, J. C.; Sieker, L. C.; Jensen, L. H. Acta Crystallogr., Section B
1973, 29, 797.

Brown, G. M.; Levy, H. A. Acta Crystallogr., Section B 1973, 29, 790.
Hynes, R. C.; Le Page, Y. J. Appl. Crystallogr. 1991, 24, 352.

53

56.

57.

58.
59.

60.
61.

62.
63.

64.

65.

66.
67.

68.
69.

70.

Thiem, J.; Kleeberg, M.; Klaska, K. H. Carbohydr. Res. 1989, 189, 65.
Quingley, G. J.; Sarko, A.; Merchessault, R. H. J. Amer. Chem. Soc. 1970,
92, 5834.

Gress, M. E.; Jeffrey, G. A. Acta Crystallogr., Section B 1977, 33, 2490.
Takusagawa, F.; Jacobson, R. A. Acta Crystallogr., Section B 1978, 34,
213.

Ham, J. T.; Williams, D. G. Acta Crystallogr., Section B 1970, 26, 1373.
Cannon, J. R.; Raston, C. L.; Toia, R. F.; White, A. H. Aust. J. Chem.
1980, 33 2229.

Chu, S. S. C.; Jeffrey, G. A. Acta Crystallogr. 1967, 23, 1038.
Foces-Foces, C.; Cano, F. l-l.; Garcia-Blanca, 8. Acta Crystallogr., Section
B 1981, 37, 1270.

Tanase, t.; Nouchi, r.; Oka, y.; Kato, M.; Nakamura, N.; Yamamura, T.;
Yamamoto, Y.; Yano, S. J. Chem. Soc., Dalton Trans. 1993,2645.

Van der Sluis, P.; Kroon, J. Acta Crystallogr., Section C (Cr. Str. Comm.)
1990, 46,2171.

Kivikoski, J.; Pitkanen, l.; Nurmi, J. Carbohydr. Res. 1992, 232, 189.

Jin, G; Fujii, S.; Tomita, K. Acta Crystallogr., Section C (Cr. Str. Comm.)
1990, 46, 1866.

Kanters, J. A.; Shouter, A.; van Bommel, M. J. Mol. Struct. 1990, 238, 403.
Kivikoski, J.; Valkomen, J.; Nurmi, J. Carbohydr. Res. 1992, 232, 53.
Lamba, D.; Segre, A. L.; Glover, S.; Mackie, W.; Sheldrick, B.; Perez, S

Carbohydr. Res. 1990, 208, 215.

71.

72.

73.

74.

75.

76.

77.

78.

79.

80.

81.

82.

Lee, C. K.; Koh, L.L. Acta Crystallogr., Section C (Cr. Str. Comm.) 1993,
49, 621.

Noguchi, K.; Okuyama, K.; Kitamura, S.; Takeo, K. Carbohydr. Res.
1992, 237, 33.

Koll, P.; Petrusova, M.; Petrus, I.; Zimmer, B.; Mort, M.; Kopf, J.
Carbohydr. Res. 1993, 248, 37.

Lindon, A.; Lee, C. K. Acta Crystallogr., Section C (Cr. Str. Comm.) 1995,
51, 1007.

Lee, C. K.; Linden, A. Carbohydr. Res. 1994, 264, 319.

Linden, A.; Lee, C. K. Acta Crystallogr., Section C (Cr. Str. Comm.)
1995,51,1012.

Prawit, H.; Mahidol, C.; Ruchirawat, S.; Prawat, U.; Tuntiwachwuttikul, P.;
Tooptakong, U.; Tayler, W. C.; Pakawachai, C.; Skelton, B. W.; White, A.
H. Phytochemistry1995, 40, 1167.

Kanie, O.; Takeda, T.; Hatano, K. Carbohydr. Res. 1995, 276, 409.
Wieczorek, M. W.; Blaszczyk, J.; Krol, B. W. Acta Crystallogr., Section C
(Cr. Str. Comm.) 1996, 52, 1193.

Brisse, F.; Merchessault, R. H.; Perez, 8.; Zugenmaier, P. J. Amer. Chem.
Soc. 1982, 104, 7470.

Kivikoski, J.; Pitkaren, l.; Valkomen, J.; Heikkila, H. Carbohydr. Res. 1992,
223, 45.

Connor, 8. 0.; Lam, L. K. T.; Jones, N. D.; Chasey, M. O. J. Org. Chem.

1976, 41, 2087.

55

83.

84.

85.
86.

87.

88.
89.

90.
91.

92.

93.

94.

95.

96.
97.

98.

Perez, 8.; Brisse, F. Acta Crystallogr., Section B 1977, 33, 2578.
Hoogendorp, J. 0.; de Kok, A. J.; Romers, C. Carbohydr. Res. 1983, 114,
169.

Moran, R. A.; Richards, G. F. Acta Crystallogr., Section B 1973, 29, 2770.
Taga, T.; lnagaki, E.; Fujmori, Y.; Nakamura, S. Carbohydr. Res. 1993,
240, 39.

Taga, T.; lnagaki, E.; Fujmori, Y.; Nakamura, S. Carbohydr. Res. 1994,
251, 203.

Jeffrey, G. A.; Huang, D. Carbohydr. Res. 1991, 222, 47.

Bequart, J.; Neuman, A.; Gillier-Pandraud, H. Carbohydr. Res. 1982, 111,
9.

Jeffrey, G. A.; Park, Y. J. Acta Crystallogr., Section B 1972, 28, 257.
Rohrer, D. C. Acta Crystallogr., Section B 1972, 28, 425.

Ferretti, V.; Bertolasi, V.; Gilli, G.; Accorsi, C. A. Acta Crystallogr., Section
C (Cr. Str. Comm.) 1984, 40, 531.

Berrnan, H. M. Acta Crystallogr., Section B 1970, 26,290.

Jeffrey, G. A.; Huang, D. Carbohydr. Res. 1990, 206, 173.

Avenel, D.; Neuman, A.; Gillier-Padraud, H. Acta Crystallogr., Section B
1976, 32, 2598.

Imberty, A.; Perez, 8. Carbohydr. Res. 1988, 181, 41.

Carter, G. T.; Phillipson, D. W.; West, R. R.; Borders, D. B. J. Org. Chem.
1993, 58, 6588.

Jeffrey, G. A.; Huang, D. Carbohydr. Res. 1991, 210, 89.

56

99.

100.

101.

102.

103.
104.

105.

Nardi, R. G.; Flippen-Anderson, J. L. Acta Crystallogr., Section C (Cr. Str.
Comm.) 1987, 43,806.

Noguchi, K.; Kobayashi, E.; Oguyama, K.; Kitamura, 8.; Takeo, K.; Ohno,
8. Carbohydr. Res. 1994, 258, 35.

Hinrichs, W.; Saenger, W. J. Amer. Chem. Soc. 1990, 112, 2789.

Ashton, P. R.; Brown, C. L.; Menzer, 8.; Nepogodiev, 8. A.; Stoddart, J.
F.; Williams, D. J. Chem. Eur. J. 1996, 2, 580.

Perez, 8.; Marchessault, R.H. Carbohydr. Res. 1978, 65, 114.
Chandrasekaran, R. Adv. Carbohydr. Chem. Biochem. 1997, 52, 311.

Molecular Simulations, Inc. Waltham, MA 02154 U. S. A.

57

Chapter 3

Design and Synthesis of Compounds with
Rigid Glycosidic Linkages

Abstract

One of the important methods for determining the three-dimensional
structures of carbohydrates is using the proton-carbon coupling constants to
calculate the critical glycosidic dihedral angle values. In this work we report the
synthesis of seven compounds that have rigid glycosidic linkages. These are:
1 ,2-O-[2-methyl-(8)-1 ,2-ethanediyl]-01-L-fucopyranoside (1 ), 4,6-O-benzyledene-
1,2-O-[2-methyI-(R)-1,2-ethanediyl]-01 -D-glucopyranoside (2), 1,2-O-[2-methyl-
(R)-1 ,2-ethanediyl]-B-L-fucopyranoside (3), 3-O-methyl-1,2-O-[2-methyl-(S)-1,2-
ethanediylj-B-D-glucopyranoside (4), 4,6-O—benzyledene-1,2-O-[2-methyl-(8)-1,2-
ethanediylj-B-D-glucopyranoslde (5), 4,6-O-benzyledene-1,2-O-[2-methyl-(8)-1,2-
ethanediyl]-[SD-galactopyranoside (6), and 3-O-benzyl-1,2-O-[2-methyI-(R)-1,2-
ethanediyl]-B-D-mannopyranoside (7). These compounds are important to
parameterize a new Karplus-type equation relating the glycosidic dihedral

angles to the vicinal proton-carbon coupling constant of carbohydrates.

59

Introduction

Previous equations by Tvaroska and coworkers‘ and Mulloy and
coworkers2 for determining glycosidic dihedral angles of carbohydrates from
vicinal proton-carbon coupling constants were parameterized by using vicinal
proton-carbon heteronuclear coupling constants measured in solution and
dihedral angles taken from crystal structures. While the solution structure of
carbohydrates is an average of several conformations, the crystal structure
represents one of the solution structures where crystal-packing forces play a role.
In addition, while our study (Chapter Two) revealed that glycosidic dihedral
angles defined by the four atoms H-C-O-C are almost always between 0 and 90
degrees, the equations by Tvaroska and coworkers‘ and Mulloy and coworkers2
are parameterized with a very small number of data points in the range between
0 and 90 degrees.

It would then be very valuable to design and synthesize compounds with
rigid glycosidic linkages with predetermined dihedral angles and measure the
interglycosidic heteronuclear coupling constants in solution. By relating the
values of the dihedral angles to the three-bond heteronuclear coupling constant,

°JCH, we can then come up with a new and more accurate Karplus type equation

for carbohydrates.

Design and Synthesis

We designed and synthesized seven compounds with rigid glycosidic

linkages for the purpose of parameterizing a new Karplus-type equation relating

60

carbohydrate glycosidic dihedral angles and vicinal proton-carbon coupling
constants. The method we chose to rigidize the glycosidic linkage is forming a
six-membered ring around and including the 02-C2-C1-Ol fragment of
monosaccharides (Figure 3.1). We prepared the seven compounds starting
from different monosaccharides, which commonly occur in nature. These
compounds have different electronic environments around the anomeric carbon
(C1) because of the different substitutions and/or configurations at the different
sites of the monosaccharides. This is important because, the electronic
environment around the glycosidic linkages may affect the heteronuclear spin —
spin coupling constant, which is transmitted through bonding electrons.

The synthesis is designed in such away that the 2-hydroxy group of the
monosaccharides acts as a nucleophile for asymmetric oxidation of the prochiral
C=C double bond of the allyI group attached to the anomeric position (Scheme
31-3m.

The set of compounds we plan to prepare include 01' and B-linked
glycosides with varying glycosidic dihedral angles. In addition, the compounds
represent the most common monosaccharide units in carbohydrates, namely,
glucose, galactose, mannose and fucose. As mentioned earlier we plan to
prepare compounds with rigid dihedral angles by forming a six-membered ring
around the 02-02-C1-O1 portion of the monosaccharides. The problem with
making the ring involving the glycosidic bonds is that the values of the glycosidic
dihedral angles we will be able to make are limited to certain values. But we

believe that the Karplus type equation parameterized for these dihedral angles

61

Figure 3.1 The ten molecules that have rigid glycosidic dihedral angles. While
molecules 8-10 are commercially available, molecules 1-7 are synthesized in

this work.

62

P

CH3 10 4 5 O
8 9 P O 2 1
HO 3 O O
3 7
2 QCH3
6 O
7 H 5 1 o
7
CH H 2 0A;
3 9 3 x0 3 8 CH3
H30 9
10 4

 

O
6 0 0‘” 1
5 or 1 4 3 OH
H 2
4 3 2 OH
OH
9 10

63

should reasonably predict the coupling constant or dihedral angle values for the
remaining range. The compounds we prepare will have around 60 or 180
degrees for the glycosidic dihedral angles. For the dihedral angles between 60
and 180 degrees, we will use the commercially available 1,6-
anhydroglucopyranose (8), 1,6-anhydrogalactopyranose (9), and 1,6-
anhydromannopyranose (10) (Figure 3.1). These three compounds cover the

range of about 110 to 165 degrees.

Preparation of a-glycosides

The a—glycosides with rigid glycosidic linkages were prepared by first
allylating the anomeric carbon by allyl alcohol using the traditional Fischer
glycosidation. This was followed by oxymercuration and intramolecular
nucleophilic attack by the 2-hydroxyl group of the monosaccharide3 to form the
six-membered ring with high or exclusive preference of the CHzHgOAc group for
the equatorial position. Finally, sodium borohydride reduction gave the final
products (Scheme 3.1 and 3.2).

The Fischer glycosidation of L-fucose gave almost exclusively the a—L-
fucoside (12) while that of D-glucose gave a mixture of a— and B—glucosides with
a quantitative yield. To separate the a— and B-glucosides and to improve their
solubility in non-aqueous solvents, they were benzyledenated4 using
benzyledene dimethyl acetal in dimethyl formamide in the presence of a catalytic

amount of p-toluene sulfonic acid (pTSA) to give allyl 4,6-O-benzyledene-D-

 

H30 Amberlite (H+)
0H »
0:410
1 1

H3O
OH

or-HO
12

1. HQ(OCOCF3)2

CH3CN
2, NaBH4
CH3
H3C
O
or—l"°
1

Scheme 3.1 Synthesis of 1 ,2-0-[2-methyI-(S)-1,2-ethanediyl]-a-L-fucopyranoside

(1)-

65

Figure 3.2 Proton and carbon nuclear magnetic resonance spectra of 1.2-0-

[2-methyl-(S)-1 ,2-ethanediyl]-or-L-fucopyranoside (1 ).

66

iLE.FP » Illin- IiPrLirLlh it 11 iriulibpbb P1» E III} lbi l Ibu'bl r?! FL 11>»!er DVD? » bl|| I
‘ldi‘ll‘11-111“ 1“ {1111- d111 1“ Ta 11 “‘1‘I‘41-l41 1.14141 1 I 11!}!1 It‘ll

 

 

 

 

 

.
h
—

-.l LillL'lLullL!

7 ~ jogj‘ illil.

 

 

 

67

OH 1. W0”

 

o Amberlite (W) o 0
OH > Ph/ K
HOHO OH O HO 0
2. PhCl-KOMe)2

13 DMF 14

1. H9(OCOCF3)2

CH30N
2. NaBH4
0 0
Ph/ <0
HO

O o
2 \VA

Me

Scheme 3.2 Synthesis of 4,6-O-benzyledene-1,2-0-[2-methyl-(R)-1,2-

ethanedin]-a -D-g|ucopyranoside (2).

68

Figure 3.3 Proton and carbon nuclear magnetic resonance spectra of 4.6-0-

benzyledene-1,2-0-[2-methyl-(R)-1,2-ethanediyl1-a -D-glucopyranoside (2).

69

'59

.. p».

Ofl ON On Ofl

bb-thiP—bth—rbhhbb

On on Oh On
bib—bhbbhhbbbbhhhb—

r3..-

 

 

 

 

On OOH Odd

Pb-b—Dbbbhbbbbbb

OOH OnH

phr—

bprrrbbb

 

 

 

 

70

glucosides. Crystallization in hexane-ethyl acetate (2:1) gave crystals of the allyl
4,6-O-benzyledene-a—D-glucoside (14). Oxymercuration of compounds 12 and
14 was then performed using mercuric trifluoroacetate in acetonitrile. Attack of
the oxymercurated double bond by the hydroxyl group at CZ position formed a
six-membered ring, creating a new chiral center, CB. The methylene mercuric
trifluoroacetate group is almost exclusively at the equatorial position because of
the 1,3-diaxial interactions between C3 and the CHzHgOAc group if it was to be
at the axial position. Finally reduction of the cyclized products with sodium
borohydride in methanol converted the CHzHgOAc group into a methyl group

affording the final products (1 and 2).

Preparation of B-glycosides

The B-glycosides (3-6) were prepared by allylation of the
monosaccharides using the Koenigs-Knorr reaction followed by oxymercuration
and intramolecular nucleophilic attack by the 2-hydroxl group of the
monosaccharides to form a six-membered ring and sodium borohydride
reduction to give the final cyclic B-glycosides (Schemes 3.3 - 3.6).

Starting from the commercially available D-glucose, D-galactose, L-fucose
and 3-O-methyl D-glucose, acetobromination5 was accomplished using acetyl
bromide in acetic acid. This activates the anomeric carbon by putting a bromine
group at it and protects the 2-, 3-, 4- and 6-hydroxyl groups with acetyl groups.

After rotaevaporation of the solvent, the remaining residue was allylated6 by allyl

71

1. CH3COBr

 

2 WOH
0” 3. NaOH
0.410 CH30H

11

H
SC ° °Z
on/
0

0H" 15
1. HQ(OCOCF3)2

CHaCN
2. NaBH4

H3C
O O
MJC H3
0W0 3

Scheme 3.3 Synthesis of 1,2-0-[2-methyl-(R)-1,2-ethanediyl]-B-L-fucopyranoside

(3)-

72

Figure 3.4 Proton and carbon nuclear magnetic resonance spectra of 1,2-0-

[2-methyl-(R)-1,2-ethanediyl]-B-L-fucopyranoside (3).

73

Od ON On Ov On on Oh on on OOfi

OH."

on."

cnd

 

P b b h b FF P b P b b t. P P P P P b P FL b b h P b b b F P b b P b P b F b h b b t— b D b DLF h P b — p b h b b b p h LP — b b P b h b
_. _ . C _ . .
an o a n n v n m
* 1P b P p b P b L p b r by n r b P P * b b h — P F b p — b P F P _ bill

 

 

 

 

 

, (:11 3:;

 

74

O 2. W OH O
OH Ang03 °\/\
HQ > HO
O OH

 

CH / ' CH OH
3 16 3 17 .
1. Hg(OCOCF3)2
CH3CN
2. NaBH4
”(@330
0 CH3

CH3’ 4

Scheme 3.4 Synthesis of 3-O-methyl-1,2-0-[2-methyI-(S)-1,2-ethanediyl]-B-D-

glucopyranoside (4).

75

Figure 3.5 Proton and carbon nuclear magnetic resonance spectra of 3-0-

methyl-1 ,2-0-[2-methyI-(S)-1 ,2-ethanedin]-B-D-glucopyranoside (4).

76

EAR O_fi On On O. on Ob Oh

D-bLLbP-llDPD PDhL-bbhbt—lhbbh-bnb

Inn O

[r h b D h b b h h

—>.

b

b

On
p

h

Pb

OOfi

r.

P

b

Odd Odd OMH

—*hLb

— D b b b p F

{zest}

 

 

77

alcohol in the presence of silver carbonate to give exclusively the allyl B-
glycosides. The acetyl groups at the 2-, 3-, 4- and 6-positions were then
deacetylated using sodium methoxide in methanol to give the allyl B-L-fucoside
(15), allyl 3-O-methyl-B-D-glucoside (17), allyl B-D-glucoside (18), allyl B-D-
galactoside (21 ).

Allyl B-D-glucoside (18) and allyl B-D-galactoside (21) were
benzyledenated by benzyledene dimethyl acetal in dimethyl formamide in the
presence of a catalytic amount of p-toluene sulfonic acid to give allyl 4.6-0-
benzyledene-B-D-glucoside (19) and allyl 4,6-O-benzyledene-B-D-galactoside
(22), respectively. The allyl B-glycosides 15, 17, 19 and 22 were then
oxymercurated using mercuric trifluoroacetate in acetonitrile. Attack of the
oxymercurated double bond by the hydroxyl group at CZ position formed a six-
membered ring, creating a new chiral center, C8. These compounds were finally
reduced with sodium borohydride similar to the allyl or-glycosides, the CHzHgOAc

group being converted into a methyl group to give the final products (3-6).

Preparation of the B-mannoside

Our attempt to prepare the B-mannoside using a similar procedure to that
of the other B-glycosides was not successful. As a result we designed a different
method to prepare the cyclic B-mannoside (7) (Scheme 3.7).

Starting from the commercially available methyl a—mannoside (23),

benzyledenation was performed by using benzyledene dimethyl acetal in

78

1 . /C/H\/3C:OOBHT

"(I/$0“ 2. 192C021 ‘ HO/§Y_:/O\/\
3 NaOH

 

 

CH30H
PhCH(OMe)2
pTSA
DMF

1. Hg(OCOCF3)2
CH30N
Ph’ <
CH3 2. NaBH4 Ph’ :0

5

Scheme 3.5 Synthesis of 4,6-O-benzyledene-1,2—0-[2-methyI-(S)-1,2-

ethanediyll-B-D-glucopyranoside (5).

79

Figure 3.6 Proton and carbon nuclear magnetic resonance spectra of 4.6-0-

benzyledene-1,2-0-[2-methyl-(S)-1 ,2-ethanedin]-B-D-glucopyranoside (5).

80

I50 OH ON On ov on O0 Ob on em OOH OHH ONH OnH OCH

[ll—lublelprppphprbL—prbbptpp—>->»—>p>»—->>-—ppprbhppppppuppy-pp—Pppppbepbppppb

. ,..

é ,_ . ._ _. . c _ . _ . _.._ . _ :

 

 

 

 

 

 

 

 

81

1 . CHacOBI'

 

 

OH
OH A92C03
H OH NaOH
20 CH30H
Ph
0
O 1. HQ(OCOCF3)2
O CH CN
0 2 3
0/9 2. NaBH4
H CH3
6

OH
OH
H OH
21
PhCH(OMe)2
pTSA
DMF
' Ph
0
O
O
H 0.. Q
22

Scheme 3.6 Synthesis of 4,6-O-benzyledene-1,2-0-[2-methyl-(S)-1,2-

ethanediyl]-B-D-galactopyranoside (6).

82

Figure 3.7 Proton and carbon nuclear magnetic resonance spectra of 4.6-0-

benzyledene-1,2-0-[2-methyl-(S)-1 ,2-ethanedin]-B-D-galactopyranoside (6).

83

 

 

OH On On ov Oh OD OD on on OOH OHH OOH OHH OOH
[Ill- D D D D b D D D D h D D D D b D D D D ~ D D D D b .P D D D h D D D D _ D D D D — D D D D _ D D D D — D D D D — D D D D - D D D D h D D D D . D D D
xi} 2
:59 H n n v n o D O m
f. D D D D b D D D D — D D D D — D D D D b D D P b D D D D P D D — D D h D D D

 

 

 

 

 

 

 

 

1 . PhCH(OMe)2

 

 

pTSA
DMF '3 HO .0.
O
> Ph‘ -\_ v
e 2. PhCHgBr P no / OMe
DMF 2 24
NaH
CI
OTHF

. Z £0
He 0 O O 1,LiAIH4 o

O
A .
H85 2 P“ 6 v
PhCHz/ - 0.5 M H2804 O

/ . OMe
7 PhCH2 25

 

Scheme 3.7 Synthesis of 3-O-benzyl-1,2-0-[2-methyl-(R)-1,2-ethanediyl]-B-D-

mannopyranoside (7).

85

Figure 3.8 Proton and carbon nuclear magnetic resonance spectra of 3-0-

benzyl-1 ,2-0-[2-methyl-(R)-1 ,2-ethanedin]-B-D-mannopyranoside (7).

86

Eh OH ON On ov On OO Oh OO OO OOH OHH OHH OnH OOH

DDDPDDDPPPhDDDDFDDlDDbrPLDPhLDPD-DDDF—DPDDFrLLr-DPDD—PDDDPPDDD—PDDD

..pbp—pi

 

 

 

 

 

 

87

dimethyl formamide in the presence of a catalytic amount of p-toluene sulfonic
acid to give methyl 4,6-O-benzyledene-a-D-mannoside. This resulted in
protection of the 4- and 6-hydroxyl groups of the methyl oc-D-mannoside. We
then protected the 3- hydroxyl group using benzyl bromide.7 in tetrahydrofuran to
get methyl 3-O-benzyl-4,6-O-benzyledene-oc-D-mannoside (24). Reaction of 24
with racemic epichlorohydrin gave the diastereomeric 25. This was followed by
lithium aluminum hydride reduction8 of the epoxide and acid hydrolysis9 to give
the final product 3-O-benzyl-1 ,2-0-[2-methyl-(R)-1 ,2-ethanedinJ-B-D-
mannopyranoside (7) after purification by flash chromatography using methanol-

chlorofon'n (1 :6).

Conclusions

Seven compounds (1-7) having rigid and predetermined glycosidic
dihedral angles were designed and synthesized starting from commercially
available monosaccharide units. Their synthesis was accomplished by allylation
(Fischer glycosidation for cur-glycosides and Koenigs-Knorr reaction for B-
glycosides), oxymercuric activation of the allylic double bond, and intramolecular
attack by the 2-hydroxy group to form a six-membered ring and finally reduction
by sodium borohydride. Synthesis of 7 required a different method as shown in
Scheme 3.7. These compounds (1-7) are important for parameterization of a
new Karplus-type equation relating the vicinal proton-carbon heteronuclear
coupling constant, 3JCH, with dihedral angles defined by H-C-O-C (see Chapter

Foun.

Experimental

General Techniques

All reagents used were reagent grade. Reaction temperatures were
measured externally. NMR spectra were recorded at 300 MHz with a Varian
spectrometer at the Max T. Rogers NMR Facility, Michigan State University.
They were obtained at 20 °C and chemical shifts are reported relative to the
residue solvent peak unless otherwise specified. Melting points were measured
using Fischer-Johns Melting Point Apparatus. Optical rotations were measured
using a Perkin Elmer polarimeter at 589 nm. Infra Red spectra were recorded
using Nicolet FTIR spectrometer. High-resolution mass spectra (HRMS) were
recorded on a JEOL HX-110-HF spectrometer using Fast Atom Bombardment
(FAB) conditions and an N-benzyl alcohol matrix. TLC was performed on pre-
coated plates of silica gel 60F, 0.20 mm thick (Merck). For detection, the plates
were sprayed with 50% H2804 (Orcinol) and heated for 3-5 minutes at 130 °C.
Flash chromatography was performed on Aldrich silica gel 60 (Merck), 200-400
mesh. Yields refer to chromatographically and spectroscopically (1H NMR)

homogeneous materials.

Synthesis
1 ,2-0-[2-methyl-(S)—1 ,2-ethanediyll-a-L-fucoside (1 ).

A mixture of L-Fucose 1.0 g and dry Dowex-50wx-8 (H‘) resin (0.5 g) in 15
mL allyl alcohol was stirred and heated under reflux (bath temperature 110 °C)

for 100 min. After cooling, the resin was filtered off, washed with anhydrous

89

ethanol (2 x 10 mL) and the combined filtrate was concentrated. The residue
was co-evaporated with a 1:1 mixture of benzene and anhydrous ethanol (2 x 15
mL) to yield 1.2 g of syrup. Then, mercuric trifluoroacetate (3.32 g) in 50 mL
acetonitrile was added drop wise in to allyl oc-D-fucoside (12) (0.81 g) in 50 mL
acetonitrile. The reaction mixture was stirred overnight at room temperature after
which 10 mL of 3M sodium hydroxide solution was added and stirring was
continued for a further 10 minutes. Sodium borohydride (0.60 9) dissolved in 10
mL of sodium hydroxide solution was added and the reaction mixture was stirred
for 3 h at room temperature. The suspension was then filtered through celite and
the clear filtrate was concentrated under vacuum to remove the solvent. The
residue thus obtained was purified by column chromatography
(chloroform/methanol = 6:1). Yield: 0.41 g (50 7.). M.p. 141-144 9C. 1H NMR
(300 MHz, D20): 5 0.92 (3H, d, J = 6.45 Hz, Me), 1.05 (1H, d, J = 6.45 Hz, Me),
3.36 (1H, dd, J = 12.30, 10.55 Hz, H7ax), 3.66 (1H, dd, J = 3.52, 10.55 Hz, H2),
3.73 (1H, d, J = 2.93 Hz, H4), 3.81 (1H, dd, J = 2.93, 12.30 Hz, H7e), 3.93 (1H,
ddd, J = 6.45, 2.93, 10.54 Hz, H8), 3.97 (1H, q, J = 6.45 H2, H5), 4.30 (1 H, dd, J
= 3.52, 10.55 Hz, H3), 4.87 (1H, d, J =3.52 Hz, H1); 13CNMR (75 MHz, 020): 6
94.5(C1), 72.73(C4), 71.74(02), 71.69(C7), 6955(05), 6430(03), 6367(08),
1577(06), 1530(09); Fl’lR (CHCla cast) 3418, 2976, 2901, 1449, 1338, 1173,
1143, 1109, 1079, 999, 960, 839, 758 cm"; [81020 = -83.939 (c 3.45, CHgOH);

FAB-HRMS (NBA): cacd CanOs [M+H]+, 205.1076, found 205.1086.

4,6-O-Benzylldene-1 ,2-0-[2-methyl-(R)-1 ,2-ethanedlyll-or-D-glucopyranoside
(2).

A mixture of D-glucose 6.0 g, allyl alcohol 25 mL) and dry Dowex-50wx-8
(H+) resin (2 g) was stirred and heated under reflux (bath temperature 110 °C) for
100 min. After cooling, the resin was filtered off, washed with anhydrous ethanol
(2 x 10 mL) and the combined filtrate was concentrated. The residue was co-
evaporated with a 1:1 mixture of benzene and anhydrous ethanol (2 x 15 mL).
The residual allyl D-glucopyranoside was directly benzylidenated with 2.5 mL of
benzaldehyde dimethyl acetal in 50 mL dimethyl formamide, by shaking for 15-20
h in the presence of catalytic amount of p-toluenesulfonic acid monohydrate.
After addition of NaHCOa (0.31 g), the solvent was then evaporated and the
residue Was treated with ice water (15 mL) and extracted with ethyl acetate (3 X
15 mL), and the organic phase washed with water, dried over anhydrous sodium
sulfate, and evaporated. Crystallization from hexane / ethyl acetate gave Allyl
4,6-O-benzylidene-a-D-glucopyranoside (2.11 g, 37% yield). Then mercuric
trifluoroacetate (2.2 g) in 100 mL acetonitrile was added drop wise in to allyl 4,6-
O-benzylidene-oc-D-glucopyranoside (14) (0.8 g) in 150 mL acetonitrile. The
reaction mixture was stirred at room temperature for 6 h after which 10 mL of 3M
sodium hydroxide solution was added and stirring was continued for a further 10
minutes. Sodium borohydride (0.4 g) dissolved in 10 mL of sodium hydroxide
solution was added and the reaction mixture was stirred for 3 h at room .
temperature. The suspension was then filtered through celite and the clear

filtrate was concentrated under vacuum at 30 °C to remove the acetonitrile. The

91

residue thus obtained was purified by column chromatography (hexane/ethyl
acetate = 1:2). Yield: 0.22 g (25 7.). M.p. 217 - 220 90. 1H NMR (300 MHz,
CDCI3): 6 4.96 (H1, d, J = 3.6 Hz), 3.71 (H2, dd, J = 3.46, 9.22 Hz), 4.49 (H3, dd,
J = 922,957 Hz), 3.50 (H4, dd, J = 957,944 Hz), 3.96 (H5, ddd, J = 4.72, 9.44,
10.30 Hz), 4.30 (H6eq, dd, J = 4.72, 10.43 Hz), 3.67 (H6ax, dd, J = 10.43, 10.30
Hz), 3.92 (H7eq, dd, J = 2.85, 11.73 Hz), 3.47 (H7ax, dd, J = 11.73, 10.51 Hz),
3.97 (H8, ddd, J = 2.85,6.10,10.51 Hz), 1.08 (H9, d, J = 6.10 Hz), 5.51 (H10, s),
7.41 (Ph ); 13CNMR(75 MHz, CDCla): 6 128.1 and 125.4 (Ph), 101.2 (010), 94.0
(C1), 80.2 (C4), 74.8 (C2), 71.4 (C7), 68.1(C6), 65.1 (03), 63.4 (05), 62.7 (08),
15.2 (09); Fun (CHCI3 cast) 3411, 2969, 2928, 2863, 1453, 1372, 1173, 1164,

1148, 1130, 1117, 1087, 1076, 1028, 1017, 969, 943, 885, 822, 750, 698 cm";

[61].,20 = +4779 (0 0.595, CHCI3).

1 ,2-0-[2-methyl-(R)-1 ,2-ethanediyl]-13-L-fucoslde (3).

L-Fucose 2.0 g was added at once into acetyl bromide under stirring at
0°C. Soon after this, cold acetic acid (8 mL) was added to the suspension.
During the next 10 min, all the sugar dissolved, with a fairly violent evolution of
HBr. After the evolution of gas stopped, the ice-bath was removed and the
reaction mixture was stirred for 1.5 hr at ambient temperature. The excess
reagents were evaporated under reduced pressure at room temperature and
then co-evaporated with toluene three times. The syrup thus obtained was
treated with allyl alcohol (15 mL) and A92003 (2.8 g). After passing through

celite, it was then rotaevaporated. The residue was then deacetylated with 0.4 %

92

NaOMe in 100 mL methanol. It was then neutralizedwith Amberlite (H*) ion
exchange resin. The resin was filtered off and the solution was evaporated
under low pressure. The crude residue obtained was directly cyclized as follows.
Mercuric trifluoroacetate (1.66 g) in 100 mL acetonitrile was added drop wise in
to allyl B-L-fuccopyranoside (0.4 g) in 150 mL acetonitrile. The reaction mixture
was stirred at room temperature for 6 h after which 10 mL of 3M sodium
hydroxide solution was added and stirring was continued for a further 10 minutes.
Sodium borohydride (0.3 9) dissolved in 10 mL of sodium hydroxide solution was
added and the reaction mixture was stirred for 3 h at room temperature. The
suspension was then filtered through celite and the clear filtrate was
concentrated under vacuum at 30 °C to remove the acetonitrile. The residue
thus obtained was purified by column chromatography (chloroform : methanol =
6:1). Yield: 0.16 g (40 %). M.p. 164 - 167 9c. 1H NMR (300 MHz, 020): 6 0.98
(3H, d, J = 6.45 Hz, Me-9), 1.12 (3H, d, J = 6.45 Hz, Me-6), 3.38 (1H, dd, J =
11.58,10.80 Hz, H7ax), 3.25 (1 H, dd, J = 7.62996 H2, H2), 3.62 (1 H, d, J = 3.51
H2, H4), 3.83 (1H, dd, J = 2.35, 11.58 Hz, H7e), 3.73 (1H, ddd, J = 6.45, 2.35.
10.80 H2, H8), 3.76 (1H, q, J = 6.45 H2, H5), 3.61 (1H, dd, J = 3.51, 9.96 H2,
H3), 4.23 (1H, d, J =7.62 Hz, H1); ‘3CNMR (75 MHz, 020): 6 98.2 (C1), 76.9
(02), 72.8 (CS), 72.8 (C8), 71.8 (C3), 71.8 (C4), 71.6 (C7), 15.4(C6), 15.2(09);
Fl' IR (CHCla cast) 3460, 2983, 2916, 1447, 1370, 1277, 1156, 1131, 1100, 1075,
1002 cm"; [61020 = 39.39 (c 0.82, CH30H); FAB-HRMS (NBA): cacd C9H1705

[M+H]*, 205.1076, found 205.1077.

93

3-O-Methyl-1,2-0-[2-methyl-(S)-1,2-ethanediyl]-8-D-glucoslde (4).
3-O-Methyl-1,2-0-[2-methyI-(S)-1,2-ethanediyl]-B—D-glucoside (4) was
prepared using a procedure similar to the one used to prepare 1,2-0-[2-methyl-
(R)-1,2-ethanediyll-B-L-fucoside (3) starting from 3-O-methyl-D-glucose. Yield:
0.15 g (26 9/6). 1H NMR (300 MHz, 00300): 6 1.00 (3H, d, J = 6.41 Hz, Me-9),
3.45 (3H, s, OMe-10), 3.07 (1 H, dd, J = 783,996 H2, H2), 3.26 (1 H, dd, J =
800,996 H2, H3), 3.34 (1H, d, J = 800,961 H2, H4), 3.40 (1H, dd, J =
12.27.10.85 Hz, H7a), 3.60 (1H, dd, J = 4.98.12.45 Hz, H6), 3.70 (1H, ddd, J =
2.49,6.41,10.85 H2, H8), 3.71 (1H, dd, J = 9.61 ,4.98 H2, H5), 3.75 (1 H, d, J =
12.45, 9.96 H2, H6’), 3.85 (1 H, dd, 2.49.12.27 Hz, H7a), 4.34 (1 H, d, J =7.83 H2,
H1); 13CNMR (75 MHz, CDaOD): 5 97.5 (C1), 78.3 (C2), 82.3 (C3), 71.4 (C4),
77.7 (OS), 60.6 (C6), 71.4 (C7), 72.2 (C8), 15.2(09), 79.7 (OMe-10); FTIR
(CHCI3 cast) 3421, 2916, 1456, 1376, 1278, 1151, 1121, 1068, 1034 cm"; [611020
= +3379 (0 0.72, CH30H); FAB-HRMS (NBA): cacd C10H1906 [M+H]*,

235.1182, found 235.1174.

4,6-O-Benzylidene-1 ,2-0-[2-methyl-(S)-1 ,2-ethanedlyl]-8-D—glucopyranoside
(5).

D-Glucose (2.0 g) was added at once into acetyl bromide under stirring at
0°C. Soon after this, cold acetic acid (8 mL) was added to the suspension.
During the next 10 min, all the sugar dissolved, with a fairly violent evolution of
HBr. After the evolution of gas stopped, the ice—bath was removed and the

reaction mixture was stirred for 1.5 hr at ambient temperature. The excess

94

reagents were evaporated under reduced pressure at room temperature and
then co-evaporated with toluene three times. The syrup thus obtained was
treated with allyl alcohol (15 mL) and A92003 (2.8 9). After passing through
celite, it was then rotaevaporated. The residue was then deacetylated with 0.4 %
NaOMe in 100 mL methanol. It was then neutralized with Amberlite (H*) ion
exchange resin. The resin was filtered off and the solution was evaporated
under low pressure. The residue obtained was directly benzylidenated as
follows. The residual allyl B-D-giucopyranoside was directly benzylidenated with
0.83 mL of benzaldehyde dimethyl acetal in 50 mL dimethyl formamide, by
shaking for 15-20 h in the presence of catalytic amount of p-toluenesulfonic acid
monohydrate. After addition of NaHCOa (0.3 g), the solvent was evaporated, the
residue treated with ice-water (15 mL) and extracted with ethyl acetate (3 X 15
mL), and the organic phase washed with water, dried over anhydrous sodium
sulfate, and evaporated. Then, mercuric trifluoroacetate (2.2 g) in 100 mL
acetonitrile was added drop wise in to allyl 4,6-O-benzyiidene-B-D-
glucopyranoside (19) (0.8 g) in 150 mL acetonitrile. The reaction mixture was
stirred at room temperature for 6 h after which 10 mL of 3M sodium hydroxide
solution was added and stirring was continued for a further 10 minutes. Sodium
borohydride (0.4 g) dissolved in 10 mL of sodium hydroxide solution was added
and the reaction mixture was stirred for 3 h at room temperature. The
suspension was then filtered through celite and the clear filtrate was
concentrated under vacuum at 30 °C to remove the acetonitrile. The residue

was then purified by column chromatography (hexane/ethyl acetate = 1:2). Yield:

95

0.25 g (31 9.). Mo 188 - 190 9c. 1H NMR (300 MHz, CDCla): 6 1.16 (3H, d, J
= 6.23 Hz, Me-9), 3.30 (1H, dd, J = 7.8.9.52 Hz, H2), 3.49 (1H, dd, J =
10.721172 Hz, H7ax), 3.85 (1 H, dd, J = 6.22, 2.93, 12.8 Hz, H8), 3.82 (1 H, dd, J
= 9.89, 10.25 H2, H3), 3.89 (1 H, 842,842, H5), 3.94 (1 H, dd, J = 11.72, 2.93 Hz,
H7e), 4.20 (1 H, d 11.7 Hz, H6), 3.61 (1 H, dd, J = 1025.842 Hz, H4), 4.39 (1 H, d,
J = 7.69 H2, H1), 5.54 (1 H, 5, H10), 7.40 (5H, m, Ph); 13CNMH (75 MHz, CDCls):
6 129.4, 128.4 (Ph), 102.38 (C10), 98.87 (C1), 81.54 (C2), 71.03 (C5), 68.68
(C3), 68.31 (C4), 72.27 (C7), 71.87 (C8), 68.65(C6), 15.2(C9); FTIR (CHCl3
cast) 3459, 2975, 2871, 1452, 1370, 1314, 1277, 1146,1120, 1094, 1066, 1055,

1029, 1006, 971, 844, 749, 699 cm"; [(11020 = +1009 (0 1.22, CHCla).

4,6-O-Benzylidene-1 ,2-0—[2-methyl-(S)-1,2-ethanedlyll-l3-D-
galactopyranoside (6).

A procedure similar to the one we used to prepare 4,6-O-Benzylidene-1,2-
O-[2-methyl-(S)-1,2-ethanediyll-B-D-glucopyranoside (5) was used to prepare
4,6-O-Benzylidene-1 ,2-0-[2-methyl-(S)-1 ,2-ethanediyll-B-D-galactopyranoside
(6) starting from D-galactose. Yield: 0.24 g (30 7.). 1H NMR (300 MHz, CDCla): 6
4.30 (H1, (1, J = 7.69 Hz), 3.60 (H2, dd, J = 769,989 Hz), 3.79 (H3, dd, J =
989,384 Hz), 4.25 (H4, dd, J = 3.84,0.91 Hz), 3.61 (H5, d, J = 1.46 Hz), 4.39
(H6, dd, J = 1.46,12.67 Hz), 4.08 (H6’, d, J = 12.67 Hz), 3.92 (H7eq, dd, J = 2.75,
11.89 Hz), 3.47 (H7ax, dd, J = 11.89, 10.25 Hz), 3.85 (H8, ddd, J =
2.75,6.41,10.25 Hz), 1.15 (H9, d, J = 6.41 Hz), 5.56 (H10,'s), 7.37 (Ph ); 13mm
(75 MHz, CDCI3): 6 129.4, 127.3, 126.6 (Ph), 101.7 (C10), 98.7 (C1), 77.3 (C2),

96

70.6 (C3), 75.9 (C4), 68.3 (C5), 69.1 (C6), 72.1 (C7), 71.7 (C8), 16.3 (C9); FTIR
(CHCl3 cast) 3463, 2973, 2871, 1713, 1452, 1401, 1370, 1331, 1315, 1288,
1250,1218,1173,1159,1142,1124,1101,1079,1060,1024,1007,973,919,
889 cm"; [61020 = +5249 (0 0.55, CHCla); FAB-HRMS (NBA): cacd C16H21Oe

[M+H]+, 309.1338, found 309.1349.

3-O-benzyl-1 ,2-0-[2-methyl-(R)-1 ,2—ethanedinJ-B-D-mannopyranoside (7).
Methyl oc-D-mannopyranoside glucopyranoside 4.0 g was directly
benzylidenated with 2.5 mL of benzaldehyde dimethyl acetal in 50 mL dimethyl
formamide, by shaking for 15-20 h in the presence of catalytic amount of p-
toluene sulfonic acid monohydrate. After addition of NaH003 (0.5 g), the solvent
was evaporated, the residue treated with ice-water (17 mL) and extracted with
ethyl acetate (3 X 20 mL), and the organic phase washed with water, dried over
anhydrous sodium sulfate, and evaporated. Then, a mixture of Methyl 4,6-0-
benzylidene-ol-D-mannopyranoside (2.5 g) and dibutyl tin oxide (2.2 g) in 100 mL
methanol was refluxed for 1 h at 60 °C. The solvent was then rotaevaporated
and the residue dissolved in 125 mL dimethyl formamide. To this solution 1.8 mL
of benzyl bromide was added and was stirred at 100 °C for 30 min. It was then
concentrated and purified by flash chromatography with ethyl acetate-hexane
(3:2) to give methyl 3-O-benzyl-4,6-O-benzylidene-oi-D-mannopyranoside (24).
Then, to a solution of 0.64 g of methyl 3-O-benzyl-4,6-O-benzylidene-ol-D-
mannopyranoside (24) in 30.0 mL of tetrahydrofuran was added 0.41 g of

sodium hydride, 22.9 mg of imidazole and 22.6 mg of tetrabutyl ammonium

97

bromide. It was then refluxed for 0.5 h and 1.6 g of epichlorohydrin was added
and refluxed overnight. To this mixture 20.0 mL of methanol was added drop
wise followed by 42.0 mL of dilute aqueous HCI (0.12 % v/v). It was then
extracted with methylene chloride (3 x 60 mL), the organic layer washed with
dilute solutions of sodium carbonate and brine and dried over anhydrous sodium
sulfate and the solvent evaporated under reduced pressure. And the residue
was purified by flash chromatography (hexane-ethyl acetate 6:1) to give 0.54 9
product (57% yield). To a solution of this epoxide (0.54 g) in 75 mL of
tetrahydrofuran at 0°C under N2 was added 0.46 mg of Lithium Aluminum
Hydride. The mixture was stirred at 0°C for 1 h and at room temperature for 5 h.
The reaction was quenched by the addition of saturated aqueous solution of
sodium sulfate until a white, granular precipitate was observed. The THF
solution was then decanted, and the salts were washed with additional THF. The
combined organic solution was dried over anhydrous sodium sulfate and
concentrated under reduced pressure to give 0.54 g (99% for the reduction step)
of 1-O-methyl-2-(2-hydroxypropyl)-3-O-benzyl-4,6-O-benzylidene-0l-D-manno-
pyranoside. Hydrolysis of methyl 2-(2-hydroxy propyl)-3-O-benzyl-4,6-O-
benzylidene-a—D-mannopyranoside was performed by refluxing overnight in 0.5
M H2804 and then purification by flash chromatography (methanol/chloroform
1/6) gave 0.1 g (22% yield) of 3-O-benzyl-1,2-0-[2-methyl-(R)-1,2-ethanediyl]-B-
D-mannopyranoside (7). 1H NMR (300 MHz, 020): 6 1.03 (3H, d, J = 6.59 Hz,
Me-9), 3.34 (1 H, ddd, J = 2.20,5.37,9.77 H2, H5), 3.27 (1 H, dd, J = 10.74,11.96
Hz, H8ax), 3.70 (1H, dd, J = 9.76977 Hz, H4), 3.87 (1H, dd, J = 2.20,12.20 Hz,

98

H6), 3.81 (1H, dd, J = 2.921196 Hz, H8e), 3.57 (1 H, dd, J = 3.66.9.76 Hz, H3),
3.74 (1H, dd, J = 5.37,12.20 Hz, H6’), 4.02 (1H, dd, J = 3.66, 1.46 Hz, H2), 4.27
(1H, ddd, J = 10.74, 2.93, 6.59 H2, H7), 4.78 (1H, d, J = 1.46 Hz, H1), 4.54 (2H,
s, CH2-10), 7.34 (5H, m, Ph); 13CNMH (75 MHz, 020): 6 128.9, 137.0 (Ph), 93.5
(C1), 71.1 (C2), 76.3 (C5), 79.2 (C3), 71.2 (C7), 66.0 (C4), 66.0 (CB), 73.0 (C10),
62.0 (C6), 15.4 (C9); FTIR (CHCI3 cast) 3419, 2871, 1496, 1454, 1367, 1274,
1208, 1173, 1071, 860, 741, 699 cm"; [611020 = -18.49 (c 1.23, CH30H); FAB-

HRMS (NBA): cacd C16H2306 [M+H]*, 311.1495, found 311.1496.
Acknowledgements

This work was supported by the Michigan State University Research

Excellence Fund.

99

References

3’19"?J

Tvaroska, I; Hricovini, M; Petrakova, E. Carbohydr. Res. 1989, 189, 359.
Mulloy, B.; Frenkiel, T. A.; Davies, D. B. Carbohydr. Res. 1988, 184, 39.
Huang, G.; Hollingsworth, R. l. Tetrahedron 1998, 54, 1355.

Pelyvés, l.; Lindhorst, T.; Thiem, J. Liebigs Ann. Chem. 1990, 761.

Koto, S.; Yoshida, T.; Takenaka, K.; Zen, S. Bull. Chem. Soc. Jpn. 1982,
55,3667.

Koto, 8.; Morishima, N.; Takenaka, K.; Kanemitsu, K.; Shimoura, N.;
Kase, M.; Kojiro, S.; Nakamura, T.; Kawase, T.; Zen, S. Bull. Chem. Soc.
Jpn. 1989, 62, 3549.

Nashed, M. A.; Anderson, L. Tetrahedm Left. 1976, 39, 3503.

Loughhead, D. G. J. Org. Chem. 1985, 50, 3931.

Tomaszewski, M. J.; Holme, K. R.; Perlin, A. S. Carbohydr. Res.

1991 ,217, 237.

100

Chapter 4

New Karplus-type Equations Relating the Glycosidic Dihedral

Angles with Vicinal Proton-Carbon Coupling Constants.

101

Abstract

We report here new Karplus-Type equations relating the vicinal proton-
carbon coupling constants, 3JHc, to the glycosidic dihedral angles 0 (H1'-C1'-OX-
CX) and w (HX-CX-OX-Cf‘). The three dimensional structures of the synthetic
model compounds 1-7 and the commercially available compounds 8-10 have
been studied by Molecular Mechanics and AM1 and PM3 Semi-empirical
calculations. The vicinal proton-carbon coupling constants, 3JHc, have been
measured in solution by two-dimensional excitation-sculptured indirect-detection
experiment (EXSIDE) NMR spectroscopy. From the least squares fitting of the
measured coupling constants and the glycosidic dihedral angles of the rigid
molecules, equations of the form: 3JHC (0) = a Cos2 (0) + b Cos (0) + c, were

obtained where 6 stands for the dihedral angles defined by the four atoms

H-C-O-C (0 and \II for carbohydrates).

102

Introduction.

The equation 3Jnc (6) = 5.7 Cos2 (0) — 0.6 Cos (0) + 0.5 (equation 1),
where 0 stands for the dihedral angle defined by the atoms H-C-O-C, has been
proposed by Tvaroska et. all.1 and a similar equation 3JHc (0) = 5.5 Cos2 (0) — 0.7
Cos (6) + 0.6 (equation 2) has also been proposed by Mulloy et. al.2 The above
equations have been used to relate the vicinal proton-carbon heteronuclear
coupling constants across the glycosidic bonds of carbohydrates to the glycosidic
dihedral angles. However, more often than not, the structure of the
carbohydrates predicted by using the above equations is different from the
structures predicted by other methods.3 We believe that the reason for the
discrepancy is the following:

i) The glycosidic dihedral angles of carbohydrates defined by the atoms H-
C-O-C is, to our knowledge, almost always in the range 0 to 80 degrees. This is
based on analysis of the crystal structure of carbohydrates from Cambridge
Structural Database. This is discussed in detail in chapter two.

Ii) Only 3 data points are used to parameterize the above equation 1 for
the region 0 to 85 degrees. The remaining 14 data points are in the range of 86
to 180 degrees. Even those 3 data points, as pointed out by authors, are “only
estimated values and therefore the precision of these data is lower".1 The
accuracy of the values of the three through-bond coupling constants is also not
given. And in the case of equation 2 only 6 of the 18 data points are in the range
of 0 to 90 degrees and the remaining 12 are in the range of 90 to 180 degrees.

In addition, equations 1 and 2 are parameterized using coupling constants

103

measured in solution and dihedral angles from crystal structures. But the
solution conformations of their model compounds and carbohydrates, in general,
may not be identical with the crystal structures.

iii) The electronic environments (electronegativity and anomeric effect,
among others) are different for the two glycosidic dihedral angles, 0 (H1'-C1'—OX—
CX) and w (HX-CX-OX-C1') and hence the magnitudes of the through-bond
coupling constants for a given dihedral angle may be different. And this
difference might require two equations to describe the dependence of the three
bond proton-carbon heteronuclear coupling constants of carbohydrates across
the glycosidic bonds to the glycosidic dihedral angles, one for each angle 0 (H1'-
C1'-OX—CX) and w (HX-CX-OX-C1 ').

More recently Cloran et. al. reported a new Karplus type equation, 3Jnc (0)
= 7.49 Cos2 (6) — 0.96 Cos (6) + 0.15 (equation 3), where 6 stands for the
dihedral angle defined by the atoms H-C-O-C, relating the vicinal proton-carbon
coupling constants to the glycosidic dihedral angles from pure theoretical
calculations of the coupling constants and geometries of their model
compounds.“'5 This equation shows a big difference with the previ0us two
equations of Tvaroska et al.1 and Mulloy et al.2 at lower and higher dihedral
angles and factors such as solvation, basis set limitations, the small set of
geometries and structures studied are suggested as possible factors for the
deviations from the above mentioned equations. The equation of Haasnoot et.

al. is a well-parameterized Karplus-type equation for co dihedral angle of

104

Figure 4.1 The ten molecules that have rigid glycosidic dihedral angles. While
molecules 8-10 are commercially available, molecules 1-7 are synthesized in this

work (Chapter 3).

105

CH: 1004 5°

 

6
O
O 6 O OH 1
O\H 1 5 01" 1 \ .
4 3 H 4 3 2 OH
2 4 3 2 OH
OH OH OH
8 9 10

106

carbohydrates relating the dihedral angles defined by H-C-C-H with the proton-
proton coupling constants for 1->6 linked sugars.6

Because of the reasons discussed above we decided to synthesize
molecules with rigid glycosidic linkages, measure their vicinal proton-carbon
coupling constants in solution, refine their dihedral angles by molecular
mechanics and semi-empirical (AM1 and PM3) calculations, and least squares fit
the dihedral angles and coupling constants to come out with new Karplus-type
equations. The synthesis of the model compounds is presented in Chapter
Three. The structural studies and the parameterization of the new Karplus-type

equation are reported in this Chapter.

Material and Methods.

Model compounds.

The model compounds 1-7 were synthesized in this work and their
synthesis and characterization is described in Chapter Three. And compounds
8-10 were commercially available and were used as is. Seven of the model
compounds are bicyclic and the remaining three are tricyclic. The model
compounds contained a and B sugars, D and L sugars, substituted and
unsubstituted sugars and the most common monosaccharide units in nature,

namely, glucose, galactose, mannose, and fucose.

107

Nuclear Magnetic Resonance (NMR) Spectroscopy.

Nuclear Magnetic Resonance spectra of compounds 1, 3, 4 and 7-10 were
obtained in deuterated water, while that of 2, 5, and 6 were obtained in
deuterated chloroform. Spectra are recorded at 30 °C. The residual solvent
peaks were used as reference. Combinations of homonuclear— and
heteronuclear- (one- and) two-dimensional NMR experiments were performed in
order to allow the assignments of the signals for the model compounds and to
measure the coupling constants. This included: double quantum filtered J-
correlated spectroscopy (DQF-COSY) spectra (Ernst et al., 1987),7 total
correlated spectroscopy (TOCSY) spectra (Bax & Davis, 1985), 3 heteronuclear
multi-quantum coherence (HMQC) experiment (Bax & Subramanian, 1986),9
heteronuclear multi-bond coherence (HMBC) experiment (Titman et al., 1989),10
and nuclear Overhauser effect (NOESY) experiment (Sangers et al., 1993).11

Analysis of the spin systems for the model compounds from the 1D NMR
spectra and from DQF-COSY and TOCSY spectra gave the connectivities. Once
the assignments are made NOESY experiment was used to determine the
various inter-nuclear distances. Spectra of the NOESY experiments were
acquired at different mixing times to address the potential problem of spin
diffusion.

The vicinal proton-carbon coupling constants were obtained from
excitation-sculptured indirect-detection (EXSIDE) experiments (Krishnamurthy et.

al., 1996)”. Excitation bandwidth selection, also known as excitation-sculpting,

was carried out with the Varian software-intemal package using a Gaussian

108

cascade 03 pulse of 14 ms duration for selective inversion of the selected
protons in compounds 1-10. A J-scaling factor of 15 was used and 16 scans of
2048 complex data points were collected for 512t1 increments. EXSIDE provides
sensitivity comparable to a nonselective gradient HMBC experiment, but with
pure absorptive line shapes. Long-range proton-carbon coupling constants were
J-scaled and measured along the F1 (carbon) dimension. In the EXSIDE
spectrum, the active coupling constant is measured with no interference from any
passive couplings, in contrast to other methods where the proton-carbon
coupling constants are measured along the F2 (proton) dimension with
overlapping homonuclear couplings. But the resonances of interest should be
well resolved from any homonuclear-coupled spin partner. The EXSIDE
experiment is especially very useful for carbohydrates since the anomeric
protons are further away from the rest of the bulk proton signals and hence can
be selectively excited. The error in measured coupling constants was i 0.1 Hz.
All of the two dimensional NMR spectra were measured on Varian lnova 600
MHz NMR spectrometer operating at 600 MHz for protons and 150 MHz for
carbons. The tables of assignment of the proton and carbon chemical shifts and
coupling constants of compounds 8 — 10 is given in Tables 4.1 - 4.3, while
compounds 1 - 7 were fully characterized and their proton and carbon chemical

shifts and coupling constants assigned in Chapter Three.

109

Table 4.1. 1H and 130 chemical shift assignments and proton-proton coupling

constants of 1,6-anhydroglucopyranose (8).

 

 

 

 

 

 

 

 

1,6-anhydro 1H (6, ppm) 3Jl-ll-l (Hz) 13C (8, ppm)
glucose (8)

1 5.40 1.17 (3JH1H2') 101.7

2 3.47 70.5

3 3.63 72.7

4 3.62 71.1

5 4.58 5.86 (3.1mm) 76.5

6 4.03 7.62 (3.145%) 65.4

6’ 3.71 5.86, 7.62 65.4

(3JH5H6'. 3JH6H6')

 

 

 

 

 

 

110

Table 4.2. 1H and 13C chemical shift assignments and proton-proton coupling

constants of 1,6-anhydrogalactopyranose (9).

 

1.6-anhydro 1H (8, ppm) 3JHH (HZ) ‘30 (5, DPT")

galactose (9)

 

 

 

 

 

1 5.37 1.46 (3.14116) 100.9
2 3.90 4.89 (3.11.2113) 73.4
3 3.76 1.71 (3.1.4.414) 71.5
4 4.01 4.89 (3JH4H5) 64.5
5 4.46 464,488 74.5

(av-1114115. 3JHsHs’)

 

6 4.28 7.81 (3JH6H6’) 63.7

 

6’ 3.61 5.12,7.81 63.7

(3JH5H6'. 3JH6H6')

 

 

 

 

 

 

111

Table 4.3. 1H and 130 chemical shift assignments and proton-proton coupling

constants of 1,6-anhydromannopyranose (10).

 

 

 

 

 

 

 

 

 

 

 

(3JH5H6'. 3JH6H6’)

 

1.6-anhydro 1H (5, ppm) 3Jl-IH (Hz) ‘30 (5, ppm)
mannose (1 O)
1 5.28 101 .3
2 3.60 4.88 (3.1mm) 66.0
3 3.88 1 .83 (3.1113144) 71.6
4 3.84 1.46 (3.11.4115) 70.3
5 4.48 5.86 (3JH5H6') 75.8
6 4.10 7.57 (3JH6H6') 64.8
6’ 3.62 5.86757 64.8

 

112

 

Molecular Mechanics and Seml-empirlcal Calculations

Molecular mechanics calculations were performed using the Alchemy
2000 program (Tripos, Inc. St. Louis, MO, USA) using the MM3 force fields.
Minimizations are performed using the conjugate gradient method.

Semi-empirical calculations were performed using the Alchemy 2000
program (Tripos, Inc. St. Louis, MO, USA). The geometries of the molecules
were optimized using the AM113 and PM3“ Hamiltonians in the Alchemy 2000
program and the geometries were re-optimized after they were perturbed.

Flgures 4.24.5 are color presentations for clarity.

Results and Discussion

We observed that sucrose has a dihedral angle of 8 degrees in the crystal
structure in the Cambridge Structural Database. Also its 2-hydroxyl group and
the C1 hydroxymethyl group are perfectly situated to form an eight membered
ring by acetal formation. Based on this information we prepared the sucrose
acetal hoping that it will serve as a model for a dihedral angle value of about zero
degrees. But the measured three-bond proton-carbon coupling constant, 2.07
Hz, is actually consistent with a W structure of the cyclooctane ring with a
dihedral angle value of around 60 degrees. As a result we parameterized the
Karplus-type equation based on our studies of compounds 1-10 which does not

include dihedral angles around 0 degrees.

113

Tvaroska et. al. used 17 data points to parameterize the Karplus-type
equation 3JHC (6) = 5.7 Cos2 (9) — 0.6 Cos (9) + 0.5, where 6 stands for the
dihedral angle defined by the atoms H-C-O-C. Of the 17 data points, 14
correspond to magnitudes of 6 between 86 and 180 degrees and of the
remaining 3, 2 are for values of 6 being 60 degrees with 3.1.40 of 2.2 and 2.6 Hz
and the remaining 1 is for a value of 9 being 10 degrees with 3JHC of 5.2 Hz. In
addition, the 3 data points that were used to parameterize their Karplus-type
equation for the range of 0 to 86 degrees are ‘only estimated values and
therefore the precision these is lower’.1 The studies of Mulloy at. al. has 18 data
points, 12 of which are above 105 degrees. Furthermore, both equations 1 and 2
were obtained by fitting crystal geometries with experimentally measured
coupling constants in solution. But the solution conformations of their model
compounds and carbohydrates, in general, may not be identical with the crystal
structures.

Our studies (Chapter two) revealed that the values of the critical dihedral
angles of carbohydrates that determine their three dimensional structure are
almost always in the range of 0 to 85 degrees. It became apparent than that the
equations of Tvaroska et. al. and Mulloy et. al. are poorly parameterized for
carbohydrates.

We used 28 data points out of which 13 are between 0 and 90 degrees
and 15 are between 90 and 180 degrees. For the range of 9 between 100 and
180 degrees our data points (Tables 4.4 and 4.5) lay close to the Karplus-type

curves of Tvaroska at. al. and Mulloy et. al. (Figure 4.2) This is not surprising

114

Table 4.4. Heteronuclear three-bond proton-carbon coupling constants of

compounds 1-5 measured from 2D EXSIDE NMR experiments and the

corresponding dihedral angles (torsions). The compound number is given in

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

bracket.
Dihedral 3JCH Torsion (MM Torsion (AM1 Torsion
calculation) calculation) (PM3
calculation)
H1C1osc5 (1) 6.837 -175.8 173.1 168.3
H1 C1 0707 (1) 1.211 -59.7 -72.1 -77.7
H7aC707C1 (1) 1 .469 59.3 68.1 71 .7
H5C505C1 (1) 1 .95 55.5 58.2 63.1
H1C101 C7 (2) 2.273 59.3 62.5 71.1
H7axC701 C1 (2) 1 .667 59.5 -59.8 -75.5
H1C105C5 (2) 6.671 176.7 -174.1 -163.1
H7aC707C1 (3) 1 .225 -61 .0 -68.6 -73.8
H7eC707C1 (3) 6.813 179.4 173.7 171 .4
H5CSOSC1 (3) 1.711 58.8 58.6 61.4
H1 C1 0707 (4) 2.197 -60.2 -63.1 -66.1
H7aC707C1 (4) 1 .21 1 60.6 67.5 74.0
H7eC707C1 (4) 7.015 -179.6 -174.8 -171.2
H1 C1 0505 (5) 1 .947 60.1 62.4 63.6
H7eC707C1 (5) 7.203 -179.1 -173.8 -171.2
H5C505C1 (5) 1 .953 -58.1 -59.6 -60.5

 

 

 

 

 

115

 

Table 4.5. Heteronuclear three-bond proton-carbon coupling constants of

compounds 6-10 measured from 2D EXSIDE NMR experiments and the

corresponding dihedral angles (torsions).

 

 

 

 

 

 

 

 

 

 

 

 

 

Dihedral 3J¢H Torsion (MM Torsion (AM1 Torsion
calculation) calculation) (PM3
calculation)
H7aC707C1 (6) 1 .225 61 .3 69.9 74.0
H5C505C1 (6) 1 .095 -60.6 -70.1 -70.2
H1C107C7 (7) 6.587 178.8 -174.7 -166.2
H1C106C6 (8) 4.883 138.3 149.7 148.5
H6C606C1 (8) 3.175 -107.5 -130.3 -130.0
H5C505C1 (8) 5.878 164.0 159.0 160.1
H1C106C6 (9) 4.871 138.9 147.6 147.1
H6C606C1 (9) 2.692 -107.5 -126.5 -127.0
H5C505C1 (9) 5.867 164.4 159.6 160.6
H1C106C6 (10) 5.398 137.6 148.6 147.2
H6C606C1 (10) 3.170 -106.2 -127.9 -127.1
H5C50501 (10) 5.605 164.3 159.2 160.7

 

 

 

 

 

 

116

 

because their equations are actually reasonably well optimized in this range of
dihedral angles because the higher percentage (14 of 17 in Tvaroska et. al. and
12 of 18 in Mulloy et. al.) of their data points belong to this range of dihedral
angles. The fact that our data points (AM1 and PM3 optimized geometries with
the experimentally measured heteronuclear coupling constants) are in good
agreement with the predictions by the equations of Tvaroska at. al. and Mulloy et.
al. for the range of dihedral values between 100 and 180 degrees gave us
confidence that the geometries obtained by the semi-empirical calculations are
reasonably reliable.

In the important range of dihedral angles for carbohydrates, 0 to 85
degrees, our data points are significantly different from the Karplus-type curves
of Tvaroska at. al. and Mulloy et. al. (Figure 4. 2). And indeed, the least square
fit of our data points (e.g. equatlon 4: 3.1110(9) = 5.597 Cos2 (9) - 0.421 Cos (9) +
0.722) gives a different prediction with better R2 values (Table 4.10).

Molecular mechanics and PM3 and AM1 semi-empirical calculations were
used to refine the structures of the already constrained bi- and tri-cyclic model
compounds (1 -— 10). Molecular mechanics optimization of the structures of
compounds 1 - 10 revealed that the dihedral angles obtained from this studies
are far way from the Karplus-type curves of Tvaroska et. al. and Mulloy et. al.
(Figure 4.2 ). Especially for the dihedral angles around 60 degrees, the dihedral
angles from the MM calculations are between 59 and 61 degrees for all

molecules even though the values of the vicinal proton-carbon heteronuclear

117

coupling constants are distributed between 2.27 and 1.09 Hz (Table 4.4 and 4.5
and Figure 4.2). In contrast the corresponding dihedral angles of the molecules
obtained from geometry optimizations by PM3 and AM1 semi-empirical
calculations are distributed between 58 and 78 degrees, which is consistent with
the distribution of the vicinal proton-carbon heteronuclear coupling constants
between 2.27 and 1.09 Hz (Tables 4.4 and 4.5 and Figures 4.3 and 4.4).
Indeed, the least squares fitting of the dihedral angles refined by the molecular
mechanics calculations to the experimentally measured vicinal proton-carbon
heteronuclear coupling constants gave a much lower R2 value than that of the
dihedral angles refined by semi-empirical calculations and the measured
coupling constants (Table 4.10). The least squares fitting of the dihedral angles
refined by semi-empirical geometry optimizations with the experimentally
measured vicinal proton-carbon heteronuclear coupling constants gave the
highest R2 value. Table 4.10 gives the coefficients of the Karplus-type equation
from least squares fitting. For each case (e.g. AM1), three equations were
obtained. These were for the 0 dihedral angle alone (AM1-0), for the w dihedral
angle alone (AM1-1y), and for both 0 and 1); angles (AM1-0w). It was clear that
the equation for the 0 dihedral angle alone (e.g. AM1-0) predicts a higher value of
the coupling constant than that for the wdihedral angle alone (AM1-1y). This
confirms our expectation that different equations are required for the 0 and w

dihedral angles. On the other hand, dividing our database into two reduces the

118

Table 4.6. Structures of 1,6-anhydroglucose (8) from Cambridge Structural
Database and from MM, AM1 and PM3 geometry optimizations. The relative
energies are from PM3 single point energy calculation of the crystal structures
and the optimized structures using MM, AM1 or PM3 geometry optimizations.
Ref code stands for the reference codes given in the Cambridge Structural

Database.

 

 

 

 

 

 

 

 

 

 

Compound H1CtOGCB H5CSOSC1 H6pCGOGC1 Ref Relative E

code (kcaI/mol)
1,6-Glcanh 140.90 164.03 -115.90 Ahglpy01 9.61
(8) 144.42 161.14 -106.67 Ahglpy10 2.39
MM 147.2 162.9 -1 18.8 Ahglpy10 0.78
AM1 144.7 162.9 -122.2 Ahglpy10 0.17
PM3 146.8 160.9 -127.2 Ahglpy10 0.0
MM 146.6 163.2 -1 17.5 Ahglpy01 0.81
AM1 146.5 161.8 -123.5 Ahglpy01 0.18
PM3 146.8 160.9 -127.1 Ahglpy01 0.0

 

 

 

 

 

 

 

119

Table 4.7. Stmctures of 1,6-anhydrogalactose (9) from Cambridge Structural
Database and from MM, AM1 and PM3 geometry optimizations. The relative
energies are from PM3 single point energy calculation of the crystal structures
and the optimized structures using MM, AM1 or PM3 geometry optimizations.

Ref code stands for the reference codes given in the Cambridge Structural

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Database.
Relative E
Compound H1 C1 06C6 H5CSOSC1 H6pC606Ct Refcode (kcal/mol)
1 ,6- 120.13 -173.36 -88.29 Ahgalp3 4.26
Galanh 133.68 156.92 -1 10.09 Ahgalp2 7.75
(9) 122.86 163.13 -108.84 Ahgalp1 5.39
MM 149.2 162.1 -122.5 Ahgalp1 0.84
AM1 143.9 161.7 -120.7 Ahgalp1 0.16
PM3 148.4 160.5 -129.4 Ahgalp1 0.0
MM 148.3 162.4 -120.8 Ahgalp2 0.83
AM1 150.6 158.7 -130.4 Ahgalp2 0.27
PM3 147.2 161.3 -127.2 Ahgalp2 0.18
MM 148.3 162.6 -120.7 Ahgalp3 0.86
AM1 143.9 161.8 -120.6 Ahgalp3 0.15
PM3 148.1 160.7 -129.3 Ahgalp3 0.0

 

 

 

 

 

 

120

 

Table 4.8. Structures of 1,6-anhydromannose (10) from Cambridge Structural
Database and from MM, AM1 and PM3 geometry optimizations. The relative
energies are from PM3 single point energy calculation of the crystal structures
and the optimized structures using MM, AM1 or PM3 geometry optimizations.

Ref code stands for the reference codes given in the Cambridge Structural

 

 

 

 

 

 

 

 

 

 

Database.
Compound H1 C1 0606 H5050501 H6pCSOGC1 Ref Relative E
code (kcaI/mol)
1,6-Mananh 142.51 165.68 -118.35 Biftuf1 6.91
(10) 144.68 161.17 -121.26 BiftufZ 6.65
MM 147.0 162.9 -117.7 Biftuf1 0.86
AM1 146.7 160.5 -124.6 Biftuf1 0.18
PM3 146.9 161.0 -126.6 Biftuf1 0.0
MM 147.2 162.3 -118.6 BiftufZ 0.71
AM1 146.4 161 .0 -124.2 Biftuf2 0.19
PM3 145.9 161.8 -124.8 Biftuf2 0.1

 

 

 

 

 

 

121

 

Table 4.9. The values of the dihedral angle C2-C3-C4-CS as a measure of the
degree of flattening of the chair conformations of the three 1,6-anhydro sugars,
1,6-anhydroglucose (8), 1,6-anhydrogalactose (9), and 1,6-anhydromannose (10)
from Cambridge Structural Database and from MM, AM1 and PM3 geometry
optimizations. The relative energies are from PM3 single point energy
calculation of the crystal structures and the optimized structures using MM, AM1

or PM3 geometry optimizations. Refcode stands for the reference codes given in

the Cambridge Structural Database.

 

 

 

 

 

 

 

 

 

REFCODE MM AM1 PM3 Crystal Relative
structure energy
(kcanol)

AHGLPY10 (8) -47.2 -34.7 -21.9 -35.1 0.0
AHGLPY01 (8) -47.9 -36.0 -22.7 -35.2 7.22
AHGALP1 (9) -46.3 -35.2 -31.9 -39.8 1.13
AHGALP2 (9) -47.0 -35.3 -37.0 -38.9 3.49
AHGALP3 (9) -47.4 -35.1 -32.3 -39.1 0.0
BIFTUF1 (10) -47.7 -35.5 -28.4 -41.0 0.26
BIFTUF2 (10) -46.0 -37.7 -31.5 -37.9 0.0

 

 

 

 

 

 

122

 

accuracy. So, at this stage we recommend that a single equation (equation 6,
AM1-0w or equation 4, AM1*) be used until the accuracy of the individual
equations (e.g. AM1-0 and AM1-w) is confirmed with more data points. Since we
don’t have data points around 0. degrees, to check how different the equations
would be if a data point around 0 degrees was present, we added one more data
point to our 28 data points. That data point is 5.6 Hz for 11 degrees.2 The
equations thus obtained are termed MM", AM1*, and PM3* in Table 4.10.

There were two crystal structures of 1,6-anhydroglucose (8) and two for
that of 1,6-anhydromannose (10) and three for 1,6-anhydrogalactose (9) in the
Cambridge Structural Database (Tables 4.6 - 4.9). The structures were similar
except for the dihedral angles involving CG that showed some variations. A
single point energy calculation of these crystal structures revealed that some of
the structures differ by as much as 7.2 kcanol. So in solution these molecules
may exist as an ensemble average of different conformations. The two crystal
structures of 1,6-anhydroglucose (8) differ by 7.2 kcal/mol and this corresponded
to compound 8 spending 99.9995 % of the time in the lower energy conformer.
And the two crystal structures of 1,6-anhydrogalactose (9) differ by energies of
3.49 and 1.13 kcanol from the third and lowest energy conformer and this
corresponded to populations of 0.23 %, 12.81 %, and 86.96 °/o for the three
crystal structures. The two crystal structures of 1,6-anhydromannose (10) differ
by energy of 0.26 kcal/mol and this means compound 8 would spend 39.16 % of

the time in one conformer and 60.842 % in the other conformer.

123

Figure 4.2. The new data points plotted against the Karplus-type curves of the
previous equations. Our data points are from the experimental coupling
constants and from AM1, PM3, and MM geometry optimizations, respectively.

This figure is a color presentation for clarity.

124

 

J H-C (Hz)

Curves of the three old equations
with the new data points

 

10

 

l - Tvaroska eqn. ‘
£ - Mulloy eqn. ,
l . AM1 datapoints ii
> . PM3 datapoints l!
l . MM datapoints l‘
% - Cloran eqn ‘

 

 

 

 

 

Torsion

 

 

125

Flgure 4.3. The new AM1 data points plotted against the Karplus-type curves
of the previous equations and the new equations. This figure is a color

presentation for clarity.

126

 

Curves of new equations (AM1) and the two
old equations with the new data points

 

 

  

 

- Tvaroska eqn.
W - Mulloy eqn.
1 o AM1-psi datapoints
1. - AM1* eqn
* o AM1-phi datapoints
ll - AM1-phi eqn.
'1 + AM1-psi eqn
y - AM1-phi-psi eqn

 

 

 

90
Torsion

127

135

180

Figure 4.4. The new PM3 data points plotted against the Karplus-type curves
of the previous equations and the new equations. This figure is a color

presentation for clarity.

128

 

Curves of new equations (PM3) and the two
old equations with the new data points

 

O

’

 

l- Tvaroska eqn. ‘
. ’ - Mulloy eqn. ll

1

l ~ PM3‘ eqn. .1
l o PM3-psi datapoints l
H o PM3-phi datapoints"
.1 - PM3-phi eqn. 1
‘3 + PM3-psi eqn.

- PM3-phi-psiveqn.

 

 

 

1 35 1 80
Torsion

 

 

 

129

 

l

Difference in predicted J values ‘
between the newequations and the
two previous equations

:jfill “ l m

0.7
l 0.6
0.5
0.4
0.3‘
0.2 ~'
0.1

  
 
    

 

l
l
l
l
l
i

Mi
p ‘IAM1'-Mulloy l
‘ PM3*-Tvaroska H
_x PM3‘-Mulloy l

Difference in J

p 150 180;
‘ Torsibn

l
l
l
l
' “(£1111 l
l
l
l
i

Q

Flgure 4.5. The difference in the predicted values of the vicinal coupling
constants between the new Karplus-type equations (AM1' and PM3*) and the
two previous equations (T varoska et. al. and Mulloy et. al.). This figure is a color

presentation for clarity.

130

Table 4.10. The values of the parameters a, b, c of the equation 3JHc (9) =
a Cos2 (6) + b Cos (9) + c and the value of R2 from a least squares fitting of the
heteronuclear coupling constants and the dihedral angles given in Tables 4.4 and

4.5. Those with better than 98.5 % R2 values are highlighted.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

a b c R2
MM-d> 2.639 -1.879 2.184 97.7
MM-‘I’ 2.270 -2.221 2.059 97.1
MM-<1>‘I’ 2.380 -2.128 2.102 97.0
PM3-(I) 6.395 0.375 0.943 98.0
PM3-‘1’ 5.793 -0.120 0.787 98.2
PM3-(I)? 5.953 -0.051 0.846 97.8
AM1-(I) 6.672 0.41 3 0.524 99.2
AM1-‘1‘ 5.926 -0.371 0.504 99.1
AM1-(NP 6.020 -0.193 0.573 98.7
MM * 4.004 -1.280 1.513 93.7
AM1* 5.597 -0.421 0.722 98.6
PM3* 5.335 -0.414 1 .020 97.7

 

 

 

 

 

131

 

Geometry optimizations starting from each of the crystal structures of 1,6-
anhydroglucose (8), 1,6-anhydrogalactose (9), and 1,6-anhydromannose (10),
using molecular mechanics and AM1 and PM3 semi-empirical calculations, gave
structures which are lower in energy than any of the crystal structures and which
are consistent between the different calculation methods. These structures are
also consistent with structures obtained from AM1 and PM3 geometry
optimizations starting from built structures (as opposed to starting from crystal
geometries). But the MM geometry optimization starting from built structures
gave structures relatively not consistent with any of the structures. The main
difference is on the dihedral angles involving C6 and the extent of flattening of
their chair conformation. While AM1 geometry optimization consistently gave the
same chair conformation as that of the crystal structures, PM3 geometry
optimizations consistently flatten the chair by about 9 degrees and the MM
optimizations do the opposite of PM3 by about 9 degrees (Table 4.9). These
results suggest that the refinement of the dihedral angles of the model
compounds using the AM1 semi-empirical methods give reliable results.

Previous studies on AM1 and PM3 calculations have shown that the PM3
semi-empirical calculations can lead to physically incorrect results, for example in
carbohydrates, because of distortions of the potential energy surface (relative
energy versus H---O distance in the intra- or intermolecular interactions)."3'16
The authors suggest that the PM3 method be used with care. Our own studies
on the model compounds 1 - 10 using the PM3 and AMtsemi-empirical

calculations revealed that the values of the vicinal heteronuclear coupling

132

constants for similar dihedral angles are more spread for the PM3 results than
the results from AM1 calculations. Furthermore, PM3 optimizations tend to
flatten the chair conformation of the 1,6-anhydro sugars (8-10) (Table 4.9). This
was also apparent in the higher R2 value for the AM1 results (equatlon 4: 901.0
(0) = 5.597 Cos2 (9) — 0.421 Cos (9) + 0.722) than the PM3 results (equation 5:
3JHC (0) = 5.335 Cos2 (8) — 0.414 Cos (8) + 1.020) (Table 4.10). As a result, we
suggest that the equation (equation 4 Le. 3Jim: (6) = 5.597 Cos2 (0) - 0.421 Cos
(9) + 0.722) obtained from the least squares fitting of the dihedral angles refined
by AM1 semi-empirical geometry optimizations to the experimentally measured
vicinal proton-carbon coupling constants, 3JHc, be used for predicting the dihedral
angles defined by the four atoms H-C-O-C (0 and w for carbohydrates) from

vicinal proton-carbon coupling constants, 3JHC, or vice versa.

Conclusions

The three dimensional structures of the synthetic model compounds 1-7
and the commercially available compounds 8-10 have been refined by Molecular
Mechanics and AM1 and PM3 Semi-empirical calculations. The vicinal proton-
carbon coupling constants, 3JHc, have been measured in solution by two-
dimensional excitation-sculptured indirect-detection experiment (EXSIDE) NMR
spectros00py. From the least squares fitting of the measured coupling constants
and the glycosidic dihedral angles of the rigid molecules, equation of the form:

3Jltc (0) = a Cos2 (8) + b Cos (9) + c, are proposed where 9 stands for the

dihedral angles defined by the four atoms H-C-O-C (0 and w for carbohydrates).

133

This new Karplus-Type equations relating the vicinal proton-carbon coupling
constants, 3.1m, to the glycosidic dihedral angles 0 (H1'-C1'-OX-CX) and \u (HX-
CX-OX-C1'), respectively have been compared to the previous equations and

shown to give a higher R2 value.
Acknowledgements

This work was supported by the Michigan State University Research

Excellence Fund.

134

References

10.

11.

12.

13.

Tvaroska, l; Hricovini, M; Petrakova, E. Carbohydr. Res. 1989, 189, 359.
Mulloy, B.; Frenkiel, T. A.; Davies, D. B. Carbohydr. Res. 1988, 184, 39.
Rundolf, T.; Kjellberg, A.; Damberg, C.; Nishida, T.; Widmalm, G. Magn.
Reson. Chem. 1998, 36, 839.

Cloran, F.; Carmichael, l.; Serianni, A. S. J. Am. Chem. Soc. 1999, 121,
9843.

Bose, B.; Zhao, 8.; Stenutz, R.; Cloran, F.; Bondo, P. B.; Bondo, G.;
Hertz, B.; Carmichael, l.; Serianni, S. J. Am. Chem. Soc. 1998, 120,
11158.

Haasnoot, C. A., De Leeuw, F. A., & Altona, C. Tetrahedron 1980, 36,
2783.

Ernst, R., Bodenhansen, G., & Wokahun, A. Principles of Nuclear
Magnetic Resonance in One and Two Dimension, 1987, pp 431-440,
Oxford University Press, Oxford.

Bax, A. & Davis, D. G. J. Magn. Reson. 1985, 65, 355.

lax, A. & Subramanian, S. J. Magn, Reson. 1986, 67, 565.

Titman, J., Neuhaus, D., & Keeler, J. J. Magn. Reson. 1989, 85, 111.
Sangers, J. K., & Hunter, B.K. Modem NMR Spectroscopy, 1993, pp 193-
197, Oxford University Press, Oxford.

Krishnamurthy, V. V. J. Magn. Reson. 1996, 121, 33.

Dewar, M. J. S.; Zoebish, E. G.; Healy, E. F.; Stewart, J. J. P. J. Am.

Chem. Soc. 1985, 107, 3902.

135

14. Stewart, J. J. P. J. Comp. Chem. 1989, 10, 221.
15. Csonka, G. l.; Angyan, J. G. J. Mol. Struct. (Theochem)1997, 39.3, 31.

16. Kallies, B; Mitzner, R. J. Mol. Model. 1995, 1,68.

136

Chapter 5

Studies on the Solution Conformations and Dynamics of
Maltose, Reduced Maltotetraose, and the Potent

Glycohydrolase Inhibitor Acarbose

137

Abstract.

We applied the rule-based method (Chapter Two) and the new Karplus-
type equations (Chapters Three and Four) that we have developed in this work in
combination with existing methods using multi-nuclear, multidimensional NMR
spectrosc0py, molecular mechanics calculations and molecular dynamics
simulations to study the solution conformations and dynamics of a family of
01-1->4 linked oligosaccharides namely, the disaccharide maltose and the
tetrasaccharide reduced-maltotetraose, and the pseudo-tetrasaccharide
acarbose, which is a potent glycohydrolase inhibitor. The study revealed that all
three oligosaccharides occupy the RR region of the (1)-W space as predicted by
the rule-based method. Also we found out that the pseudotetrasaccharide
acarbose is more flexible than maltose and reduced maltotetraose as revealed

from the NMR spectroscopy and molecular dynamics studies.

138

Introduction.

Carbohydrates provide diverse three-dimensional (3D) structural
possibilities and have several applications. For example, recent studies have
demonstrated that oligosaccharides are involved in a number of recognition
events such as cell adhesion, metastasis, fertilization and embryonic
development, amongst others."2 In view of their diverse function in a variety of
biological systems, the primary structures and 3D structures of carbohydrates are
of great interest. Due to their diversity and the inability to crystallize such
conjugates, nuclear magnetic resonance (NMR) spectrosc0py, in combination
with molecular mechanics calculations and molecular dynamics studies, is a
method of choice for their identification and conformational analysis.

A major problem when using NMR spectroscopy to study carbohydrates is
that, except for the anomeric protons, all the protons on oxygenated atoms (e.g.,
those on carbons 2-6 in aldohexoses) appear in a range of ~3.3 ppm to ~53
ppm. Hence, as the number of monosaccharide units increases, the assignment
of peaks for each proton (and carbon) becomes increasingly difficult. This
means extracting coupling constants and NOESY volumes--which are useful for
determining dihedral angles and distances, respectively-becomes very difficult.

In chapter two we have collected and analyzed the crystal structures of
oligosaccharides and proposed a simple and accurate rule based method for
predicting the conformations of oligo and polysaccharides.3 Because solution
conformations of carbohydrates may not be identical to their crystal stmcture, it

would be very useful to come out with a rule-based method for predicting

139

Figure 5.1 Schematic diagrams of maltose, reduced maltotetraose, and

acarbose.

140

 

 

HO

HO

8

 

Maltose

OH
0
OH
OH 0
HO OH
OH
b 0 0
H
0 OH
Reduced Maltotetraose
OH
OH
NH O
HO OH
OH
b 0 O
H
OH
0
Acarbose

141

HO

HO

OH
OH

OH

d OH

OH

OH

(1 OH

carbohydrate conformations in solution by analyzing a library of solution
conformations of carbohydrates. In an effort to build a database of solution
conformations of carbohydrates and to see how the rule-based method and the
new Karplus-type equations compare to existing methods and equations we
decided to study the solution conformation and dynamics of maltose, reduced
maltotetraose, and the glycohydrolase inhibitor acarbose by a combination of
nuclear magnetic resonance (NMR) spectroscopy, molecular mechanics

calculations and molecular dynamics simulations.

Material and Methods

Ollgosaccharldes

The commercially available maltose is used as it is. The reduced
maltotetraose was obtained by reduction of the commercially available
maltotetraose with sodium borohydride in methanol. Acarbose was generously
supplied by Professor Jack Preiss, Department of Biochemistry, Michigan State

University.

Nuclear Magnetic Resonance (NMR) Spectroscopy

Nuclear Magnetic Resonance spectra of the three oligosaccharides were
obtained in deuterated water. Spectra were recorded at 30 0C. The residual
solvent peaks were used as reference. Combinations of homonuclear- and

heteronuclear- (one- and) two-dimensional NMR experiments were performed in

142

order to allow the assignments of the signals for the three oligosaccharides and
to measure the coupling constants. This included: double quantum filtered
J-correlated spectroscopy (DQF-COSY) spectra (Ernst et al., 1987),4 total
correlated spectroscopy (TOCSY) spectra (Bax & Davis, 1985), 5 heteronuclear
multi-quantum coherence (HMQC) experiment (Bax & Subramanian, 1986),6 and
nuclear Overhauser effect (NOESY) experiment (Sangers et al., 1993).7

Analysis of the spin systems for the three oligosaccharides from the 1D
NMR spectra and from DOF-COSY and TOCSY spectra gave the connectivities.
The tables of assignment of the proton and carbon chemical shifts of the three
oligosaccharides are given in Tables 5.1 - 5.3. Once the assignments were
made NOESY experiments were performed to determine the various inter-
nuclear distances (Table 5.4 and 5.5). Spectra of the NOESY experiments were
acquired at different mixing times (300, 400, and 500 ms) to address the potential
problem of spin diffusion. The vicinal proton-carbon coupling constants were
obtained from excitation-sculptured indirect-detection (EXSIDE) experiments
(Krishnamurthy et. al., 1996).8 The EXSIDE spectra were measured on Varian
lnova 600 MHz NMR spectrometer operating at 600 MHz for protons and 150
MHz for carbons. And the rest of the one- and two-dimensional NMR spectra
were measured on Varian VXR 500 MHz NMR spectrometer operating at 500

MHz for protons and 125 MHz for carbons.

143

Table 5.1. 1H and 13C chemical shift assignments (in ppm) of maltose.

 

 

 

 

 

 

 

 

 

H/C Residue a Residue b
1 5.32 99.1 4.56 95.3
2 3.49 72.3 3.16 73.7
3 3.61 72.6 3.67 75.8
4 3.33 69.1 3.51 76.5
5 3.63 71 .4 3.54 74.3
6 3.76 60.1 3.82 60.4
6 3.65 60.1 3.66 60.4

 

 

 

 

 

 

 

144

Table 5.2. 1H and 13C chemical shift assignments (in ppm) of reduced

 

maltotetraose.

 

H/C Residue a Residue b Residue c Residue d

 

1 5.38 101.6 5.38 101.6 5.12 102.2 3.64 62.2

 

2 3.60 73.3 3.60 73.3 3.60 73.3 3.65 64.3

 

3 3.97 74.7 3.97 74.7 3.97 74.7 3.67 64.5

 

4 3.41 71.2 3.65 78.8 3.65 78.8 3.88 83.7

 

5 3.71 74.6 3.69 74.6 3.66 74.6 3.71 63.9

 

6 3.84 62.2 3.84 62.1 3.84 62.2 3.64 62.2

 

 

6 3.65 62.2 3.64 62.1 3.64 62.2 3.64 62.2

 

 

 

 

 

 

 

 

 

 

145

Table 5.3. 1H and 13C chemical shift assignments (in ppm) of acarbose.

 

 

 

 

 

 

 

 

 

 

 

H/C Residue a Residue b Residue c Residue d

1 3.52 57.1 5.27 101 5.35 100 4.61 96.6 5.19 92.7
2 3.77 73.8 3.56 71.6 3.53 72.2 3.23 75.0 3.51 71.6
3 3.72 73.9 3.55 73.7 3.93 74.3 3.73 77.3 3.89 74.3
4 4.00 72.1 2.46 65.7 3.60 78.0 3.59 77.9 3.62 78.1
5 -- 140 3.74 70.4 3.80 72.1 3.53 75.5 3.90 71.0
6 4.18 62.8 1.30 47.3 3.92 61.6 3.86 61.6 3.88 61.6
6 4.08 62.8 -- 47.3 3.83 61.6 3.78 61.6 3.75 61.6
7 5.86 124

 

 

 

 

 

 

 

 

 

 

 

146

 

Table 5.4. Distances calculated from NOESY experiment for reduced
maltotetraose. The letters A, B, C, and D in column 1 refer to the different
residues and the numbers refer to the protons (e.g. A1 -B4 refer to proton 1 of

residue A with proton 4 of residue B).

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Atom pairs NOESY Distances in A
Volume
A1 -B4 79.2 2.37
B1 -C4 81 .2 2.35
C1 -D4 100 2.28
B1 -C6 31 .2 2.76
A1 -B6 22.9 2.91
C1 -06 18.7 3.00
A2-A4 37.5 2.67
A4-A5 16.7 3.06
A4-A6 10.4 3.31
A3-A5 16.7 2.70

 

147

 

Table 5.5. Distances calculated from NOESY experiment for acarbose. The
letters A, B, C, and D in column 1 refer to the different residues and the numbers

refer to the protons (e.g. A1 -B4 refer to proton 1 of residue A with proton 4 of

 

 

 

 

 

 

 

 

 

 

 

 

residue B).
Atom pairs NOESY Distances in A
Volume
A6-A7 3 3.54
A1 -A7 18 2.63
A6’-A7 5 3.25
A4-A7 3 3.54
C1 -D4 61 2.15
B1 -C4 49 2.22
A7-36 6 3.16
A7-B4 1 4.25
82-34 1 6 2.68
A1 -B4 1 7 2.65

 

 

 

 

 

148

Molecular Mechanics Calculations and Molecular Dynamics Simulations
Molecular mechanics calculations and molecular dynamics simulations
were performed using the DRElDlNG force fields (Mayo et al., 1990)9
implemented in the BIOGRAF10 (Molecular Simulations lnc., Waltham, MA
02154) program. The Drieding force field includes a harmonic term for the bond
distortions, a simple harmonic function for the valence angle term, an improper
torsion term to evaluate inversions about atomic centers, a cosine expansion
torsional term, a Coulombic potential for evaluating electrostatic energy, a
Leonard-Jones 12-6 potential for van der Waals contributions and a Leonard-
Jones 1210 potential for evaluating hydrogen bonding contributions.
Minimizations were performed using the conjugate gradient method. The default
parameters given in this program for carbohydrate rings were used without
modification since they have been validated earfier (Wang et. al. 1996).11 The
calculated distance constraints, from NOESY experiments, were included as
harmonic restraints with a force constant of 10,000 kcal per mol per A2 that was
gradually relaxed to 1000 and 10 kcal per mol per A2. Grid search studies were
performed using sequential search mode in the Biograf program, with a 5
degrees step growth search from 0 to 360 degrees, for each defined dihedral

angle. Figures 5.2 and 5.3 are color presentations for clarity

Results and Discussion

The three oligosaccharides namely, maltose, reduced maltotetraose and

acarbose, all have 0t-1->4 linkages which is the linkage type of the

149

 

polysaccharide amylose. While the disaccharide maltose has one 01-1->4
linkage, the tetrasaccharides maltotetraose and acarbose each have three
0t-1->4 linkages. Maltose and maltotetraose are the building blocks of the

polysaccharide amylose. Acarbose is a potent inhibitor of a-glucosidases. It is
used for treatment of diabetes, hyperlipidaemia and obesity and is marketed as
Glucobay.

The potential energy (PE) surfaces of the three oligosaccharides in this
study were calculated using the Dreiding9 force field in the Biograf10 molecular
mechanics (MM) program. The PE surfaces provide a first estimation of the
conformational regions that are energetically accessible. While two minima
(Figure 5.2) were found for maltose and maltotetraose (the second minimum 9
kcanol above the first minimum), the acarviosine (a-b) unit of acarbose had two
minima with 1 kcal/moi energy difference and a third minimum that was at 6
kcal/moi above the lowest minimum. This indicated that while maltose and
maltotetraose should have a reasonably rigid conformation, the a-b unit of
acarbose should be more flexible because of the fact that its residue-a at the
non-reducing end is not actually a monosaccharide. The isoenergy contours in
Figure 5.2 were drawn with interpolation of 1 kcanol above the minimum.

The next task was to perform combinations of homonuclear and
heteronuclear, one- and two-dimensional NMR experiments in order to allow the
assignments of the signals for the three oligosaccharides and to extract inter-
nuclear distances. This included performing double quantum filtered J-

correlated spectroscopy (DQF-COSY) spectra,4 total correlated spectroscopy

150

Figure 5.2 Rigid residue potential energy (PE) surfaces of maltose (top) and
acawiosine (the a-b unit of acarbose) (bottom). lsoenergy contours are drawn
with interpolation of 1 kcal/moi above the minimum. Notice that while maltose
has practically a single minimum, acarviosine has two minima with 1kcal/mol
PE difference and a third minimum at 6 kcal/moi higher. The dihedral angles
of maltose are defined as (O5'-C1'-O4-C4) for 0 and (C1'-O4-C4-C5) for w.

The dihedral angles acarviosine are defined as (C7'-C1'-N4-C4) for 0 and

(C1'-N4-C4-05) for 1)]. This figure is a color presentation for clarity.

151

 

152

Figure 5.3 Solution conformations of maltose (top), reduced maltotetraose
(middle) and acarbose (bottom) as determined from constrained minimization
using distance constraints from 2D NOESY experiments. This figure is a color

presentation for clarity.

153

 

llaOSE

. coll

 

154

(TOCSY) spectra,5 heteronuclear multi-quantum coherence (HMQC)
experiment,6 and two-dimensional nuclear Overhauser effect (n.O.e.) experiment
(NOESY) .7

Analysis of the spin systems for the glycoside ring systems of the three
oligosaccharides with 1-D traces from DQF-COSY and TOCSY spectra gave the
connectivities. Once the assignments were made (Table 5.1-5.3), n.O.e.
experiments were performed to determine the various inter-nuclear distances.
Spectra of the n.O.e. experiments were acquired at different mixing times to
address the potential problem of spin diffusion. Assuming isotropic motion then,
the cross-relaxation rates and thus inter-nuclear distances between nuclei i and j,
n], for each pair wise interaction were extracted (Tables 545.5) from the volume
of the NOESY peak, Vii, using the relation rij = r... (V... / Vii)”3 where, r =
distance, V = volume, ref: reference pair of nuclei the distance between which is
fixed and known.

This was followed by constrained minimization using the inter-nuclear
distances obtained at the previous step as constraints in the force fields of
molecular mechanics calculations (the Drieding force fields9 implemented in the
BIOGRAF program) to determine the average conformation about glycosidic
linkages of these oligosaccharides in solution. For maltose the distance between
H1’ and H4 was 2.35 A. The constraint-minimized solution conformations of
maltose, reduced maltotetraose, and acarbose though similar were not identical

as shown in Flgure 5.3 and Table 5.6. Notice that the two tetrasaccharides,

155

 

 

 

 

 

 

 

 

 

 

Figure 5.4 The molecular dynamics population map of maltose, shown above, is
consistent with the PE surface. The dihedral angles are defined as (05'-C1'-O4-

C4) for 0 and (C1'-O4-C4-C5) for w.

156

Figure 5.5 Molecular dynamics population maps of reduced maltotetraose
from 200 ps simulation starting from its constraint-minimized solution structure.
The plots are for the glycosidic linkages a-b (top), b-c (middle) and c-d
(bottom). The x-axis is the 0 dihedral angle (05’-C1’-O4-C4) and the y-axis is

the v dihedral angle (C1’-O4-C4-CS), both plotted from -180 to 180°.

157

 

180 ,-_ -, .4 n f -.-_.,.1.-- ‘21,-

psi-a-b
O
l
l
l

 

-180 l
-180 0 180

180 s~- v~~+r~w~mds H ”fiz-taq'r'““‘ "‘ i A

alracse

UClll'E

 

psi-b-c
0

nd 54

84555 ‘ -180 L __ _ z,

 

180 ; ~~*- ~ ~art~—mn-«~

psi-c-d
O

 

158

Figure 5.6 Molecular dynamics Trajectories of the 0 angle of reduced
maltotetraose from 200 ps simulation starting from its constraint-minimized
solution structure. The plots are for the glycosidic linkages a-b (top), b-c
(middle) and c-d (bottom). The x-axis is time in ps and the y-axis is the 0

dihedral angle (05’-C1’-O4-C4) plotted from -180 to 180°.

159

 

 

 

 

 

 

1 e
l 2
! a
l
educed . 3
l s
.. a.
llmlzel l
pill ‘
;[hel }

 

 

3
l E
l o.
l
180 W , , 1..- ‘_ fig
0 100 200 l

Figure 5.7 Molecular dynamics Trajectories of the w angle of reduced
maltotetraose from 200 ps simulation starting from its constraint-minimized
solution structure. The plots are for the glycosidic linkages a-b (top), b-c
(middle) and c-d (bottom). The x-axis is time in ps from 0 to 200 ps and the y-

axis is the w dihedral angle (C1’-O4-C4-C5) plotted from -180 to 180°.

161

 

 

 

 

 

 

 

 

162

 

 

 

 

 

reduced maltotetraose and acarbose, showed the beginning of the helical type
conformation that is known for amylose.

Because solution conformations are ensemble averages of many
conformations, molecular dynamics simulations were performed in order to gain
insight into the dynamics of the structures of the oligosaccharides in solution.
Starting from the constraint-minimized solution conformations of the three
oligosaccharides, 200 ps molecular dynamics simulations (canonical NVT) were
performed using the Biograf molecular mechanics program.10 The results of the
molecular dynamics simulations are shown in Figures 5.4-5.10. The molecular
dynamics population maps of all the three oligosaccharides (Figures 5.4, 5.5,
and 5.8) were consistent with the potential energy surfaces (Figure 5.2). Notice
that in Figures 5.2 and 5.4 the dihedral angles are plotted from 0 to 360 degrees
and not —180 to 180 degrees and as a result the areas populated by the
oligosaccharides may look different from the other population maps that are
drawn from —180 to 180 degrees, but it is just a shift of 180 degrees in both
dihedral angles.

The molecular dynamics simulation studies revealed that, as the study of
the potential energy surface indicated, the acarviosine (a-b) unit of acarbose
showed significantly greater flexibility than all the other linkages (Figure 5.4,
5.10). In fact, the b-c and c-d units of acarbose were also slightly more flexible
than the linkages of maltose and reduced maltotetraose and this may be
associated with the greater flexibility of the a-b unit of acarbose. The molecular

dynamics trajectories of the 0 and w angles of the three oligosaccharides

163

Figure 5.8 Molecular dynamics population maps of acarbose from 200 ps
simulation starting from its constraint-minimized solution structure. The plots
are for the glycosidic linkages a-b (top), b-c (middle) and c-d (bottom). The x-
axis is the 0 dihedral angle (05,C7’-C1’-O4,N4-C4) and the y-axis is the w

dihedral angle (C1’-O4,N4-C4-C5), both plotted from -180 to 180°.

164

 

 

 

Psi a-b

 

 

   
 

l 9 ‘
.2 0 l--- 22,--- - ,
tn ‘ ‘ 1
a O
. ‘ . . 1
. .Q.‘ . ‘l
l 1,38 l
l .2 jwol. :l. ‘
'180 o 0 o;;::;‘ ‘
-180 180 I
phi-b—c ‘

.80 ~wr~~t~ .W,,.-. l

".3
l .2 0 l
(0
l ‘* ~
g l
l‘ A A
-180 L A. 1
-180 l

 

165

Figure 5.9 Molecular dynamics Trajectories of the 0 angle of acarbose from
200 ps simulation starting from its constraint-minimized solution structure. The
plots are for the glycosidic linkages a-b (top). b-c (middle) and c-d (bottom).

The x-axis is time in ps from 0 to 200 ps and the y-axis is the 0 dihedral angle

(C7’,05’-C1’-N4,04-C4) plotted from —180 to 180°.

166

)se llOl'l

bolllil

lal all?

 

  

 

 

 

  

11

l .

-180 LL—_u ‘.__., .. H.
0 100

“"190,” l

‘1
. i l l
-180 -.jeuMe W is- ~~~—‘—~ ‘
0 100 200 l
time (pa) 1

167

Figure 5.10 Molecular dynamics Trajectories of the W angle of acarbose from
200 ps simulation starting from its constraint-minimized solution structure. The
plots are for the glycosidic linkages a-b (top). b-c (middle) and c-d (bottom).

The x-axis is time in ps from 0 to 200 ps and the y-axis is the \l’ dihedral angle

(C1’-N4,04-C4-CS) plotted from -180 to 180°.

168

 

psl-a-b

 

 

 

sellol ‘
re. Tit? l
ottoml l

l antie

 

 

 

 

 

169

(Figures 5.6, 5.7, 5.9, ad 5.10) showed the variation of the dihedral angles
against time as the simulation progresses. It was evident that acarbose showed
frequent transitions between different conformations as compared to reduced
maltotetraose. Comparison of the solution conformations, the grid-search
minimum energy conformations, and the crystal structures of the three
oligosaccharides (Table 5.6) revealed that the structures are reasonably
consistent with each other. Two conformations of the a-b unit of acarbose have
been inferred to exist at neutral, acidic and alkaline pHs.‘2'13 This is consistent
with our studies as shown in Figures 5.2 and 5.9. The rule based prediction of
the dihedral angles of an a—1->4 linkage in oligosaccharides (Chapter Two) gave
98, -128 degrees for 0 and v.3 This is consistent with conformations determined
by NMR spectroscopy and molecular mechanics calculations and molecular
dynamics simulations and structures from X-ray crystallography (Table 5.6).

An important piece of structural information comes from the calculation of
dihedral angles from vicinal proton-carbon coupling constants measured from
NMR spectroscopy, or vice versa. To calculate the dihedral angles (9) between
the glycoside residues from the vicinal proton-carbon coupling constants, 3JHC,

““6 and the new (Chapter Four) Karplus-type

we used the three previous
equations which have the general form 3J0” = a cos2 (0) + b Cos (0) + c. The
3qu values measured from 2D EXSIDE experiments for the dihedral angles H1 b-
C1b-O4c-C4c and H1c-C1c-O4d-C4d of acarbose were 2.60 and 3.60 Hz,

respectively. The corresponding value for maltose was 4.88 Hz. These values

170

 

Table 5.6. Values of the glycosidic dihedral angles of the three oligosaccharides

from constrained minimization, grid search studies and from crystal structure.

 

 

 

 

Molecule Linkage Constrained Grid Crystal Structure
Minimization Search

Maltose a-b 120.56, -105.83 100, -145 124, -1088

Reduced a-b 124.34, -104.20 100, -145

maltotetraose b-c 123.27, -103.06 100, -145
c-d 148.49, -114.70 100, -145

Acarbose a-b 156.71, -120.61 160, -130 33,-151b(104,-113)°
b-c 148.86, -103.48 100, -145 110,-120 (91 ,-180)
c-d 137.38, -101.33 100, -145 121,-107 (111,-117)

 

 

 

 

 

 

 

a Ref. 17. b Ref. 18. ° PDB database tGAH.

171

Table 5.7. Comparison of the predictions of the new Karplus-type equations

l.15 and

(Chapter Four) with that of the three previous equations of Mulloy et. a
Tvaroska et. al. ‘4 and Cloran et. al.16 with the experimentally measured values
of the vicinal coupling constant, 3JHC, in Hz. The results are averages of the 2000
data points from molecular dynamics simulations that were performed starting

from constraint-minimized solution structures.

 

 

 

 

 

 

 

 

 

 

Maltose Acarbose
0 a-b 0 b-c 0 c-d
Mulloy et. al.15 4.39 2.33 3.21
Tvaroska et. al."4 4.54 2.38 3.30
Cloran ET. Alf'6 5.31 2.50 3.70
Equation 6 (AM1—0w) (This work) 5.24 2.83 3.86
Equation 4 (AM1*) (This work) 4.84 2.65 3.59
Experiment (This work) 4.88 2.60 3.60

 

 

 

 

172

 

were not consistent with a single constraint minimized solution conformation, but
were found to be more consistent with average of conformations in solution from
molecular dynamics simulations starting from the constraint minimized solution
conformation. In all the cases the vicinal proton-carbon coupling constants
predicted by the new Karplus-type equations showed better agreement with the

measured coupling constants as compared to the three previous Karplus-type

equations (Table 5.7).

Conclusions

The solution conformations and dynamics of maltose, reduced
maltotetraose, and acarbose were studied by a combination of NMR
spectroscopy, molecular mechanics calculations and molecular dynamics
simulations. The results of the rule-based method (Chapter two) and the new
Karplus-type equations (Chapters Three and Four) were compared with previous
methods and equations. The results indicate that the rule based prediction
method can give a reasonable prediction of the conformations of oligo- and
polysaccharides. The new Karplus-type equations gave results in very good and
better agreements with experiment than the previous equations. The greater
flexibility of acarbose as compared to maltose and reduced maltotetraose was
evident. This flexibility might be one of the important factors for the ability of

acarbose to bind to many enzymes.

173

Acknowledgements

This work was supported by the Michigan State University Research
Excellence Fund. We thank Professor Jack Preiss, Department of Biochemistry,

Michigan State University for generously supplying acarbose.

174

References

1.

10.
11.
12.
13.
14.

15.

Rademacher, T. W.; Parekh, R. B. & Dweck, R. A. Ann. Rev. Biochem.
1988, 57, 785.

Phillips, M. L.; Nudelman, E.; Gaeta, F. C. A.; Perez, M.; Shingel, A. K.;
Hakamori, S. ; Paulson, J. C. Science 1990, 250, 1130.

Mohammed, H.; Hollingsworth, R. l. J. Org. Chem. 2001 (accepted for
publication).

Ernst, R., Bodenhansen, G., & Wokahun, A. Principles of Nuclear
Magnetic Resonance in One and Two Dimension, 1987 pp 431-440,
Oxford University Press, Oxford.

Bax, A.; Davis, D. G. J. Magn. Reson. 1985, 65, 355.

Bax, A.; Subramanian, S. J. Magn, Reson. 1986, 67, 565.

Sangers, J. K., & Hunter, B.K. Modem NMR Spectroscopy, 1993 pp 193-
197, Oxford University Press, Oxford.

Krishnamurthy, V. V. J. Magn. Reson. 1996, 121, 33.

Mayo, S. L.; Olafson, B. D.;Goddard, W. A. J. Phys. Chem. 1990, 94,
8897.

Molecular Simulations, Inc. Waltham, MA 02154.

Wang, Y.; Hollingsworth, R. l. Biochemistry1996, 35, 5647.

Bock, K.; Pederson, H. Carbohydr. Res. 1984, 132, 142.

Bock, K.; Meldal, M.; Refn, S. Carbohydr. Res. 1991, 221, 1.

Tvaroska, l; Hricovini, M; Petrakova, E. Carbohydr. Res. 1989, 189, 359.
Mulloy, B.; Frenkiel, T. A.; Davies, D. B. Carbohydr. Res. 1988, 184,39.

175

16.

17.

18.

Cloran, F.; Carmichael, l.; Serianni, A. S. J. Am. Chem. Soc. 1999, 121,
9843.

Quingley, B. l.; Sarko, A.; Marchessault, R. H. J. Am. Chem. Soc. 1970,
92, 5834.

Brzozowski, A. M.; Davis, G. J. Biochemistry 1997, 36, 10837.

176