MASS SPECTROMETRIC PROFILING AND LOCALIZATION OF
METABOLITES IN BIOLOGICAL SAMPLES
By
CHAO LI

A DISSERTATION
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
DOCTOR OF PHILOSOPHY
Chemistry
2011

ABSTRACT
MASS SPECTROMETRIC PROFILING AND LOCALIZATION OF
METABOLITES IN BIOLOGICAL SAMPLES
By
CHAO LI
Metabolomics is a rapidly growing approach based on global analysis of all metabolites in a
tissue, fluid or organism.

This approach is based on the assumption that the metabolome serves

as a holistic measure of metabolic status.

Most metabolomic analyses have been based on bulk

sample analyses, which do not differentiate metabolites from specific tissues or cells.

Since

metabolic activity is organized in complex networks, bulk analyses involve a loss of potentially
important chemical information.

This dissertation presents development of analytical methods

to profile localized metabolites in biological tissues, and applies these to investigations of
glandular trichomes, which are epidermal plant cells that are prolific at chemical synthesis.
Microsampling of individual trichomes from wild tomato relatives of the genus Solanum was
achieved using micropipettes, and blasting of tissue from surfaces of corals was performed using
a waterpik.

Metabolite detection, identification, and quantification were performed using liquid

chromatography/mass spectrometry (LC/MS). Through accurate mass measurements and
chromatographic peak alignments, 109 metabolite signals were annotated from extracts of single
trichome types.

Methylated flavonoids represented one class of trichome metabolites, but mass

spectra alone were not adequate to distinguish isomers.

Complete assignment of isomeric

methylated myricetins was achieved by employing two recombinant O-methyltransferase

enzymes from Solanum habrochaites LA1777 for selective semisynthesis of methylated
myricetins.

Tandem mass spectrometric analysis and chromatographic retention times provided

evidence needed for structure annotation, aided by selective enzymatic incorporation of
deuterium labeled methyl groups.

These structural assignments aided demonstration of

substrate selectivity of the two methyltransferases.
To accelerate profiling of individual trichomes, a contact printing approach was developed
to transfer cells from across an entire leaf to a carbon substrate while preserving spatial
information during the transfer.

Direct and matrix-free laser desorption ionization from the

printed cells allowed chemical imaging of trichomes across an entire leaflet.

Individual

trichome resolution was achieved using smaller steps of stage position, yielding about 50 µm
resolution.

Using this scheme, the first spatially-resolve profiling of trichome metabolites

across the Solanum leaflet was achieved, and demonstrated chemical heterogeneity among
trichomes within a single type.
Localized metabolite profiling was also performed for reef coral tissues collected from a
field location near Curacao.

A stable isotope dilution method was developed for quantifying

seven betaine metabolites, which may serve as photosystem stabilizing substances, and
dimethylsulfoniopropionate (DMSP) in extracts of coral tissues.

The LC-MS protocol

employed a pentafluorophenylpropyl (PFPP) column and time-of-flight mass spectrometry.
The results demonstrate that corals accumulate multiple betaines, that betaine profiles vary
widely among different coral species, and that betaine levels are correlated with ambient light
levels.

ACKNOWLEDGEMENTS

This dissertation would be impossible without the contribution from many others.
difficult to mention them all.

It is

I will first express my sincere thanks to Dr. A. Daniel Jones, my

advisor, for his guidance, influence and support.

He has guided me through the ups and

downs of my graduate career and personal life.

I am so lucky to have such a wonderful

mentor who I will continue to benefit from for the rest of my career.
I would also like to thank my collaborators Dr. Robert Last’s group from Michigan State
University, Dr. Eran Pichersky’s group from the University of Michigan and Dr. David Gang’s
group from the University of Arizona.

It is honor to learn from them and work with them.

I am also grateful to my committee members, Dr. Gavin Reid, Dr. Merlin Bruening, and
Dr. Marcos Dantus for their constructive comments and insightful suggestion on my work.
I would like to extend my appreciation to many past and present members of Jones’s group
for their engaging discussion and their friendship, which bring joy and happiness to my life at
Michigan State University.
Completion of this dissertation would have been difficult without the moral and financial
support from my family, my parents, my parents-in-law, and my sister.

I’d like to give special

thanks to my wife Jiamin, who has tolerated the long distance separation for years but been
standing by me throughout my graduate career and has been a constant source of
encouragement and support.

iv

TABLE OF CONTENTS

LIST OF TABLES …………………………………………………………………...………. viii
LIST OF FIGURES …………………………………………………………………….…….. x
LIST OF SCHEMES …………………………………………………………………..…… xx
CHAPTER ONE
INTRODUCTION ......................................................................................................................... 1
1.1 Metabolite profiles and the quest to understand biological functions .................. 1
1.2 The Application of Metabolomics in Plant Research .............................................. 5
1.3 Analytical platforms for metabolomics ..................................................................... 8
1.4 Spatially resolved metabolite profiling, and the emerging interest in------micrometabolomics ................................................................................................... 12
1.5 Summary of Research ................................................................................................... 17
CHAPTER TWO
ANNOTATION OF SPECIALIZED METABOLITES ACCUMULATING IN INDIVIDUAL
TYPES OF GLANDULAR TRICHOMES OF THE GENUS SOLANUM .................................. 21
2.1 Specialized Metabolites in Plants ................................................................................. 21
2.2 Glandular Trichomes in the Genus Solanum ................................................................ 22
2.3 Challenges of Analysis of Trichome Specialized Metabolites ..................................... 25
2.4 Experimental Section .................................................................................................... 28
2.4.1 Materials ............................................................................................................. 28
2.4.2 Chemicals ............................................................................................................ 29
2.4.3 Trichome Collection ........................................................................................... 29
2.4.4 LC-MS Analysis ................................................................................................. 30
2.4.5 Data Processing ................................................................................................... 31
2.5 Results and Discussion ................................................................................................. 31
CHAPTER THREE
CHEMICAL IMAGING OF TRICHOME SPECIALIZED METABOLITES USING
CONTACT
PRINTING
AND
LASER
DESORPTION/IONIZATION
MASS
SPECTROMETRY ....................................................................................................................... 89
3.1 Introduction ................................................................................................................... 89
3.2 Materials and Methods .................................................................................................. 97
3.2.1 Chemicals ............................................................................................................ 97
3.2.2 Plants ................................................................................................................... 97

v

3.2.3 Sample Preparation ............................................................................................. 97
3.2.4 Laser Desorption Ionization Imaging ............................................................... 100
3.2.5 LC/MS analysis ................................................................................................. 101
3.3 Results and Discussion ............................................................................................... 102
3.4 Conclusions ................................................................................................................. 138
CHAPTER FOUR
CHARACTERIZATION OF METHYLATED FLAVONOID REGIOISOMERS USING
LC/MS/MS COUPLED WITH SEMISYNTHETIC ENZYMATIC METHYLATION ............ 140
4.1 Introduction ................................................................................................................. 140
4.2 Experimental Methods ................................................................................................ 147
4.2.1 Material ............................................................................................................. 147
4.2.1.1 Plant material .......................................................................................... 147
4.2.1.2 Chemicals ................................................................................................ 147
4.2.2 Enzyme cloning and expression........................................................................ 148
4.2.3 O-methyltransferase assay ................................................................................ 148
4.2.3.1 General methodology .............................................................................. 148
4.2.3.2 Individual enzyme assays ....................................................................... 149
4.2.3.3 Coupled enzyme assays .......................................................................... 149
4.2.4 LC/MS............................................................................................................... 149
4.3 Results and Discussion ............................................................................................... 150
4.3.1 Retention time rules .......................................................................................... 154
4.3.2 Characterization of enzymatic semisynthesis products .................................... 156
4.3.2.1 Myricetin as substrate ............................................................................. 156
4.3.2.2 3’-methylmyricetin as methyltransferase substrate ................................ 178
4.3.2.3 3’, 5’-methylmyricetin as methyltransferase substrate ........................... 198
4.3.2.4 3’,4’,5’,-trimethylmyricetin as methyltransferase substrate ................... 205
4.3.3 The origin of hydrogen radical loss from protonated O-methylflavonoids .......211
4.3.4 Application of annotation rules for identification of methylated myricetin-metabolites in S. habrochaites LA1777 trichomes ........................................... 216
4.4 Conclusions ................................................................................................................. 228
CHAPTER FIVE
DETERMINATION
OF
BETAINE
METABOLITES
AND
DIMETHYLSULFONIOPROPIONATE IN CORAL TISSUES USING LIQUID
CHROMATOGRAPHY/TIME-OF-FLIGHT MASS SPECTROMETRY AND STABLE
ISOTOPE-LABELED INTERNAL STANDARDS................................................................... 230
5.1 Introduction ................................................................................................................. 230
5.2 Materials and methods ................................................................................................ 236
5.2.1 Chemicals .......................................................................................................... 236
5.2.2 Collection and extraction of coral tissue samples ............................................. 236

vi

5.2.3 Internal standard synthesis, characterization and quantification ...................... 237
5.2.4 NMR characterization and quantification ......................................................... 238
5.2.5 LC-MS analyses of metabolites ........................................................................ 239
5.2.6 Preparation of calibration and quality control standards .................................. 240
5.2.7 Evaluation of analytical method performance .................................................. 241
5.3 Results and Discussion ............................................................................................... 243
5.3.1 Standard synthesis, characterization and quantification ................................... 243
5.3.2 Extraction of Metabolites from Coral ............................................................... 244
5.3.3 LC-MS analyses of betaines and DMSP ........................................................... 246
5.3.4 Method Performance and Validation................................................................. 255
5.4 Betaine in coral reef samples ...................................................................................... 261
CHAPTER SIX
CONCLUDING REMARKS ...................................................................................................... 267
REFERENCES .......................................................................................................................... 273

vii

LIST OF TABLES

Table II-1 List of wild tomato species and trichome types subjected to metabolite------profiling................................................................................................................................. 29
Table II-2 List of glycosylated flavonoids and methylated flavonoids detected in all trichome
extracts using LC/MS in negative ion mode. Listed fragment ions were generated using
aperture 1= 75 V aperture 1. ................................................................................................. 50
Table II-3 List of acylsugar metabolites detected in trichome extracts using LC/MS in
negative ion mode. Fragment ions were generated using aperture 1= 45 V. ..................... 54
Table II-4 List of terpenoid metabolites detected in extracts of individual trichome types
using LC/MS in negative ion mode. Fragment ions were generated using aperture 1= 75
V except for the sesquiterpene acids, which were analyzed as product ion spectra using a
QTof mass spectrometer with collision potential of 45 V. ................................................... 62
Table II-5 List of glycoalkaloid metabolites detected in extracts of individual trichome types
using LC/MS in positive ion mode. Fragment ions were generated using aperture 1 = 45
V. ........................................................................................................................................... 63
Table III-1 Mass spectra resolution (±SD, n=3) and S/N ratio (±SD, n=3) calculated by
averaging the resolution and S/N ratio of five most abundant acylsugar peaks from S.
habrochaites LA1777 extracts in positive mode and sesquiterpene acids (m/z 233) peak in
negative mode. Resolution and S/N for individual peak is taken from peak lists generated
using Launchpad software (version 2.8.1, Kratos Analytical). ............................................ 111
Table III-2 Major ions observed in laser desorption ionization (LDI) spectra of S.
habrochaites LA1777 trichome extracts on pencil lead-coated glass slides. ......................119
Table V-1 Mean recoveries of betaines and DMSP in spiked coral tissue extracts (n=4). ...... 246
Table V-2 Solvent composition used in the ternary gradient for elution of long-chain and
short-chain betaines on the pentafluorophenylpropyl silica column. Solvent A: water
adjusted to pH 3.85 with formic acid, solvent B：methanol, solvent C: 0.15% aqueous
formic acid + 10 mM ammonium acetate. .......................................................................... 255
Table V-3 Limits of detection (LODs) of betaines and DMSP determined in aqueous formic
acid and in coral tissue extract matrix. ............................................................................................ 256

viii

Table V-4 Accuracy and precision of quality control samples with addition of multiple
betaine metabolites and DMSP. .......................................................................................... 259
Table V-5 Experimental determination of matrix effect on determination of individual
betaines and DMSP in coral tissue extracts (n=3) .............................................................. 260
Table V-6 Concentrations (mean ± SE) in Madracis mirabilis collected from exposed
(unshaded) locations. Values for specific depths are listed only for compounds that
showed statistically significant differences in concentration as a function of depth. AlaB
= alanine betaine, βAlaB = β-alanine betaine, DMSP = dimethylsulfoniopropionate, GlyB
= glycine betaine, HProB = hydroxyproline betaine, ProB = proline betaine, TauB =
taurine betaine, Trig = trigonelline. n = 24, 8, 8, and 8 for the four columns. ................. 263
Table V-7 Concentrations (mean ± SE) in Madracis pharensis at 10 m. Values for exposed
and shaded colonies are listed only for compounds that showed statistically significant
differences in concentration as a function of exposure. n = 19, 10, and 9 for the three
columns. .............................................................................................................................. 264
Table V-8 Concentrations (mean ± SE) in Madracis senaria collected from exposed locations.
Values for colonies collected in morning (09:30-11:00) and afternoon (15:30-17:00) are
listed only for compounds that showed statistically significant differences in concentration
between the times.
n = 25, 7, and 18 for the three columns. Pheophytin and protein
−2
are in units of 100 × mg cm ............................................................................................. 265

ix

LIST OF FIGURES

Figure II-1 Trichomes on L. esculentum (A) and L. hirsutum (B). Roman numerals indicate
trichome type. Adapted from Reference [104]. ................................................................. 25
Figure II-2 Schematic Diagram of the Waters LCT Premier time-of-flight mass spectrometer.
The region indicated in the diagram is where the ion kinetic energies are influenced by the
potential applied to aperture 1. Adapted from Waters Micromass LCT Premier Mass
Spectrometer Operator’s Guide. For interpretation of the references to color in this and
all other figures, the reader is referred to the electronic version of this or dissertation........ 34
Figure II-3 The top 10 possible molecular formula for m/z 609.1414, a metabolite ion from S.
habrochaites LA1777, generated by Masslynx software (Waters). DBE (double bond
equivalent), i-Fit is expressed in arbitrary units, and measures the goodness of fit to the
isotopic pattern calculated for each elemental composition. Lower i-Fit values reflect a
superior fit. ............................................................................................................................ 37
Figure II-4 The 2 examples of 20 matching structures found on the ChemSpider database for
formula C27H30O16. ............................................................................................................. 38
Figure II-5 The 4 matching structures found on the ChemSpider database for formula
C28H26N4O12. ...................................................................................................................... 39
Figure II-6 Extracted ion LC/TOF MS chromatograms (XIC) for an extract of S.
habrochaites LA1777 leaf type VI trichomes (electrospray, negative): A) m/z 609 at low
CID potential (aperture 1 = 10 V); B) m/z 284 at high CID (aperture 1= 75 V); C) m/z 625
at low CID (aperture 1 = 10 V); D) m/z 300 at high CID (aperture 1=75 V). ................... 42
Figure II-7 The top 10 possible elemental formulas for fragment ion m/z 284.0340 obtained
by averaging spectra (7.5-7.9 min; aperture 1= 75 V) from LC/TOF MS analysis of an
extract of leaf type VI trichomes from S. habrochaites LA1777 (electrospray, negative
mode). ................................................................................................................................... 45
Figure II-8 A) Collision induced dissociation (CID) spectra of S. habrochaites LA1777 leaf
type VI trichome extract (retention time 7.89 min) using aperture 1= 75 V in negative
mode; B) Collision induced dissociation (CID) spectra of the same retention time using
aperture 1=75 V in positive mode; C) Proposed fragmentation for
kaempferol-3,7-O-diglucoside detected from LA1777 type VI trichome extracts in ESI
negative mode. ...................................................................................................................... 48

x

Figure II-9 Top panel: CID spectra of acylsugar S4:21 (75V aperture 1 voltage) from extract
of S. habrochaites LA1777 type I trichomes in ESI negative mode; bottom panel: proposed
(2,4,5,10)
fragment ions observed for S4:21
. .......................................................................... 52
Figure II-10 Top panel: LC/MS product ion spectrum of m/z 233 with retention time 30.77
min from S. habrochaites LA1777 type VI trichome extract (ESI, negative) generated on a
QTof mass spectrometer (45 V collision voltage); middle panel: LC/MS XIC of m/z 233
(aperture 1= 10 V) from LA1777 type VI trichome extracts in ESI negative mode, full scan
MS analysis; bottom panel: proposed fragmentation for α-santalenoic acid. ...................... 58
Figure II-11 Top panel: CID spectra of (tomatine) in ESI positive mode (aperture 1= 75 V)
from S. habrochaites LA1777 type VII trichome extract; bottom panel: proposed
fragmentation of protonated tomatine. .................................................................................. 64
Figure II-12 Quantitative distribution of glycosylated flavonoids in extracts of each type of
trichome (50 trichomes extracted in 50 µl solvent) from leaf and stem tissues of S.
habrochaites (LA1777) measured using negative mode LC/MS.
Key:
Kaempferol-dihexosides (red) and quercetin –dihexoside (yellow). Error bars indicate
standard error with n=3. ........................................................................................................ 68
Figure II-13 Quantitative distribution of methylated flavonoids in each types of trichome
(from 50 trichomes extracted in 50 µl solvent) from S. habrochaites (LA1777) measured
using negative mode LC/MS. Key: Dimethyl myricetin (blue) trimethyl myricetin (red)
tetramethyl myricetin (yellow). Error bars indicate standard error with n=3. ................... 69
Figure II-14 Quantitative distribution of terpenoid metabolites in each types of trichome
(from 50 trichomes extracted in 50 µl solvent) from S. habrochaites (LA1777) measured
using negative mode LC/MS. Key: sesquiterpene acids (blue), sesquiterpene diglucoside,
malonate ester (red), diterpene acid (yellow). Error bars indicate standard error with
n=3. ....................................................................................................................................... 70
Figure II-15 Quantitative distribution of total acylsucroses in each types of trichome (from
50 trichomes extracted in 50 µl solvent) from S. habrochaites (LA1777) measured using
negative mode LC/MS. Error bars indicate standard error with n=3. ................................ 71
Figure II-16 Quantitative distribution of glycoalkaloids in each types of trichome (from 50
trichomes extracted in 50 µl solvent) from S. habrochaites (LA1777) measured using
negative mode LC/MS. Key: tomatine (blue), dehydrotomatine (red). Error bars
indicate standard error with n=3. .......................................................................................... 72
Figure II-17 Quantitative distribution of methylated flavonoids in each types of trichome

xi

(from 50 trichomes extracted in 50 µl solvent) from S. pennellii (LA0716) measured using
negative mode LC/MS. Key: monomethyl kaempferol (blue), dimethyl kaempferol (red).
Error bars indicate standard error with n=3. ......................................................................... 74
Figure II-18 Quantitative distribution of glycosylated flavonoid (rutin) in each types of
trichome (from 50 trichomes extracted in 50 µl solvent) from S. pennellii (LA0716)
measured using negative mode LC/MS. Error bars indicate standard error with n=3....... 75
Figure II-19 Quantitative distribution of total acylsugar metabolites in each types of trichome
(from 50 trichomes extracted in 50 µl solvent) from S. pennellii (LA0716) measured using
negative mode LC/MS. Error bars indicate standard error with n=3. ................................ 76
Figure II-20 Quantitative distribution of glycosylated flavonoids in each types of trichome
(from 50 trichomes extracted in 50 µl solvent) from S. lycopersicum (M82) measured
using negative mode LC/MS. Key: Rutin (blue) and quercetin-diglucoside (red). Error
bars indicate standard error with n=3. .................................................................................. 78
Figure II-21 Quantitative distribution of methylated flavonoids in each types of trichome
(from 50 trichomes extracted in 50 µl solvent) from S. lycopersicum (M82) measured
using negative mode LC/MS. Key: monomethyl myricetin (blue), dimethyl myricetin
(red). Error bars indicate standard error with n=3. ............................................................ 79
Figure II-22 Quantitative distribution of total acylsugar metabolites in each types of trichome
(from 50 trichomes extracted in 50 µl solvent) from S. lycopersicum (M82) measured
using negative mode LC/MS. Error bars indicate standard error with n=3. ...................... 80
Figure II-23 Principal component analysis (PCA) of individual trichome metabolite A) score
plot for first and second principal. B) loadings plot including all of the detected
metabolites (“M” represents metabolites). L means leaf and S means stem. ....................... 84
Figure II-24 Principal component analysis (PCA) of individual trichome metabolite from
LA1777, displaying the scores plot for the first and third principal components. L means
leaf and S means stem. The number after the underscore means biological replicate
number. ................................................................................................................................. 87
Figure III-1 Optical image of S. habrochaites LA1777 leaf (left) showing trichomes on the
adaxial (upper) leaf surface including type VI glands (right). Adapt from Reference [140]
............................................................................................................................................... 95
Figure III-2 Transmitted light optical image of trichomes from the adaxial (upper) leaf
surface of S. habrochaites LA1777 transferred to a glass microscope slide. ..................... 105

xii

Figure III-3 Negative ion mode extracted ion LC/MS chromatogram of m/z 1076.5 showing
dehydrotomatine for (A) metabolites washed off from the S. habrochaites LA1777
trichome printed glass substrate analyzed by LC/MS. (B) Metabolites remaining on the S.
habrochaites LA1777 leaf surface after printing and flushing with extraction solvent and
analyzed by LC/MS. ........................................................................................................... 107
Figure III-4 The background MALDI and LDI mass spectra of DHB (A), CHCA (B), Pencil
lead (C) over m/z 50~1500 in the positive mode. The spectra were scaled to the same
absolute y-axis scale. .......................................................................................................... 109
Figure III-5 Laser desorption ionization (LDI) mass spectra of trichome extracts from S.
habrochaites LA1777 spotted on pencil lead-coated glass slides using (A) positive mode
(B) negative mode. Post source decay (PSD) product ion spectra of (C) acylsugar S4:21
+
precursor=m/z 715 ([M+Na] ) (D) mixed isomeric flavonoid glycosides rutin and
kaempferol-O-diglucoside ([M-H] , m/z 609). Electrospray ionization in-source decay
(ISD) mass spectra of (E) acylsugar S4:21 and (F) mixed flavonoid glycosides rutin and
kaempferol-O-diglucoside [M-H] acquired using TOF-MS and nonselective collision
induced dissociation. ............................................................................................................113
Figure III-6 LDI images of selected ions for contact printed trichomes from S. habrochaites
LA1777 leaf on pencil lead-coated glass slides: (A) ion abundances of m/z 731 (acylsugar
+
S4:21 [M+K] ) using positive mode and laser power setting = 95, 125 µm stage step, and
20 laser shot per pixel. (B) m/z 301 in negative ion mode (aglycone fragment of quercetin
glycosides), (C) m/z 233 (sesquiterpene acid [M-H] of various sesquiterpene acids) for the
same leaf sample under negative mode with laser power setting of 105, and 20 laser shots
per pixel. The key for color coding is presented at the left of each panel. ...................... 121
Figure III-7 Imaging of metabolites from S. habrochaites LA1777 leaf trichomes transferred
into the (A) 100 µm and (B) 50 µm TEM grids. Ion abundances are displayed for m/z 731
+
(acylsugar S4:21 [M+K] ), using positive mode with laser power setting = 95, 25 µm
stage step size, and 200 laser shot per pixel........................................................................ 123
Figure III-8 Left: Mass spectrometry images of selected ions for S. habrochaites LA1777
trichomes transferred to glassy carbon. (A) m/z 731 in positive mode, corresponding to
+
acylsugar S4:21 ([M+K] ), using laser power setting of 105, 125 µm stage step, and 20
laser shots per pixel. (B) m/z 301 in negative mode (fragment ion of quercetin glycosides)
for the same leaf sample using laser power setting of 115, 125 µm step size, and 20 laser
shot per pixel. Right: (A) mass spectra for two individual pixels in positive mode (B) mass
spectra for two pixels in negative mode. ............................................................................ 127
Figure III-9

Images of S. habrochaites LA1777 trichomes printed on (A) glassy carbon

xiii

surface, reflected light image (B) m/z 233 ion (sesquiterpene acids) in negative mode on
the same glassy carbon area. Step size was 25 µm with 200 laser shots per pixel. (C)
fluorescent image of S. habrochaites LA1777 trichomes printed on a glass slide using 543
nm excitation and 560 nm longpass emission. Slide was subjected to carbon coating
afterward; (D) mass spectrometry image of the same glass slide as (C) after carbon coating,
showing ion abundance map for m/z 233 in negative mode using 200 laser shots per pixel
with 25 µm step size. .......................................................................................................... 130
Figure III-10 (A) The LDI spectra of a mixture of S. habrochaites LA1777 and S. pennellii
LA0716 extracts using positive linear mode. Using a laser power=145, spectra from 200
laser shots were averaged while rastering the laser across the spot. (B) Calibration curve
+
showing ion abundance of m/z 731 (S4:21, [M+K] ) relative to the least diluted LA1777
extract as a function of 1/dilution factor of LA1777 extract (n=3, ±SE). A constant
amount of LA0716 extract was added to provide a matrix in which the LA1777
metabolites were diluted. .................................................................................................... 132
Figure III-11 (A)The schematic diagram showing the hole drilling experiment on the leaf
surface for validation of LDI metabolite profiling using LC/MS. Four areas were collected
around the leaf midvein and four areas were collected on the leaf edge. (B) The ratio of the
sesquiterpene acid detected in the midvein (m/z 233) cut to the leaf edge was measured
using two different analytical method; LC/MS and LDI. The experiment was conducted on
two different leaves. ............................................................................................................ 135
Figure III-12 The reflected light optical image of transferred S. habrochaites LA1777
trichomes on glassy carbon surface (A) taken right after the transfer. (B) after laser
ablation with 200 laser shots per pixel and laser power =145. ........................................... 136
Figure III-13 Imaging of “Spartan” logo constructed by tomatine deposited on the 10 µm of
parafilm attached on the top of glassy carbon. (A) Optical image of the Spartan logo cut
with a razor blade. (B) intensity map of tomatidine fragment ion (m/z 416) from tomatine.
Spectra were acquired using linear positive mode using laser power of 155, and 20 laser
shots per pixel. .................................................................................................................... 138
Figure IV-1 Basic structures of the main classes of flavonoids. Common O- and
C-substitution positions are indicated with an arrow. Adapted from Reference [206]. .. 142
Figure IV-2 Structure of the flavonol aglycone myricetin. The ring was labeled as large
letters in black. Positions of substitution by hydroxy or methoxy groups are numbered in
black inside the myricetin structure. The bonds on the B ring that undergo cleavage during
collision induced dissociation are numbered in red.
Characteristic multiband
fragmentations are labeled in black near the arrow heads, as proposed in Reference [206].154

xiv

Figure IV-3 A) Extracted ion chromatogram (XIC) of m/z 333 for monomethyl product from
incubation of ShMOMT2 with myricetin B) Extracted ion chromatogram (XIC) of m/z
333 for monomethyl product from incubation of ShMOMT1 with myricetin C) Extracted
ion chromatogram (XIC) of m/z 333 for monomethyl product from incubation of both
ShMOMT1 and ShMOMT2 with myricetin D）
Enhanced product ion scan (EPI) spectra of
monomethyl product from incubation of ShMOMT2 with myricetin, later annotated as
3’-methylmyricetin E）Enhanced product ion scan (EPI) spectra of monomethyl product
from incubation of ShMOMT1 with myricetin, later annotated as 4’-,methylmyricetin.
Data was acquired using Qtrap 3200 instrument using electrospray positive mode. EPI
spectra was acquired under 40V collision energy. .............................................................. 158
Figure IV-4 A) Extracted ion chromatogram (XIC) of m/z 333 for commercial standard
+
3’-methylmyricetin B) Enhanced product ion scan (EPI) spectrum for [M+H] (m/z 333)
of commercial standard 3’-methylmyricetin using 40 V collision potential. Data were
acquired using a Qtrap 3200 mass spectrometer using electrospray ionization in positive
mode. ................................................................................................................................... 163
Figure IV-5 Enhanced product ion scan (EPI) spectra of protonated dimethylated product
from A) incubation of myricetin with ShMOMT1 and SAM; later annotated as
4’,7-dimethylmyricetin and B) incubation of myricetin with ShMOMT1 and d3-SAM, later
annotated as 4’(CD3),7(CD3)-dimethylmyricetin. Data were acquired using a Qtrap 3200
mass spectrometer using electrospray positive mode and 40 V collision potential. ........... 166
Figure IV-6 Fragments of m/z 167 derived from protonated 4’,7-dimethylmyricetin that can
+
distinguish sites of methylation through deuterium labeling. Pathway involves 1,3A
cleavage with one methyl group remaining on A ring, retaining the deuterium label if
labeled in the methyl on A ring........................................................................................... 168
+

Figure IV-7 A) Extracted ion chromatogram (XIC) of m/z 347 for [M+H] of dimethylated
products from incubation of both ShMOMT1 and ShMOMT2 with myricetin; B)
enhanced product ion scan (EPI) spectrum of dimethyl product (RT 8.71), annotated as
3’,5’-dimethylmyricetin; C) enhanced product ion scan (EPI) spectrum of dimethylated
product (RT 8.95), annotated as 3’,4’-dimethylmyricetin; D）EPI spectrum of product (RT
9.34). annotated as 4’,7-dimethylmyricetin; E）EPI spectrum of product (RT 9.50)
annotated as 3’,7-dimethylmyricetin. Data were acquired using a Qtrap 3200 mass
spectrometer using electrospray positive mode and 40 V collision potential. .................... 172
+

Figure IV-8
A) Extracted ion chromatogram (XIC) of m/z 347 ([M+H] ) for
dimethylmyricetin products from incubation of ShMOMT1 with 3’-monomethylmyricetin
B) EPI spectrum of dimethylmyricetin product (RT 8.97), later annotated as

xv

3’,4’-dimethylmyricetin C) EPI spectrum of dimethyl product (RT 9.47), later annotated as
3’,7-dimethylmyricetin. Data were acquired using a Qtrap 3200 mass spectrometer using
electrospray positive mode and 40 V collision potential. ................................................... 180
+

Figure IV-9 EPI spectrum of m/z 361 ([M+H] ) for trimethylmyricetin product (RT 10.22
min) from incubation of ShMOMT1 with 3’-monomethylmyricetin, later annotated as 3’,
4’, 7-trimethylmyricetin. Data were acquired using a Qtrap 3200 mass spectrometer using
electrospray positive mode and 40 V collision potential. ................................................... 184
+

Figure IV-10
A) Extracted ion chromatogram (XIC) of m/z 347 ({M+H] ) for
dimethylmyricetin products from incubation of both ShMOMT1 and ShMOMT2 with
3’-monomethylmyricetin; B) EPI spectrum of product (RT 8.34 min), later annotated as
3,3’-dimethylmyricetin C) EPI spectrum of dimethyl product (RT 8.73 min), later
annotated as 3’,5’-dimethylmyricetin D) EPI spectrum of dimethyl product (RT 8.98min).
later annotated as 3’,4’-dimethylmyricetin E) EPI spectrum of dimethyl product (RT
9.52min). later annotated as 3’,7-dimethylmyricetin. Data were acquired using a Qtrap
3200 mass spectrometer using electrospray positive mode and 40 V collision potential. .. 187
+

Figure IV-11
EPI spectrum of m/z 350 ([M+H] ) for dimethylmyricetin product
3’-methyl-4’(d3)-methylmyricetin
from
incubation
of
ShMOMT1
with
3’-monomethylmyricetin and d3-SAM. Data were acquired using a Qtrap 3200 mass
spectrometer using electrospray positive mode and 40 V collision potential. .................... 196
Figure
IV-12
EPI
spectrum
of
m/z
350
for
dimethyl
product
3(d3)-methyl-3’-methylmyricetin from incubation of both ShMOMT1 and ShMOMT2
with 3’-monomethylmyricetin and d3-SAM. Data were acquired using a Qtrap 3200 mass
spetrometer using electrospray positive mode and 40 V collision potential....................... 197
+

Figure IV-13 A) Extracted ion chromatogram (XIC) of m/z 361 ([M+H] ) for trimethyl
product from incubation of ShMOMT1 with 3’,5’-dimethylmyricetin. B) EPI spectrum of
m/z 361 for trimethyl product (RT 9.71min), later annotated as 3’,4’,5’-trimethylmyricetin
C) EPI spectrum of m/z 361 for trimethyl product (RT 10.14 min), later annotated as
3’,5’,7-trimethylmyricetin. Data were acquired using a Qtrap 3200 mass spectrometer
using electrospray positive mode and 40 V collision potential. ......................................... 199
+

Figure IV-14 A) EPI spectra of m/z 361 ([M+H] ) for trimethylmyricetin product from
incubation of ShMOMT2 with 3’,5’-dimethylmyricetin later annotated as 3,
3’,5’-trimethylmyricetin; B) EPI spectrum of m/z 364 for trimethyl product from
incubation of ShMOMT2 with 3’,5’-dimethylmyricetin and d3-SAM, later annotated as
3(d3)-3’,5’ -trimethylmyricetin.
Data were acquired using a Qtrap 3200 mass
spectrometer using electrospray positive mode and 40 V collision potential. .................... 204

xvi

+

Figure IV-15 A) EPI spectrum of m/z 375 ([M+H] ) for tetramethylmyricetin product (RT
9.81 min) from incubation of ShMOMT2 with 3’,4’,5’-trimethylmyricetin, later annotated
+
as 3, 3’, 4’,5’-tetramethylmyricetin; B) EPI spectrum of m/z 378 ([M+H] ) for
tetramethylmyricetin product (RT 9.81 min) from incubation of ShMOMT2 with
3’,4’,5’-trimethylmyricetin
and
d3-SAM,
later
annotated
as
3(d3)
3’,
4’,5’-tetramethylmyricetin. Data were acquired using a Qtrap 3200 mass spectrometer
using electrospray positive mode and 40V collision potential. .......................................... 207
+

Figure IV-16 EPI spectrum of m/z 375 ([M+H] ) for tetramethylmyricetin product (RT 10.31
min) from incubation of ShMOMT1 with 3’,4’,5’-trimethylmyricetin, later annotated as
3’,4’,5’,7-tetramethylmyricetin. Data were acquired using a Qtrap 3200 mass spectrometer
using electrospray positive mode and 40 V collision potential. ......................................... 210
+

Figure IV-17 EPI spectrum of m/z 350 ([M+H] ) for dimethylmyricetin products from
incubation of ShMOMT2 with 3’-monomethylmyricetin and d3-SAM, later annotated as
3’-methyl-5’(d3)-methylmyricetin. Data were acquired using a Qtrap 3200 mass
spectrometer using electrospray positive mode and 40 V collision potential. .................... 214
+

Figure IV-18 EPI spectrum of m/z 353 ([M+H] ) for dimethylmyricetin product from
incubation of both ShMOMT1 and ShMOMT2 with myricetin and d3-SAM, later
annotated as 3’(d3),5’(d3)-dimethylmyricetin. Data were acquired using a Qtrap 3200 mass
spectrometer using electrospray positive mode and 40 V collision potential. .................... 215
+

Figure IV-19 A) EPI spectrum of m/z 347 ([M+H] ) for dimethylmyricetin metabolite in S.
habrochaites LA1777 leaf dip extracts, later annotated as 3’,7-dimethylmyricetin (RT 9.51
+
min); B) EPI spectrum of m/z 347 ([M+H] ) for dimethylmyricetin metabolite in S.
habrochaites LA1777 leaf dip extracts, later annotated as 3,4’-dimethylmyricetin (RT 8.45
min). Data were acquired using a Qtrap 3200 mass spectrometer using electrospray
positive mode and 40 V collision potential. ........................................................................ 219
+

Figure IV-20 A) EPI spectrum of m/z 361 ([M+H] ) for trimethylmyricetin metabolite in S.
habrochaites LA1777 leaf dip extracts, later annotated as 3, 3’,7-trimethylmyricetin, (RT
9.61 min); B) EPI spectrum of m/z 361 for trimethylmyricetin metabolites in S.
habrochaites LA1777 leaf dip extracts, later annotated as 3,3’,4’-trimethylmyricetin. (RT
9.32 min). Data were acquired using a Qtrap 3200 instrument using electrospray positive
mode and 40 V collision potential. ..................................................................................... 222
+

Figure IV-21 EPI spectrum of m/z 375 ([M+H] ) for tetramethylmyricetin metabolite in S.
habrochaites LA1777 leaf dip extracts, later annotated as 3,3’,5’,7-tetramethylmyricetin,
(RT 10.48 min). Data were acquired using a Qtrap 3200 mass spectrometer using

xvii

electrospray positive mode and 40 V collision potential. ................................................... 225
+

Figure IV-22 EPI spectrum of m/z 389 ([M+H] ) for pentamethylmyricetin metabolite in S.
habrochaites
LA1777
leaf
dip
extracts,
later
annotated
as
3,3’,4’,5’,7-pentamethylmyricetin, (RT 11.18 min). Data were acquired using a Qtrap 3200
mass spectrometer using electrospray positive mode and 40 V collision potential. ........... 227
Figure V-1

Structures of betaines and DMSP detected in extracts of coral tissues. ............... 248

Figure V-2 Retention times of alanine betaine (□), DMSP (﹡) and β-alanine betaine (▲) on a
pentafluoropropyl HPLC column over pH 3.4 to 4.3 in the aqueous mobile phase
component. .......................................................................................................................... 251
Figure V-3 Extracted ion chromatograms showing betaines and DMSP extracted from the
2
tissue covering about 0.05 mm area of skeleton in a specimen of the coral species
Madracis mirabilis collected in the Netherlands Antilles. (A): Peak #1, glycine betaine
(m/z 118, tR 1.41 min); peak #2: alanine betaine (m/z 132, tR 1.60 min); peak #3: β-alanine
betaine (m/z 132. tR 2.71 min); peak #4: DMSP (m/z 135, tR 2.36 min); peak #5: proline
betaine (m/z 144, tR 1.67 min); peak #6: GABA betaine (m/z 146, tR 6.68 min); peak #7:
hydroxyproline betaine (m/z 160, tR 1.28 min) and peak #8: aminovaleric acid betaine (m/z
160, tR 8.57 min); (B): Extracted ion chromatogram for m/z 141, which shows peaks
corresponding to chromatographically resolved internal standards alanine betaine-d9 (peak
a) DMSP-d6 (peak b) and β-alanine betaine-d9 (peak c). .................................................. 253

xviii

LIST OF SCHEMES

Scheme III-1

Sample preparation procedure for trichome contact printing. .......................... 100

Scheme IV-1 Conversions of myricetin and monomethylmyricetins by ShMOMT1 and
ShMOMT2 enzymes. Products of ShMOMT1 alone are colored red, of ShMOMT2 alone
are colored blue, and products requiring both enzymes are colored brown. ...................... 193

xix

CHAPTER ONE
INTRODUCTION

1.1

Metabolite profiles and the quest to understand biological functions

For centuries, the mysteries of living organisms have fascinated people, and a great
deal of research effort has been invested to improve our understanding about how living
things function.

The discoveries of chromosomes, genes, and DNA focused much of this

research over the past few decades upon the influence of specific genetic contributions to
biological functions [1].

In the 1980s, several teams of researchers set about to sequence

entire genomes, including those of bacteria [2], yeast [3], the model plant Arabidopsis
thaliana [4], rice [5], and ultimately, humans, in 2003 [6].

In the early 21

st

century,

numerous whole genomes have been sequenced, and the number of sequenced genomes is
expected to undergo explosive growth during the next decade owing to improvements in
the cost and speed of DNA sequencing.

However, genome sequences only provide a

starting point for understanding gene functions, because these functions are usually
inferred by comparing sequences with genes of similar sequence for which the functions
are already known.

However, small differences in DNA and protein sequences, even

single mutations can have dramatic effects on protein functions.

A new area of research

called “functional genomics”, which tries to understand the functions of unknown genes,
has emerged as genomic information and genetic technologies including expressed
1

sequence tag (EST) sequencing, microarrays, or sequencing based analysis of gene
expression (RNA-Seq) for large-scale transcriptome profiling [7].
Unlike the genome, transcriptomes, which are global profiles of mRNA levels, are
dynamic, varying with external environmental conditions.
measures and identities of which genes are actively expressed.

Transcriptomes provide
It was initially thought

that transcriptome profiling could answer most questions about gene functions.
However it was soon realized that transcriptomes do not always correlate to the content
and level of proteins in specific tissues for several reasons [8-11].

First, the rapid

degradation of mRNA and also degradation of proteins contribute to poor correlations
between levels of specific mRNAs and the proteins for which they encode.

Second,

many proteins must undergo post-translational modification to convert to their functional
forms.

This information cannot be captured from analysis of transcriptomes.

In

addition, due to alternative splicing of transcripts, a single mRNA may yield proteins with
more than one amino acid sequence, which also leads to further discrepancies between
transcriptome and proteome.

Last, but not least, many proteins become active only when

they interact with other protein or mRNA molecules.

All of these complexities give

birth to another research area, “proteomics”, which is large scale study of all proteins in a
biological system.
Despite the progress made in genomics, transcriptomics and proteomics, there are
still knowledge gaps regarding cell functions.

The “omics” data mentioned above

provide information about the potential functions of cells, and many of these functions
2

involve the synthesis, degradation, transport, and sensing of the most downstream
products of the biological cascade, which are metabolites.

Metabolites are the end

products of cellular regulatory processes and their levels reflect the combination of
genetic and environmental influences, which give a close reflection of biochemical
activity.

Analysis of metabolic snapshots serves as a valid approach for quantitative

rather than qualitative description of cellular regulation and control.

Thus the

metabolome can be regarded as the unbiased chemical phenotype of a cell or tissue and
can be used to establish effects of genes and/or environment influence cellular chemistry
[12].

This is summarized in a comment attributed to Dr. Bill Lasley of UC Davis:

“Genomics and proteomics tell you what might happen, but metabolomics tells you what
actually did happen!” [13]
Another relatively new research area called “metabolic genomics” is based on the
combination of genomics and metabolomics [14, 15].

Metabolic genomics research aims

to extend and refine our understanding of relationships between genes and cellular
biochemistry by creating more detailed functional genomics maps connecting biosynthetic
pathways, regulation of metabolite levels, with the genetic components involved.

The

ultimate goal is to reveal regulatory mechanisms of the complex interacting networks
underlying functions in whole biological systems.
Together with proteomics and transcriptomics, integration of profiling of transcripts,
proteins, and metabolites provides a more complete picture of biological functions and
dynamics.

By taking account of the metabolome, researchers have more opportunities to
3

discover the links between macromolecules, identify functional linkages between
phenotypes, and also construct dynamic networks for predicting biological behavior.
Although metabolomics alone cannot answer all questions about biological functions,
there are special merits of metabolome analysis compared to other “-omics” approaches.
Annotations of gene functions must have a definitive phenotype to differentiate function
from absence of function.

One advantage of metabolomics derives from the smaller

number of metabolites relative to the number of genes and proteins in a cell.

This

decrease in experimental complexity is exploited when the metabolome is used as a proxy
to represent phenotypes.

Second, the metabolome can often magnify changes in activity

of low abundance proteins, particularly when the protein acts as a catalyst or transporter.
In such cases, metabolites can be far more abundant than the proteins involved in their
synthesis, and this feature makes phenotypes more readily detected and distinguished.
In many cases, levels of individual or groups of metabolites are indicators of altered
biological functions.

Third, the expense of metabolite analysis is estimated to be two to

three orders of magnitude lower than global transcriptomics or proteomics [16].

In certain

scenarios, the metabolome provides a shortcut to recognize outlier phenotypes, and
metabolome data can direct other analyses, including proteome and transcript analyses,
toward those samples that exhibit phenotypes of interest.

For example, the forward

genetics approach of random mutation is one popular way to discover gene functions.
However, the vast majority of mutations do not lead to obvious morphological phenotypes,
a phenomenon called silent mutation.

In this case, metabolite screening as the starting
4

point provides a more efficient and cost-effective way to identify biochemical phenotypes
that otherwise might be missed [17].

1.2

The Application of Metabolomics in Plant Research

Plants are sessile organisms that are not capable of moving to avoid environmental
stresses such as temperature or water stress, predation by animals, or diseases caused by
pathogens.

As a result, plants largely respond to stress by changing their cellular

chemistry.

Because of this evolved behavior, plants accumulate an extensive and diverse
5

6

group of metabolites, estimated to be around 10 to 10 distinct chemicals across the plant
kingdom [18].

This huge diversity of phytochemicals indicates diversified gene

functions in plants, and these functions are often specialized in individual plant species or
lines based on the environmental stresses that specific plants encounter in their natural
habitats.

For example, there are 272 and 303 cytochrome P450 enzymes, which catalyze

metabolite oxidations, discovered in Arabidopsis and rice, respectively, compared to only
57 in humans [19, 20].
Modern research is only beginning to explore the composition of the metabolome in
individual plant species.

Despite longstanding knowledge that certain plants make

compounds of economic interest to humans, remarkably little is known about the
biosynthetic pathways leading to even those compounds of greatest interest. The
metabolome of Arabidopsis thaliana, a model plant for which the genome has been
5

sequenced, is probably the most extensively studied metabolome among all plants.
Metabolite profiles have been generated for well-defined libraries of Arabidopsis mutants
[21], but much of the chemical information gathered to date has focused on abundant
“primary” metabolites including sugars, organic acids, and amino acids that are essential
for plant survival.

Many of these mutants showed metabolic differences compared to

wild type plants, and research is heading toward discoveries of how single point mutations
change metabolite profiles even though no visible phenotypic change occurs.

Using

metabolite information as the bridge, the prediction of the functions of unknown genes
could be made by monitoring the metabolic changes caused by deletion/insertion of
unknown genes.

This approach could become more effective if supplementary analyses

such as large scale enzyme analysis [22] were combined with metabolite profiling and
gene mapping to build an unprecedented overview of plant metabolism.
Another practical application of metabolomics is for population screening and plant
breeding.

To humans, the intrinsic interest to understand plants derives from a desire to

improve food products, medical therapies, and other valuable feedstocks including
biofuels.

For the applied scientist, e.g. in the plant breeding industry, the metabolomics

approaches have already gained a great deal of attention recently [23] owing to their
potential to help define better phenotypes and assist breeding of plants for desirable traits.
Economically valuable traits of the plants such as disease resistance, nutrient content and
quality, flavor, and fragrance are based on metabolite composition, and these are traits
breeders value.

Using metabolite information to regularly select plants having better
6

disease resistance, greater nutrient content and desirable appearance (pigments) have
substantial economic value for breeders.

Some plants even contain metabolites which

have potentially health-promoting properties, e.g. Chinese herbal medicines.
The interest in linking food composition and human health has grown rapidly in
recent years.

People are paying more attention to purchasing healthier food products,

leading to introduction of the term “nutraceuticals” to describe food products that offer
potential therapeutic benefits.

Food producers and supply companies also want to

advertise their products with scientific support of claims that consuming certain foods,
such as those rich in antioxidant plant metabolites, leads to better health.

Considering

that metabolite profiling is a relatively inexpensive analytical procedure, metabolomic
analyses are finding increased use for screening natural or induced mutant plants or, to
demonstrate that genetically modified foods are functionally equivalent to “natural” foods.
One good example is the mutant or modified tomatoes [24-26].

Davuluri reported using

targeted genetic modification for generating lines of tomato accumulating more health
related carotenoid and flavonoid antioxidants.

He also used metabolite analysis to

identify novel antioxidants in transgenic tomatoes transfected with a stilbene synthase
gene from grape.
Aside from developing more nutritious foods, metabolomics can help plant industry
to engineer stronger plants that can better resist disease and predation.

Plants have

sophisticated chemical defense mechanisms that are triggered when they are subjected to
biotic or abiotic stresses.

It has suggested that a coordinated ‘metabolic reprogramming’
7

occurs in plants subjected to sudden environmental changes including both biotic and
abiotic stresses [27].

An example of where this may play important roles in the future

lies in the development of more energy-rich biofuel crops.

Such plants, which

accumulate large stores of metabolites designed as fuel precursors, are likely to be
attractive foods for animals and microbes.

Successful development of biofuel crops will

require metabolic engineering of pathogen and herbivore defenses.

In this context, it is

important to understand stress responses fully from their molecular basis, and design
crops with better stress resistance.

Metabolomics will play major a role in guiding these

research efforts.

1.3

Analytical platforms for metabolomics

A metabolite may come from more than one metabolic pathway.

Researchers can

best identify which pathways are involved in its metabolism when they assess the
metabolome as a whole.

There are several challenges to metabolome analysis that pose

greater barriers relative to genomics, transcriptomics or proteomics.

DNA and RNA are

each constructed as linear polymers from four nucleotides, so the structural diversity is
somewhat limited.

Furthermore, many oligonucleotides participate in predictable

molecular recognition events.

This selective binding, coupled with modern capabilities

to amplify DNA, makes it easy to implement high throughput platforms for analyzing
DNA or RNA.

Similarly, proteins are linear combinations of twenty amino acids, and an
8

assortment of methods has been developed to assess their primary structures and quantify
proteins.

In comparison, metabolites exhibit structural diversity that is vast relative to

oligonucleotides or proteins because there are virtually no limits on the structural features
that make up the metabolome.

Wide variations of physicochemical properties across the

metabolome make any global analysis challenging, especially with regard to structure
annotation.

Furthermore, the large range of metabolite concentrations (can differ over

several orders of magnitude) challenges even the most powerful analytical technologies.
The current state of global metabolite analysis is based on three types of platforms,
gas

chromatography-mass

spectrometry

(GC/MS),

nuclear

magnetic

resonance

spectroscopy (NMR), and liquid chromatography-mass spectrometry (LC/MS) [28-31].
NMR finds widespread use for medical metabolomics but has found limited application
for profiling of specialized plant metabolites.

NMR has the advantage of uniform

detection of hydrogen-containing metabolites without substantial artificial matrix effects,
so the magnitude of each observed signal provides an accurate measure of the true
metabolite level.

The weakness of NMR lies in the need to analyze milligram quantities

of each detected metabolite.

Therefore, NMR is a powerful approach for profiling

metabolites of high abundance [32], but lacks the capability to measure the numerous
metabolites that are orders of magnitude less abundant than the major constituents.
Quick annotation of metabolites can be achieved by comparing spectra to NMR spectra of
authentic standards.

However for previously undiscovered metabolites for which

standards are not available, explicit structure elucidation often requires purified materials
9

and multidimensional NMR, and these techniques require yet more metabolite.
GC-MS is one widely used method for metabolomics, and its use for global
metabolite profiling predates the word ‘metabolomics’. Its popularity comes from its
remarkably high separation efficiency, and its coupling with electron ionization which
provides near-universal detection.

Since electron ionization employs a standard fixed

ionization condition (70 eV electron energy), mass spectra for the same analyte are similar
across instruments and laboratories.

The availability of libraries of mass spectra and

metabolite identification software facilitates metabolite annotation.

However, the

technique is only applicable to separation and detection of volatile and chemically stable
metabolites at temperatures up to about 300℃ (e.g. alcohols, monoterpenes and esters).
Therefore chemical derivatization needs to be employed for analysis of nonvolatile or
polar metabolites, including amino acids, sugars and organic acids [33].

Derivatization

involves drying and conversion of polar functional groups to nonpolar substructures.
These processing steps can introduce variability and losses during the process of drying
and

derivatization.

Even

with

derivatization,

many

metabolites

including

glycoconjugates may still not be detected using GC/MS owing to their decomposition
during heating and evaporation in the GC injector, and such reactions can generate new
compounds that were not in the original metabolite extracts.
Application of liquid chromatography-mass spectrometry (LC/MS) for global
metabolic profiling has increased in recent years [34].

The greatest advantage of LC/MS

is that it is usually coupled with soft ionization, e.g. electrospray ionization (ESI), which
10

can preserve fragile molecules such as glycosylated flavonoids intact.

Analyses that

employ LC/MS often involve minimal sample preparation, as derivatization is usually not
necessary.

Furthermore, LC column efficiencies have improved dramatically in recent

years [35], leading to coining of the term ultrahigh performance liquid chromatography
(UHPLC).

These efficient columns have stimulated renewed interest in moderate to high

throughput LC/MS methods, and these are emerging as the preferred approach for many
metabolome analyses.

Advances in novel column chemistries (e.g. hydrophilic

interaction chromatography (HILIC) and long monolithic columns) offer the potential for
greater coverage of the metabolome over a wider range of properties.

Although

electrospray ionization (ESI) is not as universal as electron ionization (EI), a range of
alternative ionizations techniques is available to ensure ionization of a broad range of
metabolites.

These methods include atmospheric pressure chemical ionization (APCI)

and atmospheric pressure photoionization (APPI). The greatest challenge for LC/MS
based metabolite analysis comes from two points.

Unlike GC-MS, libraries of mass

spectra, particularly tandem mass spectra, for LC-MS are limited.

Because experimental

parameters differ across laboratories, standard libraries of spectra have been slow to
develop.

How to annotate metabolite structures without standard spectra for comparison

remains a challenge.

This topic has been the subject of continuing discussion [36-41].

A second challenge that confronts LC/MS-based metabolome analysis arises from the
phenomenon called “matrix effects”.

A coeluting compound(s) often changes the extent

of ionization of another analyte, and causes its signal to be reduced (or in some cases,
11

increased).

This problem is especially severe for analyses of extracts having complex

and unknown composition, because it is virtually impossible to resolve all metabolites in
a single LC separation.

1.4

Spatially resolved metabolite profiling, and the emerging interest in

micrometabolomics

Since shortly after the invention of the microscope, researchers have suspected, and
later confirmed, that biological functions are organized at the level of cells and
subcellular domains such as organelles.

During the intervening years between then and

now, the functions of these small domains have largely remained a mystery, in part
because analytical tools lacked the sensitivity, speed, and resolution needed to probe how
chemistry is organized at these levels.

Recent and dramatic technological improvements

in all of these areas have opened doors to profiling of metabolites in specific organs,
tissues, and even single cells.
Plants are considered to have about 40 different cell types [42].

Extracting and

analyzing an entire tissue of organ as a whole generates a mixture of metabolites that are
derived from multiple cell types, and is not the optimal way to study complex biological
systems.

There are important questions that still remain unanswered, such as how

different cells differentiate and communicate with each other; how cells with different
ages or developmental stages within a tissue are different in their chemistry; how
12

metabolites or enzyme substrate are transported from one tissue location to another; and
how signaling molecules trigger metabolic cascade responses throughout the whole
organism.

These questions are best answered by investigating biochemical behavior of

individual organs, tissues, and cells.

As the resolution and content of information

increases, the knowledge gained reaches closer proximity to the truth [43, 44].

So far,

metabolomics analyses have been mainly conducted on whole plants, or on whole organs
such as leaves or roots.

Only a few examples of metabolite analyses have been reported

for specific tissues [45, 46] or on specific plant cells, most notably lignified cells in
spruce bark [47, 48], level have been reported.
Analyses of a single tissue or individual cells pose daunting challenges, particularly
with regard to sample preparation.

Microsampling of individual cells is complicated by

the potential for contamination, whether the contaminants are derived from adjacent cells
of a different type, or simply because the small quantities of metabolite may be obscured
by contaminants derived from other sources.

Two major sampling methods have been

developed for microanalyses, with most focusing on collection of nucleic acids [43, 44, 49,
50].

The first involves usage of microcapillaries or micropipettes to extract cells or their

contents from tissues [49-52].

A typical experimental setup employs a microcapillary or

micropipette mounted on a microscope.

The targeted tissue or cell can be visualized

using a CCD camera, and sampling is controlled using a micromanipulator.

The second

method involves microdissection, and is sometimes referred to as Laser-Assisted
Microdissection (LAM) [53].

LAM uses focused light from a laser to cut a selected area
13

of tissue, including single cells, and the materials are collected from the sample either by
contact sampling e.g., attachment to a sticky surface or non-contact sampling, e.g. gravity
deposition, ejection by photon pressure.

LAM was initially developed for RNA isolation

[43] to measure gene expression in individual cell types.

Most recently, it has been

applied to small molecule analysis in specific plant tissues and cells [47, 48, 54, 55].
For metabolite analysis, it is important to avoid sources of contamination from the
material used in LAM such as paraffin, plastics, adhesives, dehydrating agents, and
staining tinctures, as these contain low-molecular-weight contaminants that may interfere
with metabolite profiling [56].
Two additional complications arise from microsampling, namely preservation of
unstable metabolites during the long sampling process, and the limited amount of material
present in tiny samples such as individual cells.

Many metabolites have short lifetimes

and are converted to other products by the action of various enzymes.
enzymatic transformations often requires quick freezing of tissues.

Halting these

Unfortunately, most

microsampling methods depend on the cytoplasm being in a liquid state, and when this is
the case, enzymes may be active, transforming the metabolite pool.

For this reason,

microsampling has been most successful for analyses of substances such as lipids that
undergo slow turnover.

The issue of analyte quantity has been challenging to address, as

single cells may contain only a picoliters to nanoliters of material.

Only the most

abundant metabolites, those present at low millimolar concentrations, are easily detected
because the amounts of metabolites of interest are present at attomole to femtomole
14

quantities which are below the routine capabilities of most instruments.
In terms of detection, many detection schemes have been used for analysis of single
tissues or cells, including fluorescence [57-61], electrochemical detection [62-65] and
mass spectrometry [66-86].

These are among the most sensitive analytical tools.

Single tissue or cell metabolome analysis requires technologies that offer low limits of
detection, high throughput, and also the capability to detect a wide range of analytes.
Although fluorescence and electrochemical analysis offer remarkably low detection limits,
these techniques are not universal for a wide range of metabolites, and fail to give
information that can distinguish a large number of substances from one another, as
analytes need to be fluorescent or electrochemically active to be detected.

Chemical

derivatization could be performed to introduce labels to facilitate detection, but this rarely
has provided adequate selectivity relative to background signal.

In view of these

considerations, mass spectrometry has emerged as the predominant technique for single
tissue or cell metabolome analysis owing to its near universal detection.
Mass spectrometric schemes for single cell analysis can be grouped into four
categories.

The first scheme is sampling the cell using a capillary probe, followed

introduction into the mass spectrometer using nanoelectrospray [80].

The greatest

drawback of this technique is that it is difficult to make it to be a high throughput
platform owing to the need to analyze numerous cells of a given type and because
coordination of nL/min flows from large numbers of cells is inherently slow.
second scheme employs microfluidic chips coupled with mass spectrometry.
15

The

At the front

end, the microchip sorts cells and can also quench, lyse and separate the metabolites using
electrophoretic or nanochromatographic methods.

The chip can be coupled to a variety

of ion interfaces, such as electrospray (online detection) or matrix-assisted laser
desorption/ionization (MALDI) (off-line detection).

The complete setup has not yet

been realized as a routine practice, but advances promise to introduce more robotic
operation, which is promising in terms of throughput [87-90].
sample arrays [91].

A MALDI sample plate was first coated with a solvent-repelling

coating, patterned in a checkerboard array format.
on the array.
each well.

A third scheme is called

Cell suspensions were then sprayed

An appropriate cell suspension density will yield only one to two cells in
Such samples can be analyzed using MALDI afterward [91].

This

technique could have broad application on certain types of cells such as microbes grown
in liquid culture, but it is difficult to envision how this might be implemented for other
types of cells in complex eukaryotes such as plant epidermal cells without compromising
the integrity of the metabolome.

The fourth scheme, which is believed to be most

promising, is imaging mass spectrometry [68, 69, 92-94], which can use any one of
several ionization methods including desorption electrospray, secondary ion mass
spectrometry (SIMS), or MALDI.

The biggest advantage of MALDI imaging is it

requires minimal sample preparation aside from introduction of an absorbing matrix that
can convert laser light into energy for ion formation and evaporation.

Another merit of

imaging mass spectrometry is that each pixel yields a mass spectrum, generating spatially
resolved chemical profiles.

A growing wave of interest is pushing improvements in
16

spatial resolution of imaging mass spectrometry over the past few years.

For example, a

spatial resolution of 10 µ m was reached with state-of-the-art LDI instrumentation based
on a near-infrared femtosecond laser [95], which offers potential for single cell metabolite
profiling.

With advances in laser technologies and sample preparation methods, imaging

mass spectrometry is expected to have great impact on single cell metabolite analysis that
will improve our understanding of cellular chemical heterogeneity and cell-cell
communication.

1.5

Summary of Research

As mentioned above, metabolomics is an important component of the growing
interest in systems biology, which often requires multidisciplinary collaborations
encompassing genetics, bioinformatics, molecular biology, and bioanalytical chemistry.
This dissertation presents results, in Chapters II-IV, of my role in developing and applying
analytical methods for metabolite profiling in a multidisciplinary team.

The scope of this

work has covered almost all aspects of metabolite analysis discussed above including
microsampling, high throughput screening, imaging mass spectrometry, detailed structural
elucidation, separation method development and quantitative analysis.
chapters of this dissertation are derived from two projects.

The various

One is the National Science

Foundation funded Solanum trichome project, which has involved plant geneticist,
biochemist, biologist, chemical ecologist, and bioinformaticist team members from
17

Michigan State University, the University of Michigan, and the University of Arizona.
The goals of this effort have been focused on identifying the genes and proteins involved
in the synthesis of specialized metabolites stored and excreted by Solanum trichomes.
Since the genus Solanum includes many important crop species including tomato, potato,
pepper, and eggplant, discoveries of these gene functions can open the door to molecular
level understanding of the factors that regulate metabolite synthesis, and form the
foundation for future efforts at metabolic engineering of plants and other organisms.
The second project, a collaboration with Professor Richard Hill and his group from the
Michigan State University Department of Zoology, is described in Chapter V.

This work

focused on developing and applying metabolite profiling approaches to understand the
dynamics of the coral reef metabolome under different environmental stress conditions.
My role in all of these projects has been to develop and apply sampling and analytical
methods for metabolite profiling.

Aspects of the research that are based on field

collection of coral samples and expression and assay of recombinant enzymes were
performed by collaborators identified in the individual chapters.
In Chapter II of this dissertation, the development of methodology for sampling
individual trichomes from five Solanum species is described.
throughput

LC/MS

platform,

called

multiplexed

The application of a high

collision

induced

dissociation

(multiplexed CID) for profiling metabolites in individual trichome types is reported.
The cell-specific chemical information discussed in Chapter II paved the way for our
collaborators to construct a network of genes expressed in trichomes that can be
18

associated with specific biochemical traits.

Chapter III illustrates a new technique for

matrix-free imaging mass spectrometric profiling of metabolites for Solanum trichomes.
This technique conquers several sample preparation challenges for imaging of trichome
by mass spectrometry.

In addition to showing different metabolites in different plant

species and genotypes, it is illustrated that the spatial distribution of trichome chemistry is
heterogeneous across a single tissue.

This information provides clear evidence of the

importance of incorporating spatial information into metabolite profiling.

In Chapter IV,

analytical strategies were developed and applied to demonstrate the specificity of two
trichome-containing

methyltransferase

enzymes

elucidation of regioisomeric enzyme products.

through

MS/MS-based

structure

These unknown enzyme products do not

have authentic standards available for spectra comparisons, and the mass spectrometric
data provide convincing evidence for individual structure assignments.

The approach

used in this effort employed semisynthetic preparation of products through use of the
recombinant enzymes coupled with isotopic labeling.

In Chapter V, the development of

a stable isotope dilution LC/MS method demonstrated, for the first time, the extraordinary
abundance of betaine metabolites in coral tissues, and the variation in betaine composition
across multiple species of field-collected corals.

Betaines are believed to be protective

metabolites that confer resistance to environmental stresses.

The results of this

investigation demonstrate that betaine content is regulated by environmental factors in a
way consistent with being a plastic metabolic response.
Some of the research described in this dissertation has already been published in
19

modified form.

Metabolite profiles from individual trichome types that are described in

Chapter II have appeared in a manuscript in Plant Physiology (E. T. McDowell, J. Kapteyn,
A. Schmidt, C. Li, J.-H. Kang, A. Descour, F. Shi, M. Larson, A. Schilmiller, L. An, G. A. Howe,
A. D. Jones, E. Pichersky, C. A. Soderlund, and D. R. Gang, Comparative functional genomic
analysis of Solanum glandular trichome types, Plant Physiology, 2011, 155: 524-539).

Some

of the principles of metabolite localization, established in Chapter III, have been
discussed in a recent book chapter (L. W. Sumner, D. S. Yang, B. J. Bench, B. S. Watson, C.
Li, and A. D. Jones.

‘Spatially resolved plant metabolomics’ in Biology of Plant Metabolomics,

Annual Plant Reviews, Volume 43, R. Hall, Ed., Chapter 11, pp. 344-366, 2011). Discoveries of
O-methyltransferase functions, described in part in Chapter IV, were included in another
paper (A. Schmidt, C. Li, F. Shi, A. D. Jones, and E. Pichersky, Polymethylated myricetin in
trichomes of the wild tomato species Solanum habrochaites and characterization of
trichome-specific 3′/5′ and 7/4′ myricetin O- methyltransferases, Plant Physiology, 2011, 155:
1999-2009). The methods for quantifying betaines in coral tissues described in Chapter V
have appeared in two papers (C. Li, R. W. Hill, and A. D. Jones, Determination of betaine
metabolites

and

dimethylsulfoniopropionate

in

coral

tissues

using

liquid

chromatography/time-of-flight mass spectrometry and stable isotope-labeled internal standards,
Journal of Chromatography B, 2010, 878: 1809-1816; and R. W. Hill, C. Li, A. D. Jones, J. P.
Gunn, and P. R. Frade. Abundant betaines in reef-building corals and ecological indicators of a
photoprotective role. Coral Reefs, 2010, 29: 869-880).

20

CHAPTER TWO
ANNOTATION OF SPECIALIZED METABOLITES ACCUMULATING IN
INDIVIDUAL TYPES OF GLANDULAR TRICHOMES OF THE GENUS SOLANUM

2.1

Specialized Metabolites in Plants

The plant kingdom has been estimated to synthesize more than 100,000 different
natural product metabolites, but the genetic basis for production of these compounds
remains largely unexplored.

The majority of these substances are considered specialized,

or secondary metabolites, that are not essential to plant survival, yet confer beneficial
traits that have evolved over long time scales as responses to evolutionary pressures.
Researchers are driving renewed efforts to understand the basis of interspecies and
intraspecies genetic variations responsible for biosynthesis of such a diversified suite of
metabolites.

The key approach used in these efforts involves correlating the genotype

with chemical phenotype, or composition.
synthesized in a specialized tissue.

In addition, it is typical for a metabolite to be

For plants, the most prolific tissue involved in

metabolite accumulation is the glandular trichome, which is a specialized epidermal cell
type that occurs on the surfaces of leaves, stems, and other tissues.
near-ideal model for investigations of plant metabolism.

Trichomes provide a

Multiple trichome types are

known for a variety of plant species, and trichomes may be collected using a variety of
microsampling methods that facilitate cell type-specific investigations of metabolism and
21

gene expression.
The common approach for studies of plant specialized metabolism has relied on
chemical profiling of bulk tissue extracts and interpretation of this chemistry in the
context of plant genetics, perhaps involving measures of gene expression in the extracted
tissue.

However, this strategy suffers when metabolite biosynthesis is largely confined

to a specific cell type, because genes expressed in tissues not involved in biosynthesis
may obscure the roles of the key biosynthetic genes.

Because it is not feasible to study

every individual type of tissue or cell within a plant, researchers should aim to select a
model plant system with sufficient allelic diversity, genomic tools, and distinct metabolic
phenotypes to investigate genotype and phenotype correlations.

Furthermore,

accessibility of specialized cell types provides a valuable handle for investigating
cell-specific biochemical behavior.

Based on those requirements, the genus Solanum,

which includes tomato, potato, and other crops and is known to produce an assortment of
trichome morphological types, serves as a great system to be the model for research on
specialized metabolite biosynthesis.

2.2

Glandular Trichomes in the Genus Solanum

Trichomes are epidermal cells on plant surfaces which are easy to isolate compared to
tissues on which the trichomes are attached.

As a result, it is possible to collect and

analyze RNAs, proteins, and metabolites from trichomes with minimal contamination from other
22

cells.

Tomato has recently been reassigned to the genus Solanum, and an extensive collection

of wild relatives of tomato are available.

Also, many Solanum species (tomato, eggplant, potato)

can inter-cross with each other and produce a diverse group of genetic variant hybrids
(~1500-2000 so far).

Those genetic variants offer a large number of allelic variants that can

facilitate elucidation of biosynthetic pathways.

Sequencing of the tomato genome is nearly

complete, and this information can be taken advantage of to facilitate the recognition of
candidate genes.

Finally, besides the advantages of using Solanum species for research, tomato

and its relatives are grown internationally and have important economic value that might be
improved by metabolic engineering.
Many plant trichomes store or secrete a complex array of metabolites.

It was proposed

that trichomes have many attributes important to the success of the plant [96].

The

well-accepted theory is that trichomes accumulate metabolites that protect plants against insect
herbivory.

For example, the model plant Arabidopsis produces more methyl salicylate in

trichomes after wounding and insect attack [97]. The glandular trichomes in tobacco secrete
nicotine, which is toxic to insects [98].
herbivory [99].

Trichomes of Aristolochia elegans decrease insect

Removal of trichomes from leaf surfaces makes the plant more vulnerable to

insect herbivory [100]. Ecological studies show that Arabidopsis increases trichome density
under pressure from insect herbivore attack [101].

All of the evidence above shows that

trichome provide an important protective layer of cells and metabolites.

In addition to these

defensive functions, trichomes play roles in plant responses to changes in temperature, water loss,
and gas exchange regulation [102, 103].
23

Trichomes are described as belonging to one of two categories: nonglandular and glandular
trichomes.

Since nonglandular trichomes are not known to accumulate substantial amounts of

specialized metabolites, this discussion will focus on glandular trichomes which accumulate a
diverse collection of metabolites in large quantities.

Based on Luckwill’s definition (Figure

II-1) [104], there are four types of glandular trichomes in the genus Solanum.

Type I trichomes

are defined as having a long multicellular stalk and a small glandular head; type IV trichomes
have a similar appearance, but a shorter multicellular stalk and a small glandular head; type VI
trichomes are composed of a short stalk with a larger mushroom-shaped glandular head made up
of a cluster of four cells; type VII trichomes consist of a short stalk and also a mushroom-shaped
small glandular head, but are smaller than type VI.

The feature of multiple gland types often

forms the basis for difference phenotypes between individual plants that can facilitate our
understanding of trichome functions and biosynthetic pathways in different trichomes.
The comparison between cultivated tomato and wild types indicates that many important
metabolites synthesized by the trichome have been lost during tomato domestication [105-109].
Conventional cultivated tomato crops are more susceptible to insects than are wild tomato
relatives [105, 110-112], and this has been attributed largely to differences in trichome chemistry.
One long-term goal of current studies of tomato trichomes is to discover genes responsible for
chemical defenses against insects, and this information offers the potential to reintroduce key
genes into tomato to improve insect resistance.

24

A

B

Figure II-1 Trichomes on L. esculentum (A) and L. hirsutum (B). Roman numerals indicate
trichome type. Adapted from Reference [104].

2.3

Challenges of Analysis of Trichome Specialized Metabolites

Discoveries of the genetic basis of trichome functions and biosynthetic pathways ultimately
rely on DNA sequencing and biochemical characterization of enzyme functions involved in the
metabolite biosynthesis.

However, understanding of the chemistry in different trichome types

25

is central to these research efforts, because the chemistry defines the phenotype.

Since different

cell types exhibit different chemical phenotypes, microsampling of individual trichome types
must be performed, and the chemical composition of individual trichome types must be defined
and interpreted in the context of genes expressed in each cell type.
glandular trichome (type VI) of tomato is about 100 picoliter.

The volume of the largest

Although trichomes are surface

tissues and are relatively easy to isolate, each individual trichome contains a tiny amount of
metabolites owing to its small size.

Isolation of large numbers of trichomes for chemical

analysis is labor intensive, tedious, and time-consuming.

To achieve the goal of profiling

numerous trichome types from different plants, a more rapid approach was called for.
Our efforts have aimed to gather deep metabolome information for individual trichomes.
To date, the most popular analytical method for metabolite profiling has been mass spectrometry.
The advantage of mass spectrometry is that it serves as a near-universal detection technology that
is useful for detecting a wide range of chemicals and providing information helpful for
compound identification.

In terms of structure elucidation capability, it is only second to NMR

among routinely used analytical techniques.

However, the detection limit of mass spectrometry

is several orders of magnitude lower compared to NMR, which often requires milligram
quantities of purified compound.

For nontargeted structure elucidation of metabolites by mass

spectrometry, the initial step involves determining the molecular mass and formula.

High

resolution mass spectrometers (e.g. time-of-flight, Orbitrap, Fourier Transform-Ion Cyclotron
Resonance) provide mass measurements with errors of less than 5 ppm, and are employed to
limit the range of molecular formulas consistent with a measured ion mass.
26

More detailed

structure information can be deduced from the masses of fragment ions generated from MS/MS
experiments.
A big challenge for metabolome analysis is encountered because a single analysis (e.g. of a
trichome extract), yields more than one thousand detectable ion masses based on the current
technologies.

To generate MS/MS spectra for individual metabolites to gain structure

information, large number, perhaps thousands, of MS/MS experiments need to be conducted.
Performing traditional MS/MS scans for all of the metabolite molecular masses is not feasible
due to enormous number of peaks detected.
metabolite in each sample is severely limited.

For trichome microsampling, the quantity of
Tandem mass spectrometric analyses of complex

mixtures has been advanced by the advent of data dependent MS/MS, which involves real-time
spectrum evaluation, selection of an ion for further MS/MS analysis, and generation of a product
ion MS/MS spectrum for the computer-selected ion.

This approach increases the throughput of

MS/MS experiments. However, the drawback of data dependent MS/MS is that it can only
fragment a few (e.g. up to five) of the most abundant ions within a certain time window that is
largely defined by chromatographic peak widths.

This makes it likely that low abundance

metabolites that coelute with more abundant compounds will not be selected for MS/MS analysis,
and structural information for the less abundant compounds will be lacking. This issue can be
addressed by repeated analyses, but such an approach is not feasible when extracts are limited in
sample size. Defining a trichome’s metabolic phenotype requires deep and comprehensive
profiling of as many metabolites as can be measured. Generation of fragment mass information
for both high- and low-abundance metabolites is not feasible for small sample sizes owing to the
27

limitations described above. To overcome those technical limitations and improve analytical
throughput, the Jones laboratory developed a time-of-flight mass spectrometry (TOF-MS)
method called multiplexed collision induced dissociation (multiplexed CID), which provides
quasi-simultaneous molecular mass information for metabolites and also nonselectively
fragments all of the ions in a single mass spectrometry analysis.

Its nonselective nature avoids

the limitations of data dependent MS/MS, and generate fragment ion information for low
abundance metabolites.

This analytical approach was applied for metabolite profiling of

individual trichomes as described in more detail below.

Even when we have limited volume of

trichome extracts, extensive and comprehensive generation of molecular and fragment ions can
be achieved in a single analysis.

2.4

2.4.1

Experimental Section

Materials

Seeds for Solanum lycopersicum M82, S. habrochaites LA1777, S. pennellii LA0716, S.
arcanum LA1708, and S. pimpinellifolium LA1589 were purchased from the C.M. Rick Tomato
Genetics Resource Center at the University of California-Davis.

The seeds were grown in a

growth chamber for 3 to 4 weeks, depending on the species, then seedlings were transferred to
pots and grown in the same growth chamber until analysis.

Plant growth was distributed among

multiple locations (East Lansing, MI, Ann Arbor, MI, and Tucson, AZ).
28

2.4.2

Chemicals

Acetonitrile (HPLC grade), 2-propanol (HPLC grade), methanol (HPLC grade), and formic
acid (88% aqueous solution) were purchased from VWR Scientific.

2.4.3

Trichome Collection

A summary of the trichome types collected from each plant species is presented in Table
II-1.

Table II-1

List of wild tomato species and trichome types subjected to metabolite profiling.
Species
S. pennellii (LA0716)
S. lycopersicum (M82)
S. habrochaites (LA1777)

S. arcanum (LA1708)
S. pimpinellifolium
(LA1589)

Trichome Types
leaf type IV (L4), leaf type VI (L6)
leaf I (L1), leaf VI (L6)
leaf I (L1), leaf IV (L4), leaf VI (L6), leaf VII
(L7)
stem I (S1), stem IV (S4), stem VI (S6)
leaf VI (L6)
leaf VI (L6)

To collect the type I and IV glands from the various Solanum species, microscissors were
used to cut the gland off from the stalk.
glass pipettes [113, 114].

Type VI gland heads were collected using the stretched

Type VII trichome glands were collected by agitating with glass

beads in buffer to abrasively remove trichomes from the leaf surface [115, 116].

29

S. habrochaites type VII trichomes were first transferred to a microcentrifuge tube and
centrifuge at 3,000g for 30 s and the pellet was extracted with extraction solvent
(2-propanol:CH3CN:water 3:3:2, v/v/v).

For the remaining trichome types, 50 picked

trichomes were immediately transferred to 50 µL ice-cold extraction solvent.

The gland/solvent

mixture was then sonicated for 5 min followed by vortexing for 30 s and centrifugation at 5,000g
for 10 min.

The supernatant was then transferred to a glass autosampler vial equipped with a

100-µL glass insert.

2.4.4

LC-MS Analysis

LC-MS analyses of isolated gland metabolites were performed using a model LCT Premier
mass spectrometer (Waters) coupled to an LC-20AD pump and SIL-5000 autosampler
(Shimadzu). All analyses utilized electrospray ionization in negative ion mode and employed a
custom-packed Supelco Discovery Bio C18 column (1 × 150 mm, 3-µm particles; 300 Å).
Mobile phase gradients were based on solvent A (0.15% aqueous formic acid) and solvent B
−1

(methanol), with a total flow rate of 0.05 mL min .

The gradient consisted of initial condition

95% A, a linear gradient to 50% A at 5 min, a linear gradient to 5% A at 33 min, and then to 0%
A at 35 min. The mobile phase was held constant until 38 min, when the composition returned
to the original composition.

Spectra were acquired using three quasi-simultaneous conditions:

high energy (aperture 1, 75 V), medium energy (aperture 1, 45V), and low energy (aperture 1, 10
V) in the instrument transit lens.
30

2.4.5

Data Processing

Processing of mass spectra for elemental composition analysis was performed using
MassLynx v 4.1 software (Waters).

For nontargeted metabolite profiling, Waters MarkerLynx

software was employed for peak detection, integration, retention time alignment, and export of
peak areas organized by mass-retention time pairs.

Principal component analysis was

performed using the online resource at http://www.metaboanalyst.ca/.

2.5

Results and Discussion

Details of the multiplexed CID technique have been discussed in the Ph.D. dissertation of
Feng Shi from our lab before [117].

The technique is explained briefly here.

The ion optics

element aperture 1, located behind the orifice of the ion source has relatively high pressure (~
-3

10

mbar) owing to gas leaking from the ion source (Figure II-2).

The voltage applied on the

aperture 1 can be manipulated by the instrument control software, and this potential controls the
ion kinetic energies as they pass through this region of the mass spectrometer.

Low aperture 1

potentials will deposit low energies per ion-molecule collision, and will preserve most of the
pseudomolecular ion signals.

Elevated aperture 1 voltages raise the ion kinetic energies, and

subsequent ion-molecule collisions deposit sufficient internal energy to cause fragmentation.
The comparison of the multiplexed CID spectra with MS/MS product ion spectra and the
similarity of the CID spectra got from those two approaches has been established [117].
31

Due to

the fast spectrum acquisition rate of the TOF instrument (20000 transients/sec), the optics can
switch between multiple voltage values in a single analysis, and spectra acquired under each
condition are stored as separate functions.

This approach enables the molecular ion and

fragment ions to be obtained in separate data acquisition functions in a single analysis.

The

more important feature of this multiplexed CID technique is that it does not rely on a single set
of CID parameters to yield fragment ion spectra across a diverse range of metabolites.

Instead,

multiple conditions are employed with the intent that one set of parameters will generate useful
CID spectra without need for an additional analysis.

The most basic MS/MS scan uses a fixed

collision energy (although some instruments offers ramping energy scan option), which may be
not optimal without the pre-examination of the energy dependence of fragmentation.

The

multiplexed CID technique can easily generate as many as 5 mass spectra in parallel with
different collision energies (CE), and this approach does not require any compound-dependent
optimization.

Furthermore, endogenous metabolites often present across wide concentration

ranges within and between samples, and the multiplexed CID approach reduces signals from
high-abundance metabolites into the linear dynamic range of the instrument.
An additional advantage of the multiplexed CID technique is an improvement in dynamic
quantitative range.

Functions with high CE can knock down saturated signals to within the

linear dynamic range of the instrument.

At the same time, low CE scan will still keep the low

abundant metabolites detectable.
Although mass spectrometry often fails to yield all the information needed for unambiguous
structure assignment of novel metabolites, annotation of detected metabolites can be made based
32

upon combinations of accurate measurements of molecular and fragment ions.

When combined

with known reactions involved in specialized metabolism, this information can lead to
assignments of candidate structures, even though isomeric metabolites may not be
distinguishable by traditional mass spectrometric methods alone.

An example of annotation of

a metabolite detected as a negative ion at m/z 609 in a leaf dip extract of S. habrochaites LA1777
is presented here.

33

Figure II-2 Schematic Diagram of the Waters LCT Premier time-of-flight mass spectrometer. The region indicated in the
diagram is where the ion kinetic energies are influenced by the potential applied to aperture 1. Adapted from Waters Micromass
LCT Premier Mass Spectrometer Operator’s Guide. For interpretation of the references to color in this and all other figures, the
reader is referred to the electronic version of this or dissertation.

34

The TOF mass spectrometer determined the ion at m/z 609.1414 in the lowest Aperture 1
potential data acquisition function.

The assignment of the molecular ion was confirmed by the
-

-

appearance of peaks assigned as [M+formate] and [M+chloride] at m/z 655.15 and 645.12
respectively.

In the spectrum generated using an intermediate CID voltage function (45 V), the

formate and chloride adducts did not survive, and are presumed to eliminate formic acid and HCl
respectively.

The combination of this information confirmed the molecular ion assignment. A

list of possible molecular formulas was generated using Waters MassLynx software using: 1) a
conservative ± 20 ppm mass window; 2) assuming the ion is an even electron ion, which is most
common with electrospray ionization; 3) limiting elemental composition to include only C, H, N,
O, P, S.

For this m/z 609 ion, the range of atoms allowed for each element were set as follows:

C 1-50; H 1-100; N 0-4 (it is rare to find more than 4 N in a hydrophobic metabolite); O 0-20; P
0-2 (more than 1 P usually means a polyphosphate, and most of these are polar and not expected
to be retained on reverse phase chromatography except the lipophilic prenyl pyrophosphates); S
0 (the isotope pattern did not indicate presence of sulfur).

The software evaluated 1801

possible molecular formula based on the criteria 2), 3) and returned a list of 49 molecular
formulas that satisfied ± 20 ppm mass window.

Formulas were then ranked according to

goodness of fit of the abundances of isotopolog ions using the MassLynx iFIT score, and the 10
formulas with the highest scores were selected (Figure II-3) and searched against the online
database ChemSpider (http://www.chemspider.org) to generate a list of potential structures.
Two of the ten formulas gave hits in ChemSpider.

The formula C27H30O16 yielded 20 possible

isomers (total 193 hits), all of which were glycosylated flavonoids (Figure II-4).
35

The other

formula giving matches was C28H26N4O12 with 4 hits (Figure II-5).

Since mass spectrum

-

exhibited an abundant [M-H] ion, it was concluded that the molecule is likely to have one or
more acidic functional groups.

Three of the ChemSpider structures for C28H26N4O12 lacked

acidic functional group, so only hit number 2, which has a carboxylic acid group, was considered
a possible candidate.

36

Figure II-3 The top 10 possible molecular formula for m/z 609.1414, a metabolite ion from S. habrochaites LA1777, generated by
Masslynx software (Waters). DBE (double bond equivalent), i-Fit is expressed in arbitrary units, and measures the goodness of fit
to the isotopic pattern calculated for each elemental composition. Lower i-Fit values reflect a superior fit.

37

Figure II-4 The 2 examples of 20 matching structures found on the ChemSpider database for
formula C27H30O16.

38

Figure II-5 The 4 matching structures found on the ChemSpider database for formula
C28H26N4O12.

39

Figure II-5 (cont’d)

O

O
O

N

N

O

O

N N

O

O
O

O

O

O
O

40

At this stage, information about masses of fragment ions, generated using the higher
Aperture 1 voltage, was incorporated into the metabolite annotation process.

The averaged

mass spectrum across the entire elution profile for m/z 609 (using Aperture 1=75 V) showed the
presence of two fragment ions at m/z 284 and 300.

By aligning the extracted ion chromatogram

of fragment ions with pseudomolecular ion m/z 609, the m/z 284 fragment was assigned to the
same compound as m/z 609 owing to matching chromatographic elution profiles for these two
masses (Figure II-6). Two partially resolved peaks in the XIC for m/z 609 are apparent, with
retention times of 7.56 and 7.89 min.

Mass measurements for these two peaks showed values

of m/z 609 that only differ by 0.8 milliDaltons.

Both peaks displayed a dominant fragment ion

of m/z 284 at elevated Aperture 1 potential, which suggests they are isomers.

The accurate

mass for this fragment ion is m/z 284.0340 for the later eluting peak (theoretical m/z = 284.0326
-

for C15H8O6 , discussed in detail below).

Within the spectra of these peaks, an additional

fragment ion at m/z 300 was observed, but its elution profile did not match the elution for m/z
609.

Instead, the retention time profile of this m/z 300 fragment ion reached a maximum at

7.67 min, and matches the profile of m/z 625.

Both pseudomolecular and fragment ion masses

are 16 Da heavier than the m/z 609 and 284 pair, suggesting an additional oxygen atom in the
former.

41

A

B

C

D

Figure II-6 Extracted ion LC/TOF MS chromatograms (XIC) for an extract of S. habrochaites LA1777 leaf type VI trichomes
(electrospray, negative): A) m/z 609 at low CID potential (aperture 1 = 10 V); B) m/z 284 at high CID (aperture 1= 75 V); C) m/z
625 at low CID (aperture 1 = 10 V); D) m/z 300 at high CID (aperture 1=75 V).

42

To identify possible elemental formulas for the m/z 284 fragment ion, criteria 1) and 3)
mentioned above (mass match window and elemental composition limits) were applied within
MassLynx software to generate a list of candidate formulas for the fragment ion.

Limitation of

ion type was extended to include both odd- and even-electron ions because some metabolites,
particularly those that can stabilize a radical by delocalization, fragment under CID conditions to
generate odd-electron fragments.

The elemental composition calculator generated 530 potential

formulas within the elemental formula constraints, but only 25 results lay within 20 ppm of the
measured mass.

Ten elemental formulas with the highest scores of isotopolog pattern fitting

were considered and these are displayed in Figure II-7.

This list was filtered further by not

allowing any increases in the number of an element during fragmentation, and limiting the
increase in DBE upon fragmentation to no more than +2.
-

For pseudomolecular ion

-

C27H29O16 , the radical anion fragment C15H8O6 (radical ion) is the only formula that meets
these criteria.

This formula also has the lowest DBE per carbon atom, with a value of 0.8.

Formulas with DBE/#C much greater than this are unlikely because such formulas leave limited
opportunities to accommodate hydrogens and other elements, and metabolites with such low
hydrogen counts are rare in nature.
The two molecular ions proposed above were subjected to a search of the ChemSpider
database, which returned numerous structures matching the candidate formulas.

For the

C28H26N4O12 options, no simple fragmentation scheme can account for the fragment ions
observed under any CID conditions.

In contrast, all 20 structures reported for C27H30O are

glycosylated flavonoids with a variety of aglycone and sugar moieties.
43

Plants in the genus

Solanum are already known to make a range of flavonoid aglycones, including kaempferol,
quercetin, and myricetin [118, 119], which only differ in the number of hydroxyl groups. These
findings point to the likelihood that these isomeric metabolites are glycosylated flavonoids.

44

Figure II-7 The top 10 possible elemental formulas for fragment ion m/z 284.0340 obtained by averaging spectra (7.5-7.9 min;
aperture 1= 75 V) from LC/TOF MS analysis of an extract of leaf type VI trichomes from S. habrochaites LA1777 (electrospray,
negative mode).

45

The fragment ion of m/z 284 matches the mass of kaempferol anion radical.

For

kaempferol glycosides, the observed fragment ion mass can be explained by homolytic cleavage
of the glycosidic bond to give the kaempferol anion radical with m/z 284 (Figure II-8). The
formation of a single aglycone radical anion fragment suggests that a diglycoside is attached to a
single position, probably at the 3-position.

However, kaempferol-3,7-di-O-glucoside has been

annotated in tomato species before [120, 121], and in the absence of an authentic standard,
should be considered as a possibility.

The positive mode spectra shows two sequential losses of

162 Da, indicating two hexoses attached to the aglycone (Figure II-8), and would be consistent
with kaempferol-3-7-di-O-glucoside.

The minor and earlier eluting isomer was putatively

annotated as kaempferol-3-sophoroside, which has a diglycoside on the 3-position, based on
previous publication [122].
Annotations of other metabolites were performed following similar procedures.

It is worth

emphasizing that, unlike DNA or protein sequencing, putative annotation of metabolites involves
many manual steps that are not easily automated because the chemical diversity of metabolites is
far greater than the diversity in building blocks of proteins and oligonucleotides.

As a result,

metabolite identification remains one of greatest bottlenecks studying the development of
metabolomics.

Fortunately, mass spectrometric methods that provide rapid determination of

pseudomolecular and fragment ion masses provide the most powerful information for aiding
metabolite identification.
There are four major classes of metabolites annotated from LC/MS analyses performed of
individual trichomes types, with 109 peaks analyzed.
46

These metabolites can be divided into

four different groups: a) flavonoids, b) acylsugars, c) terpenoids, and d) glycoalkaloids.

The

occurrence of these metabolites is described in more detail below.
A variety of flavonoids were detected in the extracts of individual trichome types.

The

flavonoids are polyphenolic compounds found in plants, with more than 4000 diversified
structures having been catalogued.

Most are conjugated to sugar molecules.

flavonoids play in plants remain uncertain.

The roles that

Some proposed mechanisms include protection

against UV-B radiation and defense against pathogen attack [123, 124].
analyses, two subclasses of flavonoids were annotated.

47

Based on the LC/MS

2 glu

A

B
glu
glu

C

Figure II-8 A) Collision induced dissociation (CID) spectra of S. habrochaites LA1777 leaf
type VI trichome extract (retention time 7.89 min) using aperture 1= 75 V in negative mode; B)
Collision induced dissociation (CID) spectra of the same retention time using aperture 1=75 V
in positive mode; C) Proposed fragmentation for kaempferol-3,7-O-diglucoside detected from
LA1777 type VI trichome extracts in ESI negative mode.

48

The first group consists of flavonoid glycosides.

The characteristic fragment ion observed

for all flavonoid glycosides is derived from cleavage at the glycosidic bond as shown in Figure
II-8.

The aglycones for flavonoid glycosides were either kaempferol or quercetin, as

established using accurate fragment mass measurements as described above.

The second

subclass of flavonoids consists of methylated flavonoid aglycones which will be discussed in
chapter IV in detail.

The methylated flavonoids were O-methylated aglycones of myricetin and

kaempferol but methylated quercetins were not found in this investigation.

The flavonoids

detected in individual trichome types are presented in (Table II-2).
Acylsugars comprise the most abundant and diverse group of specialized metabolites
detected in trichomes.
sugar core.

Acylsugars are viscous lipids with various acyl chains attached on the

It constitutes a significant portion of leaf biomass in the wild type tomatoes [125].

Due to the potent anti-insects activity of acylsugars, increasing acylsugar production has long
been a major target of tomato breeding [126-129].

Acylsugars have also been recognized as

potential feedstocks for food and some cosmetic applications [130, 131].

So far, the

mechanism of acylsugar biosynthesis and factors that regulate this biochemistry are not
completely understood yet.

49

Table II-2 List of glycosylated flavonoids and methylated flavonoids detected in all trichome
extracts using LC/MS in negative ion mode. Listed fragment ions were generated using
aperture 1= 75 V aperture 1.
Glycosylated flavonoid
annotation

[M-H]

Quercetin-3-rutinoside

609

655

Major
fragment ions
(m/z)
300

Kaempferol-hexosides

609

655

284

Quercetin-dihexoside

625

671

300

Methyl flavonoid
annotation

[M-H]

monomethyl myricetin
dimethyl myricetin

331
345

Major fragment
ions
(m/z)
316
330, 315

trimethyl myricetin

359

344, 329, 314

tetramethyl myricetin

373

358, 343, 328, 313

monomethyl
kaempferol
dimethyl kaempferol

299

284

M82 (type I, type VI)
LA1777 (both leaf and stem type I,
IV, VI, VII)
LA1777 (both leaf and stem type I,
IV, VI, VII)
LA1777 (both leaf and stem type I,
IV, VI, VII)
LA0716 (type IV, VI)

313

298, 283

LA0716 (type IV, VI),

-

-

[M+formate]

-

Observed
species and types
M82 (type VI),
LA0716 (type VI)
LA1777 (both leaf and
stem type I, IV, VI)
LA1589 (type VI)
LA1708 (type VI)
LA1777 (both leaf and
stem type I, IV, VI),
M82 (type VI)

Observed
species and types

A large number of acylsugars were detected in extracts of individual trichome types.
Under energetic CID conditions, the characteristic fragmentation of acylsugars involves
sequential losses of acyl chains from the sugar core as illustrated in Figure II-9.

Since

acylsugars lack functional groups that readily gain or lose protons, acylsugars are predominantly

50

-

observed as [M+HCOO] in negative mode electrospray ionization mass spectra.

Upon CID,

such acylsugar ions generate a series of fatty acid anion fragments that are observed at the low
mass end of the CID spectra.

From their masses, it is possible to establish which acyl chains

were attached to the sugar core.
acyl chains.

In Figure II-9, m/z 87, 101 and 171 indicates C4, C5, and C10

The presence of acetate ester groups is not apparent owing to the low ion

transmission of the mass spectrometer at low masses, but can be established from mass
differences between the deprotonated acylsugars (which form under slightly elevated CID
energies) and fragment ions in the high mass region, e.g. from the 42 Da difference between m/z
691

649.

The exact positions of acyl chain substitutions cannot be assigned explicitly from

mass spectra because cleavage of sugar carbon-carbon bonds, which would help indicate
positions of esters, have not been observed under any CID conditions.
To simplify naming of acylsugar metabolites, we developed a nomenclature as follows: a
single letter corresponding to the base sugar (G = glucose; S = sucrose) followed by the
designation of number of fatty acyl groups and total number of acyl carbon atoms.

For example,

a sucrose tetraester acylated with C2, C4, C5, C10 fatty acids as mentioned above would be
designated as S4:21

(2,4,5,10)

.

A compilation of acylsugars detected in individual trichome types

is presented in Table II-3, and the differences between trichomes are discussed in more detail
below.

51

[M-H]

-

C10

C4
C5

C2

Figure II-9 Top panel: CID spectra of acylsugar S4:21 (75V aperture 1 voltage) from extract of S. habrochaites LA1777 type I
(2,4,5,10)
trichomes in ESI negative mode; bottom panel: proposed fragment ions observed for S4:21
.

52

Figure II-9 (cont’d)

53

Table II-3 List of acylsugar metabolites detected in trichome extracts using LC/MS in negative
ion mode. Fragment ions were generated using aperture 1= 45 V.
Acylsugar

Fatty acid ester
groups

[M+formate]

-

Major fragment ions
(m/z)

(m/z)
G 3:12
G 3:13

C4, C4, C4
C4, C4, C5

435
449

G 3:14

C4, C5, C5

463

G 3:15

C5, C5, C5

477

G 3:16

C4, C4, C8

491

G 3:17

C4, C5, C8
C4, C4, C9

505

G 3:18

C4, C4, C10

519

G 3:19

C4, C5, C10

533

G 3:20

C5, C5, C10
C4, C4, C12

547

G 3:21

C4, C5, C12

561

G 3:22

C5, C5, C12

347, 213,143, 125, 87
361, 315, 227, 143, 125,
101, 87
375, 329, 227, 143, 125,
101, 87
375, 329, 227, 143, 125,
101
347, 301, 213, 125, 143,
87
361, 315, 227, 125, 143,
101, 87
347, 301, 213, 125,
143,157, 87
347, 301, 213, 143, 125,
171, 87
361, 315, 227, 143, 171,
101, 87
347, 301, 213, 143,125,
171, 101
375, 329, 227, 143,125,
199, 87
361, 315, 227, 143, 199,
101, 87
375, 329, 227, 143, 199,
101

575

54

Observed
species and
types

LA 0716 (type
IV, VI)

Table II-3 (cont'd)
S 3:12

C4, C4, C4

S 3:13

C4, C4, C5

S 3:14

C4, C5, C5

S 3:15

C5, C5, C5

551, 481, 411, 341, 87

597
611
625

565, 495, 411, 341, 101, 87
579, 495, 411, 341, 101, 87
593, 425, 407, 341, 323,101

639

S 3:18

C4, C4, C10

S 3:19

C4, C5, C10

S 3:20

C5, C5, C10
C4, C4, C12

681
695

709

S 3:21

C4,C5,C12
723

S 3:22

C5,C5,C12
737

S 3:24

C4, C10, C10

S 3:25

C5, C10, C10

S 4:14
S 4:15

C2,C4,C4,C4
C2,C4,C4,C5

765
779
639
653

LA1777 (both
leaf and stem type
I, IV, VI, VII)

LA1777 (both
leaf and stem type
I, IV, VI, VII)
M82 (type I, type
VI)
LA1777 (both
leaf and stem type
I, IV, VI, VII)

635, 565, 481, 411, 341, 171,
87
565,495,425,407,341,171,
101
663, 579, 509, 425, 407, 341,
LA1777 (both
323, 171, 101
leaf and stem type
663, 593, 481, 397, 341, 199,
I, IV, VI, VII)
87
M82 (type I, type
VI)
677, 593, 495, 425, 341, 199,
LA1777 (both
101, 87
leaf and stem type
I, IV, VI, VII)
691, 607, 509, 425, 341, 199,
LA1777 (both
101
leaf and stem type
I, IV, VI, VII)
M82 (type I, type
VI)
719, 649, 565, 395, 411, 341,
171, 87
LA1777
(both
733, 629, 579, 425, 341, 171, leaf and stem type
I, IV, VI, VII)
101
593, 551, 87
607, 565, 101, 87

55

Table II-3 (cont'd)
S 4:16

C2,C4,C5,C5

667

S 4:17

C2,C5,C5,C5

681

S 4:20

C2,C4,C4,C10

723

S 4:21

C2,C4,C5,C10

737

S 4:22

C2,C4,C4,C12

751

S 4:23

C2,C4,C5,C12

765

S 4:24

C2,C5,C5,C12

779

621, 579, 537, 495, 393, 341,
LA1777 (both
323, 101, 87
leaf and stem type
I, IV, VI, VII)
635, 551, 509, 425, 407, 341,
LA1777 (both
323, 101
leaf and stem type
I, IV, VI, VII)
M82 (type I, type
VI)
677, 635, 481, 411, 393, 341,
LA1777 (both
323, 101, 87
leaf and stem type
I, IV, VI, VII)
691, 649, 495, 393, 341, 323,
171, 101, 87
709, 663, 481, 393, 341, 323,
LA1777 (both
305, 199, 87
leaf and stem type
I, IV, VI, VII)
M82 (type I, type
VI)
719, 677, 495, 411, 341, 323,
LA1777 (both
199, 101, 87
leaf and stem type
I, IV, VI, VII)
733, 691, 509, 425, 407, 341,
LA1777 (both
323, 199, 101
leaf and stem type
I, IV, VI, VII)
M82 (type I, type
VI)

Trichomes also accumulate significant quantities of terpenoid metabolites.
derived from the five carbon isoprene substructural unit.

Terpenoids are

The terpene biosynthesis pathways are

important to plant growth because some plant growth regulators such as gibberellins are
terpenoid metabolites.

Structures of numerous other important metabolites also incorporate

intermediates of the terpene biosynthesis such as the phytol tail group in chlorophylls.

56

A large

proportion of interest in terpenes has focused on volatile terpenes, primarily the mono- and
sesquiterpenes [113].

A variety of sesquiterpenes accumulating in S. habrochaites LA1777

were reported before [132-134], including carboxylic acid derivatives of α-santalenoic and
bergamotenoic acids.

Analyses of extracts of this accession using LC/MS in negative ion mode

revealed three major isomers having m/z 233.15, consistent with these acids.
assignment was based on exact mass and fragments (Figure II-10).

Their putative

Most notable is the

formation of fragment ions at m/z 98 and 99, which are consistent with fragmentation at the
allylic carbon-carbon bond.

The former fragment ion is an unusual odd-electron fragment

derived from an even-electron precursor.
allylic radical by delocalization.

Its formation is attributed to stabilization of the

Several additional terpenoid acids from S. habrochaites

LA1777 were also annotated based on accurate mass measurements and the calculated relative
mass defect [135].
Despite the importance of terpenoid metabolites across plant, animal, and microbe species,
tools for recognizing terpenoids based on negative ion mass spectra have been slow to develop.
Terpenoid core structures are largely based on alkane and alkene moieties that are resistant to
charge-directed fragmentation in negative ion mass spectra.

It is proposed that a strong

indicator that a metabolite is terpenoid is that these metabolites have moderately high hydrogen
content.

Owing to the positive mass defect of the hydrogen atom, such molecules have high

positive “relative mass defects”.

A brief discussion of the importance of mass defects is in

order. Mass defect is defined as the difference between true mass and the integer, or nominal
mass.
57

[M-H]-

Figure II-10 Top panel: LC/MS product ion spectrum of m/z 233 with retention time 30.77 min from S. habrochaites LA1777 type
VI trichome extract (ESI, negative) generated on a QTof mass spectrometer (45 V collision voltage); middle panel: LC/MS XIC of
m/z 233 (aperture 1= 10 V) from LA1777 type VI trichome extracts in ESI negative mode, full scan MS analysis; bottom panel:
proposed fragmentation for α-santalenoic acid.

58

Figure II-10 (cont’d)

59

Figure II-10 (cont’d)

For example, the hydrogen atom (accurate mass: 1.00794 u) has positive mass defect of
0.00794 u.

The mass of

12

C defines the mass scale at exactly the integer mass, therefore carbon

atoms do not contribute to mass defect.

Most elements heavier than carbon such as oxygen

(accurate mass: 15.9994 u) have masses slightly less than their nominal masses owing to their

60

nuclear binding energies.

1

Hydrogen atoms ( H) have the highest positive mass defect because

there is no nuclear binding energy.

Furthermore, the number of hydrogen atoms in many

metabolites exceeds the number of atoms of any other type (excluding carbon), and therefore the
mass defect largely reflects the number of hydrogen atoms in a molecule.

The elemental

compositions of metabolites derived from specific biochemical pathways largely reflect their
precursors, particularly with regard to hydrogen content.

The notion of a relative mass defect,

which is calculated by dividing the total mass defect of the molecule by its molecular mass,
provides a measure that is strongly correlated with the %H by weight for that molecule.

Since

terpenes are synthesized from C5H8 building blocks, they have higher positive relative mass
defect than metabolites such as flavonoids which are derived from precursors of lower hydrogen
content.

For example, the flavonoid kaempferol (C15H10O6; 3.52% H), has molecular mass of

286.0477 u, and the sesquiterpene acid α-santalenoic acid (C15H22O2, 9.46% H) has molecular
mass of 234.1625 u.

The relative mass defect for the former is 0.0477/286.0477=166 ppm,

whereas the more hydrogen-rich santalenoic acid has a relative mass defect of 693 ppm.

In

extracts of S. habrochaites LA1777, an unknown metabolite with nominal m/z 319 in negative
ion mode was annotated as a diterpenoid based on the following factors.

a) its relative mass

defect of 765 ppm suggested a hydrogen content similar to the sesquiterpene acids. b) the
deprotonated ion underwent loss of water upon CID, but further fragmentation was not observed,
as is typical for terpenoids c) the accurate mass measurement (m/z 319.2441) suggests it has one
more isoprene unit C5H8 and addition of H2O relative to the LA1777 sesquiterpene acids.
There are also several sesquiterpene metabolites decorated with sugars and malonate esters as
61

listed in Table II-4.
Table II-4 List of terpenoid metabolites detected in extracts of individual trichome types using
LC/MS in negative ion mode. Fragment ions were generated using aperture 1= 75 V except for
the sesquiterpene acids, which were analyzed as product ion spectra using a QTof mass
spectrometer with collision potential of 45 V.
Terpenoid metabolites

Observed
species and types

(m/z)

Major fragment
ions
(m/z)

Sesquiterpene acids
(α-santalenoic,
bergamotenoic acids)

233

99, 98

LA1777 (both leaf
and stem type I, IV,
VI, VII)

Sesquiterpene diglycoside,
malonate ester

649

605, 563, 401

LA1777 (both leaf
and stem type I, IV,
VI, VII)

Sesquiterpene diglycoside,
malonate diester

717

629, 587, 545, 383

LA1777 (both leaf
and stem type I, IV,
VI, VII)

Sesquiterpene diglycoside,
malonate ester, acetate ester

691

647, 605, 443, 401

LA1777 (both leaf
and stem type I, IV,
VI, VII)

Sesquiterpene
monoglycoside, malonate
ester

487

443, 401

LA1777 (both leaf
and stem type I, IV,
VI, VII)

diterpene acid

319

301

LA1777 (both leaf
and stem type I, IV,
VI, VII)

[M-H]

-

62

Trichomes of wild tomatoes also accumulate two glycoalkaloids.

Tomatine, which is the

most well known alkaloid in tomato was detected in LA1777 type VII leaf trichome extracts but
trace amounts in other trichome types.

Annotation was based on the accurate mass

measurements, the fragmentation which largely occurs at the glycosidic bonds (Figure II-11),
and comparisons to authentic standards.

A related compound which has similar retention time

and neutral losses as tomatine but is 2 Da lighter than tomatine is annotated as dehydrotomatine.
Glycoalkaloids detected in various trichome types are presented in Table II-5.

Table II-5 List of glycoalkaloid metabolites detected in extracts of individual trichome types
using LC/MS in positive ion mode. Fragment ions were generated using aperture 1 = 45 V.
Alkaloid glucoside

[M+H]

-

tomatine

(m/z)
1034.5

dehydrotomatine

1032.5

Major fragment ions
(m/z)

Observed
species and types

903, 740, 578, 560, 416,
398, 273, 255

LA1777 (leaf
VII)

901. 738, 576, 558, 396,
271

LA1777 (leaf
VII)

63

[M+H-xylose]+

Figure II-11 Top panel: CID spectra of (tomatine) in ESI positive mode (aperture 1= 75 V) from S. habrochaites LA1777 type VII
trichome extract; bottom panel: proposed fragmentation of protonated tomatine.

64

Figure II-11 (cont’d)

HO
HO
O m/z 903
HO
OHOH
m/z 578
O
O O
OH
HO O m/z 740 O
O
HO
OH
O
HO
OH
OH
m/z 560

H
H
H H H
H

H

ON
H2
HO

H

H
H
H HH
H m/z 255

65

ON -H2O
H
H
H 2
H H
H m/z 416

-H2O
HO

H
H HH
H
m/z 273

m/z 398

Since S. arcanum (LA1708) and S. pimpinellifolium (LA1589) have only one trichome type
that was collected, and minimal levels of metabolites were detected in those two species, except
for glycosylated flavonoids.

As a result, the discussion below focuses on the conventional

tomato S. lycopersicum (M82) and its wild relatives S. habrochaites LA1777 and S. pennellii
LA0716.
An intraspecies comparison of trichome composition was first conducted using LC/MS
profiling.

For S. habrochaites LA1777, we both stem and leaf trichomes were collected,

extracted, and analyzed using LC/TOF MS.

From Figure II-12 to Figure II-16, metabolite

profiles of stem and leaf trichomes of the same morphological types are almost indistiguishable.
Type VI trichomes accumulate abundant glycosylated flavonoids (Figure II-12).

However,

methylated flavonoid aglycones, are predominantly found in type I and IV trichomes (Figure
II-13).

None of these extracts revealed the presence of flavonoids modified by both

methylation and glycosylation, and it is hypothesized that each of these modification reactions
precludes

the other.

These biochemical phenotypes suggest

the possibility that

glycosyltransferases are highly expressed in type VI glands, whereas O-methyltransferases are
abundant in types I and IV.
Terpenoid metabolites exhibit a strong selectivity of accumulation in type VI trichomes
(Figure II-14).
terpenoids.

LA1777 is the only accession studied that accumulates substantial amount of

For this reason, the catalog of genes expressed selectively in type VI glands

presents a promising source of information regarding candidates for terpenoid biosynthesis genes.
For acylsugars the selectivity is reversed.

Type I and IV glands accumulate at least 3-fold more
66

acylsugar per trichome than types VI and VII (Figure II-15).

Type VII trichomes comprise the

only type among all plant species and gland types that accumulates substantial amounts of
glycoalkaloid (Figure II-16).

The type VII trichome in LA1777 grows closest to the epidermal

pavement cells, which also store tomatine.

The possibility that sampling of type VII glands

also sampled the pavement cells cannot be excluded.

67

-

Intergrated XIC peak area for [M+formate]

4.0E+04

3.0E+04

2.0E+04

1.0E+04

0.0E+00
Leaf
Leaf
Leaf
Type I Type IV Type VI

Leaf
Type
VII

Stem
Stem
Stem
Type I Type IV Type VI

Figure II-12 Quantitative distribution of glycosylated flavonoids in extracts of each type of trichome (50 trichomes extracted in 50
µl solvent) from leaf and stem tissues of S. habrochaites (LA1777) measured using negative mode LC/MS. Key:
Kaempferol-dihexosides (red) and quercetin –dihexoside (yellow). Error bars indicate standard error with n=3.

68

Intergrated XIC Peak Area for [M-H]

-

5.0E+04
4.0E+04
3.0E+04
2.0E+04
1.0E+04
0.0E+00
Leaf
Leaf
Leaf
Type I Type IV Type VI

Leaf
Type
VII

Stem
Stem
Stem
Type I Type IV Type VI

Figure II-13 Quantitative distribution of methylated flavonoids in each types of trichome (from 50 trichomes extracted in 50 µl
solvent) from S. habrochaites (LA1777) measured using negative mode LC/MS. Key: Dimethyl myricetin (blue) trimethyl
myricetin (red) tetramethyl myricetin (yellow). Error bars indicate standard error with n=3.

69

-

Intergrated XIC Peak Area for [M-H]

5.0E+05
4.0E+05
3.0E+05
2.0E+05
1.0E+05
0.0E+00
Stem
Type I

Stem
Type IV

Stem
Leaf Type Leaf Type Leaf Type
Type VI
I
IV
VI

Figure II-14 Quantitative distribution of terpenoid metabolites in each types of trichome (from 50 trichomes extracted in 50 µl
solvent) from S. habrochaites (LA1777) measured using negative mode LC/MS. Key: sesquiterpene acids (blue), sesquiterpene
diglucoside, malonate ester (red), diterpene acid (yellow). Error bars indicate standard error with n=3.

70

-

Intergrated XIC peak area for [M+formate]

1.0E+07
8.0E+06
6.0E+06
4.0E+06
2.0E+06
0.0E+00
Leaf
Leaf
Leaf
Leaf
Type I Type IV Type VI Type
VII

Stem
Stem
Stem
Type I Type IV Type VI

Figure II-15 Quantitative distribution of total acylsucroses in each types of trichome (from 50 trichomes extracted in 50 µl solvent)
from S. habrochaites (LA1777) measured using negative mode LC/MS. Error bars indicate standard error with n=3.

71

-

Intergrated XIC peak area for [M +form ate]

2.0E+03
1.5E+03
1.0E+03
5.0E+02
0.0E+00
Stem
Type I

Stem
Stem
Type IV Type VI

Leaf
Type I

Leaf
Leaf
Leaf
Type IV Type VI Type VII

Figure II-16 Quantitative distribution of glycoalkaloids in each types of trichome (from 50 trichomes extracted in 50 µl solvent)
from S. habrochaites (LA1777) measured using negative mode LC/MS. Key: tomatine (blue), dehydrotomatine (red). Error bars
indicate standard error with n=3.

72

For S. pennellii LA0716, metabolites are less diversified compared to S. habrochaites
LA1777, as the former did not accumulate terpenoids or alkaloids detected by LC/MS.

In

contrast to LA1777, LA0716 stored higher levels of methyl flavonoids in type VI glands with
lesser amounts in type IV (Figure II-17).

Type VI glands in the latter accession also contain

more acylsugars, predominantly acylglucoses, and glycosylated flavonoids (Figure II-18, Figure
II-19).

In virtually all aspects, type VI glands in LA0716 are more biosynthetically prolific than

type IV trichomes in the same species.

73

-

Intergrated XIC peak area for [M-H]

4.0E+04
3.0E+04
2.0E+04
1.0E+04
0.0E+00
Type IV

Type VI

Figure II-17 Quantitative distribution of methylated flavonoids in each types of trichome (from 50 trichomes extracted in 50 µl
solvent) from S. pennellii (LA0716) measured using negative mode LC/MS. Key: monomethyl kaempferol (blue), dimethyl
kaempferol (red). Error bars indicate standard error with n=3.

74

-

Integrated XIC peak area for [M+formate]

3.0E+02

2.0E+02

1.0E+02

0.0E+00
Type IV

Type VI

Figure II-18 Quantitative distribution of glycosylated flavonoid (rutin) in each types of trichome (from 50 trichomes extracted in
50 µl solvent) from S. pennellii (LA0716) measured using negative mode LC/MS. Error bars indicate standard error with n=3.

75

-

Integrated XIC peak area for [M+formate]

5.0E+04
4.0E+04
3.0E+04
2.0E+04
1.0E+04
0.0E+00
Type IV

Type VI

Figure II-19 Quantitative distribution of total acylsugar metabolites in each types of trichome (from 50 trichomes extracted in 50
µl solvent) from S. pennellii (LA0716) measured using negative mode LC/MS. Error bars indicate standard error with n=3.

76

Levels of specialized metabolites in trichomes from the conventional tomato S.
lycopersicum (M82) were substantially less than the other accessions with a few exceptions.
However, the distribution of specific metabolites classes is similar to S. habrochaites LA1777.
Type VI glands accrue more glycosylated flavonoids than type I trichomes (Figure II-20),
however type I accumulates more methylated flavonoids and acylsugars than type VI trichome
(Figure II-21, 22).

77

-

Intergrated peak area for [M+formate]

3.0E+05

2.0E+05

1.0E+05

0.0E+00
Type I

Type VI

Figure II-20 Quantitative distribution of glycosylated flavonoids in each types of trichome (from 50 trichomes extracted in 50 µl
solvent) from S. lycopersicum (M82) measured using negative mode LC/MS. Key: Rutin (blue) and quercetin-diglucoside (red).
Error bars indicate standard error with n=3.

78

-

Integrated XIC peak area for [M-H]

1.2E+04

9.0E+03

6.0E+03

3.0E+03

0.0E+00
Type I

Type VI

Figure II-21 Quantitative distribution of methylated flavonoids in each types of trichome (from 50 trichomes extracted in 50 µl
solvent) from S. lycopersicum (M82) measured using negative mode LC/MS. Key: monomethyl myricetin (blue), dimethyl
myricetin (red). Error bars indicate standard error with n=3.

79

-

Integrated XIC peak area for [M+foramte]

1.2E+06

8.0E+05

4.0E+05

0.0E+00
Type I

Type VI

Figure II-22 Quantitative distribution of total acylsugar metabolites in each types of trichome (from 50 trichomes extracted in 50
µl solvent) from S. lycopersicum (M82) measured using negative mode LC/MS. Error bars indicate standard error with n=3.

80

The systematic comparative analysis of different species and types of trichome for Solanum
species has not been reported before.

The research described in this Chapter reports

determination of metabolite identities and quantities for individual types of trichomes in several
Solanum accessions.

Integration of this information with genomic sequence information and

assessment of mRNA transcript levels provides a foundation for determining where specific
specialized metabolic processes occur.

Comparisons of metabolite profiles in five accessions

were made based on annotations of 109 mass spectral features that were extracted from these
profiles using the automated peak detection, integration, and alignment capabilities of
MarkerLynx software, which reports each feature as a mass-retention time pair for which a peak
area is tabulated.
text file.

This database of metabolite levels was exported and organized further as a

Despite this leap in our understanding of trichome chemistry, there are still thousands

of unidentified peaks for which it is not feasible to make manual comparisons.

One useful tool

for visualizing data of such complexity employs principal component analysis (PCA), which is
an unsupervised linear technique for reduction of dimensionality.

The central idea is to reduce

large dimensional data to a few principal components that represent a large proportion of the data
variance.

Graphic representation of much of the variance can be accomplished by generating

two- or three-dimensional plots of the scores and loadings of the various measurements onto the
principal components.

The data were first scaled by dividing each peak area by the standard

deviation of that peak area calculated across the entire data set.

After scaling, all metabolites

have a standard deviation normalized to one, and therefore further processing is based on
correlations instead of covariances.

In another word, all metabolites become equally weighted
81

without bias to metabolites based on their abundances.

In the PCA scores plot derived from the

scaled peak areas for metabolites from individual trichome types (Figure II-23A), it is evident
that the species differences are more apparent than trichome type differences.

The different

types of trichomes cluster together, however, the different accessions are well separated.
II-23B presents the PCA loadings plot including all of the metabolites.

Figure

The loadings plot

displays each metabolite as a single data point with the position corresponding to the loadings on
each principal component.

For the scores plot, each data point represents an individual sample,

and the scores are calculated by summing the loadings for all of the individual metabolites on the
respective principal components.

For example, the left upper corner in the loadings plot

contains those metabolites that are more abundant in M82, S. Arcanum, and S. pimpinellifolium
compared to other species because M82, S. arcanum, and S. pimpinellifolium samples are
situated on the upper left corner of the corresponding scores plot.

The right side of the loadings

plot is dominated by those metabolites that are abundant in LA1777, and in turn, LA1777
samples are situated on the right side of the scores plot.

Since representing thousands of

metabolites on the loadings plot would compromise labeling of all individual metabolites with
their names, metabolites were labeled as M (number) to display all of the metabolites.
The loadings plot of trichome metabolite data indicates that only a small set of metabolites
(those located in the upper left corner) are distinguishing features of cultivated tomato (M82), S.
arcanum LA1708, and S. pimpinellifolium LA1589, regardless of trichome types.

The

similarity of M82 to S. arcanum and S. pimpinellifolium is consistent with a previous report
[136], which also recognized that specialized metabolites are less abundant in these accessions
82

than in the high acylsugar lines S. pennellii LA0716 and S. habrochaites LA1777.

The

differences in metabolite profiles may explain the lower resistance of the cultivated tomato to
herbivores compared to wild type [137-139].

83

A

Figure II-23 Principal component analysis (PCA) of individual trichome metabolite A) score
plot for first and second principal. B) loadings plot including all of the detected metabolites
(“M” represents metabolites). L means leaf and S means stem.

84

Figure II-23 (cont’d)

B

Metabolite profiles for S. habrochaites LA1777, the only accession that has all types of
glandular trichomes (type I, IV, VI, VII), was analyzed by PCA separately from other accessions
to highlight differences between trichome types (Figure II-24).

It is evident from the PCA

scores plot that type I and type IV cluster together regardless of whether the trichomes were

85

collected from stem or leaf tissues.
and IV.

However, type VI profiles are well separated from types I

Again, differences between stem and leaf trichomes profiles are small, as leaf and stem

profiles cluster together on the PCA scores plot.

The morphologies of type I and type IV glands

are similar, with both having a small gland tip and a long multicellular stalk. Results from this
metabolite profiling show for the first time that the chemical content of these two trichome types
is remarkably similar.

Transcript analyses also failed to find significant differences between

these two types [140].

It should be considered that type I and type IV may be different

developmental stages of the same trichome type with one more developed than the other.
Another finding from the PCA scores plot is the location of type VII trichome profiles between
types I/IV and type VI (Figure II-24).

Since most of the metabolites observed in type VII

trichomes were also detected in other types of trichome which lie at greater heights above the
leaf surface, it is suspected that exudates from the other types may have contaminated type VII
trichomes before or during sampling.

Two exceptions are the glycoalkaloids tomatine and

dehydrotomatine, which exclusively present in the type VII trichomes.

Except for these two

metabolites, other annotated substances do not exclusively present in only one type of trichome.
Metabolites profiles indicate that levels of terpenoid metabolites are present at ~100-fold greater
levels per trichome than types I and IV (Figure II-14). Cross-contamination alone cannot
account for such differences.

It is proposed that the transcripts of type VI trichomes ought to be

valuable for identifying candidate genes for terpenoid metabolite biosynthesis.

86

Figure II-24 Principal component analysis (PCA) of individual trichome metabolite from
LA1777, displaying the scores plot for the first and third principal components. L means leaf
and S means stem. The number after the underscore means biological replicate number.

The integration of metabolomic and transcriptomic data is ongoing at this point, and an
initial report has been published [140].

At this stage, elucidation of metabolic pathways is

incomplete and this may be attributed to one or more of the following:
a) the RNA profiling is not deep enough to cover low abundant transcripts;

87

b) some metabolites of interest may not be synthesized in their entirety within trichome,
and import of key precursors into the trichome may be important;
c) the metabolite is synthesized by trichome, but may only be synthesized at a
developmental stage earlier than mRNA collection;

key transcripts involved in

metabolite biosynthesis may no longer be present when trichomes were collected.
To answer questions b) and c), future investigations should collect trichomes at multiple
developmental stages for both RNA and metabolite profiling.

Awareness of this important

aspect of experimental design represents a new challenge to the field of metabolomics, but is
consistent with the dynamic nature of metabolite biosynthesis.
However, sampling of multiple cell types at different stages of development and different
time points will cause a dramatic increase in the workload of tissue collection and metabolite
analysis.

In addition, for trichome analysis, when plant tissues are young and trichomes are at

an early stage of development, the high density of trichomes prohibits manual collection of
individual trichome types.

Therefore, a different and high-throughput scheme is needed to

achieve the goal of dynamic metabolite analysis which includes localization of metabolite
profiles at locations across plant tissues.

In Chapter III, a high throughput imaging mass

spectrometry scheme for profiling leaf trichomes without manual microsampling is presented,
and this approach could provide an appealing alternative that can address these scientific issues.

88

CHAPTER THREE
CHEMICAL IMAGING OF TRICHOME SPECIALIZED METABOLITES USING
CONTACT PRINTING AND LASER DESORPTION/IONIZATION MASS
SPECTROMETRY

3.1

Introduction

Biological functions are organized at molecular, subcellular organelle, cellular, tissue, and
organismal levels, but the low levels of cellular constituents have provided a barrier to
understanding and exploiting genomic information for control of biological functions.
Measurements of the entire suite of metabolites, known as metabolomics, has emerged as a
growing research area which is trying to answer how living systems respond to environmental
stimuli or genetic modifications [141-143].

Most investigations of metabolite profiles have

focused on whole tissue or whole organism levels, and for these studies, metabolite information
represents averages over many cells and tissues and is dominated by contributions from the most
abundant cell types [44].

However, as documented in Chapter II, metabolite accumulation is

often organized in specialized cell types, such as glandular trichomes in plants. To investigate
the functional roles of specialized cells and address the important questions such as the
mechanism of cell differentiation and cell-cell signaling, studying individual cells is necessary.
Several recent reports suggest chemical heterogeneity of cells, and it is likely that even cells of
the same cell type may differ in metabolic functions depending on cell environments and stages
89

of development.

These findings point to the necessity of individual cell analysis if research is

to fully understand mechanisms that regulate metabolic dynamics [144-146].
A variety of approaches have been explored to address the challenges of sampling and
detecting metabolites in single cells.

For sampling, microcapillaries or micropipettes coupled

with micromanipulation under a dissecting microscope [50, 147, 148] and laser-capture
microdissection (LCM) [149] are successfully used for single cell isolation.

However these

sampling methods offer low throughput if many cells need to be isolated, as is desirable because
metabolite profiles change with organism growth and development.

For individual cell analysis,

flow cytometry [150], microfluidics [151], capillary electrophoresis (CE) [152] coupled with
electrochemical (ECD) [153], laser-induced fluorescence (LIF) detection [154] or mass
spectrometry have been widely used because the detection schemes offer low levels of detection
necessary for single cell analysis.

Of these analytical methods, mass spectrometry offers

near-universal detection of an assortment of metabolites, and is most appropriate for nontargeted
metabolite profiling.

Mass spectrometry based single cell analysis was pioneered in the late

1990s with peptide and biomarker analysis, as described by the Sweedler group at the University
of Illinois [155].

More recently, the laboratory of Renato Zenobi demonstrated metabolite

profiling of yeast with single cell sensitivity [156].

A new nanostructure-initiator mass

spectrometry having the capability to analyze metabolites from single cells has also been
reported by the Siuzdak laboratory at Scripps [71].
To develop a comprehensive understanding of cell metabolic heterogeneity, metabolite
profiles need to be measured for many individual cells.
90

Mass spectrometry imaging (MSI)

provides an approach for multiplexing cellular profiling by fast scanning through an area
containing multiple cells, acquiring chemical information for individual pixels across the entire
area.

Several MSI approaches have been demonstrated in recent years, with matrix assisted

laser desorption/ionization (MALDI) imaging finding extensive use in profiling proteins, lipids,
and other metabolites in animal tissues.

More recently, a few examples have been reported for

localization of natural products in biological tissues from a range of organisms [157-159].
However, the MALDI imaging scheme has a few limitations, 1) sophisticated sample preparation
is often required, 2) signals from the MALDI matrix often interfere with detection of low
molecular mass (< 500 Da) metabolites, and 3) application of the laser-absorbing matrix to
tissues has suffered from lack of uniformity, and the matrix deposition process may compromise
localization of some metabolites.
The mechanisms of desorption and ionization for MALDI are complex and continue to be
subjects of discussion and debate.

The most accepted theory proposes a two step process:

desorption/ablation and ionization [160].

During cocrystallization with matrix, the analyte

molecules are surrounded and isolated by the excess matrix molecules, which have strong
absorption at the laser wavelength.

Upon laser irradiation, the matrix absorbs energy which is

rapidly converted into internal and translational energy, i.e. heating.

This heating leads to

explosive evaporation and ejection of the matrix and analyte molecules into gas phase.

The

ejected materials consist of a combination of desorbed molecules, condensed clusters, droplets or
chunks.

The phase explosion has been successfully simulated by hydrodynamic models [161]

and molecular dynamic calculation [162].

Ionization that results from this process is even less
91

understood than the desorption process.

Two major models have been purposed.

One model

assumes that the matrix molecules were first ionized by direct photoionization, usually through a
multiphoton process [163].

Although the ionization potential is somewhat lower in solid crystal

compared to gas phase and the energy gap can be compensated by thermal energy, at least two
laser photons (3.68 eV for 337 nm laser) are required for the direct photoionization of matrix
molecules (> 7 eV) [164].

The flux of laser photon during a typical MALDI experiment is too

low for resonant two-photon absorption to be efficient.

Instead, it has been proposed that the

interaction of Π electrons between aromatic molecules in condensed phases leads to energy
pooling, which involves combining energy from two electronically excited molecules into a
single molecule [165].

After primary ionization of the matrix molecules, transfer of charged

groups such as protons takes place as a result of collisions between two matrix molecules.
Subsequent collisions between ionized matrix and analyte in the high density plume lead to a
secondary proton transfer reaction and protonation of the analyte [166].

This second model was

purposed by Karas and coworkers, and is referred to as the “lucky survivor” model [167].

It

has also been proposed that some analytes exist in ionized forms in the matrix as they are in
solution.

The model assumes the desorption process breaks the lattice into small clusters.

Some clusters can have excess positive charge or negative charge.

Most of the ionized analytes

are subsequently reneutralized by collisions with counterions within the plume and only the
lucky survivor can be detected by mass spectrometry.
Analytes can also be desorbed/ionized from many semiconductor materials surfaces
including silicon and graphite. Proposed mechanisms of analyte desorption and ionization
92

under matrix free laser desorption ionization (LDI) conditions are largely speculative, with
limited experimental evidence.
laser wavelength [168].

In such experiments, ion signal was found to be independent of

It was explained that only absorption of a photon that exceeds the band

gap energy of the semiconductor materials is required.

A thermal desorption process has been

proposed to be involved in desorption ionization of analytes under these experimental conditions
[169, 170].

Rapid heating can cause fluctuation in electron densities, and the region of intrinsic

instability that results may lead to redistribution of energy into translational energies, leading to
evaporation of analytes.

Subsequent ionization could involve various ion-molecule reactions

including proton transfer, disproportionation reactions, and cation or anion attachment.

Such

rapid heating will increase bond distances and facilitate charge separation from interfacial water
or the acid –OH group on the semiconductor surfaces [171-173].

Gas phase ion-molecule

reactions between proton or cation donors and analytes can form ions by attachment of protons
or cations such as alkali metal ions.
Water has been developed as alternative option as an endogenous matrix for mass
spectrometry imaging [83, 174].

However the atmospheric pressure matrix assisted laser

desorption/ionization (AP-MALDI) system, which is necessary to prevent water evaporation
during plant tissue imaging, normally has lower ion transmission efficiency than vacuum
MALDI, and may be limited to detection of abundant metabolites.

Recently, an infrared laser

ablation electrospray ionization (LAESI) system was developed that employs a fiber tip to focus
radiation onto a small (~ 30 µm) spot followed by electrospray assisted ionization of desorbed
neutrals, and this approach has been applied for profiling of constituents in single cells [74, 175],
93

however the instrument design and parameter optimization required special expertise, and this
report did not demonstrate formation of mass spectrometry images.

Desorption Electrospray

Ionization (DESI) has been used as a new alternative approach for imaging mass spectrometry.
The resolution is generally around few hundred µm owing to dispersion of the primary ion beam
at atmospheric pressure, and this approach has yet to demonstrate resolution of single cells [176].
Although 40 µm resolution has been achieved more recently, such performance places
restrictions on the geometry of the surface, and uniform surface texture is required which is not
always feasible for some tissues [177].

It has been reported that great effort was required to

adjust instrument parameters to achieve optimal signal [178].

Secondary ion mass spectrometry

(SIMS) imaging provides another alternative ionization method, and can give superior spatial
resolution at 50 nm which is adequate for profiling individual cells [179, 180].

However SIMS

requires ultrahigh vacuum, and special sample preparation for fresh tissues is needed that may
involve freezing and water evaporation.

SIMS deposits much higher energy per unit area than

MALDI or DESI, and leads to extensive formation of fragment ions.

In addition, the ions

ejected by SIMS are formed only from a small area, and the abundances of metabolites in such
small areas may preclude measurements except for the most abundant surface species.
Although much effort has been devoted to method development for imaging mass spectrometry,
most successful examples have employed sectioned animal tissue or tissues with flat surfaces.
Examples of MS imaging of tissues having irregular shapes are scarce.
Plant trichomes (Figure III-1) are specialized epidermal cells that protrude from surfaces of
plant tissues such as stems and leaves, and they are found on approximately one third of all
94

vascular plant species.

In earlier reports, trichomes from the model plant Arabidopsis thaliana

were chosen as a representative model for investigating cell differentiation [181, 182].
Trichomes are also prolific accumulators of specialized metabolites used by plants as chemical
defenses.

They are considered as one of the most effective “chemical factories” in the plant

kingdom [183].

Figure III-1 Optical image of S. habrochaites LA1777 leaf (left) showing trichomes on the
adaxial (upper) leaf surface including type VI glands (right). Adapted from Reference [140].

Current research focuses on understanding the mechanisms that govern metabolite
accumulation in tomato trichomes because trichome chemicals serve as the first layer of
protection against insects and other herbivores.

Agricultural crop losses to insect damage

represent a large economic loss to the worldwide economy, and the selective breeding of crops
that are resistant to insects has great potential impact on yields of numerous crops.
One fascinating feature of tomato trichomes is that they occur, in the genus Solanum, with
four distinct morphological types [184]. These specialized cell types provide a useful and
95

accessible model system in which different cell types are expected to accumulate different
metabolites, and these differences promote discoveries of gene functions relevant to trichome
biochemistry.
Most analytical methods developed for profiling of metabolites in individual cell types are
laborious because of the need to selectively sample specific cell types.
accomplished by microcollection of individual cells under a microscope.

This can be

However, the number

of cells that can be collected in a short time is limited, perhaps to a few hundred cells per day.
Before this work was initiated, most knowledge of trichome chemistry had come from
nonselective mechanical removal of multiple trichome types, often following freezing of the
plant tissue.

The resulting chemical profiles have represented mixtures of trichome types, and

have not exploited the diversity in cell types [185-187].

Furthermore, these efforts have pooled

trichomes from multiple locations across plant tissues, so no spatial information about metabolite
abundances could be obtained.

Imaging mass spectrometry offers great potential for high

throughput metabolomic analysis [83].

Therefore, we explored the application of imaging mass

spectrometry for localizing trichome metabolite profiles across plant tissues.

Herein, we

present a carbon material based printing method for high spatial resolution imaging mass
spectrometry profiling of tomato trichome metabolites without the need for tedious
microsampling.

This sensitive scheme is capable of profiling specialized metabolites of

thousands of individual trichomes level on a single tomato leaf surface.

The high spatial

resolution capability offers the potential to explore chemical heterogeneity across a population of
individual trichomes within a plant tissue.

To our knowledge, this is the first report of mass
96

spectrometry imaging on irregular shaped tissues.

3.2

Materials and Methods

3.2.1

Chemicals

Acetonitrile, 2-propanol, methanol, dichloromethane, formic acid (88% aqueous solution)
were purchased from VWR Scientific. α-Cyano-4-hydroxycinnamic acid (CHCA) and
2,5-dihydroxybenzoic acid (DHB) and tomatine were obtained from (Sigma-Aldrich, St. Louis,
MO, USA).

3.2.2

Plants

Seeds for plants were obtained from the Tomato Genetics Resource Center at the University
of California-Davis. Plants used for this study were grown in a greenhouse at Michigan State
University.

3.2.3

Sample Preparation

A dissecting glass microscope slide was etched using HF or mechanically ground with
rotating grinding by the Glass Shop of the Michigan State University Department of Chemistry.
97

A 2B pencil purchased from the Spartan Bookstore at Michigan State University was used to
draw a layer of graphite on one side of the glass slide. The coverage of surface by the graphite
was checked by examining the glass slide using transmitted light under a microscope.

The

glass slide was placed in a drying oven at 80˚C overnight, as removal of moisture strengthened
the interaction of graphite with the glass surface.

The slide was then immersed in Milli-Q water,

dichloromethane and methanol for 1 hr each, and dried under a gentle stream of nitrogen.

The

slide was placed on a block of dry ice to chill the surface as a prelude to transfer of trichomes.
Such cooling helps quench enzyme activity in the trichome and preserves the chemical
composition during sample processing.
A fresh leaflet from S. habrochaites LA1777 was cut by razor blade from the base of the
leaflet (petiolule).

Since the analysis time for an imaging experiment increases with the square

of the sample linear dimension, leaflets were chosen smaller than 3x3 cm to limit the time
needed for analysis.

The symmetric leaflet on the other side of the petiole which was not used

for imaging was collected as well, and used to estimate leaflet thickness for construction of a
parafilm mask.

A mask was constructed by either stretching or folding a sheet of parafilm until

the film thickness matched the thickness of the second leaflet.

A rectangular region was

removed from the middle of the parafilm mask, and the leaflet was positioned in the center of
this recessed area.

A graphite-coated glass slide was gently placed above the leaflet, and a 500

ml Pyrex beaker containing 200 ml water was placed on top of the slide to provide a
reproducible amount of pressure for the contact transfer of trichomes from tissue to slide.

Mass

calibrant reference solutions were spotted onto the corners of the slides for calibrating the TOF
98

mass axis before each analysis.

The glass slides, containing trichomes, were attached to the

MALDI sample stage using adhesive tape.

The procedures are illustrated in Scheme III-1.

Some conductive material (e.g. aluminum foil) was used to ensure electrical contact between the
coated slide surface and the metallic sample stage.

For imaging of single trichomes, the

trichomes were transferred onto the surface of a glassy carbon plate or a bare glass slide.

Use

of an uncoated glass slide aids visualization of trichomes under optical microscopes using
transmitted or reflected light or fluorescent imaging.

A target imaging area was marked with

liquid paper beforehand to provide position references, and reflected light optical images were
collected by a CCD camera mounted on the microscope.

Before laser desorption ionization

(LDI) imaging of trichomes on glass slides was performed, the slide was subjected to carbon
coating under vacuum at the Michigan State University Center for Advanced Microscopy.

99

Scheme III-1

3.2.4

Sample preparation procedure for trichome contact printing.

Laser Desorption Ionization Imaging

A MALDI-TOF mass spectrometer (AXIMA cfr-plus, Shimadzu, Columbia, MD) used
was equipped with a 337 nm N2 laser delivering 3 ns pulses.

In the MSI experiment, data

were acquired in either positive or negative ionization mode using the linear time of flight
detector.

In some cases, the same sample was analyzed twice using different polarity of ion

detection without further sample preparation.
laser shots per pixel.

The slide surface was first interrogated using 20

For whole leaflet imaging, the stage step size was set to 125 µm. For

single trichome analysis, the laser step size was set to 25 µm, and signals from 200 laser shots

100

per pixel were combined and stored for each pixel.

Ion signals between m/z 50 and 1500 were

collected. Metabolites were identified by first matching spectra with those obtained for leaf
dip extracts using liquid chromatography/mass spectrometry (LC/MS).

The LC/MS analyses

provided mass measurements with about 5 ppm accuracy. Annotation of metabolites was
performed through searching of molecular masses against an assortment of public metabolite
databases (ChemSpider, eMolecules, CSLS), and promising candidates were screened further
by interpreting masses from nonselective CID spectra.

Post-source decay (PSD) LDI mass

spectra were acquired for further confirmation of metabolite structures.

All LDI imaging data

acquisition and processing was performed using Launchpad software (version 2.8.1, Kratos
Analytical).

The raw mass spectrometry imaging (MSI) data were exported to Biomap

(Novartis) software for further processing.

3.2.5

LC/MS analysis

Leaflet metabolites were extracted by dipping the leaflets in 2 ml isopropanol: CH3CN:
H2O (3:3:2 v/v/v) for 1 min.

Extracts were centrifuged at 10000g for 10 min at 4 °C, and the

supernatants were analyzed without further processing.

Separations were performed on a

Supelco Bio-C18 column (1 × 150 mm, 3 µm particles) held at 30 °C that was interfaced to an
LCT Premier mass spectrometer (Waters, Milford, MA).

A solvent gradient was executed

based on 0.15% aqueous formic acid (A) and methanol (B) as follows: initial condition 5% B;
linear gradient to 50% B (5 min); 95% B (33 min); 100% B (35 min); hold at 100% B until 38
101

-1

min; return to 5% B (43 min). The flow rate was 0.1 ml·min , and the injection volume was 10
µL.

The LCT Premier mass spectrometer was operated using electrospray ionization (ESI) and

under the control of MassLynx version 4.1 software. The ESI conditions were as follows:
capillary voltage, 3200 V; cone voltage, 10 V; source temperature 90 ºC; desolvation gas flow,
300 L/h; desolvation gas temperature, 200 ºC; cone gas flow, 20 L/h. Detection was performed in
both positive and negative ion modes over m/z 50-1500 using centroided peak acquisition and
dynamic range enhancement.

Aperture 1 voltages were 10, 45, 75 V for three functions to

generate molecular ion and fragment ions at the same run.
the pseudomolecular ions by matching of retention time.

3.3

The fragment ions were aligned to
.

Results and Discussion

The diversity of epidermal cell types on leaves and stems of tomato and its wild relatives
presents challenges for research that aims to understand the mechanisms governing trichome
chemistry.

Complicating matters further are trichome fragility and the morphological

differences between trichome types.

In the wild tomato S. habrochaites (formerly known as

Lycopersicon hirsutum), examination of the leaf surface quickly reveals the long hair-like type I
and IV trichomes (Figure III-1).

Closer examination at higher magnification reveals the type VI

glands, which consist of bundles of four cells on a short stalk.

To obtain a clear view of the

type VI glands, one must adjust the microscope focus to a level below the tops of the longer type
I and IV glands, which are less obvious in the microscope image owing to the limited depth of
102

focus.

Probing of these different glands with optical or ion beams must account for this height

heterogeneity.

Furthermore, due to the variation of the stalk height of trichome, laser or ion

beams might be easy to reach taller trichome but not the short one because of the potential block
from the taller trichome.

And also ions ejected from shorter trichome can also be blocked by

adjacent taller trichome to prevent it for detection.
A recent review presented a critical evaluation of various sample preparation techniques
used for mass spectrometry imaging [188].

Much of the discussion centered on tissue fixation,

sectioning, and deposition of MALDI matrix materials as preludes to mass spectrometry imaging
analyses.

All of the matrix deposition approaches suffer from disadvantages of varying severity,

particularly with regard to even deposition of matrix across a tissue surface.

Furthermore, the

approaches that have been used for MALDI imaging have largely focused on detection of
proteins or lipids [189, 190] rather than soluble metabolites that more easily diffuse during
sample preparation.

Such concerns are particularly important for imaging of trichomes, which

are fragile structures that secrete viscous mixtures of metabolites, and are easily broken from
tissues during handling.
When imaging of trichomes is conducted using a laser for metabolite ionization and
volatilization, no single laser focal plane will be optimal for all trichomes on a tissue surface.
One promising approach to address this problem involves physical transfer of trichomes onto a
single horizontal plane, presented in the form of a substrate that absorbs the laser radiation.

A

similar strategy for “tissue printing” of cells from various plant and animal tissues onto polymer
surfaces was reviewed in 1994, and included microscopic, biochemical, and immunochemical
103

detection methods [191].

Even subcellular printing resolution has been achieved [192].

Tissue printing has also been successfully applied to live neuronal cells for investigation of
synaptic communication [193] and human surgical specimens for tumor marker profiling [194].
A few chemical printing methods have been reported for MALDI imaging including localization
of proteins, lipids, and agrochemicals in biological materials [195-200]. To my knowledge,
these analyses have all involved transfer of tissue sections to solid substrates.
fragility often makes sectioning of tissues impractical.

Trichome

Instead, the experiments described in

this Chapter employed contacting printing, and successfully and selectively transferred trichomes
onto a solid substrate, e.g. glass slide (Figure III-2).

The optical image reveals the type VI

trichome structure has been retained after the transfer.

Type I/IV trichomes lack external cell

walls, yet the transfer has retained the visible outline of the boundaries of the gland.

104

Figure III-2 Transmitted light optical image of trichomes from the adaxial (upper) leaf
surface of S. habrochaites LA1777 transferred to a glass microscope slide.

In the course of this study, some concerns were expressed that epidermal cells other than
trichomes might also be transferred during the contact printing process, and that printing force
might break trichomes and spread metabolites over a large area.

To achieve greater

reproducibility, a 500 ml glass beaker containing 200 ml of water was placed on top of the glass
2

slide to provide a consistent force (~ 500 N/m ) for the printing.

The optimal force was

determined by visual inspection of the transferred trichomes on the glass slide under the
dissecting microscope.

No evidence of transfer of type VII trichomes was found.

These

glands have much shorter stalks and are of lower abundance, and may be protected from transfer
105

by the taller type I, IV, and VI glands.
We used detection of cell-specific metabolite markers to determine whether the transferred
trichomes were contaminated by metabolites that leaked from epidermal pavement cells.

Based

on the LC/MS profiling of metabolites from pooled trichomes of a single gland type, the
glycoalkaloid dehydrotomatine was not found in type I, IV, or VI trichomes, but was present in
type VII trichome and was abundant in extracts of the entire leaf.

It is proposed that this

metabolite can serve as a marker of pavement epidermal cells and type VII glands.

In order to

make a more quantitative evaluation of the efficiency of transfer of trichomes from leaf surfaces
to the bare glass slide substrate, trichomes on the glass slide were extracted using 3:3:2 (v:v:v)
acetonitrile: isopropanol: water solvent with ultrasonication.

This extract was analyzed using

LC/MS, and the results were compared to analysis of extracts of the leaf tissue after trichome
printing.

Consistent with the result presented above, dehydrotomatine was not present at

detectable levels in the slide surface extract, but was abundant in the leaf dip extracts for three
replicate tissue samples (Figure III-3).

This finding suggests that transfer of pavement

epidermal cells to the glass slide was negligible.

106

A

B

Figure III-3 Negative ion mode extracted ion LC/MS chromatogram of m/z 1076.5 showing
dehydrotomatine for (A) metabolites washed off from the S. habrochaites LA1777 trichome
printed glass substrate analyzed by LC/MS. (B) Metabolites remaining on the S. habrochaites
LA1777 leaf surface after printing and flushing with extraction solvent and analyzed by
LC/MS.

The selection of graphitic carbon for the substrate surface was made based on its ease of
manipulation, efficient absorbance of laser light, inertness and ease of cleaning, and availability
in numerous forms including pencil lead, graphite paint, and graphite plates.

These materials

often contain impurities that contribute background peaks in LDI mass spectra.

To minimize

these, the pencil lead coated slide was rinsed with multiple organic solvents and water before
trichome transfer.

After substrate cleaning, background ions are fewer in number and lower in

abundance than in spectra obtained using the traditional MALDI matrix compounds
α-Cyano-4-hydroxycinnamic acid (CHCA) and 2, 5-dihydroxybenzoic acid (DHB), between m/z
100 to 1500 (Figure III-4).

For comparisons across these matrices, different laser powers were

employed for the various matrices to achieve approximately the same total ion current for each

107

matrix.

These laser power settings were; DHB = 105; CHCA = 85; pencil lead = 90.

These

laser power values are in arbitrary units, with the full range of 0-180 corresponding to laser
output of 0-6 mW.

When spectra were generated using negative ion mode, the pencil lead

background showed carbon cluster peaks with m/z= 12n (n = an integer; up to m/z 180).

In our

study, most specialized metabolites in tomato trichomes have molecular masses greater than 200
Da, so the carbon cluster signals did not interfere with peak detection.

The low mass carbon

cluster peaks in negative mode can also be used to calibrate the instrument mass axis at low
masses.

108

A

B

C

Figure III-4 The background MALDI and LDI mass spectra of DHB (A), CHCA (B), Pencil
lead (C) over m/z 50~1500 in the positive mode. The spectra were scaled to the same absolute
y-axis scale.

109

Pencil lead is manufactured in different grades that vary in the ratio of graphite to clay
binders.

One may anticipate that composition will influence the dynamics of the laser

desorption ionization process.

To investigate the effect of pencil lead composition, the

analytical performance was evaluated using several different pencil types (H, 2H, 4H, B, 2B, 3B,
4B, 6B, and 8B).

Aliquots (~ 1 µl) of trichome extracts from S. habrochaites LA1777 were

applied to pencil lead coated slides for comparison.

With increasing graphite content from H to

8B, the laser power required to achieve a base peak of 50 mV decreased.

Comparisons of

signal-noise-ratio and mass resolution under laser power yielding the same base peak signal
abundance were made, using several different acyl sugar metabolites (in positive mode) and
sesquiterpene acids (m/z 233 in negative mode) were evaluated.

In a typical experiment,

spectra generated from 50 laser shots were summed over multiple locations on the target, and the
mass resolution and signal-to-noise ratio were calculated by the instrument software.

Pencil

leads B, 2B and 3B (low graphite content) gave the best mass resolution among the tested pencil
leads when laser power was adjusted to yield approximately the same ion abundances.
pencil lead yielded the greatest signal-to-noise ratios (Table III-1).

2B

As a result of these findings,

pencil lead 2B (74% graphite; http://www.blurtit.com/q9883867.html) was chosen for all
subsequent analyses.

110

Table III-1 Mass spectra resolution (±SD, n=3) and S/N ratio (±SD, n=3) calculated by
averaging the resolution and S/N ratio of five most abundant acylsugar peaks from S.
habrochaites LA1777 extracts in positive mode and sesquiterpene acids (m/z 233) peak in
negative mode. Resolution and S/N for individual peak is taken from peak lists generated using
Launchpad software (version 2.8.1, Kratos Analytical).
Pencil types
H
2H
4H
B
2B
3B
4B
6B
8B

Resolution (FWHM)
(positive mode/negative mode)
982±131/1475±287
531±273/1943±213
1223±253/2912±432
2038±445/2653±372
2164±387/2830±311
2295±541/3013±306
1338±379/2173±448
1607±314/2150±337
1635±445/2089±356

S/N ratio
(positive mode/negative)
1593±576/2305±384
2207±711/2645±469
2010±785/2216±453
2900±862/2357±432
2875±693/4103±326
2268±782/2177±333
2362±847/2983±355
2126±1020/2468±594
1590±899/2298±513

The observed peaks in LDI mass spectra of trichome extracts spotted onto the pencil
lead-coated slide were first compared to spectra obtained using LC/MS analysis of the same
extracts.

Putative annotation of metabolites was performed by the following: accurate LC-TOF

MS measurements of pseudomolecular ions generated a list of candidate molecular formulas,
which were searched against several chemical and natural products databases (ChemSpider,
eMolecules, CSLS).

Masses of fragment ions had also been generated using multiplexed CID

on the LC-TOF spectrometer or as product ion spectra on the Q-Tof spectrometer.

Putative

structures were proposed based on the goodness of matching of predicted fragment ion masses
with experimental MS/MS spectra and also prior knowledge of plant metabolites in the genus
Solanum [117].

Ions observed in LDI mass spectra were assigned by comparing their m/z

values with ions observed in LC/MS analysis.

111

In some cases, product ion MS/MS spectra were generated using post-source decay (PSD)
to confirm assignment to structural classes (Figure III-5C and 5D).

In the former case, the

prominent ion at m/z 715 generated in positive ion mode for an extract of S. habrochaites
LA1777 trichomes was selected for transmission by the ion gate of the Axima TOF mass
spectrometer, and ions reaching the reflectron detector appear in the spectrum.

Based on

+

comparisons to the LC-TOF MS data, this was tentatively annotated as [M+Na] of acylsugar
S4:21.

In the PSD spectrum (Figure III-5C), in addition to the precursor ion (m/z 715) that has

not fragmented, an additional product ion is evident at m/z 399, corresponding to a loss of a
neutral fragment of 316 Da.

This ion corresponds to the sodiated triacylglucose fragment that

contains C2, C4, and C5 ester groups.

Complementary fragment information was obtained

from the LC-TOF CID spectrum (Figure III-5E) which shows a major fragment at m/z 317,
corresponding to the fructose moiety with a C10 ester group attached.

The combined results

point to S4:21 being a sucrose with C2, C4, and C5 groups on the glucose ring, and C10 on the
fructose ring. The differences in information arise from different precursor ions.

For the

LC-TOF MS analysis, the CID process is non-selective, so all ions are activated by collision,
+

with most fragment ions being generated from [M+NH4] ions (m/z 710), whereas the protons in
the sodium adduct have limited mobility.

Its fragmentation is inefficient under the LDI

conditions, and is largely confined to formation of the sodiated triacylglucose ion.

One

important finding from this experiment is the localization of the long chain fatty acid on the
fructose ring, which has not been reported in published descriptions of Solanum acylsugars.

112

A

B

Figure III-5 Laser desorption ionization (LDI) mass spectra of trichome extracts from S.
habrochaites LA1777 spotted on pencil lead-coated glass slides using (A) positive mode (B)
negative mode. Post source decay (PSD) product ion spectra of (C) acylsugar S4:21
+
precursor=m/z 715 ([M+Na] ) (D) mixed isomeric flavonoid glycosides rutin and
kaempferol-O-diglucoside ([M-H] , m/z 609). Electrospray ionization in-source decay (ISD)
mass spectra of (E) acylsugar S4:21 and (F) mixed flavonoid glycosides rutin and
kaempferol-O-diglucoside [M-H] acquired using TOF-MS and nonselective collision induced
dissociation.

113

Figure III-5 (cont’d)

C

[M-H]

-

D

114

Figure III-5 (cont’d)

E

N H 4+
OH
OH
O
O
O
O O OH
O OH
O O
O
O
O

OH
O
O
O
O O
O
O
m /z 3 5 9
HO

O OH
O
O OH

m /z 7 1 0
m /z 3 17

F

[M-H]

115

-

PSD mass spectra were generated in negative ion mode for the same S. habrochaites
LA1777 trichome extract.

The product ion (PSD) spectrum of m/z 609 showed two fragments

at m/z 284 and 300 (Figure III-5D).

These correspond to radical anion fragments for the

flavonoid aglycones kaempferol and quercetin, respectively, and demonstrate resolution of two
isomeric flavonoid glycosides in the trichome extract.

This finding stands in contrast with the

findings reported in Chapter II, in which all of the m/z 609 flavonoid glycosides were
kaempferol derivatives.

We attribute this difference to separate samplings of plants grown

under different conditions (growth chamber vs. greenhouse).
One of the challenges that analysts face when performing MALDI imaging lies in the
choice of a suitable matrix that can convert laser light into liftoff and ionization.

It is common

dogma that acidic matrices such as CHCA or DHB are effective for generating positive ions
because they can donate protons to analytes, whereas 9-aminoacridine is preferred for negative
mode because it is a proton scavenger.

As a consequence, a single matrix may perform poorly

when it is desirable, as is usually the case, to analyze materials in both positive and negative ion
modes.

Binary mixtures of matrices mixture have been used to generate information about

metabolites in both positive and negative ion modes [201].
complicates further the matrix application process.

However, such an approach

By using a solid support such as carbon as

the light-absorbing medium, LDI generates ions for a wide range of trichome metabolites in both
positive and negative modes.

An important consequence of this finding is that the laser can be

rastered over the printed substrate to generate spectra for one polarity over the entire area,
followed by a repeat analysis with the polarity inverted.
116

The key issue that needed to be

addressed was the depletion of analyte with successive laser shots for a single pixel.

Based on

some preliminary data regarding signal depletion, experiments were performed using 20 laser
shots per pixel in each mode.
Four classes of major metabolites were detected in both LC/MS of individual trichome
types from S. habrochaites LA1777 and LDI analyses of trichome extracts and from tissue
printing with imaging: acylsugars, flavonoids, alkaloids, and sesquiterpenoid acids (Table III-2).
-

In LDI analyses, flavonoid and sesquiterpenoid acids could be readily detected as [M-H] ions in
negative mode, as they are weak acids.

Acylsugar metabolites are not particularly basic or

acidic, but since sodium and potassium are abundant ions in plant tissues, acylsugars are
+

+

observed in positive ion mode mainly as [M+Na] and [M+K] . Pseudomolecular ions of
acylsugars were not observed in negative ion mode, and this finding highlights the need to
perform analyses in both modes.

In the lower m/z region of the negative ion spectra, a series of

ions matching the masses of short fatty acid anions (e.g. m/z 87 for C4, m/z 101 for C5) was
-

observed. Since acylsugars do not readily form [M-H] ions and since the short chain fatty
acids were not observed in LC/MS analyses, it is suspected that displacement or dissociation
reactions in the plume are responsible for generation of the fatty acid anions from acylsugar
metabolites.
tomatine.

One surprising finding arose from the observed behavior of the glycoalkaloid

Tomatine has one basic nitrogen group, and was expected to be detected as [M+H]

in trichome extracts and printed glands.

However, in spectra of trichome extracts, peaks at m/z

1056 and 1072 were observed, but signals corresponding to [M+H]
(Figure III-5A).

+

Two lines of reasoning may explain this phenomenon.
117

+

were not discernable
First, both sodium and

potassium are abundant in plant tissues, and the large carbohydrate moiety is expected to have
high affinity for these alkali metal cations.

In addition, the high internal energies anticipated
+

from LDI from carbon surfaces may lead to more extensive fragmentation of [M+H] owing to
proton mobility.

118

Table III-2 Major ions observed in laser desorption ionization (LDI) spectra of S. habrochaites
LA1777 trichome extracts on pencil lead-coated glass slides.
Polarity

Major Ion
+

+

617/633

+

+

631/647

+

+

645/661

+

+

659/675

+

+

673/689

+

+

687/703

+

+

701/717

+

+

715/731

+

+

729/745

+

+

743/759

+

+

757/773

+

+

771/787

+

+

1056/1072

S 4:14

Positive

[M+Na] /[M+K]

S 4:15

Positive

[M+Na] /[M+K]

S 4:16

Positive

[M+Na] /[M+K]

S 4:17

Positive

[M+Na] /[M+K]

S 4:18

Positive

[M+Na] /[M+K]

S 4:19

Positive

[M+Na] /[M+K]

S 4:20

Positive

[M+Na] /[M+K]

S 4:21

Positive

[M+Na] /[M+K]

S 4:22

Positive

[M+Na] /[M+K]

S 4:23

Positive

[M+Na] /[M+K]

S 4:24

Positive

[M+Na] /[M+K]

S 4:25

Positive

[M+Na] /[M+K]

tomatine

Positive

[M+Na] [M+K]

Sesquiterpene acid

Negative

[M-H]

Rutin

Negative

[M-H]

Kaempferol-O-diglucoside

Negative

[M-H]

Quercetin-O-diglucoside

Negative

[M-H]

119

m/z

-

233

-

609

-

609

-

625

Imaging of printed trichomes from the upper surface of a leaflet from S. habrochaites
LA1777 reveal distribution of some major metabolites throughout the leaf. Again, data were
generated in both positive and negative ion modes for the same tissue.

Data for Figure III-6A

were generated using positive mode with 20 laser shots per pixel, and then the polarity was
switched for negative mode acquisition, also with 20 laser shots per pixel.

Figure III-6 shows

the color-coded intensity map of pseudomolecular ion for one acylsugar (Figure III-6A), the
aglycone fragment of a group of flavonoid glycosides (Figure III-6B), and a deprotonated
sesquiterpene acid (Figure III-6C).

For all three m/z images, the midvein area running up and

down the center of the leaflet generates stronger signal than the rest of the tissue area.

Visual

inspection of the leaflet before printing showed higher trichome density around the midvein.
Whether the strong signal intensities near the midvein largely reflect trichome densities or
differences in composition of individual trichomes has yet to be established.

It is not practical

to sample individual trichomes around the midvein because the trichome density is so high as to
preclude microsampling of individual glands.

The images presented in Figure III-6 present a

wide view of metabolite abundances across an entire leaflet.

Based on findings reported in

Chapter II, acylsugars predominantly accumulate in type I and IV glands, whereas flavonoid
glycosides and sesquiterpene acids are abundant in type VI glands.

The wide view does not

distinguish contributions of individual trichomes to plant tissue chemistry, and highlight the need
to probe individual gland chemistry.

Such efforts are described below.

120

A

C

B

Figure III-6 LDI images of selected ions for contact printed trichomes from S. habrochaites
LA1777 leaf on pencil lead-coated glass slides: (A) ion abundances of m/z 731 (acylsugar S4:21
+
[M+K] ) using positive mode and laser power setting = 95, 125 µm stage step, and 20 laser
shot per pixel. (B) m/z 301 in negative ion mode (aglycone fragment of quercetin glycosides),
(C) m/z 233 (sesquiterpene acid [M-H] of various sesquiterpene acids) for the same leaf sample
under negative mode with laser power setting of 105, and 20 laser shots per pixel. The key for
color coding is presented at the left of each panel.

To assess how much metabolite was removed by laser ablation, 30 pixels selected to be
representative of different regions of the leaflet were chosen.

Initial data acquisition employed

laser power = 95, which is the same condition used for imaging.
single profile using 5 shots per pixel.

Then 20 shots were fired on each spot to assess the

ablation caused by typical imaging conditions.
using another 5 shots at the same spot.

Spectra were acquired into a

Afterwards, the spectrum profile was generated

This second sampling yielded 87±8% (N=30) signal

intensity compared to the first 5-shot profile.

The results suggest that the metabolites were

slightly depleted by normal spectrum generation parameters.

Changes in relative abundances of

various metabolite ions were judged to be minimal following the 20-shot ablation.

121

For individual trichome analysis, the spatial resolution of imaging should resolve objects
smaller than 50 µm (the approximate diameter of the type VI trichome and about the average
distance between two trichomes).

For the Axima cfr-plus instrument, details regarding the

laser spot size were unknown when these experiments were conducted, but the spot diameter
was estimated to be approximately 100 µm.

The spatial resolution of the LDI imaging scheme

was first tested by printing trichomes into transmission electron microscope (TEM) grids with
different spacing between grid elements.
pencil scratched substrate.

The TEM grids were mounted on the top of the

Plant leaves were pressed against it several times and trichomes

were positioned into the space between the grids.

LDI imaging was conducted after the grid

was removed from the substrate because of concerns that the thickness of the TEM grid might
interfere with illumination of the trichomes by the laser and with ion transmission from the
surface to the mass analyzer. The diameter of the laser spot is estimated to be about 100 µm,
which is larger than the smallest grid we tested (50 µm).

To analyze the object which is

smaller than the size of the laser spot, complete depletion [202] with a step size of 25 µm and
laser power of 105 was employed.

The resulting 100 µm grid can be clearly resolved, but the

smaller spacing of the 50 µm grid hindered trichome transfer to the substrate surface.
Nevertheless, the resulting image suggests that spatial resolution approaching 50 µm has been
achieved (Figure III-7).

122

A

B

Figure III-7 Imaging of metabolites from S. habrochaites LA1777 leaf trichomes transferred
into the (A) 100 µm and (B) 50 µm TEM grids. Ion abundances are displayed for m/z 731
+
(acylsugar S4:21 [M+K] ), using positive mode with laser power setting = 95, 25 µm stage step
size, and 200 laser shot per pixel.

Pencil lead coating of substrate surfaces is a fast and simple sample preparation method for
chemical imaging of leaf metabolites over a wide tissue area.

However, when higher spatial

resolution is needed for individual trichome imaging, small defects in the carbon coverage can
compromise the analysis.

A closer examination of the pencil lead-coated slide under the optical

transmission microscope showed numerous spots of diameter less than 30 µm where no pencil
lead had been deposited.

For comparison, we evaluated deposition of the MALDI matrix

CHCA using electrospray deposition.

Again, the coating lacked uniform coverage unless a very

thick layer (10-100 µm) of matrix was deposited.

To minimize the effects of surface coating

defects, several other substrates were explored for individual trichome imaging.
commercially available graphite plate (SPI supplies) was tested as a substrate.

123

First, a

The porous

structure of the graphite plate gave high surface area which offers potential enhancement of the
laser desorption/ionization efficiency [203].
for pencil lead coated glass slides.

However, in our test, the ion signal was less than

The extracts from S. habrochaites LA1777 yielded 50 mV

signal for acylsugar ion m/z 731 (S4:21) using 90 laser power and 2B pencil lead coated glass
slide.

For the same extracts, a laser power setting of 135 was required to obtain the same signal

using a graphite plate.

As a second alternative, a thin layer of carbon paint (Ted Pella, Inc.)

used for electron microscopy was painted on the back side of a stainless steel MALDI plate.
The ingredients present in the carbon paint yielded more background ion signals at low mass
range than rinsed pencil lead.

To remedy this, the carbon paint coating was heated with a gas

flame for 5-10 min to remove the impurities.

Background was reduced after the heating except

for one major background peak that has the same mass (m/z 233) in the negative mode as
sesquiterpene acid metabolites present in trichomes.
pursued further.

For this reason, this approach was not

Third, glassy carbon plate widely used for electrochemistry gave minimal

background, and was judged to be a promising alternative.

Glassy carbon was recently found to

have structural features resembling fullerenes [204] and has been used for small molecule
analysis by laser desorption ionization [205].

Fullerenes have also proven to be a successful

matrix for laser desorption ionization [206, 207].
The surface of glassy carbon can be chemically etched or mechanically ground to increase
the surface area, which should increase the ionization yield because the metabolites would be
spread in a thinner film within the illuminated area.

However, for transfer printing, the rough

surface may decrease spatial resolution of LDI analyses because spreading of metabolites would
124

be more favorable. The choice of a solid laser-absorbing material is best considered as a trade
off between ionization efficiency and spatial resolution. To illustrate this point, consider that
each type VI gland contains about 100 pL of cell sap. The gland head is fragile and the sap can
readily leak from the trichome upon application of mechanical force.

For this reason, to

achieve higher spatial resolution desired for single gland imaging, the glassy carbon surfaces
described below were not subjected to any etching procedures.
Figure III-8 shows LDI images obtained in both positive and negative modes of trichomes
transferred from a small (5 x 10 mm) leaflet of S. habrochaites LA1777 to a glassy carbon
surface.

The images document metabolite composition with spatial resolution of individual

trichomes.

Further examination of the mass spectra generated at different locations across the

leaflet suggested a heterogeneous distribution of metabolites.

In the positive mode spectra, an

example spectrum selected from a bright spot on the m/z 731 image near the midvein shows a
series of ions separated by 14 Da that correspond to potassium cationized acylsugars.

The ion

at m/z 731 corresponds to acylsucrose S4:21, and its abundance is suggestive of locations of type
I and IV trichomes, which are responsible for acylsugar accumulation, as discussed in Chapter II.
A different pixel farther away from the midvein also displays a strong m/z 731 signal, but shows
abundance of S4:21 ions greater than those for other acylsugar homologs.

Since the fatty acyl

groups are believed to be derived from various amino acids, this finding suggests that metabolic
conversion of amino acids to specialized metabolites has a significant dependence on trichome
location within leaf tissue.

Similar heterogeneity was observed in the negative ion image,

which revealed a dominance of flavonoids, detected as m/z 301, the quercetin aglycone anion,
125

near the midvein, but greater prominence of sesquiterpene acids (m/z 233) in regions closer to
the edge of the leaflet.

Both of these metabolites are selectively accumulated in type VI

trichomes [140], so this finding reflects biochemical differences, in a specific trichome type, at
various locations across the tissue.

Whether this reflects different access of trichomes to

metabolite precursors in phloem near the midvein remains to be determined, and further
investigation into the factors that control accumulation of metabolites in individual trichomes is
recommended.

126

A

Figure III-8 Left: Mass spectrometry images of selected ions for S. habrochaites LA1777
trichomes transferred to glassy carbon. (A) m/z 731 in positive mode, corresponding to
+
acylsugar S4:21 ([M+K] ), using laser power setting of 105, 125 µm stage step, and 20 laser
shots per pixel. (B) m/z 301 in negative mode (fragment ion of quercetin glycosides) for the
same leaf sample using laser power setting of 115, 125 µm step size, and 20 laser shot per pixel.
Right: (A) mass spectra for two individual pixels in positive mode (B) mass spectra for two
pixels in negative mode.

127

Figure III-8 (cont’d)

B

The differences in metabolic phenotypes between individual type VI trichomes is further
illustrated in Figure III-9, which shows a reflected light microscope image of the transferred
trichomes on a glassy carbon surface (Figure III-9A) and the corresponding LDI image of m/z
233 in negative mode.

To address the challenges of aligning optical and mass spectrometry

images, four position indicators were applied to the substrate using liquid paper, which is easily
visualized using the camera in the MALDI instrument. These reference points lie outside the
area shown in the images in Figure III-9.

These images demonstrate that metabolite signals are
128

obtained with spatial resolution of individual trichomes.

To further confirm the physical

integrity of trichomes after the transfer printing process, a fluorescent microscope image was
obtained (Figure III-9C), which shows fluorescence is confined to the interior of the type VI
trichomes.

To obtain this image, it was not feasible to use a carbon substrate.

However, a thin

film of carbon was deposited afterward using vacuum evaporation and deposition, in the same
manner

as

is

common

in

sample

preparation

for

scanning

electron

(http://www.huck.psu.edu/facilities/electron-microscopy-up/faq/sample-preparation),
employs carbon thickness on the order of 10-50 nm.

microscopy
which

The LDI image (Figure III-9D) was

obtained by transferring the carbon coated slide into the MALDI instrument.

To obtain higher

spatial resolution, complete ablation of each pixel was performed [202] using a laser power
setting of 145.

Due to the small stage step size and the relatively slow laser repetition rate (10

Hz), the imaging of an entire leaflet at single gland resolution would take about 15 days to
complete, so the imaging was only conducted over a small area.

As was the case with the

glassy carbon substrate, substantial differences in the sesquiterpene acid signal at m/z 233 were
observed for trichomes adjacent to one another on the leaf surface.

This striking heterogeneity

of trichome chemistry provides a compelling reason for performing chemical imaging of
trichomes.

If all trichomes on a leaf surface were chemically identical, the rationale for

imaging them would fade.

129

A

B

C

D

Figure III-9 Images of S. habrochaites LA1777 trichomes printed on (A) glassy carbon
surface, reflected light image (B) m/z 233 ion (sesquiterpene acids) in negative mode on the
same glassy carbon area. Step size was 25 µm with 200 laser shots per pixel. (C) fluorescent
image of S. habrochaites LA1777 trichomes printed on a glass slide using 543 nm excitation
and 560 nm longpass emission. Slide was subjected to carbon coating afterward; (D) mass
spectrometry image of the same glass slide as (C) after carbon coating, showing ion abundance
map for m/z 233 in negative mode using 200 laser shots per pixel with 25 µm step size.

130

Despite the growing number of published reports using mass spectrometry imaging for
protein and metabolite localization [208-211], validation of the performance of mass
spectrometry imaging has been scarce.

Several issues are relevant to validation of MS imaging,

but most center around the effect of the cellular matrix on energy deposition and ionization
efficiency.

To explore matrix effects, a set of experiments examined LDI using the glassy

carbon substrate using a mixtures of trichome extracts from two wild tomato species, S. pennellii
LA0716 and S. habrochaites LA1777.

The former was employed as a cellular matrix, and its

amount was maintained in excess relative to the latter, and in constant quantity. The acylsugar
profiles of these plants are easily distinguished by their molecular masses, as LA0716
accumulates acylglucoses, and LA1777 accumulates acylsucroses that are nearly 200 Da heavier.
Varying amounts of LA1777 extract were diluted into the LA0716 extract, with the proportions
normalized to the LC/MS extracted ion chromatogram peak areas for the most abundant
acylsugar from each plant, and these were assumed to have equivalent response factors.
Mixtures were prepared over the range of 0.01 to 1.0 for LA1777:LA0716, and an aliquot was
deposited into 2 mm diameter wells cut into a parafilm mask using the tip of a Pasteur pipet.
The amount of material deposited into each well is estimated to correspond to approximately 100
trichome equivalents per well.

LDI mass spectra were generated from 200 laser shots by

manually moving the stage to sample across the well.

The signal intensity for LA1777

acylsugar S4:21 at m/z 731 was determined for triplicate wells for each mixture, and the
relationship between signal and LA1777 concentration is presented in Figure III-10.

The

response approaches linearity over the 100-fold concentration range, suggesting that quantitative
131

comparisons in MS imaging should be reliable when experimental conditions are designed and
executed properly.

A

LA0716

LA1777

Figure III-10 (A) The LDI spectra of a mixture of S. habrochaites LA1777 and S. pennellii
LA0716 extracts using positive linear mode. Using a laser power=145, spectra from 200 laser
shots were averaged while rastering the laser across the spot. (B) Calibration curve showing ion
+
abundance of m/z 731 (S4:21, [M+K] ) relative to the least diluted LA1777 extract as a
function of 1/dilution factor of LA1777 extract (n=3, ±SE). A constant amount of LA0716
extract was added to provide a matrix in which the LA1777 metabolites were diluted.

132

Figure III-10 (cont’d)

B

Ion abundance of m/z 731
relative to the least diluted
LA1777 extract (%)

100
80
60
40
20
0

0
20
40
60
80
100
Concentration of LA1777 extracts (after dilution) relative
to the least diluted LA1777 extracts (%)

133

Another validation experiment involved comparing the LDI imaging signal of a metabolite
per unit area around midvein to the same signal near the leaf edge.

Two leaf samples of similar

size from the same S. habrochaites LA1777 plant were collected.

Glass Pasteur pipettes were

used to sample a core in the leaf tissue from the midvein area and leaf edge area, with the
removed tissue being contained within the pipette tip.
spots near the leaf edge were sampled for each leaf.

Four spots near the midvein and four
After removing the plant tissues, pipette

tips were broken off, and dropped into extraction solvent (acetonitrile:water:2-propanol 3:3:2) to
extract metabolites using ultrasonication for each spot separately.

One set of extracts was

analyzed using LC/MS, and the peak areas were integrated for the extracted ion chromatogram
for m/z 233.

This signal corresponds to multiple isomers of sesquiterpene acids, which are type

VI trichome specific metabolites.

Choice of this signal avoids contributions from cells other

than trichomes which are removed during the coring process.

Corresponding cores from the

second leaf were analyzed using LDI, employing the same 2 mm diameter wells described above.
The spectra for each spot were averaged for 200 laser shots per well, and the signal for m/z 233
was tabulated.

A comparison of the ratio of m/z 233 signal near the midvein relative to the

signal near the leaf edge using the two analytical methods is presented in Figure III-11.

Both

methods confirm the greater abundance of sesquiterpene acids per unit tissue area near the
midvein, and this finding supports the observation in the LDI imaging results (Figure III-6C).

134

Folds of difference between the
midvein and leaf edge

A

2.5

LC/MS

LDI

2
1.5
1
0.5
0
B

1

Figure III-11 (A) The schematic diagram showing the hole drilling experiment on the leaf
surface for validation of LDI metabolite profiling using LC/MS. Four areas were collected
around the leaf midvein and four areas were collected on the leaf edge. (B) The ratio of the
sesquiterpene acid detected in the midvein (m/z 233) cut to the leaf edge was measured using
two different analytical method; LC/MS and LDI. The experiment was conducted on two
different leaves.

135

In some cases, optical imaging of printed trichomes showed liquid in areas outside the
boundaries of individual trichomes. One question that needs to be answered is whether
metabolites leaked from trichomes during the tissue printing process, or whether laser irradiation
of trichomes leads to ejection of metabolites following rupturing of the trichome cell wall.

In

either case, the integrity of metabolite localization might be compromised. For individual
trichome imaging, pre-checking of liquid leaking using a dissecting microscope can help select
areas where leakage is minimal.

After laser ablation, examination of the residual material on

the substrate showed only tiny amounts of residue outside the trichome, but the gland structure
appears intact (Figure III-12).

A
B

B

Figure III-12 The reflected light optical image of transferred S. habrochaites LA1777
trichomes on glassy carbon surface (A) taken right after the transfer. (B) after laser ablation
with 200 laser shots per pixel and laser power =145.

136

One additional question to be addressed is whether LDI imaging can be performed on thick
biological tissues.

Typical dissected tissue in imaging experiment has been sectioned to around

10 µm thickness, or approximately the dimension of a eukaryotic cell.

There are some previous

reports about mass spectrometric imaging where tissue sections have been placed directly onto
the substrate [152].

One attempt to extend the application of this experimental scheme involves

determining whether a thin barrier between substrate and metabolite would still allow desorption
and ionization of the metabolite to occur.

To test this, a film of parafilm was stretched across

the substrate to about 10 µm thickness which is close to the thickness of a typical tissue section
for MALDI imaging.

Another layer of parafilm cut with a “Spartan” logo was placed on the

top the first parafilm and pressed against it to prevent leakage between the two layers of parafilm.
A methanolic solution (100 µM) of the glycoalkaloid tomatine, which lacks a chromophore at the
laser wavelength of 337 nm was pipetted into the recessed logo area, and LDI imaging was
performed over the entire area.

The imaging result is presented in Figure III-13.

The results

demonstrate that tomatine ions, predominantly the protonated aglycone fragment at m/z 416 can
be ejected from the surface of the thin film even when a film provides a barrier to material
transmission.

No evidence was found to suggest that tomatine had leaked through to the

substrate surface because no tomatine signals were detected after firing the laser at the glassy
carbon substrate after the parafilm was removed.

Control experiments were performed by

placing another parafilm and tomatine film on a stainless steel MALDI target plate.

No

tomatine metabolite signals were detected, even using the highest laser power setting available
for this instrument.
137

A

B

Figure III-13 Imaging of “Spartan” logo constructed by tomatine deposited on the 10 µm of
parafilm attached on the top of glassy carbon. (A) Optical image of the Spartan logo cut with a
razor blade. (B) intensity map of tomatidine fragment ion (m/z 416) from tomatine. Spectra
were acquired using linear positive mode using laser power of 155, and 20 laser shots per pixel.

3.4

Conclusions

The carbon based substrate coupled with contact printing sample preparation successfully
addresses the challenges for imaging trichomes using mass spectrometry.

For trichome imaging,

the traditional imaging scheme that involves mounting tissue on a substrate, and coating with
matrix, is not possible owing to the fragile trichomes and their incompatibility with fixation.
The contact printing scheme possesses several layers of merit:

1) sophisticated sample

preparation steps are not needed, 2) metabolite analysis does not suffer from interference from
matrix or other additives, 3) the technique allows selective imaging of trichomes without

138

interference from other epidermal cell types.

The approach provides spatial resolution of

individual trichomes across much of the leaf surface for plants in the genus Solanum.

The

simplicity of sample preparation makes such experiments accessible to a broader range of
researchers with limited technical expertise in sample preparation.
This printed transfer scheme offers other novel applications.

By using the printed transfer

method, two slides can be coated, and trichomes from both leaf surfaces can be transferred to
separate substrate surfaces in a single step.

This enables more complete analysis of a single leaf

or other plant tissues. Another potential application is for fingerprint imaging.

Fingerprint

imaging using mass spectrometry has drawn attention recently in the areas of forensics or
security [212].

The inexpensive carbon based substrate can be readily used to collect

fingerprint for imaging mass spectrometry purposes.

Last but not the least, this carbon based

substrate imaging scheme is useful for remote sampling.

Since the carbon substrate can be

easily frozen, stored and shipped, it can be used to sample fresh sample in remote area and
transfer spatially preserved chemical information to the laboratory for analysis.

In summary,

our work first time extended imaging mass spectrometry to specialized epidermal cells without
using dissection or tissue fixation.

The printed transfer method coupled with carbon based

substrate provides a complementary sample preparation method for imaging mass spectrometry
and provides potential novel applications.

139

CHAPTER FOUR
CHARACTERIZATION OF METHYLATED FLAVONOID REGIOISOMERS USING
LC/MS/MS COUPLED WITH SEMISYNTHETIC ENZYMATIC METHYLATION

4.1

Introduction

Plants synthesize a diverse group of specialized metabolites that contribute to their survival
and reproduction, with different plant species accumulating distinct sets of compounds [213].
Plants evolve to synthesize new compounds through the effects of mutations on functions of
biosynthetic enzymes, and subtle mutations can modify enzyme functions in ways that are
challenging to predict from DNA sequences alone.

For this reason, there is renewed interest in

determining functions of biosynthetic enzymes, and these efforts ultimately require a better
understanding of plant chemistry.
One family of specialized metabolites that occurs widely across the plant kingdom is the
flavonoids, of which more than 4000 different metabolites within this class are known.

The

diversity of flavonoid structures provides for a variety of physical, chemical, and biological
properties [214] (Figure IV-1).

Flavonoid metabolites such as anthocyanins are responsible for

coloration in flowers, fruits, and other tissues, and have been proposed to confer protection from
the damaging effects of ultraviolet radiation.

In addition, flavonoids also play roles as

regulators of transport of the plant hormone auxin [215], serve as signaling molecules in
plant-symbiont interactions [216], control pollen tube growth and functionality which are
140

important in plant reproduction [217], and are precursors of antibiotic phytoalexins that
strengthen plant defenses against pathogens [218, 219].

141

Figure IV-1 Basic structures of the main classes of flavonoids. Common O- and C-substitution positions are indicated with an
arrow. Adapted from Reference [220].

142

Flavonoids have several hydroxyl groups that can be modified by reactions including
methylation, acetylation, glycosylation, and sulfonation.

Modification of the basic flavonoid

structure serves an important role in determining where the different modified forms are found
within the cell.

For example, glycosylated flavonoids are water soluble and mostly present in

the cytosol. O-methylation of flavonoids confers greater lipophilicity [221] and often precludes
glycosylation. As a consequence of their physical properties, methylated forms of flavonoids
are often secreted on leaf surfaces where they protect plant tissues from UV-B damage [222].
In addition, O-methylation enhances the antimicrobial activity of the flavonoid by increasing the
lipophilicity and decreasing the reactivity of the phenolic hydroxyl groups toward oxidation, as
flavonoid oxidation can generate intermediates that trigger reactions involved in mutagenicity
[214].
Plants convert flavonoids to their methyl ethers through catalysis by O-methyltransferase
enzymes, which use S-adenosylmethionine (SAM) as the methyl donor.

Most plant genomes

contain numerous methyltranferase genes, and they exhibit diverse substrate specificity.

Their

functions include modification of phenolic acids in the synthesis of lignin precursors,
methylation of DNA and RNA, and methylation of numerous metabolites including the plant
hormones salicylic acid and jasmonic acid. Although the methyl transfer reaction mechanism is
the same for all phenolic hydroxyl groups, O-methyltransferase enzymes (OMTs) demonstrate
significant substrate and stereochemical selectivity [223].

The study of OMT specificity and

selectivity can help researchers better understand the functionality of methylation in plants.
Besides that, regioselective OMTs have potential use for production of regiospecific methylated
143

compounds that are unavailable in nature, including precursors of novel drugs when chemical
synthesis is not feasible [224-226].
The study of O-methyltransferase specificity requires first elucidating their products’
structures. Although there are many reviews about flavonoid structure analysis [220, 227-232],
techniques that can identify the position of methylation on flavonoids have been slow to develop.
So far, most methylated flavonoid analysis relies on isolation of milligram quantities followed by
NMR characterization [233-236]. In planta, methylated flavonoids usually exist at low levels
and as mixtures of isomers, often making isolation from plant tissue impractical.

A few reports

have described use of chromatographic and spectroscopic techniques to characterize methyl
flavonoids [237-240]. However, spectroscopic methods other than NMR are largely useful for
pinpointing methylation at a specific position, e.g., using UV to pinpoint whether 5 position is
methylated or not [241], but are inadequate when multiple methylations have occurred.

To

reliably distinguish and identify isomeric methylated flavonoids, comparisons to authentic
standards are preferred, but the lack of authentic standards for most methylated flavonoids
prevents detailed structure confirmation necessary to prove regioselectivity of methylation by
OMTs. This has resulted in a significant gap in our knowledge about OMT functions and the
biological importance of individual methylated flavonoids.
Mass spectrometry offers a sensitive and information-rich method for structure elucidation
that is especially useful when only small amounts or impure substances are available.

Mass

spectra generated using electron ionization (EI) has been used to characterize methylated
flavonoids [242-245], but still requires authentic standards or libraries derived from such
144

standards for comparison.

Recently, several groups have investigated the fragmentation

behavior of a few methylated flavonoid standards using soft ionization and tandem mass
spectrometry [246-251].

Although some differences between the collision induced dissociation

(CID) spectra have been observed for different isomers, the range of compounds has been
somewhat limited, and the differences between isomers have been subtle.

At the

commencement of the research described below, comparisons to authentic standards were still
needed for definitive identification of multiply-methylated flavonoids.
Several research teams have concluded that it is nearly impossible to identify positions of
methylation on flavonoids from mass spectrometry alone [220, 227-232].

The basis for this

conclusion is that in CID spectra of methylated flavonoids, loss of the O-methyl group, often as
the neutral CH3 radical, takes place with a lower threshold collision energy than other
fragmentation pathways.

Subsequent fragmentation of carbon-carbon bonds generate fragments

specific to certain portions of the flavonoid ring system, but since such fragments may have
already lost methyl groups, the resulting fragment ions whose masses are consistent with no
methyl substituent do not provide reliable evidence that the methyl group is absent.
In many cases, definitive structure elucidation requires synthesis of the proposed compound
because spectroscopic methods, including MS and NMR, often yield ambiguous results when
authentic standards are not available.

For many methylated flavonoids, the presence of

numerous hydroxyl groups with similar reactivities makes such synthesis challenging, as
selective introduction of methyl groups on a specific position requires position-selective
chemical modification [252-261].

As an alternative, synthesis of methylated flavonoid
145

standards can be achieved by reaction of a flavonoid substrate with an O-methyltransferase
enzyme of known function and regioselectivity.

Unfortunately, the gaps in our understanding

of OMT function and selectivity are large, but provide the rationale for much of the efforts at
understanding biosynthetic enzyme functions that are at the heart of this dissertation.
As part of a collaborative effort with the laboratory of Professor Eran Pichersky at the
University of Michigan, we recently identified two O-methyltransferases from the wild tomato
relative S. habrochaites LA1777 [262].

Assignment of these enzymes as OMTs was initially

based on DNA sequence similarity to other OMTs, but proof of function requires biochemical
evaluation of OMT substrate selectivity and product identification. The characterization of the
functions of these two enzymes was guided by the chemical characterization of their enzyme
products in vivo, through LC/MS profiling of metabolites described in Chapter II, and in vitro
using recombinant OMTs expressed in E. coli. These efforts were steered by the observation
that metabolites from the trichomes of this plant had molecular masses consistent with
polymethylated myricetin.

146

4.2

Experimental Methods

4.2.1

Material

4.2.1.1

Plant material

Seed stock of S. habrochaites LA1777 was obtained from the C.M. Rick Tomato Genetic
Resource Center (TGRC, University of California Davis, USA). Seeds were germinated at
room temperature on filter paper, and transferred to soil approximately 5 days after germination.
Plants were maintained under controlled light conditions of 16 hours light and 8 hours dark.
The temperature gradient was maintained between 24 °C and 18 °C (light:dark, respectively).

4.2.1.2

Chemicals

S-Adenosyl-L-methionine was purchased from Sigma-Aldrich (St. Louis, MO, USA) and
d3-labeled SAM was purchased from C/D/N isotopes (Quebec, Canada).

Flavonoid and

methylated flavonoid substrates included myricetin, laricitrin (3’-methylmyricetin), syringetin
(3’,5’-dimethylmyricetin), and 3’,4’,5’-trimethylmyricetin.

Myricetin was purchased from

Sigma-Aldrich and the remaining substrates were purchased from Extrasynthese (Benay, France).
Methanol, 88% formic acid, and acetonitrile were purchased from VWR Scientific.

147

4.2.2

Enzyme cloning and expression

Recombinant ShMOMT1 and ShMOMT2 enzymes were cloned from cDNA made from S.
habrochaites leaf RNA, expressed in E.coli BL21 Star (DE3) pLysS cells (Invitrogen), purified
by anion exchange chromatography, and assayed by Adam Schmidt of the Pichersky laboratory
at the University of Michigan.

SDS-PAGE was used to establish the degree of purity of the

active fractions.

4.2.3

O-methyltransferase assay

4.2.3.1

General methodology

Enzyme assays were performed by Adam Schmidt, using a Continuous Extraction Assay
Method (CEAM) designed to optimize product accumulation for mass spectrometric analysis.
Assays were performed in 1.5 mL glass vials (Supelco Analytical, 27080-U) in a final volume of
500 µL (aq). A layer of 500 µL 100% ethyl acetate was carefully applied over the aqueous
assay volume to serve as a non-polar extraction phase.

Briefly, the system provides an aqueous

phase (where enzyme reaction takes place) and non-aqueous ethyl acetate phase, with an
interface that promotes preferential partitioning of methylated products to the upper organic
phase. As products are removed from the reaction mixture by diffusion into the non-polar
phase, positive equilibrium is maintained and substrate turnover is enhanced.
148

At the end of

each incubation, the upper organic phase was collected, evaporated to dryness using a SpeedVac,
and redissolved in 50 µL of methanol:water 1:1 (v/v).

4.2.3.2

Individual enzyme assays

Assays for each enzyme contained approximately 23 µg (ShMOMT1) or 32 µg (ShMOMT2)
of purified protein, 500 µM substrate (methyl group acceptor), 500 µM SAM or d3-SAM
(methyl group donor), and all were buffered with 50 mM Tris (pH 7.5).

4.2.3.3

Coupled enzyme assays

ShMOMT1 and ShMOMT2 were combined, 1:1 (v/v), in a final volume of 500 µL (aq).
Each assay contained 500 µM substrate (methyl group acceptor), 500 µM SAM or d3-SAM,
and all were buffered with 50 mM Tris (pH 7.5).

4.2.4

LC/MS

Samples were analyzed on a QTRAP™ 3200 mass spectrometer from Applied
Biosystems/MDS Sciex (Concord, Ontario, Canada) coupled to a Shimadzu LC-20AD binary
pump and SIL-HTC autosampler.

Separation was achieved by using Thermo BetaBasic C18

column (150 mm × 1.0 mm, 5 µm) at temperature of 30 °C.
149

The mobile phase was A: 0.5%

formic acid and B: 0.5% formic acid in 60% methanol+ 40% acetonitrile.
was added to decrease the peak tailing [263, 264].

0.5% formic acid

The mixture of methanol and acetonitrile is

reported to provide superior chromatographic selectivity in separations of flavonoids [265].

A

15 min reverse phase gradient at flow rate of 0.100ml/min was used for separation. The liner
gradient elution program was as follows: 10% B for 0.3 min, then an immediate step up to 40%
B followed by a linear increase to 100% B from 0.31 to 8.5 min, followed by a hold at 100% B
for 2.5 min. At 11 min, the solvent returned to 10% B, and the column was equilibrated for
4 min before the next injection.
with a TurboIonSpray source.

The mass spectrometer was operated in the positive ion mode
Enhanced product ion scanning (EPI) with dynamic fill time was

used to generate product ion MS/MS spectra using 40 V collision potential (parameter CE).
Scan rate was set to 4000amu/sec. The other ionization parameters were as follows: curtain gas
(CUR), 10; ion source gas 1 (GS1), 12; ion source gas 2 (GS2), 30; source temperature (TEM),
400 °C; entrance potential (EP), 10 V; CAD gas, HIGH; capillary potential 5500 V.

The mass

spectrometer and the HPLC system were controlled by Analyst 1.4.2 software from Applied
Biosystems/MDS Sciex.

4.3

Results and Discussion

Kaempferol, quercetin, and myricetin derivatives are metabolites reported in tomato tissues
[266].

However, reports of methylated forms of flavonoids in tissues of any tomato or its

relatives are scarce [267], and are nonexistent for trichome profiling. Transcripts of two genes
150

were found expressed in the glandular trichomes of Solanum habrochaites LA1777, and were
designated as methyltransferases ShMOMT1 and ShMOMT2 based on sequence homology to
other methyltransferase genes.

Six metabolites detected in extracts of isolated S. habrochaites

LA1777 trichomes were initially assigned as putative methylated myricetin aglycones based
upon their masses and retention times.

In the cases of dimethyl- and trimethyl-derivatives, two

putative isomers of each were observed.

To annotate these methylated myricetin metabolites,

commercial standards of methylated flavonol aglycones were incubated with recombinant
methyltransferases to generate products of more extensive methylation.

All these products are

subjected to LC/MS/MS analysis.
Only three methylated myricetin derivatives were available from commercial suppliers:
3’-methylmyricetin, 3’,5’-dimethylmyricetin, 3’,4’,5’-trimethylmyricetin.

In light of the

limited availability of standard reference compounds and the low levels of methylated myricetin
metabolites in trichome extracts, alternative strategies needed to be developed to annotate
methylated flavonoid metabolites.

It was decided to exploit the expected selectivity of

recombinant O-methyltransferase enzymes with tandem mass spectra to assign structures to
methylated metabolites.

Recombinant ShMOMT1 and ShMOMT2 enzyme were exploited to

produce methylated products of myricetin and its methylated analogs.

Products of these

enzymatic reactions could then be compared to trichome metabolites to determine whether they
would match retention times and product ion spectra. The first experiment involved incubation
of myricetin substrate with each individual enzyme (ShMOMT1 or ShMOMT2) and separately
with a mixture of both enzymes.

Annotation of enzyme reaction products was accomplished
151

using a process of elimination, which starts from exhaustively considering all product structures
consistent with observed pseudomolecular masses, and then various criteria were applied to
eliminate candidate structures that were not consistent with retention time and mass
spectrometric data.
The benefit of using enzyme-catalyzed reactions for metabolite semisynthesis arises from
the regioselectivity of most enzymes.

In many cases, addition of a single methyl group is the

major reaction product, and this fact allows more straightforward assignment of methyl group
position.

The methylated positions in the precursors are preserved in the enzyme products,

which reduces the number of the possible isomers that need to be considered.

For example, a

completely unknown trimethylmyricetin would have 20 possibilities based on random
combinations of three methyl groups over six potential positions of methylation on myricetin
hydroxyl groups.

However, if a trimethyl myricetin compound is synthesized by reaction of a

known dimethylmyricetin with a methyltransferase, only 4 isomers are possible.

The success of

this approach then depends on observations of isomer-specific MS/MS spectra that can
discriminate among the four isomers, but might not distinguish all 20 isomers products by
random methylation.
Another benefit of using enzymes for metabolite semisynthesis is that selective
incorporation of CD3 groups at selected positions can be achieved by using trideuterated
co-substrate S-adenosylmethionine-d3.

The end products consist of specific substitutions of

methyl and d3-methyl groups at different positions.

Such compounds are not available from

natural sources and are also not readily synthesize, again, owing to the presence of multiple sites
152

for nonselective alkylation.

These partially labeled methylated myricetins are valuable for

distinguishing fragmentation behavior of specific methyl group using tandem mass spectrometry,
and results from these investigations are described below.
Elimination of candidate metabolite structures is a progressive, and often iterative process.
Any structures already identified from previous steps can be used as references to eliminate
unsuitable candidates in the next step.

Comparisons of retention times and MS/MS spectra of

products from different enzymatic reactions were also used to facilitate candidate elimination or
sometimes make conclusive annotation. Fragment ions observed in the MS/MS product ion
spectra are annotated following previous proposed nomenclature [268] (Figure IV-2).

153

OH

1,3 A+

OH
3'4'
B 5'

0 O 1

HO
7A
5

C
4

OH

O

OH

2
3 OH
3

0,2 B+
1,3 B+

0,4 B+
Figure IV-2 Structure of the flavonol aglycone myricetin. The ring was labeled as large
letters in black. Positions of substitution by hydroxy or methoxy groups are numbered in
black inside the myricetin structure. The bonds on the B ring that undergo cleavage during
collision induced dissociation are numbered in red. Characteristic multiband fragmentations
are labeled in black near the arrow heads, as proposed in Reference [220].

4.3.1

Retention time rules

Liquid chromatographic retention times are features that aid elimination of unsuitable
candidate structures of metabolites, when compared with other isomers and authentic standards.
Methylation converts phenolic groups from weak acids to neutral ethers, and increases the
hydrophobicity of the molecule.
Methylation on different hydroxy groups exerts different degrees of increase of the
molecule’s hydrophobicity, largely because many phenolic groups on myricetin and other
154

flavonols can participate in intramolecular hydrogen bonding.

Hydroxyl groups on the 3- and

5- positions can form hydrogen bonds with the C ring carbonyl group.

Upon methylation at

positions 3 or 5, the intramolecular hydrogen bond is removed, and this frees the electrons on the
carbonyl oxygen to interact more with mobile phase and decreasing retention time.
Methylation of hydroxyls involved in intramolecular hydrogen bonding may be expected to
increase retention on reversed phase HPLC column by a lesser degree than methylation of other
hydroxyl groups.

It was observed in kaempferol and quercetin [238] that adding one extra

methyl on 5 position almost has negligible effect on chromatographic retention times.
Compared to methylation at the 5 position methylation, 3 position methylation increases
retention time more than methylation at the 5 position [238-240]. Positions 3’, 4’ and 5’ form
even weaker intramolecular hydrogen bond with one another.

Therefore, 3’; 4’ or 5’

methylation increases retention time more than 3 or 5 methylation. The 7 position phenolic
group cannot form intramolecular hydrogen bonds based on molecular geometry.

Therefore,

methylation at position 7 increases the retention time more than for the other myricetin isomers.
Also, it was reported before that 7 position has the highest gas phase acidity among the six
hydroxyl groups [269], but this information is more relevant to mass spectrometric behavior than
HPLC retention times.
Based on previous reports using reverse phase chromatography, the effect of individual
methyl groups on retention of myricetin analogs are 7>3’(5’) ≈4’>3>5 [238-240].

The principle

of linear free energy relationships suggests that adding methyl groups will have consistent effects
on the free energy of partitioning between mobile and stationary phases.
155

However, this rule

assumes that all hydroxyl groups are equivalent.

Instead, methylation of hydroxyl groups that

have differential contributions to solute-solvent interactions are expected to have different
retention time shifts upon methylation, and these shifts should be similar as long as the chemical
properties of the group being modified are similar. Following this retention time rule (we name
it rule #1 in our study), the retention times of isomers can be compared with any chemically
similar standards, and this information can be combined with knowledge of chromatographic
retention of already identified enzyme products to suggest methylation positions.

However,

retention time is only conservatively used as a tool to assist the elimination process but final
assignments were mostly based on interpretation of MS/MS spectra.

Comprehensive

investigation of MS/MS product ion spectra provides additional information for metabolite
annotation.

4.3.2

Characterization of enzymatic semisynthesis products

4.3.2.1

Myricetin as substrate

The LC/MS analysis of the products derived from incubating myricetin with ShMOMT1
showed one peak (RT: 8.15 min) in the extracted ion chromatogram (XIC) for monomethylated
myricetin (m/z 333; Figure IV-3B).

This compound elutes slightly later than the

3’-methylmyricetin standard (RT: 8.03 min; Figure IV-4A).

The product ion spectrum (Figure

IV-3E) is distinct from that of the 3’-methylmyricetin standard (Figure
156

IV-4B) in that the

+

[M+H] , m/z 289, 259, 167, and 165 abundances are substantially lower than for
3’-methylmyricetin standard.
3’-methylmyricetin.

Therefore, this monomethylmyricetin was assigned to not be

Since 3- or 5- mono-O-methylmyricetins are expected to elute earlier than

3’-methylmyricetin, the remaining candidates could be 4’- or 7-monomethylmyricetins.

Final

annotation was made using further enzymatic processing of this metabolite as described below.

157

A

Time, min
Figure IV-3 A) Extracted ion chromatogram (XIC) of m/z 333 for monomethyl product from incubation of ShMOMT2 with
myricetin B) Extracted ion chromatogram (XIC) of m/z 333 for monomethyl product from incubation of ShMOMT1 with
myricetin C) Extracted ion chromatogram (XIC) of m/z 333 for monomethyl product from incubation of both ShMOMT1 and
ShMOMT2 with myricetin D）Enhanced product ion scan (EPI) spectra of monomethyl product from incubation of ShMOMT2 with
myricetin, later annotated as 3’-methylmyricetin E）
Enhanced product ion scan (EPI) spectra of monomethyl product from incubation
of ShMOMT1 with myricetin, later annotated as 4’-,methylmyricetin. Data was acquired using Qtrap 3200 instrument using
electrospray positive mode. EPI spectra were acquired under 40V collision energy.

158

Figure IV-3 (cont’d)

B

Time, min

159

Figure IV-3 (cont’d)

C

160

Figure IV-3 (cont’d)

D
[M+H]

+

1,3A

161

+

Figure IV-3 (cont’d)

E

+

1,3A

[M+H]

162

+

A

Figure IV-4 A) Extracted ion chromatogram (XIC) of m/z 333 for commercial standard 3’-methylmyricetin B) Enhanced product
+
ion scan (EPI) spectrum for [M+H] (m/z 333) of commercial standard 3’-methylmyricetin using 40 V collision potential. Data
were acquired using a Qtrap 3200 mass spectrometer using electrospray ionization in positive mode.

163

Figure IV-4 (cont’d)

B
[M+H]

+

1,3A

164

+

This monomethylmyricetin enzyme product can be methylated further by reaction catalyzed
by ShMOMT1, and incubations gave one dimethylmyricetin product (RT 9.3 min) detected as
+

m/z 347 (353 in the incubation using d3-SAM as the cofactor; both as [M+H] ).

Enhanced

product ion MS/MS spectra of m/z 347 yielded fragment ion m/z 167 using unlabeled SAM as
substrate (Figure IV-5A), whereas incubations that employed d3-SAM as substrate yielded a
product heavier by 6 Da, consistent with two CD3 groups.

The EPI MS/MS spectrum of m/z

353 yields a fragment at m/z 170 which is 3 Da heavier than m/z 167 observed using unlabeled
SAM, indicating that m/z 167 fragment ion has one methyl group (Figure IV-5).

Among the

higher mass product ions, we only observe one CH3 loss (loss of 15 Da) from the protonated
molecule.

This finding suggests that one methyl group is far more labile than the other. The

fragment at m/z 167 was reported to arise from a characteristic retro-Diels–Alder reaction (RDA)
+

that is annotated as a 1,3A cleavage for flavonoids having A-ring methyl substitution
[270](Figure

IV-6).

The m/z 167 fragment ion peak is the characteristic ion indicative of A

ring methylation. (we name it as rule #2)

165

A
+

1,3A

[M+H]

+

Figure IV-5 Enhanced product ion scan (EPI) spectra of protonated dimethylated product from A) incubation of myricetin with
ShMOMT1 and SAM; later annotated as 4’,7-dimethylmyricetin and B) incubation of myricetin with ShMOMT1 and d3-SAM, later
annotated as 4’(CD3),7(CD3)-dimethylmyricetin. Data were acquired using a Qtrap 3200 mass spectrometer using electrospray
positive mode and 40 V collision potential.

166

Figure IV-5 (cont’d)

B
+

1,3A

[M+H]

167

+

Figure IV-6 Fragments of m/z 167 derived from protonated 4’,7-dimethylmyricetin that can
+
distinguish sites of methylation through deuterium labeling. Pathway involves 1,3A cleavage
with one methyl group remaining on A ring, retaining the deuterium label if labeled in the
methyl on A ring.

There are two hydroxyl groups on the A-ring of myricetin, at the 5- and 7-positions, that
could potentially undergo methylation.

The appearance of the fragment at m/z 167 in the

methylated myricetins suggests that one of these is methylated (Figure IV-5), and none of the
enzyme incubations or plant extracts yielded product ions at m/z 181 that would point to
methylation at both positions [271].

The mass spectra alone do not allow for definitive

assignments as 5-methyl or 7-methyl substitutions.

However, comparisons of the HPLC

retention of the enzymatic (ShMOMT1) dimethylmyricetin product (RT 9.3 min) with
3’,5’-dimethylmyricetin standard (RT 8.7 min), show later elution of the enzymatic product. The
longer retention time is consistent with methylation at the 7-position because the 7-hydroxyl
cannot participate in intramolecular hydrogen bonding, whereas methylation of the 5-hydroxy
group would disrupt intramolecular hydrogen bonding, and would lead to a smaller increase in
reversed phase retention upon methylation.
168

The monomethylated myricetin generated from ShMOMT1 showed a fragment at m/z 153
consistent with no methylation of the A-ring, and the dimethylmyricetin product was concluded
to have undergone methylation at the 7-position.

Based on the discussion of chromatographic

retention of methylated myricetins described above, it was already concluded that the
monomethylmyricetin product was not methylated at the 3- or 3’- positions.
information, only one isomer is possible, specifically the 4’,7-dimethylmyricetin.

From this
Since methyl

group migrations during MOMT reactions have not been established (nor are they expected), the
monomethylated product (RT 8.15 min) is assigned as 4’-monomethylmyricetin.

One

interesting aspect of the ShMOMT1 product mixture is that the enzyme is capable of adding a
methyl group to the 7- position, but no 7-monomethylmyricetin was observed, which would have
been evident from the m/z 167 indicator of A-ring methylation.

From this information, it is

concluded that ShMOMT1 catalyzed selective methylation of myricetin at the 4’-position, and
the selective formation of 4’,7-dimethylmyricetin suggests that 4’-monomethylmyricetin may be
a better substrate for 7-position methylation than unmodified myricetin.
Using the second recombinant methyltransferase enzyme cloned from S. habrochaites
LA1777 (ShMOMT2) for incubation with myricetin, only one chromatographic peak observed in
+

the extracted ion chromatogram (XIC) for [M+H] corresponding to monomethylmyricetin at
m/z 333 is observed (RT 8.04 min; Figure IV-3A).

Both retention time and enhanced product

ion spectra are indistinguishable from 3’-methylmyricetin standard (Figure IV-3A and Figure
IV-4).

Therefore, the sole monomethylated product of ShMOMT2 is assigned as

3’-methylmyricetin. No incubations of ShMOMT2 with myricetin yielded detectable amounts
169

of products with more than one methyl group.
+

It is notable that one additional feature in the product ion spectra for [M+H] of the 3’- and
4’- isomers of monomethylmyricetin distinguishes these isomers (Figure

IV-3D and E).

The

residual abundance of protonated 4’-methylmyricetin in the product ion spectrum is nearly 4-fold
lower than for 3’-methylmyricetin, when normalized to the base peak abundance.

In both cases,

the base peak arises from loss of methyl radical, and this distinguishing characteristic suggests
that different methyl groups exhibit differential susceptibility to homolytic bond cleavage
following collisional excitation.

More detailed investigations into this phenomenon are

described in Section 4.3.2.2.
Myricetin was incubated both with individual ShMOMT1 and ShMOMT2 enzymes
separately, as well as in mixtures containing both enzymes, to generate as many products as
possible to: (1) establish enzyme functions, and (2) elucidate characteristic fragment ions for
different structures through a combinatorial synthetic approach.

In incubations that employed
+

both enzymes, a broader chromatographic peak for m/z 333 ([M+H] for monomethylmyricetins)
was observed, with a split peak top consistent with two or more overlapping isomers (Figure
IV-3C).

It was anticipated that the mixture of enzymes would form mixtures of methylated

myricetins containing methyl groups at both the 3’- and 4’- positions based upon the activities of
the individual enzymes, and LC/MS results confirmed that both were formed in comparable
amounts.

These two isomers could not be completely resolved using an assortment of HPLC

columns and mobile phase compositions, and this is consistent with earlier observations that 3’and 4’-methylated isomers of quercetin were the isomers most difficult to separate [238-240].
170

Incubations of myricetin with enzyme mixtures (ShMOMT1 and ShMOMT2) yielded four
dimethylmyricetin isomers that were resolved using LC/MS, and enhanced product ion spectra
for all four peaks are presented in Figure IV-7.

Chromatographic peaks 3 (RT 9.34 min) and 4

(RT 9.50 min) both gave the m/z 167 fragment ions characteristic of a single methyl group on
Ring A.

Based on the earlier discussion, it is concluded that the A-ring substituent is on the

7-position oxygen. Since the two enzymes yield 3’- and 4’- monomethylmyricetins as initial
products, and since ShMOMT1 exhibited catalytic activity toward further methylation at the
7-position, it is concluded that the two later eluting peaks are 3’,7-dimethylmyricetin and
4’,7-dimethylmyricetin.

The direct annotation of peak 3 (RT 9.34 min) was made as

4’,7-dimethylmyricetin because its retention time and MS/MS spectrum matched the product
(4’,7-dimethylmyricetin) generated from ShMOMT1 alone.

Annotation of peak 4 was

completed later through use of incubation of 3’-methylmyricetin with enzymes, and is discussed
in

greater

detail

in

section

4.3.2.2.

In

summary,

peak

4

was

annotated

3’,7-dimethylmyricetin and peak 3 (RT 9.34 min) was annotated as 4’,7-dimethylmyricetin.

171

as

A

+

Figure IV-7 A) Extracted ion chromatogram (XIC) of m/z 347 for [M+H] of dimethylated products from incubation of both
ShMOMT1 and ShMOMT2 with myricetin; B) enhanced product ion scan (EPI) spectrum of dimethyl product (RT 8.71),
annotated as 3’,5’-dimethylmyricetin; C) enhanced product ion scan (EPI) spectrum of dimethylated product (RT 8.95), annotated as
3’,4’-dimethylmyricetin; D）EPI spectrum of product (RT 9.34). annotated as 4’,7-dimethylmyricetin; E）EPI spectrum of product
(RT 9.50) annotated as 3’,7-dimethylmyricetin. Data were acquired using a Qtrap 3200 mass spectrometer using electrospray
positive mode and 40 V collision potential.

172

Figure IV-7 (cont’d)

B

[M+H]

+

1,3A

173

+

Figure IV-7 (cont’d)

C
+

1,3A

[M+H]

174

+

Figure IV-7 (cont’d)

D

+

1,3A

[M+H]

175

+

Figure IV-7 (cont’d)

E

[M+H]

+

1,3A

176

+

For the two early eluting dimethylmyricetin products generated from the mixed enzyme
incubation (Figure IV-7A), the fragment of m/z 167 that is characteristic of A-ring methylation
was not observed (Figure

IV-7B and C), indicating no A-ring substitution.

Combining the information about the other metabolites, the following constraints are in
place for both of the two early eluting dimethylmyricetin isomers derived from incubations with
myricetin as substrate: 1) no methyl on A- ring; b) at least one or more methyl groups on the 3’or 4’- positions; 3) 3’ is equivalent to 5’ owing to symmetry.

Within these constraints, there are

four possibilities for these two dimethylmyricetin isomers: (3’,5’), (3’,4’), (3,4’) and (3,3’).
The first peak (RT 8.71 min) matched retention time and fragmentation pattern of authentic
3’,5’-dimethylmyricetin.

Based on the chromatographic retention times, the best match of latter

peak was 3’,4’-dimethylmyricetin because it elutes about 0.2 min later (RT 8.95 min) than the
3’,5’-isomer, and the 3,3’- and 3,4’-isomers would be expected to elute earlier owing to
intramolecular hydrogen bonding.

All of the four dimethylmyricetin products are consistent

with the enzyme selectivity observed in the individual enzyme incubations that demonstrated
methylations at 3’-, 4’-, and 7- positions of myricetin.
Several features in the MS/MS spectra of these two 3,3’- and 3,4’-isomers distinguish them
from the 3’,7- and 4’-7-dimethylmyricetins described above.
feature, lies is the abundance of higher mass product ions.

The first, and most prominent

For both of 3,3’- and 3,4’-isomers,

the most abundant product is m/z 331, which is 16 Da lower in mass than the precursor.

In

contrast, the dominant product ion for the later eluting isomers (3’,7 and 4’,7) was m/z 332, an
odd-electron fragment corresponding to loss of methyl radical (15 Da).
177

The loss of 16 Da in the

3,3’- and 3,4’-isomers can be explained either by successive losses of CH3• and a hydrogen atom,
or a concerted elimination of CH4, which would require a hydrogen migration.

A more

complete and systematic exploration of the structural features responsible for these differences is
presented in subsequent sections (Section 4.3.3) of this Chapter.

4.3.2.2

3’-methylmyricetin as methyltransferase substrate

The catalytic activity of ShMOMT2 yielded 3’-O-methylmyricetin as its only product.
Because only one methylated product was formed, and since commercial sources of this product
were available, more extensive investigations of ShMOMT1 functions could be explored.
Incubations of ShMOMT1 with 3’-methylmyricetin and SAM yielded two peaks in extracted ion
chromatograms corresponding to protonated dimethylmyricetins (m/z 347; Figure IV-8).

The

later eluting peak, (RT 9.47 min) gave the characteristic fragment ion at m/z 167 which indicates
methylation at either the 5- or 7- positions. As discussed above, this was annotated as
3’,7-dimethylmyricetin and not 3’,5-dimethylmyricetin based on its late elution as compared to
the 3’,5’-dimethylmyricetin standard (RT 8.73 min).

The retention time and product ion

spectrum is indistinguishable from the fourth peak (RT 9.50 min) in the mixed enzyme
incubation described in Section 4.3.2.1 (Figure IV-7E), and supports its assignment as
3’,7-dimethylmyricetin also.

The first chromatographic peak (RT 8.97 min; Figure IV-8)

yielded fragment at m/z 153 but not m/z 167, which indicates no A-ring substitution.
Considering the starting substrate is 3’-methylmyricetin, the remaining possibilities are (3’,5’),
178

(3’,4’) or (3,3’) isomers.

Comparison of the product ion spectra of this isomer with an

authentic standard of 3’,5’-dimethylmyricetin led to its elimination as a candidate.
two possibilities are left which are (3’,4’) and (3,3’).

Then only

Since its retention time and product ion

spectra matched the second peak in the mixed enzyme incubation described in Section 4.3.2.1, it
was identified as 3’,4’-dimethylmyricetin (Figure IV-7C).

179

A

+

Figure IV-8 A) Extracted ion chromatogram (XIC) of m/z 347 ([M+H] ) for dimethylmyricetin products from incubation of
ShMOMT1 with 3’-monomethylmyricetin B) EPI spectrum of dimethylmyricetin product (RT 8.97), later annotated as
3’,4’-dimethylmyricetin C) EPI spectrum of dimethyl product (RT 9.47), later annotated as 3’,7-dimethylmyricetin. Data were
acquired using a Qtrap 3200 mass spectrometer using electrospray positive mode and 40 V collision potential.

180

Figure IV-8 (cont’d)

B
+

1,3A

[M+H]

181

+

Figure IV-8 (cont’d)

C

[M+H]

+

1,3A

182

+

One trimethylmyricetin product was generated from incubations of 3’-methylmyricetin with
ShMOMT1 (RT 10.22 min; Figure IV-9).

The presence of m/z 167 in the product ion spectrum

is indicative of A-ring methylation, which is consistent with methylation in both 4’- and 7positions as these were formed independently.

Furthermore, this isomer elutes later than

3’,4’,5’-trimethylmyricetin standard (RT: 9.70 min), which is consistent with its assignment as
3’,4’,7-trimethylmyricetin.

183

+

1,3A

[M+H]

+

+

Figure IV-9 EPI spectrum of m/z 361 ([M+H] ) for trimethylmyricetin product (RT 10.22 min) from incubation of ShMOMT1
with 3’-monomethylmyricetin, later annotated as 3’, 4’, 7-trimethylmyricetin. Data were acquired using a Qtrap 3200 mass
spectrometer using electrospray positive mode and 40 V collision potential.

184

Incubations of 3’-methylmyricetin with ShMOMT2 or with mixtures of ShMOMT1 and
ShMOMT2 yielded two chromatographic peaks in XIC (m/z 347) corresponding to
dimethylmyricetins using ShMOMT2 incubation and four analogous chromatographic peaks in
the products from the mixed enzyme incubations.

Since the two dimethylmyricetins from

ShMOMT2 incubation were also observed in the mixed enzyme incubation, coverage of the
complete range of isomers is accomplished via analysis of the mixed enzyme products (Figure
IV-10).

Considering that 3’-methylmyricetin is the starting reagent, five dimethylmyricetin

isomers are theoretically possible: (3’,4’), (3’,5’), (3,3’), (3’,5) and (3’,7).

An exploration of

the retention times and product ion spectra was undertaken to annotate the incubation products.
The chromatographic retention of the (3’,7) isomer should be greatest, as discussed above, and
this isomer would be expected to yield a fragment at m/z 167 that is characteristic of a
monomethyl-subsituted A-ring.

The latest eluting isomer (RT 9.50 min) did indeed show this

fragment, and was therefore annotated as 3’,7-dimethylmyricetin.

The second eluting isomer

(RT 8.73 min) showed a strong match with retention and product ion spectra of the commercially
available standard 3’,5’-dimethylmyricetin, and was annotated as such.

The possibility that the

third eluting isomer (Figure IV-10D) was 3’,5’-dimethylmyricetin was then excluded.

Because

the m/z 153 fragment indicates no A-ring methylation, and its late retention time (RT 8.98 min)
argues against the 3,3’-substitution (which should elute earlier).

The only remaining candidate

is 3’,4’-dimethylmyricetin, and this finding is supported by of its match of retention time and
MS/MS spectrum with the 3’,4’-dimethylmyricetin product formed by ShMOMT1 (which
exhibits 4’- methyltransferase activity) methylation of 3’-methylmyricetin (Figure IV-8). The
185

remaining and earliest-eluting isomer (RT 8.34 min) is annotated from the remaining options,
3,3’- or 3’,5-dimethylmyricetin.

As discussed above, the 5-methyl isomer is expected to give
+

the m/z 167 fragment peak through 1,3 A cleavage.

However, this isomer only shows a

fragment at m/z 153, consistent with no A-ring methylation.
3,3’-dimethylmyricetin.

186

Therefore, it is assigned as

A

+

Figure IV-10 A) Extracted ion chromatogram (XIC) of m/z 347 ({M+H] ) for dimethylmyricetin products from incubation of both
ShMOMT1 and ShMOMT2 with 3’-monomethylmyricetin; B) EPI spectrum of product (RT 8.34 min), later annotated as
3,3’-dimethylmyricetin C) EPI spectrum of dimethyl product (RT 8.73 min), later annotated as 3’,5’-dimethylmyricetin D) EPI
spectrum of dimethyl product (RT 8.98min). later annotated as 3’,4’-dimethylmyricetin E) EPI spectrum of dimethyl product (RT
9.52min). later annotated as 3’,7-dimethylmyricetin. Data were acquired using a Qtrap 3200 mass spectrometer using electrospray
positive mode and 40 V collision potential.

187

Figure IV-10 (cont’d)

B

[M+H]

+

1,3A

188

+

Figure IV-10 (cont’d)

C

+

[M+H]

1,3A

189

+

Figure IV-10 (cont’d)

D
+

1,3A

[M+H]

190

+

Figure IV-10 (cont’d)

E

[M+H]

+

1,3A

191

+

From these isomer annotations, it is also concluded that incubations of 3’-methylmyricetin
with ShMOMT2 alone yield two isomers: 3,3’- and 3’,5’-dimethylmyricetin, and this finding
demonstrates that ShMOMT2 can methylate at the 3- and 5’ (equivalent to 3’)-positions.

A

summary of the enzymatic conversions of myricetin and the enzymatic monomethylmyricetins is
presented in Scheme IV-1.

One novel aspect of this discovery is the methylation at the

3-position by ShMOMT2, and this activity is considered rare among plant O-methyltransferases.

192

Scheme IV-1 Conversions of myricetin and monomethylmyricetins by ShMOMT1 and
ShMOMT2 enzymes. Products of ShMOMT1 alone are colored red, of ShMOMT2 alone are
colored blue, and products requiring both enzymes are colored brown.

As mentioned above in Section 4.3.2.1, product ion MS/MS spectra suggested that the
fragmentation behavior of methyl groups varied based upon the position of the methyl group.
However, to establish different methyl group behaviors in molecules containing multiple methyl
groups, a tag or indicator of which methyl group is present in a product ion is needed.
193

Selective

methylation of flavonoids using common chemical methylation reactions does not offer adequate
selectivity to be useful for chemical synthesis without invoking complex schemes for protecting
and deprotecting flavonoid functional groups.

For this reason, enzymatic syntheses, which are

usually more selective, were employed to introduce deuterated methyl groups in a limited range
of positions on myricetin.
In incubations conducted using ShMOMT1 with 3’-methylmyricetin and deuterium labeled
d3-SAM, the 3’-methyl-4’-d3-methylmyricetin contains one unlabeled and one labeled methyl
group.

+

The surprising finding in the product ion spectrum for [M+H] is that it gives a barely

detectable peak corresponds to loss of unlabeled methyl radical or CH4 (m/z 335 or 334
respectively), as was observed for the unlabeled compound as described above.

Instead, only

loss of labeled methyl radical plus hydrogen atom (m/z 331, loss of 19 Da) was observed (Figure
IV-11).

Two explanations regarding this phenomenon are postulated.

First, the initial

fragment ion from loss of the 3’-O-methyl group may undergo more rapid conversion to other
ionic products than occurs following 4’-O-methyl loss. However, loss of methyl radical or
methane (CH3•, -15 Da or CH4, -16 Da) are the base peaks in product ion spectra from
commercial standards of 3’-methylmyricetin and 3’,5’-dimethylmyricetin, respectively, at the
same collision potential (40 V).
hypothesis.

These findings argue against the unstable intermediate

The second postulated option suggests that the rates of losses of methyl radical

from the 3’- and 4’-positions are quite different, with loss of the 4’-position CD3 group much
faster than loss of the 3’- methyl radical when both are present in the same molecule. While
kinetic isotope effects are anticipated, the magnitude of the kinetic differences would be difficult
194

to explain based on kinetic isotope effects alone.
As was the case for the 3’-methyl-4’-(d3)-methyl isomer (Figure

IV-11), the enhanced

product ion spectra of 3(d3)-methyl-3’-methylmyricetin show no CH3 radical loss (m/z 335) but
only CD3 loss (m/z 332; Figure IV-12).

As discussed above, this position-selective loss

indicates the methyl group on the 3-position is more labile under these CID conditions relative to
the 3’ position.
The findings of this experiment demonstrate that differences in chemical behavior of methyl
groups in methylated phenolic metabolites may provide a distinguishing feature useful for isomer
discrimination.

These kinds of isomers have been exceptionally difficult to resolve using mass

spectrometry alone.

Enzymatic synthesis of compounds having both labeled and unlabeled

methyl groups are the key for this discovery because such compounds are not available in nature
and also challenging to make by chemical synthesis. Understanding the reactivity of O-methyl
groups and its dependence upon position of substitution provides a critical tool for metabolite
structure identification, and its importance will be discussed more below.

195

+

1,3A

[M+H]

+

+

Figure IV-11 EPI spectrum of m/z 350 ([M+H] ) for dimethylmyricetin product 3’-methyl-4’(d3)-methylmyricetin from
incubation of ShMOMT1 with 3’-monomethylmyricetin and d3-SAM. Data were acquired using a Qtrap 3200 mass spectrometer
using electrospray positive mode and 40 V collision potential.

196

+

1,3A

[M+H]

+

Figure IV-12 EPI spectrum of m/z 350 for dimethyl product 3(d3)-methyl-3’-methylmyricetin from incubation of both ShMOMT1
and ShMOMT2 with 3’-monomethylmyricetin and d3-SAM. Data were acquired using a Qtrap 3200 mass spectrometer using
electrospray positive mode and 40 V collision potential.

197

4.3.2.3

3’, 5’-methylmyricetin as methyltransferase substrate

As described above, the catalytic activity of ShMOMT2 alone can convert myricetin to
3’-methyl- and 3’,5’-dimethylmyricetins.

Since trichome extracts yielded an assortment of

more extensively methylated myricetins, the possibility that these might form from actions of
multiple

methyltransferase

enzymes

was

explored

further.

Incubation

of

3’,5’-dimethylmyricetin with ShMOMT1 yielded two chromatographic peaks with masses (m/z
+

361; [M+H] ) corresponding to trimethylmyricetin isomers.

The first isomer (RT 9.70 min)

matched retention time and product ion spectra with 3’,4’,5’-trimethylmyricetin, and was
annotated as such.

For the second isomer (RT 10.14 min), its product ion spectrum (Figure

IV-13) gives the characteristic A-ring methylated product ion of m/z 167.

Since its retention

time is longer than 3’,4’,5’-trimethylmyricetin standard (RT 9.70 min) the only option is
3’,5’,7-trimethylmyricetin.

198

A

+

Figure IV-13 A) Extracted ion chromatogram (XIC) of m/z 361 ([M+H] ) for trimethyl product from incubation of ShMOMT1
with 3’,5’-dimethylmyricetin. B) EPI spectrum of m/z 361 for trimethyl product (RT 9.71min), later annotated as
3’,4’,5’-trimethylmyricetin C) EPI spectrum of m/z 361 for trimethyl product (RT 10.14 min), later annotated as
3’,5’,7-trimethylmyricetin. Data were acquired using a Qtrap 3200 mass spectrometer using electrospray positive mode and 40 V
collision potential.

199

Figure IV-13 (cont’d)

B
[M+H]

+

1,3A

200

+

Figure IV-13 (cont’d)

C
[M+H]

+

1,3A

201

+

Incubations of 3’,5’-dimethylmyricetin with ShMOMT2 yielded evidence of only a single
trimethylmyricetin (RT 9.21 min) which eluted earlier than 3’,4’,5’-trimethylmyricetin standard
(RT: 9.70 min).

The product ion spectrum of this isomer yielded a peak at m/z 153 but not at

m/z 167, pointing to no substitution on either the 5- or 7-positions.

The only option consistent

with these observations is 3,5’,3’-trimethylmyricetin (Figure IV-14), and this structure provides
evidence of a second ShMOMT2 substrate that is methylated at the 3-position.

Product ion

+

MS/MS spectra from [M+H] of the trimethylmyricetin product generated from incubations
using d3-SAM showed a prominent neutral loss of 18 Da (CD3 radical; m/z 346) but no peak
corresponding to neutral loss of 15 Da (CH3 radical; m/z 349) (Figure IV-14B).

Again, this

fragmentation is consistent with the finding described above that when 3- and 3’- (or 5’-) methyl
groups are present on the same molecule, the direct methyl loss from the protonated molecule
comes almost exclusively from the 3- methyl position.
To further understand the formation of more extensively methylated myricetins,
3’,5’-dimethylmyricetin was incubated with a combination of ShMOMT1 and ShMOMT2.
Unlike the incubation with ShMOMT2 alone, the mixture of enzymes generated two
trimethylmyricetin isomers. These peaks are assigned as 3’,5’,7-trimethylmyricetin, which is
presumed to be formed by the ShMOMT1 capacity to methylate at the 7-position, and
3,5’,3’-trimethylmyricetin discussed above as a product of ShMOMT2 activity.

202

A

+

1,3A

[M+H]

+

+

Figure IV-14 A) EPI spectra of m/z 361 ([M+H] ) for trimethylmyricetin product from incubation of ShMOMT2 with
3’,5’-dimethylmyricetin later annotated as 3, 3’,5’-trimethylmyricetin; B) EPI spectrum of m/z 364 for trimethyl product from
incubation of ShMOMT2 with 3’,5’-dimethylmyricetin and d3-SAM, later annotated as 3(d3)-3’,5’ -trimethylmyricetin. Data were
acquired using a Qtrap 3200 mass spectrometer using electrospray positive mode and 40 V collision potential.

203

Figure IV-14 (cont’d)

B

+

1,3A

[M+H]

204

+

4.3.2.4

3’,4’,5’,-trimethylmyricetin as methyltransferase substrate

The results described above demonstrated that ShMOMT1 can introduce methyl groups at
the 4’ position of myricetin, and ShMOMT2 can introduce methyls at the 3’- and 5’-positions.
The combination of both enzymes can form 3’,4’,5’-trimethylmyricetin, and its further
transformation was investigated, since authentic substrate was commercially available and this
substrate has all three B-ring positions methylated.

Incubation of 3’,4’,5’-trimethylmyricetin
+

with ShMOMT2 yielded only one chromatographic peak with [M+H] of m/z 375 (RT 9.81 min)
consistent with a tetramethylmyricetin.

The product ion MS/MS spectrum exhibited a strong

m/z 153 fragment ion, which indicates no A-ring methylation (Figure

IV-15).

The only

remaining possibility is 3,3’,4’,5’-tetramethylmyricetin, again demonstrating the capacity of
ShMOMT2 to methylate the 3-position hydroxyl.

In incubations using d3-SAM as the

co-substrate, the counterpart product of 3-d3-methyl-3’,4’,5’-trimethylmyricetin gave a major
neutral loss of 18 Da (CD3; m/z 360) followed by a successive neutral loss of 15 Da (CH3; m/z
345) but only a miniscule peak corresponding to loss of 16 Da (CH3 plus H) from the protonated
molecule.

As observed for less methylated myricetins, when 3’- (or 5’-), 4’-, and 3- O-methyl

groups are all present, the dominant methyl loss from the protonated molecule comes from the 3position (Figure IV-15B).

The minor peak that corresponds to neutral loss of 16 Da (CH3+H;

m/z 362) is believed to come from the 4’- position methyl radical loss plus an additional
hydrogen atom, from sources yet to be discussed.

Because we already observed when 4’ and 3’

(or 5’) methyl groups are present, direct methyl loss from the protonated molecule mainly comes
205

from the 4’ position (Figure IV-11) unless a methyl group is also present at the 3-position
oxygen.

We conclude that the 3-O-methyl group is even more labile than 4’-O-methyl.

Based

on all of the trideuteromethyl analogs investigated for this project, we conclude when protonated
polymethyl flavonoids undergo collisional activation, the rate of loss of methyl direct from the
protonated molecule follow the ranking 3 > 4’ >> 3’(5’) >>7.

The rate of loss of methyl groups

provides a valuable tool that can differentiate isomers formed by methylation (Rule #3).

206

A
[M+H]

+

1,3A

+

Figure IV-15 A) EPI spectrum of m/z 375 ([M+H] ) for tetramethylmyricetin product (RT 9.81 min) from incubation of
ShMOMT2 with 3’,4’,5’-trimethylmyricetin, later annotated as 3, 3’, 4’,5’-tetramethylmyricetin; B) EPI spectrum of m/z 378
+
([M+H] ) for tetramethylmyricetin product (RT 9.81 min) from incubation of ShMOMT2 with 3’,4’,5’-trimethylmyricetin and
d3-SAM, later annotated as 3(d3) 3’, 4’,5’-tetramethylmyricetin. Data were acquired using a Qtrap 3200 mass spectrometer using
electrospray positive mode and 40V collision potential.

207

+

Figure IV-15 (cont’d)

B

[M+H]

+

1,3A

208

+

The 3’,4’,5’-trimethylmyricetin was also subjected to incubations with ShMOMT1, and
only one chromatographic peak was observed to correspond to a tetramethylmyricetin, though in
+

low yield. This single product, gives m/z 167 in the MS/MS product ion spectrum of [M+H] ,
an indication of a single methyl group on the A-ring (Figure IV-16).
3’,4’,5,5’- or 3’,4’,5’,7-tetramethylmyricetin.

The two possibilities are

The enzymatic reaction product (RT 10.31 min)

eluted later than 3, 3’,4’,5’-tetramethylmyricetin, (RT 9.81 min), and this is consistent with
3’,4’,5’,7-tetramethylmyricetin.

This

finding

methylation at the 7-position, as discussed above.

is

consistent

with

ShMOMT1-catalyzed

Incubation of the 3’,4’,5’-trimethylmyricetin

with the mixture of ShMOMT1 and ShMOMT2 yielded two tetramethylmyricetin products,
which are the same as those generated by incubations with the individual enzymes.

209

[M+H]

+

+

1,3A

+

Figure IV-16 EPI spectrum of m/z 375 ([M+H] ) for tetramethylmyricetin product (RT 10.31 min) from incubation of ShMOMT1
with 3’,4’,5’-trimethylmyricetin, later annotated as 3’,4’,5’,7-tetramethylmyricetin. Data were acquired using a Qtrap 3200 mass
spectrometer using electrospray positive mode and 40 V collision potential.

210

4.3.3

The origin of hydrogen radical loss from protonated O-methylflavonoids

In six authentic standards and products of enzymatic methylation of myricetin and its
methylated analogs: 3’,5’-dimethylmyricetin (Figure IV-7B), 3’,4’-dimethylmyricetin (Figure
IV-7C), 3’,4’,7-trimethylmyricetin (Figure IV-9), 3’,4’,5’-trimethylmyricetin (Figure IV-13B),
3’,5’,7-trimethylmyricetin (Figure IV-13C) and 3’,4’,5’,7-tetramethylmyricetin (Figure IV-16),
a major product ion corresponding to loss of methane (16 Dalton loss) from the protonated
molecule is more abundant than loss of CH3 (15 Dalton) using 40 V collision potential.

This

behavior stands in contrast to all other methylated myricetin derivatives studied, for which the
product from loss of CH3 was far more abundant than the 16 Da neutral loss.

Interestingly, all

six compounds that show a significant 16 Da neutral loss have two or more than two methyl
substitutions on the B-ring, therefore it is concluded that that the apparent loss of methane is
characteristic of compounds with two or more methyl groups on the B ring of myricetin.
However, two methylated myricetins are exceptions to this rule:

3,3’,5’-trimethylmyricetin

(Fig IV-14A) and 3, 3’,5’,7 tetramethylmyricetin (Figure IV-21) , both of which have 3-position
methyl groups which are more labile to loss of CH3 than all B-ring methyl groups as discussed
above. Therefore, it is suggested that a prominent loss of 16 Da from methylated myricetins is
characteristic of two or more B-ring methyl groups and no substitution at the 3-position oxygen.
The mechanism that leads to loss of 16 Da from methylated myricetins with two B-ring
methyl groups remains uncertain.

Product ion spectra for the entire set of methylated

211

+

myricetins showed no evidence of loss of hydrogen radical from [M+H] using CID potentials
ranging from 25 to 80 V with 5 V increments. Subsequent MS
+

3

analysis of

+

[M+H] —>[M+H-15] —> product ions also failed to yield detectable fragments resulting from
a hydrogen atom loss in the second dissociation. Based on these results, the possibility of a
concerted loss of CH4 cannot be discounted.

If the elimination of CH4 were concerted, the

origin of the additional hydrogen must be within range of migrating to the carbon of the methyl
group eliminated, such as: 1) the hydrogen involved in protonation, 2) a hydrogen atom from a
hydroxyl group, most likely the 3- and 4’ or 5’ positions which are nearest; 3) carbon-bound
hydrogen on the B-ring (e.g. 2’ or 6’ positions); 4) hydrogen from a nearby O-methyl group.
To address the source of the additional hydrogen in the 16 Da neutral loss from protonated
methylated myricetins, product ion spectra were generated for the deuterated analogs described
above.

An

analog

with

a

single

deuterated

methyl

group

on

the

B-ring,

3’-methyl-5’(d3)-methylmyricetin yielded neutral losses of 16 Da (-CH4 or –CH2D•, m/z 334),
17 Da (most likely -CH3D, m/z 333), and 19 Da (-CD3H, m/z 331) from the protonated molecule
(Figure IV-17).

In the corresponding unlabeled compound, the product from loss of CH4 was

more abundant than the product generated by loss of CH3•.

Both losses of 17 and 19 Da from

the deuterated analog indicate that at least one deuterium from the 5’(d3)-methyl group was lost
in the neutral product.

Similar results were also observed for other deuterated methylmyricetins

including 3’(d3),5’(d3)-dimethylmyricetin which gave a prominent neutral loss of 20 Da (-CD4,
m/z 333) from the protonated molecule (Figure IV-18).

212

These results provide compelling

evidence that methyl hydrogen are a source of hydrogen loss, either as a concerted loss of CH4
or successive losses of CH3• and H•.
lost hydrogen atom.

However, methyl hydrogens are not the only source of the

For example, the 16 Da neutral loss product (m/z 334) from

3’-methyl-5’(d3)-methylmyricetin corresponds to loss of CH4, but no methyl hydrogens on the 5’
position methyl group are available to donate a proton, as all are deuteriums.
hydrogen must come from other sources.

Therefore, this

Since protonated 3’,4’,5’-trimethylmyricetin also

yields a prominent product from loss of CH4 (Figure IV-13B), a B-ring hydroxyl group is not a
prerequisite for generating loss of CH4.

There is little evidence that B-ring free hydroxyl

groups are the source of the final hydrogen, since no loss of 16 Da was observed for 3’-methylor 4’-methylmyricetin (Figure

IV-3), both of which have two B-ring hydroxyls.

Likewise,

numerous methylmyricetins with hydroxyl groups on A- and C-rings did not yield loss of CH4,
suggesting that these are also not the hydrogen source.

Using similar logic, all of these

myricetin derivatives had protons at the 2’- and 6’-positions, and these therefore cannot show
correlation with the occurrence of CH4 loss.

Having discounted the likelihood that the 2’- and

6’- hydrogens and hydroxyl hydrogens are important contributors in the neutral loss of CH4 from
3’-methyl-5’(d3)-methylmyricetin, attention should turn to the proton involved in ionization.

213

+

1,3A

[M+H]

+

+

Figure IV-17 EPI spectrum of m/z 350 ([M+H] ) for dimethylmyricetin products from incubation of ShMOMT2 with
3’-monomethylmyricetin and d3-SAM, later annotated as 3’-methyl-5’(d3)-methylmyricetin. Data were acquired using a Qtrap 3200
mass spectrometer using electrospray positive mode and 40 V collision potential.

214

Figure IV-17 (cont’d)

+

1,3A

[M+H]

+

+

Figure IV-18 EPI spectrum of m/z 353 ([M+H] ) for dimethylmyricetin product from incubation of both ShMOMT1 and
ShMOMT2 with myricetin and d3-SAM, later annotated as 3’(d3),5’(d3)-dimethylmyricetin. Data were acquired using a Qtrap 3200
mass spectrometer using electrospray positive mode and 40 V collision potential.

215

In summary, the product ion MS/MS spectra of an assortment of methylated myricetin
derivatives demonstrated that neutral loss of CH4 is diagnostic for myricetin derivatives that
have two or more B-ring O-methyl groups are present, provided that no methyl is situated on the
3-position (Rule #4).

It is anticipated that structurally similar flavonoids such as quercetin

derivatives may follow these rules of annotation owing to their similar chemical properties.

4.3.4

Application of annotation rules for identification of methylated myricetin

metabolites in S. habrochaites LA1777 trichomes

In extracts of glandular trichomes from the wild tomato S. habrochaites LA1777, several
compounds were detected using LC/MS profiling that exhibited molecular and fragment masses
consistent with assorted methylated myricetins.

The low levels of these compounds in

trichome extracts precluded use of techniques such as NMR, therefore structure elucidation was
performed using LC/MS/MS.

For putative dimethylmyricetins, chromatographic peaks in the

XIC (m/z 347) suggested two isomers of dimethylmyricetin. The more abundant of these (RT
9.51 min) matched the enzymatic product 3’,7-dimethylmyricetin in both retention and
fragmentation pattern (Figure IV-19).

The earlier eluting but less abundant metabolite (RT

8.45 min) gave a product ion spectrum showing m/z 153 but not m/z 167, suggesting no
methylation on the A-ring (5- or 7-positions).

The absence of a neutral loss of 16 Da (-CH4)

argues against both methyl groups being on the B-ring (Rule #4).

By comparison of

chromatographic retention with products of enzymatic methylations, the 3,3’-dimethylmyricetin
216

was eliminated (Rule#1).

Since this metabolite eluted later than the 3,3’- isomer (RT 8.34

min), the only remaining possibility is 3,4’-dimethylmyricetin.

Here it is intriguing that a

substantial difference in the pattern of methyl group losses was observed in comparisons of
product ion MS/MS spectra of 3,4’-dimethylmyricetin and 3,3’-dimethylmyricetin generated
using enzymatic synthesis (Figure IV-10B).

For 3,3’-dimethylmyricetin, the dominant

fragment ion arises from loss of a single CH3• (m/z 332) with a substantially less abundant peak
for losses of two CH3 groups (m/z 317).

In contrast, the 3,4’-dimethyl isomer yields m/z 317

(loss of two CH3 groups) as the dominant fragment ion.

Consistent with our proposed Rule #

3 (lability of methyl groups), the more facile loss of the second methyl argues for substitution at
the 4’ position rather than 3’.

217

A
[M+H]

+

+

1,3A

+

Figure IV-19 A) EPI spectrum of m/z 347 ([M+H] ) for dimethylmyricetin metabolite in S. habrochaites LA1777 leaf dip extracts,
+
later annotated as 3’,7-dimethylmyricetin (RT 9.51 min); B) EPI spectrum of m/z 347 ([M+H] ) for dimethylmyricetin metabolite in
S. habrochaites LA1777 leaf dip extracts, later annotated as 3,4’-dimethylmyricetin (RT 8.45 min). Data were acquired using a
Qtrap 3200 mass spectrometer using electrospray positive mode and 40 V collision potential.

218

Figure IV-19 (cont’d)

B

+

1,3A

[M+H]

219

+

Trichome extracts of S. habrochaites LA1777 also yielded evidence for two
trimethylmyricetin isomers (XIC of m/z 361; Figure IV-20).

The later eluting

chromatographic peak (RT 9.61 min) shows the m/z 167 fragment indicative of a single methyl
substitution on an A-ring oxygen (Rule #2).
enzymatic

synthesis

products

The retention time for this peak lies between

3,3’,5’-trimethylmyricetin

3’,4’,7-trimethylmyricetin (RT 10.22 min).

(RT

9.21

min)

and

No authentic standard was available that matched

this product. Based on the relatively long chromatographic retention time, we conclude the
A-ring substituent is a methyl at the 7-position (Rule #1). The product ion MS/MS spectrum
did not yield any major loss of CH4 from the protonated molecule, indicating either a methyl at
the 3-position or less than two methyls on the B-ring (Rule #4).

From this information, two

candidates remained: 3,3’,7- or 3,4’,7-trimethylmyricetin. As discussed above, the 4’-methyl
group is more labile upon CID than the 3’- methyl.

Relative abundances of the residual

precursor ion to product ions attributed to losses of one and two CH3 groups are useful
diagnostics of the lability of the methyl substituents (Rule #3).

For this more abundant

trimethylmyricetin isomer, relative abundances match 3,3’-dimethylmyricetin (Figure IV-10B)
better

than

3,4’-dimethylmyricetin

3,3’,7-trimethylmyricetin.

(Figure

IV-19B),

so

it

was

annotated

as

The less abundant isomer (RT 9.32 min) yielded m/z 153 rather

than m/z 167 peaks indicating no A-ring substituent (Rule #2).

It elutes earlier than the

3’,4’,5’- isomer (RT: 9.7 min) but slightly later than 3,3’,5’-trimethylmyricetin (9.21 min).
Excluding candidates with methylation at the 5- position (as these are expected to elute earlier)
(Rule #1), and two standards, the commercial 3’,4’,5’- isomer and the enzyme product
220

3,3’,5’-trimethylmyricetin which did not match retention time or fragmentation, the only
remaining option was 3,3’,4’-trimethylmyricetin.

Although two methyl groups are situated on

the B-ring, the methyl group on the 3- position is most labile and dominates the product ion
spectrum, precluding the 16 Da loss that would occur in its absence.

221

A

[M+H]

+

+

1,3A

+

Figure IV-20 A) EPI spectrum of m/z 361 ([M+H] ) for trimethylmyricetin metabolite in S. habrochaites LA1777 leaf dip extracts,
later annotated as 3, 3’,7-trimethylmyricetin, (RT 9.61 min); B) EPI spectrum of m/z 361 for trimethylmyricetin metabolites in S.
habrochaites LA1777 leaf dip extracts, later annotated as 3,3’,4’-trimethylmyricetin. (RT 9.32 min). Data were acquired using a
Qtrap 3200 instrument using electrospray positive mode and 40 V collision potential.

222

Figure IV-20 (cont’d)

B

+

1,3A

[M+H]

223

+

The trichome extract yielded a single chromatographic peak (RT 10.48 min) corresponding
to a tetramethylmyricetin for which the product ion MS/MS spectrum was generated (Figure
IV-21).

The m/z 167 fragment ion indicates one methyl substituent on the A-ring (Rule #2).

This metabolite eluted later than enzyme product 3,3’,4’,5’-tetramethylmyricetin (9.81 min),
indicating that the A-ring methylation is on the 7- position (Rule #1).

The absence of a

fragment corresponding to loss of CH4 eliminates the 3’,4’,5’,7- isomer, and based on the
relative abundances of the unfragmented precursor and successive methyl loss peaks, the best
match was 3,3’,5’,7-tetramethylmyricetin (Rule #3, 4).

224

[M+H]

+

+

1,3A

+

Figure IV-21 EPI spectrum of m/z 375 ([M+H] ) for tetramethylmyricetin metabolite in S. habrochaites LA1777 leaf dip extracts,
later annotated as 3,3’,5’,7-tetramethylmyricetin, (RT 10.48 min). Data were acquired using a Qtrap 3200 mass spectrometer using
electrospray positive mode and 40 V collision potential.

225

The LC/MS analysis of the S. habrochaites LA1777 trichome extract also showed
evidence of a single pentamethylmyricetin (RT: 11.18 min) for which the MS/MS product ion
spectrum was obtained (Figure IV-22).
substituent on A-ring (Rule #2).

Again, the fragment at m/z 167 indicates one methyl

Only one isomer is possible with this constraint --

3,3’,4’,5’,7-pentamethylmyricetin.

226

[M+H]

+

+

1,3A

+

Figure IV-22 EPI spectrum of m/z 389 ([M+H] ) for pentamethylmyricetin metabolite in S. habrochaites LA1777 leaf dip extracts,
later annotated as 3,3’,4’,5’,7-pentamethylmyricetin, (RT 11.18 min). Data were acquired using a Qtrap 3200 mass spectrometer
using electrospray positive mode and 40 V collision potential.

227

4.4

Conclusions

Structure elucidation of methylated flavonoid regioisomers has long presented enormous
technical challenges.

In fact, a key 2004 review commented on this subject, “Although

isomeric compounds show a different fragmentation behavior, the exact position of the methoxy
group could not be defined without comparison with standards or NMR analysis.” (Cuyckens
2004).

Since levels of these metabolites are often too low for isolation of milligram quantities,

improved strategies for mass spectrometric structure annotation have been necessary.
By combining the tools of selective enzymatic semisynthesis, HPLC, and tandem mass
spectrometry, selective synthesis was achieved, including regioselective introduction of
deuterated methyl groups onto the myricetin backbone.

Combining these techniques allowed

experiments to determine substrate selectivity of recombinant O-methyltransferase enzymes
whose functions had not previously been established.

In addition, MS/MS spectra of

selectively labeled methylmyricetins allowed contributions of specific groups to observed
fragmentation chemistry to be established.
A series of rules were presented for annotating methylated myricetins, which are:
Rule #1 [238-240]:

Methylation increases chromatographic retention time differentially

based on the position of the methyl group following the order 7>3’(5’) ≈4’>3>5;
Rule #2:

The number of methyl groups on the A-ring is evident from characteristic

fragment ions (m/z 153 means 0; m/z 167 means 1; and m/z 181 means 2 methyl groups);
Rule #3:

Methyl groups exhibit differential lability upon collisional activation, with the
228

rates of methyl loss following the order 3>4’>>3’(5’)>>7;
Rule #4:

Characteristic neutral loss of 16 Da (loss of CH4) is diagnostic for two or more

methyl substituents on the B-ring, and the absence of a methyl group at position 3.
Application of these rules does not require any derivatization which might introduce
unanticipated artifacts or complicate analyses of other metabolites.

This strategy of using

LC/MS analysis coupled to recombinant enzyme assays may be applicable to other methylated
flavonoids or even other enzyme catalyzed chemical isomers study.

229

CHAPTER FIVE
DETERMINATION OF BETAINE METABOLITES AND
DIMETHYLSULFONIOPROPIONATE IN CORAL TISSUES USING LIQUID
CHROMATOGRAPHY/TIME-OF-FLIGHT MASS SPECTROMETRY AND STABLE
ISOTOPE-LABELED INTERNAL STANDARDS

5.1

Introduction

Corals are in decline worldwide [272, 273], and one of the most important causes of this
decline is described as ‘bleaching’—a reduction or loss of populations of zooxanthellae, which
are algae that live in symbiotic relationships within coral tissues.

Bleaching has been attributed

to stresses of elevated water temperature and high light intensity acting in a synergistic manner
[274-276].

Seawater temperatures only a few degrees Celsius above traditional norms have

been suggested to be responsible for coral bleaching [272, 273, 277].

When corals experience

bleaching, the primary insult is often to the complex of proteins known as photosystem II or to
pathways of electron flow downstream from photosystem II in the coral algal symbionts
[274-278].

Urgent attention deserves to be focused on factors the influence in vivo levels of

metabolites with potential to protect photosystem II in corals against stresses of high temperature
and high irradiance. In several organisms, a group of endogenous metabolites called betaines
have been reported to stabilize photosystem II functions [279, 280].

230

Betaines of amino acids are quaternary ammonium metabolites derived from enzymatic
methylation at the amino group nitrogen.

The related analog dimethylsulfoniopropionate

(DMSP) is a tertiary sulfonium metabolite produced by a wide range of higher plants and marine
algae, and also found in (sometimes produced by) bacteria and animals [281-283].

Each of

these metabolites possesses a functional group with a permanent positive charge and an acidic
moiety such as a carboxylic acid that combine to confer zwitterion characteristics at
physiological pH. The most firmly established function of betaines in animals is as osmolytes
[284, 285], a role for which they are well poised because of their compatibility with normal
structures and functions of macromolecules.

In addition, in vascular plants and free-living

algae, glycine betaine was found to play a key role in protecting photosystem II against a number
of stresses, of which high temperature and high irradiance are noteworthy [286-290].

These

effects are sufficiently established that multiple crops (or plants used as models in crop research)
have been genetically engineered to increase betaine expression as a way of enhancing resistance
to photosystem stress [291-296].
From preliminary investigations, accumulation of betaines to millimolar concentrations in
coral tissues has been found.

This finding suggests that corals synthesize high concentrations

of betaines to resist the damaging effects of stressful environments, including those with elevated
temperatures.

In this context, Professor Richard Hill of the Michigan State University Zoology

Department collected coral samples from a field location near Curacao to investigate whether
betaine levels showed correlations with intensity of solar radiation, with depth in water serving

231

as an influence on light intensity.

To achieve this goal, a reliable and accurate analytical

method for quantification of betaines from coral tissues needed to be developed.
One of the challenging aspects of analysis of betaines and DMSP lies in their lack of
useful chromophores, and their chemical structures have permanently charged groups
that preclude gas chromatographic separation unless they are partially demethylated and
converted to volatile derivatives. In the 1980s, analyses of betaines and DMSP relied on
qualitative or semi-quantitative colorimetric tests that employed planar chromatography
and Dragendorff's reagent [297, 298].

Because these methods are limited in their

sensitivity, selectivity and quantitative accuracy, and ability to resolve multiple betaines,
knowledge of the identities and absolute concentrations of betaines in biological tissues
remained elusive. Pyrolytic dealkylation has allowed betaines and DMSP to be detected
as their dealkylated forms using gas chromatography [299-307].

However these

methods suffer from potential non-specificity, particularly if the dealkylated forms are also
present in the native state. The reaction conditions used for dealkylation may convert
other endogenous compounds to trimethylamine and dimethylsulfide after alkaline
treatment.

As a consequence, such methods may overestimate levels of tissue betaines

and DMSP. For direct analysis of betaines or DMSP, chromophoric derivatives were
developed for HPLC/UV detection [308-311]. However, during the use of this approach,
the abundance of numerous other acidic metabolites compromises the selectivity of
detection and analytical accuracy, and laborious extraction and sample preparation
procedures are needed to remove matrix interferences.
232

Similar derivatization schemes

were developed using capillary electrophoresis and UV detection, but are subject to
similar complications [312-315].
The permanent cationic moieties of betaines and DMSP make mass spectrometry an
attractive method for their detection, and the advent of soft ionization techniques provides
a convenient means for determining their molecular masses and for efficient and
mass-selective detection [316, 317].

Betaines in plant extracts were analyzed using

electrospray ionization MS using off-line chromatographic clean-up [318], but the
procedure employed strong basic conditions (6 M NH3) that could cause artifacts in the
measurement of metabolites such as DMSP and β-alanine betaine that are subject to
base-induced decomposition.

While a recent

report described use of

an

ultraperformance liquid chromatography (UPLC)-MS method using a reversed phase
column, and the procedure was applied to measure DMSP in marine algae [319], this
approach relied on derivatization with a hydrophobic fluorescent tag to ensure adequate
chromatographic retention on a reversed phase column and low level detection.
Successful HPLC separations of betaines and DMSP have been reported using ion
exchange (IC) chromatography [309, 320, 321]. However, coupling of IC with MS is
complicated by mobile phase salts employed for IC separations owing to suppression of
electrospray ionization by salts. These earlier achievements in the application of MS
advanced the analysis of polar betaine metabolites and DMSP, but they highlight a need
for improved separations of underivatized polar metabolites.

233

In recent years, numerous stationary phases have been developed for separations of
polar metabolites [322-324]. In an earlier report from the Jones laboratory, the presence
of cationic nitrogen-containing functional groups contributed to greater retention on a
pentafluorophenylpropyl (PFPP) column using water-acetonitrile mixtures as mobile
phase.

Such retention of cationic analytes was more pronounced at high organic

composition, and was attributed to simultaneous electrostatic interactions of analytes with
silanol residues and hydrophobic or polar interactions of analytes with the
pentafluorophenylpropyl group [325].

Mobile phase pH was found to be a critical

parameter for selectivity of separation of basic compounds on PFPP column owing to
pH-dependent shifts in ionization of silanol groups and analytes [325]. Such prior
research offered prospects that betaines and DMSP might be separated using HPLC on
PFPP columns and coupled to a mass spectrometer for metabolite identification and
quantification.
In this chapter, describe a non-targeted quantitative method is described for
measuring multiple betaines and DMSP in extracts of small biological samples without
derivatization using HPLC-TOF-MS and a pentafluorophenylpropyl (PFPP) column for
separation.

This approach involves minimal sample processing, gentle processing conditions,

and provides chromatographic resolution of isomeric and isobaric analytes with stable
isotope-labeled internal standards for accurate quantification. In addition, this chapter
presents an alternative synthetic scheme for preparation of deuterium-labeled DMSP for
use as an internal quantification standard.
234

235

5.2

Materials and methods

5.2.1

Chemicals

Glycine betaine hydrochloride, β-alanine, γ-aminobutyric acid (GABA), iodomethane,
iodomethane-d3, 3-mercaptopropanoic acid, and 5-aminovaleric acid were purchased from
Sigma-Aldrich (St. Louis, MO, U.S.A.).

L-alanine and pivalic acid were obtained from Fluka.

Proline was purchased from Applied Science.

Glycine, ammonium acetate, methanol, 88%

formic acid, and acetonitrile were purchased from VWR Scientific, and D2O was purchased from
Isotec, Inc.

Unless otherwise stated, all other reagents were analytical or HPLC grade and were

used without further purification.

5.2.2

Collection and extraction of coral tissue samples

Samples of whole corals were collected from Curaçao (Netherlands Antilles) and frozen as
soon as possible in liquid nitrogen or dry ice.

Coral samples were stored in dry ice during

shipping and stored in a -80°C freezer afterward until sample extraction.

Pieces of coral (~1-9

2

cm surface area) were removed from the freezer, held at room temperature for 10-15 minutes,
and transferred into a small polyethylene bag. Tissue was first blasted from the coral surface
using a high-pressure stream of ice-cold distilled water from a Water-pik.

The volume of the

water with blasted tissue (collectively termed the blastate) was measured in a graduated cylinder.
236

Typical volume of blastate was about 50 mL.

Then the blastate was immediately mixed 2:1

(v/v) with methanol and acidified to 3% formic acid.
suspend any cellular debris.

The treated blastate was vortexed to

A 596-µL aliquot was promptly removed, and 4.0 µL of a stock

solution containing a mixture of eight deuterated internal standards was added.

Based on

results from preliminary coral tissue extractions, the internal standard was added to yield 60 µM
(36 nmol) for deuterated betaines of glycine and proline and 6.0 µM each (3.6 nmol) for the
remaining six deuterated standards. Extraction was facilitated by sonicating the extracts in an
ice bath for 10 min followed by centrifugation at 8000xg for 5 min and removal of the
supernatant. The pellet was then resuspended in the same extraction solvent by vortexing, and
centrifugation was repeated, to improve extraction efficiency.
and evaporated to dryness using a SpeedVac.

The supernatants were combined

The dried residue was then redissolved in 600 µL

of initial HPLC mobile phase (1% methanol/99% water, adjusted to pH 3.85 with formic acid) in
preparation for LC-MS analysis.

5.2.3

Internal standard synthesis, characterization and quantification

Betaine standards that were not commercially available were synthesized following
procedures described previously [326].

Betaine standards were synthesized by reaction of 1

mmol of amino acid and 16 mmol of CH3I in 20 ml 1:1 (v:v) methanol/water mixture plus 1 g
of KHCO3.

Reaction vessels were protected from light using aluminum foil, and reactions

were allowed to proceed for 16 hours at 25 ˚C.
237

Deuterium labeled betaine standards were

made by incubating the same amino acids with CD3I using the same amounts of reagents and
reaction conditions.

The reaction solvent was then evaporated under a stream of N2 gas and

the residue was partitioned using 5 ml of chloroform with 5 ml of distilled water.
aqueous layer, which contains betaines, was collected.

The upper

Labeled and unlabeled standards of

dimethylsulfoniopropionate (DMSP) were synthesized by the reaction of CH3I with
3-mercaptopropanoic acid in the same fashion but with 0.03 g of NaOH and 0.17 g of KH2PO4
as buffer.

To stabilize the base-sensitive DMSP after solvent partitioning, the aqueous layer

was promptly adjusted to pH~3 by addition of formic acid.

All of the standards were stored at

-20 ˚C until further use.

5.2.4

NMR characterization and quantification

1

Synthetic standards were characterized by H-NMR spectroscopy using a Varian Inova
300 MHz spectrometer.

For synthetic standard quantification, 1.0 mg of pivalic acid was

added to 1.00 ml of synthetic standard solutions and co-dried in a drying oven at room
temperature.

Dried residue was redissolved in 99.9% D2O.

1

A H water suppression pulse

was employed to suppress the water resonance to aid accurate integration of methyl proton
resonances. Peak areas of the methyl group proton resonance from pivalic acid and methyl
group proton resonances from betaines were integrated, and used to calculate betaine
concentrations.

Concentrations of synthetic deuterium-labeled betaines were determined by

LC-MS by comparisons to responses from unlabeled betaines.
238

Quantification was performed

by comparing peak areas of extracted ion chromatograms of known concentrations of unlabeled
betaine and DMSP standards with responses for deuterium-labeled betaine and DMSP solutions.
After establishing concentrations of stock solutions of deuterium-labeled betaines and d6-DMSP,
known amounts, ranging from 1.6 to 34 nmol of deuterated betaines and d6-DMSP, were added
to each coral tissue extract, and isotope dilution was employed to quantify betaines and DMSP
in these extracts.

5.2.5

LC-MS analyses of metabolites

Three Shimadzu LC-20AD HPLC pumps and SIL-5000 autosampler were fitted with a
pentafluorophenylpropyl

HPLC

column (150 × 2.1

mm, 5 µm particles,

Supelco

Discovery-HS-F5) that was coupled to a Waters LCT Premier orthogonal acceleration
time-of-flight mass spectrometer.

The column temperature was maintained at 60 °C.

Separations employed a ternary gradient based on solvents (A) water, adjusted to various pH
values for method optimization ranging from 3.4 to 4.3 by addition of formic acid, (B) methanol,
and (C) 0.15% aqueous formic acid containing 10 mM ammonium acetate.

The total flow rate

was 350 µL/min and the injected volume was 10 µL. Metabolite detection was conducted using
electrospray ionization (ESI) in positive mode.

Nebulization gas flow was 350 L/h at 350 ℃,

and the source temperature was 100 °C. The capillary and the cone potentials were set to 3100
and 10 V, respectively.

The mass spectrometer was operated in V optics mode, tuned to give

mass resolution (FWHM) of 5000.

Mass spectra were acquired at a rate of 1 s/spectrum using
239

dynamic range enhancement, and employed a 0.05 s interscan delay.

Metabolite quantification
+

was performed by calculating extracted ion chromatograms (XICs) for [M+H] ions for each
zwitterionic betaine and DMSP using a mass window of 0.2 Da.

5.2.6

Preparation of calibration and quality control standards

Since no coral tissues were found to be devoid of betaines, it was concluded that
determining limits of detection by spiking a blank coral tissue extract with standard betaines at
low levels was not feasible.

Instead, extracts of field-collected coral tissues were subjected to

addition of unlabeled betaine standards at different levels.

To accomplish this, calibration

standard solutions were prepared by adding NMR-standardized betaine and DMSP standards
into pH 3.85 formic acid solution (the initial mobile phase used for LC separation) to give eight
different concentrations ranging from 0.1-300 µM.

Aliquots of stock solutions of deuterated

internal standards (IS) were added to give concentrations of 60 µM each of glycine betaine-d9
and proline betaine-d6, and 6.0 µM for the remaining deuterated standards.

Low-, medium-

and high-level quality control (QC) spiked solutions were prepared separately to contain 0.3, 20,
and 250 µM of unlabeled standards and the same concentrations of deuterated standards as
described above.

Triplicate Madracis senaria tissue extracts were spiked with unlabeled

standards to reach 20 or 250 µM of unlabeled betaines and DMSP in the extracts to serve as QC
extracts.

240

5.2.7

Evaluation of analytical method performance

Peak integration, generation of calibration curves, and calculations of analyte levels were
performed using the QuanLynx routine within MassLynx v. 4.1 software (Waters).

Calibration

curves were based upon triplicate analyses and seven non-zero metabolite levels, and linear
regression of ratios of XIC peak areas of unlabeled metabolites to XIC peak areas for deuterated
internal standards (Au/AIS) against levels of the unlabeled metabolites.
Limits of detection (LOD) were defined as signal-to-RMS noise of 3 for each XIC, and low
limit of quantification (LLOQ) was defined as the lowest concentration that yielded signal
meeting all of the following criteria: S/N ≥ 5, accuracy between 80-120% of the true value, and
relative standard deviation ≤15%.

Two different measures of LOD were calculated.

Instrumental LODs were determined by adding standards to pH 3.85 aqueous formic acid and
measuring S/N for different metabolite concentrations, and extrapolating to calculate the
concentration for which S/N = 3.

LODs for coral tissue extracts were determined by spiking

extracts with deuterated internal standards, quantifying betaines and DMSP, measuring S/N
ratios for each metabolite peak, and extrapolating to the metabolite concentration corresponding
to S/N = 3.
Effects of the coral extract matrix on method accuracy and precision were evaluated by
preparing and analyzing QC samples consisting of a coral tissue (M. senaria) extract with a
cocktail of metabolites (20 or 250 µM) added both pre- and post-processing and three QC
metabolite cocktail solutions (0.3, 20, and 250 µM for each metabolite) prepared in pH 3.85
241

aqueous solution.

Five technical replicate analyses were performed each day for all QC

solutions over a 3-day period.
To evaluate metabolite recoveries, coral tissue blastate was homogenized by vortexing and
divided into three 500-µL aliquots.

The same amount of unlabeled betaine cocktail (either 20

or 250 µM) was added to one aliquot before the processing steps of extraction, centrifugation,
solvent evaporation, and redissolution.

The same amount of betaine cocktail was added to a

second aliquot after processing, and the third aliquot did not receive any unlabeled betaines.
The cocktail of deuterated internal standards was added to each after processing. Recovery (R)
was calculated for each metabolite using the following equation:

R =1−

Apost / AIS − post − Apre / AIS − pre
Apost / AIS − post − Aendog / AIS −endog

where Apost is the analyte peak area for unlabeled betaine introduced after sample processing,
Apre is analyte peak area spiked before sample processing, Aendog is peak area of endogenous
analyte (no spike), and AIS-post, AIS-pre, and AIS-endog are peak areas of deuterated internal
standard for the corresponding solutions.

Four technical replicates were analyzed by LC/MS

for each solution.

242

5.3

Results and Discussion

5.3.1

Standard synthesis, characterization and quantification

Before

this

report,

the

standard

method

employed

for

synthesis

of

dimethylsulfoniopropionate (DMSP) has involved reaction of dimethylsulfide with acrylic acid
[327].

A drawback of this synthetic approach is posed by the volatility and odor of

dimethylsulfide. Here, a new synthetic method is described that avoids using dimethylsulfide,
and allows for synthesis of deuterated DMSP using the same labeled reagent (CD3I) as was used
for synthesis of deuterated betaines.

Synthesis of DMSP was achieved in this case by

methylation of 3-mercaptopropanoic acid using an excess of methyl iodide.

The synthesis was

carried out by mixing an aqueous solution of 3-mercaptopropanoic acid in pH 7.0 phosphate
buffer with an equal volume of methanol, followed by addition of CH3I.

Reaction conditions

avoid exposing DMSP to alkaline conditions, which promote its rapid dealkylation to
dimethylsulfide and acrylate [328].
The carboxylic group on DMSP or betaines offers a secondary potential site of reaction with
iodomethane to form methyl esters.

To ensure that such undesirable methyl ester formation had

not occurred to a significant extent, characterization of all synthetic products was performed
using both LC-MS and NMR.

A single peak in the extracted ion chromatogram for the

protonated zwitterion was observed for each synthetic standard.

NMR spectra of all synthetic

standards matched expected spectra without showing detectable singlet proton resonances around
243

3.7 ppm corresponding to -OCH3, confirming that methyl esterified products were below limits
of detection (<< 1% of total), even in the case of the DMSP synthesis.
Separation of betaines and DMSP was achieved on the PFPP column using a ternary
gradient based on water and methanol, with the aqueous pH adjusted to exploit differences in
pKa values for the carboxylic groups of the various metabolites.

The more hydrophobic

betaines such as 5-aminovaleric acid betaine required both a higher organic concentration and
addition of 10 mM ammonium acetate to the mobile phase, owing to the strong retention of
hydrophobic cationic analytes at high proportions of methanol.

Using 100% methanol, 30

minutes of isocratic elution failed to elute 5-aminovaleric acid betaine from the column.

Such

retention behavior is consistent with contributions of ion-exchange mechanisms at high organic
contents as observed in an earlier study [325].

Incorporation of ammonium acetate in the initial

mobile phase decreased retention and resolution of early-eluting betaines, therefore this additive
was added later in the gradient.

5.3.2

Extraction of Metabolites from Coral

Several extraction methods were tested on corals before selecting the method described
above.

Overnight extractions of pieces of coral skeleton with tissue were judged to extract

metabolites from deep in the coral skeleton, and thus were presumed to sample residues left from
dead tissues.

Repeated extractions of whole coral pieces with water/methanol mixtures

suggested a slow and persistent release of betaines from the coral skeleton or dead tissues within
244

the skeleton, with levels in the third extraction declining to about 25% of betaine relative to the
first extraction.

To focus analyses on tissues alive at the time of coral harvesting, an approach

was used employing a Water-pik, commonly used in coral studies, for selective removal of coral
tissues from the skeleton surface to minimize contributions from the bulk coral skeleton and to
generate information about betaine levels in tissues alive at the time of sample collection [329].
Various mixtures of methanol and water were tested as potential blasting solutions, with
methanol proving to be incompatible with the Water-pik construction. Cold water blastate was
immediately mixed with methanol and acidified with formic acid to quench metabolism and
minimize metabolite degradation because although most betaines are stable during extraction,
DMSP and β-alanine betaine are susceptible to decomposition to acrylate under basic conditions
[328, 330].
The recoveries for all target betaines and DMSP from coral tissue extracts are summarized
in Table V-1.

Since most betaines exhibit chemical stability and high aqueous solubility, the

double extraction gives high recovery for analytes from 82 -104% .

The minor losses of DMSP

(82-92% recovery) during sample processing do not present problems for quantitative analysis,
but highlight the value of using isotope-labeled DMSP as an internal standard.

245

Table V-1 Mean recoveries of betaines and DMSP in spiked coral tissue extracts (n=4).
Compound

Recovery (%)
Low Spike
(20 µM)
95
90
104
82
98
90
88
102

Glycine betaine
Alanine betaine
β-Alanine betaine
DMSP
GΑΒΑ betaine
Proline betaine
Hydroxyproline betaine
Aminovaleric acid betaine

5.3.3

High Spike
(250 µM)
95
88
91
92
91
94
87
94

LC-MS analyses of betaines and DMSP

Initial efforts to analyze betaines using an assortment of reversed phase HPLC columns (C4
and C18) and mobile phases provided unsatisfactory results owing to limited retention using all
conditions.

While addition of perfluoroheptanoic acid as an ion pairing agent afforded

acceptable retention, it was decided to pursue separations that would not risk long-term
instrument contamination by ion pairing reagents.
As an alternative, both ion-exchange and hydrophobic retention mechanisms offered
by the PFPP stationary phase were exploited, and metabolite separations capitalized on
subtle differences in pKa of the carboxylic acid moiety to achieve separation of metabolite
isomers and isobaric analytes. By adjusting the pH of mobile phase, analytes including
internal standards were resolved, allowing separation of multiple betaines and DMSP

246

metabolites (Figure V-1) in coral samples, their detection using LC-MS, and quantification
by stable isotope dilution.

Betaines and DMSP have groups with a permanent positive

charge that confer increased water solubility and yield poor retention using reversed phase LC
separations.

Though retention of such metabolites can often be achieved using anionic

ion-pairing reagents, such reagents often suppress ionization in mass spectrometry.
Pentafluorophenylpropyl (PFPP) columns offer enhanced retention of cationic analytes using
water-organic mobile phases.

Unlike alkyl silica phases that minimize ionic mechanisms of

retention [331], PFPP columns exploit a combination of ionic and hydrophobic retention
mechanisms that provide suitable retention of betaines and DMSP.

247

Figure V-1

Structures of betaines and DMSP detected in extracts of coral tissues.

248

+

All tested analytes yielded abundant [M+H] ions (based on M corresponding to the neutral
zwitterionic form), but several analytes are isomers or isobaric and require chromatographic
separation.

In addition to the two isomers of alanine betaine, three deuterium-labeled standards

all give signals at a common nominal mass: namely, alanine betaine-d9, β-alanine betaine-d9,
+

DMSP-d6 are all detected as [M+H] at m/z 141.

Subtle differences in charge and solubility

were exploited to resolve these compounds (Figure V-2).

Both alanine betaine and β-alanine

betaine eluted in mobile phases containing < 50% methanol, which corresponds to “reverse
phase”-like conditions using PFPP columns [325].

Systematic adjustment of the aqueous

solvent pH to a point near the reported pKa values, yielded chromatographic resolution that
could not be achieved through hydrophobic interactions alone [331].

It is anticipated that the

greater distance between the carboxylic group and the electron-withdrawing positive charge in
β-alanine betaine and DMSP would make their carboxylic acid groups less acidic than betaines
of alpha amino acids, and therefore have a higher proportion in neutral form within specific pH
conditions. Adjusting pH of the aqueous mobile phase component to lie between estimated
carboxylic acid pKa values of alanine betaine, β-alanine betaine, and DMSP enhanced retention
differences.

Since reference pKa values for all three compounds were not found, the optimal

pH for resolving these metabolites was determined empirically.

Retention of alanine betaine

was unchanged from pH 3.4 to 4.3 (Figure V-2). Carboxylic acid pKa values were estimated to
be ~2 for betaines of α-amino acids, so the specified retention behavior of alanine betaine is
consistent with the carboxylic acid group being largely in anionic form at pH 3.4-4.3.

For

DMSP and β-alanine betaine, the greater distances between the cation and carboxylic acid groups
249

are expected to result in less acidic carboxylic acid groups.

As mobile phase pH increases, an

increase in retention of both β-alanine betaine and DMSP is observed, but retention decreases
upon further pH increase (Figure V-2).

The increased retention is attributed to an increase in

ionization of surface silanol groups, followed by a decrease in retention as the carboxylic group
becomes ionized at pH>4.

The optimized retention is consistent with a hydrophobically

assisted ion-exchange mechanism [332, 333], in which both ion-ion interaction and hydrophobic
interaction are important for the retention.

Mobile phase pH should be regarded as the critical

parameter that can be adjusted to optimize separation of these three compounds. Optimal
separation of these three metabolites was achieved at pH 3.85 for the aqueous component of the
mobile phase. DMSP, alanine betaine and β-alanine betaine were well resolved using this
solvent composition (Figure V-3).

Adjusting column temperatures over the range of 30-60 °C

had negligible effect on resolution of these three analytes.

250

Retention time (min)

3
2.5
2
1.5
1
3.25

3.75

4.25
pH

Figure V-2 Retention times of alanine betaine (□), DMSP (﹡) and β-alanine betaine (▲) on a
pentafluoropropyl HPLC column over pH 3.4 to 4.3 in the aqueous mobile phase component.

Binary gradients based on aqueous formic acid (pH 3.85) and methanol failed to elute
aminovaleric acid betaine, which is more hydrophobic than the other detected metabolites.
Such behavior is attributed to the greater hydrophobicity and the cation-exchange retention,
which is more pronounced at high organic mobile phase content.

To elute this metabolite

without compromising the separation described above, a ternary gradient was implemented to
elute long-chain and short-chain betaines in a single HPLC separation (Table V-2).

This

gradient was based on incorporation of a second aqueous component, after elution of the more
hydrophilic betaines, that incorporated 10 mM ammonium acetate added to water, with
subsequent adjustment to pH 3.85 with formic acid.

The ammonium acetate provides a

mechanism for displacement of the ionic retention of aminovaleric acid betaine by ammonium

251

cation.

This gradient allowed for elution and resolution of all of the betaine analytes using a

15-min gradient.
Detection of betaines and other metabolites employed time-of-flight mass spectrometry with
high mass accuracy and high mass resolution. Narrow mass-window (m/z 0.1) extracted ion
chromatograms filtered out signal from ions with the same nominal mass, and yielded clean
chromatograms for coral extracts that were useful for metabolite quantification.

This approach

aided nontargeted metabolite profiling for corals and other marine organisms, providing accurate
mass measurements for metabolite discovery with simultaneous quantification of targeted
metabolites using stable isotope dilution.

In addition to the anticipated betaines and DMSP

metabolites described above, additional metabolites (Figure V-1) taurine betaine (observed m/z
168.0727, theoretical m/z 168.0689) and trigonelline (observed m/z 138.0555, theoretical m/z
138.0584) were observed in coral extracts and confirmed based on accurate mass measurements
and separate offline MS/MS analyses.

252

Figure V-3 Extracted ion chromatograms showing betaines and DMSP extracted from the
2
tissue covering about 0.05 mm area of skeleton in a specimen of the coral species Madracis
mirabilis collected in the Netherlands Antilles. (A): Peak #1, glycine betaine (m/z 118, tR 1.41
min); peak #2: alanine betaine (m/z 132, tR 1.60 min); peak #3: β-alanine betaine (m/z 132. tR
2.71 min); peak #4: DMSP (m/z 135, tR 2.36 min); peak #5: proline betaine (m/z 144, tR 1.67
min); peak #6: GABA betaine (m/z 146, tR 6.68 min); peak #7: hydroxyproline betaine (m/z 160,
tR 1.28 min) and peak #8: aminovaleric acid betaine (m/z 160, tR 8.57 min); (B): Extracted
ion chromatogram for m/z 141, which shows peaks corresponding to chromatographically
resolved internal standards alanine betaine-d9 (peak a) DMSP-d6 (peak b) and β-alanine
betaine-d9 (peak c).

253

Figure V-3 (cont’d)

254

Table V-2 Solvent composition used in the ternary gradient for elution of long-chain and
short-chain betaines on the pentafluorophenylpropyl silica column. Solvent A: water adjusted
to pH 3.85 with formic acid, solvent B：methanol, solvent C: 0.15% aqueous formic acid + 10
mM ammonium acetate.

5.3.4

Method Performance and Validation

Although there are clear and comprehensive guidelines for validating quantitative methods
in the context of administration of exogenous xenobiotics, there are far more questions than
answers regarding how to validate methods for endogenous metabolites, especially for materials
other than human plasma samples [334].

These challenges are particularly acute for studies of

field-collected marine invertebrates because tissue availability is limited by the remoteness of
sampling locations.

Furthermore, the evidence from this study suggests that endogenous

betaine levels are high in all coral samples collected, and no blank matrix was available that is
free of betaines that could be used for spiking experiments to determine limits of detection.
255

To address a desire to evaluate method performance, calibration curves were constructed
using 7 different concentrations for each betaine and DMSP in aqueous solution instead of
sample matrices.

2

Linear responses (r ≥ 0.99) were obtained over metabolite concentrations

ranging from 100 nM to 300 µM.

These curves were used to estimate analytical limits of

detection (LOD) for standards prepared in aqueous formic acid (pH 3.85) and using spiked coral
tissue extracts (Table V-3).

Based on this approach, it is estimated that LODs for all tested

betaines and DMSP ranged from 6-50 nM in redissolved extracts of coral tissue, which
2

corresponds to endogenous metabolite concentrations measured in <0.05 mm of sample tissue
using the current protocol.

Table V-3 Limits of detection (LODs) of betaines and DMSP determined in aqueous formic
acid and in coral tissue extract matrix.
Compound

Glycine betaine
Alanine betaine
β-Alanine betaine
DMSP
GΑΒΑ betaine
Proline betaine
Hydroxyproline betaine
Aminovaleric acid
betaine

LOD (nM)
(in aqueous
solvent)
30
30
50
8
6
30
8
6

LOD (nM)
(coral tissue
extract)
20
30
40
15
12
30
6
11

Method intra-day and inter-day accuracy and precision are summarized in Table V-4.

All

of the quality control analyses met the criteria that accuracy between 80-120% of the true value,

256

and relative standard deviation ≤15%, demonstrating that the method is robust and reproducible.
The data from the two quality controls made in coral extracts suggests that using the aqueous
calibration curve with stable isotopic labeled standards can accurately and precisely quantify the
sample in coral matrix.
All coral tissue extracts contained one or more abundant betaines, and blank field-collected
coral tissues were not available for evaluating method performance using procedures normally
employed for exogenous xenobiotics, as commonly practiced in a pharmaceutical analysis
context.

To address effects of the coral tissue extract matrix on instrument response and

quantitative performance of the method, the matrix effect (ME) was calculated as

ME =

A post − Aendog
Asspike

where Apost is metabolite peak area for post-process standard addition, Aendog is the peak area of
endogenous metabolite without spike, and Asspike is metabolite peak area for spiked aqueous
solvent without tissue extract.

After addition of a betaine cocktail to a coral tissue extract at

two different metabolite levels (20 and 250 µM), the processed extracts were analyzed and the
effect of matrix on analyte signal was calculated (Table V-5).

The decrease in analyte signal

caused by the matrix was never greater than 11% for any of the metabolites. This surprisingly
small matrix effect is attributed to the high betaine concentrations found in corals relative to
other cellular constituents, which allowed blasting of tissue quantities of a few milligrams or less
into a blastate volume of about 50 mL. This extensive dilution reduces concentrations of matrix
constituents, including betaines and DMSP, by 2-3 orders of magnitude.
257

Substances that might

suppress ionization are also diluted, and have minimal effects on ionization of betaines and
DMSP.

258

Table V-4 Accuracy and precision of quality control samples with addition of multiple betaine
metabolites and DMSP.
Intra-day (n=5)
Compound
Glycine betaine

Alanine betaine

β-Alanine betaine

DMSP

GΑΒΑ betaine

Proline betaine

Nominal
concentration
(µM)
0.3
20
250
20 (coral matrix)
250 (coral matrix)
0.3
20
250
20
250
0.3
20
250
20 (coral matrix)
250 (coral matrix)
0.3
20
250
20 (coral matrix)
250 (coral matrix)
0.3
20
250
20 (coral matrix)
250 (coral matrix)
0.3
20
250
20 (coral matrix)
250 (coral matrix)

Accuracy
(%)
89
93
95
106
103
91
106
98
91
94
116
98
97
93
109
84
89
95
95
97
94
106
102
97
93
91
94
97
108
108

259

Inter-day (n=3)

CV (%)

Accuracy
(%)

CV (%)

6
4
4
6
2
5
6
4
6
9
8
5
5
3
13
3
6
4
5
5
5
7
5
7
10
7
5
4
8
4

92
91
96
103
106
95
106
95
94
97
112
95
96
92
106
90
93
96
92
97
91
112
108
95
91
93
93
99
107
105

5
5
7
6
3
4
3
5
5
6
8
8
5
2
12
6
7
6
5
7
7
4
5
10
12
7
4
6
3
6

Table V-4 (cont’d )
Hydroxyproline
betaine

Aminovaleric acid
betaine

0.3
20
250
20 (coral matrix)
250 (coral matrix)
0.3
20
250
20 (coral matrix)
250 (coral matrix)

90
96
106
98
103
108
96
96
92
102

4
4
7
3
6
9
7
4
9
4

91
93
108
94
99
109
95
94
93
94

6
4
8
4
6
11
6
6
10
5

Table V-5 Experimental determination of matrix effect on determination of individual
betaines and DMSP in coral tissue extracts (n=3).
Compound

Glycine betaine
Alanine betaine
β-Alanine betaine
DMSP
GΑΒΑ betaine
Proline betaine
Hydroxyproline betaine
Aminovaleric acid betaine

QC sample (20 µM)

QC sample (250 µM)

Matrix
Effect
(%)
89
90
93
93
95
93
93
97

Matrix
Effect
(%)
90
92
92
94
94
93
94
96

CV (%)
6
4
3
7
5
4
5
5

CV (%)
8
6
7
8
7
7
5
6

Since subtle changes in mobile phase pH have critical effects on chromatographic resolution
of the two alanine betaine isomers and DMSP, the stability of freshly prepared mobile phase
solvent was also tested. Three replicate 250 ml volumes of solvent A were prepared at pH 3.85
and stored in sealed glass bottles at room temperature.

The bottles were opened to the air every

12 h, and the solution pH was measured using a calibrated pH meter. The rate of pH drift was

260

calculated to be 0.02±0.01 pH units/24 h, and reached pH 3.97±0.03 after 7 days storage.
Since baseline chromatographic resolution of DMSP and the two alanine betaine isomers was
achieved over the pH range of 3.8-4, it is recommend that fresh mobile phase solution A be
prepared at least once a week.

5.4

Betaine in coral reef samples

Betaines were the most abundant metabolites detected within the 50–1,500 Dalton range of
molecular masses from coral extracts.

Tissue molar concentrations of betaines were estimated

by assuming a model tissue thickness of 1 mm.

Taking this approach, 10% of the tissue shows

molar concentration of a individual betaine in mmol per liter of tissue volume and summed tissue
molar concentrations of all betaines estimated in this way were 12 to 204 (mean = 75) mmol per
liter of tissue volume in 110 coral samples.

Studies of other photosynthetic organisms indicate

that tissue molar concentrations of this magnitude are sufficient for stabilization of protein and
membrane functions by betaines [293-296].

Operative subcellular betaine concentrations in

Curacao coral tissues are probably substantially higher than those estimated here because the
estimation assumes a homogenous subcellular distribution, whereas betaines are probably, in fact,
more concentrated in some subcellular regions than others.
Although the biological function of betaine in coral is yet to determined, this study shows
strong evidence that corals substantially modulate tissue concentrations of betaines in response to
environmental conditions.
261

For study of patterns of variation in betaine concentrations in field populations of corals,
Madracis mirabilis was the focal species because its is common at a wide range of depths, where
it grows almost always in fully exposed (unshaded) locations [335], so that (on a given reef) the
depth of a colony is an excellent proxy for the irradiance the colony experiences.

The

concentrations of several betaines: glycine betaine (GlyB), alanine betaine (AlaB), proline
betaine (ProB), and hydroxyproline betaine (HproB) varied inversely with depth (Table V-6).
The second species Madracis pharensis occurs in both exposed and highly shaded locations at
most depths shows concentrations of several betaines averaged significantly higher in exposed
than shaded colonies (Table V-7).

Furthermore, Madracis senaria, which is the most exposed

colonies available at three depths (5, 10, and 20 m) at two times (morning and afternoon) shows
GlyB, AlaB, and HProB were more concentrated in afternoon (15:30-17:00 local time) - than
morning (09:30-11:00 local time) -collected specimens (Table V-8).
The metabolite profiling of betaine shows the correlation of betaine level with irradiance
level, which calls for focused consideration of the full range of roles that betaines play.

The

deeper understanding of betaine function may give opportunities to prevent the decline of coral
worldwide.

262

Table V-6 Concentrations (mean ± SE) in Madracis mirabilis collected from exposed
(unshaded) locations. Values for specific depths are listed only for compounds that showed
statistically significant differences in concentration as a function of depth. AlaB = alanine
betaine, βAlaB = β-alanine betaine, DMSP = dimethylsulfoniopropionate, GlyB = glycine
betaine, HProB = hydroxyproline betaine, ProB = proline betaine, TauB = taurine betaine, Trig
= trigonelline. n = 24, 8, 8, and 8 for the four columns.
−2

concentration (100 × µmol cm )
compound

all colonies
sampled

colonies at
5m

colonies at
10 m

colonies at
20 m

DMSP

6.01 ± 0.23

GlyB

39.8 ± 2.6

52.6 ± 3.2

37.2 ± 1.9

29.5 ± 3.4

AlaB

3.55 ± 0.18

4.05 ± 0.32

3.64 ± 0.27

2.96 ± 0.24

βAlaB

2.23 ± 0.45

ProB

1430 ± 71

1640 ± 112

1520 ± 94

1140 ± 92

HProB

2.73 ± 0.31

3.32 ± 0.46

3.16 ± 0.65

1.71 ± 0.28

TauB

23.7 ± 2.7

Trig

4.58 ± 0.39

5.90 ± 0.51

4.04 ± 0.34

3.79 ± 0.88

263

Table V-7 Concentrations (mean ± SE) in Madracis pharensis at 10 m. Values for exposed
and shaded colonies are listed only for compounds that showed statistically significant
differences in concentration as a function of exposure. n = 19, 10, and 9 for the three columns.
−2

concentration (100 × µmol cm )
compound

exposed colonies

shaded colonies

DMSP

all colonies
sampled
7.05 ± 0.82

GlyB

38.7 ± 3.2

43.7 ± 5.0

33.2 ± 3.2

AlaB

2.99 ± 0.25

3.36 ± 0.32

2.58 ± 0.37

βAlaB

5.96 ± 1.9

ProB

1020 ± 87

1190 ± 127

829 ± 84

HProB

4.56 ± 2.5

TauB

16.4 ± 2.1

Trig

2.70 ± 0.33

264

Table V-8 Concentrations (mean ± SE) in Madracis senaria collected from exposed locations.
Values for colonies collected in morning (09:30-11:00) and afternoon (15:30-17:00) are listed
only for compounds that showed statistically significant differences in concentration between
the times. n = 25, 7, and 18 for the three columns. Pheophytin and protein are in units of
−2
100 × mg cm .
−2

concentration (100 × µmol cm )
compound

morning-collec
ted colonies

afternoon-colle
cted colonies

DMSP

all colonies
sampled
8.84 ± 0.67

GlyB

55.1 ± 4.1

35.9 ± 6.1

62.6 ± 4.0

AlaB

2.27 ± 0.20

1.63 ± 0.26

2.51 ± 0.23

βAlaB

0.601 ± 0.21

0.861 ± 0.083

0.546 ± 0.072

0.983 ± 0.099

TauB

24.0 ± 2.4

14.4 ± 1.3

27.8 ± 2.9

Trig

11.2 ± 0.94

8.50 ± 0.82

12.3 ± 1.2

ProB
HProB

27.2 ± 3.4

In summary, the pentafluorophenylpropyl HPLC separation provides fast and effective
analyses of betaines and DMSP with high accuracy typical of isotope dilution quantitative
methods.

This method avoids analyte derivatization and employs an approach that is simple,

rapid, and easy to conduct.

This is beneficial for analysis of unstable metabolites such as

DMSP, and is suitable for analysis of field-collected tissues that are frozen upon collection.
The metabolite profiling of betaine metabolite by LC/MS shows the most comprehensive
coverage of betaine metabolites in coral reef samples so far.

A question aroused from this study

is the subcellular distribution of betaines and the mechanisms that regulate betaine accumulation.
In an animal–alga symbiosis, the subcellular sites of metabolite accumulation (animal or algal)

265

can differ from sites of metabolite biosynthesis.

To fully understand the betaine function and

mechanism, the localization of betaine relative to the photosynthesis protein need to be studies
therefore subcellular analysis is necessary.

Based on the analyses of coral extracts, this LC/MS

method routinely detected multiple betaines at levels corresponding to extraction of as little as
2

0.05 mm area of tissue.

The detection limits achievable (below 100 nM in extracts) highlight

the potential for using this method to perform spatially resolved profiling analyses of betaine
metabolites in coral tissue if suitable microsampling methods can be developed.

266

CHAPTER SIX
CONCLUDING REMARKS

Compared to the more established approaches typical of modern genomics, transcriptomics,
and proteomics, metabolomics is currently practiced by fewer research teams, but its application
is growing at a rapid pace.

Much of the appeal of metabolomics derives from the close

relationship between cellular metabolite composition and phenotypes.

One important aspect

that one learns when doing metabolomics is an awareness of the variability between individual
samples and between individual portions within a single tissue.

This variation can probably be

attributed to localized variation in environment, as the availability of nutrients and other
regulatory molecules need not be uniform across a tissue.

As an important consequence of this

variability, it follows that metabolome analyses need to be performed on large numbers of tissues,
using organisms grown under various environmental conditions and treatments, and by collecting
subsamples from different cell populations within a tissue or organism.

These factors drive an

emerging principle of metabolite profiling – specificially that such analyses must be performed
faster, ought to cover the metabolome more deeply, and must profile metabolites in smaller local
domains within a tissue.
Identification of novel metabolites remains one of the great bottlenecks in metabolome
analysis, and this limitation is particularly acute when sample sizes are small or when metabolite
levels are low. Chapter II of this dissertation presented the application of a new technology
called multiplexed collision induced dissociation, which generates mass spectra using several
267

collision energies, to profile and identify metabolites from glandular trichomes of single gland
types.

This technology has several layers of merit.

The TOF mass spectrometry platform is

inexpensive compared to other instruments that provide high mass resolution and accuracy, and
the multiplexed CID approach yields such information for both molecular and fragment ions
which are important for identification of novel metabolites.

Multiplexed CID yields

nonselective activation and fragmentation of all ions, and does not require pre-knowledge of
metabolites, nor does it involve data-dependent acquisition, which can compromise comparison
between samples.

This approach also eliminates compound optimization steps, and can extend

the dynamic range for quantifying high level metabolites [117].

With this methodology,

information about thousands of metabolites was captured in a single analysis.

By using this

approach, metabolite profiles were used to make intra- and interspecies comparisons for
trichome metabolites in specific cell types.

Metabolite profiles revealed the differences in

abundances of bioactive compounds in trichomes including acylsugars, glycoalkaloids,
flavonoids, and several novel terpenoid related metabolites.

When combined with transcript

data for trichomes collected from the same tissues, the resulting database has helped identify
candidate genes that may be responsible for regulating biosynthesis of specific metabolites.
Research to confirm these gene functions is continuing in the laboratories of collaborators Robert
Last and Cornelius Barry at Michigan State University, and Eran Pichersky at the University of
Michigan.

One important discovery from metabolite profiles of isolated trichomes is the

recognition that discrepancies between transcript and metabolite data point to the need to
investigate the dynamic behavior of both the transcriptome and metabolome.
268

Since metabolites may be synthesized at confined locations in a tissue such as a single cell
or organelle, bulk analysis of large tissue samples often fails to differentiate biochemical
behavior of cell types.

As a result, metabolite profiles from whole tissues represent multiple

cell types, and the behavior of individual cell types may be lost in the noise from the tissue
metabolome.

In Chapters II and V, sampling methods were described to collect metabolites in

individual trichomes and also from epidermal tissues from field-collected corals.

The resulting

metabolite profiles, generated using mass spectrometric methods, provide some of the first
evidence for the abundance of certain metabolites in specific tissues and cell types, and this
information is expected to serve as the foundation for engineering more productive biological
production of valuable chemicals and as the basis for understanding coral responses to
environmental stresses.
The need for rapid and localized metabolite profiling was addressed in Chapter III, which
presented a scheme capable of high throughput profiling of individual trichomes using contact
printing and laser desorption ionization mass spectrometry from carbon substrates.

These

advances have provided some of the clearest details regarding how individual cells from a single
tissue vary in chemical composition.

The observation that individual glands differ substantially

in composition provides a powerful reason to propel further development of metabolome
imaging mass spectrometry, or ‘micrometabolomics’.

At the outset of this project, nearly all

imaging mass spectrometry reports focused on analyses of animal tissues which were usually
sectioned to yield a flat layer of tissue.

Such an approach was not appropriate for imaging

trichomes on plant tissue surfaces owing to their fragility and scarcity on a single sectioned plane.
269

The contact printing method addresses several technical challenges for mass spectrometric
imaging of trichome metabolites.

By rastering the laser across the printed cells, trichome

chemical imaging is performed across the whole leaf tissue, while preserving spatial information
about metabolite location.

This approach avoids the tissue deformation encountered when fresh

tissues are placed in vacuum, and also eliminates the interfering signals coming from cells below
the trichomes.

The graphite substrate employed for tissue printing gives a clean background

spectrum using laser desorption ionization, and the cleaner spectra simplifies interpretation of
metabolite mass spectra.

The simple sample preparation method avoids damaging the integrity

of metabolite localization which is often encountered when applying matrix to cells.

The

images generated in this research reached 50 µm resolution, which is suitable for single gland
analysis.

Chemical images of trichomes showed heterogeneous metabolite distribution across

the surface of Solanum leaflet tissues.

The factors driving this phenomenon remain unclear, but

these findings may indicate that different glands were at different developmental stages at the
time of sampling.
Some of the challenges encountered in distinguishing isomeric metabolites were addressed
in Chapter IV, which focused on identification of isomeric methylated flavonoids.

By

combining enzymatic semi-synthesis using recombinant enzymes, selective incorporation of
deuterium labels using enzymatic synthesis, and retention times and detection of isomer-selective
fragment ions using MS/MS, an entire suite of methylated flavonoid isomers were distinguished.
The success of this approach illustrates the power of combining multiple instrumental and
biochemical tools to accelerate solutions to the problems of metabolite identification.
270

Due to the diversified physicochemical properties of metabolites, comprehensive
metabolome coverage often requires nontraditional column separations, particularly for polar
hydrophilic metabolites.

In Chapter V, a novel separation scheme was developed for separating

an assortment of betaine metabolites extracted from reef corals collected from marine field
locations.

The separation of several hydrophilic isobaric and isomeric metabolites was

achieved through combination of judicious choice of mobile phase pH and an unusual stationary
phase chemistry.

Accurate profiling of betaines and the related metabolite DMSP, a precursor

of atmospheric dimethylsulfide, was achieved using stable isotope dilution methodology.

One

important finding of this work has been that tissue levels of betaines correlate with light levels
across multiple coral species, and may represent a metabolic adaptation of corals to high light
stress.
Although metabolomics technologies and applications have advanced rapidly over the past
decade, numerous substantial challenges remain.

One major challenge is the lack of reference

compounds to support structure identification, and this remains one of the great bottlenecks for
metabolomics research.

Second, analytical instruments and protocols now generate chemical

information at remarkably fast rates.

For example, between the Michigan State University

Mass Spectrometry Core and the Jones laboratory, approximately 20,000 metabolite profiles are
generated each year, with hundreds to thousands of compounds measured per profile. The
primary goal of metabolome analysis is to answer biological questions, but this can rarely be
accomplished from metabolite profiles alone.

Combining metabolome data with modern tools

of genetics, bioinformatics, and molecular biology promises to pose even greater interpretation
271

and analysis challenges, and building interdisciplinary teams that attack these issues from
multiple directions should be pursued to accelerate biological discoveries.

Although we remain

far away from a holistic understanding of plant metabolism, the technology-driven progress
made in a short time has enlightened the path towards this ultimate goal.

272

REFERENCES

273

REFERENCES

1.

Eisenstein, E., Gilliland, G.L., Herzberg, O., Moult, J., Orban, J., Poljak, R.J.,
Banerjei, L., Richardson, D., and Howard, A.J. (2000). Biological function made
crystal clear - annotation of hypothetical proteins via structural genomics. Current
Opinion in Biotechnology 11, 25-30.

2.

Blattner, F.R., Plunkett, G., Bloch, C.A., Perna, N.T., Burland, V., Riley, M.,
ColladoVides, J., Glasner, J.D., Rode, C.K., Mayhew, G.F., et al. (1997). The
complete genome sequence of Escherichia coli K-12. Science 277, 1453-1462.

3.

Goffeau, A., Barrell, B.G., Bussey, H., Davis, R.W., Dujon, B., Feldmann, H.,
Galibert, F., Hoheisel, J.D., Jacq, C., Johnston, M., et al. (1996). Life with 6000
genes. Science 274, 563-567.

4.

Kaul, S., Koo, H.L., Jenkins, J., Rizzo, M., Rooney, T., Tallon, L.J., Feldblyum, T.,
Nierman, W., Benito, M.I., Lin, X.Y., et al. (2000). Analysis of the genome
sequence of the flowering plant Arabidopsis thaliana. Nature 408, 796-815.

5.

Yu, J., Hu, S.N., Wang, J., Wong, G.K.S., Li, S.G., Liu, B., Deng, Y.J., Dai, L.,
Zhou, Y., Zhang, X.Q., et al. (2002). A draft sequence of the rice genome (Oryza
sativa L. ssp indica). Science 296, 79-92.

6.

Austin, C.P. (2003). The completed human genome: Implications for chemical
biology. Current Opinion in Chemical Biology 7, 511-515.

7.

Wang, Z., Gerstein, M., and Snyder, M. (2009). RNA-Seq: A revolutionary tool
for transcriptomics. Nature Reviews Genetics 10, 57-63.

8.

Gygi, S.P., Rochon, Y., Franza, B.R., and Aebersold, R. (1999). Correlation
between protein and mRNA abundance in yeast. Molecular and Cellular Biology
19, 1720-1730.

9.

Anderson, L., and Seilhamer, J. (1997). A comparison of selected mRNA and
protein abundances in human liver. Electrophoresis 18, 533-537.

10.

Ross, J. (1995). Messenger-rna stability in mammalian-cells. Microbiological
Reviews 59, 423-450.

274

11.

Guhaniyogi, J., and Brewer, G. (2001). Regulation of mRNA stability in
mammalian cells. Gene 265, 11-23.

12.

Sumner, L.W., Mendes, P., and Dixon, R.A. (2003). Plant metabolomics:
Large-scale phytochemistry in the functional genomics era. Phytochemistry 62,
817-836.

13.

Schmidt, C.W. (2004). Metabolomics: What's happening downstream of DNA.
Environmental Health Perspectives 112, A410-A415.

14.

Feist, A.M., Herrgard, M.J., Thiele, I., Reed, J.L., and Palsson, B.O. (2009).
Reconstruction of biochemical networks in microorganisms. Nature Reviews
Microbiology 7, 129-143.

15.

Libourel, I.G.L., and Shachar-Hill, Y. (2008). Metabolic flux analysis in plants:
From intelligent design to rational engineering. Annual Review of Plant Biology
59, 625-650.

16.

Fiehn, O., and Spranger, J. (2003). Use of metabolomics to discover metabolic
patterns associated with human diseases. In Metabolic Profiling: Its Role in
Biomarker Discovery and Gene Function Analysis (Springer), pp. 199-215.

17.

Hall, R.D. (2006). Plant metabolomics: From holistic hope, to hype, to hot topic.
New Phytologist 169, 453-468.

18.

Dixon, R.A., and Strack, D. (2003). Phytochemistry meets genome analysis, and
beyond. Phytochemistry 62, 815-816.

19.

D'Auria, J.C., and Gershenzon, J. (2005). The secondary metabolism of
Arabidopsis thaliana: Growing like a weed. Current Opinion in Plant Biology 8,
308-316.

20.

Matsumoto, T., Wu, J.Z., Kanamori, H., Katayose, Y., Fujisawa, M., Namiki, N.,
Mizuno, H., Yamamoto, K., Antonio, B.A., Baba, T., et al. (2005). The map-based
sequence of the rice genome. Nature 436, 793-800.

21.

Beale, M.H., and Sussman, M.R. (2011). Metabolomics of Arabidopsis thaliana.
In Biology of Plant Metabolomics (Wiley-Blackwell), pp. 157-180.

22.

Gibon, Y., Blaesing, O.E., Hannemann, J., Carillo, P., Hohne, M., Hendriks,
J.H.M., Palacios, N., Cross, J., Selbig, J., and Stitt, M. (2004). A robot-based

275

platform to measure multiple enzyme activities in Arabidopsis using a set of
cycling assays: Comparison of changes of enzyme activities and transcript levels
during diurnal cycles and in prolonged darkness. Plant Cell 16, 3304-3325.
23.

Stewart, D. (2011). Crops and tasty, nutritious food – How can metabolomics help?
In Biology of Plant Metabolomics (Wiley-Blackwell), pp. 181-218.

24.

Giovinazzo, G., D'Amico, L., Paradiso, A., Bollini, R., Sparvoli, F., and DeGara,
L. (2005). Antioxidant metabolite profiles in tomato fruit constitutively
expressing the grapevine stilbene synthase gene. Plant Biotechnology Journal 3,
57-69.

25.

Bino, R.J., de Vos, C.H.R., Lieberman, M., Hall, R.D., Bovy, A., Jonker, H.H.,
Tikunov, Y., Lommen, A., Moco, S., and Levin, I. (2005). The
light-hyperresponsive high pigment-2(dg) mutation of tomato: Alterations in the
fruit metabolome. New Phytologist 166, 427-438.

26.

Davuluri, G.R., van Tuinen, A., Fraser, P.D., Manfredonia, A., Newman, R.,
Burgess, D., Brummell, D.A., King, S.R., Palys, J., Uhlig, J., et al. (2005).
Fruit-specific RNAi-mediated suppression of DET1 enhances carotenoid and
flavonoid content in tomatoes. Nature Biotechnology 23, 890-895.

27.

Broeckling, C.D., Huhman, D.V., Farag, M.A., Smith, J.T., May, G.D., Mendes, P.,
Dixon, R.A., and Sumner, L.W. (2005). Metabolic profiling of Medicago
truncatula cell cultures reveals the effects of biotic and abiotic elicitors on
metabolism. Journal of Experimental Botany 56, 323-336.

28.

Dunn, W.B., and Ellis, D.I. (2005). Metabolomics: Current analytical platforms
and methodologies. Trac-Trends in Analytical Chemistry 24, 285-294.

29.

Lindon, J.C., Holmes, E., and Nicholson, J.K. (2003). So what's the deal with
metabonomics? Metabonomics measures the fingerprint of biochemical
perturbations caused by disease, drugs, and toxins. Analytical Chemistry 75,
384A-391A.

30.

Plumb, R.S., Stumpf, C.L., Gorenstein, M.V., Castro-Perez, J.M., Dear, G.J.,
Anthony, M., Sweatman, B.C., Connor, S.C., and Haselden, J.N. (2002).
Metabonomics: The use of electrospray mass spectrometry coupled to
reversed-phase liquid chromatography shows potential for the screening of rat
urine in drug development. Rapid Communications in Mass Spectrometry 16,
1991-1996.

276

31.

Want, E.J., Cravatt, B.F., and Siuzdak, G. (2005). The expanding role of mass
spectrometry in metabolite profiling and characterization. Chembiochem 6,
1941-1951.

32.

Defernez, M., Gunning, Y.M., Parr, A.J., Shepherd, L.V.T., Davies, H.V., and
Colquhoun, I.J. (2004). NMR and HPLC-UV profiling of potatoes with genetic
modifications to metabolic pathways. Journal of Agricultural and Food Chemistry
52, 6075-6085.

33.

Desbrosses, G.G., Kopka, J., and Udvardi, M.K. (2005). Lotus japonicus
metabolic profiling. Development of gas chromatography-mass spectrometry
resources for the study of plant-microbe interactions. Plant Physiology 137,
1302-1318.

34.

Lenz, E.M., and Wilson, I.D. (2007). Analytical strategies in metabonomics.
Journal of Proteome Research 6, 443-458.

35.

MacNair, J.E., Patel, K.D., and Jorgenson, J.W. (1999). Ultrahigh pressure
reversed-phase capillary liquid chromatography: Isocratic and gradient elution
using columns packed with 1.0-µm particles. Analytical Chemistry 71, 700-708.

36.

Verhoeven H.A., Ricde Vos C.H., Bino R.J., Hall R.D. (2006). Plant
Metabolomics (Heidelberg, Germany: Springer Verlag), pp. 33-48.

37.

Hill, D.W., Kertesz, T.M., Fontaine, D., Friedman, R., and Grant, D.F. (2008).
Mass spectral metabonomics beyond elemental formula: Chemical database
querying by matching experimental with computational fragmentation spectra.
Analytical Chemistry 80, 5574-5582.

38.

Liao, W.T., Draper, W.M., and Perera, S.K. (2008). Identification of unknowns in
atmospheric pressure ionization mass spectrometry using a mass to structure
search engine. Analytical Chemistry 80, 7765-7777.

39.

Bocker, S., and Rasche, F. (2008). Towards de novo identification of metabolites
by analyzing tandem mass spectra. Bioinformatics 24, I49-I55.

40.

Benton, H.P., Wong, D.M., Trauger, S.A., and Siuzdak, G. (2008). XCMS2:
Processing tandem mass spectrometry data for metabolite identification and
structural characterization. Analytical Chemistry 80, 6382-6389.

277

41.

Giavalisco, P., Kohl, K., Hummel, J., Seiwert, B., and Willmitzer, L. (2009). C 13
isotope-labeled metabolomes allowing for improved compound annotation and
relative quantification in liquid chromatography-mass spectrometry-based
metabolomic research. Analytical Chemistry 81, 6546-6551.

42.

Martin, C., Bhatt, K., and Baumann, K. (2001). Shaping in plant cells. Current
Opinion in Plant Biology 4, 540-549.

43.

Kehr, J. (2003). Single cell technology. Current Opinion in Plant Biology 6,
617-621.

44.

Kehr, J. (2001). High resolution spatial analysis of plant systems. Current
Opinion in Plant Biology 4, 197-201.

45.

Moco, S., Capanoglu, E., Tikunov, Y., Bino, R.J., Boyacioglu, D., Hall, R.D.,
Vervoort, J., and De Vos, R.C.H. (2007). Tissue specialization at the metabolite
level is perceived during the development of tomato fruit. Journal of
Experimental Botany 58, 4131-4146.

46.

Fait, A., Hanhineva, K., Beleggia, R., Dai, N., Rogachev, I., Nikiforova, V.J.,
Fernie, A.R., and Aharoni, A. (2008). Reconfiguration of the achene and
receptacle metabolic networks during strawberry fruit development. Plant
Physiology 148, 730-750.

47.

Li, S.H., Schneider, B., and Gershenzon, J. (2007). Microchemical analysis of
laser-microdissected stone cells of Norway spruce by cryogenic nuclear magnetic
resonance spectroscopy. Planta 225, 771-779.

48.

Schneider, B., and Holscher, D. (2007). Laser microdissection and cryogenic
nuclear magnetic resonance spectroscopy: An alliance for cell type-specific
metabolite profiling. Planta 225, 763-770.

49.

Nelson, T., Gandotra, N., and Tausta, S.L. (2008). Plant cell types: Reporting and
sampling with new technologies. Current Opinion in Plant Biology 11, 567-573.

50.

Brandt, S.P. (2005). Microgenomics: Gene expression analysis at the
tissue-specific and single-cell levels. Journal of Experimental Botany 56,
495-505.

51.

Tomos, A.D., and Sharrock, R.A. (2001). Cell sampling and analysis (SiCSA):

278

Metabolites measured at single cell resolution. Journal of Experimental Botany 52,
623-630.
52.

Verscht, J., Tomos, D., and Komor, E. (2006). Sugar concentrations along and
across the Ricinus communis L. hypocotyl measured by single cell sampling
analysis. Planta 224, 1303-1314.

53.

Xu, B.J. (2010). Combining laser capture microdissection and proteomics:
Methodologies and clinical applications. Proteomics Clinical Applications 4,
116-123.

54.

Balestrini, R., and Bonfante, P. (2008). Laser Microdissection (LM): Applications
to plant materials. Plant Biosystems 142, 331-336.

55.

Schutze, K., Niyaz, Y., Stich, M., and Buchstaller, A. (2007). Noncontact laser
microdissection and catapulting for pure sample capture. In Laser Manipulation of
Cells and Tissues (San Diego: Elsevier Academic Press Inc), pp. 649-673.

56.

Angeles, G., Berrio-Sierra, J., Joseleau, J.P., Lorimier, P., Lefebvre, A., and Ruel,
K. (2006). Preparative laser capture microdissection and single-pot cell wall
material preparation: A novel method for tissue-specific analysis. Planta 224,
228-232.

57.

An, G.H., Suh, O.S., Kwon, H.C., Kim, K., and Johnson, E.A. (2000).
Quantification of carotenoids in cells of Phaffia rhodozyma by autofluorescence.
Biotechnology Letters 22, 1031-1034.

58.

Berg, J., Hung, Y.P., and Yellen, G. (2009). A genetically encoded fluorescent
reporter of ATP:ADP ratio. Nature Methods 6, 161-166.

59.

Chang, H.T., and Yeung, E.S. (1995). Determination of catecholamines in single
adrenal-medullary cells by capillary electrophoresis and laser-induced native
fluorescence. Analytical Chemistry 67, 1079-1083.

60.

Fehr, M., Frommer, W.B., and Lalonde, S. (2002). Visualization of maltose uptake
in living yeast cells by fluorescent nanosensors. Proceedings of the National
Academy of Sciences 99, 9846-9851.

61.

Fuller, R.R., Moroz, L.L., Gillette, R., and Sweedler, J.V. (1998). Single neuron
analysis by capillary electrophoresis with fluorescence spectroscopy. Neuron 20,
173-181.

279

62.

Lan, H., and Kennedy, R.T. (1995). Exploring single-cell dynamics using
chemically-modified microelectrodes. Trac-Trends in Analytical Chemistry 14,
158-164.

63.

Cannon, D.M., Winograd, N., and Ewing, A.G. (2000). Quantitative chemical
analysis of single cells. Annual Review of Biophysics and Biomolecular Structure
29, 239-263.

64.

Kozminski, K.D., Gutman, D.A., Davila, V., Sulzer, D., and Ewing, A.G. (1998).
Voltammetric and pharmacological characterization of dopamine release from
single exocytotic events at rat pheochromocytoma (PC12) cells. Analytical
Chemistry 70, 3123-3130.

65.

Zhang, B., Adams, K.L., Luber, S.J., Eves, D.J., Heien, M.L., and Ewing, A.G.
(2008). Spatially and temporally resolved single-cell exocytosis utilizing
individually addressable carbon microelectrode arrays. Analytical Chemistry 80,
1394-1400.

66.

Rubakhin, S.S., Greenough, W.T., and Sweedler, J.V. (2003). Spatial profiling
with MALDI MS: Distribution of neuropeptides within single neurons. Analytical
Chemistry 75, 5374-5380.

67.

Page, J.S., Rubakhin, S.S., and Sweedler, J.V. (2000). Direct cellular assays using
off-line capillary electrophoresis with matrix-assisted laser desorption/ionization
time-of-flight mass spectrometry. Analyst 125, 555-561.

68.

Zimmerman, T.A., Rubakhin, S.S., Romanova, E.V., Tucker, K.R., and Sweedler,
J.V. (2009). MALDI mass spectrometric imaging using the stretched sample
method to reveal neuropeptide distributions in Aplysia nervous tissue. Analytical
Chemistry 81, 9402-9409.

69.

Holscher, D., Shroff, R., Knop, K., Gottschaldt, M., Crecelius, A., Schneider, B.,
Heckel, D.G., Schubert, U.S., and Svatos, A. (2009). Matrix-free UV-laser
desorption/ionization (LDI) mass spectrometric imaging at the single-cell level:
Distribution of secondary metabolites of Arabidopsis thaliana and Hypericum
species. Plant Journal 60, 907-918.

70.

Jun, J.H., Song, Z.H., Liu, Z.J., Nikolau, B.J., Yeung, E.S., and Lee, Y.J. (2010).
High-spatial and high-mass resolution imaging of surface metabolites of
Arabidopsis thaliana by laser desorption-ionization mass spectrometry using
colloidal silver. Analytical Chemistry 82, 3255-3265.

280

71.

Northen, T.R., Yanes, O., Northen, M.T., Marrinucci, D., Uritboonthai, W., Apon,
J., Golledge, S.L., Nordstrom, A., and Siuzdak, G. (2007). Clathrate
nanostructures for mass spectrometry. Nature 449, 1033-U1033.

72.

Shimizu, M., Ojima, N., Ohnishi, H., Shingaki, T., Hirakawa, Y., and Masujima, T.
(2003). Development of the single-cell MALDI-TOF (matrix-assisted laser
desorption/ionization time-of-flight) mass-spectroscopic assay. Analytical
Sciences 19, 49-53.

73.

Mizuno, H., Tsuyama, N., Harada, T., and Masujima, T. (2008). Live single-cell
video-mass spectrometry for cellular and subcellular molecular detection and cell
classification. Journal of Mass Spectrometry 43, 1692-1700.

74.

Shrestha, B., and Vertes, A. (2009). In situ metabolic profiling of single cells by
laser ablation electrospray ionization mass spectrometry. Analytical Chemistry 81,
8265-8271.

75.

Lapainis, T., Rubakhin, S.S., and Sweedler, J.V. (2009). Capillary electrophoresis
with electrospray ionization mass spectrometric detection for single-cell
metabolomics. Analytical Chemistry 81, 5858-5864.

76.

Ostrowski, S.G., Kurczy, M.E., Roddy, T.P., Winograd, N., and Ewing, A.G.
(2007). Secondary ion MS imaging to relatively quantify cholesterol in the
membranes of individual cells from differentially treated populations. Analytical
Chemistry 79, 3554-3560.

77.

Piehowski, P.D., Davey, A.M., Kurczy, M.E., Sheets, E.D., Winograd, N., Ewing,
A.G., and Heien, M.L. (2009). Time-of-flight secondary ion mass spectrometry
imaging of subcellular lipid heterogeneity: Poisson counting and spatial
resolution. Analytical Chemistry 81, 5593-5602.

78.

Tobias, H.J., Pitesky, M.E., Fergenson, D.P., Steele, P.T., Horn, J., Frank, M., and
Gard, E.E. (2006). Following the biochemical and morphological changes of
Bacillus atrophaeus cells during the sporulation process using bioaerosol mass
spectrometry. Journal of Microbiological Methods 67, 56-63.

79.

Koek, M.M., Bakels, F., Engel, W., van den Maagdenberg, A., Ferrari, M.D.,
Coulier, L., and Hankemeier, T. (2009). Metabolic profiling of ultrasmall sample
volumes with GC/MS: From microliter to nanoliter samples. Analytical Chemistry
82, 156-162.

281

80.

Mizuno, H., Tsuyama, N., Date, S., Harada, T., and Masujima, T. (2008). Live
single-cell metabolomics of tryptophan and hHistidine mMetabolites in a rat
basophil leukemia cell. Analytical Sciences 24, 1525-1527.

81.

Fung, E.N., and Yeung, E.S. (1998). Direct analysis of single rat peritoneal mast
cells with laser vaporization ionization mass spectrometry. Analytical Chemistry
70, 3206-3212.

82.

Shimizu, M., Levi-Schaffer, F., Ojima, N., Shingaki, T., and Masujima, T. (2002).
A single-cell matrix-assisted laser desorption/ionization time-of-flight
mass-spectroscopic assay of the cell-maturation process. Analytical Sciences 18,
107-108.

83.

Gholipour, Y., Nonami, H., and Erra-Balsells, R. (2008). Application of pressure
probe and UV-MALDI-TOF MS for direct analysis of plant underivatized
carbohydrates in subpicoliter single-cell cytoplasm extract. Journal of the
American Society for Mass Spectrometry 19, 1841-1848.

84.

Rubakhin, S.S., and Sweedler, J.V. (2008). Quantitative measurements of cell-cell
signaling peptides with single-cell MALDI MS. Analytical Chemistry 80,
7128-7136.

85.

Fletcher, J.S., Rabbani, S., Henderson, A., Blenkinsopp, P., Thompson, S.P.,
Lockyer, N.P., and Vickerman, J.C. (2008). A new dynamic in mass spectral
imaging of single biological cells. Analytical Chemistry 80, 9058-9064.

86.

Berman, E.S.F., Fortson, S.L., Checchi, K.D., Wu, L., Felton, J.S., Wu, K.J.J., and
Kulp, K.S. (2008). Preparation of single cells for imaging/profiling mass
spectrometry. Journal of the American Society for Mass Spectrometry 19,
1230-1236.

87.

Amantonico, A., Urban, P.L., and Zenobi, R. (2009). Facile analysis of
metabolites by capillary electrophoresis coupled to matrix-assisted laser
desorption/ionization mass spectrometry using target plates with polysilazane
nanocoating and grooves. Analyst 134, 1536-1540.

88.

Amantonico, A., Urban, P.L., Oh, J.Y., and Zenobi, R. (2009). Interfacing
microfluidics and laser desorption/ionization mass spectrometry by continuous
deposition for application in single cell analysis. Chimia 63, 185-188.

89.

Mellors, J.S., Jorabchi, K., Smith, L.M., and Ramsey, J.M. (2010). Integrated

282

microfluidic device for automated single cell analysis using electrophoretic
separation and electrospray ionization mass spectrometry. Analytical Chemistry
82, 967-973.
90.

Mellors, J.S., Gorbounov, V., Ramsey, R.S., and Ramsey, J.M. (2008). Fully
integrated glass microfluidic device for performing high-efficiency capillary
electrophoresis and electrospray ionization mass spectrometry. Analytical
Chemistry 80, 6881-6887.

91.

Urban, P.L., Jefimovs, K., Amantonico, A., Fagerer, S.R., Schmid, T., Madler, S.,
Puigmarti-Luis, J., Goedecke, N., and Zenobi, R. High-density micro-arrays for
mass spectrometry. (2010). Lab on a Chip 10, 3206-3209.

92.

Chughtai, K., and Heeren, R.M.A. (2010). Mass spectrometric imaging for
biomedical tissue analysis. Chemical Reviews 110, 3237-3277.

93.

Fletcher, J.S. (2009). Cellular imaging with secondary ion mass spectrometry.
Analyst 134, 2204-2215.

94.

Rubakhin, S.S., and Sweedler, J.V. (2007). Characterizing peptides in individual
mammalian cells using mass spectrometry. Nature Protocols 2, 1987-1997.

95.

Coello, Y., Jones, A.D., Gunaratne, T.C., and Dantus, M. (2010). Atmospheric
pressure femtosecond laser imaging mass spectrometry. Analytical Chemistry 82,
2753-2758.

96.

Larkin, J.C., Brown, M.L., and Schiefelbein, J. (2003). How do cells know what
they want to be when they grow up? Lessons from epidermal patterning in
Arabidopsis. Annual Review of Plant Biology 54, 403-430.

97.

Chen, F., D'Auria, J.C., Tholl, D., Ross, J.R., Gershenzon, J., Noel, J.P., and
Pichersky, E. (2003). An Arabidopsis thaliana gene for methylsalicylate
biosynthesis, identified by a biochemical genomics approach, has a role in
defense. Plant Journal 36, 577-588.

98.

Laue, G., Preston, C.A., and Baldwin, I.T. (2000). Fast track to the trichome:
Induction of N-acyl nornicotines precedes nicotine induction in Nicotiana repanda.
Planta 210, 510-514.

99.

Fordyce, J.A., and Agrawal, A.A. (2001). The role of plant trichomes and
caterpillar group size on growth and defence of the pipevine swallowtail Battus

283

philenor. Journal of Animal Ecology 70, 997-1005.
100.

Zvereva, E.L., Kozlov, M.V., and Niemela, P. (1998). Effects of leaf pubescence
in Salix borealis on host-plant choice and feeding behaviour of the leaf beetle,
Melasoma lapponica. Entomologia Experimentalis et Applicata 89, 297-303.

101.

Mauricio, R., and Rausher, M.D. (1997). Experimental manipulation of putative
selective agents provides evidence for the role of natural enemies in the evolution
of plant defense. Evolution 51, 1435-1444.

102.

Klich, M.G. (2000). Leaf variations in Elaeagnus angustifolia related to
environmental heterogeneity. Environmental and Experimental Botany 44,
171-183.

103.

Schreuder, M.D.J., Brewer, C.A., and Heine, C. (2001). Modelled influences of
non-exchanging trichomes on leaf boundary layers and gas exchange. Journal of
Theoretical Biology 210, 23-32.

104.

Luckwill, L. (1943). The genus Locopersicon, an historical, biolgical, and
taxonomic survey of the wild and cultivated tomatoes (Aberdeen: The University
Press), pp. 1-44.

105.

Rodriguez, A.E., Tingey, W.M., and Mutschler, M.A. (1993). Acylsugars of
lycopersicon-pennellii deter settling and feeding of the green peach aphid
(homoptera, aphididae). Journal of Economic Entomology 86, 34-39.

106.

Oghiakhe, S. (1997). Trichomes and resistance to major insect pests in cowpea,
Vigna unguiculata (L.) Walp.: A review. Discovery and Innovation 9, 173-178.

107.

Medeiros, A.H., and Tingey, W.M. (2006). Glandular trichomes of Solanum
berthaultii and its hybrids with Solanum tuberosum affect nymphal emergence,
development, and survival of Empoasca fabae (Homoptera : Cicadellidae).
Journal of Economic Entomology 99, 1483-1489.

108.

Zhang, X.M., Thacker, R.R., and Snyder, J.C. (2008). Occurrence of
2,3-dihydrofarnesoic acid, a spidermite repellent, in trichome secretions of
Lycopersicon esculentum x L-hirsutum hybrids. Euphytica 162, 1-9.

109.

Besser, K., Harper, A., Welsby, N., Schauvinhold, I., Slocombe, S., Li, Y., Dixon,
R.A., and Broun, P. (2009). Divergent regulation of terpenoid metabolism in the
trichomes of wild and cultivated tomato species. Plant Physiology 149, 499-514.

284

110.

Puterka, G.J., Farone, W., Palmer, T., and Barrington, A. (2003).
Structure-function relationships affecting the insecticidal and miticidal activity of
sugar esters. Journal of Economic Entomology 96, 636-644.

111.

Chao, W.S., Serpe, M.D., Anderson, J.V., Gesch, R.W., and Horvath, D.P. (2006).
Sugars, hormones, and environment affect the dormancy status in underground
adventitious buds of leafy spurge (Euphorbia esula). Weed Science 54, 59-68.

112.

Nonomura, T., Xu, L., Wada, M., Kawamura, S., Miyajima, T., Nishitomi, A.,
Kakutani, K., Takikawa, Y., Matsuda, Y., and Toyoda, H. (2009). Trichome
exudates of Lycopersicon pennellii form a chemical barrier to suppress
leaf-surface germination of Oidium neolycopersici conidia. Plant Science 176,
31-37.

113.

Schilmiller, A.L., Schauvinhold, I., Larson, M., Xu, R., Charbonneau, A.L.,
Schmidt, A., Wilkerson, C., Last, R.L., and Pichersky, E. (2009). Monoterpenes in
the glandular trichomes of tomato are synthesized from a neryl diphosphate
precursor rather than geranyl diphosphate. Proceedings of the National Academy
of Sciences 106, 10865-10870.

114.

Kang, J.H., Shi, F., Jones, A.D., Marks, M.D., and Howe, G.A. (2009). Distortion
of trichome morphology by the hairless mutation of tomato affects leaf surface
chemistry. Journal of Experimental Botany 61, 1053-1064.

115.

Gang, D.R., Wang, J.H., Dudareva, N., Nam, K.H., Simon, J.E., Lewinsohn, E.,
and Pichersky, E. (2001). An investigation of the storage and biosynthesis of
phenylpropenes in sweet basil. Plant Physiology 125, 539-555.

116.

Fridman, E., Wang, J.H., Iijima, Y., Froehlich, J.E., Gang, D.R., Ohlrogge, J., and
Pichersky, E. (2005). Metabolic, genomic, and biochemical analyses of glandular
trichomes from the wild tomato species Lycopersicon hirsutum identify a key
enzyme in the biosynthesis of methylketones. Plant Cell 17, 1252-1267.

117.

Shi, F. (2009). Profiling of specialized metabolites in glandular trichomes of the
genus solanum using liquid chromatography and mass spectrometry. In
Department of Chemistry, Michigan State University, Doctoral Thesis (East
Lansing).

118.

Stewart, A.J., Bozonnet, S., Mullen, W., Jenkins, G.I., Lean, M.E.J., and Crozier,
A. (2000). Occurrence of flavonols in tomatoes and tomato-based products.
Journal of Agricultural and Food Chemistry 48, 2663-2669.

285

119.

Tokusoglu, O., Unal, M.K., and Yildirim, Z. (2003). HPLC-UV and GC-MS
characterization of the flavonol aglycons quercetin, kaempferol, and myricetin in
tomato pastes and other tomato-based products. Acta Chromatographica. 13,
196-207.

120.

Le Gall, G., DuPont, M.S., Mellon, F.A., Davis, A.L., Collins, G.J., Verhoeyen,
M.E., and Colquhoun, I.J. (2003). Characterization and content of flavonoid
glycosides in genetically modified tomato (Lycopersicon esculentum) fruits.
Journal of Agricultural and Food Chemistry 51, 2438-2446.

121.

Le Gall, G., Colquhoun, I.J., Davis, A.L., Collins, G.J., and Verhoeyen, M.E.
(2003). Metabolite profiling of tomato (Lycopersicon esculentum) using H-1
NMR spectroscopy as a tool to detect potential unintended effects following a
genetic modification. Journal of Agricultural and Food Chemistry 51, 2447-2456.

122.

Ferreres, F., Taveira, M., Pereira, D.M., Valentao, P., and Andrade, P.B. Tomato
(Lycopersicon esculentum) seeds: New flavonols and cytotoxic effect. Journal of
Agricultural and Food Chemistry 58, 2854-2861.

123.

Li, J.Y., Oulee, T.M., Raba, R., Amundson, R.G., and Last, R.L. (1993).
Arabidopsis flavonoid mutants are hypersensitive to uv-b irradiation. Plant Cell 5,
171-179.

124. Dixon, R.A., and Paiva, N.L. (1995). Stress-induced phenylpropanoid metabolism.
Plant Cell 7, 1085-1097.
125.

Fobes, J.F., Mudd, J.B., and Marsden, M.P.F. (1985). Epicuticular
lipid-accumulation on the leaves of lycopersicon-pennellii (corr) darcy and
lycopersicon-esculentum mill. Plant Physiology 77, 567-570.

126.

Bonierbale, M.W., Plaisted, R.L., Pineda, O., and Tanksley, S.D. (1994). QTL
analysis of trichome-mediated insect resistance in potato. TAG Theoretical and
Applied Genetics 87, 973-987.

127.

Chortyk, O.T., Kays, S.J., and Teng, Q. (1997). Characterization of insecticidal
sugar esters of Petunia. Journal of Agricultural and Food Chemistry 45, 270-275.

128.

Lawson, D.M., Lunde, C.F., and Mutschler, M.A. (1997). Marker-assisted transfer
of acylsugar-mediated pest resistance from the wild tomato, Lycopersicon
pennellii, to the cultivated tomato, Lycopersicon esculentum. Molecular Breeding
3, 307-317.

286

129.

McKenzie, C.L., Weathersbee, A.A., and Puterka, G.J. (2005). Toxicity of sucrose
octanoate to egg, nymphal, and adult Bemisia tabaci (Hemiptera : Aleyrodidae)
using a novel plant-based bioassay. Journal of Economic Entomology 98,
1242-1247.

130.

Hill, K., and Rhode, O. (1999). Sugar-based surfactants for consumer products
and technical applications. Fett-Lipid 101, 25-33.

131.

Chortyk, O.T., Severson, R.F., Cutler, H.C., and Sisson, V.A. (1993). Antibiotic
activities of sugar esters isolated from selected nicotiana species. Bioscience,
Biotechnology, and Biochemistry 57, 1355-1356.

132.

Coates, R.M., Denissen, J.F., Juvik, J.A., and Babka, B.A. (1988). Identification
of alpha-santalenoic and endo-beta-bergamotenoic acids as moth oviposition
stimulants from wild tomato leaves. Journal of Organic Chemistry 53, 2186-2192.

133.

Frelichowski, J.E., and Juvik, J.A. (2001). Sesquiterpene carboxylic acids from a
wild tomato species affect larval feeding behavior and survival of Helicoverpa zea
and Spodoptera exigua (Lepidoptera:Noctuidae). Journal of Economic
Entomology 94, 1249-1259.

134.

Li, L., Zhao, Y.F., McCaig, B.C., Wingerd, B.A., Wang, J.H., Whalon, M.E.,
Pichersky, E., and Howe, G.A. (2004). The tomato homolog of
Coronatine-insensitive1 is required for the maternal control of seed maturation,
jasmonate-signaled defense responses, and glandular trichome development. Plant
Cell 16, 783-783.

135.

Stagliano, M.C., DeKeyser, J.G., Omiecinski, C.J., and Jones, A.D.
Bioassay-directed fractionation for discovery of bioactive neutral lipids guided by
relative mass defect filtering and multiplexed collision-induced dissociation.
Rapid Communications in Mass Spectrometry 24, 3578-3584.

136.

Peralta, I.E., and Spooner, D.M. (2001). Granule-bound starch synthase (GBSSI)
gene phylogeny of wild tomatoes (Solanum L. section Lycopersicon [Mill.] Wettst.
subsection Lycopersicon). American Journal of Botany 88, 1888-1902.

137.

Kumar, N.K.K., Ullman, D.E., and Cho, J.J. (1995). Frankliniella-occidentalis
(thysanoptera, thripidae) landing and resistance to tomato spotted wilt tospovirus
among lycopersicon accessions with additional comments on thrips tabaci
(thysanoptera, thripidae) and trialeurodes-vaporariorum (homoptera, aleyrodidae).
Environmental Entomology 24, 513-520.

287

138.

Pico, B., Diez, M., and Nuez, F. (1998). Evaluation of whitefly-mediated
inoculation techniques to screen Lycopersicon esculentum and wild relatives for
resistance to tomato yellow leaf curl virus. Euphytica 101, 259-271.

139. Martins Santana, F., Ribeiro, S.D., Moita, A.W., Moreira, D.J., and Giordano, L.D.
(2001). Sources of resistance in Lycopersicon spp. to a bipartite
whitefly-transmitted geminivirus from Brazil. Euphytica 122, 45-51.
140.

McDowell, E.T., Kapteyn, J., Schmidt, A., Li, C., Kang, J.H., Descour, A., Shi, F.,
Larson, M., Schilmiller, A., An, L.L., et al. (2011). Comparative functional
genomic analysis of solanum glandular trichome types. Plant Physiology 155,
524-539.

141.

Fiehn, O. (2002). Metabolomics - the link between genotypes and phenotypes.
Plant Molecular Biology 48, 155-171.

142.

Buchholz, A., Hurlebaus, J., Wandrey, C., and Takors, R. (2002). Metabolomics:
Quantification of intracellular metabolite dynamics. Biomolecular Engineering 19,
5-15.

143.

Nicholson, J.K., Lindon, J.C., and Holmes, E. (1999). 'Metabonomics':
Understanding the metabolic responses of living systems to pathophysiological
stimuli via multivariate statistical analysis of biological NMR spectroscopic data.
Xenobiotica 29, 1181-1189.

144.

Bahar, R., Hartmann, C.H., Rodriguez, K.A., Denny, A.D., Busuttil, R.A., Dolle,
M.E.T., Calder, R.B., Chisholm, G.B., Pollock, B.H., Klein, C.A., et al. (2006).
Increased cell-to-cell variation in gene expression in ageing mouse heart. Nature
441, 1011-1014.

145.

Levsky, J.M., and Singer, R.H. (2003). Gene expression and the myth of the
average cell. Trends in Cell Biology 13, 4-6.

146.

Kaern, M., Elston, T.C., Blake, W.J., and Collins, J.J. (2005). Stochasticity in
gene expression: From theories to phenotypes. Nature Reviews Genetics 6,
451-464.

147.

Karrer, E.E., Lincoln, J.E., Hogenhout, S., Bennett, A.B., Bostock, R.M.,
Martineau, B., Lucas, W.J., Gilchrist, D.G., and Alexander, D. (1995). In-situ
isolation of messenger-rna from individual plant-cells - creation of cell-specific
cdna libraries. Proceedings of the National Academy of Sciences of the United

288

States of America 92, 3814-3818.
148.

Brandt, S., Kloska, S., Altmann, T., and Kehr, J. (2002). Using array hybridization
to monitor gene expression at the single cell level. Journal of Experimental
Botany 53, 2315-2323.

149.

EmmertBuck, M.R., Bonner, R.F., Smith, P.D., Chuaqui, R.F., Zhuang, Z.P.,
Goldstein, S.R., Weiss, R.A., and Liotta, L.A. (1996). Laser capture
microdissection. Science 274, 998-1001.

150.

Katsuragi, T., and Tani, Y. (2001). Single-cell sorting of microorganisms by flow
or slide-based (including laser scanning) cytometry. Acta Biotechnologica 21,
99-115.

151.

Chao, T.C., and Ros, A. (2008). Microfluidic single-cell analysis of intracellular
compounds. Journal of the Royal Society Interface 5, S139-S150.

152.

Nelson, A.R., Allbritton, N.L., and Sims, C.E. (2007). Rapid sampling for
single-cell analysis by capillary electrophoresis. In Laser Manipulation of Cells
and Tissues (San Diego: Elsevier Academic Press Inc), pp. 709-722.

153.

Schulte, A., and Schuhmann, W. (2007). Single-cell microelectrochemistry.
Angewandte Chemie-International Edition 46, 8760-8777.

154.

Yang, Q., Zhang, X.L., Bao, X.H., Lu, H.J., Zhang, W.J., Wu, W.H., Miao, H.N.,
and Jiao, B.H. (2008). Single cell determination of nitric oxide release using
capillary electrophoresis with laser-induced fluorescence detection. Journal of
Chromatography A 1201, 120-127.

155.

Li, L.J., Garden, R.W., and Sweedler, J.V. (2000). Single-cell MALDI: A new tool
for direct peptide profiling. Trends in Biotechnology 18, 151-160.

156.

Amantonico, A., Oh, J.Y., Sobek, J., Heinemann, M., and Zenobi, R. (2008). Mass
spectrometric method for analyzing metabolites in yeast with single cell
sensitivity. Angewandte Chemie-International Edition 47, 5382-5385.

157.

Simmons, T.L., Coates, R.C., Clark, B.R., Engene, N., Gonzalez, D., Esquenazi,
E., Dorrestein, P.C., and Gerwick, W.H. (2008). Biosynthetic origin of natural
products isolated from marine microorganism-invertebrate assemblages.
Proceedings of the National Academy of Sciences of the United States of America
105, 4587-4594.

289

158.

Yang, Y.L., Xu, Y.Q., Straight, P., and Dorrestein, P.C. (2009). Translating
metabolic exchange with imaging mass spectrometry. Nature Chemical Biology 5,
885-887.

159.

Shroff, R., Vergara, F., Muck, A., Svatos, A., and Gershenzon, J. (2008).
Nonuniform distribution of glucosinolates in Arabidopsis thaliana leaves has
important consequences for plant defense. Proceedings of the National Academy
of Sciences 105, 6196-6201.

160.

Knochenmuss, R. (2006). Ion formation mechanisms in UV-MALDI. Analyst 131,
966-986.

161.

Vertes, A., Irinyi, G., and Gijbels, R. (1993). Hydrodynamic model of
matrix-assisted laser-desorption mass-spectrometry. Analytical Chemistry 65,
2389-2393.

162.

Zhigilei, L.V., Leveugle, E., Garrison, B.J., Yingling, Y.G., and Zeifman, M.I.
(2003). Computer simulations of laser ablation of molecular substrates. Chemical
Reviews 103, 321-347.

163.

Ehring, H., Karas, M., and Hillenkamp, F. (1992). Role of photoionization and
photochemistry in ionization processes of organic-molecules and relevance for
matrix-assisted laser desorption ionization mass-spectrometry. Organic Mass
Spectrometry 27, 472-480.

164.

Land, C.M., and Kinsel, G.R. (2001). The mechanism of matrix to analyte proton
transfer in clusters of 2,5-dihydroxybenzoic acid and the tripeptide VPL. Journal
of The American Society for Mass Spectrometry 12, 726-731.

165.

Ludemann, H.C., Redmond, R.W., and Hillenkamp, F. (2002). Singlet-singlet
annihilation in ultraviolet matrix-assisted laser desorption/ionization studied by
fluorescence spectroscopy. Rapid Communications in Mass Spectrometry 16,
1287-1294.

166.

Wang, B.H., Dreisewerd, K., Bahr, U., Karas, M., and Hillenkamp, F. (1993).
Gas-phase cationization and protonation of neutrals generated by matrix-assisted
laser desorption. Journal of The American Society for Mass Spectrometry 4,
393-398.

167.

Karas, M., Gluckmann, M., and Schafer, J. (2000). Ionization in matrix-assisted
laser desorption/ionization: singly charged molecular ions are the lucky survivors.

290

Journal of Mass Spectrometry 35, 1-12.
168.

Kruse, R.A., Li, X.L., Bohn, P.W., and Sweedler, J.V. (2001). Experimental
factors controlling analyte ion generation in laser desorption/ionization mass
spectrometry on porous silicon. Analytical Chemistry 73, 3639-3645.

169.

Nordstrom, A., Apon, J.V., Uritboonthal, W., Go, E.P., and Siuzdak, G. (2006).
Surfactant-enhanced desorption/ionization on silicon mass spectrometry.
Analytical Chemistry 78, 272-278.

170.

Luo, G.H., Chen, Y., Siuzdak, G., and Vertes, A. (2005). Surface modification and
laser pulse length effects on internal energy transfer in DIOS. The Journal of
Physical Chemistry B 109, 24450-24456.

171.

Law, K.P. Laser desorption/ionization mass spectrometry on nanostructured
semiconductor substrates: DIOS (TM) and QuickMass (TM). International
Journal of Mass Spectrometry 290, 72-84.

172.

Law, K.P. Surface-assisted laser desorption/ionization mass spectrometry on
nanostructured silicon substrates prepared by iodine-assisted etching.
International Journal of Mass Spectrometry 290, 47-59.

173.

Alimpiev, S., Grechnikov, A., Sunner, J., Karavanskii, V., Simanovsky, Y., Zhabin,
S., and Nikiforov, S. (2008). On the role of defects and surface chemistry for
surface-assisted laser desorption ionization from silicon. The Journal of Chemical
Physics 128, 19.

174.

Nemes, P., and Vertes, A. (2010). Atmospheric-pressure molecular imaging of
biological tissues and biofilms by laesi mass spectrometry. Journal of Visualized
Experiments, e2097.

175.

Vertes, A., and Shrestha, B. (2010). Direct analysis of single cells by mass
spectrometry at atmospheric pressure. Journal of Visualized Experiments, e2144.

176.

Takats, Z., Wiseman, J.M., Gologan, B., and Cooks, R.G. (2004). Mass
spectrometry sampling under ambient conditions with desorption electrospray
ionization. Science 306, 471-473.

177.

Kertesz, V., and Van Berkel, G.J. (2008). Improved imaging resolution in
desorption electrospray ionization mass spectrometry. Rapid Communications in

291

Mass Spectrometry 22, 2639-2644.
178.

Costa, A.B., and Cooks, R.G. (2008). Simulated splashes: Elucidating the
mechanism of desorption electrospray ionization mass spectrometry. Chemical
Physics Letters 464, 1-8.

179.

Chandra, S., and Morrison, G.H. (1995). Imaging ion and molecular-transport at
subcellular resolution by secondary-ion mass-spectrometry. International Journal
of Mass Spectrometry and Ion Processes 143, 161-176.

180.

Strick, R., Strissel, P.L., Gavrilov, K., and Levi-Setti, R. (2001). Cation-chromatin
binding as shown by ion microscopy is essential for the structural integrity of
chromosomes. Journal of Cell Biology 155, 899-910.

181.

Marks, M.D., Wenger, J.P., Gilding, E., Jilk, R., and Dixon, R.A. (2009).
Transcriptome analysis of arabidopsis wild-type and gl3-sst sim trichomes
identifies four additional genes required for trichome development. Molecular
Plant 2, 803-822.

182.

Breuer, C., Kawamura, A., Ichikawa, T., Tominaga-Wada, R., Wada, T., Kondou,
Y., Muto, S., Matsui, M., and Sugimoto, K. (2009). The trihelix transcription
factor gtl1 regulates ploidy-dependent cell growth in the arabidopsis trichome.
Plant Cell 21, 2307-2322.

183.

Wagner, G.J., Wang, E., and Shepherd, R.W. (2004). New approaches for studying
and exploiting an old protuberance, the plant trichome. Annals of Botany 93,
3-11.

184.

Simmons, A.T., and Gurr, G.M. (2005). Trichomes of Lycopersicon species and
their hybrids: Effects on pests and natural enemies. Agricultural and Forest
Entomology 7, 265-276.

185.

Yerger, E.H., Grazzini, R.A., Hesk, D., Coxfoster, D.L., Craig, R., and Mumma,
R.O. (1992). A rapid method for isolating glandular trichomes. Plant Physiology
99, 1-7.

186.

Wienkoop, S., Zoeller, D., Ebert, B., Simon-Rosin, U., Fisahn, J., Glinski, M., and
Weckwerth, W. (2004). Cell-specific protein profiling in Arabidopsis thaliana
trichomes: Identification of trichome-located proteins involved in sulfur
metabolism and detoxification. Phytochemistry 65, 1641-1649.

292

187.

Zhang, X.G., and Oppenheimer, D.G. (2004). A simple and efficient method for
isolating trichomes for downstream analyses. Plant and Cell Physiology 45,
221-224.

188.

Kaletas, B.K., van der Wiel, I.M., Stauber, J., Dekker, L.J., Guzel, C., Kros, J.M.,
Luider, T.M., and Heeren, R.M.A. (2009). Sample preparation issues for tissue
imaging by imaging MS. Proteomics 9, 2622-2633.

189.

Chaurand, P., Norris, J.L., Cornett, D.S., Mobley, J.A., and Caprioli, R.M. (2006).
New developments in profiling and imaging of proteins from tissue sections by
MALDI mass spectrometry. Journal of Proteome Research 5, 2889-2900.

190.

Murphy, R.C., Hankin, J.A., and Barkley, R.M. (2009). Imaging of lipid species
by MALDI mass spectrometry. Journal of Lipid Research 50, S317-S322.

191.

Varner, J.E., and Ye, Z.H. (1994). Tissue printing. Faseb Journal 8, 378-384.

192.

Jacobsen, J.V., and Knox, R.B. (1973). Cytochemical localization and antigenicity
of alpha-amylase in barley aleurone tissue. Planta 112, 213-224.

193.

Kotecha, S.A., Eley, D.W., and Turner, R.W. (1997). Tissue printed cells from
teleost electrosensory and cerebellar structures. Journal of Comparative
Neurology 386, 277-292.

194.

Gaston, S.M., Soares, M.A., Siddiqui, M.M., Vu, D., Lee, J.M., Goldner, D.L.,
Brice, M.J., Shih, J.C., Upton, M.P., Perides, G., et al. (2005). Tissue-print and
print-phoresis as platform technologies for the molecular analysis of human
surgical specimens: Mapping tumor invasion of the prostate capsule. Nature
Medicine 11, 95-101.

195.

Todd, P.J., Schaaff, T.G., Chaurand, P., and Caprioli, R.M. (2001). Organic ion
imaging of biological tissue with secondary ion mass spectrometry and
matrix-assisted laser desorption/ionization. Journal of Mass Spectrometry 36,
355-369.

196.

Caprioli, R.M., Farmer, T.B., and Gile, J. (1997). Molecular imaging of biological
samples: Localization of peptides and proteins using MALDI-TOF MS. Analytical
Chemistry 69, 4751-4760.

197.

Sjovall, P., Lausmaa, J., Nygren, H., Carlsson, L., and Malmberg, P. (2003).
Imaging of membrane lipids in single cells by imprint-imaging time-of-flight

293

secondary ion mass spectrometry. Analytical Chemistry 75, 3429-3434.
198.

Mullen, A.K., Clench, M.R., Crosland, S., and Sharples, K.R. (2005).
Determination of agrochemical compounds in soya plants by imaging
matrix-assisted laser desorption/ionisation mass spectrometry. Rapid
Communications in Mass Spectrometry 19, 2507-2516.

199.

Prideaux, B., Atkinson, S.J., Carolan, V.A., Morton, J., and Clench, M.R. (2007).
Sample preparation and data interpretation procedures for the examination of
xenobiotic compounds in skin by indirect imaging MALDI-MS. International
Journal of Mass Spectrometry 260, 243-251.

200.

Vidova, V., Novak, P., Strohalm, M., Pol, J., Havlicek, V., and Volny, M. (2010).
Laser desorption-ionization of lipid transfers: Tissue mass spectrometry imaging
without MALDI matrix. Analytical Chemistry 82, 4994-4997.

201.

Zhong, G., and Lin, H. (2007). A binary matrix for background suppression in
MALDI-MS of small molecules. Analytical and Bioanalytical Chemistry 387,
1939-1944.

202.

Jurchen, J.C., Rubakhin, S.S., and Sweedler, J.V. (2005). MALDI-MS imaging of
features smaller than the size of the laser beam. Journal of the American Society
for Mass Spectrometry 16, 1654-1659.

203.

Chen, Y.F., Chen, H.Y., Aleksandrov, A., and Orlando, T.M. (2008). Roles of
water, acidity, and surface morphology in surface-assisted laser
desorption/ionization of amino acids. Journal of Physical Chemistry C 112,
6953-6960.

204.

Harris, P.J.F. (2004). Fullerene-related structure of commercial glassy carbons.
Philosophical Magazine 84, 3159-3167.

205.

Huang, J.P., Yuan, C.H., Shiea, J., and Chen, Y.C. (1999). Rapid screening for
diuretic
doping
agents
in
urine
by
C60-assisted
laser-desorption-ionization-time-of-flight mass spectrometry. Journal of
Analytical Toxicology 23, 337-342.

206. Michalak, L., Fisher, K.J., Alderdice, D.S., Jardine, D.R., and Willett, G.D. (1994).
C60-assisted laser desorption-ionization mass-spectrometry. Organic Mass
Spectrometry 29, 512-515.

294

207.

Hopwood, F.G., Michalak, L., Alderdice, D.S., Fisher, K.J., and Willett, G.D.
(1994). C60-assisted laser desorption/ionization mass-spectrometry in the
analysis of phosphotungstic acid. Rapid Communications in Mass Spectrometry 8,
881-885.

208.

Seeley, E.H., and Caprioli, R.M. (2011). MALDI imaging mass spectrometry of
human tissue: Method challenges and clinical perspectives. Trends in
Biotechnology 29, 136-143.

209.

Rauser, S., Deininger, S.O., Suckau, D., Hofler, H., and Walch, A. Approaching
MALDI molecular imaging for clinical proteomic research: Current state and
fields of application. Expert Review of Proteomics 7, 927-941.

210.

Fernandez, J.A., Ochoa, B., Fresnedo, O., Giralt, M.T., and Rodriguez-Puertas, R.
(2011). Matrix-assisted laser desorption ionization imaging mass spectrometry in
lipidomics. Analytical and Bioanalytical Chemistry 401, 29-51.

211.

Shanta, S.R., Kim, Y., Kim, Y.H., and Kim, K.P. (2011). Application of MALDI
tissue imaging of drugs and metabolites: A new frontier for molecular histology.
Biomolecules & Therapeutics 19, 149-154.

212.

Ifa, D.R., Manicke, N.E., Dill, A.L., and Cooks, G. (2008). Latent fingerprint
chemical imaging by mass spectrometry. Science 321, 805-805.

213.

Pichersky, E., and Lewinsohn, E. (2011). Convergent evolution in plant
specialized metabolism. Annual Review of Plant Biology 62, 549-566.

214.

Middleton, E., and Kandaswami, C. (1993). The impact of plant flavonoids on
mammalian biology: Implications for immunity, inflammation and cancer. In The
Flavonoids: Advances in Research Since 1986 (London: Chapman and Hall), pp
619-652.

215.

Jacobs, M., and Rubery, P.H. (1988). Naturally-occurring auxin transport
regulators. Science 241, 346-349.

216.

Long, S.R. (1989). Rhizobium-legume nodulation - life together in the
underground. Cell 56, 203-214.

217.

Ylstra, B., Touraev, A., Moreno, R.M.B., Stoger, E., Vantunen, A.J., Vicente, O.,
Mol, J.N.M., and Heberlebors, E. (1992). Flavonols stimulate development,
germination, and tube growth of tobacco pollen. Plant Physiology 100, 902-907.

295

218.

Dixon, R.A., Dey, P.M., and Lamb, C.J. (1983). Phytoalexins - enzymology and
molecular-biology. In Advances in Enzymology and Related Areas of Molecular
Biology (John Wiley & Sons, Inc), pp 1-136.

219.

Dakora, F.D., and Phillips, D.A. (1996). Diverse functions of isoflavonoids in
legumes transcend anti-microbial definitions of phytoalexins. Physiological and
Molecular Plant Pathology 49, 1-20.

220.

Cuyckens, F., and Claeys, M. (2004). Mass spectrometry in the structural analysis
of flavonoids. Journal of Mass Spectrometry 39, 1-15.

221.

Ibrahim, R.K., Deluca, V., Khouri, H., Latchinian, L., Brisson, L., and Charest,
P.M. (1987). Enzymology and compartmentation of polymethylated flavonol
glucosides in chrysosplenium-americanum. Phytochemistry 26, 1237-1245.

222.

Cuadra, P., and Harborne, J.B. (1996). Changes in epicuticular flavonoids and
photosynthetic pigments as a plant response to UV-B radiation. Zeitschrift Fur
Naturforschung C-a Journal of Biosciences 51, 671-680.

223.

Kim, B.G., Jung, B.R., Lee, Y., Hur, H.G., Lim, Y., and Ahn, J.H. (2006).
Regiospecific flavonoid 7-0-methylation with Streptomyces avermitilis
O-methyltransferase expressed in Escherichia coli. Journal of Agricultural and
Food Chemistry 54, 823-828.

224.

Kim, B.G., Shin, K.H., Lee, Y., Hur, H.G., Lim, Y., and Ahn, J.H. (2005). Multiple
regiospecific methylations of a flavonoid by plant O-methyltransferases expressed
in E-coli. Biotechnology Letters 27, 1861-1864.

225.

Frick, S., Ounaroon, A., and Kutchan, T.M. (2001). Combinatorial biochemistry
in plants: The case of O-methyltransferases. Phytochemistry 56, 1-4.

226.

Walle, T. (2007). Methoxylated flavones, a superior cancer chemopreventive
flavonoid subclass? Seminars in Cancer Biology 17, 354-362.

227.

de Rijke, E., Out, P., Niessen, W.M.A., Ariese, F., Gooijer, C., and Brinkman,
U.A.T. (2006). Analytical separation and detection methods for flavonoids.
Journal of Chromatography A 1112, 31-63.

228.

Bloor, S.J. (2001). Overview of methods for analysis and identification of
flavonoids. In Flavonoids and Other Polyphenols (San Diego: Academic Press
Inc), pp. 3-14.

296

229.

Stobiecki, M. (2000). Application of mass spectrometry for identification and
structural studies of flavonoid glycosides. Phytochemistry 54, 237-256.

230.

Merken, H.M., and Beecher, G.R. (2000). Measurement of food flavonoids by
high-performance liquid chromatography: A review. Journal of Agricultural and
Food Chemistry 48, 577-599.

231.

Robards, K., and Antolovich, M. (1997). Analytical chemistry of fruit
bioflavonoids - A review. Analyst 122, R11-R34.

232.

March, R., and Brodbelt, J. (2008). Analysis of flavonoids: Tandem mass
spectrometry, computational methods, and NMR. Journal of Mass Spectrometry
43, 1581-1617.

233.

Kim, H., Moon, B.H., Ahn, J.H., and Lim, Y. (2006). Complete NMR signal
assignments of flavonol derivatives. Magnetic Resonance in Chemistry 44,
188-190.

234.

Lambert, M., Staerk, D., Hansen, S.H., Sairafianpour, M., and Jaroszewski, J.W.
(2005). Rapid extract dereplication using HPLC-SPE-NMR: Analysis of
isoflavonoids from Smirnowia iranica. Journal of Natural Products 68,
1500-1509.

235.

Park, Y., Moon, B.H., Lee, E., Hong, S., Lee, S., and Lim, Y. (2008). The H-1 and
C-13 NMR data of 19 methoxyflavonol derivatives. Bulletin of the Korean
Chemical Society 29, 81-84.

236.

Lee, S., Park, Y., Moon, B.H., Lee, E., Hong, S., and Lim, Y. (2008). Substitution
effect of hydroxyl groups on the H-1 and C-13 chemical shifts in
hydroxyflavonols. Bulletin of the Korean Chemical Society 29, 1597-1600.

237.

Markham, K.R. (1989). A reassessment of the data supporting the structures of
blumea-malcolmii flavonols. Phytochemistry 28, 243-244.

238.

Greenham, J., Williams, C., and Harborne, J.B. (1995). Identification of lipophilic
flavonols by a combination of chromatographic and spectral techniques.
Phytochemical Analysis 6, 211-217.

239.

Greenham, J., Harborne, J.B., and Williams, C.A. (2003). Identification of
lipophilic flavones and flavonols by comparative HPLC, TLC and UV spectral
analysis. Phytochemical Analysis 14, 100-118.

297

240.

Stefova, M., Stafilov, T., Kulevanova, S., Stefkov, G., and Bankova, V.S. (2007).
QSRR of flavones: Evaluation of substituent contributions to RP HPLC retention.
Journal of Liquid Chromatography & Related Technologies 30, 1035-1049.

241.

Mabry, T.J., Markham, K.R., and Thomas, M.B. (1970). The Systematic
Identification of Flavonoids (New York: Springer).

242.

Kingston, D.G. (1971). Mass spectrometry of organic compounds .6.
electron-impact spectra of flavonoid compounds. Tetrahedron 27, 2691-2700.

243.

Berahia, T., Gaydou, E.M., Cerrati, C., and Wallet, J.C. (1994).
Mass-spectrometry of polymethoxylated flavones. Journal of Agricultural and
Food Chemistry 42, 1697-1700.

244.

Guidugli, F.H., Ardanaz, C.E., Kavka, J., Garibay, M.E., and Josephnathan, P.
(1986). Further mass-spectral studies of flavones. Organic Mass Spectrometry 21,
117-124.

245.

Hedin, P.A., and Phillips, V.A. (1992). Electron-impact mass-spectral analysis of
flavonoids. Journal of Agricultural and Food Chemistry 40, 607-611.

246.

Justesen, U. (2001). Collision-induced fragmentation of deprotonated
methoxylated flavonoids, obtained by electrospray ionization mass spectrometry.
Journal of Mass Spectrometry 36, 169-178.

247.

Cren-Olive, C., Deprez, S., Lebrun, S., Coddeville, B., and Rolando, C. (2000).
Characterization of methylation site of monomethylflavan-3-ols by liquid
chromatography/electrospray ionization tandem mass spectrometry. Rapid
Communications in Mass Spectrometry 14, 2312-2319.

248.

Ma, Y.L., Van den Heuvel, H., and Claeys, M. (1999). Characterization of
3-methoxyflavones using fast-atom bombardment and collision-induced
dissociation tandem mass spectrometry. Rapid Communications in Mass
Spectrometry 13, 1932-1942.

249.

Wang, K.W., and Zhang, H. (2009). Positive electrospray ionization tandem mass
spectrometry of polymethoxylated flavones. Rapid Communications in Mass
Spectrometry 23, 2107-2111.

250.

Zhou, D.Y., Chen, D.L., Xu, Q., Xue, X.Y., Zhang, F.F., and Liang, X.M. (2007).
Characterization of polymethoxylated flavones in Fructus aurantii by liquid

298

chromatography with atmospheric pressure chemical ionization combined with
tandem mass spectrometry. Journal of Pharmaceutical and Biomedical Analysis 43,
1692-1699.
251.

Wang, D.D., Wang, J., Huang, X.H., Tu, Y., and Ni, K.Y. (2007). Identification of
polymethoxylated flavones from green tangerine peel (Pericarpium Citri
Reticulatae Viride) by chromatographic and spectroscopic techniques. Journal of
Pharmaceutical and Biomedical Analysis 44, 63-69.

252.

Bouktaib, M., Lebrun, S., Atmani, A., and Rolando, C. (2002). Hemisynthesis of
all the O-monomethylated analogues of quercetin including the major metabolites,
through selective protection of phenolic functions. Tetrahedron 58, 10001-10009.

253.

Boers, F., Deng, B.L., Lemiere, G., Lepoivre, J., DeGroot, A., Dommisse, R., and
Vlietinck, A.J. (1997). An improved synthesis of the anti-picornavirus flavone
3-O-methylquercetin. Archiv Der Pharmazie 330, 313-316.

254.

Cren-Olive, C., Lebrun, S., and Rolando, C. (2002). An efficient synthesis of the
four mono methylated isomers of (+)-catechin including the major metabolites
and of some dimethylated and trimethylated analogues through selective
protection of the catechol ring. Journal of the Chemical Society-Perkin
Transactions 1, 821-830.

255.

Jurd, L. (1958). Plant polyphenols. V. selective alkylation of the 7-hydroxyl group
in polyhydroxyflavones. Journal of the American Chemical Society 80,
5531-5536.

256.

Jurd, L. (1962). Selective alkylation of polyphenols .1. use of diphenylmethylene
as a protective grouping for o-dihydroxyflavones. Journal of Organic Chemistry
27, 872-875.

257.

Jurd, L. (1962). Selective alkylation of polyphenols .2. methylation of 7-, 4'-, and
3'-hydroxyl groups in flavonols. Journal of Organic Chemistry 27, 1294-1297.

258.

Picq, M., Prigent, A.F., Chabannes, B., and Pacheco, H. (1984). O-alkylation of
quercetin and the synthesis of 3,7,3',4'-tetra-o-ethyl-5-[h-3]-o-ethylquercetin.
Tetrahedron Letters 25, 2227-2230.

259.

Yoo, H., Kim, S.H., Lee, J., Kim, H.J., Seo, S.H., Chung, B.Y., Jin, C., and Lee,
Y.S. (2005). Synthesis and antioxidant activity of 3-methoxyflavones. Bulletin of
the Korean Chemical Society 26, 2057-2060.

299

260.

Li, H.J., Luan, X.H., and Zhao, Y.M. (2004). Facile synthesis
3-O-methylquercetin. Chinese Journal of Organic Chemistry 24, 1619-1621.

261.

Kim, J., Park, K.S., Lee, C., and Chong, Y. (2007). Synthesis of a complete series
of O-methyl analogues of naringenin and apigenin. Bulletin of the Korean
Chemical Society 28, 2527-2530.

262.

Schmidt, A., Li, C., Shi, F., Jones, A.D., and Pichersky, E. Polymethylated
myricetin in trichomes of the wild tomato species solanum habrochaites and
characterization
of
trichome-specific
3'/5'and
7/4'-myricetin
o-methyltransferases. Plant Physiology 155, 1999-2009.

263.

Crozier, A., Jensen, E., Lean, M.E.J., and McDonald, M.S. (1997). Quantitative
analysis
of flavonoids
by reversed-phase high-performance liquid
chromatography. Journal of Chromatography A 761, 315-321.

264.

Merken, H.M., and Beecher, G.R. (2000). Liquid chromatographic method for the
separation and quantification of prominent flavonoid aglycones. Journal of
Chromatography A 897, 177-184.

265.

Geng, H.M., Zhang, D.Q., Zha, J.P., and Qi, J.L. (2007). Simultaneous HPLC
determination of five flavonoids in Flos Inulae. Chromatographia 66, 271-275.

266.

Moco, S., Bino, R.J., Vorst, O., Verhoeven, H.A., de Groot, J., van Beek, T.A.,
Vervoort, J., and de Vos, C.H.R. (2006). A liquid chromatography-mass
spectrometry-based metabolome database for tomato. Plant Physiology 141,
1205-1218.

267.

Ferreres, F., Taveira, M., Pereira, D.M., Valentao, P., and Andrade, P.B. (2010).
Tomato (Lycopersicon esculentum) seeds: New flavonols and cytotoxic effect.
Journal of Agricultural and Food Chemistry 58, 2854-2861.

268.

Ma, Y.L., Li, Q.M., VandenHeuvel, H., and Claeys, M. (1997). Characterization
of flavone and flavonol aglycones by collision-induced dissociation tandem mass
spectrometry. Rapid Communications in Mass Spectrometry. 11, 1357-1364.

269.

Musialik, M., Kuzmicz, R., Pawlowski, T.S., and Litwinienko, G. (2009). Acidity
of hydroxyl groups: An overlooked influence on antiradical properties of
flavonoids. Journal of Organic Chemistry 74, 2699-2709.

270.

Stevens, J.F., Wollenweber, E., Ivancic, M., Hsu, V.L., Sundberg, S., and Deinzer,

300

of

M.L. (1999). Leaf surface flavonoids of Chrysothamnus. Phytochemistry 51,
771-780.
271.

Reddy, N.P., Reddy, B.A.K., Gunasekar, D., Blond, A., Bodo, B., and Murthy,
M.M. (2007). Flavonoids from Limnophila indica. Phytochemistry 68, 636-639.

272.

Carpenter, K.E., Abrar, M., Aeby, G., Aronson, R.B., Banks, S., Bruckner, A.,
Chiriboga, A., Cortes, J., Delbeek, J.C., DeVantier, L., et al. (2008). One-third of
reef-building corals face elevated extinction risk from climate change and local
impacts. Science 321, 560-563.

273.

Hoegh-Guldberg, O., Mumby, P.J., Hooten, A.J., Steneck, R.S., Greenfield, P.,
Gomez, E., Harvell, C.D., Sale, P.F., Edwards, A.J., Caldeira, K., et al. (2007).
Coral reefs under rapid climate change and ocean acidification. Science 318,
1737-1742.

274.

Warner, M.E., Fitt, W.K., and Schmidt, G.W. (1999). Damage to photosystem II in
symbiotic dinoflagellates: A determinant of coral bleaching. Proceedings of the
National Academy of Sciences 96, 8007-8012.

275.

Fitt, W.K., Brown, B.E., Warner, M.E., and Dunne, R.P. (2001). Coral bleaching:
Interpretation of thermal tolerance limits and thermal thresholds in tropical corals.
Coral Reefs 20, 51-65.

276.

Lesser, M.P., and Farrell, J.H. (2004). Exposure to solar radiation increases
damage to both host tissues and algal symbionts of corals during thermal stress.
Coral Reefs 23, 367-377.

277.

Hill, R., Larkum, A.W.D., Frankart, C., Kuhl, M., and Ralph, P.J. (2004). Loss of
functional Photosystem II reaction centres in zooxanthellae of corals exposed to
bleaching conditions: Using fluorescence rise kinetics. Photosynthesis Research
82, 59-72.

278.

Jones, R.J., and Hoegh-Guldberg, O. (2001). Diurnal changes in the
photochemical efficiency of the symbiotic dinoflagellates (Dinophyceae) of corals:
Photoprotection, photoinactivation and the relationship to coral bleaching. Plant,
Cell and Environment 24, 89-99.

279.

Papageorgiou, G.C., and Murata, N. (1995). The unusually strong stabilizing
effects of glycine betaine on the structure and function of the oxygen-evolving
photosystem-ii complex. Photosynthesis Research 44, 243-252.

301

280.

Schiller, H., and Dau, H. (2000). Preparation protocols for high-activity
Photosystem II membrane particles of green algae and higher plants, pH
dependence of oxygen evolution and comparison of the S-2-state multiline signal
by X-band EPR spectroscopy. Journal of Photochemistry and Photobiology B:
Biology 55, 138-144.

281.

Blunden, G., and Gordon, S.M. (1986). Betaines and their sulphonio analogues in
marine algae. Progress in Phycological Research 4, 39-80.

282.

Csonka, L.N., and Hanson, A.D. (1991). Prokaryotic osmoregulation - genetics
and physiology. Annual Review of Microbiology 45, 569-606.

283.

Rhodes, D., and Hanson, A.D. (1993). Quaternary ammonium and tertiary
sulfonium compounds in higher-plants. Annual Review of Plant Physiology and
Plant Molecular Biology 44, 357-384.

284.

Anthoni, U., Christophersen, C., Hougaard, L., and Nielsen, P.H. (1991).
Quaternary ammonium-compounds in the biosphere - an example of a versatile
adaptive strategy. Comparative Biochemistry and Physiology - Part B:
Biochemistry & Molecular Biology 99, 1-18.

285.

Yancey, P.H. (2005). Organic osmolytes as compatible, metabolic and
counteracting cytoprotectants in high osmolarity and other stresses. Journal of
Experimental Biology 208, 2819-2830.

286.

Yang, G.P., Rhodes, D., and Joly, R.J. (1996). Effects of high temperature on
membrane stability and chlorophyll fluorescence in glycinebetaine-deficient and
glycinebetaine-containing maize lines. Australian Journal of Plant Physiology 23,
437-443.

287. Allakhverdiev, S.I., Hayashi, H., Nishiyama, Y., Ivanov, A.G., Aliev, J.A., Klimov,
V.V., Murata, N., and Carpentier, R. (2003). Glycinebetaine protects the
D1/D2/Cytb559 complex of photosystem II against photo-induced and
heat-induced inactivation. Journal of Plant Physiology 160, 41-49.
288.

Allakhverdiev, S.I., Kreslavski, V.D., Klimov, V.V., Los, D.A., Carpentier, R., and
Mohanty, P. (2008). Heat stress: An overview of molecular responses in
photosynthesis. Photosynthesis Research 98, 541-550.

289.

Klimov, V.V., Allakhverdiev, S.I., Nishiyama, Y., Khorobrykh, A.A., and Murata,
N. (2003). Stabilization of the oxygen-evolving complex of photosystem II by

302

bicarbonate and glycinebetaine in thylakoid and subthylakoid preparations.
Functional Plant Biology 30, 797-803.
290.

Hema, R., Senthil-Kumar, M., Shivakumar, S., Reddy, P.C., and Udayakumar, M.
(2007). Chlamydomonas reinhardtii, a model system for functional validation of
abiotic stress responsive genes. Planta 226, 655-670.

291.

McNeil, S.D., Nuccio, M.L., and Hanson, A.D. (1999). Betaines and related
osmoprotectants. Targets for metabolic engineering of stress resistance. Plant
Physiology 120, 945-949.

292.

Sakamoto, A., and Murata, N. (2002). The role of glycine betaine in the protection
of plants from stress: Clues from transgenic plants. Plant, Cell and Environment
25, 163-171.

293.

Prasad, K., and Saradhi, P.P. (2004). Enhanced tolerance to photoinhibition in
transgenic plants through targeting of glycinebetaine biosynthesis into the
chloroplasts. Plant Science 166, 1197-1212.

294.

Shirasawa, K., Takabe, T., and Kishitani, S. (2006). Accumulation of
glycinebetaine in rice plants that overexpress choline monooxygenase from
spinach and evaluation of their tolerance to abiotic stress. Annals of Botany 98,
565-571.

295.

Yang, X.H., Wen, X.G., Gong, H.M., Lu, Q.T., Yang, Z.P., Tang, Y.L., Liang, Z.,
and Lu, C.M. (2007). Genetic engineering of the biosynthesis of glycinebetaine
enhances thermotolerance of photosystem II in tobacco plants. Planta 225,
719-733.

296.

Chen, T.H.H., and Murata, N. (2008). Glycinebetaine: An effective protectant
against abiotic stress in plants. Trends in Plant Science 13, 499-505.

297.

Blunden, G., Elbarouni, M.M., Gordon, S.M., McLean, W.F.H., and Rogers, D.J.
(1981). Extraction, purification and characterization of dragendorff-positive
compounds from some british marine-algae. Botanica Marina 24, 451-456.

298.

Grieve, C.M., and Grattan, S.R. (1983). Rapid assay for determination of
water-soluble quaternary ammonium-compounds. Plant and Soil 70, 303-307.

299.

Hitz, W.D., and Hanson, A.D. (1980). Determination of glycine betaine by
pyrolysis-gas chromatography in cereals and grasses. Phytochemistry 19,

303

2371-2374.
300.

Howard, A.G., Freeman, C.E., Russell, D.W., Arbab-Zavar, M.H., and
Chamberlain, P. (1998). Flow injection system with flame photometric detection
for
the
measurement
of
the
dimethylsulphide
precursor
beta-dimethylsulphoniopropionate. Analytica Chimica Acta 377, 95-101.

301.

Karsten,
U.,
Wiencke,
C.,
and
Kirst,
G.O.
(1990).
The
beta-dimethylsulphoniopropionate (DMSP) content of macroalgae from antarctica
and southern Chile. Botanica Marina 33, 143-146.

302.

Kiene, R.P., and Slezak, D. (2006). Low dissolved DMSP concentrations in
seawater revealed by small-volume gravity filtration and dialysis sampling.
Limnology and Oceanography-Methods 4, 80-95.

303.

Niki, T., Fujinaga, T., Watanabe, M.F., and Kinoshita, J. (2004). Simple
determination of dimethylsulfide (DMS) and dimethylsulfoniopropionate (DMSP)
using solid-phase microextraction and gas chromatography-mass spectrometry.
Journal of Oceanography 60, 913-917.

304.

Reed,
R.H.
(1983).
Measurement
and
osmotic
significance
of
beta-dimethylsulphoniopropionate in marine macrolagae. Marine Biology Letters
4, 173-181.

305.

Smith, G.C., Clark, T., Knutsen, L., and Barrett, E. (1999). Methodology for
analyzing dimethyl sulfide and dimethyl sulfoniopropionate in seawater using
deuterated internal standards. Analytical Chemistry 71, 5563-5568.

306.

Visscher, P.T., Kiene, R.P., and Taylor, B.F. (1994). Demethylation and cleavage
of dimethylsulfoniopropionate in marine intertidal sediments. Fems Microbiology
Ecology 14, 179-189.

307.

Yang, G.P., Levasseur, M., Michaud, S., and Scarratt, M. (2005). Biogeochemistry
of dimethylsulfide (DMS) and dimethylsulfoniopropionate (DMSP) in the surface
microlayer and subsurface water of the western North Atlantic during Spring.
Marine Chemistry 96, 315-329.

308.

Gorham, J. (1984). Separation of plant betaines and their sulfur analogs by
cation-exchange
high-performance
liquid-chromatography.
Journal
of
Chromatography 287, 345-351.

304

309.

Gorham, J. (1986). Separation and quantitative estimation of betaine esters by
high-performance liquid-chromatography. Journal of Chromatography 361,
301-310.

310.

Gorham, J., and McDonnell, E. (1985). High-performance liquid-chromatographic
method for the separation and estimation of choline, glycinebetaine aldehyde and
related-compounds. Journal of Chromatography 350, 245-254.

311.

Rajakyla, E., and Paloposki, M. (1983). Determination of sugars (and betaine) in
molasses by high-performance liquid-chromatography - comparison of the results
with those obtained by the classical lane eynon method. Journal of
Chromatography 282, 595-602.

312.

Nishimura, N., Zhang, J.H., Abo, M., Okubo, A., and Yamazaki, S. (2001).
Application of capillary electrophoresis to the simultaneous determination of
betaines in plants. Analytical Sciences 17, 103-106.

313.

Zhang, J.H., Nagahama, T., Abo, M., Okubo, A., and Yamazaki, S. (2005).
Capillary electrophoretic analysis of dimethylsulfoniopropionate in sugarcane and
marine algal extracts. Talanta 66, 244-248.

314.

Zhang, J.H., Nishimura, N., Abo, M., Okubo, A., and Yamazaki, S. (2001).
Determination of 3-dimethylsulfoniopropionate in marine algae by low-pH
capillary electrophoresis. Bunseki Kagaku 50, 819-824.

315.

Zhang, J.H., Nishimura, N., Okubo, A., and Yamazaki, S. (2002). Development of
an analytical method for the determination of betaines in higher plants by
capillary electrophoresis at low pH. Phytochemical Analysis 13, 189-194.

316.

Rhodes, D., Rich, P.J., Myers, A.C., Reuter, C.C., and Jamieson, G.C. (1987).
Determination of betaines by fast-atom-bombardment mass-spectrometry identification of glycine betaine deficient genotypes of zea-mays. Plant
Physiology 84, 781-788.

317.

Bonham, C.C., Wood, K.V., Yang, W.J., Nadolskaorczyk, A., Samaras, Y., Gage,
D.A., Poupart, J., Burnet, M., Hanson, A.D., and Rhodes, D. (1995).
Identification of quaternary ammonium and tertiary sulfonium compounds by
plasma desorption mass-spectrometry. Journal of Mass Spectrometry 30,
1187-1194.

318.

Wood, K.V., Bonham, C.C., Miles, D., Rothwell, A.P., Peel, G., Wood, B.C., and

305

Rhodes, D. (2002). Characterization of betaines using electrospray MS/MS.
Phytochemistry 59, 759-765.
319.

Wiesemeier, T., and Pohnert, G. (2007). Direct quantification of
dimethylsulfoniopropionate (DMSP) in marine micro- and macroalgae using
HPLC or UPLC/MS. Journal of Chromatography B-Analytical Technologies in
the Biomedical and Life Sciences 850, 493-498.

320.

Colmer, T.D., Corradini, F., Cawthray, G.R., and Otte, M.L. (2000). Analysis of
dimethylsulphoniopropionate (DMSP), betaines and other organic solutes in plant
tissue extracts using HPLC. Phytochemical Analysis 11, 163-168.

321.

Zhang, J.J., and Zhu, Y. (2007). Determination of betaine, choline and
trimethylamine
in
feed
additive
by
ion-exchange
liquid
chromatography/non-suppressed
conductivity
detection.
Journal
of
Chromatography A 1170, 114-117.

322.

Needham, S.R., and Brown, P.R. (2000). The high performance liquid
chromatography electrospray ionization mass spectrometry analysis of diverse
basic pharmaceuticals on cyanopropyl and pentafluorophenylpropyl stationary
phases. Journal of Pharmaceutical and Biomedical Analysis 23, 597-605.

323.

Needham, S.R., Brown, P.R., Duff, K., and Bell, D. (2000). Optimized stationary
phases for the high-performance liquid chromatography-electrospray ionization
mass spectrometric analysis of basic pharmaceuticals. Journal of Chromatography
A 869, 159-170.

324.

Needham, S.R., Jeanville, P.M., Brown, P.R., and Estape, E.S. (2000).
Performance of a pentafluorophenylpropyl stationary phase for the electrospray
ionization high-performance liquid chromatography-mass spectrometry-mass
spectrometry assay of cocaine and its metabolite ecgonine methyl ester in human
urine. Journal of Chromatography B 748, 77-87.

325.

Bell, D.S., and Jones, A.D. (2005). Solute attributes and molecular interactions
contributing to "U-shape" retention on a fluorinated high-performance liquid
chromatography stationary phase. Journal of Chromatography A 1073, 99-109.

326.

Chen, F.C.M., and Benoiton, N.L. (1976). New method of quaternizing amines
and its use in amino-acid and peptide chemistry. Canadian Journal of
Chemistry/Revue Canadiene de Chimie 54, 3310-3311.

306

327.

Chambers, S.T., Kunin, C.M., Miller, D., and Hamada, A. (1987). Dimethylthetin
can substitute for glycine betaine as an osmoprotectant molecule for
Escherichia-Coli. Journal of Bacteriology 169, 4845-4847.

328.

Dacey, J.W.H., and Blough, N.V. (1987). Hydroxide decomposition of
dimethylsulfoniopropionate to form dimethylsulfide. Geophysical Research
Letters 14, 1246-1249.

329.

Johannes, R.E., and Wiebe, W.J. (1970). Method for determination of coral tissue
biomass and composition. Limnology and Oceanography 15, 822-824.

330.

Bouchereau, A., Duhaze, C., Martin-Tanguy, J., Guegan, J.P., and Larher, F.
(1999). Improved analytical methods for determination of nitrogenous stress
metabolites occurring in Limonium species. Journal of Chromatography A 836,
209-221.

331.

Yamamoto, F.M., and Rokushika, S. (2000). Retention properties of the
fluorinated bonded phase on liquid chromatography of aromatic hydrocarbons.
Journal of Chromatography A 898, 141-151.

332.

Neue, U.D., Phoebe, C.H., Tran, K., Cheng, Y.F., and Lu, Z.L. (2001).
Dependence of reversed-phase retention of ionizable analytes on pH,
concentration of organic solvent and silanol activity. Journal of Chromatography
A 925, 49-67.

333.

Yang, X.Q., Dai, J., and Carr, P.W. (2003). Analysis and critical comparison of the
reversed-phase and ion-exchange contributions to retention on polybutadiene
coated zirconia and octadecyl silane bonded silica phases. Journal of
Chromatography A 996, 13-31.

334.

Houghton, R., Horro, P.C., Ward, I., and Macarthur, R. (2009). Generic approach
to validation of small-molecule LC–MS/MS biomarker assays. Bioanalysis 1,
1365-1374.

335.

Vermeij, M.J.A., and Bak, R.P.M. (2002). How are coral populations structured by
light? Marine light regimes and the distribution of Madracis. Marine Ecology
Progress Series 233, 105-116.

307