Citation
Oleanane triterpene saponins from semen hippocastani

Material Information

Title:
Oleanane triterpene saponins from semen hippocastani
Creator:
Perone, Peter A
Publication Date:
Language:
English
Physical Description:
xvi, 103 leaves : ; 28 cm

Subjects

Subjects / Keywords:
Horse chestnut -- Seeds ( lcsh )
Saponins ( lcsh )
Escins ( lcsh )
Genre:
bibliography ( marcgt )
theses ( marcgt )
non-fiction ( marcgt )

Notes

Bibliography:
Includes bibliographical references (leaves 101-103).
Statement of Responsibility:
by Peter A. Perrone.

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
227794352 ( OCLC )
ocn227794352
Classification:
LD1193.L46 2007m P47 ( lcc )

Full Text
OLEANANE TRITERPENE SAPONINS FROM SEMEN HIPPOCASTANI
by
Peter A. Perrone III
B.S., Metropolitan State College of Denver, 1998
A thesis submitted to the
University of Colorado at Denver and Health Sciences Center
in partial fulfillment
of the requirements for the degree of
Master of Science
Chemistry
2007


This thesis for the Master of Science in Chemistry
Degree by
Peter A. Perrone III
Has been approved
By
Date


Perrone, Peter Anthony III (M.S. Chemistry)
Oleanane Triterpene Saponins from Semen Hippocastani
Thesis directed by Professor John Lanning
ABSTRACT
The seed of the common horse chestnut tree (Aesculus hippocastanum,
L.) has been widely used in traditional and modem herbal medications for the
treatment of disorders of the circulatory system. The active chemical
constituents responsible for these medicinal properties are a group of
approximately thirty saponin isomers collectively known as escins. Two
individual isomers were isolated from the escin mixture by preparative high
performance liquid chromatography. The identities and purity of each
compound was confirmed using liquid chromatography, mass spectrometry,
and nuclear magnetic resonance. Based on the experimental evidence, the
compounds were determined to be 21 -tigloyl-22-acetylprotoescigenin-3 -O- [|3-
D-glucopyranosyl-( 1 2)]-[P-D-glucopyranosyl-( 1 *4)]-fl-D-glucuronic acid
pyranoside (escin la), and 21-angeloyl-22-acetylprotoescigenin-3-0-[j)-D-
glucopyranosyl-( 1 >2)]-[fl-D-glucopyranosyl-( 1 -+4)]-P-D-glucuronic acid
pyranoside (escin lb). In addition, a qualitative liquid chromatography mass
spectrometry method was developed to identify the structural similarities that
are common to all escin isomers contained in horse chestnut seed extract. The


mass spectral fragmentation pattern for each isomer showed that the
polysaccharide fragment generated from each escin isomer gave intense
signals at either mJz 493.5 or m/z 523.5. These signals were then used to
develop a quantitative liquid chromatography mass spectrometry method. The
method was designed to determine the total amount of escin in herbal
medications containing horse chestnut. The quantitative method was shown to
be linear within the concentration range of 0.01 to 0.19 mg/mL, with a
correlation coefficient (r2) of 0.9967.
This abstract accurately represents the content of the candidates thesis. I
recommend its publication.
Signei
JohhLanning


TABLE OF CONTENTS
Figures...............................................................ix
Tables...............................................................xiv
Abbreviations.........................................................xv
Chapter
1. Introduction.................................................1
2. Background...................................................3
2.1 Botanical Characteristics of Aesculus Hippocastanum..........3
2.2 Medicinal Uses of Semen Hippocastani.........................4
2.2.1 Historical Uses..............................................4
2.2.2 Anti-inflammatory Effects of HCSE............................5
2.2.3 HCSE for the Treatment of Chronic Venous Insufficiency.......5
2.2.4 HCSE in the Treatment of Edema...............................6
2.3 Chemical Composition of Semen Hippocastanum..................7
2.3.1 Elemental Analysis...........................................8
2.3.2 Analysis of the Sugar Components of Escin Saponins...........8
2.3.3 Analysis of the Aglycone.....................................9
2.3.4 Stuctural Determination.....................................10
2.3.5 Properties of Escin Mixtures................................12
2.3.6 HPLC Purification of Individual Escins......................12
v


2.3.6.1 Escinla.......................................................13
2.3.6.2 Escin lb......................................................14
3. Saponin Analysis Techniques...................................16
3.1 Drying, Extraction, and Isolation.............................16
3.2 High Performance Liquid Chromatography (HPLC).................17
3.2.1 Types of HPLC.................................................17
3.2.2 Analytical and Preparative Techniques.........................18
3.3 Separation Theory.............................................19
3.3.1 Resolution....................................................19
3.3.2 Experimental Parameters.......................................21
3.3.3 Selectivity Factor............................................22
3.3.4 Column Efficiency.............................................23
3.4 Detection of Saponins.........................................24
3.4.1 Mass Spectrometer Detection...................................24
3.4.2 Electrospray Ionization.......................................25
3.4.3 Ion Traps.....................................................25
3.4.4 Ion Traps and MS.............................................26
4. Project Goals.................................................27
5. Experimental..................................................29
5.1 Materials and Reagents........................................29
5.2 Instrumentation...............................................29
vi


6. Results and Discussion......................................31
6.1 Initial Investigation.......................................31
6.1.1 Solubility and Melting Point................................31
6.1.2 FTIR Analysis...............................................31
6.1.3 LC-MS Analysis..............................................32
6.2 Isolation of Individual Escins..............................35
6.2.1 Analytical Scale Method Development.........................35
6.2.2 Column Selection............................................36
6.2.3 Mobile Phase................................................36
6.2.4 Load Study..................................................39
6.2.5 Large Scale Preparative HPLC Purification...................42
6.2.5.1 Fraction Analysis...........................................43
6.2.5.2 Result of Large Scale Isolation.............................45
6.2.6 Final Preparative HPLC Purification.........................45
6.2.6.1 Column Evaluation..........................................45
6.2.6.2 Final Purification of Compound 1............................48
62.63 Final Purification of Compound 2............................49
6.3 Characterization of Isolated Compounds......................50
6.3.1 HPLC-DAD Analysis of Compound 1.............................50
6.3.2 HPLC-DAD Analysis of Compound 2.............................54
6.3.3 Mass Spectrometry Analysis..................................56
vii


6.3.3.1 MS" Analysis of Compound 1...................................57
6.3.3.2 MSn Analysis of Compound 2...................................61
6.3.4 Nuclear Magnetic Resonance Spectroscopy......................65
6.3.4.1 NMR Analysis of Compound 1...................................66
6.3.4.2 NMR Analysis of Compound 2...................................71
6.4 Qualitative and Quantitative Analysis of Herbal Products
Containing HCSE.............................................75
6.4.1 Qualitative Analysis.........................................72
6.4.1.1 Result of Qualitative Analysis...............................82
6.4.2 Quantitative Analysis........................................84
6.4.2.1 Instrumental Conditions......................................85
6.4.2.2 Standard Analysis............................................87
6.4.2.3 Sample Analysis..............................................90
7. Conclusion...................................................97
Bibliography..........................................................101
VUl


LIST OF FIGURES
Figure
1. Line Drawing of the Leaves, Flowers, and Fruit of the Horse
Chestnut Tree...............................................4
2. Major Sugar Residues Found After Acid Hydrolysis Escin
Mixture.....................................................9
3. The Molecular Structure of the Olean-12-en Skeleton.........10
4. The Basic Molecular Structures and Percentages of the Major
Escin Isomers as Proposed by Wulff and Tschesche...........11
5. Molecular Structure of Escin la, as proposed by Yoshikawa et. al.... 13
6. Molecular Structure of Escin lb, as proposed by Yoshikawa et. al..... 14
7. Chromatogram Showing Two Peaks with a Resolution of 1.0.....20
8. FTIR Spectrum of p-Escin....................................32
9. TIC Generated Through the LC-MS analysis of P-Escin.........33
10. Mass Spectrum of P-Escin from 14 to 18 Minutes (Figure 9)...34
11. Acid Study: A) Flow 1 ml/min 63:37 MeOH:0.1 % Acetic;
B) Flow 0.5 ml/min 63:37 MeOH:0.1% Formic;
C) Flow 0.5 ml/min 63:37 MeOH:0.1% TFA..........38
12. Load Study: A) 0.05 mg Injection;
B) 2.3 mg Inection;
IX


C) 4.5 mg Injection;
D) 9.1 mg Injection..............................41
13. Preparative Chromatogram, 3 gram injection....................42
14. Preparative Fraction Analysis.................................44
15. Column evaluation: A) Phenomenex Luna Cl 8 (2),
250 x 21.2 mm, 10 pm particle size
B) Phenomenex Luna Cl 8 (2),
150 x 21.2 mm, 5 pm particle size
C) Columns A + B........................47
16. Preparative Purification of Compound 1........................48
17. Preparative Purification of Compound 2........................50
18. HPLC Chromatogram of Compound 1...............................51
19. UV Spectrum at RT = 34.8 minutes..............................52
20. Peak Purity Plot as Calculated for Peak at 34.8 Minutes.......53
21. HPLC Chromatogram of Compound 2...............................54
22. UV Spectrum at RT = 38.9 Minutes..............................56
23. Peak Purity Plot as Calculated for Peak at 38.9 Minutes.......54
24. ESI+ Infusion of Compound 1 at 10 pL/min......................58
25. ESI+ infusion of Compound 1 at 10 pL/min. MS of the
m/z 1153.5....................................................59
26. Compound 1 ESI+ infusion at 10 pL/min. MS/MS 991.5............60
x


27. Compound 1 ESI+ infusion at 10 pL/min. MS/MS 523.5.........61
28. ESI+ Infusion of Compound 2 at 10 pL/min...................62
29. ESI+ Infusion of Compound 2 at 10 pL/min. MS2 of mh 1153.5.63
30. Compound 2 ESI infusion at 10 pL/min MS3 653.36............64
31. Compound 2 ESI Infusion at 10 pL/min MS3 523.5.............65
32. 'H NMR of Compound 1.......................................66
33. !H NMR Specram of Compound 1 From the Region of
4.8 to 7.8 ppm............................................67
34. The Molecular Structure of Tiglic Acid and the *H NMR
Spectrum of Compound 1 From the Region of 0.4 to 2.5 ppm...68
35. *H NMR of Compound 2.......................................71
36. The Molecular Structure of Angelic Acid and the *11 NMR
Spectrum of Compound From the Region of 0.4 to 2.6 ppm.....72
37. TIC Chromatogram of P-Escin................................77
38. Mass Spectrum of Peaks at 10.6 and 11.65 Minutes..........78
39. Mass Spectrum of Peaks at 20.6,23.4 minutes...............79
40. Mass Spectrum of the Peaks at 23.0,23.4,26.65,28.7 Minutes.80
41. Mass Spectrum of the Peaks at 26.7,27.6,29.4, and 30.4
Minutes...................................................81
42. Mass spectrum of peaks at peaks at 46.5 and 47.9 minutes...82
43. Mixed Chromatogram of a 0.2 mg/mL P-Escin Standard with the
xi


TIC (top), the Extracted Ion at m/z 523.5 (middle), and
Extracted Ion at m/z 493.5 (bottom).........................87
44. Calibration Curve for the Extracted Ion at m/z 523.5........89
45. Calibration Curve for the Extracted Ion at m/z 493.5........89
46. Calibration Curve for the Combined Ions.....................89
47. Mixed Chromatogram of Powdered HCS With the TIC (top),
the Extracted Ion at m/z 523.5 (middle), and Extracted Ion at
m/z 493.5 (bottom)..........................................90
48. Mixed Chromatogram of Sample 05-0537 With the TIC (top),
the Extracted Ion at m/z 523.5 (middle), and Extracted Ion
at m/z 493.5 (bottom).......................................91
49. Mixed Chromatogram of Sample 05-0538 With the TIC (top),
the Extracted Ion at m/z 523.5 (middle), and Extracted Ion at
m/z 493.5 (bottom)..........................................92
50. Mixed Chromatogram of Sample 07-0355 With the TIC (top),
the Extracted Ion at m/z 523.5 (middle), and Extracted Ion at
m/z 493.5 (bottom)..........................................93
51. Mixed Chromatogram of of Sample 07-0356 With the TIC (top),
the Extracted Ion at m/z 523.5 (middle), and Extracted Ion at m/z
493.5 (bottom).............................................94
52. Mixed Chromatogram of Sample 07-0357 With the TIC (top),
Xll


the Extracted Ion at m/z 523.5 (middle), and Extracted Ion at m/z
493.5 (bottom)..............................................
.95


LIST OF TABLES
Tables
1. The Acids Studied for Adjusting the pH of the Mobile Phase..36
2. MS Conditions...............................................55
3. 13C NMR Data for Compound 1.................................66
4. 13C NMR Data for Compound 2.................................70
5. HPLC Conditions............................................ 73
6. MS Conditions...............................................74
7. P-Escin Peak Analysis.......................................83
xiv


LIST OF ABBREVIATIONS
Ac: Acetyl
ACN: Acetonitrile
amu: atomic mass unit
APCI: Atmospheric Pressure Chemical Ionization
API: Atmospheric Pressure Ionization
13C: Carbon thirteen
C: Carbon
C: Degrees Celsius
CID: Collision Induced Dissociation
DAD: Diode Array Detection
ESI: Electrospray Ionization
EtOH: Ethanol
FTIR: Fourier Transform Infrared
GC: Gas Chromatography
H2O: Water
HCS: Horse Chestnut Seed
HCSE: Horse Chestnut Seed Extract
HPLC: High Performance Liquid Chromatography
KBr: Potassium Bromide
xv


LC: Liquid Chromatography
MeOH: Methanol
mg: Milligram
pL: Microliter
mL: Milliliter
MS: Mass Spectrometry
MS/MS: Mass Spectrometry/Mass Spectrometry
MS": If n = 2 then MS/MS. If n = 3, then MS/MS/MS. Repeat as necessary,
m/z: mass to charge ratio
N: Number of theoretical plates
nm: Nanometer
NMR: Nuclear Magnetic Resonance
R: Resolution
RT: Retention Time (usually in minutes)
TDS: Total Dissolved Solids
TFA: Triflouroacetic Acid
TIC: Total Ion Chromatogram
TLC: Thin Layer Chromatography
UV: Ultraviolet
V: Volt
xvi


1.
Introduction
Plants have been used for thousands of years as medicines for the
treatment of a range of diseases, and today plants continue to play a key role
in the medicinal systems of many cultures. According to the World Health
Organization (WHO), over 60% of the worlds population use plants as their
primary source of medicine.1 The use of standardized herbal supplements has
recently increased in both Europe and the United States. For example,
standardized horse chestnut seed extract (HCSE) is the most frequently
prescribed medication in Germany for the treatment of chronic venous
insufficiency. However, herbal supplements containing HCSE are complex
mixtures of many different compounds, and the quantity of active constituents
can vary considerably between different manufacturerers.3 Regulations
controlling the manufacture and prescription of herbal medications like HCSE
vary from country to country. In the United States, herbal medicines are
regulated under the Dietary Supplement Health and Education Act of 1994,
which requires the use of good manufacturing procedures, but does not require
the same quality assurance procedures used in the manufacture of
pharmaceutical drugs as mandated by the Food and Drug Administration. In
Germany, herbal medications are evaluated for safety and efficacy by a group
of doctors and pharmacists known as the Commission E, who publish a series
1


of monographs that approves the use of an herbal medication and provides
information about the herbs pharmacological activity and possible side
effects. In order to ensure the consistent quality of herbal medications,
harmonization of each countrys regulations is needed, and quality goals need
to be defined. These criteria should include the use of validated analytical
methods that enable the identification and quantification of the active
chemical compounds found in herbal medications like HCSE.
2


2. Background
2.1 Botanical Characteristics of Aesculus Hippocastanum
The common horse chestnut tree (Aesculus hippocastanum, L.) is a
member of the Sapindaceae family of flowering plants. There are
approximately 25 species of aesculus trees and shrubs found in temperate
regions throughout the world. A phylogenetic study of the evolutionary
relationships among aesculus species indicates the genus originated in East
Asia before the end of the tertiary period, and diverged into several different
species that moved into Europe and North America.4 The European species, A.
hippocastanum, is native to Albania and northern Greece. It is a large tree that
can grow up to 30 meters high, and is cultivated in many countries to provide
shade and ornamentation. The horse chestnut tree has numerous branches,
with leaves that grow along the stem in pairs opposite each other and spread
outward palmately. The flowers, which bloom from April through July, are
typically white colored with four or five petals, and can be arranged as upright
branched clusters. The fruit of the horse chestnut tree is composed of a large,
sometimes spiny husk, which contains as many as four brown seeds. An
illustration of the leaves, flowers, and fruit of the horse chestnut is shown in
Figure 1.
3


Figure 1. Line Drawing of the Leaves, Flowers, and Fruit of die Horse
Chestnut Tree. (Reprinted From http://plants.usda.gov)
2.2 Medicinal Uses of Semen Hippocastani
2.2.1 Historical Uses
Mattioli, an Italian doctor and botanist of the 16th century, was the first
European to ascribe medicinal properties to the seed of the horse chestnut tree
attributing astringent qualities to the seed.5 In 1896 the French physician
Artault de Vevey described numerous medicinal properties of the horse
chestnut seed, including its use in the treatment of circulatory problems and as
4


an anti-inflammatory agent.6 Kings American Dispensatory (1898)
recommended the use of horse chestnut seed for the treatment of rectal
irritation and hemorroids.7 Currently, the WHO recommends the use of
semen hippocastani (horse chestnut seed) for a variety of medical conditions,
including as an anti-inflammatory agent, in the treatment of chronic venous
insufficiency (CVI), and for the reduction of edema and bruising.8
2.2.2 Anti-inflammatory Effects of HCSE
Many of the medicinal effects of HCSE can be attributed to its ability
to modulate the inflammatory response to stress. The inflammation response
involves a cascade of events, characterized by neutrophil activation and
adhesion, which leads to an increase in capillary permeability, and results in
the effusion of fluid into the interstices of cells contained in tissue spaces or
into body cavities. One of the mechanisms by which HCSE has been shown to
modulate inflammation is by decreasing the expression of adhesion molecules
in neutrophils during the early stages of the inflammatory response.9 This
mechanism can prevent tissue damage associated with acute inflammation
because neutrophils release the free radical superoxide anion, which can
disrupt the structure of biological membranes through radical oxidation.
2.23 HCSE for the Treatment of Chronic Venous Insufficiency
Inflammation of a vein is called phlebitis, which can result in painful
swelling and stiffness in the affected limb. Phlebitis occurs when the veins are
5


unable to carry oxygen-depleted blood back to the heart. This results in a
condition known as chronic venous insufficiency, or CVI. The primary
medicinal use of HCSE is in the treatment of CVI, where numerous double-
blind, randomized, placebo-controlled studies have demonstrated its
therapeutic equivalence to other medical therapies.10 A study comparing
HCSE to compression stocking therapy in reducing the symptoms of CVI
showed treatment with HCSE resulted in a significant decrease in the volume
and circumference of the lower leg.11 One mechanism by which HCSE exerts
its beneficial effect on patients with CVI is through increased venous tone,
and animal veins exposed to HCSE show an increase of up to 50% in
contractile strength.12 Compounds isolated from HCSE have been shown to
stimulate the production of prostaglandins,13 which improve venous tone by
enhancing the constricting effect of noradrenaline.14
2.2.4 HCSE in the Treatment of Edema
The inflammation response can also lead to a decrease in the oxygen
supply to cells that line veins and arteries, leading to a reduction in adenosine
triphosphate, and to the release of platelet activating factor. This cascade of
events increases capillary permeability, which allows fluid to effuse into the
surrounding tissues. This results in a distension of these tissues, and
compression of nerve endings causing pain and discomfort, which is a
condition known as edema. A randomized placebo controlled study of
6


patients with edema confirmed the clinical effectiveness of treatment with
HCSE.15 HCSE has been shown to have a positive effect in the treatment of a
wide variety of conditions associated with the edema, including the reduction
of swelling associated with minor injuries and bruising, as well as post-
operative soft tissue swelling.16 Several biological mechanism have been
proposed to account the reduction of edema in patients treated with HCSE.
Primarily, HCSE has been shown to alter vascular permeability, which
normalizes fluid exchange between the capillary wall and connective tissue.17
Also, the active constituents found in HCSE have been shown to reduce the
degradation of the proteoglycans that are known to protect capillary walls
from enzymatic break down following stress.18
2.3 Chemical Composition of Semen Hippocastanum
Chemical analysis of A. hippocastanum began in the mid-nineteenth
century, during which time it was determined that saponins were one of the
major components of the horse chestnut seed.19 Saponins are amphiphilic
molecules that are typically found to occur as a mixture of closely related
isomers, each of which is composed of a hydrophobic aglycone attached to
one or more sugar moieties. The amphiphilic nature of saponins gives them
soap like properties, and one of the earliest identification techniques was to
shake an aqueous solution and look for the formation of a stable foam. Other
common properties of saponins are their toxicity to fish, and their ability to
7


destroy red blood cells. Plants synthesize saponins for a variety reasons,
including increased resistance to fungal infections, as a defense against
>JA
insects, and even to suppress the growth of competing plants. The saponins
originally isolated from the seed of the horse chestnut were initially called
aescins (because they were isolated from the plant genus aesculus), but this
was later changed to the currently accepted name of escins.
23.1 Elemental Analysis
In 1931 Winterstein21 investigated the elemental composition of the
escin aglycone. He began his analysis by extracting A. hippocastanun seeds in
alcohol, precipitating the escins with acetone, and finally using dialysis to
obtain an amorphous white powder. This product was characterized by its
piscicidal and hemolytic activity, which are both characteristics of saponins.
The melting point was determined to be 220 to 230 C, and combustion
analysis showed the empirical formula to be C53H88O27. Hydrolysis products
included a sapogenin with empirical formula C35H58O7 and an organic acid
with empirical formula C5H8O2 (believed to be tiglic or angelic acid).
233 Analysis of the Sngar Components of Escin Saponins
In 1953 Jermstad and Waaler 22 investigated the sugar moieties of
escin saponins extracted from A. hippocastanum. As one of the most abundant
types of biological molecules, saccharides were once thought to function
mainly as a source of energy. However, as a component of glycosides,
8


saccharides have numerous biological activities. Many plants store reactive
secondary metabolites as inactive glycosides, and then use enzymes to remove
the saccharide moiety when the compound is needed. Jermstad and Waaler
isolated escins from horse chestnut seeds by separating the methanol extract
on an aluminum oxide column and then purifying the saponin fraction using
dialysis. Acid hydrolysis of this product, followed by paper chromatography,
showed the major escin sugar residues to be glucuronic acid, glucose, and
xylose (Figure 2).
OH OH
OH OH OH
beta-D-glucuronopyranosyl acid beta-D-glucopyranosyl beta-D-Xylopyranosyl
Figure 2. Major Sugar Residues Found After Acid Hydrolysis Escin Mixture
233 Analysis of the Aglycone
In 1957 Jeger et al. proposed that escins are composed of a group of
structurally related triterpene glycosides, all of which are built up from an
oleanane skeleton, which is shown in Figure 3.
9


Figure 3. The Molecular Structure of the Olean-12-en Skeleton
Because the oleanane skeleton can be hydroxylated at numerous positions,
oleanane derivatives comprise over 50% of saponin aglycones,24 and are
commonly found in angiosperms like the horse chestnut tree. Triterpene
saponins are frequently found in much greater amounts than other secondary
metabolites, and the horse chestnut seed can contain up to 13% escin.
23.4 Structural Determination
Later work by Wulff and Tschesche25 confirmed that the two oleanane
derivatives found as aglycones in escins are protoescigenin and barrintogenol
C, which were found to be present as aglycones in the escin mixture at an 8 to
2 ratio, respectively. Protoescigenin is an oleanane derivative that is
hydroxylated at the C-3, C-16, C-21, C-22, C-24 and C-28 positions, while the
barringtogenol C derivative is hydroxylated at the C-3, C-16, C-21, C-22, and
C-28 positions. These two aglycones were found to be esterified at the C-21
position with either angelic or tiglic acid, and esterified at the C-22 position
10


with acetic acid. The C-3 position was found to be derivatized with a D-
glucuronic acid that is linked to a D-glucose, and either a D-glucose or D-
xylose. Although WulfF and Tschesche were unable to purify the individual
saponins from the escin mixture, they were able to purify the hydrolysis
products of the mixture, and then use NMR spectroscopy to confirm the
structures of the aglycones and positions of the esterifying acids and
trisaccharide side chain. The basic molecular structure, as well as the relative
percentage of each major isomer, is summarized in Figure 4.
Compound__________________________R Groups
Ri Ra R4 r5 Relative %
1 Tigloyl Acyl H OH Glucose 15
2 Angeloyl Acyl H OH Glucose 23
3 Tigloyl Acyl H OH Xylose 6
4 Angeloyl Acyl H OH Xylose 9
5 Tigloyl Acyl H OH Galactose 7
6 Angeloyl Acyl H OH Galactose 6
Figure 4. The Basic Molecular Structures and Percentages of the Major Escin Isomers as
Proposed by Wulff and Tschesche.25
11


23.5 Properties of Escin Mixtures
Although the majority of the escin mixture was found to be esterified
with tiglic or angelic acid at the C-21 position and esterified with acetic acid
at the C-22 position, under certain conditions the acyl group located at the
C-22 position can shift to the C-28 position. These structural isomers are
called isoescins, and when approximately 60% of the escin mixture has been
converted to isoescins the physical and biological properties of the mixture
also changes. Therefore, escin mixtures are characterized as either P-escin or
a-escin. The naturally occurring form is called P-escin, and is composed of a
mixture of C-21 and C-22 diesters. P-escin is not soluble in water, and has a
high hemolytic index, a-escin is composed of a 4:6 mixture of P-escin and
isoescin, is soluble in water, and has a low hemolytic index.
23.6 HPLC Purification of Individual Escins
Because of the complex nature of the escin mixture, isolation and
purification of the individual escin isomers has proven to be difficult.
However, Yoshikawa, et al.26 used modem liquid chromatographic techniques
to purify the major components of the escin mixture. After extracting the
seeds in methanol, a portion was passed through an anion exchange column to
remove sugars and lipids. The methanol eluate was separated by adsorption
chromatography to give the pure escin mixture. Finally, individual escins were
12


separated using reverse phase preparative high performance liquid
chromatography.
23.6.1 Escin la
The main component, called escin la, was characterized as 21-tigloyl
22-acetylprotoescigenin-3-0-[|}-D-glucopyranosyl-(l>2)]-[p-D-
glucopyranosyl-(l>4)]-p-D-glucuronic acid pyranoside. The molecular
structure of escin la is shown in Figure 5.
1
Figure 5. Molecular Structure of Escin la, as proposed by Yoshikawa et. al.26
The molecular structure of escin la shows the olean-12-en skeleton
hydroxylated at the C-3, C-16, C-21, C22, C-24, and C-28 positions to give
the protoescigenin aglycone. A trisaccharide chain composed of one
glucuronic acid and two glucoses is attached to the C-3 position of the
13


aglycone. Finally, the molecule is esterified with tiglic acid and acetic acid at
the C-21 and C-22 positions, respectively.
23.6.2 Escin lb
Another major component of the escin mixture, Called escin lb, was
determined to be 21-angeloyl-22-acetylprotoescigenin-3-0-[|$-D-
glucopyranosyl-(l^2)]-[P-D-glucopyranosyl-(l>4)]-P-D-glucuronic acid
pyranoside. The molecular structure of escin lb is shown in Figure 6.
Figure 6. Molecular Structure of Escin lb, as proposed by Yoshikawa et al
The molecular structure of escin lb is similar to that of escin la, and shows the
olean-12-en skeleton hydroxylation at the C-3, C-16, C-21, C22, C-24, and C-
28 positions, with a trisaccharide chain composed of one glucuronic acid and
two glucoses attached to the C-3 position, and is esterified with acetic acid at
14


the C-22 position. However, the aglycone is esterified with angelic acid at the
C-21.
15


3. Saponin Analysis Techniques
3.1 Drying, Extraction, and Isolation
The isolation of escins from semen hippocastanum follows several
steps commonly used to isolate saponin mixtures, beginning with drying the
plant material, initial extraction with alcohol or aqueous alcohol, liquid-liquid
partitioning, and finally precipitation of the saponin mixture.27 Typical drying
methods include air-drying, oven-drying, vacuum-drying, and lyophilization.
After drying is complete, the sample is ground to a powder suitable for
extraction. The extraction process begins with a careful choice of solvents and
extraction techniques, because it is not only necessary to free the desired
material from the plant matrix, but care must also be taken to prevent sample
degradation. The material obtained from the extraction procedure will usually
be a complex mixture of substances, and the next step in the isolation process
is to separate and concentrate the desired compounds through liquid-liquid
partitioning or solid phase extraction. Because saponins usually occur as a
mixture of closely related isomers, the purification of individual saponins can
be a challenging task, and in many instances the final purification will involve
multiple liquid chromatography steps.
16


3.2 High Performance Liquid Chromatography (HPLC)
High performance liquid chromatography (HPLC) has become one of
the most frequently used separation techniques for the isolation and
quantification of saponin mixtures. HPLC is a type of column
chromatography in which the components of a mixture are distributed
between two phases, one of which is called the stationary phase, while the
other one is called the mobile phase. The liquid mobile phase is pumped
through a column filled with small spherical particles, and the components of
the sample are separated based on their differential migration through the
column. Because individual compounds will be more or less strongly retained
by the stationary phase, the migration rate of each compound will vary, and
the sample components will be separated into discrete bands. The basic
components of an HPLC system consist of a mobile phase pump, a sample
injection system, a column, a detector, and typically some type of computer
software to control the hardware and record the data.
3.2.1 Types of HPLC
There are many types of HPLC separations that are based on different
mechanisms, and utilize different mobile and stationary phases. Reversed
phased HPLC is a separation technique based on the differences in sample
polarity, and where the stationary phase is less polar than the mobile phase.
17


Reversed phase separations typically involve the use of columns packed with
silica gel particles coated with a non-polar stationary phase like Ci8, and
mobile phases consisting of mixtures of water and methanol or water and
acetonitrile. Therefore, compounds that can be separated by reversed phase
HPLC should be soluble in polar solvents, but hydrophobic enough to be
partially retained by the column. Retention times will generally be greater for
compounds that are relatively more hydrophobic, while compounds that have
a greater overall relative polarity will elute from the column first. Because of
its versatility and low cost, reversed phase chromatography has become one of
the most commonly used analytical and preparative HPLC techniques.
3.2.2 Analytical and Preparative Techniques
Although both analytical and preparative HPLC operate under the
same chemical principles and utilize similar instrumentation, the goals of each
technique are different. The goals of the analytical separation are to separate
and quantify the components of a mixture, whereas the goal of preparative
chromatography is to collect the eluted compounds and recover them at a
much higher purity than the starting material. These differences lead to several
practical experimental considerations, including differences in sample loading
and mobile phase consumption. Sample concentration and injection volume
are dramatically increased for preparative chromatography, and sample
solubility in the mobile phase becomes an important factor. Mass and volume
18


overload can have a detrimental impact on the HPLC separation, and
optimization of these parameters using analytical scale HPLC is
recommended. Because the mobile phase velocity is substantially increased,
one of the major operating costs in preparative chromatography is mobile
phase consumption. Although acetonitrile is the solvent most commonly used
in analytical reverse phase HPLC its cost is prohibitive for preparative
separations, while methanol is relatively inexpensive and is the solvent most
commonly used in preparative HPLC. In either case, the successful
chromatographic separation involves making a compromise between
chromatographic resolution, sample capacity, and the total analysis time.
33 Separation Theory
33.1 Resolution
The goal of liquid chromatography is to separate the components of a
sample into distinct bands or peaks as they migrate through the column. The
extent to which the compounds are separated can be measured using the
following equation
Wi+W2
Where R is the resolution, ti and t2 are the retention times of the retained
components, and Wi and W2 are the widths of each peak measured at the
baseline. The resolution equation is only valid if the two peaks of interest
19


retain a Gaussian shape as they migrate through the column, and can be used
to determine the percent error when measuring the area under a peak, or to
estimate the percent purity of each compound. A typical chromatogram is
shown in Figure 7.
Figure 7. Chromatogram Showing Two Peaks with a Resolution of 1.0.
(Reprinted from http://www.wikipedia.org/)
The chromatogram in Figure 7 shows the detector response plotted as function
of time. The resolution between the two peaks has a calculated value equal to
one, and if the mobile phase is collected from the start of the first peak to the
valley between the two peaks the compound will be 98% pure.29
20


33.2 Experimental Parameters
The resolution between two adjacent bands can also be expressed in
terms of several experimental parameters, the capacity factor (k), the
selectivity (a), and the number of theoretical plates (N):
The capacity term is related to the relative strength of the mobile phase as
compared to the stationary phase, and determines how long the compounds
are retained on the column. The capacity factor can be determined using the
equation
Where t* is the retention time of the peak of interest and to is the retention time
of a non-retained peak. When the compound is not retained the capacity term
is zero, but as k increases the total analysis time will also increase. Changing
the experimental parameters effecting k alters the ratio of the compound in
the stationary phase to the amount in the mobile phase, which can be changed
by making changes to the strength of the mobile phase. This is most easily
accomplished by lowering or raising the percentage of organic modifier.
21


333 Selectivity Factor
The selectivity factor is a measure of the differences in retention times
between two adjacent bands, and can be determined by using the following
equation
kB
a = f-
kA
where k'e is the capacity factor of the highly retained compound, and k'A is the
capacity factor of the less retained compound. The selectivity can be adjusted
by changing the mobile phase chemistry. The simplest way to change o is to
change the type of organic solvent used, for example, by switching from a
polar aprotic solvent like acetonitrile to a polar protic solvent such as
methanol. Alternatively, different types of buffers can be used, the pH of the
mobile phase can be altered, or ion pairing agents can be used. However,
when analyzing compounds with ionizable functional groups the pH of the
mobile phase is most commonly varied. Because aqueous mobile phases are
used in reversed phase HPLC, the pH must be adjusted so that ionizable
compounds are in their neutral state, otherwise these compounds will not be
retained by the non-polar stationary phase. Although many HPLC separations
are carried out at ambient temperatures, elevated or lowered temperatures are
22


also commonly employed to enhance separation or to ensure reproducible
retention times.
33.4 Column Efficiency
The column efficiency term can be calculated by determining the
number of theoretical plates using the following equation:
Where N is the number of theoretical plates, W is the peak width at the
baseline, tr is the retention time of the peak of interest. This equation
compares the width of the peak to the length of time the component has been
on the column. Thus, an efficient column keeps the bands from spreading and
gives very narrow peaks. The value for N is a useful measure of the
performance of the chromatographic column because the value of N is
independent of the retention time. Therefore, the value of N calculated for a
peak at 3 minutes will be the same as the value calculated for peaks emerging
at 10 minutes. The number of theoretical plates is proportional to column
length, and doubling of the length usually results in a doubling of the retention
time and an increase in the separation between bands. The column diameter is
also important, and the typical column diameters for analytical separations are
between 2 and 4 mm, while for preparative work, diameters of 8 to 50 mm are
23


more appropriate because the sample loading capacity of a column increases
in proportion to the square of the diameter.
3.4 Detection of Saponins
Many saponins lack a UV chromophore, or have UV maxima only at
lower wavelengths. This limits the choice of mobile phase solvents, as well as
the use of acid or base modifiers. In addition, mobile phase gradients can
cause severe baseline drift at lower wavelengths, which makes quantification
difficult. Although the most common detector used in HPLC is the UV
detector, most UV absorption spectra are not specific enough to identify
individual saponins in a mixture, and retention times must be compared to
standards. Therefore, hyphenated techniques coupling HPLC with MS have
become more common. When MS/MS or MS" is used, full identification of
individual saponins is possible.
3.4.1 Mass Spectrometer Detection
The most commonly used HPLC detectors utilize ultraviolet detection
(UV). However, HPLC systems are increasingly being used in conjunction
with mass spectrometry detection. Using this type of system, compounds can
be identified using molecular weights and fragmentation patterns. Initially, the
major technical challenge involved in using MS with HPLC was in the
interface between the two systems. The HPLC system operates at atmospheric
pressure, while the MS operates under a vacuum. The HPLC eluent must be
24


changed from a neutral liquid to a charged gas, which must occur at the
interface.
3.4J Electrospray Ionization
One of the most commonly used LC-MS interfaces is electrospray
ionization (ESI). This is a soft ionization technique that is well suited for
relatively large polar compounds. In ESI, the HPLC eluent is passed through
capillary tubing to which a high electrical potential has been applied. Nitrogen
gas is allowed to flow around the tube, and the particles exit the tip of the tube
as an aerosol of charged droplets. As the mobile phase solvent evaporates the
charge density increases, causing the droplets to divide until only single ions
exist. These ions then pass through a heated capillary and into the vacuum
region of the MS detector, where they are accelerated and focused by a series
of charged cones and lenses.
3.4.3 Ion Traps
Ion trap MS instruments are widely used mass analyzers because of
their relatively low cost and relatively small size. Ion traps have a resolution
of 1 atomic mass unit (amu) and a mass limit of4,000 amu. The fundamental
working principles of the ion trap are the same as for quadrupole MS systems,
and ion traps have been described as a linear quadrupole in which one rod is
bent into a closed loop, while two cone shaped electrodes form endcaps above
and below the ring. Unlike linear quadrupoles, all of the generated ions are
25


pulsed into the trap, where helium gas is used to decrease the kinetic energy of
the ions and focus them into the center of the trap. The ions are released one at
a time by increasing the amplitude of the radio frequency (rf) voltage until
ions of a specific mass to charge ratio (m/z) are excited and then ejected
through the endcaps and onto a conversion dynode.
3.4.4 Ion Traps and MS
Ion traps can also be used to perform mass spectrometry/mass
spectrometry (MS/MS or MS"), in which specific ions are isolated, and then
fragmented through collision induced dissociation (CID) with helium gas. To
perform MS" experiments, specific ions are selected by adjusting the rf
frequency on the endcaps to selectively eject all other ions. Using a resonating
frequency that matches the isolated mass, a voltage is then applied to excite
the ion, which begins to oscillate and collide with the helium gas. After
numerous collisions, the internal energy of the selected ions increase until
fragmentation is induced. The fragment ion can then be selected and
fragmented, providing detailed structural information about the compound of
interest.
26


4. Project Goals
The current research project has been undertaken as a collaborative
study between Chromadex Analytics and the University of Colorado at
Denver and Health Sciences Center with the overall goal of developing an
analytical method or methods to ensure the identity and strength of herbal
supplements containing horse chestnut seed (HCS) or horse chestnut seed
extract (HCSE). These herbal supplements are primarily used as a treatment
for chronic venous insufficiency, and the active constituents believed to be
responsible for this pharmacological effect are a group of structural isomers
known collectively as escins. Quantification of these compounds is difficult
due to their sensitivity and lack of chromophores with strong absorptivity
coefficients, which would allow for UV detection without derivatization.
Therefore, the goal of the current research is to develop an analytical method
that would enable the quantification of escins in herbal products utilizing
liquid chromatography with mass spectrometry detection. As a prerequisite to
achieving this goal, the following objectives were determined to be necessary:
1. Isolate and purify individual escin isomers horn a mixture of escin
saponins.
2. Confirm the identity and purity of these compounds so they can be
used as reference standards.
27


3. Develop a qualitative LC-MS method to determine the MS
fragmentation patterns common to all escin isomers.
4. Develop a quantitative LC-MS method for the analysis of herbal
products containing HCS or HCSE.
28


5.
Experimental
5.1 Materials and Reagents
Horse chestnut seed (dried and ground) was obtained from Pacific
Botanicals (Grants Pass, OR). Commercially available reagent grade P-escin,
previously purified from A. hippocastanum, was obtained from Chromadex
Inc. (Secondary Standard, > 95% purity, Irvine, CA). Acetic acid (ACS grade,
99.7% min.), formic acid (ACS grade, 98%min.), methanol (HPLC grade,
99.8% min.), and acetonitrile (HPLC grade, 99.8% min.) were obtained from
EMD Chemicals (Gibbstown, NJ). HPLC grade water was obtained from a
Milli-Q Academic reverse osmosis system from Millipore (Billerica, MA).
The KBr used for the IR analysis was purchased from Sigma-Aldrich (FTIR
grade, 99+%, St. Louis, MO). Bakerbond Cis silica was obtained from
Mallinckrodt Baker, Inc.(Phillipsburg NJ, USA). Technical grade methanol
was purchased from Uni-Bar The solvent used in the NMR analysis was
Pyridine-d5 from Sigma-Aldrich (99+% deuterium, St. Louis, MO).
5.2 Instrumentation
The analytical HPLC systems used were Agilent 1100 series (Santa
Clara, CA) that were equipped with a vacuum degasser, an autosampler
injection system, a thermostated column oven, and a binary pump with a
quaternary low pressure mixing valve. MS analysis was carried out using a
29


LCQ-Deca ion-trap mass spectrometer from Thermo-Finnigan (San Jose, CA)
equipped with an electrospray ionization source. A Branson 3210 Ultrasonic
Water Bath (Danbury, CT) and a Mettler Toledo AX205 (Columbus, OH)
were used for the preparation of standards and samples. A Waters HPLC
system (Milford, MA) operating with the Waters 600 controller, 600 binary
pump, and 996 DAD was used for preparative purification. A SepTech
preparative HPLC system (Varian, Palo Alto, CA), using a SepTech GP900
controller, SepTech VWD, and 40 mL manual loop injector was used for
large scale purification. Each purified fraction was then added to an open
column packed with 15 grams of 50 pm diameter Bakerbond Cig silica from
Mallinckrodt Baker, Inc. (Phillipsburg, NJ). Individual fractions were then
dried under vacuum using a Buchi R205 Rotavapor (Switzerland). Melting
points were obtained using a Mel-Temp II from Laboratory Devices (U.S.A.)
Infrared spectra were obtained using a Paragon 500 Fourier transform infrared
(FTIR) spectrophotometer from Perkin- Elmer (Waltham, MA). The 'H NMR
and 13C NMR spectra of purified compounds were obtained at the University
of Colorado at Boulder NMR laboratory using a Varian 500 MHz (Palo Alto,
CA) and the data were processed using NUTS software (Livermore, CA).
30


6.
Results and Discussion
6.1 Initial Investigation
6.1.1 Solubility and Melting Point
Chromadex P-escin (Secondary Standard, > 95% purity), that had been
isolated from A. hippocastanum, was evaluated to confirm its identity and
purity. The material was observed to be a crystalline white powder, that was
found to be soluble in methanol and acetonitrile, but was insoluble in water,
acetone, and tetrahydrofuran. The melting range was determined to lie within
an average range of 221 to 225 C (with decomposition). The solubility and
melting point confirm that the material is P-escin and not a-escin, which is
soluble in water and has a melting range of225 to 227 C, and that could have
'll)
been formed though exposure to moisture or heat.
6.1.2 FTIR Analysis
Infrared spectra were obtained by mixing approximately 5 mg of P-
escin with 200 mg KBr using a mortar and pestle. The mixture was then
formed into a clear pellet using a die and anvil set. The FTIR spectrum
showed two strong absorption bands at 3429 and 1083 cm*1 corresponding to
frequencies produced by primary alcohols, a carbonyl absorption band at 1726
cm*1, and an alkene absorption at 1642 cm*1, as shown in Figure 8.
31


Figure 8. FTIR Spectrum of P-escin
The FTIR spectrum of P-escin shown if Figure 7 is a match when compared to
the infrared spectrum of P-escin obtained by Voigtlander and Rosenberg.
6.1.2 LC-MS Analysis
An exploratory LC-MS analysis of the material was carried out using
the Agilent 1100 series HPLC system connected to the LCQ-Deca ion-trap
mass spectrometer equipped with the electrospray ionization source. Samples
were prepared by dissolving approximately 1 mg of P-escin in 1 mL of
methanol, which were then analyzed by injecting a 5 pL aliquot. The method
utilized a Waters Symmetry Cig (150 x 4.6 mm, 3.5 pm particle size) column
equilibrated at 40 C and a flow rate of 1.0 mL/min. Mobile phase A consisted
0.1% formic acid in H2O, and mobile phase B consisted of acetonitrile
(ACN). The mobile phase conditions began at 5% B increasing to 95% B over
32


20 minutes, then holding at 95% B for 5 minutes, and finally decreasing to 5%
B over 5 minutes. The total ion chromatogram (TIC) from an entire gradient
run was used to tentatively identify the masses of all the components in the
sample, and to give an estimate of the percentage of each component. The
resulting chromatogram generated from the TIC trace is shown in Figure 9.
Figure 9. TIC Generated Through the LC-MS analysis of p-escin
The TIC generated from the initial LC-MS gradient shows two major peaks
with retention times of 15.2 and 16.5 minutes, and mass spectrum with strong
signals at m/z 1153.6, as shown in Figure 10.
33


FigurelO. Mass Spectrum of (J-escin from 14 to 18 Minutes (Figure 9)
A similar chromatogram and mass spectra were obtained by Pietta et al.32
who identified the two major peaks as escin la and escin lb. Therefore, the
signal at m/z 1153.6 was tentatively determined to correspond to the sodium
adduct of these two components of the P-escin mixture, escin la and escin lb,
both of which have an exact mass of 1130.6 amu. The mass spectrum shown if
Figure 10 also shows a strong signal at m/z 1123.5, which corresponds to the
sodium adduct of escins Ha and lib, both of which have an exact mass of
1100.5 amu, and if similar chromatographic conditions are used, they always
elute together with escins la and lb.33 Therefore, based on the experimentally
obtained chromatograms and mass spectra, as well as the chromatograms and
mass spectra obtained from P-escin in chemistry literature the two major
peaks seen in the TIC were tentatively identified as escin la co-eluting with
34


escin Ha (retention time of 15.3 minutes), and escin lb co-eluting with escin
lib (retention time of 16.5 minutes).
Also, the sodium adduct of the protoescigenin aglycone, esterified at
the C-21 position by either tiglic or angelic acid, can be seen at m/z = 653.6.34
The signal at m/z 523.4 corresponds to a polysaccharide chain composed of a
glucuronic acid and two glucose molecules, while the signal at m/z 493.4 is
characteristic of the sodium adduct of a polysaccharide chain composed of a
glucose, a xylose, and a glucuronic acid.35
6.2 Preparative Isolation of Individual Escins
6.2.1 Analytical Scale Method Development
Although the extraction, isolation, and characterization of escins as a
group of structurally related isomers had been reported as early as 1931 by
Winterstein21, it was not until 1994 that Yoshikawa et al.,26 purified the major
constituents of the saponin mixture using preparative reverse phase
chromatography. However, many of the important parameters of the final
purification were not reported, such as column dimensions, the loading and
injection volume, or flow rate of the mobile phase. Also, details of the
recovery of pure compounds from the mobile phase were not reported, which
is important because escins are sensitive to acidic conditions as well as heat.36
Therefore, many of the parameters involved in the preparative isolation of
35


individual escins needed to be evaluated and optimized using analytical scale
chromatography.
6.2.2 Column Selection
In order to determine the optimum preparative separation conditions,
the first step is to determine the appropriate mode of chromatography based
on the physical properties of the samples. Because escins are small molecules
with a molecular weights of less than 2000 that contain ionic functional
groups and are soluble in methanol-water mixtures, reverse phase HPLC was
selected. Therefore, an analytical column with the similar packing material
was chosen during the initial method development, with the goal of gradually
scaling up the process for preparative and large scale HPLC purification.
6.2.3 Mobile Phase
Methanol was chosen as the organic component of the mobile phase,
and because escins are ionizable compounds containing multiple hydroxyl
groups as well as a carboxylate moiety, the mobile phase pH will have to be
adjusted using a volatile acid that has a UV cutoff of less the 220 nm (because
the location of the cut points for the collection of each compound will be
determined by UV response). Therefore, several commonly used acids were
evaluated to see which one would provide the best peak shape and give the
best resolution between analytes. The acids studied and their physical
properties are summarized in Table 1.
36


Table 1. The Acids Studied for Adjusting the pH of the Mobile Phase
Acid M.W. b.p. pKa UV cutoff (% 10 mM
Acetic 60.05 118.0 4.74 210
Formic 46.03 100.8 3.75 210
Triflouroacetic 114.02 72.4 0.3 210
The Agilent 1100 series HPLC system was used for the initial
analytical chromatographic evaluation. The chromatographic conditions were
optimized using a Phenomenex Luna Cl8 (2), 250 x 4.6 mm, 5 pm particle
size column at a temperature of 25 C, and UV detection at 210 nm. The
results of this acid study are graphically demonstrated in Figure 11.
37


Figure 11. Acid study: A) Flow 1 mL/min 63:37 MeOH:0.1% Acetic; B) Flow 0.5 mL/min
63:37 MeOH:0.1% Formic; C) Flow 0.5 mL/min 63:37 MeOH:0.1% TFA
38


The results were compared for peak shape and overall resolution of the major
components using similar chromatographic conditions. As shown in Figure
11, the use of either acetic acid or formic acid gave broad peaks with
significant tailing, and failed to adequately separate the major group of peaks.
However, using TFA sharp symmetrical peaks were observed, and partial
separation of the major components was achieved. Therefore, TFA was
chosen as the acid modifier that would be used for the preparative HPLC
mobile phase.
6.2.4 Load Study
Sample loading is an important parameter in preparative
chromatography, because efficiency and throughput increase as increased
amounts of sample are loaded onto the column, although an increase in the
sample loading decreases resolution. Semi-purified fractions are collected, and
then later re-injected at a lower load ratio to get higher purity.35 The ratio of
the total dissolved solids (TDS) in the sample to the quantity of silica
contained in the column is defined as the load ratio. As the load ratio is
increased, the stationary phase can no longer absorb analytes from the mobile
phase, peaks become distorted, and resolution is compromised. Therefore, a
load study was conducted on an analytical scale, whereby increasing amounts
of sample were injected onto the column, and the peak shape and resolution
are evaluated. The HPLC system and column were the same as described in
39


section 6.2.3, with a mobile phase consisting of 65% methanol and 35% water
both containing 0.05% TFA at a flow rate of 0.3 mL/min. Using these
conditions, 0.05 mg, 2.3 mg, 4.5 mg, and 9.1 mg of fl-escin was injected onto
the column. The results are shown in Figure 12.
40


Figure 12. Load Study: A) 0.05 mg Injection; B) 2.3 mg Injection; C) 4.5 mg Injection; D)
9.1 mg Injection
41


The load study revealed that high sample loading was possible, because even
when injecting 9.1 mg sample at a load ratio of 231 mg silica to lmg total
dissolved solids (TDS), the critical pairs had a resolution greater than 0.5,
which would still enable each major peak to be collected at 85% purity (if the
mobile phase is collected from the start of the first peak to the valley between
the two peaks).29
6.2.5 Large Scale Preparative HPLC Purification
Large scale purification was performed on the SepTech preparative
HPLC system with the variable wavelength detector (VWD) set at 215 nm,
and using a 500 x 76.2 mm column packed with 1140 grams of
Chromatorex Cis coated silica gel (15 pm particle size). The mobile phase
was composed of 65% MeOH and 35% H2O (both containing 0.05% TFA),
and using a flow rate of200 ml per minute. The sample was prepared by
dissolving 3 grams of {3-escin sequentially beginning with 7 mL of MeOH
containing 0.05% TFA, then adding 4 mL of 80% MeOH with 20% H2O
containing 0.05% TFA, and finally 9 mL of 65% MeOH with 35% H2O
containing 0.05% TFA to give a total of 20 mL of 80% MeOH with 20% H2O
containing 0.05% TFA. The mobile phase flow rate was decreased to 50 mL
per minute and total 20 mL sample volume was injected, to give a load ratio
of 380 mg silica to 1 mg of TDS. The resulting chromatogram is shown in
Figure 13.
42


Esctab + IIa
Figure 13. Preparative Chromatogram, 3 Gram Injection
The chromatogram shows the characteristic major peaks seen in the analytical
LC-MS analyses (Figure 9), and believed to composed of escin la and Ha
followed by escins lb and lib. Fractions were collected every minute starting
at 152 minutes and ending at 230 minutes. Each fraction was then analyzed
for purity by analytical scale HPLC.
6.2.5.1 Fraction Analysis
Analysis of each fraction was done using the Agilent 1100 series
HPLC system equipped with UV detection at 220 nm. The analysis method
utilized a Waters Symmetry C18 (150 x 4.6 mm, 3.5 pm particle size) column,
which was maintained at a temperature of 20 C. The mobile phase was held
isocratic at 65% MeOH (containing 0.025% TFA) and 35% H2O (containing
0.025% TFA) and a flow rate of 0.7 mL/min. A plot of the peak area of each
43


component versus the fraction collection time was constructed. The result is
Figure 14. Preparative Fraction Analysis
The fraction analysis indicates that each band did not retain a Gaussian shape,
and severe tailing of each compound occurred in the column during the
purification process. This could be the result of the high injection volume, or
from laminar flow in the sample loop during the injection process. However,
during the preparation of the mobile phase a precipitate formed, which was
later determined to be a polymer contaminant in the technical grade MeOH
used during the process. This polymer was found to have a detrimental effect
on the column, and continued use inevitably resulted in a rising and noisy
baseline with poor resolution between peaks and distorted peak shape.
44


Based on the fraction analysis, appropriate fractions were pooled, and
then diluted with water until the total volume of the combined fractions
contained 80% water. In order to remove the TFA, each diluted fraction was
then passed through a Dynamax 250 x 21.4 mm preparative column packed
with 15 pm particle size Cjg stationary phase. The column was equilibrated at
20% MeOH with 80% water at a flow rat of 50 mL/min. The samples were
then added to the column (as mobile phase), and water was then passed
through the column until the TFA had been removed, after which the samples
were eluted with MeOH. The fractions were then dried under vacuum using a
Rota-Vap rotary evaporator with the temperature bath set to 20 C.
6.2.5.2 Result of Large Scale Isolation
Using the large scale HPLC method described, 336 mg of compound 1
(labeled as CDXA-37.89.4) was obtained, and 200 mg of compound 2
(labeled as CDXA-37.89.8) was obtained. These compounds were further
purified using smaller diameter preparative HPLC columns.
6.2.6 Final Preparative HPLC Purification
6.2.6.1 Column Evaluation
The Waters HPLC system was used for the final purification of each
compound. Several columns were evaluated, including a Phenomenex Luna
Cl8(2) column with dimensions 250 x 21.2 mm and 10 pm particle size, as
well as a Phenomenex Luna C18(2) column with dimensions 150 x 21.2 mm
45


and 5 particle size. The mobile phase used to evaluate each column was
composed of 65% methanol and 35% water containing 0.05% TFA, and using
a flow rate of 10 mL per minute. During the initial evaluation, an injection
volume of 500 pL of methanol containing 50 mg p-escin was used. The results
show that the resolution was poor when using either column singly, but when
the two columns were connected in series the chromatography closely
resembled the analytical runs, as shown in Figure 15.
46


0.
-i--1--1--1---1-1i----1--1--1i---1 i i i I I t~
*0.00 40.00 00.00 00.00
Figure 15. Column evaluation. A) Phenomenex Luna C18 (2), 250 x 21.2 mm, 10 pm particle
size B) Phenomenex Luna C18 (2), 150 x 21.2 mm, 5 pm particle size C)
Columns A + B
47


6.2.6.2 Final Purification of Compound 1
The Waters HPLC system was used for the final purification of
compound 1 (labeled as CDXA-37.89.4). The mobile phase was composed of
60% methanol and 40% water containing 0.05% TFA at a flow rate of 10 mL
per minute. The Phenomenex Luna Cl8(2) column (250 x 21.2 mm, 10 pm
particle size) was connected in series with the Phenomenex Luna Cl8(2)
column (150 x 21.2 mm, 5 pm particle size). All 336 grams of sample were
dissolved in 1 mL MeOH, to give a load ratio of 210 mg silica to 1 mg of
sample. The resulting chromatogram is shown in Figure 16.
Figure 16. Preparative Purification of Compound 1.
48


One fraction was collected from 168 minutes to 169 minutes. In order to
remove the TFA from the sample, open column flash chromatography was
used. The column was packed with 15 grams of 50 pm Bakerbond C18 silica
which had been pre-wetted with 100 mL MeOH and conditioned with 300 mL
of a mixture of 84% H2O to 16% MeOH. After addition of the diluted fraction
to the column, water was added until the TFA had been removed. The
adsorbed compound was then eluted using 100 mL of MeOH, and dried using
a Buchi Rotavapor with the temperature bath set at 20 C. The dried sample
weighed 132 mg.
6.2.6.3 Final Purification of Compound 2
The chromatographic conditions described in the previous section
were used to purify compound 2 (labeled as CDXA-37.89.8), with the
exception that the mobile phase was changed to 65% methanol and 35% water
containing 0.05% TFA. Using these conditions, 1 mL of MeOH containing
200 mg of compound 2 was injected, to give a load ratio of 353 mg silica to 1
mg of TDS. The resulting chromatogram is shown in Figure 17.
49


Figure 17. Preparative Purification of Compound 2.
One fraction was collected from 98 minutes to 112 minutes, and the TFA was
then removed as described in section 6.2.6.2. After drying in the Buchi
Rotavapor, the sample weighed 72.7 mg.
6.3 Characterization of Isolated Compounds
6.3.1 HPLC-DAD Analysis of Compound 1
The purity of compound 1 and and compound 2 was determined using
the Agilent 1100 series HPLC system equipped with diode array detection
(DAD). The method utilized a Waters Symmetry C18 (150 x 4.6 mm, 3.5 pm
particle size) equilibrated at 20 C and a flow rate of 0.5 mL/min. Mobile
50


phase A consisted of 0.025% TFA in H2O, and mobile phase B consisted of
0.025% TFA in methanol. The mobile phase conditions began at 60% B
increasing to 65% B over 30 minutes, then holding at 65% B 15 minutes, then
increasing to 85% B over 5 minutes, and holding isocratic for 5 minutes. A 5
pL of a 1 mg/mL solution of escin la dissolved in methanol was injected and
the result is shown in Figure 18.
Figure 18. HPLC Chromatogram of Compound 1
The results show that the major component in the chromatograph to be 91 %
pure based on UV absorbance at 220 nm. The UV profile of this peak, taken at
34.8 minutes, is shown in Figure 19.
51


Figure 19. UV Spectrum at RT = 34.8 Minutes.
The UV absorption spectrum seen in Figure 19 shows two broad absorbance
shoulders at 200 nm and at 220 nm. This type of UV absorption spectra is
characteristic of compounds that lack chromophores with strong molar
absorptivities.
Peak purity was also determined using the Agilent ChemStation
software. This technique ensures that there are no co-eluting impurities hidden
under the peak of interest by comparing the differences between individual
spectra and the average spectrum across the peak. A purity value can be
assigned by comparing the similarities between the spectra, and when the
differences between spectra are small the peak is considered to be pure. The
peak purity was determined to be 99.8%, and the plot of the chromatographic
peak, similarity curve, and threshold curve is shown Figure 20.
52


Figure 20. Peak Purity Plot as Calculated for Peak at 34.8 Minutes.
The peak purity plot seen in Figure 20 is typically seen for chemically pure
chromatographic peaks. The similarity curve is determined by evaluating the
spectral differences between all spectra throughout the peak, which are seen to
remain close to zero from 34.3 to 35.1 minutes. The threshold curve is a
graphic display of the increase in spectral differences caused by background
noise, which is seen to increase only at the front and back of the peak, as
expected for a chemically pure peak. The peak purity curve, representing the
difference between the similarity and threshold curves, is shown to remain
close to unity throughout the peak. Therefore, the peak eluting at 34.8 minutes
in Figure 18 was determined to be chemically pure.
53


63.2 HPLC-DAD Analysis of Compound 2
A 5 pL injection of a 0.75 mg/ml of compound 2 was also analyzed
using the method described in section 6.3.1, and the result is shown in Figure
21.
Figure 21. HPLC Chromatogram of Compound 2
The results show that the major component in the chromatograph to be 93%
pure (based on UV absorbance at 220 10 nm). The UV profile of the peak at
38.9 minutes was also taken, as shown in Figure 22.
54


Figure 22. UV Spectrum at RT = 38.9 Minutes
The absorption spectrum shown in Figure 22 is similar to that obtained from
the compound 1, and shows the two broad absorbance shoulders at 200 nm
and at 220 nm, which is also characteristic of compounds that lack
chromophores with strong molar absorptivities.
The purity of the peak eluting at 38.9 minutes was calculated to be
99.9%, and a plot of the chromatographic peak, similarity curve, and threshold
curve is shown Figure 23.
55


Figure 23. Peak Purity Plot as Calculated for Peak at 38.9 Minutes.
The peak purity plot seen in Figure 23 is similar to the purity plot generated
for compound 1, and also represents the plot of a chemically pure
chromatographic peak seen in Figure 21 with a RT of 38.9 minutes.
6.33 Mass Spectrometry Analysis
The MS data for compound 1 were obtained by infusing a 1 mg/mL
solution in methanol into the LCQ-Deca ion-trap equipped with the
electrospray ionization source in positive mode. Because positive electrospray
ionization (ESI) is a soft ionization technique, the mass spectrum obtained
shows a strong signal generated by the molecular ion peak associated with
sodium. However, fragmentation of the molecule can occur in the source, and
these fragments are also observed as sodium adducts. When the ESI source is
combined with an ion trap mass spectrometer, it becomes possible to perform
56


MS" experiments, whereby the parent ion is isolated, and then sequentially
fragmented. The data obtained from these MS" experiments are then used as
an aid in structural determination.
The electrospray ionization (ESI) source was used with a sheath gas
flow rate of 60 (arbitrary units) and an auxiliary gas flow rate set to 0
(arbitrary units). The spray voltage was set to 5 KV, the capillary temperature
was set to 350 C, and the capillary voltage was set to 7 volts. The MS
conditions are summarized in Table 2.
Table 2. MS Conditions
Source Conditions Set Points
Mode and Polarity ESI Positive
Sheath Gas Flow Rate (arb) 60
Aux. Gas flow Rate (arb) 0
Spray Voltage (KV) 5.00
Capillary Temp (C) 350.00
Capillary Voltage (KV) 7.00
Tube Lens Offset (V) 0.00
Initially, the mass spectrometer was set to scan the full range between 100 and
2000 amu, with a scan time of 3 microscans and a maximum injection time of
200 ms using automatic gain control.
633.1 MS" Analysis of Compound 1
Using the MS setting described above, a 1 mg/mL solution of
compound 1 was analyzed. The spectra show a strong signal at m/z 1153.46,
which is characteristic of the sodium adduct of several escin isomers that have
57


molecular weights of 1130.5, including escin la. The mass spectrum of
compound 1 is shown Figure 24.
Figure 24. ESI+ Infusion of Compound 1 at 10 pL/min
The scan setting was changed to isolate the mJz 1153.5 ([M+Na]*) in the ion
trap, where a radio frequency corresponding the isolated molecule is applied.
This causes increased collisions with the helium damping gas used to focus
ions in the center of the trap. This process is referred as collision induced
disassociation (CID), the extent of which can be controlled by adjusting the
intensity of the resonating frequency and the time it is applied. After isolating
mJz 1153.5 with an isolation width of 1, normalized collision energy of 27%
was applied for 30 milliseconds. The result is shown in Figure 25.
58


The spectrum shows the isolated sodium adduct of the molecular ion at m/z
1153.5, with strong signals at m/z 991.4,653.4, and 523.5. The signal at m/z
991.4 is indicative of the sodium adduct of the neutral fragment generated
through the loss of a terminal deoxyhexose sugar group. The other two
fragments at m/z 523.1 and 653.4 are generated by the fiagmentation of the
trisaccharide chain and the protoescigenin aglycone, respectively.
After isolating the signal at m/z 1153.5, the signal at 991.5 was
isolated using MS3, a normalized collision energy of 30%, and an activation
time of 30 millisecond. The result is shown in Figure 26.
59


Figure 26. Compound 1 ESI+ Infusion at 10 pL/min. MS3 of m/z 991.5
The mass spectrum shows a small signal at 991.47 followed by a stronger
signal at m/z 829.33 due to the loss of a second deoxyhexose group from the
isolated molecular ion. The signal at 653.3 is exactly 176 m/z units less than
the signal at 829.3, indicating the loss of a deoxyglucopyranuronic acid. The
fragmentation pattern indicates the presence of two terminal hexose sugars
attached to a glucuronic acid group, which is in turn connected to the
aglycone.
After establishing that the signal at m/z 653.3 is generated through
fragmentation of the major compound, the peak seen at m/z 523.5 was isolated
and fragmented, the result is shown in Figure 27.
60


The spectra show the isolated m/z 523.5 ion representing the sodium adduct of
the sugar chain, which is composed of three sugars, one glucuronic acid and
two glucose. The major signal at m/z = 505.06 represents the loss of water and
the signal at m/z = 361.07 is the further loss of a glucopyranosyl.
633.2 MS" Analysis of Compound 2
The initial full spectrum scan of Compound 2 shows a strong signal at
m/z 1153.42. This was determined to be the sodium adduct of a molecule with
a mass of 1130.55 amu, corresponding to a molecular formula of C55H86O24,
which is consistant with the molecular formulas of several known escin
isomers. The mass spectrum is shown in Figure 28.
61


Figure 28. ESI+ Infusion of Compound 2 at 10 pL/min
The mass spectrum of Compound 2 also contained strong signals at m/z
989.44,653.47, and 523.26. To determine which signals were generated
through fragmentation of the signal at m/z 1153.42, and were not due to the
presence of contaminants, the molecule responsible for the signal at m/z
1153.42 was isolated in the ion trap and fragmented successively using the
inherent MS" of the instrument. Using an isolation width of 1.0 m/z, a
normalized collision energy of 27%, and an activation time of 30
milliseconds, the fragmentation pattern shown in Figure 29 was obtained.
62


Figure 29. ESI+ Infusion of Compound 2 at 10 pL/min. MS2 of m/z 1153.5
Fragmentation of the signals at m/z 1153.4 generated fragments at m/z 991.3,
653.36, and 523.1.These signal at m/z 991.3 was generated through the loss of
a single terminal glucose sugar from the trisaccharide chain.
The signal at 653.4 was generated by the aglycone after fragmentation
of the entire trisaccaride moiety, while the signal at m/z 523.1 was generated
by the trisaccaride fragment. The signal at m/z 653.4 was isolated and
fragmented using MS3 as shown in Figure 30.
63


51707^0 4000- JOOO- 3000^ 5400- 3MO-; MWK 2500 2000- 2400- 2200- t {mo- 1000- 1400- 1200- ooo- 000- 200- 01-31 ITT 0 03-0 >( M. 11? mm 31151 50037 00 } 50037 00 1175 06-1000 GO; 200^2 403-04 10002 j 221 JO 274.02 2J2 | 011.22 544.00 SO 12 053 03205 05 55200
200 200 300 200 400 400 500 000 OOO 000 TOO 750
Figure 30. Compound 2 ESI Infusion at 10 pL/min MS3 of mh 653.36
The mass spectrum shows the sodium adduct of the aglycone at m/z 653.36 as
well as two strong signals at mJz 593.31 and m/z 553.32. The signal at m/z
593.31 represents the loss of a CH3COOH group from the aglycone skeleton,
while the signal at 553.32 represents the loss of a tigloyl or angeloyl group
from the aglycone.
The signal at m/z 523.10 was isolated and fragmented, showing a
similar pattern as the fragments obtained from the isolation of this signal
during the analysis of compound 1 as shown in Figure 31.
64


The fragmentation pattern of the isolated mass at m/z 535.10 is exactly the
same as the fragmentation pattern for escin la, confirming that both
compounds contain a sugar chain composed of one glucuronic acid and two
glucose. The major signal at m/z = 505.06 represents the loss of water and the
signal at m/z = 361.07 is the further loss of a glucopyranosyl moiety.
63.4 Nuclear Magnetic Resonance Spectroscopy
Because saponins are large molecules containing multiple hydroxyl
groups, the one dimensional !H NMR spectrum is complex, and characterized
by numerous proton resonances with overlapping of the monosaccharide sugar
protons between 4 and 5 ppm. Therefore, classical methods of elucidating the
molecular structure of saponins necessarily began with acid hydrolysis, which
was followed by NMR analysis of the aglycone and sugars separately.
65


However, with the advent of modem NMR techniques, including two
dimensional correlation spectroscopy (COSY) and ^-detected heteronuclear
multiple-bond correlation (HMBC), structural elucidation has become
possible using the intact molecule. The first reported analysis of individual
escins using these modem NMR techniques was reported in 1994 by
Yoshikawa et al.38 Consequently, the structural confirmation of several escin
isomers can be accomplished by comparison of one dimensional C NMR
and 'H NMR to these previously published results.
6.3.4.1 NMR Analysis of Compound 1
The H NMR and 13C NMR spectra of compound 1 were obtained
using a Varian 500 MHz instrument, and the data were processed using NUTS
software. As seen in Figure 32, the *H NMR spectrum of compound 1 is
Figure 32. H NMR of Compound 1
66


However, the broad singlet at 5.39 ppm is characteristic of the H-12 olefinic
methine proton, which is indicative of the basic aglycone skeleton of an
oleanane (P-amyrin) triterpene.39 Also, tigloyl and acyl substitution was
indicated by the presence of two signals, one at 86.6 (1H, d, 21-H) and the
other at 5 6.3 (1-H, d, 22-H), as shown in Figure 33.
There are also several distinctive signals from P-amyrin methyl groups at
80.65 (3H, s, Me-25) and 80.79 (3H, s, Me-26).40 The strong signal at 81.34
can also be attributed to several methyl groups on the P-amyrin skeleton (6H,
s, Me-29 and Me-30). The molecular structure of tiglic acid and the *H NMR
spectrum of compound 1 from the 0.4 to 2.5 ppm region is shown in Figure
34.
67


Figure 34. The Molecular Structure of Tiglic Acid and the H NMR Spectrum of Compound
1 From the Region of 0.4 to 2.5 ppm
Tiglic acid was confirmed by the presence of the signals at 87.13 (1H, q, 3-
H) in Figure 33, as well as the signals at 1.97 (3H, s, 5-H3), and 1.65 (3H, d,
4-H3)41 shown in Figure 34. The data from the *H NMR spectra, when
compared to the literature data for isolated escin compounds, were found to be
in close agreement with the data for the known compound Escin la.38
The 13C NMR spectrum of compound 1 in pyridine-ds is shown in
Table 3.
68


Table 3. 13C NMR Data for Compound 1
Position (8, ppm) Position (8, ppm)
Sapogenol moiety Sugar moiety
C-l 38.9 3-O-fV-D-glucuronic acid
C-2 27.1 C-l' 105.2
C-3 91.6 C-2' 80.1
C-4 442 C-3' 76.9
C-5 56.6 C-4' 82.3
C-6 19 C-5' 76.2
C-7 33.7 C-6' 172.4
C-8 40.4 2'-0-p-D-glucopyranosy)
C-9 471 C-l" 104.7
C-10 37 C-2" 76.2
C-ll 24.5 C-3" 78.6
C-12 123.6 C-4" 70.2
C-13 143.4 C-5" 78.8
C-14 42.2 C-6" 62.1
C-15 35.1 4'-0-P-D-glucopyranosyl
C-l 6 68.5 c-r 105.2
C-17 48.4 C-2" 75.3
C-18 40.6 C-3" 78.6
C-19 47.7 C^' 72
C-20 36.8 0 1 l/l 3 79
C-21 79.9 C-6" 62.9
C-22 74.7 21-O-acyl moiety
C-23 22.9 Tigloyl C-l"" 168.6
C-24 63.8 Tigloyl C-2"" 130
C-25 16 Tigloyl C-3"" 137.5
C-26 17.2 Tigloyl C-41" 14.8
C-27 27.9 Tigloyl C-5"" 13
C-28 64.2 22-O-acetyl moiety
C-29 30.1 C-l"" 171.5
C-30 20.7 C-2" 21.4
The presence of a double bond at the 12-position on the oleanane skeleton was
supported by the presence of a signal at 8143.4, corresponding to a quaternary
olefinic carbon, and a signal at 8123.6, corresponding to a methine olefinic
69


carbon. The presence of four hydroxylated methines were indicated by the
signals at 891.6 (C-3), 68.5 (C-16), 79.9 (C-21), and 74.7 (C-22), as well as
signals for two hydroxylated methylenes at 863.8 (C-24) and 64.2 (C-28). All
of these signals are characteristic of a protoaescigenin aglycone.
The identity of compound 1 was determined by combining the data
obtained from the HPLC-UV analysis, the MS molecular mass determination
and fragmentation patterns, as well as the *H NMR and 13C NMR spectra. The
molecular mass of compound 1 was determined to be 1130.55 amu,
corresponding to a molecular formula of C55H86O24, which is consistent with
the molecular formulas of several known escin isomers, one of which is escin
la. In addition, the 'H NMR and 13C NMR spectra of compound 1 obtained in
the experiment were similar to the spectral data for escin la found in the
literature. Therefore, on the basis of foregoing evidence, the sample labeled as
compound 1 was determined to be escin la: 21-tigloyl-22-
acetylprotoescigenin-3-0-[fl-D-glucopyranosyl-( 1 >2)]-[P-D-glucopyranosyl-
(14)]-P-D-glucuronic acid pyranoside.
63.4.2 NMR Analysis of Compound 2
The 'H NMR spectra of compound 2 was similar to theH NMR
spectrum obtained from compound 1, Showing multiple overlapping hydroxyl
signals in the region between 4 and 5 ppm, and angeloyl and acyl substitution
70


of the aglycone was indicated by the presence of two signals, one at 86.6 (1H,
d, 21-H) and the other at 86.2 (1-H, d, 22-H) as shown in Figure 33.
Figure 35. H NMR of Compound 2
Angelic acid was confirmed by the presence of the signals at 86.00 (1H, q,
3-H) shown in figure 33, and the signals at 82.03 (3H, s, 5-H3), and 2.12
(3H, d, 4-H3)41 shown in Figure 36.
71


Figure 36. The Molecular Structure of Angelic Acid and the *H NMR Spectrum of
Compound From the Region of 0.4 to 2.6 ppm
11
The C NMR spectrum of compound 2 in pyridine-^ was also
consistent with the structure observed in the *H NMR spectrum, as shown in
Table 4.
72


Table 4.13C NMR Data for Compound 2
Position (8, ppm) Position (5, ppm)
Sapogenol moiety Sugar moiety
C-l 38.9 3-O-P-D-glucuronic acid
C-2 27 C-l' 105.2
C-3 91.6 C-2' 80.1
C-4 44.2 C-3' 76.9
C-5 56.6 C-4 82.3
C-6 19 C-5' 76.2
C-l 33.7 C-6' 172.4
C-8 40.4 2'-0-|)-D-glucopyranosyl
C-9 47.2 C-l 104.8
C-10 36.8 C-2" 76.2
C-ll 24.5 C-3" 78.6
C-I2 123.6 C-4 70.2
C-13 143.4 C-5 78.8
C-14 42.2 O b, 3 62.1
C-15 35.7 4'-0-f)-D-glucopyranosyl
C-l 6 68.5 c-r 105.2
C-17 48.5 C-2 75.4
C-18 40.6 C-3" 78.6
C-19 47.6 C-4' 72
C-20 36.8 C-5m 79
C-21 79.4 C-6" 62.9
C-22 74.8 21-0-acyl moiety
C-23 22.9 Angeloyl C-l1 168.4
C-24 63.8 Angeloyl C-2 129.5
C-25 16 Angeloyl C-3 137.7
C-26 17.2 Angeloyl C^ 16.4
C-27 27.9 Angeloyl C-5" 21.6
C-28 64.2 22-O-acetyl moiety
C-29 30 C-l"" 171.4
C-30 20.8 C-2"" 21.4
73


The presence of a double bond at the 12-position on the oleanane skeleton was
supported by the presence of a signal at 5143.4, corresponding to a quaternary
olefinic carbon, and a signal at 5123.6, corresponding to a methine olefinic
carbon. The presence of four hydroxylated methines at 591.6 (C-3), 68.5 (C-
16), 79.4 (C-21), and 74.8 (C-22), as well as two hydroxylated methylenes at
563.8 (C-24) and 64.2 (C-28). All of these signals are characteristic of a
protoaescigenin aglycone
The identity of compound 2 was based on the UV profile, the MS data,
and the data from the 1H NMR and 13C NMR spectra. The molecular mass of
compound 2 was determined to be 1130.42 amu, which corresponds to a
molecular formula of C55H86O24. The identity of compound 2 was further
supported by the *H NMR and 13C NMR spectra which were in agreement
with the literature values for escin lb.41 Therefore, on the basis of foregoing
evidence, the sample compound 2 was identified as escin lb: 21-angeloyl-22-
acetylprotoescigenin-3-0-[p-D-glucopyranosyl-(l2)]-[P-D-glucopyranosyl-
(1 *4)]-P-D-glucuronic acid pyranoside.
74


6.4 Qualitative and Quantitative Analysis of Herbal Products
Containing HCSE
Many herbal products containing Horse Chestnut seed use an extract
standardized to 20% escins, which is a saponin mixture composed of
approximately thirty structural isomers. Current quality control methods that
monitor for escins in herbal products utilize gravimetric testing, photometric
determination, or TLC analysis. Using the fragmentation patterns observed for
escin la and escin lb, a qualitative HPLC/ESI/MS method was developed to
determine a common fragment ion or ions, and these common ions were used
to quantitate escins in herbal products.
6.4.1 Qualitative Analysis
The purpose of the qualitative HPLC method was to separate each
escin isomer as fiilly as possible, therefore, two Waters Symmetry Cl8
columns (150 x 4.6 mm, 3.5 pm particle size) were connected in series. The
columns were equilibrated at 30 C, and the mobile phase flow rate was set at
0.5 mL/min. Mobile A consisted of H2O with 0.1% formic acid and mobile
phase B consisted of ACN with 0.1% formic acid. The initial mobile phase
conditions started at 30% B and changed linearly to 50% B over 40 minutes,
held at 50% B for 5 minutes, then returning to 40% B over 5 minutes, and
then held constant for 5 minutes in order to re-equilibrate the HPLC system
.75


and the analytical columns to the initial conditions. The gradient profile is
summarized in Table 5.
Table 5. HPLC Conditions
Time, minutes % A %B
0 60 40
40 50 50
45 50 50
50 60 40
Mass spectrometer conditions were optimized by continuously infusing a 0.02
mg/mL solution of P-escin, dissolved in methanol, at 10 pJL/min. into a three
way T-connecter with the HPLC mobile phase at a flow of 0.5 mL/min. The
MS conditions are summarized in Table 6.
Table 6. MS Conditions
Source Conditions Set Points
Mode and Polarity ESI Positive
Sheath Gas Flow Rate (arb) 80
Aux. Gas flow Rate (arb) 20
Spray Voltage (KV) 5.00
Capillary Temp (C) 350.00
Capillary Voltage (KV) 7.00
Tube Lens Offset (V) 0.00
The liquid chromatography-mass spectrometry (LC-MS) method was
used to analyze a 1 pL injection of a 1.6 mg/mL solution of P-escin, which
shows multiple peaks, each one corresponding to the molecular weight of a
76


known escin compound. The total ion chromatogram (TIC) is shown in Figure
37.
Figure 37. TIC Chromatogram of 0-escin
The TIC shows over 25 individual peaks that have mass spectra characteristic
of escin saponins. Each spectrum was characterized by the sodium adduct of
the molecular ion, as well as the two characteristic sodium adducts of the
sugar and sapogenin fragments. Representative spectra are shown in Figures
38 through 42. The mass spectrum of the peaks at 10.63 and 11.65 minutes is
shown in Figure 38.
77


Figure 38. Mass Spectrum of Peaks with RTs of 10.63 and 11.65 Minutes
The spectrum shows a strong signal at m/z =1113.43. The two fragment ions
at m/z 613.5 and m/z 523.5 are indicative of the aglycone and sugar fragments,
respectively. The sodium adduct of the aglycone is seen at m/z = 613.5,
indicates that the aglycone is doubly acylated, and the signal at m/z 523.5
indicates that the polysaccharide chain is composed of a two glucopyranoside
residues and a glucuronic acid.34
78


The mass spectrum of the peaks at 20.60 and 25.39 minutes is shown in
Figure 39.
I1ID.'7K2 T ctSIFijll 480000- id 6-633 RT 20 90-20 69 AV 6 SO 144 16 7 Til ( 100 00 7000 00, sz U7 417 m 2003 21 13-22 66 NI116E6 114 36 Mta B7B47 Hi i -TrrJ b" 4B 1140 4B 114140 1173 IB | T55
n3MM04MM040D700400MO 1MB 110O 1M 1300 1MB 1B00 1B0D 1700 1MB 1000 3000
Figure 39. Mass Spectrum of Peaks with RTs of 20.60 and 23.4 Minutes
The spectrum shows a strong signal at m/z = 1141.5. The two fragment ions at
m/z 641.5 and m/z 423.5 are indicative of the aglycone and sugar fragments,
respectively. The sodium adduct of the aglycone is seen at m/z = 641.5,
indicates that the aglycone is esterified by isobutyric acid at the C 21 position,
and the signal at 523.28 indicates that the polysaccharide chain is composed
of a two glucopyranoside residues and a glucuronic acid. The mass spectrum
of the peaks at 23.00,25.39,26.65, and 28.70 minutes is shown in Figure 40.
79


Figure 40. Mass Spectrum of the Peaks with RTs of23.0, 23.4, 26.65,28.70 Minutes
These spectrum shows a strong signal at m/z = 1123.42. The two fragment
ions at m/z 653.49 and m/z 493.17 are indicative of the aglycone and sugar
fragments, respectively. The sodium adduct of the aglycone is seen at m/z =
653.5, indicates that the aglycone is esterified at the C 21 position by either
tiglic or angelic acid, and the signal at m/z 493.17 indicates that the
polysaccharide chain is composed of a glucose, a xylose, and a glucuronic
acid. The mass spectrum of the peaks at 26.78,27.60,29.44, and 30.38
minutes is shown in Figure 41.
80


Figure 41. Mass Spectrum of the Peaks with RTs of26.78,27.60,29.44, and 30.38 Minutes
The spectrum shows a strong signal at m/z = 1137.47, indicating a molecular
weight of 1114.47. The two fragment ions at m/z 637.48 and m/z 523.25 are
indicative of the aglycone and sugar fragments, respectively. The sodium
adduct of the aglycone is seen at m/z = 637.48, indicates that the aglycone is
derived from barringtogenol C skeleton that is esterified with either tiglic acid
or angelic acid at the C-21 position and acylated at the C-22 position, and the
signal at m/z 523.25 indicates that the polysaccharide chain is composed of a
two glucopyranoside residues and a glucuronic acid. The mass spectrum of the
peaks at 46.5 and 47.9 minutes is shown in Figure 42.
81


Figure 42. Mass Spectrum of Peaks with RTs of 46.5 and 47.9 Minutes.
The spectrum shows a strong signal at m/z = 1193.46, indicating a molecular
weight of 1170.46. The two fragment ions at m/z 693.46 and m/z 523.22 are
indicative of the aglycone and sugar fragments, respectively. The sodium
adduct of the aglycone is seen at m/z = 693.46, indicates that the aglycone is
esterified at the C-21 and the C-22 by either tiglic acid or angelic acid, and the
signal at m/z 523.25 indicates that the polysaccharide chain is composed of a
two glucopyranoside residues and a glucuronic acid.
6.4.1.1 Result of Qualitative Analysis
The analysis of the mass spectra for each of the peaks seen in Figure
37, including the strongest signal observed as well as the major fragment ions,
is summarized in Table 7.
82


Table 7. P-escin Peak Analysis
Retention Time, min. [Aglycone + Na]+ Aglycone Molar Mass, amu [Trisacch aride + Na]+ Trisacc haride Molar Mass, amu [M + Naf Molar Mass, amu
10.4,11.4 613.5 590.5 493.3 470.3 1083.4 1060.5
10.6,11.7 613.5 590.5 523.5 500.5 1113.4 1090.5
12.01 613.5 590.5 523.5 500.5 1097.5 1074.5
17.4,18.9, 20.1 641.5 618.5 493.3 470.3 1111.5 1088.5
20.6,23.4 641.5 618.5 523.5 500.5 1141.5 1118.5
23.0,23.4, 26.7,28.7 653.5 630.5 493.3 470.3 1123.5 1100.5
23.6,26.0, 26.9,29.0 653.5 630.5 523.5 500.5 1153.5 1130.5
23.1,25.4, 26.7,28.7 653.5 630.5 493.5 470.3 1123.5 1100.5
23.6,26.0, 26.9,29.0 653.5 630.5 523.5 500.5 1153.5 1130.5
26.7,27.6, 29.4,30.4 637.5 614.5 523.5 500.5 1137.5 1114.5
46.5,47.9 693.5 670.5 523.2 500.2 1193.5 1170.5
The qualitative LC-MS analyses of P-escin shows that although the
mixture is composed of over thirty isomers with at least 10 different molecular
weights, each individual escin has a polysaccharide chain composed of either
two glucopyranoside residues and a glucuronic acid, or a glucose, a xylose,
and a glucuronic acid with a molecular weight of either 500.5 or 470.5 amu
respectively. Therefore, these common fragments can be used to quantify
escins in herbal products.
83


6.4.2 Quantitative Analysis
An HPLC/ESI/MS method has been developed to quantitate the total
escin content in herbal products containing HCSE. The method was designed
to be used in a quality control laboratory using commercially available
reagents and common instrumentation. Commercially available (3-escin was
used as the reference standard, and in order to ensure the analysis of numerous
samples in a short period of time, the extraction procedure was designed as
simply as possible with no clean up procedure prior to LC-MS analysis.
The standards and samples were weighed directly into volumetric
flasks, MeOH was added, the solution was sonicated for 30 minutes, and
finally filtered through a 0.45 pm filter prior to analysis. Isopropyl alcohol
was added to the mobile phase to prevent bacterial growth, and to decrease the
column re-equilibration time required when using gradient chromatography.
TFA was chosen as the acid modifier and ACN was chosen as the organic
component of the mobile phase because this combination gave sharp,
symmetrical escin peaks that were partially separated from the other
components of the herbal mixture, but were not resolved from each other.
Instead, all of the escins were forced into a small region of the overall method.
Although the individual escin compounds were not well resolved using this
methodology, it enabled the integration of the escins peaks as a group.
84