 

MSU ‘

LIBRARIES
.—,_.

 

RETURNING MATERIALS:
P1ace in book drop to
remove this checkout from
your record. FINES w111
be charged if book is
returned after the date
stamped be1ow.

 

 

 

 

 

 

 

EVALUATION OF NOVEL AND EXISTING IONIZATION
METHODS FOR COMPLEX MIXTURE ANALYSIS

by

Daniel B. Kassel

Submitted to

Michigan State University

in partial fulﬁllment of the requirements

for the degree of

DOCTOR OF PHILOSOPHY

Department of Chemistry

1988

ABSTRACT

EVALUATION OF NOVEL AND EXISTING IONIZATION
METHODS FOR COMPLEX MIXTURE ANALYSIS

by

Daniel B. Kassel

A number of existing and novel ionization methods are evaluated as to their utility for
enhancing the type and quantity of information that can be derived for the analysis of
mixtures when used in conjunction with conventional mass spectrometric methods.
Mixtures of organic acids isolated from urine are targeted for such analyses. A new
desorption/ionization method, K+ ionization of desorbed species (K+IDS), is shown to
be an attractive method for analyses of these complex mixtures and is useful for
differenu'ating between various disease states and healthy conditions, without the need for
extensive work-up of physiological samples and gas chromatographic analysis. K+IDS
is based on them'rionic emission of alkali materials for the production of gas phase alkali
ions. Adduct ion formation occurs as the species being analyzed is rapidly heated and
desorbed into the gas phase. A dual filament K+IDS probe is described which has
replaced a single-filament design and has led to a 30-40 fold improvement in sensitivity
of the technique. Furthermore, the implementation of this design has resulted in better
quantitative representations of complex mixtures of those compounds considered

marginally volatile. Chemical ionization mass spectrometry (CIMS) using ammonia as

the reagent gas is evaluated as to its capacity for providing quantitative information for the
analysis of complex mixtures of organic acids. This technique is found to be superior to
conventional electron impact (El) ionization for experiments involving the use of stable—
isotope tracers to study various aspects of metabolism. Feasibility experiments are
performed using stable-isotope tracers for a means of differentiating between disease-
states that are clinically indistinguishable. A novel ionization method based on carbon
tetrachloride pretreatment for electron capture negative ion mass spectrometry is
described. The technique involves reactions that take place on ion source surfaces to
generate free radical intermediates that exhibit very high cross-sections for electron
attachment. Lower limits of detection are observed for several classes of compounds
when this ionization method is employed. The production of molecular adduct ions
having unique isotopic patterns allows for unequivocable determinations of molecular
weights of unknowns and adduct ion formation is found not to be competitive with low-
energy fragmentation pathways observed under conventional electron capture-negative

ion conditions.

In memory of my father, Fred L. Kassel, whose
spirit continues to guide me towards excellence

in clinical chemistry and all that I pursue in life.

iv.

WEISS

It has been a long, but truly enjoyable four and half years spent at Michigan State
University in pursuit of my PhD. degree. Joe, Bob, and Mark, you guys have been
truly great friends. How could I ever forget the sunglasses and scarf incident in our
first term of analytical chemistry, late nights at PT O'Malleys, tubing, pre-games with
are famous concoction, and of course, the Froggers! These times I will always
remember, even the unpleasant ones, such as being buried by an avalanche of snow
following my initiation to siding at Crystal Mtn.!

I owe a great deal of thanks to members of the mass spectrometry facility (Brian,
Mike, Mel, and Doug) for all their help and suggestions. Special thanks go to the
members of both the Allison and Wastson groups. The commaraderie that existed in
the laboratory made being in the chemistry building that much more pleasant. Karen
and Kathleen, ”old faithful" is now solely in your hands. Keep on milking it for
everything its worth. I wish you both most successful careers in chemistry. Gary,
what can I say. You've been a real good friend. Remember, Boston is only a short
distance away (short enough even for week-end tripsl). Jim, I'll never forget the trip
to Florida, Dukbo, and the eye-opening experiences. Of course, I can't forget the
kindness of all the graduate ofﬁce personnel! To Val, Kelly, Anna, and Ellen, thanks
for the friendship, conversation, lemon drops, and use of the copier.

My scientific development in the past years has been a result of the guidance and
inspiration of my mentors and colleagues. I give a lot of credit to David Pinkston,
whom I respect tremendously as a person and chemist, and of whom I have tried to

model my graduate studies career. To my mentor John Allison, for whom I have the

utmost repect and admiration as both a person and and research adviser, I cannot thank
enough. I attribute much of my growth as a scientist to your tremendous support and
encouragement, the insightful discussions that we have had, and the conﬁdence you
have have had in me. I will always be grateful for these things, and am hopeful that
our paths will cross in the future. Dr. Watson, you have shown a lot of conﬁdence in
my abilities and for that I am grateful. I have enjoyed greatly my association with you
and have learned much from you, not the least of which, the ability to write scientiﬁc
documents. I appreciate your encouragement, and I hope my association with one of
your mentors will be as fruitful for me as it was for you. Dr. Sweeley, you have been
an incredible inspiration to me and I only hope that I can achieve as much success as
you in bioanalytical chemisu'y. I honestly hope that we can, in the future, continue to
collaborate on projects of mutual interest. I have learned a great deal from you and look
forward to keeping in contact.

Most importantly, I am grateful to my family (Mom, J-Lo, Deb, and Paul) for all
their encouragement, support, understanding, friendship, and love. The relationship
that I have developed with my family over the past years is deeply cherished. It is
ironic that the loss of a loved one is often the necessary energy for creating such strong
bonds between family members. I hope the relationship between my family and its
members continues to grow, and that success and happiness comes in all their

endeavers.

TABLE OF CONTENTS

Page
List of Tables ............................................................................. x.
List of Figures ................................................... , ......................... xi.
List of Schemes .......................................................................... xviii.
Chapter 1: Complex Mixture Analysis .................................................. 1
Introduction .................................................................. 2
Metabolic profiling ........................................................... 3
Chapter 2: Desorption/Ionization (DI) and K+IDS Theory ........................... 14
K+IDS theory ................................................................. 19
Operative mechanism of K+le ............................................ 22
Experimental considerations ................................................. 24
Initial results utilizing K+IDS for mixture analysis ....................... 29

Chapter 3: A Novel DI method, K+IDS, for the Rapid Screening of Urine for Organic

Acidemias ..................................................................... 44
Introduction .................................................................... 44
Experimental .................................................................. 47

Chapter 4:

Chapter 5:

Chapter 6:

Chapter 7:

Results of mixture analysis utilizing "dual-ﬁlament” probe ............. 49

K+IDS for urinary organic acid profiling .................................. 54
Conclusions ........................................................ 68
Extension of the Utility and Applications of K+IDS ..................... 75
Organic synthesis on K+IDS emitter surface .............................. 76
K+IDS analyses of cardiac glycosides ..................................... 86
Utility of K+IDS for peptide mapping ..................................... 91
Future studies utilizing strengths of K+IDS .............................. 110
Final comments of K+IDS ................................................ 114
Chemical Ionization Theory ............................................... 115
Chemical ionization mass spectrometry (CIMS) ......................... 115
Quantitative mixture analysis using CIMS ................................ 130
Results of quantitative metabolic proﬁling analysis using CIMS ...... 136
Experimental .................................................................. 138
Results and Discussion ...................................................... 139
Utilizing Ammonia-CI for Stable—Isotope Tracer Studies ................ 165
Introduction ................................................................... 165
Isotope-ratio and isotope-dilution mass spectrometry ................... 168
Investigation of Cell Metabolism Using Labelled Glucose ............. 175
Introduction .................................................................. 175
Experimental ................................................................. 178

viii.

Results and Discussion ..................................................... 180

Conclusions .................................................................. 199
Chapter 8: ECNI—MS with CCl4 ”pretreatment” ...................................... 203
Introduction .................................................................. 203
A novel ionization method for ECNI mass spectrometry ............... 206
Experimental .................................................................. 207
Results and discussion on simple mixtures ............................... 209
Final Remarks ............................................................................. 213
Appendix 1: ................................................................................ 216

"Utility and Limitations of Electron Ionization and Chemical
Ionization for the Determination of Position and Extent of

Labelling for lsO—and 13c- Containing Permethylated Alditols by
Gas Chomatography Mass Spectrometry” ........................... , 216

"Urinary Metabolites of L-Threonine in Type 1 Diabetes
Determined by Combined Gas Chromatography Chemical
Ionization Mass Spectrometry" .......................................... 222

"Utility of Ion Source Pretreatment with Chlorine-Containing
Compounds for Enhanced Performance in Gas Chromatography
Negative Ionization Mass Spectrometry" ............................... 228

Appendix 2: Table of molecular weights of unknowns determined from ammonia

CI and electron impact ionization ......................................... 256

References ................................................................................. 260

Table 1.

Table 2.

Table 3.

Table 4.
Table 5.

Table 6.

Table 7.

Table 8.

Table 9:

Table 10:

LIST OF TABLES

Twenty-ﬁve most abundant organic acids in MSUD urine identiﬁed from
GC and GC-MS analyses ........................................................ 55

Major organic acid metabolites identiﬁed from GC—MS data for (a) lactic
acidemia, (b) isovaleric acidemia, (c) healthy; and (d) methylmalonic
acidemia urines .................................................................... 62

Comparison of relative intensities from Na+IDS analysis and peak areas
from GC analysis for the ten most abundant organic acids in methylmalonic
urine ................................................................................ 73

Fragments produce from protonated linear peptides (1) upon undergoing
collisionally activated dissociation (CAD)-MS-MS ........................... 95

List of structures, formulae, molecular weights and retention indeces for
components of synthetic organic acid mixture .............................. 151

Comparison between actual concentrations of several two and three
component organic acid mixtures with integrated peak areas obtained from
GC-(NH3-CD-MS and GC-EI-MS analyses ............................... 154

Comparison of integrated areas from El and NH3-CI analyses on an eleven

component organic acid mixture. Actual concentrations are shown in the
lower half of this table .......................................................... 157

Identiﬁed metabolites of both cell ﬁbroblasts and cell medium. Peak
number 43-glutamic acid. Peak number 47-unknown (MW=437), believed
to be a ﬁve-carbon aminoribose sugar ...................................... 182

13C-Enrichment observed into various metabolite pools from ﬁbroblast
cells and cell medium for (a) Cytochrome oxidase, (b) Pyruvate
Dehydrogenase, and (c) control lines ........................................ 187

Fragment ions observed and number of trimethylsilyl groups associated
with each fragment from the electron impact ionization mass spectrum of the
d9-TMS and do-TMS derivatives of Unknown (MW =437) .............. 194

X.

Figure 1.

Figure 2.

Figure 3.
Figure 4.

Figure 5.

Figure 6.

Figure 7.
Figure 8.

Figure 9.

Figure 10.

Figure 11.

Figure 12.

Figure 13.

LIST OF FIGURES

Two-dimensional graphical representation of the metabolic status of an

individual from paper chromatographic analysis ................................ 6
Procedures for carrying out the isolation and derivatization of urinary
organic acids for analysis by GC-MS ........................................... 10
70 eV EI-MS mass spectrum of 2-hydroxyisovaleric acid (di-TMS) ....... 12
Thermodynamic considerations for alkali ion emission ....................... 20

Pr0posed mechanism for adduct ion formation of desorbing species,
ABCD, from alkali aluminosilicate surface ..................................... 23

Modiﬁed direct insertion probe (DIP) for use as a thermionic emitter for

K+IDS analyses .......................................................... . ........ 25
Schematic diagram of emission current power supply ...................... 27
"Typical" K+IDS tuning ﬁle depicting K+ ion abundance .................. 28

Na+IDS mass spectrum of ﬁve common organic acids (underivatized) 32

K+IDS averaged mass spectrum ofa 5:1:1:l mixture of C-12, C-14, C-15,
and C-16 free fatty acids; operating current, I, = 3.0 A ..................... 34

K+IDS mass spectrum of same fatty acid mixture obtained at I = 4.0 A.. 36

Analyte desorption and alkali ion emission proﬁles vs. time utilizing a
single-ﬁlament K+IDS probe ................................................... 37

Selected ion monitoring (SIM) comparison of methyl stearate (400 ng)
using (a) single-ﬁlament and (b) dual-ﬁlament K+IDS probe ............... 4o

Figure 14.

Figure 15.

Figure 16.

Figure 17.

Figure 18.

Figure 19.

Figure 20.

Figure 21.

Figure 22.

Figure 23.

Figure 24.

Figure 25.

Detection limits for methyl stearate by K+IDS using dual-ﬁlament probe..4l

Analyte desorption and alkali ion emission profiles vs. time utilizing a
dual-filament K+IDS probe ..................................................... 43

Modiﬁed dual-ﬁlament K+IDS Probe (P). Filament 1 (F1) is electrically
heated using resistive wire (W) and is the alkali ion emitter. Filament 2 (F2)
is the analyte ﬁlament and solely experiences radiative heating ............ 48

Total ion current (TIC) proﬁle (upper insert) and K+IDS mass spectrum of
myristic acid (CH3(CH2)12C O 2 H) .............................................. 5 1

Na+IDS mass spectrum of an equirnolar mixture of C-12, C-13, 014, C-
16, and C-17 free fatty acids ................................................... 53

Na+IDS averaged mass spectrum of a complex mixture of organic acids,
fatty acids, and steroids analyzed simultaneously ............................ 55

Simulated Na+IDS molecular weight proﬁle of the 25 most abundant
organic acids present in urine of a patient with Maple Symp Urine Disease
(MSUD) ........................................................................... 58

Experimentally obtained Na+IDS averaged mass spectra of organic acids
from same patient with MSUD .................................................. 59

Averaged Na+IDS mass spectra of organic acids from (a) healthy,
(b) lactic acidosis, and (c) isovaleric acidemia urines ........................ 61

TIC proﬁle of the trimethylsilyl derivatives of organic acids isolated from
isovaleric acidemia urine .......................................................... 64

Simulated Na+ IDS molecular weight proﬁle of the 25 most abundant
organic acids present in urine of patient with isovaleric acidemia ........... 65

Averaged Na+IDS mass spectra of urinary organic acids isolated from
patient with methylmalonic acidemia ............................................ 67

xii.

Figure 26.

Figure 27.

Figure 28.

Figure 29.

Figure 30.

Figure 31.

Figure 32.

Figure 33.

Figure 34.

Figure 35.

Figure 36.

Figure 37.

Figure 38.

TIC proﬁle of trimethylsilyl derivatives of organic acids isolated from a
second methylmalonic urine ...................................................... 70

Averaged Na+IDS mass spectra of the underivatized organic acids from the
same methylmalonic acidemia urine ............................................. 71

K+IDS analyte TIC desorption proﬁles of C-12, -13, -l4, -16, -17 fatty

acids generated by K+1DS from a ”contaminated" ﬁlament surface. Peak
l-free fatty acids; peak 2- fatty acid salts ...................................... 78

Averaged K+IDS mass spectra of the fatty acid mixture from (a) scan 19 of
analyte desortption proﬁle 1 and (b) scan 29 of analyte desortption proﬁle
2 .................................................................................... 79

K+IDS TIC analyte desorption proﬁles for stearic acid (10 ug) obtained at
(a) I: 3.1 A, (b) 1= 3.55 A, and (c) I = 3.8 A ............................... 82

Three-dimensional mass spectra of the two analyte desorption proﬁles for
stearic acid. Mass axis from 280-380 11; scan axis = 35-80 ................. 83

Structure and assignment of major ions from thermal decomposition
fragments of digitonin ............................................................ 87

K+IDS mass spectrum of digitonin ............................................ 88

K+IDS mass spectrum of (a) digoxin and (b) assignment of ions formed
from thermal decompostion fragments..: ....................................... 91

Fast atom bombardment (FAB) mass spectrum of leucine-enkephalin in
glycerol ............................................................................ 93

FAB MS-MS collisionally activated dissociation (CAD) mass spectrum of
leucine-enkephalin ................................................................. 97

Structure and possible thermal decomposition fragments for leucine-

enkephalin arising from a Na+IDS analysis .................................... 99

Na+IDS mass spectrum of Leucine-Enkephalin ............................. 102

xiii.

Figure 39.

Figure 40.

Figure 41.

Figure 42.

Figure 43.

Figure 44.

Figure 45.

Figure 46.
Figure 47.

Figure 48.

Figure 49.

Figure 50.

Figure 51.

Figure 52.

Structure and possible thermal decomposition fragments for methionine-
enkephalin arising from a Na+IDS analysis ................................. 103

Na+IDS mass spectrum of methionine-enkephalin ......................... 104

Structure and possible thermal decomposition fragments for the octa-peptide
(Val-His-Leu-Thr-Pro-Val-Glu-Lys) arising from a Na+IDS analysis....106

Na+IDS mass spectrum of Val-I-Iis-Leu-Thr—Pro-Val-Glu-Lys ............ 107

Cs+IDS mass spectrum of PPG 425 obtained on the JEOL PIX-100
HRMS ............................................................................. 109

(a) FAB mass spectrum and (b) "Na+IDS by FAB" mass spectrum of
arachidonic acid ................................................................. lll

"K+IDS by FAB" mass spectrum of glycerol obtained using a split fab

probe tip ........................................................................... l 13
Methane-CI mass spectrum of 2-hydroxyisobutyric acid (di-TMS) ...... 119
Relationship of P(E) and k(E) for unimolecular decomposition ............ 120

Distribution of internal energy in protonated analyte molecule following
exothermic proton transfer ...................................................... 122

Chemical ionization mass spectra of ammonia obtained at various ion source
pre 8 s ures .......................................................................... l 25

Plot of ion source pressure vs. relative intensities of pseudo-molecular ions
for 2-hydroxyisobutyric acid (di-TMS)...: .................................. 128

SIM of p—hydroxybenzoic acid (di-TMS) comparing sensitivity obtained at
optimum ion source pressure using (a) ammonia and (b) methane ....... 137

Comparison of true CI ion source pressure vs. source housing
pressure ........................................................................... 140

xiv.

Figure 53.

Figure 54.

Figure 55.

Figure 56.

Figure 57.

Figure 58.

Figure 59.

Figure 60.

Figure 61.

Figure 62.

Figure 63.

"Typical” chemical ionization tuning ﬁle using perﬂuorotributylamine
(PFI‘BA) as the standard calibrant; m/z 69, (CF3 ); m/z 219 ( 4F9 +); m/z
414 (C8F16N +) .................................. . ............................... 142

Effect CI ion source pressure has on overall sensitivity for (a) p-
hydroxybenzoic acid (di-TMS), and (b) Maleic acid (di-TMS). m/z 261

(Mm); m/z 278 (MM-14+) ..................................................... 143

NHa-CI mass spectrum of (a) m-hydroxybenzoic acid (di-TMS) and (b) p-
hydroxybenzoic acid (di-TMS) ................................................ 145

Dependence of fragment ions and abundances of 2-hydroxyisobutyric acid
(di-TMS) on methane ion source pressure .................................. 146

Flow chart representing scheme for determining the quantitative strengths of
NI-I3-CI relative to EI-MS ...................................................... 149

Comparison of quantitative strengths of El and NH3-CI for the analysis of

2-hydroxyisobutyric and glycolic acids. (a) Reconstructed mass
chromatograms showing isobaric ion interference for El, and (b)
comparison of NHa-CI and 70 eV EI mass spectra ......................... 153

TIC proﬁles of the trimethylsilyl derivatives of an organic acid synthetic
mixture obtained by (a) GC-EI-MS and (b) GC-(N113-0)-MS ........... 156

Comparison of TIC proﬁles of a mixture containing maleic acid (di-
TMS) using NHa-CI at (a) P = 200, and (b) P = 320 mtorr ...... 159

““3 ""3

Comparison of TIC proﬁles of derivatized urinary organic acids from a
diabetic dog obtained using (a) NH3-CI, and (b) E1 ionization ............ 161

Repeat analyses on urinary organic acids by NHa-CI demonstrating
eproducibility from one run to another ....................................... 162

TIC proﬁle for naturally occuring amino acids in a diabetic dog plasma
obtained by (a) NH 3-CI, (b) E1, and (c) electron—capture negative ion

(ECNI) mass spectrometry ..................................................... 163

XV.

Figure 64.

Figure 65.

Figure 66.

Figure 67.

Figure 68.

Figure 69.

Figure 70.

Figure 71.

Figure 72.

Figure 73.

Figure 74.

Scheme for the clinical and biochemical investigation of patients with
suspected congenital lactic acidoses .......................................... 177

GC-EI-MS TIC proﬁles of metabolites found in (a) control, (b) cytochrome
oxidase deﬁcient, and (c) pyruvate dehydrogenase deﬁcient ﬁbroblast
cells ................................................................................ 18 l

GC—EI-MS TIC proﬁles of metabolites excreted into (a) control, (b)
cytochrome oxidase (CytOx) deﬁcient, and.(c) pyruvate dehydrogenase
(PDH) deficient cell medium ........................................ . .......... 183

NHB-CI mass spectrum of glucose (penta-TMS) obtained at the onset of
pulse-labelling. Isotope enrichment observed at rn/z 564 (MNH4+) ..... 186

NH3-CI mass spectrum of lactic acid (di-TMS) from control ﬁbroblasts
collected during the pulse-labelling period .................................... 189

Plot of 13C-enrichment vs. collection time for lactic acid in control,
Cytochrome Oxidase, and PDH ﬁbroblast cells .............................. 190

NHa-CI mass spectrum of glutamic acid from (a) cytochrome oxidase, and
(b) pyruvate dehydrogenase ﬁbroblasts ....................................... 192

Plot of l3C-enrichment vs. collection time for glutamic acid in control,
CytOx, and PDH ﬁbroblast cells ............................................... 193

Proposed su'ucture and fragment ions for unknown (MW =437) presumed
to be an aminoribose sugar metabolite ......................................... 196

Plot of l3C-enrichment vs. collection time for Unknown (MW =437) in
control, CytOx, and PDH ﬁbroblast cells ..................................... 198

TIC proﬁle of the trimethylsilyl derivatives of a synthetic mixttn'e of organic

acids obtained using (a) EI-MS, (b) conventional ECNI-MS, and (c) carbon
tetrachloride "pretreatment" prior to ECNI analysis ........................ 210

xvi.

Figure 75. Comparison of TIC proﬁles of an equimolar mixture of three fatty acid
methyl esters, (C-14, C-16, and C-18) obtained using (a) conventional
ECNI-MS, and (b) ECNI-MS following ion source pretreatment with
CCI 4 ......................................................................... 212

xvii.

Scheme

Scheme

Scheme

Scheme

Scheme

2

LIST OF SCHEMES

...................................................................................... 89

...................................................................................... 96

...................................................................................... 100

...................................................................................... 121

..................................................................................... 185

xviii.

Chapter 1: Complex Mixture Analysis

Introduction:

The ability to accurately characterize complex biological mixtures, both
qualitatively and quantitatively, has been a problem which has plagued both chemists
and biochemists for years. Numerous approaches to achieving these results have been
attempted, perhaps the most widely used analytical approach being that of
chromatography.

As early as 1951, analytical methods such as paper chromatography1 were used
to separate complex mixtures of biological origin. With the advent of high
performance liquid chromatography (HPLC) and gas-liquid chromatography (GLC),
the scope of many biomedical and biochemical studies requiring mixture analysis
expanded. The choice of the chromatographic technique was governed by the
complexity and type(s) of compounds contained within the mixture. HPLC offered
the advantage of the direct analysis of polar non-volatile samples. Analyses of
relatively simple mixtures utilizing HPLC were shown to be readily achieved.
However, for more complex mixtures, such as those of biological origin, HPLC
analyses were often hindered by less than adequate chromatographic resolution than
could be achieved. Furthermore, successful combinations of this technique with other
analytical methods, such as mass spectrometry, are still being evaluated and in the
developmental stage. Difﬁculties are still encountered in achieving the optimum
interface for such a combination. Gas chromatography, on the otherhand, has made
tremendous advances in the past two decades, speciﬁcally in the area of enhanced

column chromatography performance as a result of the introduction of capillary

2

column technology. Even with the tremendous resolving capabilities of narrow bore
capillary columns, shortcomings still were evident with these methods when used
alone for accurate characterization of complex mixtures. Since identiﬁcation of
unknowns relied on the availability of chromatographic libraries (which contain
accurate retention times, retention indices, etc.), identiﬁcation of compounds not
contained within these libraries was impossible. Introduction of a variety of gas
chromatographic detectors (e.g., thermionic ('I'ID), thermal conductivity (TCD), etc.)
frequently simpliﬁed mixture analysis by demonstrating selective responses for
speciﬁc compound classes, yet were limited in that data for identifying unknowns was
not available.

However, not until 1956, when Morrel £111.. 2 demonstrated the potential power
of combining gas chromatography with mass spectrometry (CC-MS) did the ability to
elucidate the identity of unknowns from complex mixtures appear to be possible.

3 showed one of the earliest applications of GC-

Later, in 1971 Homing and Homing
MS-DS, (DS = data system) for urinary metabolite identiﬁcation. To date, this
combination has proven to be perhaps the most powerful analytical method available
for characterizing complex mixtures. Like all analytical methods, though, GC—MS
suffers from some drawbacks in its utility as a "universal" method for mixture
analysis. Even with ”hyphenated" techniques such as these, some prior knowledge of
the mixture to be analyzed is required. For thermally unstable and/or nonvolatile
compounds, chemical derivatization of the mixture is required prior to analysis by GC
and GC-MS. Knowledge of the origin, and/or the principal components of the
mixture to be analyzed is thus required such that the most suitable derivatizing agent
and appropriate stationary phases can be chosen. Derivatization and work-up

procedures can often be tedious. Isolation and derivatization of steroids from urine,

for example, require in excess of 24 hours for reactions to go to completion.

3

Furthermore, GC-MS analysis times can be exceedingly long and data handling and
manipulation laborious.

These drawbacks are more than offset, though, by the absolute strengths of this
combined technique for the analysis of very complex mixtures, especially those of
biological origin which may contain more than 150 components. Even for mixtures
as complex as these, GC-MS can accurately provide both quantitative and qualitative
information. When the mass spectrometer is Operated in the electron impact
ionization (EI) mode, the detector simulates closely that of the FID detector, being
responsive to all compounds that are introduced into the mass spectrometer in the gas
phase. When the mass spectrometer is operated in one of several other modes (e.g.,
chemical ionization (CI), negative ion CI (NICI), etc.) the detector can now simulate
the behaviors of more selective GC detectors, such as the TID, for example,
demonstrating selective responses to individual components of the mixture. These
added dimensions to the mass spectrometer make it ideal as a detector for the gas

chromatograph.

Metabolic profiling:

This thesis speciﬁcally deals with the analysis of complex mixtures of organic
acids, and other metabolites isolated from physiological fluids such as urine. Mixtures
of organic acids have been chosen for a number of reasons. Mixtures associated with
metabolic proﬁling are truly complex. In physiological ﬂuids such as urine, it is not
uncommon to ﬁnd more than 150 metabolites of varying concentration. Since any
one of the metabolites found in the physiological fluids could be of considerable
biological signiﬁcance, it is critical that they be characterized sufﬁciently, and it is

therefore important to have available an appropriate analytical method which will

4

allow for the most extensive qualitative and quantitative analysis of each component
contained within the mixture. Most work done in the area of organic acids analysis (in
x110.) has been performed using GC-MS with electron impact ionization. A number
of limitations are associated with El yet only very sparse data can be found on
alternative ionization approaches for characterizing these complex mixtures, limiting
the capabilities of this hyphenated technique to date for metabolic proﬁling analyses.
Attention will be given to this point later in this section. Information about the
composition of urinary organic acids from healthy and disease-state patients may
reveal insights into various metabolic pathways and possible inborn errors of
metabolism. This thesis is in part devoted to the investigation of some known
metabolic disorders for the purpose of obtaining a greater amount of information on
the relationship of speciﬁcally excreted metabolites for healthy and disease-state
conditions through the use of stable-isotope tracers and alternative ionization
methods, such as GC-CIMS.

These types of analyses collectively fall under the name, or category of
W. This term, ﬁrst coined by Homing and Homing3 has been
given many deﬁnitions, but may loosely be deﬁned as the analysis of components of
physiological fluids (e.g., plasma, urine) for determining subtle and/or signiﬁcant
differences in the proﬁles of control (reference) and subject (disease-state) groups, as
related to the composition of the particular ﬂuid. This deﬁnition is quite general, and
encompasses a large number of biochemical applications. As a screening technique,
several components of the physiological fluid under study are measured and the
concentrations of these measured metabolites are compared to some set of reference
ranges of concentrations in which they are normally found. Screening, or multi-
component analysis can often be achieved for mixtures of various physiological

metabolites, such as organic acids, steroids, and amino acids, simultaneously.

5

Metabolic proﬁling analysis, on the otherhand, involves the measurements of not only
the absolute concentrations of individual components. but more importantly the
relative relationship of these analytes to one another in a single analysis. In addition,
the analytes measured in a proﬁling analysis are typically metabolically related and
analyses are performed on one particular class of compounds such as isolated urinary
organic acids. When the goal of analysis is to determine certain relationships
between excreted metabolites and speciﬁc metabolic pathways, more elegant
experiments, such as stable-isotope uacer experiments, are performed in order to
determine whether some correlation exists between the administered precursor and
some speciﬁcally excreted metabolite(s). An excellent review by Sweeley and
coworkers4 describes the recent developments in the area of metabolic proﬁling.

The earliest graphical representation depicting the metabolic patterns of healthy
individuals was developed by Williams1 shown in Figure l. The positions of the
vectors (labelled 1-31) represented the types of compounds contained within the
physiological ﬂuid under investigation; the length of those vectors directly reﬂected
the concentration of the individual components. Gas chromatograms soon became the
accepted analytical tool for elucidating differences in the metabolic compositions of
"normal" individuals. Homing and Homings demonstrated the enormous power of
gas chromatography for the analysis of very complex mixtures of organic acids from
urine, and the derivatization procedures required to make this class of compounds
amenable to the technique. GC found such wide-spread use in such studies for a
number of reasons. The complexity of the number and amounts of several organic
acids from a complex medium (such as urine) required a technique which (a) allowed
for adequate separation of the compounds in order that identiﬁcation could be made
and (b) a large enough dynamic range such that components which were found in

exceedingly low (or high) levels of concentration could be accurately quantiﬁed from

6

the remaining components of the mixture. Flame ionization detection (FID) for GC
appeared to yield a sufﬁciently large dynamic range for the analysis of these types of
compounds. In addition, rapid advances in capillary column chromatography made it

 

 

31
30
29 5
28 1 2
27
26 ——
25 -. f 5
24 , 3 7
22 9
17 15 12
10
20 18 13

Adapted from Williams, 12.1. with
permission of the author (ref. 1).

Figure 1: Two-dimensional representation of the metabolic status of an
individual derived from paper chromatographic analysis.

possible to completely separate greater than 100 organic acids from urine when
megabore capillary colums have been employed.‘5 More recently, unique methods
for characterizing differences between the metabolic patterns of healthy and disease-
state individuals have been introduced. Sweeley, e131, have converted the qualitative

and quantitative data contained within metabolic proﬁles into musical frequencies.7

Intensities of the musical notes generated are correlated with the concentrations of the
components of the urine specimen. Subtle differences in metabolic proﬁles which are
difﬁcult to differentiate visually (via gas chromatograms) may be shown to be more
easily assessed.

However, it was not until the combination of gas chromatography—mass
spectrometry was successfully achieved did the real potential for metabolic proﬁling
analyses become evident. No one technique seems to parallel the enormous strength
of this hyphenated technique for such studies. Perhaps this is why so few laboratories
enjoy the ability to conduct research in this area, since the cost associated with
obtaining GC-MS systems still precludes their wide-spread availability for most
hospitals and clinical centers.

Following the breakthrough by Homing, 9111.5, and one of the ﬁrst published
results utilizing GC—MS-DS for organic acid metabolic proﬁling analysis by Gates, et
a1M8

between healthy and "disease-state" individuals based on either subtle or large

several research groups began exploring the possibility of differentiating

differences in their metabolic patterns, as a consequence of some genetic disorder.
This has made possible the identiﬁcation of several inborn errors of metabolism.
Differences in urinary organic acid patterns from one individual to another were
assumed to be attributed to differences in the genetic make-up of the various
individuals. This was inferred from metabolic proﬁles obtained from identical twins
which were essentially indistinguishable]. Advances in molecular biology through
the ability to isolate speciﬁc enzymes from ﬁbroblasts and determine their activity in
"controls" and patients has now proven that a genetic basis exists for these classes of
inherent errors of metabolism. Typically, inborn errors of metabolism are manifested

in deﬁciencies in one or more enzymes that regulate carbohydrate, amino acid, lipid,

8

and/or other biosynthetic pathways. Deﬁciencies in these enzymes result in
accumulation of a number of species, including amino acids in plasma and organic
acids in plama and urine. Enzyme deﬁcient individuals are frequently diagnosed
based on monitoring for elevated levels of organic acids (relative to references ranges
in which the metabolites are normally found) excreted in mine. The identiﬁcation of
these enzyme disorders has led to their classiﬁcation as the W.

Organic acids, in addition, have been found to accumulate in amniotic ﬂuid,9’lo

cerebral spinal ﬂuid (CSF), '1 and other ﬂuids.12’13

Most metabolic proﬁling studies have involved the analysis of the m organic
acids using GC-MS. The urinary organic acids are an attractive class of compounds
since they are typically excreted in relatively high concentrations in urine, due to their
relatively high water solubility and, because urine is a physiological ﬂuid which is
readily available and easily obtained. Furthermore, isolation and derivatization
procedures required to make mixtures of organic acids amenable to gas phase analysis
are relatively straight-forward, not nearly as time-consuming as those procedures
required for the isolation and derivatization from other physiological media.

A drawback, however, to urinary organic acid analysis is that this physiological
ﬂuid is very much affected by dietary intake, and other factors (e.g., strenuous
exercise),14 which can alter organic acid composition in urine. In some cases, false
positives for various organic acidurias can be inferred from the urinary organic acid
metabolic proﬁles assuming no other clinical observations are available. When
special care is taken in obtaining urine specimens, however, these problems are
sufﬁciently minimized.

GC-MS has been responsible for the diagnosis of more than 50 heritable disorders
of metabolism (i.e., the organic acidurias).15’16 Lactic acidemia, the most

frequently diagnosed organic aciduria, is differentiated from ”healthy" and other

9

"disease-state" conditions based on severely elevated levels of lactic acid in both
plasma and urine. A number of enzyme deﬁciencies have been attributed to the
excessive build-up of lactic acid in urine, including partial pyruvate dehydrogenase
deﬁciency, pyruvate carboxylase deﬁciency, and many others. In fact, over 15
enzyme deﬁciencies have been correlated with the lactic acidemias alone! 15
Differential diagnosis of these speciﬁc enzyme defects has been difﬁcult based on
comparison of organic acid proﬁles.

The procedures required for carrying out urinary organic acid analysis are
described in Figure 2. A complete set of isolation procedures and various
derivatizing methods may be found in Qgganic Acids in Man."
Bis(trimethylsilyl)triﬂuoroacetamide (BSTFA)/Pyr (4:1) has been used most
frequently for the analysis of organic acids and Other components of urine due to the
simplicity of the derivatization reaction. BSTFA/Pyr derivatizes carboxylic and
hydroxyl moieties to their trimethylsilyl ester and ethers, respectively. Derivatization
leads to better thermal stability of the organic acids which is a requirement for gas
chromatographic analyisis. Trimethylsilylation also leads to an increase in the
molecular weight of the compound, as 72u is added to the molecule for each
carboxylic or hydroxyl group that experiences derivatization. This can be a problem
when mass spectrometers with limited mass range capabilities are utilized. A number
of other derivatizing agents has been employed, including diazomethane, which
derivatizes carboxylic groups to the methyl esters, as to effectively offset the
sometimes undesirable dramatic increase in the molecular weight that accompanies
trimethylsilylation. Following various derivatizing procedures, GC-MS analyses of
the urinary organic acids, as stated earlier, are most frequently performed in the El
ionization mode. Since identiﬁcation of a particular organic aciduria is dependent on

the elevated levels of one or more organic acids relative to the "normal range"

10

 

[PHYSIOLOGICAL FLUID (ex. URINE]

 

\L l. 1111!:anthde

J, 2. DEAR Sephadex, Fyr. Acetate

\L 3. Lyophilize

 

 

I ORGANIC ACID RE SIDUE—l

 

\L 1. Hydroxylamine-hydrochloride

\L 2. B STFAIPyr or Methylation

 

[GAS CHROMANGRAPHT] ——

 

analysis time
i — 1.0-1.5 h

7‘ [MASS SPECTROMETRY I

 

 

 

 

 

 

Figure 2. Procedures for carrying out the isolation and derivatization of urinary
organic acids for analysis by GC-MS.

concentrations, it is obvious that a a technique which quantitatively represents the
components of the mixture be employed. EI-MS has been found to be a quite
reproducible and useful method based on its capacity to provide quantitative
information from a mixture, due to the fact that ionization cross-sections for most
molecules are nearly equivalent by conventional El. 18 However, there are a number
of inherent problems associated with EI-MS which makes this method of analysis
often undesirable for speciﬁc organic acid proﬁling applications. The EI mass spectra
of the trimethylsilyl (TMS) esters of organic acids are generally dominated by ions
that are structurally insigniﬁcant. A typical EI mass spectrum of 2-hydroxyisovaleric
acid (di-TMS) is shown in Figure 3. The major ions in the mass spectrum, m/z =73
[Si (05);] and m/z = 147 [(CH 3)3 Si - o = Si (0492*), are ions associated with the
trimethylsilyl derivatizing agent, and are structurally insigniﬁcant. Only a very weak
ion signal for the pseudormolecular ion (m/z = 247; [M-CH3+] ) is obtained using
electron impact ionization. This behavior is typical for all derivatized organic acids.
A second drawback to EI-MS is that the resultant mass spectra are often unnecessarily
complex. The capacity for accurately quantifying the components of a mixture
becomes exceedingly difﬁcult when the efﬂuent from the gas chromatograph contains
coeluting species that are presented to the mass spectrometer and analyzed by
conventional EI. In addition, when the goal of the metabolic proﬁling analysis is to
gain insight into speciﬁc metabolic pathways, it is generally necessary to perform
stable isotope-, or radioisotope-labelled tracer experiments. A known, labelled
precursor is administered to the subject under investigation and the ﬂux of label (e.g.,
13C) from that precursor is measured for incorporation into various metabolites. The
levels of enrichment observed in various metabolites is typically quite low. As a

result, these kinds of experiments require relatively abundant ions that are unique to

l2

 

 

 

 

 

 

. 147
5 . a.
,5 "7 114-051’
5
1
. M,
60 ' -- It 1 160 m “zoo a a. zoo

mlz

Figure 3. 70 eV EI-MS mass spectrum of 2-hydroxyisovaleric acid (di-TMS)

 

13

the metabolite such that accurate levels of enrichment may be measured. The mass
spectra of organic acids under El conditions are typically devoid of abundant high
mass ions that are indicative of the analyte.

A more general concern regarding GC—MS analyses of these complex mixtures,
arises from the desire to rapidly screen for enzyme disorders, since the organic
acidurias can manifest themselves as very serious illnesses in the patients. As was
shown in Figure 2, the work-up and derivatizing procedures for organic acids can be
quite extensive. Derivatization procedures can also lead to an increase in the number
of components of the mixture being analyzed. Furthermore, GC-MS analyses may
take between 1.0-1.5 hours for complete characterization of a mixtm’e. When the goal
of the proﬁling study is to obtain a rapid screen or "ﬁngerprint" for a particular
organic aciduria, there is an obvious need to obviate the work-up procedures and
time-consuming analyses.

This thesis invesitgates the utility of a novel desorption/ionization (DI) method,
K+IDS, as a very rapid, and extremely attractive alternative to GC—MS analyses for
the ability to distinguish between a number of organic acidurias. Chemical ionization
methods, such as NH3-CI, are evaluated as to their capacity to provide as sastifactory
quantitative alternative methods to El for GC-MS analyses. In addition, this thesis
will demonstrate that the quality and quantity of information which can be derived
from a conventional metabolic proﬁling analysis can be enhanced signiﬁcantly when
chemical ionization methods are used in conjunction with EI mass spectral data for
providing answers to speciﬁc metabolic proﬁling questions. The proper choice of
ionization method will be shown to depend speciﬁcally on the type of proﬁling
question which is to be addressed.

14

Chapter 2: Desorption/Ionization (DI) and K+IDS theory

Prior to the mid 1970's, most mass spectrometric analyses required that the
analyte be introduced in the gas phase prior to ionization and detection. This meant
that the compound of interest either had to exhibit a relatively high degree of
volatility or that some suitable chemical derivatization reaction be performed to
enhance the volatility and thermal stability prior to mass spectrometric analysis.
Sample heating techniques through the use of direct insertion probes (DIP) found
success for both conventional EI and CI studies, but still required that the analyte be
volatile, so that gas-phase ionization reactions could occur. Analyses utilizing
chemical ionization mass spectrometry were becoming commonplace, thanks to the
ground work of Field and Munson,19 because intense peaks representing molecular
weight information were shown to be obtained readily . In conjunction with the
structural information available by El, identiﬁcations of unknowns could now be
made more accurately and with a greater degree of conﬁdence. With these gas phase
ionization methods it was found, though, that not all compounds gave rise to
molecular ions in their mass spectra; some compounds were either considered too
”non-volatile" for these gas-phase ionization reactions to occur or were found to
readily undergo fragmentation to thermodynamically favored decomposition
products. These concerns led to the investigation of ionization methods which could
more effectively deal with the problems associated with non-volatile compound
analysis. Field ionization (F1) 20 and ﬁeld desorption (FD)21 techniques were
shown to be quite useful methods for generating molecular ions having very low
internal energies as a consequence of their respective ionization processes, and the

mass spectra of non-volatile compounds utilizing these two techniques were generally

15

very simple to interpret. These two methods began to expand the range of
compounds amenable to mass spectrometric analyses. In addition, rapid-heating
methods, such as flash vaporization techniques, found success in that compounds
previously considered to be non-volatile could, under the appropriate conditions,
desorb intact into the gas phase. Beuhler showed through rapid heating techniques,
that thermally labile compounds such as peptides could desorb from inert teﬂon
surfaces. The rates of desorption and decomposition were found to be competitive,
but conditions could be attained in which desorption of the intact analyte molecule
was favored over a thermodynamically favorable decomposition pathway. 22

Barber and coworkers developed the novel desorption/ionization method of fast
atom bombardment (FAB)23 in 1981 and demonstrated that mass spectrometry could
be used to yield mass spectra of compounds which were previously considered
thermally labile and unamenable to mass spectrometric analysis. The technique of
FAB is now considered one of the most important desorption/ionization methods
available to the mass spectrometrist. Fast atom bombardment ionization involves the
deposition of a thermally labile compound (e.g., oligosaccharide; peptide) onto a
probe tip containing a polar matrix (e.g., glycerol) which is used to solubilize the
analyte species. A high energy beam of fast atoms (or ions), is focused onto the probe
and the energy associated with the bombardment process is typically sufﬁcient for
desorption of the analyte to occur whereby it may participate in gas phase ion
molecule reactions with the co-desorbing ions from the matrix. Other matrices have
been introduced. such as dithiothreitol/dithioerythritol (D'I'I‘:DTE), thioglycerol, and
others which have allowed for the analysis of a wide range of compound types by
FAB-MS.24'25 However, a number of inherent problems associated with FAB
have warranted the investigation and implementation of other, new

desorption/ionization methods in mass spectrometry. The implementation of some of

16

these methods has been a result of the need to overcome some of the inherent
problems associated with FAB. The need for a matrix in FAB in the desorption
process to produce long-lived ion signals (a) limits the number and type of thermally
labile compounds that can be analyzed based on solubility constraints, and (b)
unnecesarily complicates mass spectra due to severe matrix ion interferences. In
addition, the technique has shown only little promise for quantitative analysis.

The analytical utility of all desorption/ionization methods lies in their ability to
produce abundant molecular ions. Positive-ion FAB mass spectra are typically
characterized by intense protonated molecules, which are presumed to arise through
gas phase proton transfer from the more acidic glycerol molecules. Compounds can
be induced to form molecular adduct ions with appropriate cationic species. In FAB,
LDMS, F1 and FD mass spectrometry. formation of [M+ alakali ion]+ has been
observed for mixtures of alkali salts and organic compounds. For FAB-MS,
cationization is made possible by pre-mixing glycerol (or other matrix) with an
appropriate alkali salt (e.g., Na], KI, etc.). The stability of these molecular adduct
ions is an attractive feature, since mass spectra are relatively clean, and mixtures can
often be characterized accurately due to the simplicity of the mass spectra that result.
The stability of the alkali ion adducts lies predominantly in the fact that the charge
associated with the molecular adduct species is localized on the alkali ion, due to its
low recombination energy (identical in magnitude to its ionization potential which is
low relative to most organic compounds).

The pioneering work in the area of cationization of organic compounds is a result
of the elegant studies done by Rollgen and coworkers.“'28 Early experiments in
their laboratory utilized 10 11M diameter tungsten ﬁeld ionization emitters coated with
a thin layer of alkali salt, most typically Lil. Low heating currents were-applied to

the emitter and Li". was found to be generated in copius amounts in the presence of

1?

moderate electric ﬁelds. Organic compounds, such as carbohydrates, straight-chain
alcohols and others, were desorbed through slower heating methods and introduced
into the gas phase. Reactions between organic compounds and Li+ on the surface of
the emitter occurred, resulting in the formation of cationized molecular species in the
presence of electric ﬁelds just below the threshhold level required for ﬁeld ionization.
Alkali ion attachment was shown to readily occur for numerous organic compounds,
both volatile and thermally labile.

In addition, Rollgen 5131,29 described thermal desorption of ions such as
[M+Na+] from heated mixtures of alkali salts and organic compounds on metal
surfaces. One of their approaches incorporates a two-ﬁlament probe. A source of
gas-phase alkali ions is generated through rapid heating of an alkali salt, such as KI,
Lil, or NaI (contained within a silica gel matrix) from one emitter wire, and gas phase
addition reactions are achieved with organic compounds which desorb from the
second wire. The types of ions observed in the mass spectra primarily consist of
alkali ion adducts.

A number of novel desorption/ionization methods have been proposed to increase
the scope and capabilities of mass spectrometry in the area of thermally labile
compound analysis. The technique of thermally-assisted (TA)-FAB has been found
to effectively eliminate matrix interference through methods of background
subtraction for certain classes of compounds30 . The technique of continuous ﬂow
FAB, developed by Caprioli, our, 31 in which the efﬂuent from an HPLC column is
deposited directly onto a FAB target containing a solution of glycerol and water has
shown enormous potential as a reduction in chemical background is observed,
possibly as a consequence of the fact that glycerol is diluted to a large extent in water.
In addition, methods such as laser desorption and plasma desorption mass

spectrometry (PDMS) are ﬁnding much greater use, especially for the analysis of high

mass compounds such as high molecular weight polymers and biopolymcrs, ”'33

due to the wealth of molecular weight information which is available from the mass
spectra obtained using these ionization techniques. However, short-comings exist
with all of these ionization methods which limit, to some extent, their utility for mass
spectrometric analysis. Not of least concern is that implementation of many of these
techniques requires sophisticated instrumentation resulting in relatively expensive
methods for thermally labile compound analysis. The ideal desorption/ionization
method would be amenable to all thermally-labile compounds, thus making it
"universal" as an ionization method while being W.

This research has been focused on the continued development and
characterization of a novel desorption/ionization method developed in this laboratory,
called potassium ion ionization of desorbed species (K+ IDS) which continues to
show promise in being responsive to these two criteria mentioned above. K+IDS has
shown tremedous potential in its universality as an ionization method for both volatile
and thermally labile compounds. The technique of K+ IDS has found applications,
too, in areas as varied as industrial polymer and biopolymer analysis, and complex
mixture analysis (including organic acids, and other marginally-volatile compounds).
This ionization method is based on the production of gas phase metal ions (alkali ions
are most frequently used) which are used to ionize thermally labile compounds which

are desorbed into the gas phase by rapid heating.

19

K+ IDS theory

K+ IDS is based on the process of thermionic emission for the production of gas
phase alkali ions. Bewlett and Jones demonstrated a simple method for producing

34 that could be used for mass spectrometric analysis. This

metal ions in the gas phase
method involved the use of thermionic emission materials, consisting of metal oxides
contained within an aluminosilicate matrix. Metals ions have been found to be
produced in copious amounts when glass materials and/or an alkali oxide-containing
matrix is heated and biased appropriately. Temperatures greater than 700-800 'C
have been proposed as the lower temperature limit for alkali ion emission to occur.
For thermionic emission to occur, the energy available in the solid must be sufﬁcient
such that a positive ion can be formed and detected in the gas phase.

Shown pictorially in Figure 4 are the thermodynamic processes important in alkali
ion emission from a aluminosilicate matrix. The overall process for alkali ion
emission can be considered a combination of the heat of vaporization of the molecule
(X A)’ the ionization energy (LE) of that molecule, and the work function ((1)) of the
surface. The thermodynamic values of the energy of diffusion (ED), the energy
associated with the migration of an atom through the material, and X A’ the energy
required to transfer an atom (or molecule) from the aluminosilicate surface to the gas
phase, are considered negligibly small for metal oxides contained within this
aluminosilicate framework.” Ion scattering experiments have shown that potassium
migrates readily to the surface of these aluminosilicate matrices. The observation of
only small ion signals from other atoms indicates a high surface population of
potassium ions“. FAB analyses of these metal oxide containing aluminosilicate

mixtures show that only the metal ion is formed to an appreciable extent in the gas

phase, suggesting that potassium can migrate relatively freely to the surface of the

20

 

 

 

 

 

 

-r (4)
if A t-) 2‘ t :2
LP. (K)
Xx 2 o
1’1)
M K20 : A1203: Si 20 Surface

Figure 4: Thermodynamic considerations for K+ ion emission.

matrix. The emission of alkali ions from the surface, therefore, may be approximated
as only being dependent on its ionization energy and the work function of the

material., as shown in equation 1,
E = LE. - (b, (1)

where (b, in this example, represents the work function of the potassium
aluminosilicate surface and LE. (K) is 99.8 kcal mole'l . The work function of the
material, in simple terms, represents the energy required to remove an electron from
its surface. Typically, these work functions are on the order of a few eV energy. Low

work function materials, such as aluminosilicate materials containing cesium can

21

more readily emit electrons under similar conditions than a high work function
surface, such as an aluminosilicate material containing sodium. Bombick gives an
exhaustive discussion of the types of work functions and their dependences on
temperature. In addition, the properties of these materials are found to be affected
dramatically by their chemical environment 37

Based on the properties of these glass materials, it is possible to construct an
alkali ion emitter that can be used to produce a high ﬂux of alkali ions (or other metal
ion) in the gas phase which can be used in ion molecule reactions. This is the basic
premise of the K+IDS technique. Experimental considerations and theoretical aspects
of the technique will be discussed in the following section's.

Early work in this laboratory used thermionic emitters for CIMS experiments in
which alkali ions were employed as the reagent ions for these gas-phase reactions. A
thermionic emitter containing this glass mixture was designed such that it could be
introduced into the ion source of any mass spectrometer having a DIP inlet. The
probe used in this laboratory is a modiﬁed DIP probe for an HPS985 quadrupole
mass spectrometer. Potassium ions were found to be produced in c0pius amounts if
the probe was heated to sufﬁciently high temperatures and biased positively with
respect to the ion source housing. Introduction of sample gas, either through a
”batch" inlet system or through the gas chromatograph at the same time that
thermionic emission was occuring could lead to gas-phase alkali ion addition
reactions. Bombick, Pinkston and Allison showed that addition reactions involving
volatile analyte molecules (e.g., acetone) and these gas-phase alkali ions required that
( 1) the kinetic energy associated with the K+ ions emitted from the surface fall into a
narrow "bias" window and (2) a third body collision was required for the stabilization
of the newly formed adduct ion. 38 This "bias” window, typically a 2-5 V positive

bias on the thermionic probe tip relative to the ionization source housing, is necessary

22

for adduct ion formation to occur. It was determined that under conditions in which
the bias voltage applied to the probe falls outside the range of this ”bias window,"
surface ionization reactions can be overwhelmingly favored over gas-phase alkali ion
attachment reactions for certain classes of compounds (e.g., amines).

These observations led Bombick and coworkers to investigate the types of ions
that might be produced when a thermally labile compound was place directly onto the
K+ glass. The types of ions observed, when the alkali ion thermionic emitter
containing thermally labile sample was heated, appeared to be a combination of K+-
CI and thermal degradation products. The overwhelming majority of ions observed in
the mass spectra of these compounds were K+ adducts which are presumed to occur
following desorption of analyte molecules into the gas phase; hence, the name,

potassium ionization of desorbed species (K+ IDS).
Operative mechanism of K+IDS

As mentioned, K+IDS is a novel desorptionflonization developed in our
laboratory that is based on thermionic emission materials for producing gas phase
alkali ions which can be used as a source of reagent ions in reactions with most
organic compounds (including thermally labile and volatile molecules) desorbed from
a surface. As mentioned, the best proposed mechanism to date involves the
desorption of analyte from a solid surface followed by gas phase alkali ion
attachment. This mechanism has been further supported based on the implementation
of a new probe design which was required for the analysis of mixtures of marginally-
volatile compounds. This point will be discussed in detail. As the alkali

aluminosilicate matrix containing analyte solution is rapidly heated, one of two

23

processes can occur. If the compound is sufﬁciently thermally stable, intact
desorption of analyte can occur from that surface. However, if the compound is
considered thermally labile, a number of thermal decomposition reactions can occur
on the sm'face prior to desorption into the gas phase. The processes are summarized
in Figure 5 below. A thermally labile species, denoted as ABCD, is coated onto the
surface of the alkali aluminosilicate bead. When heated to temperatures above that
required for alkali ion emission, K+ (or other alkali ion) ionization and desorption of
the alkali ion into the gas phase occurs. Concurrently, desorption of the thermally
labile species, AB-CD occurs, in some form into the gas phase (i.e., as the intact
molecule ABCD®, or as a thermal decomposition product, AB®, where (g)
represents the gas phase) and alkali ion addition reactions with desorbed species are
readily observed. It has been mentioned that adduct ion formation requires a third

body for collisional stabilization of the initially formed adduct ion complex.

ABCD. +K
AB. 1' K
CD. +K .

    
 
 

Ki-

  
        

 
 
 

\
\\\\\\

\\\\\\
\\\\\

    

 

\

Figure 5: Proposed mechanism for adduct ion formation of desorbing species,
ABCD, from alkali aluminosilicate surface.

24

The Operative process envisioned is similar to the one proposed for both

secondary ion mass spectrometry (SIMS) and FAB, in which a selvedge region is
believed to exist directly above the surface. 39 This selvedge region in a K+ IDS
experiment is believed to be a high pressure region consisting Of desorbed neutrals
and K"’ ions. Bombick, m1.” demonstrated that by increasing the pressure in the
ion source using a relatively inert gas, such as N2, an increase in the detectability Of
analyte ions could occur, lending further support to the requirements for third-body

stabilization.
Experimental Considerations

A direct insertion probe for an HP5985 quadrupole mass spectrometer was
modiﬁed by Bombick m], to yield the "original" K+ IDS probe shown in Figure 6.
The tip of the probe consists Of a two-holed ceramic rod through which a rhenium
wire (0.007 " diameter) is inserted and looped. An alkali aluminosilicate mixture
containing potassium in the form Of KNO3 (converts tO K20 when heated) is applied
tO the end Of the "looped" rhenium wire ﬁlament as a slurried mixture suspended in
MeOH. The bead is then conditioned under a bunsen burner for a few minutes to
cause the alkali glass material to conform to a porous, ceramic-like glass. Electrical
connections between the rhenium wire and extended K+IDS probe are made using
low resistive copper wire.

Rapid and stable heating currents are achieved using a low current power supply
designed by Martin Rabb, Electronics Shop, Department of Chemistry, Michigan
Statee University. The probe tip may be biased relative tO ion source housing (or ion

focusing lenses) through an internal variable low voltage source contained within the

25

/. / alkali 0mm

"3 __/ A110,: 31,0 :lt.o

 

 

 

  
  

GC capillary dine!
Interface column

 

 

 

Power Supply
0A1 IS 5.!“

   

 

  

__J

High Volt, tow
Current P8.
L (9) 97'

 

 

 

 

Figure 6. Modiﬁed direct insertion probe (DIP) for use as a thermionic emitter
for K+IDS analyses

26

current source. The schematics of this current power supply are shown in Figure 7 as
reference. The emission current source is a current power supply which is capable of
delivering up to 5 amps DC. The current output is ﬂoating on a bias supply which is
referenced to the reference voltage input. The current ampliﬁer compares the voltage
generated by the load current ﬂowing through a 0.1 Ohm resistor with the voltage set
on the current control and via a power transistor, drives the current to the desired
value. A current limit transistor senses part Of the voltage across the two 0.1 Ohm
resistors which are in series with the load current and limits the output tO a controlled
value. Current may be preset to a desired value prior to being turned on by a front
panel toggle switch. The bias voltage may be adjusted tO a maximum Of :1: 15V DC or
:1: 200 V DC, with lO-hi and polarity toggle switches and front panel potentiometer.
The probe is inserted into the source Of the mass spectrometer via the direct
insertion inlet as shown in Figure 6 and again conditioned by slowly heating the probe
tip in approximately 0.5A increments up tO c.a. 0.25 A above maximum operating
current for analysis. Alkali ion emission is Optimized by: (a) varying the bias
potential to the K+IDS probe; (b) probe positioning; and perhaps most importantly (c)
tuning the ion focusing lenses and repeller.
A typical tuning ﬁle achieved after conditioning Of the bead is shown in Figure 8.

Peaks corresponding to m/z =39 and 41 (K+ and its natural isotope) are Optimized for
best peak shape and abundance. Mass axis calibration is generally achieved under EI
conditions using a high mass calibrant which readily forms low mass fragment ions,
so that calibration can be achieved over a wide mass range. The mass Of the
potassium ion in this tuning ﬁle was calibrated to m/z = 39.05 u. Also implied from
the tuning ﬁle is that at low electron multiplier (EM) settings (0V 5 EM 5 3000) it is

possible to generate substantial potassium ion current. It does not necessarily follow

27

in ’ ' .
”to-l m . m,
3:" hi it) ’ I ma

 

 

 

 

(Unﬁt) Jﬂmw

'3 “-800

I. [@211

 

        

 

 

 

 

 

 

 

 

 

 

Figure 7. Schematic diagram Of emission current power supply.

28

owls}:

 

 

8am 3%
”an no.3
.28 no.2.”

 

 

IJ\J
\

~IBEO§
RIENUMSO”

Ewe" m20~

32 7 lhmmmmo g mag
380." I Z—(O mg mag

_ Na IEEO 92¢.

Q». I 7:50 932

 

 

onus):

o a >958 205

o t 23 29mg

8" .. 8 >53"

3 .. ea. $5 «25 En
a...” a co 800m 29

9. .. E 50355

e .. cc g

3;.“ ” A82 55 ES .

29

that alkali ion emission current increases with increased probe current. Typically, as
power supply current is increased, a change in bias voltage is required to optimize
potassium ion signals. Currents used above 4.0A have been found to signiﬁcantly
decrease the life time of existing glass beads, yet as will show later, these higher
currents can result in beneﬁts in the analysis of certain classes of organic compounds.

Following optimization of K+ emission, the probe is removed from the source
and the compound to be analyzed is deposited onto the emitter as a solution in
methanol, hexane, or acetone. These transport solvents are generally chosen due to
their case of evaporation. Evaporation of aqueous solutions are difﬁcult and time-
consuming, and frequently poorer sensitivity has been observed when water is used as
the transport medium.

Bombick gives examples of many types of compound classes that are amenable to
the K+ IDS ionization method,” including analysis of pharmaceuticals,
oligosaccharides, and polymers, to name only a few. Discussed in this thesis are the
results of experiments which have utilized this nearly "universal" ionization method
for (a) determining the quantitative and qualitative strengths of the technique for
complex mixture analysis, (b) improving upon existing detection limits to make the
technique competitive with other ionization methods for similar compound analyses,
and (c) further deﬁning the boundaries of this ionization method (as to the types and

numbers of compounds amenable to analysis using the technique).
Initial results utilizing K+IDS for mixture analysis

The potential of the K+IDS technique for yielding an accurate and quantitative

representation of the components of a mixture has previously been suggested.40

30

Bombick and Allison demonstrated that the K+IDS technique yields mass spectra that
accurately reﬂect the molecular weight distribution of low molecular weight
oligomeric components of a polymeric mixture. For example, the analysis of the
K+IDS mass spectrum of poly(propylene glycol) 725 (average MW = 725), yielded
an average molecular weight Of 710, based on the intensities of the (M+K)+ ions of
the individual components in the complex mixture. Typically, several mg of
polymeric material were coated onto the probe tip and resultant TIC proﬁles Of these
polymers showed gaussian type behavior, with extensive tailing, due to desorption of
analyte from both "hot" and "cold" spots on the ceramic rod. Typically a mixture of
thermal degradation products and desorption products were observed in the K+ IDS
mass spectra. Thermal degradation products were Observed generally at late times
during analysis and are believed to have resulted from desorption from these "cold"
spots on the thermionic emitter. Averaging of mass spectra over the entire elution
proﬁle could not be used for determining the relative concentrations of the
components of the mixture. Typically, one region within the entire polymer
desorption proﬁle was found to most accurately reﬂect the components, and their
relative concentrations of the mixture.

These initial results on the analysis of polymers suggested that the K+IDS
technique might be useful for mixture analysis. Part of the goals outlined in this
thesis are to further investigate the capabilities of K+IDS for providing "batch"
analyses of mixtures. Of speciﬁc interest are the organic acids, and other metabolites
of physiological ﬂuids. The following chapter will concentrate on results utilizing the
K+IDS approach for rapid characterization of several of the organic acidurias. The
problem of sensitivity constraints, which to date limited the analytical utility of the
technique for biochemical analysis had to be addressed, since the concentration of

organic acids in urine are not likely to exceed several tens of rig/ml urine at best and

31

generally fall in the range of low ng - 11g] ml urine concentration.

Polymers , as mentioned, were found to undergo both thermal degradation and
intact desorption and this has been found to depend on the amount of material and
heating currents used in their analysis. Situations in which both intact analyte ions
and thermal degradation products are observed in K+IDS mass spectra are non-ideal
for mixture analysis since spectra can become exceedingly complicated and
quantitation of the components difﬁcult. Thus, for mixture analysis, the K+IDS mass
spectra ideally would be nearly as simple as possible, yielding only one peak for each
component of the mixture.

The K+ IDS technique was evaluated as to whether it would be capable of
yielding [M+K]+ information on compounds typically excreted in urine, such as
organic acids, fatty acids and amino acids. Experiments using the single-ﬁlament
probe required the deposition of several 11g of organic acid onto the K+IDS probe tip.
Experiments utilizing Operating currents Elm 2.75 A (c.a. 1000 'C) resulted in no
analyte ion signals. Only when the probe was heated to currents above this
temperature were analyte ions of the form [M+K]+ produced and detected. This was
a unique requirement since all K+IDS mass spectra obtained by Bombick used
heating currents between 2.0A and 2.5A. Subsequent studies revealed that these
higher power supply currents were required to generate a high ﬂux of potassium ions
more rapidly. This situation allowed for better spatial overlap to occur between the
desorbing analyte and K+ ions available in the gas phase.

Fortunately, K+IDS mass spectra Of volatile compound classes such as short-
chain carboxylic acids, amino acids, fatty acids, and others are indeed quite ”clean”.
Shown in Figure 9 is the K+IDS mass spectrum of a mixture containing ﬁve
commonly occurring organic compounds, of varying concentrations. The mass

spectra are very clean, exhibiting only [M+K]+ adduct ions for each species.

32

UQ-wahnﬁmudMnﬂ
JMk-Phthyphnﬁxukadd
M3-Homovanillieaeﬂ
[Mai-No") (“3+NOIIW4'PW3W“—|
M’s-Palnriticdtl
(”4+N°+l [M5+NO+]

 

‘ [H‘+No*]

 

 

 

 

 

 

 

130 150 170 190 210 230 250 270 290'

Figure 9. Na+IDS mass spectrum of five common organic acids
(underivatized)

33

However, the ion currents associated with these compounds are quite short-lived. TIC
proﬁles are considerably shorter in duration than those Obtained for thermally labile
compounds, due to smaller sample loads (relative to polymers) and low temperatures
required for volatilization of the compounds. Nevertheless, these types of compounds
exhibit the ideal behavior that was needed for the analysis of complex mixtures by
K+IDS.

Several simple mixtures of organic acids, fatty acids, and steroids were prepared
in order to be analyzed by the K+IDS technique. Shown in Figure 10 are averaged
mass spectra ofa 5:1:l:l mixture of C-12, 14, 15, and 16 free fatty acids obtained in
one K+IDS analysis. Four important observations regarding the types and quantities
of ions observed in the mass spectra are: (1) only [M+K+] ions are generated for
each component of the mixture, (2) the K+IDS mass spectra do not accurately reﬂect
the relative concentrations of the components of the mixture, (3) higher mass ions
increase in intensity at later times during the K+IDS analysis, and (4) when higher
heating rates are used (I 2 3.5A), the relative intensities of the species are slightly
improved to more closely represent the true mixture composition. This type of
behavior has been observed not only for fatty acid mixtures but for mixtures of other
marginally-volatile compound classes, such as short-chain organic acids and amino
acids.

These results were discouraging since analysis of polymeric materials suggested
that relative ion intensities could be used to accurately reﬂect the composition of
complex mixtures, whereas simple mixtures of organic acids and fatty acids now
appeared to show preferential response towards the higher mass species, since higher
mass ions were observed to contain the larger fraction of total ion intensity. The

concept of preferential ionization needed to be considered since there are a number of

35

reports in literature which suggest that adduct ion formation does not necessarily
occur with equal probability for similar compounds, even isomer-s}l
However, this preferential ionization towards higher mass molecules (for the
homologous series of straight-chain fatty acids) might be a result of differing degrees
of volatility of the organic components. This proposed explanation for the results
stemmed from a number of experimental observations. As brieﬂy mentioned, when
higher heating rates are used (i.e., I=3.5-4.0A) the relative ion intensities appear to
more closely resemble the true composition of the mixture. This behavior is shown
pictorially in Figure 11. The preferential ionization of higher mass ions is proposed
to be a result of the lower molecular weight Species desorbing from the glass surface
at such rapid rates that the majority of these analyte molecules are evacuated from the
ion source prior to production of a sufﬁcient [K+] ion concentration needed in the gas
phase to participate in alkali ion addition and stabilization reactions. A steadyrstate
K+ ion concentration may be attained more rapidly when utilizing higher heating
rates. Theoretical plots of alkali ion emission and analyte desorption proﬁles as.
time for thermally labile compounds such as polymers were described by Bombick
and are shown in Figure 12. The importance of these plots is that desorption of
analyte occurs prior to alkali ion emission, and only where the two curves overlap
may addition reactions occur. Veriﬁcation of this type of behavior for the analysis of
fatty acids, and other volatile compounds was made. Desorption proﬁles of fatty
acids were generated by placing a solution of fatty acidsonto the K+IDS probe tip,
rapidly heating without a bias potential and forming ions by electron ionization. It
was found that at low source temperatures ('1‘ source S 40 'C), ions were formed

Willi-51311141199115]! (slightly less than 1 860) Upon application of probe current.
Potassium ion emission proﬁles, on the Other hand, were obtained by inserting the

36

 

[M (+10“

IM3+KF
‘ Mylo"

“42m“

RELATIVE ABUNDANCE

 

 

 

“I 16. 18. 28. 22. 24. 26. 28. 38' 32.

Figure 11. K+1DS mass spectrum of same fatty acid mixture obtained at I = 4.0 A

 

INTENSITY

37

#-

Species Desorbed by
Ropld Heating

\

K

K’ Emission

 
    

 

 

 

 

 

 

 

 

 

 

 

 

 

TIME

Figure 12. Analyte desorption and alkali ion emission proﬁles \8 . time utilizing
a single-ﬁlament K+IDS probe *

38

probe without analyte into the mass spectrometer, applying the same current and now
adding the appropriate bias voltage to "push" potassium ions from the surface. K+
emission was found to reach a steady-state approximately 6 sec following the
application of current. These results suggest that addition reactions are proceeded by
desorption of a majority of analyte prior to sufﬁcient alkali ion emission. For the
higher molecular weight species, desorption occured at a slightly slower rate than for
the lower molecular weight species, resulting in better overlap between the analyte
and K+ emission proﬁles. These observations were alsoOOf utility in explaining why
such large sample loads were required using the "original" probe design.

These observations led to the consideration of a modiﬁcation Of the existing probe
so that the analyte desorption and alkali ion emission proﬁles would better overlap for
all species contained within the mixture, with the intent of improving both the
sensitivity and quantitative nature of the technique in the analysis of these marginally
volatile compounds. A dual-ﬁlament K+ IDS probe was utilized to try to effectively
eliminate the inherent problem of poor overlap between analyte desorption and alkali
ion emission proﬁles associated with the original design. Of note is that a similar

42 who demonstrated

design modiﬁcation was presented by Rollgen and coworkers
the use of a two-ﬁlament design mounted on a common "push-rod." The question of
sensitivity was addressed ﬁrst since the amount of organic compound required to
produce appreciable ion signals using the original probe design were such that it
precluded the analytical capacity for mixture analysis of components of biological
interest. Bombick demonstrated that the lower limit of detection for polyphenyl ether
was approximately 500 ng at a signal-to—noise ratio (S/N) of ﬁve. When a modifying

gas such as N 2 was introduced into the ion source at high pressures (c.a. l torr) the

detection limits were lowered to approximately 250 ng (S/N =5). The presence of this

39

modiﬁer gas further supported the proposed gas-phase mechanism in which a third-
body was required for stabilization, since lower detection limits resulted.
Implementation of the dual-ﬁlament probe modiﬁcation has resulted in an
approximate 30— to 40-fold enhancement in detection limits for the technique.
Methylstearate is typically used for determining the relative sensitivities of ionization
techniques such as CI and El, where speciﬁcations for full scan detection limits in the
El mode are typically lng for methyl stearate. Results of selected ion monitoring
(SIM) analyses on methyl stearate shown in Figure 13 are compared using the
original and modiﬁed probe designs. Comparison of peaks heights (H) and peak
areas (A) of the [M+K]+ ions (tn/z =337) illustrate the enhancement in detectability.
Selected ion monitoring on the natural isotope peak of methyl stearate, m/z =338,
further demonstrated this tremendous improvement in sensitivity. The amount of
methyl stearate required to achieve a S/N = 5 using the dual-ﬁlament probe has been
determined to be 3ng (SIM) and 10ng (full scan, 50-450u/ 0.5 sec). A calibration
curve monitoring SIM ion counts vs. ng methyl stearate onto the bare rhenium wire
is shown in Figure 14. The detection limits for this class of compounds is certainly
comparable to other desorption/loniation methods. In FAB, for example, 1-5 11g
material are "typically" required for analysis. However, it must be pointed out that
the ion signals achieved by FAB are much longer-lived than those Obtained using the
K“ IDS technique.

Reasons for the dramatic increase in sensitivity can be attributed to better overlap
in analyte and alkali ion desorption proﬁles. By placing the analyte solution onto a
second ﬁlament which now only experiences radiative heating from its close
proximity to the potassium emitter, the onset of desorption of volatile analyte from

this ﬁlament is slightly slower than in the original design. K“ emission has also been

40

 

A: Peak Area
H = Peak Height

 

 

 

 

 

H: 5100
A = 26000

 

 

 

Scan time

Figure 13. Selected ion monitoring (SIM) comparison of methyl stearate (400 ng)
using (a) single-ﬁlament and (b) dual-ﬁlament K+IDS probe

Ion 'eounts' for ml: .- 337

41

 

 

 

 

 

20000-
16000—-
12000—-
8000-
4000— l l lﬁ'
0 IO 20 30 40
l I l l l I l I I 7
o 100 200 300 400 500

ng Hethyl Stearate

Figure 14. Detection limits for methyl stearate by K+IDS using dual-ﬁlament
probe

42

found to occur more rapidly from the free emitter surface. In the original probe
design, K+ ions were required to migrate not only through the aluminosilicate
framework, but in addition, through the bulk of analyte sample that was deposited
onto the aluminosilicate material. By removing this second migration constraint for
the K+ ions, emission into the gas-phase is achieved more rapidly. The result has
been a reduction in the amount of time required for K+ emission to reach a steady-
state value (c.a. 4 sec). Comparison of desorption proﬁles using the one and two-
ﬁlament designs is shown in Figure 15. As can be seen, better overlap is achieved
when the two-ﬁlament design is implemented resulting in a greater fraction of
desorbing analyte molecules undergoing collisions with gas—phase alkali ions.

The improvements in detection limits obtained using the modiﬁed probe design
has now allowed this K+ IDS technique to be utilized in the analysis of complex
mixtures of organic acids isolated from urine specimens of both healthy and disease-
state individuals. The following chapter gives a description of the currently used
probe design and demonstrates the capabilities of the K+ IDS technique for providing
both qualitative and quantitative information for complex mixtures. The strengths of
the K+IDS technique are demonstrated using appropriate "real-life" examples in the
area of metabolic proﬁling. A brief discussion is presented ﬁrst on the need for
evaluating alternative ionization methods in the area of metabolic proﬁling. The
results summarized in the following chapter are as they appear in a manuscript that
has recently been accepted for publication in Biomedical and Environmental Mass

Spectrometry .

43

Species Desorbed
from Re Wire

 

 

 

 

 

 

 

 

 

 

I ”’
I .
Ii ’ K+ Emission from
"VI "Free" Surface
II ’
I
I
I
I
l
I
II ,1
I /;.
I
I
I .
TIME

Figure 15. Analyte desorption and alkali ion emission proﬁles is . time utilizing
a dual-ﬁlament K+IDS probe

44

Chapter 3: A Novel DI Method, K+IDS, for the Rapid Screening of

Urine for Organic Acidemias

Introduction

The gas chromatograph-mass spectrometer (GC-MS) combination has been a
powerful tool for the screening of physiological ﬂuids for disease-state identiﬁcation
and characterization. Inbom errors of metabolism in which amino acid, lipid, and/or
carbohydrate pathways may be altered, frequently lead to physiological states known
as the organic acidurias (or acidemias) -characterized by elevated concentrations of
one or more organic acid in plasma or urine. More than 50 organic acidemias have
been identiﬁed, and a number Of reviews have recently appeared on the diagnostic
utility of GC-MS in this ﬁeld. “5’43 In general, identiﬁcation of the organic
acidemias requires the ability to determine that one or more organic acids are excreted
at levels above those that exist in "healthy" individuals. One of the most widely
reported disorders of this type is lactic acidemia. A number of enzyme defects have
been associated with the lactic acidemias, and result in the accumulation of lactic acid
in both plasma and urine.“’15 Diagnosis of this condition is frequently made by
monitoring for excessive levels of this metabolite by GC-MS, as well as by
monitoring the excretion of 2-hydroxybutyric acid.44

Of considerable interest are those organic acidemias resulting from disorders in

45 lsovaleric acidemia, for

46

branched-chain amino acid (BCAA) metabolism.
example, is the result of isovaleryl-COA dehydrogenase deﬁciency, which affects
the oxidation of leucine. This enzyme deﬁciency leads to the accumulation of
isovaleric acid, which itself is neurotoxic to experimental animals. "Stores" of

glycine are normally available to conjugate with a-hydroxyisovaleric acid to form the

45

more water soluble metabolite, isovalerylglycine. In fact, orally administered doses
of glycine are frequently part of the dietary regime for treating this condition. It is
this metabolite that is frequently observed to be excreted in 2000-fold excess relative
to "healthy” conditions and disease-state identiﬁcation is generally made on the basis
of the elevation of this metabolite in urine. In Maple Syrup Urine Disease
(MSUD),47 which results from branched-chain ketoacid dehydrogenase deﬁciency,
elevated levels of 2-hydroxyisova1eric, 2-hydroxybutyric, and 2-hydroxy-3-
methylvaleric acids have been observed, in addition to increased valine, leucine. and
isoleucine plasma concentrations.

A number of alternative approaches to GC-MS has been suggested for situations
in which the goal of analysis is to identify the speciﬁc metabolic disorders involving
the organic acidemias. Analyses that utilize GC-MS are time-consuming since: (a)
derivatization of the isolated organic acids from urine is required to make the mixture
amenable to analyses using gas chromatography, and (b) long GC analysis times are

48 The

required to achieve suitable separation of the organic acid components.
protocol for isolation and derivatization of organic acids for gas chromatographic
analysis was shown to be quite extensive (see Figure 2). Numerous attempts have
been made to simplify the work-up procedures of organic acids by bypassing the gas
chromatograph. Alternatives to GC-MS for screening for organic acidurias have
included LC-lvrs,49 MS-only,50 trivia,51 and TLC-MS 52.

Recent attempts to distinguish between the metabolic proﬁles of various disease-
state and healthy individuals on a more rapid time-scale include direct exposure
probe-chemical ionization-MS. In these studies, the mixtures of organic acids are
introduced directly into the mass spectrometer. Heating, rather than derivatization, is

used to volatilize the analyte mixture. Issachar and Yinonso demonstrated that this

MS-only method using isobutane chemical ionization could qualitatively differentiate

46

between disease-state and healthy urinary organic acid proﬁles. Although this
approach yielded a very rapid alternative to GC-MS analysis, it could not be used to
accurately quantify the components of the mixture. Organic acids had to be
seaparated into ”volatile" and "non-volatile" classes. Volatile compounds were
considered those for which isobutane-CI mass spectra could be obtained at ion source
temperatures of 70 'C. Non-volatiles required source temperatures of 190 'C for their
vaporization to occur. As a result, two individual analyses were required for each
isolated organic acid fraction. This situation posed .problems since operating
conditions (i.e., reagent gas pressure) could not be maintained rigorously constant.
As a result, chemical ionization mass spectra were subject to ﬂuctuations in ion
source pressure and these variations have been found to have a pronounced inﬂuence
on relative intensities of ions in the chemical ionization mass spectra.

Tandem mass spectrometry (MS-MS) has also been suggested as a viable
alternative to GC-MS for complex mixture analysis. Hunt and Giordani53
investigated the potential of this technique for disease-state identiﬁcation of the
organic acidurias. Although this method certainly provides a much more rapid
approach, and often yields data as informative as that obtained by GC-MS, the cost
associated with such an experiment may preclude its widespread use for routine
clinical diagnoses. ‘

K+IDS is evaluated as an extremely rapid and effective "batch" mass
spectrometric approach for differentiating various organic aicdurias. For the analysis
of organic acids from urine, no derivatization is required. The potential of this

method as an alternative to GC-MS for characterizing and differentiating complex

mixtures of organic acids in urine is described.

47

Experimental

Procedures for a typical K+IDS experiment have been described elsewhere."0 In
previous K+IDS experiments, nonvolatile compounds or mixtures were deposited
directly onto a thermionic emitter glass bead. When rapidly heated, analyte
desorption and K+ emission occurred from the same surface, with K+ attachment to
the desorbed neutrals occurring (in the gas phase) to yield the species observed in the
resulting mass spectrum. The probe design initially employed has been modiﬁed as
mentioned in the last chapter; the new design is schematically shown in Figure 16.
The probe consists of two ﬁlaments; one supports the alkali-aluminosilicate mixture,
and the other supports the sample. Filament 1 is resistive‘ly heated and appropriately
biased to generate K+ ions in the gas phase. Filament 2 is positioned a few
millimeters from Filament l and is not electrically heated. Samples experience
radiative heating from Filament 1 and desorb from the surface. The beneﬁts of
incorporating such a conﬁguration for K+ IDS analyses have recently been discussed.
Separation of the analyte from the K+ emitter has resulted in substantial
improvements in sensitivity and much improved quantitative capabilities.

No matrix (e. 3., glycerol) is required for K+IDS, unlike other
desorption/ionization (DI) methods such as fast atom bombardment (FAB).
Appropriate volatile solvents, such as MeOH, CH3CN, and others are used solely to
assist in the transport of analyte to Filament 2. Once transport has been achieved, the
solvent is evaporated and the probe is inserted into the mass spectrometer, rapidly
heated, and the resulting mass spectra collected.

No source modiﬁcations were required to the quadrupole mass spectrometer.
K+IDS analyses were performed in the E1 source conﬁguration, with introduction of

the K+IDS probe through the direct insertion probe inlet. Ion source temperature was

48

New Dual-Filament K+1DS Probe Design

p
S

 

 

 

 

 

 

 

Low current,
Variable Hi h
Voltage, P. .

 

Figure 16. Modiﬁed dual-ﬁlament K+IDS Probe (P). Filament 1 (F1) is
electrically heated using resistive wire (W) and is the alkali ion
emitter. Filament 2 (F2) is the analyte ﬁlament and solely experiences
radiative heating

49

maintained at 40 'C for analysis of the underivatized organic acids. No electron
ﬁlament was employed. A high ﬂux of alkali ions (Na+ or K4) was generated by
rapidly heating the alkali-aluminosilicate mixture to temperatures of 800-1000 'C.
The mass spectrometer was scanned from 50-350 11 at a rate of 0.4 sec/scan. The
mass spectra shown here were obtained by averaging over the entire elution proﬁle
(usually 20 scans or less).

Urine samples from organic acidemic patients and from age-matched controls
were obtained from Meridian Instruments, Inc., Okemos, MI. The urinary organic
acids were quantitatively extracted onto a column of DEAE Sephadex. The

54 A 1.0ml

preparation of the ion exchange column has been described elsewhere.
aliquot of urine at pH = 7 was passed through the ion exchange column at a ﬂow rate
of approximately 0.1 drop/sec. The column was washed thoroughly with doubly-
distilled water and acidic components were eluted off with pyridinium acetate (1.5M).
The collected organic acid fraction was concentrated by lyophilization. The residue

was then dissolved in approximately 2 ml of acetonitrile and approximately 5-10 ul of

the resultant solution was deposited onto the K+IDS probe.

Results of mixture analysis utilizing "dual-filament" probe

In K+IDS, the analyte or analyte mixture is rapidly heated. The mass spectra that
result show that analyte molecules fall into one of three categories with respect to
their behavior upon rapid heating. Type I compounds exhibit desorption rates that are
much higher than degradation rates (at the temperature required for K+ emission by
the K+IDS probe). Upon heating, type I molecules desorb intact, and the (M+K)+

adduct ion is observed in the resulting mass spectrum. Only molecular weight

50

information is provided in the mass spectrum. Type 11 compounds have competitive
desorption and decomposition rates at the temperatures used in K+IDS. Intact analyte
molecules, M, and thermal degradation products, Di’ appear in the gas phase above
the heated surface leading to (M+K)+ and (Di+K)+ ions in the mass spectrum,
yielding molecular weight and structural information. Type III molecules are those
for which degradation and desorption rates are strongly temperature dependent within
the temperature range accessible in the K+IDS experiment. For this class of
compounds, the temperature variable can be controlled to favor desorption or
degradation.

K+IDS analyses of mixtures of type II compounds yield exceedingly complex
mass spectra that would be difﬁcult to interpret for quantitative information. If
mixtures of type 1 and/or type III compounds are analyzed by K+IDS, experimental
conditions can be selected such that each component (Mi) will lead to ion current at
only one m/z value (as well as its natural isotope peaks) corresponding to the
(Mi+K)+ Species. If rates of (Mi+K)+ formation are approximately independent of
the chemical nature of Mi’ the resulting mass specuum will be a "molecular weight
proﬁle" of the mixture. The following sections show that: a) organic acids behave as
type I compounds; b) K+IDS spectra of synthetic mixtures represent the relative
concentrations of the species present; and c) K+IDS spectra can be obtained for
complex mixtures of organic acids isolated from urine. From these ﬁrst studies, we
suggest that these spectra can be used, in many cases, for the rapid screening of urine
samples for the identiﬁcation of organic acidurias.

The results of a K+IDS analysis of myristic acid, CH 3(C112)]2COOH, are shown
in Figure 17, in order to demonstrate behavior typical of the organic acids
incorporating the new probe design. Upon rapid heating, the desorption proﬁle is

fairly short-lived, as shown in the upper insert. Desorption of analyte generally occurs

51

088“

.anoaxwgnﬁv so. a:

 

Firs

eapaaeaaaéoazﬁaaaeaupagcgesioe "spasm
N)...—

8~ 3N emu can 8’ 8p 3.. our
b—qubE—pani—npbn—i-l rb—L nP—P bn—bpbbi
rm
1
V

.u

A

8 9 3

- it.» INI

I.

.3

N

R.

u

+035

 

 

 

 

 

0F

 

 

 

 

52

over 15 scans. Rapid heating of the alkali emitter to approximately 1000 'C results in
the desorption of the intact acid. Alkali ion attachment yields an (M+K)+ ion at mlz
267 as the only ion observed at mlz values greater than those for the K+ ions at mlz
39 and 41.

There are, of course, many chemical events that can occur when a mixture is
heated, even if thermal degradation does not occur. One possibility is reaction
involving two or more components, yielding species not originally present in the
mixture. To test this for organic acids, a synthetic mixture of ﬁve fatty acids was
studied. In the analysis of fatty acids, organic acids, and other analytes, it was found
that better sensitivity is observed when Na+ is used in place of K+. The experiments
that follow will all utilize Na” (i.e., Na+IDS). The only difference is that the
thermionic emission material is an aluminosilicate material made with sodium oxide.
Figure 18 shows the Na+IDS mass spectrum for a nearly equimolar mixture (c.a. 4
nmoles each) of ﬁve fatty acids. The (M+Na)+ ions are observed for each component
of the mixture. The ﬁve ions are Of approximately equal abundance, reﬂecting the
equirnolar nature of the mixture. A more accurate representation of the mixture is
obtained from the data that results from integrating over the entire elution proﬁle.
The integrated areas for each of the ﬁve (Mi+Na+) ions over the entire reconstructed
mass chromatograms reveal deviations from the mean relative concentrations that
are no greater than 10 per cent. NO ions suggesting reaction between these
components are observed. Thus, it is believed that the Na+le analysis of this simple
mixture shows that the Na+le spectra of more complex mixtures will reﬂect the
distribution of the components of the mixture separated on the basis of their
molecular weights, thereby providing a molecular weight proﬁle. In fact, this

behavior is observed even for a complex synthetic mixture of steroids, organic acids,

53

do. as 8e 2.02. .2
.O .36 «To «70 no cause Became a. Co 8.68... are warez "a 85m

 

 

 

 

 

 

NE
8» 8a 8a 8" can 8a
b I b —lth P b h — P n (m htt— .1 n m b tH DHthlbtﬁrlttpit.
11+ ._ __ __. ._.
-m
can
u
IA
3
M
.1.
a
N
H no _ _ .m
32.2
f Lozeaé k .
Lozeésa

Loy—Ks: .

 

Luz... £5.

 

 

54

and fatty acids. The same ﬁve fatty acids discussed earlier were added to a mixture of
low molecular weight organic acids and steroids. The combined mixture was placed
onto the sample ﬁlament holder and a K+IDS analysis performed. The results of such
an experiment are summarized in Figure 19. The relative intensities of the fatty acid
mixture are maintained, even in the presence of other simultaneously desorbing
species. There is no evidence in the mass spectra of mixtures of this type for
chemical reaction between species. Molecular weight information is generated solely

for individual components of the synthetic mixture.
K+IDS for urinary organic acid profiling:

For simple mixtures, K+IDS (or Na+IDS) mass spectra not only represent the
individual components of the mixture, but also reﬂect the relative concentrations of
these species. This information is available on a very rapid time-scale. The results
presented above suggest that this technique may be extended to the analysis of
organic acids in urine, for the rapid differential diagnosis of various metabolic
disorders.

Patients exhibiting the condition of MSUD are found to excrete abnormally high
levels of 2-hydroxyisovaleric, 2-hydroxy-3-methylvaleric, and 2- and 3-
hydroxybutyric acids in urine. Data from GC and GC—MS analysis of the derivatized
organic acids of MUSD urine were obtained in collaboration with Chamberlin, e;
31.55 and are summarized in Table 1. The most abundant organic acids in the
mixture are identiﬁed and their concentrations relative to the internal standard, tropic

acid, are given. Based on the information presented in Table 1, it is possible to

55

 

u \E
can can. can ONN OHN can on. O.—
E)? t . . .__.. _ ﬁe .

z z _ l ..

m i

H.003...» u an I

«+32. 2

Petra t

H... 92.95.: . . ~+02+n3~ .-

 

 

5.8835315? Eugene
iguana—ea gnaoonuoguoeanougmeﬂz "323E

 

 

 

 

 

 

 

 

 

 

 

 

 

EONVGNIIHV EAILV’IEX

56

8%....- .
lee 82.38... no. lo: cal-:8 39:95:02.8 .332.- uewﬂoww A13

.8223a55n2oeoé.§.u383

.9. .38 .8... 3:3... so... 39...... o... 3.6 8:33... 25.3. 3
discloses-55:52.5

:- has. so. one. .935. its... 1. o. :39. 388 o... 8....> 3

a.» uv
n a an non :3 08.3qu .vn
2 S o: 2.52.: . .3
a... an men 2.— 0389-33926 .3
N. m an o— NVm o—«ool—hﬂoanhnoL—vhntv .— N
nu mm mm M on n 3.30.3... on
we ha mm— 02 Seesebnn an
3 me mu «2 2.8.3.0264. O." 3
h? vohN nmn an" 0133-0 bu
ad mm o2 oeo1s_2.oe.n.o.o._.« 2
3. 8 M2 o2 c.aoouaﬁxooeé 2
e. m u n v.— oS 33.—553553... z
a; 2 n1 2: Secohﬁueoeé 2
«.n S 3 a: 35.5.5.2 2
3 an a: m: 9.33.5 3
"a Gav — on mg m oto—akiiuor. canon @—
an on a u 3 mm a ate—55355.. ones a
3 9 ma an e: etc—86.8.." a
we can a n a as .33 U... h
c.a. 3. 52 1: 3.5313334” c
a.» mm mm— 42 passages..." m
a... a... 2“ n2 oceaeoaxé v
N. r 3. n: ca 9:35 n
a? ea 3 WM 2 “Wm..— u
a a reassess m3 inure“ w

exaggeaooeﬁ 8a
8a. can: a sec. oaaeo .583. .8... use. 2.3.... ._ use.

57

construct a simulated molecular weight proﬁle of the major components in the MSUD
urine, as would be expected in the Na+IDS mass spectrum. For each compound
whose concentration and molecular weight is known, 23u is added to simulate Na+
attachment to the neutral molecule. All species of the same molecular weight will be
indistinguishable in such a representation. The relative intensities of all components
of the same molecular weight are added, giving a summed intensity for that particular
mlz value. This simulated Na+IDS spectrum can then be compared to the
experimentally obtained one. Shown in Figure 20 is this reconstructed composite
mass spectrum reﬂecting the identiﬁed compounds from CC and GC—MS analysis.
Although isobaric compounds may not be differentiated in such a representation,
information may still be gleaned from perusal of the molecular weight distribution of
the organic acids from this urine. The designate ion for this metabolic condition is
found at mlz 141, corresponding principally to [2-hydroxyisovaleric acid + Na+].
Confirming ions for this disease state are at mlz 127 and 153, corresponding to Na+
adducts of 2-, and 3-hydroxybutyric acids, and 2-keto-3-methylvaleric acid,
respectively, based on their intensities relative to the Na+ adduct of 2-
hydroxyisovaleric acid.

The Na+lDS molecular weight proﬁle for the organic acids isolated from the
urine of a patient suffering from MSUD is shown in Figure 21. As expected, the
major organic acid metabolite is 2-hydroxyisovaleric acid, leading to the signal at mlz
141 (the [M+Na+] adduct). The confirming ions for the disease-state are also evident
in the mass spectrum. The similarities between the simulated Na+ IDS mass spectrum
of the mixture of organic acids from this urine based on the available GC and GC-MS
data and the experimentally obtained Na+IDS mass spectrum are quite striking.

Slight differences in the two proﬁles might be attributed to the fact that not all of the

58

.353

garagmsﬁzaiﬁuaataﬁsagaﬁaas

.Ssegnuoaeoépaauvaaaosoa ”8+2 3:33.... ”89.5.—

o& 8, 92 on. 8. . o:

 

...4 ._.. .J_._p._...ﬂ .ﬂ ._.A ..~
3“ 0:.

 

 

 

 

 

 

hﬂr

 

S;

r

I

r

T I
MISNSLNI SALLV'IBH

 

 

59

£552 65 “Evan «88 89a

 

 

 

 

 

 

 

.32 3.38838“. 35 Bang. «8+2 8:38 >38; "Roam
N):
8N on." 9%.. awe 0%.. o1. OT
-quuum— .1 -.J.— — an— cu; J? u qu——dm———:Anq1—‘—=w:Mni—m-— - a _1 1134— n—ll- u q n— 4

ﬂ . app.

NON r

8.. ﬁ

3..
w:

 

 

 

 

MISNSLNI §A|1V138

60

components from GC and GC-MS analysis have been identiﬁed. Unknowns cannot
be included in the simulated molecular weight proﬁle since no molecular weight
information is available for these species to date.

Obviously a molecular weight proﬁle of a complex mixture of organic. acids will
not allow isobaric species to be individually identiﬁed. Thus, quantifying species, on
an absolute scale, is not always possible. For the diagnosis of the organic acidurias,
however, there is typically one component present at a concentration that is extremely
elevated in urine, which may be used to identify the disease state. The relative
concentration differences between a "healthy"-state and disease-state is often the
information of interest when screening for organic acidurias. Targeting for a speciﬁc
urinary organic aciduria based on the elevation of a particular organic acid allows this
direct method to be useful for differential diagnosis.

The average Na+lDS mass spectrum of the urinary organic acid from healthy,
lactic acidotic, and isovaleric acidemic individuals are shown in Figure 22 (a-c),
respectively. Inspection of these molecular weight proﬁles certainly suggests that
disease-state and healthy urinary organic acid proﬁles can be differentiated easily.

The lactic acidosis condition is characterized by excessive levels of lactic acid in
both plasma and urine. Pyruvate dehydrogenase deﬁciency, an enzyme defect
frequently associated with lactic acidemia, causes accumulation of this organic acid in
urine. The Na+IDS mass spectrum of acids isolated from this urine indeed suggests,
that lactic acid which appears as the Na+ adduct at mlz 113 is the major excreted
metabolite. Summarized in Table 2 are the prominent organic acids from the Na+IDS
mass spectrum and their relative concentrations (obtained from available GC and GC-
MS data). The relative concentrations of the organic acids are in good agreement, in

general, with the ion abundances observed in the Na+IDSomass spectrum.

6|

 

' “3 (A)

 

 

 

 

 

 

 

 

 

s ‘ m
.m 121
113
. (a)
l
E 127 m
a. 202
r
LA I
‘ 182 (c)

RELATIVE

 

 

 

 

‘ 202 218
in. nrJ'. L: I. 1.1 '_ I'L...l.4.l.- . _ ‘
110 1J0 1&0 ' ' '1; 1 210 230
Ml:

Figure 22. Averaged Na+IDS mass spectra of organic acids from (a) healthy,
(b) lactic acidosis, and (c) isovaleric acidemia urines

62

6.323» 358.: a... 3 «>238 822' 52233.39 2:3 can 832.» 932953.. a...
no 53 92.00 can DU :5...— eocmEcote 9.... «Eva 35MB .3.— 82a> 5:828:00 “_u

«cm
mu—
m:
m:

«an
3
ha—

n:

mun
«an
Nan
«3

was
«an
a:
h:
n:

fag

can 325%:
no 3.3.:_u.m._.oE.n.axe._eb-.n
8m— o_:_oo=m 622-515»:
a . £36 6:2...
own 3.52::
E. 35926 die—gazeuebﬁn Juno—geaixecggé
we Subgeaixeegné .e_..b=aem_.~xe.€3.n
drainage—.26 .utbanmxeceagé

3 9:36 .9221—
8... argue...— .935
8 3.59::
can camera—reign—
cvn 3.59:5. 63855.5
8 3.593— .335
S— 3335
. no 353.5 ate—ghueuemaé Juno—gonmaxeSb—é
o2 92.33.92.525,” .otbanhxecvhgé
cow emeeiecnmxeeemaé 6:85 .9225
«gun—Eu g

65.5 £835»? s «38 8.3.5 “533.. .38 :8 05
a. 3% mzoo 52.. Raga 85258. 28 as? 3.22 .N 23

A3

A3

2:

3

63

Similarly, the isovaleric acidemia condition may be identiﬁed easily from
extremely elevelated levels of isovalerylglycine excreted in urine. The reconstructed
total ion current (TIC) proﬁle (Figure 23) from the GC-MS analysis of the derivatized
organic acids of the isovaleric acidemia urine shows the massive accumulation of this
metabolite. The Na+IDS mass spectrum of the underivatized organic acids from the
same urine reveals an identical situation. The base peak in the Na+IDS mass
spectrum is due to isovalerylglycine. (mlz of [M+Na+] = 182). The major organic
acids identiﬁed, and their relative concentrations are also summarized in Table 2.

The results are not surprising, since numerous reports on this condition suggest
that isovalerylglycine may be excreted at levels more than ZOOO-fold over those in
healthy individuals. To confirm that the Na+IDS mass spectra provide reasonable
quantitative representations of the actual organic acid compositions, a simulated
molecular weight proﬁle was constructed for the isovaleric acidemia urine, in a
manner identical to that described for the MSUD urine. The "simulated" molecular
weight proﬁle in Figure 24 is compared with the experimentally obtained Na+IDS
mass spectrum (see Figure 20). In general, the proﬁles are in good agreement.

The results from a Na+IDS analysis of the control urine reveals the major
metabolite of this urine to be hippuric acid (mlz of [M+Na+] = 202). Again, the
results from the Na+le analysis parallel closely those obtained from GC and GC-
MS.

The feasibility experiments presented here suggest that the K+IDS (or Na+ IDS)
method is capable of differentiating between some of the more commonly occurring
organic acidurias such as lactic acidemia, isovaleric acidemia, MSUD, and controls.
However, one might expect that this method would fall short in its utility for

distinguishing between those disease conditions in which the principal metabolites are

64

 

J TIC

+— Isovalerylglyc In.

 

 

 

 

_u .00. . A k .44 0‘ AA..- o.“ A..-A“- “b~‘4

340 540 740 940 1140 1340 1540 1740 1940

 

 

Time (see)

Figure 23. TIC proﬁle of the trimethylsilyl derivatives of organic acids isolated
from isovaleric acidemia urine

65

a3? 3838a 55 503385 $243803“?
“advance 8.08 mm 05 we each. “ammo? .3329: warez 3898mm 13a 0.59m

 

NE.
emu 05 8.. our . amp our 0:
— Vb n n _ n - 1n nit—IL - b — b n n n — b1 h min b b b lm'P #
- -1- __ 7 _3 a-.. ___l. _—.. 141. -7
m3

 

N2.

 

 

 

MISNalNI SAILV‘ISH

66

of the same nominal molecular weight (but each arising from different metabolic
pathways). A suitable example describes the capacity of the Na+le technique to
differentially diagnose methylmalonic acidemia and MSUD. In the two cases, the
major metabolites, methylmalonic acid and 2-hydroxyisovaleric acid, respectively,
have molecular weights of 118 Daltons, leading to the (M+Na+) adduct at mlz 141 in
a Na+IDS analysis.

Can these two disease-states be differentiated, even though the major metabolites
associated with the conditions are of the same nominal mass? The answer is found by
examining not only the designate ion for the diseases but also the conﬁrming ions
expected for each disorder. In the case of MSUD, these species have been identiﬁed,
and their intensities relative to the peak at mlz 141 have been discussed. For
methylmalonic acidemia, on the other hand, these compounds are of only relatively
low concentration and do not lead to prominent ions for this disease-state. The
Na+IDS mass spectrum of the organic acids isolated from the methylmalonic
acidemia urine are shown in Figure 25. Other ions in addition to methylmalonic acid
that have been identiﬁed in the Na+IDS mass spectrum are summarized in Table 2.

In cases where the differences may not be entirer obvious, K+IDS may be
utilized in mass spectrometers that provide additional features. For example, if
information is required on isobaric ions, a high resolution mass spectrum may be of
utility. If true isomers must be differentiated, K+IDS as an ionization method with
subsequent MS/MS analysis may be a viable approach. The feasibility of the K+IDS
experiment on simple mixtures using a high resolution JEOL I~IX110 double-focusing

mass spectrometer is currently being evaluated.

 

67

 

 

.NON

S;

 

AllSNalNl SALLV'IBH

 

 

68

Conclusions

Over 50 organic acidurias have been identiﬁed and only a few of the more
commonly occurring organic acidurias have been shown to be quickly and easily
identiﬁed using the desorption/ionization method of K+ IDS. As a screening
procedure, this method appears to be a viable, low-cost and extremely rapid
alternative to GC—MS. Although not as useful for studying metabolic pathways, when
combined with more advanced systems such as high resolution mass spectrometers,
these questions may be addressed. Future work will deal with extending the dynamic
range of this method for such analyses of complex mixtures.

The ﬁnal question that must be answered is whether or not these K+ IDS
molecular weight proﬁles are may representative of the components of a complex
mixture. Comparisons between simulated molecular .weight proﬁles and those
obtained experimentally suggested that K+ IDS mass spectra do indeed represent
fairly accurately the components of a mixture. Simulated molecular weight proﬁles
were constructed based on results of the known, 25 most abundant species from
available GC data and compared to the experimentally obtained K+ IDS spectra.
However, a number of unknowns exist in the gas chromatographic libraries of the
organic acids. These unknowns can often be fairly concentrated components of urine,
and when their concentrations are great, these unknowns must be taken into account.

The best way to go about elucidating the identity of the unknowns from CC
analyses is to combine GC with mass spectrometric detection. In the production of
simulated molecular weight proﬁles the molecular weight of the unknowns is what is
desired. As will be shown in the next few chapters, EI mass spectra of derivatized

organic acids frequently are devoid of molecular ions. The technique of NHS-CI

69
mass spectrometry is capable of providing abundant molecular weight information on
organic acids.

To better evaluate the quantitative capabilities of K+ IDS (Na+IDS) for metabolic
proﬁling, the urinary organic acids from another patient suffering from
methylmalonic acidemia were isolated and lyophilized. The organic acid residue was
dissolved in approximately 2ml of acetonitrile. One half of the solution was
evaporated to dryness and derivatized with BSTFA/Pyr to form the trimethylsilyl
derivatives. The other half was stored for "batch" analysis by Na +IDS.

The derivatized organic acids were analyzed by both EI and NHa-CI in order to
obtain molecular weight information on the major organic acid constituents. Shown
in Figure 26 is the reconstructed TIC chromatogram of the derivatized organic acids.
The major component, as expected was methylmalonic acid (di-TMS). A major
component which eluted late during the GC-MS analyses was found to have a
(derivatized) molecular weight of 570 from NH3-CI results. This compound was
found to have a total of ﬁve TMS groups, as determined from analyses using the
deuterated analog of BSTFA. From these data, the underivatized molecular weight of
species was determined to be 210 11. In addition, a number of molecular weight
determinations were made on other unknowns contained in the mixture. This
information was useful for comparing the molecular weights of all species contained
within the mixture with the molecular adduct ions observed as a result of the
susbsequent Na+IDS analysis. The GC-MS analysis was followed by the "batch"
Na+IDS analysis on the other isolated organic acid fraction. The averaged Na+IDS
mass spectra of the underivatized organic acids is shown in Figure 27. The base peak
in the mass spectrum was the [M+Na+] adduct ion of methylmalonic acid and other
isobaric species were found to contribute, to a small extent, to the overall abundance

of this ion at mlz 141. The unknown having a MW = 210 was also observed as the

'70

Azanzgaoaaaagueuvsnxa
. 82a 358m .38 as? do 8%53 3338.5 .6 eaea on. ”em BEE

coom

camp 00—" com 005 000 can bop

....._...._t..._...._...._....

 

 

 

 

a?

 

 

 

 

e_¢0saeaaxeevl« c.a.-4

/ r
e—eo_costlzon n

/ 07.2.3.3...
o.sa_o c.a.-ssh r

0 _ t_. 0.80.23th

 

; /-

 

 

 

 

 

 

 

o.eo.2§.xa.oa

 

7|

“anaeaaauaaaawnapaxuaaa.
“a:axeaaaszagaceugasgauuazxaee.anxw.aaamantgaunseuz‘ “Kwanzaa

N_\\cc

onN onN opN our our on. on" OFF. 00

a, ._k_ a. a. grad. 5...? M___..,_m___=1. u.....___u__a.:l ii

i.

 

 

 

 

 

 

> t
W I
on" I 32 0.0.3.0 0.500%.0
cues—e: \
a e_eee305su
o_sa~o .o.eoes:p r.

 

k
occooo.oeoo:.o

o_so.o»au05va£
.e.e..np«»nwu

 

 

 

 

72

sodium adduct ion at mlz 233. The Na+IDS mass spectra in general accurately
reflected the true composition of organic acids in this disease-state condition.
Integrated areas are compared to relative ion intensities for the most abundant organic
acids in this mixture and are summarized in Table 3.

The one major differences in these two proﬁles is thath large peak corresponding
to phosphoric acid was observed in the reconstructed TIC chromatogram but no ion
corresponding to the sodium adduct (i.e., mlz =125) was found in the Na+ IDS mass
spectrum. A possible explanation for this difference is as follows. Results from
negative ion surface ionization mass spectrometry have suggested that compounds
such as these can readily decompose on "hot" surfaces, resulting in the formation of
the P03. and P205. anions. 37’“ These species arise from decomposition on the
surface followed by electron attachment from a surface of relatively low work
function (i.e., CszOzA1203:SiOz). Na+ adducts of these neutral decomposition
products were not observed and would not be expected to undergo addition reactions
with alkali ions.

The capabilities of 16le for mixture analysis has been demonstrated for
compound classes as varied as polymers, synthetic mixtures of fatty acids and
complex mixtures of urinary organic acids. In addition, the relative concentrations of
fatty acid components have been found to be unaffected by the presence of other
compound types in a K+IDS analysis. The ability to derive quantitative information
from mixtures containing several different classes of compounds appears to be
possible using this technique. Future studies utilizing K+IDS for the diagnosis of
organic acidurias might include a "true” batch mixture analysis of the components of
urine. Work-up procedures might only involve the precipitation and acidiﬁcation of

inorganic salts without isolation of any particular class of compounds prior to their

73

Table 3. Comparison of relative intensities from Na+IDS analysis and peak
areas from GC analysis for the ten most abundant organic acids rn

 

 

methylmalonic urine
Part 0 Compound Amﬂc_ Rel. Am mlz mmn’l Rel. Int.

1 Lactic 43500 9 us

1 Z-hydroxybulyric 41000 7 127

3 methylmalonic 501000 141

4 succlnic S3000 mo 14! 100
s sliced: 111000 20 m rs
6 2-deoxytetronlc ' 47000 e ' 143 it
7 threonlc I 18000 2! 159

I erythronlc 221000 39 159 3‘
9 citric 275000 49 215 es
lo UNK MW e 210 116000 at 233 20

74

lyophilization. Differentiation of organic acidurias by 16le might be achieved
because these conditions are typically characterized by extremely elevated
concentrations of one or more components in urine. The main problem with such an
analysis is that urine has a high water content. As was alluded to earlier, water
appears to be an undesirable transport medium for K+IDS studies, independent of the
alkali oxide material utilized.

The real strength of this technique lies in its capabilities for producing a rapid
"ﬁngerprint" proﬁle for various physiological conditions. It has not been developed
to the extent to which it could be used exclusively for diagnosing these organic
acidurias. Therefore, at present time, other methods, such as gas chromatographic
analysis would be required in order that unequivocable diagnoses be made. In
general, the K+IDS technique shows great promise for mixture analysis. As a
desorption/ionization method that is free from matrix interferences and requires no
modiﬁcations to existing mass spectrometers, this method will no doubt ﬁnd greater
acceptance in the area of analytical chemistry. The present design has improved
dramatically the detection limits of the technique but work still remains in improving

the dynamic range of the technique.

75

Chapter 4: Extension of the Utility and Applications of K" IDS

Research efforts have also been concentrated on better characterizing the K+IDS
technique as to its "universality” and/or limitations for organic compound analysis.
These types of studies have all been performed in the context of how the K+IDS
ionization technique compares to other commercially available DI methods. A major
emphasis of the research in this laboratory has been to demonstrate that K+IDS may,
in many cases, produce comparable mass spectra, and be a very cost-effective
alternative to other more complicated and DI methods. Recall, K+ IDS can be
implemented on instruments as simple as table-top quadrupole mass spectrometers.
The only requirement is that the instrument be equipped with a direct insertion probe
inlet system. Typically, other D/I techniques require sophisticated instrumentation.
In addition, techniques such as FD and F1, for example, require a great deal of
expertise in the production of the ionization emitters required for non-volatile

analyses.

76

Organic synthesis on the K+IDS emitter surface

Desorption/ionization methods such as LDMS and FAB frequently utilize a salt
(e.g., NaI) as part of the target containing analyte for the purpose of producing
molecular adduct ion species that are frequently required in order to confirm the
molecular weight of unknowns. Cationization of analyte species occurs in the gas
phase. 57 Frequently, reactions between the two species can occur in which labile (or
acidic) hydrogen atoms are replaced with alkali atoms, r leading to ions of the type
[M+ 2A -H]+, where A = Li, Na, or K. This phenomenom has been discussed for
reactions involving glycerol and various alkali salts by fast atom bombardment. 57
Under certain conditions, the K+IDS technique has been found to produce similar
species and is capable of desorbing salts into the gas phase whereby they undergo K+
ionization.

When a salt solution that is deposited onto the K+IDS probe is rapidly heated,
cationization of the desorbed salt may occur. Salts of the type R-COzNa have been
found to desorb intact upon rapid heating and readily undergo adduct ion attachment
to form the [R-COzNa + Alkali ion]. This process is generally not observed using
other desorption/ionization methods. The desorption of these salts, though, is not
always complete and memory effects can carry over into subsequent analyses. These
memory effects are easily eliminated by taking care in the cleaning of the analyte
ﬁlament before another analyte solution is applied. Unique chemistry can occur on
the surface of a emitter that has not been totally free from salt contamination.

A disulfonated, biphenyl sodium salt was submitted for analysis using the
technique of K+IDS since no molecular weight information was attained using other

ionization methods, such as FAB. The results of analysis by K+ IDS yielded

77

molecular weight information on the molecule through K+ attachment to the desorbed
species. Following the analysis of this organic salt by the K+IDS technique using
the two-ﬁlament design, a mixture of ﬁve [:12 fatty acids (C-12,13,l4,16, and 17)
was placed onto the analyte ﬁlament and analysis by K+IDS was performed. The
results of analysis by K+IDS showed two distinct analyte desorption proﬁles, the
second desorption proﬁle occurring approximately 6 scans (c.a. 3 sec) following the
desorption of the ﬁrst species. Heating rates employed in this experiment were
approximately 250‘C/sec as the ﬁnal current chosen was I = 3.0 A (ﬁnal emitter
temperature approximately 1100-1200 'C). The power supply was turned on before
the tenth scan; the analyte ﬁlament was presumed to continue heating long past the
time of the second analyte desorption. The TIC proﬁle and reconstructed mass
chromatograms of mlz 309 and 331 (from heptadecanoic acid) are shown in Figure 28
below. The reconstructed mass chromatogram of mlz 309 represented the [M+K+]
ion for heptadecanoic acid; m/z 331 represented the [ (M+ Na - H) + K+] ion. A
comparison of mass Spectra obtained from the individual analyte desorption proﬁles
revealed that all the components of the free fatty acid mixture experienced
competitive salt formation prior to gas-phase K+ ionization. The averaged mass
spectra are compared in Figure 29 a-b. The mass spectrum of the ﬁve fatty acids
obtained from the ﬁrst desorption proﬁle ) is represented predominantly by the
[M+K+] ions, as labelled M1’ M2, ..., M 5.‘ The averaged mass spectra from the
second desorption proﬁle showed that each of the ions were offset by 22 11, suggesting
a hydrogen-sodium exchange. This process is described in more detail for palmitic

acid as shown in equation 2.

CH3(CH2) 14(10211 + ”Na" ----------- -> CH3(CH2) 1400211111 + "H” (2)

. 0335.": “00.3
80.3.2 g3.33§s§..§m9+2 .3358»

:- 67 .3- .2- .20 co 8.02.. 80.5.3 0:. e53. mar-M "895»...

C. 0.

78

 

 

 

 

 

 

 

 

 

 

 

 

 

on em 0
a I
.3. as r
.23.. taco... ”-

n L b i ii
men .03 laced ..

P p 1
w
_r
I

«an .«3 530..

 

 

79

agsgﬁaaceasaéizoapasggaﬁ
coasonevaﬁeugaeeegcoeessaauam8+ueom§>< "0.5mm

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

as» one can on» 0% 2.0 one onu 03 00‘
Pulirptu phi-trust » run-p» rip-rplrip.»

A Fart; — _ d _ _1 # . _l_— r

/ _ r

a HMS-«2+ S: . .l

+§+£52+w§ + ﬁ-

3 -

baa-.203. +39..an +Eé..z+€_ 8-8-3523 -

on» 00» can can 000 2.0 000 one 30 0!. .

_.-.._..-._.. -..- .TIL-.1.FILL-IL...-

_ _— q— 1.1 _ _ t

Luna ....

+97% .-

+U~+va LOT-NE 000 I
+§+nE 8-3 I vowEo>< 38m

 

 

 

 

80

Sodium, present in some form on the emitter surface readily undergoes an exchange
reaction to form the fatty acid salt. The formation of the salt species is not believed to
occur through a gas phase exchange reaction since two distinct desorption proﬁles
were observed for the clean mixture originally deposited onto the K+IDS probe.

These results were quite surprising. TIC chromatograms revealed that salt
formation followed by K+ addition was competitive with [M+K+] formation for the
free fatty acids. In addition, the mass spectra of the fatty acid salts revealed no
preferential surface chemistry between any one of the 5 fatty acids since relative ion
intensity ratios comparing the free fatty acids and the fatty acid salts were
maintained. Fatty acid salt formation was believed to be a result of a reaction on the
surface of the analyte filament that was contaminated by the previously analyzed
disulfonated, biphenyl salt. Presumably this salt did not completely desorb and
coating of this salt to the rhenium filament surface was sufficient for the facile
exchange reaction to occur.

Similar results for salt formation have been obtained when alkali glass materials
other than the aluminosilicate alkali material are used. KZCO3 has been used and
investigated as an alternative glass material for K+IDS analyses. In general, these
materials are not nearly as long-lived as the aluminosilicate glass materials, though
K+ formation from 1(2CO3 occurs quite readily and K+ ion currents are quite large.
K2003 and other materials (e.g. CH2=O) are being investigated for the purpose of
enhancing the local K+ ion "plasma" and local ion source pressure (just directly
above the potassium emitter material). Recall, higher local ion source pressures were
found to improve detection limits of the K+IDS technique. Several mg of K2003
were deposited onto the filament containing a thin coating of Na20:8i20:A1203 to

determine what effect, if any, the addition of this salt had on ion formation by

Bl
Na+IDS.

Stearic acid (C17H35002H, c.a. 10 p g) was deposited onto the second filament
and the Na+le analysis was performed. At low mlz values, only K+ was formed in
high yield, apparently overshadowing Na+ ion production from the aluminosilicate
mauix. Similar to the results described for the fatty acid mixture in Figure 29 above,
two analyte desorption proﬁles were observed when this compound was applied to the
"clean" rhenium filament. Again, these two analyte desorption profiles were found
to include both the free and the salt forms of the C-18 fatty acid. Furthermore, the
extent to which stearic acid had undergone alkali exchange to form the fatty acid salt
was dependent on the operating current chosen for analysis. Salt formation was found
to be inversely proportional to applied heating rates to the K2CO3 emitter, as shown
in Figure 30 a-c. When an operating current of I=3.8A was employed in the analysis
of stearic acid, only a small fraction of the free fatty acid was converted to the fatty
acid salt. Intermediate free fatty acid and fatty acid salt desorption was found when
operating currents of I=3.35-3.55A were employed. Salt formation was found to be
favored for slower heating rates, but typically occurred at the expense of total ion
abundance for the analysis, as a dramatic reduction in signal intensity was observed
for identical sample loads when comparing TIC "counts" for the three analyses.

The mass spectra of the two analyte peaks observed for the K+IDS analysis
performed at I=3.1A were compared. Only the [M+K]+ adduct ion, mlz = 323, was
observed for the free acid form, corresponding to the ﬁrst analyte signal. The second
analyte signal was composed of the fatty acid salt, mlz [M+2K~H]+ = 361 as well as a
number of other ions, including the [(M+Na-H)+K+], mlz = 345, and the free acid
potassium adduct, at mlz 323. The individual mass spectra obtained for both analyte
proﬁles are best represented by the three-dimensional mass spectral map of Figure 31
delineated in the following manner: the X-axis being deﬁned by a 280-380 u mass

82

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

, (A) l- 3.1A

l “Ilia =60“.

“ I‘!.“l Iluiﬁfijl‘i ‘L‘I‘Iili 113111}.Uﬁ‘lﬁhl.1111‘;1 11".

‘ (B) I = 3.35 A

. 100 % = 21168.

"r1ﬁ..vva.l.v1111.1111L1111l11twlrmlrvvvirvvvlwvvj;w

1 (C) l= 3.8 A

. 100 % = 30144.

”T 1.1—1 T ‘I..l.‘ T1 1LT 1 “I 1"" T1 Il‘ I l1 1.1 “I 1]”! “I Il‘T
Scan Number

Figure 30. K+IDS TIC analyte desorption proﬁles for stearic acid (10 pg)
obtained at (a) I= 3.1 A, (b) I = 3.55 A, and (c) I = 3.8 A

83

‘ CCU”

e8 389E gouge 8353 23 05 we «been» «a 30%

on

DD
0”
DD
Oh
DB
00

.owﬁmlauag Bowmémaaoungasz .335
3ng

"\6

COM Ohm COM 0“” 0V” 0”” ON” 0“” 00” CON 0.“

 

0v.
0v.

 

. ﬁt"
..HOC

84

window; the Y-axis, represented by the normalized ion intensities; and the Z—axis
represented by the scan number (correlated with time and temperature). Early in the
analysis (c.a. scans 35-50) only the [M+K+] ion of mlz 321 was observed for stearic
acid. Between scans 60-70, a mixture of [M+K+] and [M+2K-H]+ ions were
observed. Clearly, rates for salt formation and desorption were now found to be
competitive with those for free acid desorption. Unlike the situation described earlier
for the ﬁve component fatty acid mix, in which the second peak was solely
represented by the adduct ions of the salt species, the second analyte peak was not
only represented by the salt species, but by the free form of the acid, as well. These
two species found in the second scan region were separated in time as well, thus
lending support to a surface reaction rather than a gas-phase process. Possible
explanations for this behavior are proposed which support a surface mechanism. In
both the examples, salt desorption was found to occur at late times during analysis. A
possible explanation for this behavior is that salt formation on the surface requires
time. Conversion of the free acid form to the salt form may be nearly exothermic.
Secondly, salt formation and desorption require high temperatures. Temperature

studies have been performed in this laboratory58

to determine how rapidly the second
ﬁlament heats based solely on the radiative heating that is experienced from the
electrically heated alkali ion emitter. These studies have revealed that the second
ﬁlament continues increasing in temperature for several minutes after the alkali ion
emitter reaches its maximum temperature. Salts of carboxylic acids are known to
have higher heats of sublimation than the corresponding free acids. The TIC
chromatograms suggest that higher temperatures must be required for desorption of

the salt species.

For stearic acid, a K2003 solution was applied to the alkali emitter surface in the

85

form of slurried suspension. It is likely that upon rapid heating of alkali emitter that
sputtering of K2003 to the analyte ﬁlament occurred which allowed this salt to
participate in hydrogen-potassium exchange reactions with the fatty acid. Regardless
of the ﬁnal temperature chosen for the alkali emitter, sputtering no doubt occurred to
some extent. When lower heating rates were employed, the analyte molecules ( i.e.,
stearic acid) resided on the ﬁlament surface for a slightly greater amount of time than
when higher heating currents were employed (I=3.8A). The increased residence time
on the ﬁlament surface using low heating currents (I=3.1A) might be sufﬁcient for
reactions between K2C03 and stearic acid to occur. It is pr0posed that using high
heating rates, the residence time of stearic acid molecules on the surface was
sufﬁciently short-lived such that salt formation from the free acid could not occur.

Another explanation for salt formation may be that desorption of analyte is
occurring from the bulk of the sample ﬁrst and only when the surface monolayers are
reached does the desorption of salt occur. This must be considered as a possible
operative mechanism for the situation involving the deposition of the free fatty acid
mixture onto the contaminated sodium sulfonate surface. Reactions between analyte
and alkali atom (in some form) would presumably have a greater time to interact due
to desorption from the bulk prior to desorption from the surface monolayers.

The scope of compounds amenable to the K+IDS method has been extended to .
the analysis of salts, as they are found to be capable of desorbing intact into the gas
phase. The utility of K+IDS in the analysis of these salts might be unparalleled
Frequently, compounds must either be converted to a free form, or extracted from
buffer solution prior to analysis by other desorption ionization methods For GC—MS
analyses, these salts must be converted to their free form prior to derivatization and
subsequent analysis by GC-MS. The reactions observed on the surfaces of these

K+IDS emitters might offer a method for performing in situ derivatizations, or

86

determining what functional groups are present. Here, salt formation gives the
indication that the compound under analysis contains a carboxylic moiety. This type

of behavior would not be observed for oleﬁns, for example.
K+ IDS analyses of cardiac glycosides

Cardiac glycoside are of interest to the biomedical community due to their
accepted use in treating conditions of arrythmia and congestive heart failure.59
Numerous mass spectrometric techniques have been shown to provide both structural
and molecular weight information on these molecules. These include FAB-MS, 60
FD—MS, 61 Fl-MS, 62 and LD-MS. 63 K+IDS is demonstrated here as a low-cost
alternative DI method for obtaining both molecular weight and structural information
on these thermally labile compounds. 1

The potential of K+le for cardiac glycoside analysis was realized when
digitonin (a widely used drug for treating arrythmias) was deposited onto the bare
rhenium wire of the dual-ﬁlament K+1DS probe and the mass spectrum obtained was
rich in structural information. Although the molecular weight of this compound was
beyond the mass range of the HP quadrupole mass spectrometer, structural
information was still available from the numerous fragment ions produced that were
indicative of both the steroid (aglycone) and sugar moieties of the molecule. The
structure and mass assignments for the major ions formed from thermal
decomposition fragments of digitonin in the K+IDS analysis are shown in Figure 32.
The molecular weight of this compound is 1228, beyond the mass capabilities of the
instrument. All averaged K+IDS mass spectrum of digitonin is shown in Figure 33.

A number of fragment ions are observed in the mass spectrum which assist in

elucidating the structure of this compound based on mass spectrometric analysis. The

87

”BK? m/z 487

um? mlz 649

 

 

 

 

 

+11 +K" '
mlz 811
\ 01on C330“
0 o
o -o no on
on
HO OH
0“ mlz 943
\ +H+K+ |
-H+K
CHon
CHon HO 0 mlz 363
° -o
on
no on
on

 

+
mlz 201 /'H+K

Figure 32. Structure and assignment of major ions from thermal decomposition
fragments of digitonin -

88

 

disguogg 3.8 weak "mm ”Sum

 

 

 

 

 

 

«w.

010

ea

 

 

 

 

 

«ON

 

 

aochmnav mm

89

most abundant peaks in the mass spectrum arise from cleavage along the sugar
backbone of the molecule with corresponding ﬁ-hydrogen shifts (i.e., a 1,2
elimination) to yield the ions shown in Figure 32. 1,2-Eliminations at the ether
linkage of the sugar backbone results in the formation of two neutral fragments, both
of which are available for K+ attachment. These 1,2 eliminations requiring the

breaking and reforming of two bonds, as shown in Scheme 1 below.

C1120}! 63203 cnzorr cazorr
O O o O
HO J K, OH ——) no on / + H on
H OH H
Scheme 1

Cleavage between the aglycone (steroid) moiety and sugar backbone also involves
a 1,2-elimination and subsequent alkali ion attachment gives rise to the peak
corresponding to mlz 48‘]. This kind of fragmentation mechanism has been observed
for several classes of compounds‘when analyzed by the K+IDS technique. The
production of stable neutral species as a result of 1,2-eliminations is a
thermodynamically favored process. The reaction in which two neutral products are
formed upon cleavage of the intact molecule through this two step mechanism is
frequently found to be exothermic. The activation energy required for this reaction
pathway is relatively low since the thermal energy supplied is sufﬁcient to produce

these stable decomposition products.

90

A similar cardiac glycoside, digoxin, has a molecular weight within the mass
range of the HPS985 quadrupole mass spectrometer. The structure and K+IDS mass
spectrum of digoxin are shown in Figure 34. For this lower molecular weight
compound a molecular adduct ion is observed in addition to the numerous
decomposition products. Cleavages along the sugar backbone occur readily, giving
rise to intense peaks separated by 130u, that correspond to neutral losses of the
polyhydroxy deoxyhexose moieties, of the form, C 6H1003' Again, cleavages occur
as a result of 1,2-eliminations to form neutrals which readily undergo alkali ion
attachment. The respective neutral fragments generated are not found to form
potassium adducts to the same extent. This is best shown for the cleavage between
the 82 and S3 sugar moieties of digoxin, which produce peaks at mlz 559 and 299 in
the K+IDS mass spectrum. The intensity relationship between the two ions is not
equivalent, even though both neutrals are formed. This type of behavior might be
useful for trying to determine certain branch point linkages in cardiac glycosides.
Costello and coworkers have investigated unique fragmentation pathways and the
fragment ion intensities associated with these reactions for branched~chain cardiac
glycosides that are used extensively in Chinese medicine with the technique of FAB
MS-MS. 64 We are currently evaluating whether K+IDS might be useful for
determining exact branched-sites directly for these cardiac glycosides without the
need for MS-MS analyses.

The presence of both molecular weight information and abundant structural
information available from a single K+IDS mass spectrum makes this technique an
attractive alternative to FAB, in which MS-MS is required in order to generate similar
fragment ions. Interestingly, the the types of ions observed in the K+IDS analyses of
this class of compounds parallel closely those ions observed by Marshall and

63

coworkers using laser desorption FF-lCR-MS. The amounts of sample required to

91

 

(A) g «I

 

 

 

 

 

 

‘0. alo. 3o. ' '.l°°V V“ Imi' ' I 1 I I ,rrI I I .0. I I a.
M
(8)
0:60er C4'H640‘4
I"! ..W e 780

molt“) s mlz are
ttn-uzolorl a mi: act

 

Ill/t mlz
80 ...," 559
r’
- 0
0|! 0M 4‘“.

 

 

 

-..o‘. M’! /
ml: 42’

Figure 34. K+IDS mass spectrum of (a) digoxin and (b) assignment of ions
formed from thermal decompostion fragments

92

achieve acceptable signal to backgroud has been demonstrated to be nearly equivalent
for the two techniques. A much more detailed study on the applications of K+IDS to
cardiac glycoside analysis has been completed in collaboration with a fellow graduate
student. The manuscript is in preparation for publication in Organic Mass

Spectrometry .65
The utility of K+IDS for peptide mapping

Martin and Biemann66 offer an excellent review on the present state of peptide
sequencing by mass spectrometry. A number of approaches have been used for
obtaining either sequence information or molecular weight information on high
molecular weight peptides. PDMS using time—of-flight mass (T OF) spectrometry is a
technique capable of yielding molecular weight information on peptides consisting of
more than 200 amino acid residues. FAB, on the other hand, has been used almost
exclusively for providing sequencing information on smaller peptide fragments (c.a.,
520 amino acids in length). The real strength of mass spectrometry in the area of
protein chemistry lies in its ability to rapidly and accurately sequence small peptide
molecules that are not amenable to other sequencing techniques such as 13de

67 FAB-MS of these small peptide fragments gives rise to abundant

degradation.
protonated molecules with very little fragment ion information. The mass spectrum
of Leucine-Enkephalin (T yr-Gly-Gly-Phe-Leu) obtained by FAB-MS using a JEOL
HX—I 10 double focusing mass spectrometer is shown in Figure 35. The only ions
for the peptide molecule above background noise are the protonated parent and

molecular adduct ion species, [M+Na+] and [M+K+], the latter two ions a result of a

mix of Nal/KI being added to the glycerol matrix. The lower molecular weight ions

93

 

 

.383»

3§§3§§AQB.§83§§£ "anon—um

 

+516 a

+5 + as a

 

 

 

 

 

aonvamav anrrv'riltl

94

observed in this mass spectrum are primarily the solvated protonated glycerol
molecules of the form [Gly+l—I]+, [ZGly+l-I]+, [3Gly+rr]" , and [4Gly+H]+.
Sequence information is obtained by collisionally activated dissociation (CAD) of the
protonated molecule. In FAB MseMS analyses the protonated molecule undergoes
decomposition upon reaction with some collision gas (e.g., He) to form a number of
fragments that are detected by linked-scanning techniques. 68 The fragment ions in
the FAB MS-MS mass spectrum are indicative of cleavages of peptide bonds from
both the C- and N-teminus ends of the molecule. Roepstorff developed a
nomenclature scheme for the fragment ions produced from protonated linear—chain
peptides following CAD. 69 The types of fragment ions most commonly observed
are summarized in Table 4. (not shown are immonium ions and W-ions, which can
also arise from CAD of the protonated molecule). 70 A pictorial representation of
these fragment ions will be presented shortly.

A-, B-, and C-ions all originate from the N-terminus of the peptide chain. X-, Y-,
and Z-ions, on the other hand, all originate from the C-terminus. The most commonly
observed peptide fragment ions are of the B- and Y-types.’ The intensity of these ions
relative to the other fragment ions, though, is dependent on the amino acid
composition of the peptide molecule. Y-type ions result from cleavage of the peptide
bond and the fragment ions formed contain the C-terminus of the molecule. These
ions generally result from proton transfer and corresponding hydrogen shift to the
nitrogen functionality from the OL—carbon adjacent to the carbonyl group of the amino
acid residue as shown in scheme 2 for the general case of a tripeptide above. The ions

originating from the carboxy terminus are commonly designated as Y" ions, since the

95

Table 4. Fragments produce from protonated linear peptides (I) upon
undergoing collisionally activated dissociation (CAD)-MS-MS

a a It
[Hﬁ—CH—CO—(M—Cﬂ-CO),-N‘I-CH-COOH + a]

i + it
H-(HN-CH—co),,-NH=CH

'ﬂ
8 “a
I I
H—(HN-CH-co),,-m-CH-Cao+
be
It a,
t I +
H—(HN—Cti-co),,-~H-cn—co-m,

ca

(Reproduced from Mass Spectrom. Rev.

66

§

a, a
I I
+co-m-or-co-(m-cu-co),,ou

x.
it, a .
KEN-CH-CO-(m-CH-CO)MOH

Y.

a, a
I I
+CH-CO-(NH-CH-CO),,.OH

1.

with permission of Martin, 5211.)

96

Yrt
0 H 0
II I ll
NH:— CH—c—NH—(f—c H—CIJH—COZH
l- n' n
B
1
Scheme 2

fragment resulting from cleavage between the peptide bond is found to contain two
additional hydrogen atoms. B-type ions, on the other hand result from cleavage
between peptide bond and the fragment ion formed contains the N-terminus of the
molecule. Sequential cleavages along the backbone of the molecule originating from
both the C- and N-terrnini allow for one to accurately deduce the correct amino acid
sequence for the molecule. .

The FAB MS-MS mass spectrum of Leucine-Enkephalin is shown in Figure 36.
A number of fragment ions are observed which are used for determining the amino
acid sequence of this molecule. The abundance of these ions is quite low, though,
relative to the protonated molecule. The most intense fragment peak (mlz 278,
corresponding to the B3 ion) is only two per cent the intensity of the protonated
molecule. A number of A, B, and Y-type ions are also observed in the FAB MS-MS
mass spectrum of Leu-Enkephalin which assist in the determination of the true amino
acid sequence. The major drawback to doing an MS-MS analysis on the protonated
parent is that a signiﬁcant portion of total ion current is dispersed amongst a number
of decomposition products upon collisionally activated dissociation of the parent ion
and the ion signals associated with these fragment ions are weak. However, matrix

interferences due to glycerol are eliminated when MS-MS analyses are performed,

97

333823
eggagvsgoagoaagsmz.§m§ ”was

6%.. 8m 8w 8.. 8m 8m 8m 8m .8“

 

 

 

 

 

Lma

“ﬂﬁﬂﬂ'ﬂ30 “3383.600

98

and as a result the SIN for these fragment ions is more than sufﬁcient for
identiﬁcations to be made.

The applications of K+le for peptide sequencing has been implied from earlier
studies using this technique on hexaglycine and the tri-peptide, Alanyl-Leucyl-

37 The results of such studies were encouraging from the standpoint of the

Glycine.
ability of the technique to provide both molecular weight and structural information in
one analysis. These studies have now been extended to more complex peptides,
including Leucine—Enkephalin and others.

Small molecular weight peptides have been studied by K+IDS to detemine this
techniques' ability to directly produce both molecular weight and structural
information simultaneously. Peptide molecules fall into the category of Type II
molecules, as discussed in chapter 3. The lability of peptide bonds is such that little
energy is required for bond-breaking to occur. By controlling the heating rates,
however, it is possible to increase or decrease the relative amount of decomposition
products generated in a K+IDS analysis. K+IDS analyses result in the desorption of
both the intact molecule and structurally signiﬁcant decomposition products.
Leucine-Enkephalin ('I‘yr-Gly-Gly-Phe-Leu, MW = 555) is discussed for the purpose
of demonstrating the capabilities of the K+ IDS technique in the analysis of low
molecular weight peptides. The structure of Leu-Enkephalin and the masses of the
structurally signiﬁcant ions that may be formed following alkali ion attachment are
summarized in Figure 37.

The Roepstorff nomenclature scheme has been maintained for identifying the
cleavages along the backbone of the peptide molecule. As an example, cleavage of

the first peptide bond originating from the C-terminus of the molecule results in the

formation of two radical species, designated as B2 and Y3. These radical species have

 

Possible Fragment Ions tn the Nt’ms Mass 8
Leuclne Bukephallu ('l‘yr-Gly-GIy-Phe-Leu,

[if

 

 

99

 

 

 

‘
355$)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

NH— 08— tm - - ca — - £_

' I ‘ " " l “I"

03 ca (ca
1 ' '

08

Q8 c / C c / ©—: 8

g / 8 8
3 I
.' j 8 8 I .Q
/ .
ﬂ ‘2 ‘t ‘ ‘3
“‘15 r t .
P335331: Nominal 'm ‘h "“3 ° ":1 3:223: Nominal In uh [Ii-a. o In.)
‘t '3' l” X. m m
I. I“ 187 y. .3. ‘3,
c. m m 1‘ m m
A: 188 8|. ‘3 888 ’2.
I, 221 2“ '3 1n 3..
‘2 an ass 2 m m
‘. 85. 8” ﬂ. 3“ 3”
I . 210 set v, :34 m
m
c I m zI m m
A, m m x, m «a
'4 425 «I v. m m
C 4 «I «a z. m m
A; m m z. m sea
a; m m

 

 

 

 

 

 

 

Figure 37. Structure and possible thermal decomposition fragments for leucine-

enkephalin arising from a Na+lDS analysis

 

100

nominal molecular weights of 221 and 334u, respectively. If no hydrogen shift was
observed for these species prior to their desorption into the gas phase, these [M+Na +1
adducts would be observed at mlz 244 and 357, respectively. However, the
formation of these two radical species is unfavored, from a thermodynamic
standpoint. A method for making this reaction exothermic has been proposed. As
was described earlier for the cardiac glycosides, cleavages at these labile bonds
frequently involve 1,2-eliminations to produce very stable neutral molecules, and this
mechanism would be expected for the peptides as well. The general case for the 1,2-
elimination reactions is shown in Scheme 3. The two neutral species formed are very
similar in nature to the B- and Y-type fragment ions observed for the tri-peptide
described earlier. Hence, the two neutral Species above represent the B-type and Y-
type fragments. Upon desorption of these neutrals and susbsequent gas-phase alkali

ion attachment, B- and Y-type alkali ions are formed.

R" o

12'" NH ('3 P: NH CH co H
I \J\ I. 2
H R

I
R'"—r~tH—c=c=o + lilH—CH—Cozﬂ
H 12'

Scheme 3

101

The Na +IDS mass spectrum obtained following deposition of 3 pg of Leucine-
Enkephalin onto the bare rhenium wire of the dual-ﬁlament K+IDS probe is shown in
Figure 38. The [M+Na+] ion is observed at mlz=578 and is of relatively low
abundance. The alkali adduct of the Y3 ion is found at mlz 358, lu above that
reported in Figure 37 (designated as [Y3'+ Na+]). Alternatively, the alkali adduct of
the B2 ion is observed at mlz 243, l u less than the the value reported in Figure 37
(designated as [Bz-H+Na+]). This is consistent with the 8-H shift (i.e., a 1,2-
elimination reaction) from the B2 containing fragment, as shown in Scheme 3. A
number of other decomposition products are formed from both the C- and N-termini
of the molecule. As can be seen clearly, the amount of information that is available
on this pentapeptide by K+IDS is more than sufﬁcient for determining the aminogacid
sequence. All of the Y-type ions are observed in the mass spectrum, the most intense
being the [Y3'+Na+]. The majority of ions observed in the K+IDS mass spectrum of
Leu-Enkephalin are the cationized species. However, low mass surface ionization
products are also observed. The peak observed at mlz 91 is believed to be the C7H7+
ion from the R-chain of phenylalanine. Other prominent ions in the K+IDS mass
spectrum of this molecule are a result of neutral losses of 1120 and NH3 which are
also observed as decomposition products.

Methionine-Enkephalin (Tyr-Gly-Gly-Phe-Met) provides as another example of
the strengths of K+le in the analysis of these compounds. The structure and masses
of possible fragment ions that are commonly observed are summarized in Figure 39
(and are analagous to those described for Leu-Enkephalin). The Na+ IDS mass
spectrum of Met-Enkephalin is shown in Figure 40. A peak for the molecular adduct
ion, [M+Na]+, observed at m/z = 596 is relatively abundant. In fact, the intensity of
this peak is dependent on the heating rate used for the analysis. The base peak in the

102

i=2n8—amoanx. .0 Co 8.58% 2!: warez "an 8.35

N\E

000 can can 00* Get 00* con can can o¢N oou cop our on 0*

 

.r..._......_: ._
_. _ a

_ m
n 31.3 a +23 ”5
8

 

 

 

 

m

a

 

 

Heezemeu

 

"+02... .«5 "+33 Ya

 

reigns-u

 

 

{no

_

 

r..-

 

emaocqoxculoemoaoq mo Eatooqa unoE marmoz

Kitsualul onyepu

103

Possible Fragment ions in the Na’lDS Mata Spectrum of

Methionine-Eukephaiia ('l‘yr-Giy-Gly-Plle-Met, MW - 573)

 

 

 

 

 

 

 

 

 

 

 

O 0 O 0
I I I 5 I
surfs—c—am—cni-c—Nn—cnf c—rm—t'nl— -tta—t':n—c-
8
(Im
on .
D l to e esTgaata 9
17:: :3" Nominal maaa m/a [frag 9 Na 1 Prune-t Nominal maaa Ila [ll-a. 9 "O I
A, m m x, m I”
I, m m v, m m
c, m 202 z. m 156
A; m m x1 :2: m
a, 221 244 v, 295 3"
ca 230 zsa z a m 303
A, 250 m x, 3" 4”
a . m m' v, 353 ’7‘
c m m z m m
A.. m m lt4 m «a
a ‘ 425 g“ y‘ «a «2
C ‘ 440 a” Z ‘ 394 417
A; sza ssl Vt 572 5”

 

 

 

 

 

 

 

Figure 39. Structure and possible thermal decomposition fragments for

methionine-enkephalin arising from a Na+IDS analysis.

 

104

Own own

 

 

 

33.558382355933932 earner

 

 

 

 

\

_ a
N+oz+ Ne; ~+oz+. e;
. m

~+02+Il Nm~

 

 

.I—F

 

 

105

mass spectrum of Methionine-Enkephalin is also the [ira'maﬁ species. Again,
major decomposition products correspond to cleavage of the peptide bond, resulting
in the production of predominantly B-type and Y-type ions. A number of low
intensity peaks have been observed in the K+IDS mass spectra and are not fully
characterized at this time.

Analyses by K+IDS on longer chain peptide fragments, such as bradykinin (Arg-
Pro-Pro-Gly-Phe-Ser-Pro—Phe-Arg) and another commonly studied peptide, Val-His-
Leu-Thr-Pro-Val-Glu-Lys (VHLTPVEK), have not been as informative. The
structure and possible sodium adduct ions that may be found in a typical Na+IDS
mass spectrum of VHLTPVEK are summarized in Figure 41. For some reason,
K+IDS analyses have been unsuccessful in generating discernible molecular adduct
ions for these compounds. The Na+IDS mass spectrum of the peptide VHLTPVEK
is shown in Figure 42. A few ions that were expected to be observed are indeed
present in the Na+le mass spectrum of this compound. The base peak in the mass
spectrum corresponds to the [32' - H+Na+] ion observed at mlz 259. In addition, a
number of C-type ions are also identiﬁed in the lower mass region of the mass
spectrum. The [C3'+Na+] ion is observed at mlz 389. Two intense peaks in the mass
specu'um, labelled at mlz 301 and 359 have not been assigned to a typical Roepstorff
designation.

It is clear that the types of ions observed for this peptide are not at all what one
would expect from a K+IDS analysis based on the results for both Leu-Enkepahlin
and Met-Enkephalin. No molecuar adduct ion is observed. The unique feature of this
peptide and bradykinin is the presence of proline residues. Perhaps the rigidity of
this ring system is such that peptide bond cleavage does not occur readily. Initial

results indicate that sequence information can be obtained up to the proline residue,

106

4115 «9+2 . 88 23a Aaq.a~o.1>.o&aﬁ.8..aﬁ.a>v

 

 

 

 

 

 

 

3&8. 3838383688888: 186538... 83583.... “Spam
a. to: e.~ 2.. .2 e .>
8

2a.. to: can e: to: e~> :. to: tau a: to: can
«.8 to: eo~ 5 to... to» a: to: too 2. to: to.
.3 to: tu~ to to: to» a: to: emu 5 to: too
.2 to: an 2: en: to» 2.. e2 eeo 5. e1 ten
.3 to: can .2 e42 eo> . .3 to... too 2» to... tan
a: to: ta~ a: ta: ta» oz to: one 8a to: eon
3. to: e .~ .2 e! e .> .2 to: t .o z. to... e .o
56 n3. 58 a «a g 53 an

«:2 snow 1!:

_ 2
...je an" .3 563 xx
..6 new .99 mm 39 no «no new .95 .8

_ _ _ .
HOIWISOIIZIn-vlrlblﬂz IWISO IEIW ZIWI-IOISZIn—VIE IEIWIIOIEIn-vluﬂo

O O O O O O O O

 

.ammamvaaeoawacwaqmameaeameasaaxwanae«apnea "seesaw"

 

 

 

N\E .

ooh 0N5 . 0mm ovm com com cum one owe 00+

. . ._ .t . . ..:. ._=_. 2: L . .. E. Lie}:

. .“_
n+ez+mmu
"+33?“ “+23%: m or x
“eastenu

7 o o 11
m 2:. . con own omu oeu oou of of on 3.
1
7702.003“ 1
m I
on: nee/Lela»: 3» 7.2.13 r.

 

 

 

— +o 23.7mm“

 

 

 

 

 

108

but little information is available beyond this peptide linkage. FAB MS-MS on
VHLTPVEK also has been found to be less informative than it is for other small-
chain peptides. Little fragmentation has been observed in the CAD-MS mass spectra
of this compound. The K+ IDS technique does show promise, though, in the analysis
of some peptide molecules. It is difﬁcult, though, to assess the capabilities of the
technique for amino acid sequencing since we have been limited to using a
quadrupole mass spectrometer for all studies. The upper mass limit of the quadrupole
instrument available in our laboratory is 1000 u.

Under investigation is the possibility of implementing the K+IDS technique on
the high resolution JEOL HXI 10 mass spectrometer. Currently, studies involving
K+IDS on the JEOL PIX-110 are limited in that a direct chemical ionization (DCI)
probe must be used. The speciﬁcations for the upper current limit of the probe is 1.5
A, which is just sufficient for K+emission to occur. In addition, the current probe
design employs only one ﬁlament. For K+IDS to be implemented effectively on the
high resolution, high mass instrument, a dual-probe design similar to the one currently
employed for use on the HP 5985 quadrupole is needed. Initial results of a K+IDS
analysis of poly(propylene)glycol (PPG) (average molecular weight, 425) in which
the maximum current of I=1.5A was used to desorb the polymer from a K+IDS
emitter in the presence of a high ﬂux of cesium ions from the single ﬁlament DCI
probe are encouraging. Cs+ was found to be generated in copius amounts under very
unusual operating conditions. Desorption of the polymer in the presence of this high
flux of Cs+ resulted in the formation of a number of [M+Cs+] adducts ion of
relatively high abundance. Shown in Figure 43 is the 05+ IDS TIC proﬁle and mass
spectrum of PPG 425. All ions observed in this mass spectrum are Cs+ adducts of the

- .35.:
as xx 3E 2.. 8 83:3 3. 93 cc 8889 .3. ”9+6 "995E

N\t

ALLLA‘

. L _j I J l

 

 

 

 

 

 

 

 

 

 

 

sum :2. new new now 8m 8w 8d
a u s a w _
_ _ m m . __
on":
.9:
m _
an: t:
.wu:
mm
H: k
n... A +6 + E ...Aomnmov. 2: c

 

 

ff'

V v v

mud-c...” quart-“Icon

T v V

 

IIO

oligomeric species of this PEG mixture, and have the formula [H(OCH2CH 2012-)-
nOH 4» C54], where n=l-10. The results parallel quite closely those obtain using
K+IDS for the analysis of PPG 425 on the quadrupole mass spectrometer. An
analysis of Leu-Enkephalin was made in an identical manner, by rapidly heating the
peptide from the direct chemical ionization (DCI) wire in the presence of a high ﬂux
of Cs+. [M+Cs+] adduct of B- and Y-type ions were observed, and the amino acid

sequence could indeed be deduced from interpretation of the complex mass spectrum.
Future studies utilizing the strengths of K+IDS

These results are encouraging since it induce the possibility of using the K+IDS
technique for the analysis of much higher molecular weight species. Compounds
such as digitonin and other high molecular weight thermally labile compounds may
be shown to give rise to abundant molecular-type ions in the K+IDS analysis.

In addition, a novel approach to using alkali aluminosilicate materials for mass
spectrometry is under investigation. This technique has been named "K+IDS by
FAB". K+ (or Na+) ions can be generated in the gas phase from bombardment of
high energy (c.a. 10 kV) atoms on to a surface containing an appropriate alkali
~ aluminosilicate material. If a sample is placed directly onto a surface coated with the
alkali glass material, and is subjected to high energy fast Xe atoms, molecular adduct
ions of the form [M+A+], and [M+2A-H]+, where A=alkali ion, can be formed. This
may be an attractive method for obtaining molecular weight information on
compounds that are not found to give rise to protonated. molecules in conventional
FAB analyses. A comparison of the conventional FAB mass spectrum of arachidonic

acid and that obtained by the "K +IDS by FAB" approach, shown in Figure 44 a-b.

ooaeucn ecu-op...

tun-oust) C("De—Q'

 

 

 

 

 

 

 

 

1
‘ D
‘ D
aa‘ :
‘ I
< +
so [M-Hl 1
1
. " mmﬁ I
o‘ :
‘ D
. g .
i . n .
1 ‘t I i j ' ' “J “ ‘i p :
é~4 I, I , ; , ;- Wt,:_i"-__;J'£Lu..‘f‘; In"! 1“, “9,. Ir m . . i
so we we as . an see as 4a
m
1
I
3, [M+2Na-H]+ .
aa- .
q p
63- b
i + i
a [Gly+NI+] IM+Nn ] ,
. C
1 b
an- .
5' 1“ 1 283 I Has. " iaae ’
M:

Figure 44. (a) FAB mass spectrum and (b) "Na+IDS by FAB” mass spectrum of

arachidonic acid .

 

 

112

This type of approach is also being investigated to help clarify and support the
proposed gas-phase mechanism for ion formation in FAB-MS. 71 A normal FAB
probe-tip has been modiﬁed to accomodate an alkali aluminosilicate mixture on one
side of the probe-tip while the other is coated with appropriate analyte solution, in
this case, glycerol. The two surfaces are physically separated by a ﬁnite distance,in
order to ensure that mixing does not occur between the two surfaces, even if subjected
to a high energy fast atom beam. The probe tip and mass spectrum obtained for
glycerol using this split-probe design is illustrated in Figure 45. Molecular weight
information in the form of the protonated glycerol molecules and K+ adduct ions is
observed.

This type of approach would be attractive for those compounds for which no
protonated molecule is observed in the conventional FAB mass spectrum.
Musselman and coworkers72 showed that cholic acid, when analyzed by
conventional FAB, produced no MH+ ion, but the highest mass ion in the mass
spectrum was the [M+Gly]+ adduct ion. If this were an unknown, the ion might be
presumed to be the protonated molecule and thus inaccurate molecular weight
determinations would be made. Addition of AgNO3 to the glycerol matrix resulted in
attachment of a Ag+ ion to the analyte which was useful for determining the exact
molecular weight of the species due to the distinct isotope pattern associated with the
silver ion. The question remained as to whether adduct ion formation was occurring
in solution or in the gas-phase for attachment. An elegant way for determining
whether or not species such as cholic acid actually desorb into the gas phase prior to
adduct ion formation is to incorporate the "K+IDS by FAB" design. By placing
cholic acid on one portion of the K+IDS probe tip and the alkali aluminosilicate

material on the other, formation of [cholic acid + alkali ion] adducts would suggest

113

Aug 59 a uni: @3390 3803.33 82.39 8.8 can

 

 

 

 

 

E

8h 8' Con CON
U cauév 92.00.;
g OAUCSV
U .263
.. .368
A

:28 once->18?!
U
..A mums
U 8.1.3.85 I30... 2.3.03

 

 

du 38.—
.3 «8+3. .9. 26E

8“

 

.398

 

OASOGV

' E 7

mm SMIV'BU

V l1 7

V '

 

 

114

a gas phase mechanism is operative in FAB analyses. This experiment is currently

being pursued.
Final comments on K+IDS

This DI technique continues to show promise in its applicability to a wide range
of organic compounds. Sensitivity problems initially associated with the technique
have been curtailed, yet work still remains in optimizing the performance of the
current probe design. A better understanding of the temperature relationship between
the two ﬁlaments currently employed, is needed and is currently under
investigations8 Results of such studies may induce a design alteration that will
lower even further the detection limits associated with the technique. A method for
generating lower detection limits might be achieved on a mass spectrometer equipped
with dual insertion probes. In this experiment, the potassium emitter could be
inserted into the ion source and heated to a point at which the highest ﬂux of K+ ions
could be generated. A second direct insertion probe containing the analyte could be
inserted and the thermally labile compound ﬂash vaporized using rapid heating
techniques. Assuming a localized high pressure region could be achieved very near
the two DIP probes, best overlap could be achieved for analyte desorption and alkali

emission, leading to the most optimum sensitivity of the technique.

115

Chapter 5: Chemical Ionization Theory

Chemical ionization mass spectrometry (CIMS)

Chemical ionization (CI) mass spectrometry is based on ion/molecule reactions
for gas phase chemical analysis. As early as 1916, reactions involving proton transfer
between H; and H2+ were observed by Dempster. 73 The analytical utility of CI-MS
was not really exploited until the early 1960's, when Field and Munson discovered
that gas phase ion molecule reactions involving methane at high pressures (c.a. l
torr) could be used to produce unique chemistry. 74 Shown in equations 3-4a, 4b
are the major reactant ions that occur following electron impact ionization of CH 4 in a

high pressure source:

6.
+ 4-
CH4 -------------> at, .CH3 .etc.(+2c-) (3)
CH + + CH ............. ..> + + . (4a)
4 4 CH5 CH3
+ 4-
CH3 + CH 4 ------------- —> c2115 (4b)

Electron impact on neutral methane molecules results in .the production of a number
of ions, mainly CH3+ and CH :1 These ions then undergo reactions with neutral
reagent gas to produce CH; and C2“ 5+ in varying degrees and the relative
abundance of these two ions becomes approximately equivalent as source pressure is

increased above 0.2 torr. These reagent ions are then utilized as chemical ionization

116

reagents for gas-phase proton transfer reactions. In addition, a small fraction of CH 4
is converted to the C3“,+ ion via a reaction between CZH; and H2, and this
species is solely observed to participate in adduct ion formation (i.e., [M+ C3H5+l
formation) reactions with analyte, M.

The strengths of methane-CI for numerous analytical applications lie in the fact
that the two product ions, CH 5+ and C2H 5+, are relatively strong acids, from a
Bronsted-Lowry deﬁnition. Since most organic compounds of biological interest,
including organic acids, steroids, and amino acids act as weak Bronsted acids, 75
proton transfer from the more acidic CH 5+ molecule is expected to occur fairly
readily. Shown in equations 5-6 are the general cases for reactions involving the

methane reagent ions with organic compounds. Equations Sa-Sc show some reactions

that occur for analyte molecule, M, with CH 5+ :

 

(:Hs+ + M > MH‘“ + neutral, AHm (5a)
+* 4»
MH --------------> MH (5b)
MB” ............... > Fragment ions (5C)
+ +
C2H5 + M --------------- > [M- C2H5] (63)
(3211; + M ---------------> MB“I (6b)
NIH” --------------- > MH+ (6c)
MH” ............... > Fragment ions (6d)

Ion formation is governed by direct proton transfer from the Bronsted acid species
to the more basic species. The energy associated with such a proton-transfer reaction,
shown in eq. 5a, is the ARM and is directly related to the difference in proton
afﬁnities of the conjugate bases of the reacting species: In general, these proton

transfer reactions are exothermic (i.e., AH rxn. S 0), though under certain conditions,

H7

endothermic proton transfer reactions have been observed. Typical proton afﬁnity
(PA) values for methane have been reported as 130.5 kcal mole'l. 7‘ For the reacting
analyte, proton afﬁnities can vary signiﬁcantly. For example, amino-containing
compounds, such as triethylamine are reported to have PA = 230 kcal mole". 75 in
contrast, ethers, such as diethylether, exhibit proton afﬁnities that are substantially
lower (c.a. 199.4 kcal mole'1 ). 77 For most organic compounds, nevertheless, their
proton aﬁinities exceed that of CH 4 by several kcal mole.1 of energy. Therefore, the
product ion formed in the initial reaction, eq. 5a, will have a maximum amount of
excess internal energy equal to the difference in proton afﬁnity between the reagent
gas and the neutral analyte (M). Generally, the energy difference in the proton
afﬁnities of the two reacting species is distributed between the two reacting species,

so that the internal energy (E. ) associated with the newly formed protonated

rnt
molecule may be expressed as Elm (MI~I+) S PA (M) - PA (CH 4). The product ion
may be stabilized by a collision with neutral reagent gas, if the difference in proton
afﬁnity between M and CH4 is sufﬁciently small, giving rise to the protonated
molecule observed in eq. 5b. On the other hand, if this energy difference is large,
then the excited product ion may be more likely to undergo decomposition to a set of
fragment ions, typically through some low-energy pathway (e.g., elimination of
stable neutral molecules). Therefore, the exotherrnicity of reaction (which is
determined by relative differences in proton afﬁnities) governs to a large extent the
amount of fragmentation resulting from the chemical ionization experiment.
Examples of CI mass spectra of various compound classes (e.g., amines, alcohols,
hydrocarbons) may be found in a detailed review by Harrison. 73

Equations 6a-6b suggest that other methane reagent ions may participate in proton

transfer or adduct ion-formation reactions. In an analogous manner to that described

for the CH; ion interaction with neutral molecule, M, C2115+ can react by proton

”8

transfer to the more basic species, M. Again, depending on the exothermicity
associated with this hydrogen transfer reaction, either stabilization to the protonated
molecule, or decomposition to lower energy fragment ions can occur. A frequently
observed reaction for C2H5+, and even C 3H5+, with many organic molecules is the
formation of the stable adduct ion. Adduct ion formation typically occurs for those
analyte molecules that exhibit proton affinities very near that of the reagent ion. The
PA (C2114) has been shown to be approximately 165-170 kcal/mole. The higher PA
(C2H4) more closely parallels the PA values for most organic molecules, and as a
result, adduct ion formation involving C2H5+ is favored over proton transfer. Shown
in Figure 46 is the methane-CI mass spectrum of the trimethylsilyl (TMS) derivative
of 2-hydroxyisobutyric acid. In addition to the weak ion signal for the protonated
molecule observed at mlz 249, two peaks of relatively high abundance are observed
at mlz 277 and mlz 289, corresponding to [M+-C2H5]+ and [M+C3H5]+, respectively.
The exothermicity associated with the proton transfer and adduct ion formation
reactions, as suggested, determines to a large extent the amount of fragmentation that
will be observed in a chemical ionization mass spectrum.

The amount of fragmentation, however, is also governed by the kinetics of these
reactions. A brief kinetic description of the important processes in electron ionization
mass spectrometry is required ﬁrst, and is summarized here. In EI-MS, molecular
ions, M+', are formed as a result of electron impact with a 70 eV particle beam on
gaseous analyte molecules. A distribution of internal energies ranging from the
lowest energy required to form the molecular ion (i.e., the ionization potential of the
molecule) to an internal energy several eV above that required for ion formation will
be observed, since several eV of energy can be transferred to the analyte molecule
following the ionization event. 79 This distribution of internal energy in the

molecular ion is generally represented by a unique probability function, P(E), for the

ll9

. .653 32 gig...“ co 58% 38 6.8.32 .3 39a

u\E

con 00N 00N oeN ONN CON omp.omp oer ON" 00; on on at

Ll :a_::_4.}L::_:: _.:._........_}..:_.:.;
m .

+2: ,
22.3 ..

 

 

 

 

 

 

YT

 

 

 

+~uzuo+3
«vaot+zav

 

 

 

 

120

molecule, called the Wahraﬁig diagram, 80 as shown in Figure 47.

hr? /

a + n?——) 3+
a AP“) , .

M'—-> A
5 u

 

 

 

 

AP(B"’)

 

---pb----

l

/ Internal Energy —-)
IP (K)

Figure 47. Relationship of P(E) and k(E) for unimolecular decomposition.

Decomposition of M+' to a fragment ion, A+ requires an amount of energy equal to
the sum of the IP (M+) and the bond dissociation energy, D (M-A), required to form
A from M. The fragment ion, A+, is generally observed at an energy slightly above
this lower limit, and this is called the appearance potential. The appearance potential
(AP) of A+ from M+' is the amount of absorbed energy at which the rate constant for
reaction is 10‘5 sec'1 (i.e., lifetimes are 10'6 sec or less). This value has been

determined to be the time required for an ion to leave an ion source in which it was

121

formed, in order that it may be detected. In E1, the extent to which fragmentation
occurs is dependent on the amount of internal energy associated with the molecular
ion and the rates of unimolecular decomposition to a set of fragment ions. The rate
constants for decomposition to a set of fragment ions are affected by the precursor
ion’s internal energy. Molecules which readily stabilize both a radical and positive
charge, such as unsaturated ring—containing compounds, will be more likely able to
'stabilize this excess internal energy, leading to less fragmentation. Fragmentation is
also determined by the rate constants of reaction for the decomposition of the excited

state of the molecular ion as shown in Scheme 4 below :

These rates, Itl and k2 are unique to the fragment ions, A+ and B+, respectively,
and will depend on the amount of internal energy associated with the molecular ion.
The lifetime of the excited molecular ion, which determines its stability, as stated
earlier, is the limiting factor in determining the amount .of fragmentation observed.
For higher internal energies, as shown in Figure 48, the fragment ion B+ may be more
favored than decomposition to A+, and the point at which the rate curves intersect is
the appearance potential, AP (B +, M).

An analogous situation may be Operative for the chemical ionization experiment.
Equation 5a showed that the formation of the protonated molecule upon reaction of

neutral analyte with an acidic reagent ion, such as 01;, will have a distribution of

122

internal energies. As discussed earlier the maximum energy, E . is the difference

max
in [PA(M) - PA(CH4)]. This behavior is described pictorially in the upper curve of

Figure 48.

 

 

 

 

 

 

8 p
ﬁt 2 Emu

Inf—9 3*
5.? AHA’) +
v P +
.u 2 \ .MH -—) A

Pl \
AP ( 3*)

 

 

0 Internal Energy (e?) ——>

Figure 48. Distribution of internal energy in protonated analyte molecule
following exothermic proton transfer.

Decomposition to a set of fragment ions has been shown to be dependent on the
amount of excess energy associated with the MB” complex. The more facile the
fragmentation process, the lower the energy requirement, and as a result, the faster the
rate of dissociation to fragment ions. The rate of dissociation will be affected,
though, by how rapidly the protonated molecule can dissipate its excess energy,

specifically through collisional stabilization. Shown in a Wahraftig-type diagram,

123

Figure 48, are the effects of low ron source pressure k collision

(P2) on the stabilzation of the protonated molecule. For

(P1) and high ion
source pressure k collision
Nil-1+. to undergo decomposition to an observed A+ ion, an amount of energy equal
to the AP (A+, M) is required. If collisional stabilzation can occur in a sufﬁciently
short time, insufﬁcient internal energy will be available to A+ for fragmentation to
occur. Thus, upon increasing the ion source pressure from P1 to P2, a smaller fraction
of MB” will be converted to the fragment ion species, A+, and a greater fraction of
MP1” will be stabilized to the MH+ ion. The dependence of rate of reaction on ion
source pressure is thus an important variable in determining the amounts of
fragmentation observed in a chemical ionization experiment. Alcohols, for example,
are found to readily undergo decomposition to the [M-HZO]+ ion from the excited,
protonated molecule. These rates of reaction are so fast that even at high ion source
pressures, in which the collision frequency is high, stabilization of the Nil-1+. does not
occur before elimination of H20 occurs. For many CI experiments, though,
decomposition reactions do not proceed at such a facile rate, and therefore will be
dependent on the how rapidly the NIH” complex can be collisionally stabilized. If
the protonated molecule is capable of undergoing several stabilizing collisions in a
very short time (less than that required for fragmentation) with other neutral reagent
molecules, the amount of excess energy associated with the excited protonated
molecule will be dissipated, and as a result, an insufﬁcient amount of energy will be
available for bond-breaking to occur.

Methane has perhaps been chosen most frequently for most CI applications due to
the relatively low PA of the CH4 molecule and the general reactivity that is observed,

81 82

including reactions with alkanes, aromatics, and alcohols.” However, a

plethora of reagent gases are found in literature which are described for particular CI

experiments. For positive ions, these reagents include hydrogen,84

124

tetrarrrethylsilane,85 diethylamine,“S and isobutane.” Research investigations in
this laboratory have used ammonia ClrMS for enhanced qualitative and quantitative
information in the analysis of organic acids. The proton afﬁnity of NH3 is quite high
relative to methane; values are reported between 202-205 kcal mole’l, 88’”
implying that this chemical ionization reagent is a very weak Bronsted acid, and will
protonate only the most basic compounds, such as amines. The reagent ions formed
upon electron impact of NH3 are summarized in eq. 7. At pressures above c.a. 300
mtorr of NH3, solvation of the ammonium ion about neutral NI-I3 molecules readily
occurs, even though rates for clustering reaction are quite slow relative to the

90

bimolecular proton transfer reactions, due to the fact that a third body collision is

N113 + e’ .............. > NH; . NH4+ (7a)
NH4+ + N113 -------------- > H(Nl‘i3)n+ , n = 2-4. (713)

required for stabilization of the solvated molecule. The extent to which ammonia will
solvate about itself to produce the ammonium cluster ions is pressure dependent, and
as will be shown later, the proper choice of operating pressure is important for the
quantitative studies undertaken. A typical chemical ionization mass spectrum of
ammonia is shown in Figure 49 below. At relatively low ion source pressures, as.
measured by a thermocouple probe positioned near the CI source and pressure
measured with a Hastings gauge, the NH4+ ion predominates. As ion source pressure
is increased to greater than 0.2 torr, solvated ions of the form [2NH3+H]+ are the
most abundant reagent ions. Keough and DeStefano91 reviewed recently the

reactivity of analyte molecules with NH3 reagent ions. A number of equations

125

 

126

representing the possible reactions betweenanalyte molecule, M, and NH4+ are
summarized below. If the PA of M is greater than the PA (NH3), then ions
characteristic of the molecular weight of the analyte may be formed, as shown in eqs.

8a and 8b. If however, PA (M) is less than PA M3): then electrophilic attachment

mun; .......... > [M+Hl“ +NH3 (8a)
[M+H]+ + NH3 ---------- > [MNH4]+ (3b)
M + NH + > M' 'H +1.
4 __________ [ . . ”NH3] (9a)
NH3
[M°--H~-NH3]+* .......... > [MoNH4]+ - ------- > Fragments (9b)
[M' "H- . -Nl-{3]+* .......... > M+ NH4+ (10)

of NH 4+ to the analyte may occur, as shown in eq. 9a and 9b. As for cases observed
with methane-CI, the process most likely to be observed will depend on the relative
difference in PA between analyte, M, and NH3.

Analogous to the situation described for methane, the stability of the initially
formed [M-NH4]+. complex is dependent on the excess internal energy associated
with the complex and it is this which is important in determining whether or not the
complex will be sufﬁciently stabilized by mild collision with neutral NH3 molecules,
or whether or not the excess internal energy is sufﬁcient to cause bond dissociation to

either the protonated molecule , fragment ions, or the initial reactant ions. When PA

I27

(M) is signiﬁcantly lower than that of NH3 it is likely that the complex will dissociate
back to the initial reactants, resulting in no net reaction (eq. 10). A lower limit value
of PA = 188 kcal/mole has been reported for autodissociation to occur.92

The ammonium adduct ion has been ascribed to electrophilic attachment of NH 4+
to the neutral analyte with stabilization occurring by collision with other neutral
reagent gas molecules. This product typically arises when the PA (M) is intermediate
to that required for eq. 8, and that for eq. 10 to occur. Often, when there is slight
exothermicity associated with this adduct ion formation, it is possible for the newly
formed adduct ion to dissipate some of this excess energy through elimination of
neutral species. Several fragmentation pathways describing this behavior are found in

93 and mechanisms for the formation of these ions are discussed in detail

literature,
elsewhere. 94

Little attention, however, has been given to reactions of analyte, M, with the
reagent ions of the form HOST-13);, where n= 2-4. For organic acids analyzed by
NH3-CI, increasing the source pressure leads to an increase in the abundance of the
NM,

hydroxyisobutyric acid (di-TMS) in the Figure 50. At relatively low ion source

]+ ion relative to [Ml-I]+ ion. This behavior is shown for 2-
pressures (0.2-0.3 torr), the intensity of the protonated molecule is nearly equivalent
to the adduct ion, [M+NH4] +. The abundance of the protonated molecule decreases
rapidly with higher reagent gas pressure. This behavior is typical for almost all the
organic acids that have been analyzed to date. Literature values of the proton
afﬁnities of chemically similar compounds (e.g., CH3COCH2C113) are very near that
of ammonia; PA (cascocrlzcnp = 201.3 kcal more". The small difference in
these two values suggests that proton transfer and adduct ion formation reactions may
be somewhat competitive. Two possibile explanations for the observed increase in

adduct ion formation may be considered: (a) increased ion source pressure leads

96110

128

'9' [M+I-lllon
‘ + [M+NH4llon
50- -- [M-CHallon

 
     
  

 

 

0 TYT'IIUIUU rIUYrI'I'U

0.5 1.0 1.5 2.0 2.5 3.0 3.5

Pressure (x 10 ee +4) -

Figure 50. Plot of ion source pressure us. relative intensities of pseudo-molecular
ions for 2-hydroxyisobutyric acid (di-TMS)

129

to a greater number of neutral reagent molecules which are available to neutralize
more effectively the [M--H«NH3]+. intermediate and/or (b) the ions of the form
H(NH3)n+ may actually participate to some extent in the direct formation of the
molecular adduct ion species. Little attention has been placed on these solvated ions

and their immediate effect on analyte ion formation. Kebarle”

suggests that the
solvated ammonium ions do not participate, to an appreciable extent, in proton
transfer reactions since the thermochemical stability of solvated ions increases with
increased clustering, and as a result, proton abstraction from the solvated ion becomes
more difﬁcult. The question as to the extent to which adduct ion formation is affected
by increased abundance of solvated ammonium ions has not been fully addressed.
One method for determining the role these solvated ions play in the formation of
the [M-NH‘J+ adduct is to actually monitor a decrease in abundance of HmHa) n+
ions as a function of eluting analyte from a gas chromatograph. This technique has
come to be known as reactant ion monitoring (RIM). This technique has found wide-
spread use for determining relative response factors in chemical ionization

96’” This point will be taken up later in the section on quantitation

experiments.
using chemical ionization mass spectrometry. A RIM experiment was performed on
malic acid (tri-TMS) and adipic acid (di-TMS) in order to determine the effect of
these solvated ions on [M+N113+] ion formation. The results of this reactant ion
monitoring analysis suggested that the ammonium adduct ions of the form H(NHB)2+
and HCNH3)3+ do not have a direct impact on the equilibrium concentration of the
molecular adduct ion species. Two possible explanations for this behavior are
considered. (1) Steric effects do not allow the "bulky" clustered ammonium reagent

ions to come in sufﬁciently close proximity to the analyte molecule for ammonium

attachment to occur since attachment to an analyte requires an available electron lone

I30

pair. For most esters, b0th proton transfer and ammonium ion attachment have been
found to be thermodynamically favored at the carbonyl oxygen. The same site of
reaction is presumed for trimethylsilyl derivatives of organic acids.”'100 (2) The
bond strength of the hydrogen atom to the solvated ammonia molecules is too strong
for the analyte to overcome. As Kebarle suggests, the hydrogen atom is bound tightly
through solvation of ammonia molecules that the bond strength is much greater than

the proton afﬁnity of the analyte for adduct ion transfer to occur.
Quantitative mixture analysis using CIMS

Chemical ionization mass spectrometry has been used routinely in analytical
chemistry for providing quantitative information on known compounds. Accurate
measurements of the amount of a particular substance present in an often complex
medium may be made. A number of approaches to attaining quantitative results on
unknowns have been described (using not only CI, but also El, and others). Perhaps
the most widely used method has been to add a stable-isotopically labelled analog of
known concengatign to a mixture containing the species to be measured. This is the
basis of isotope-dilution mass spectrometry. The amount of unknown present may be
deduced from a comparison of the mass spectrometric response of the unknown and
the known labelled analog. Care must be taken in using stable isotopically labelled
substrates, though, for quantitative determinations. The prevalence of isotope-effects

is a well documented phenomenomml"102

Deuterium, for example, can cause
severe isotOpe effects. Rate constants for reactions comparing the prOtium and
deuterium analogs (measured mass spectrometrically) can differ by a several orders
of magnitude. In addition, protonation and fragmentation reactions can also be

different for the unlabelled and labelled analogs. This point will be addressed more

131

fully in a later section.

Problems associated with chemical ionization, especially when combined with gas
chromatography have been described in detail by Millard.103 The capacity of
chemical ionization mass spectrometry as a successful quantitative analytical method
for mixture analysis has been the subject of much debate. Munson addresses the

104 Much of the

question of quantitation by chemical ionization in a recent review.
discussion has been summarized here.

In order to determine the quantitative capabilities of this technique it is ﬁrst
important to determine the sensitivity of the method and how it may differ with
structural changes in molecules (e.g., polarity). Methane-CI will be used for the
folowing discussions since it is the most common chemical ionization reagent gas,
although the discussion is valid for any other reagent gas which is involved in proton-
transfer reactions.

As was discussed earlier, reactions of the analyte with either CH; and/or C 2H;
can result in both the formation of the protonated molecule and a variety of fragment
ions, depending on the exothermicity of the reaction. The following equations
represent the reactions of these two reagent ions with the analyte, M, where 2 Pi

represents the total number of ions produced (i.e., both fragment ions and molecular

ions) by the ion molecule reactions described earlier. The rates of

k
17

CH5+ + M > mi (11)

 

k29

C211; +M »‘ > ZPi (12)

 

reaction for equations (11) and (12) are determined by the rate constants for proton

132

transfer . The rate equation governing these two processes may be given in the form
of a differential equation, which may be approximated as a difference, since
conversion of reactant ions to product ions is small (i.e., [reagent ions] >> [product
ions], c.a. 1000:l) as shown in Equation 13 below. Since no product ions are present

initially, the difference
cur?i l/dt=l It” [015*] +1.29 [czngl 1 (M1 (13)

is equivalent to the ﬁnal concentration of the added sample. In addition, the ion
currents measured in the mass spectrometer are representative of the relative
concentrations of the samples being analyzed, that is, 2 Pi may be approximated by
Eli , where Ii is the intensity of ion with molecular weight of i. "This equation
assumes that transmission of all ions through the ion source, and detection of the ions
is equivalent for all ions, which is generally valid. Upon integration the reaction can

be rewritten (eq. 14), where t refers to the residence time of the reactant ions:

21, = t l "n (1917 +5, (1929 l [M] (14)

This parameter is typically the most diﬁ'rcult to measure accurately for determining
rates of reaction (which accurately define sensitivities in CI). Two approaches have
been most frequently employed for determining this parameter accurately. The
residence time of the reagent ions in the ion source can be determined by studying
reactions of known rate. Secondly, these values have been found to be estimated very
accurately from drift velocity measurements.

Theoretical rate constants (k) for reactions involving methane have been

133

determined from Langevin ion induced-dipole theory, 105406

107

average Dipole

which more accurately estimates rate constants for polar
108

Theory (ADO).
molecule-ion interactions, and other more recent theories. Experimentally
obtained results, in general, compare favorably with theoretically obtained rate
constants. In addition, experimentally obtained values have been found to differ by no
more than two orders of magnitude for compounds as different as acetone and steroid
molecules. For a homologous series of aliphatic hydrocarbons, experimentally
obtained rate constants have been determined to be nearly identical (c.a. 1.0 x 10'9
cm3mol '1 sec'l). Determinations of rate constants from experimental data require
knowledge of the ion intensity relationship between reagent ion and product ion and
residence time of the reagent ion. Field ml, demonstrated that the rate constants for
ion molecule reactions of methane could be determined from this precursor-product

107 Small variations in the rate constants are attributed to changes in

ion relationship.
structures of the neutral compounds. These variations can be related to the mass of
the ions (and molecules), dipole moments and their polarizabilities.

For determination of rate constants from experimentally obtained data, it is critical
that the ion source pressure being maintained constant throughout analysis. This
becomes an especially important consideration when analysis of mixtures is made by
GC-MS and relative concentrations of unknowns are to be determined accurately.
Munson demonstrated that rate constants for a variety of ion molecule reactions could
be estimated fairly accurately through RIM, as mentioned earlier. RIM is analogous
to selected ion monitoring (SIM). Instead of monitoring ions which are indicative of
the analyte under investigation, RIM monitors a depletion in reagent ion
concentration as ion molecule reactions take place in the ion source. Rate constants

for eluting species from a gas chromatograph may determined based on the

knowledge that as a component elutes from the gas chromatograph, a decrease in

134

reagent ion concentration will be observed (presuming ion molecule reactions occur).
As an example, Munson and coworkers showed that the rate constant for the reaction
C2H5+ with an eluting species, B, may be obtained from manipulation of eq. 14, as
shown in eq. 15, below. Assuming the peak shape of a eluting species is nearly
triangular, the rate constant for reaction may be modiﬁed to the form shown in eq. 16,

where t2--tl is deﬁned as the width at base of the chromatographic peak, F is the
k = 1n 1 - Al + I + 2Et 1
{ C2Hs I (Czns )o H ( 5)
k=-F(t2-tl)ln{l-AIC2H5+/ICZHS+]max/2Et (16)

helium gas flow rate, and AI C211; is the peak minimum observed for the reactant ion

C2H5+ as the species elutes from the gas chromatograph. Since t is known with the
least amount of accuracy, it is customary to determine [elm rates of reaction, by
comparing the rates for two components eluting from the gas chromatograph. In this
way, the residence time of the reagent ion may be eliminated from the reaction. It is
evident from this equation that small ﬂuctuations in reagent gas pressure will affect
the measured rate constants. Recall that slight differences in ion source pressure can
result in dramatic variations in the abundances of reagent ions. As a result, inaccurate
measures of relative rate constants and relative concentrations will be made. A
method for achieving such a regulated ion source pressure has been to incorporate a
pressure regulator.

The importance of calculating these relative rate constants lies in the fact that
these values are an accurate measure of the relative molar sensitivities of the gas
chromatographic effluents. The dearth of data on experimentally obtained rate

constants for reaction and those obtained by theoretical calculations are generally in

135

good agreement. In addition, when comparing homologous series of compounds by
the RIM method, results suggest that Cl can indeed be used accurately to quantitate
the components of a mixture.

There has been a great deal of concern about the utility of chemical ionization
mass spectrometry for the analysis of complex mixtures of organic acids (and others)
isolated from physiological ﬂuids. It is suggested that due to the diverse chemical
nature and complexity of the isolated organic acids, rate constants for reaction may
differ substantially from one component to the next. For complex mixtures such as
organic acids in urine it is not uncommon to have coeluting species in both GC and
GC-MS analyses. Obviously, by GC alone, no quantitative information can be
derived for the analytes. However, by GC-MS, especially when using a relatively
"reliable" method such as E1, the concentrations of the coeluting species can often be
determined based on the intensity relationship of analyte ions. If coeluting species
give rise to an identical fragment ion in the El mass spectrum, the capacity to
accurately quantitate the two species will be markedly decreased.

Chemical ionization is evaluated as to whether the quantitative relationship of
coeluting species can be accurately reﬂected. The ability of the technique is further
tested when one of the two (or more) coeluting species is in large excess relative to
the other eluting species (a situation not at all uncommonly observed in metabolic
proﬁling analyses). It has been suggested that preferential ionization of the major
eluting species will occur, leading to an inaccurate quantitative representation of the
co-eluting species. In addition, a rapid depletion of CI reagent ions available in the
ion source is assumed as a concentrated analyte elutes from a GC column, thus
causing ﬂuctuations in reagent ion source concentration. As was shown in eqs. 15
and 16, this can lead to variations in calculated rate constants, and result in poorly

represented relative molar sensitivites.

136

Results of quantitative metabolic profiling analysis using CIMS

Experiments have been performed to identify whether CI-MS accurately reﬂects
the composition of mixtures exhibiting the type of behavior just mentioned. A
portion of this thesis investigates chemical ionization mass spectrometry and its
capacity for providing quantitative information on complex mixtures of organic acids
in complex media (e.g., urine). The following sections do not give a rigorously
detailed treatment of CI as a quantitative method (i.e., by yielding calculations of rate
constants for reactions) but the results presented suggest clearly that, when conditions
are held rigorously constant in a CI experiment, the ionization method is capable of
producing results that parallel those obtained from GC-EI-MS.

The results to be presented are obtained using ammonia as the reagent gas. There
are a number of reasons for this choice of reagent gas. Comparison of results using
methane, isobutane, and ammonia (at Optimized ion source pressure for best
sensitivity) on simple mixtures of derivatized organic acids revealed that Nl-Is-CI is
nearly three-fold more sensitive than CH 4-CI for the analysis of these types of
compounds. SIM results on p-hydroxybenzoic acid (di-TMS) are shown in Figure 51
a-b. Isobutane-CI also has been investigated. Typically, ion yields are found to be
5-10 times lower than those obtained using NH3-CI for these compounds. In
addition, results on simple mixtures suggest that the most accurate reﬂection of the
composition of the mixture is obtained when NHa-CI is used. This point will be
taken up in greater detail shortly.

The TMS derivatives of organic acids are an interesting class of compounds to
investigate for quantitative studies involving ammonia since these ester moieties

typically exhibit proton afﬁnities (experimentally determined and are based on

137

 

P optimum (NH3) = 320 mtorr

 

T!

 

 

A = 230000

 

'-----O

 

 

 

 

 

1!

 

 

 

 

 

, m
P Optimum (CH4) = 250 mtorr
A = 76000
kWW ---J-K. 1,,
‘7

 

 

 

Figure 51. SIM of p-hydroxybenzoic acid (di-TMS) comparing sensitivity

. obtained at optimum ion source pressure using (a) ammonia and (b)
methane

138

responses relative to other CI reagent gases, such as i-butane, methane, etc.) that are
very near, but usually slightly lower than that of ammonia, itself. For many of these
organic acids relatively intense [M+H]+ ions may be observed in the CI mass
spectrum even though the proton aﬂ'mities of these’analytes are presumed to be less
than that of NH3. Explanations for this result may be based on the fact that in a
molecule, some sites may be more basic than others, affecting to some degree the site
at which proton attachment may occur. Form a thermodynamic standpoint, the
carbonyl oxygen may be sufﬁciently basic for either proton transfer or NH4+
attachment to occur. In any case, the differences in overall proton afﬁnities between
reagent gas and derivatized organic acids is sufﬁciently small such that only
[M+NH4]+ and [M+l-I]+ ions are observed in the CI’ mass spectrum of these

compounds.
Experimental

All experiments were performed using a Hewlett Packard 5985 GC/MS/DS
equipped for capillary column chromatography. In most cases a 25m HP-S megabore
capillary column (i.d. 0.52mm) was used. A non-vitreous deactivated capillary
column (0.11 mm i.d.) was interfaced to the gas chromatograph and mass
spectrometer, providing a split ratio of approximately 5:1. The helium ﬂow rate using
the megabore column was adjusted to approximately Sml/min. When wide-bore
capillary columns were employed (0.32mm i.d.) the helium pressure was maintained
at approximately 1-2 mein flow rates. Reagent gas pressure was regulated by use
of a Granville-Phillips Pressure Regulator (model it AC—7096).

Source temperature for E1 and CI analyses was maintained at 150 'C for the

139
analysis of derivatized organic acids. In the chemical ionization mode, the operating
ion source pressures were maintained between 100-500 mtorr, though best sensitivity
was typically obtained near 300 mtorr. The electron energy was maintained at 120-
150 eV and emission current was 325 11A. For electron ionization, the electron
energy was 70 eV and emission current was 300 M.

Ultra high purity methane (> 99.9 %) and ultra high purity ammonia (> 99 %)
were obtained from Matheson Chemical Co., Joliet, 11. Organic acid standards were
obtained from Sigma Chemical Co. TMS-derivatives were made using
bis(trimethylsilyl)triﬂuoracetamide (BSTFA) purchased from Regis Chemical Co.

Urine samples were supplied from Meridian Instruments, Okemos, MI.

Results and Discussion

The proper choice of operating pressure is important in any chemical ionization
experiment. Ion source pressures are typically measured (quite inaccurately) based on
ion guage measurements, which are typically located far from the ionization region
for CI analyses. Figure 52 shows plots of ion guage pressure readings vs. pressures
obtained from the postioning of a thermocouple at the entrance of the CI volume with
measurements of pressure made using a Hastings guage for ammonia and methane.
These values are more accurate measures of the CI ion source pressure.

Furthermore, tuning of the instrument at one particular pressure for optimum
calibrant signal does not always ensure that this same pressure is the optimum
pressure for gas chromatographic analysis. The best approach to achieving the most
reliable chemical ionization results is obtained following a standard experimental
design. For proper calibration of the instrument for CI analyses,

perﬂuorotributylamine (PFI'BA) is used. Tuning is achieved using methane as a

TRUE CI VOLUME PRESSURE (in u.“ )

140

'1 I -Methane

 

loo—g

 

 

o IIIIIIIITIIIIIIUTTTIITTIIIIITITIITIIITﬁ

0 1.0 2.0 3.0 4.0

SOURCE PRESSURE (in tort, x 10 EE 4 )

Figure 52. Comparison of true CI ion source pressure vs. source housing
pressure

141

reagent gas, regardless of whether or not the analyses to follow will use ammonia,
isobutane, methane, or some other reagent gas. A typical tuning ﬁle is shown in
Figure 53 below. The relative peak heights of m/z 69 (CF35, 219 (c4133 , and 414
(C8F16N+) are manipulated by optimally tuning the ion lenses and repeller. It is
important that; (a) the instrument be tuned daily when operating in the CI mode (since
performance degradation occurs more rapidly in this ionization mode than in the El
mode) and (b) tuning ﬁles do not change dramatically from one day to another. That
is, the relative ion intensity ratios for the calibrant should remain nearly identical if
CI results are to be used for quantitation.

Once these tuning procedures have been accomplished, a series of standards
should be run to detennine by GC-MS to determine the optimum operating pressure
for analysis. Shown in Figure 54a and 54b below is the relationship between iOn
source pressure and ion counts for maleic acid (di-TMS) and p-hydroxybenzoic acid
(di-TMS). Typically, optimum sensitivity is achieved at an ion source pressure of
300 mtorr. However, as ion source degradation occurs over time, experience has
shown that higher source pressures are needed to obtain optimum ion signals.

More importantly, for quantitation, is to investigate the relationship between
analyte ion signals and ion source pressures. Molecular weight information, in the
form of the protonated molecule and molecular adduct ion species, are observed in the
NH3-CI mass spectra of most derivatized organic acids. Figure 54 also shows how
ion source pressure affects the intensity relationship of these two ions. In most cases,
the ammonium adduct ion, [M+NH4]+, reaches a maximum abundance at about a
pressure of 300 mtorr. This information is important for the reason that the intesity

relationship between the protonated molecule and molecular adduct ion species can

often be used to differentiate isomers. As an example, the mass spectra of p-hydroxy

L42

ﬁzammuv 3.. {a
"rang an as “rang .% as use? 265.. o... .- 2E
agesaeoéba ma... 5 «and 8e§§ 355 .3}... an Bum

 

 

 

e3 3% 5e 2 2..
R o a: 82 2 an
are 8" a: 3.8
anus“ mnmaxanamdrgmm_ muzaaanuuVMHuuaanc. wwcﬁﬁqddnua.
3.. 3.. «mm 1 m3 on l we
. d
\\ 3N ..n a): a. u a}:
.5. u as
uremmmmoooq 27555205
o2 tgmomsa own-23 22mg
Each «22 angréen
Sugarcane—o 3?. 3.3.2 o: 1:938;ng
$83 .220 82 8...: a: :3 80829
a. t EEO :2... me .. S 50:33

3.54.092... , «TEE.
88rm0<50>2m Sz<~42§~ Emma

11C

TIC

Figure 54.

143

mm (A)

 

 

 

O~meﬁm

0.5 1.0 1.5 2.0 2.5 3.. 3.5

 

 

NH3 Pressure (x 10 EE +4)
300001 (B)

1 4a- M261
250001 _y M278
15000
10000-1

4

d

1 \-

o 1 z“

0

.5 1.0 1.5 2.0 2.5 3.0 3.5.

Reactanl Gas Pressure (x 10 ea 04)

Effect CI ion source pressure has on overall sensitivity for (a) p-
hydroxybenzoic acid (di-TMS), and (b) Maleic acid (di-TMS). mlz

261 <MH”); mlz 278 (MNHJ)

I44

benzoic (di-TMS) and m-hydroxybenzoic (di-TMS) acids, are shown in Figure 55 a-
b. The differences in peak heights of the MH+and MNH4+ ions allow the two
isomers to be differentiated fairly readily, and the ﬁdelity of the ion intensity ratios
seems to hold over a wide range of ammonia pressures.

When methane CI has been employed for the analysis of the same class of
compounds, the spectra appear be much more pressure dependent. As an example,
shown in Figure 56 is the affect methane reagent gas pressure has on the ions formed
in the analysis of 2-hydroxybutyric acid (di-TMS). At higher source pressures the
protonated molecule and molecular adduct ion, [M+ 02H 51+, begin to increase in
abundance relative to the lower mass fragment ions. The lifetimes of these
pseudomolecular ions are affected by the increased ion source pressure, as was
discussed earlier. Increased rates of collisions between neutral reagent molecules
and analyte ions causes more rapid stabilization of the ion. As a result,
decomposition will not occur sufficiently fast compared to the collisonal stabilization
reactions. A second factor for observing an increased abundance of molecular adduct
ions is that the concentration of [C2H5+] is increased in the ion source and more of
this reagent ion is available for both proton transfer and attachment reactions.
Difference in proton afﬁnities between analytes and methane is sufficiently large that
even C2H5+ can play a large role in governing proton transfer reactions. As a result,
spectra are inﬂuenced to a larger degree by high ion source pressures, where the
[C2H5+] concentration is enhanced.

As mentioned, an overall enhancement in sensitivity has been observed when
ammonia is used rather than methane. The reason for this enhancement in sensitivity
may be a result of the dramatic affect ion source pressure has on governing the rates

of reactions in the ion source. The initial reaction observed in the methane-CI

I45

AmscﬁaéuﬂioxlcxgwnaSH

ﬁe?a§.§§§a§§é33§a§0&i "nuns-E

o¢n oun con onN ooN O+N CNN oou cop cow 0‘; ON— 00— on

0” 0:20.:

 

*H 86.,“ m I

 

11“ l1 '1 J

 

 

 

«9:32; omoncoaXxounxclq so Eauuooqu nooE ~0ln22

Otn own oon ooN omN o¢N CNN OON our 00— 0‘; ON— cow on

_bb’bh”bb
1‘

._

01.1203

 

 

en 0:21:

enxeau r

 

 

 

[usuaatq asrmax’

madam] among.

Fraction 01110

146

0.30 -

¢+++¢
?
§

 

O

 

 

o'w t t ' ' I ' ' ' ' I V I I I r U v Vﬁ'

I
0-5 1-5 2.5 3.5 4.5
Pressure (x 10 ee +4) '

Figure 56. Dependence of fragment ions and abundances of 2-hydroxyisobutyric
acid (di-TMS) on methane ion source pressure

I47

experiment is proton transfer from the acidic CH; ion to the less acidic organic
compound. Under ammonia-CI conditions, the most likely initial reaction is the
formation of the [M+NH4]+.ion, as was described in eq. 9a. These initial reactions
occur at appriximately the same rate. Under high ion source pressures, stabilization
of the analyte ion occurs readily, leading to the formation of the [M+NH4]+ ion in the
CI mass spectrum.

It was mentioned earlier that Nils-CIMS analyses more accurately reﬂect the
quantitative composition of synthetic mixtures of organic acids than does CH 4-CI,
and for this reason ammonia has been chosen for the quantitative studies. These
experimental observations are based on a presumed spread in the proton afﬁnities for
the complex mixtures of organic acids; the effect that methane reagent ions have on
relative molar sensitivities and spectral variations is large relative to ammonia. Recall
that the PA (NH3) = 205 kcal mole'l, PA (CH 4) = 130.5 and PA (c2114) = 160.5.
For most organic acids, proton affinities are believed to be much greater than those of
the methane and ethylene. Therefore, they will act as Breasted bases in gas phase
reactions with methane reagent ions. Differences in proton afﬁnities of the organic
acid molecules, due to differences in their polar character, for example, will result in
slightly different reactivities with methane. Though literature values of rate constants
for proton transfer reaction appear not to differ by more than a factor of two to three
for most organic compounds, slight variations in reactivity will be observed, causing
an inaccurate representation of the relative concentrations of the components of the
mixture by GC-CIMS.

For reactions involving ammonia, on the otherhand, results suggest that a
electrophilic attachment reactions play a major role in governing the types and

quantities of ions produced in the NH 3-Cl mass spectrum. For organic acids, the site

I48
of electrophilic attachment has been shown to be at the carbonyl oxygen of the

99’1” which is a common feature to all organic acids.

carboxylic acid moiety,
Only for cases in which the carbonyl oxygen is sterically hindered will ammonium
ion attachment be potentially less favorable. This situation might occur for the
ketoacids (i.e., organic acids containing a keto group at the a-carbon to the carboxylic
group) that have been derivatized with hydroxylamine.-HC1 followed by

trimethylsilylation to form the (Me3)3Si02C-C=N-O-Si(Me moiety. The bulky

3’3
trimethylsilyl group might create some steric interference for ammonium ion
attachment to occur. However, results on complex mixtures of organic acids isolated
and derivatized from urine extracts do not reveal signiﬁcant variations in relative
response factors for these species using ammonia-CI.

Studies involving ammonia-CI for determining the quantitative strengths of this
technique relative to electron impact ionization (the preferred method for organic acid
metabolic proﬁling analyses) are the focus of the following section. In order to
determine the relative capabilities of NH 3-CI as a suitable chemical ionization reagent
gas, an experimental procedure had to be devised which would take into account
conditions for which CI methods have been previously considered incapable of
yielding quantitative information (e.g., situations in which component A is found in
large excess relative to a coeluting species, B). A flow chart representing the types of
mixtures analyzed and compared by b0th EI and NH3-CI are shown in Figure 57.
This flow chart represents a number of synthetic organic acid mixtures that have been
prepared and analyzed by both EI and NH3-CI mass spectrometry. Small mixtures
containing coeluting organic acids and components added in large excess have been
studied. More complex mixtures varying in concentration from low ng to pg

quantities have also been compared using both El and NHa-CI with the long-term

goal of determining the quantitative strenghts of this CI method in the analysis of the

149

Synthetic Mixtures Used for Determining Quantitative
Strenths of Ammonia-CI Relative to 70 eV EI-MS

 

LN]

1 A+B a: Coeiuting species

C 2 Competitor
LA + B + q

S an Internal Standard
1 N C”

L A+B+C+S j

i

I Organic Acid Mix?

ﬂ

/ \

PLASMA URINARY
AMINO ACIDS ORGANIC ACIDS

 

 

 

 

 

 

 

 

 

 

 

Figure 57. Flow chart representing scheme for determining the quantitative
strengths of NH 3-CI relative to EI-MS

150

urinary organic acids. A number of goals have been outlined for the use of ammonia-

CI as a routine ionization method in metabolic proﬁling studies. They are as follows:

1. Determine the effect interferent or analyte species have on the
ability to obtain qualitative and quantitative information on the
analyte of interest.

2. Show the quantitative strengths of the technique and also
those situations in which they exceed the capabilities of
conventional El.

3. Show reproducibility of technique from one experiment to the next.

4. Show that an enhancement in sensitivity may be achieved over El.
when comparing only those ions that are indicative of the analyte.

Synthetic mixtures of organic acids commonly found in urine were prepared.
Table 5 shows the organic acids used in preparing both simple and more complex
synthetic mixtures. Mixtures of some of these organic acids were prepared to
contained at the least, (a) coeluting species, (b) intentionally added competing organic
acids (that is, an organic acid added in varying concentration to the coeluting species
with a retention time very near that of the principal analytes), and (c) interferents
(e.g., byproducts of derivatization procedures) in order to simulate as closely as
possible conditions typically observed in the analysis of real urine samples.

A two-component mixture of the coeluting species 2-hydroxybutyric (l40ng/ttl)
and glycolic acids (IOOng/ttl) was prepared and analyzed by both conventional EI and
NH3-CI. Earlier it was mentioned that there are cases in which coeluting species can
give rise to identical ions in conventional EI mass spectra. This experiment was
designed to demonstrate that in some cases ammonia-CI can provide more accurate

determinations of the relative concentrations of the components of a mixture than EI-

Table 5. List of structures, formulae, molecular weights and retention indeces
for components of synthetic organic acid mixture

Acid: A

Chemical ormula: c‘awo,
Molecular Weight: 14‘
Retention Index: 1501

Acid: 61

Chemi Formula:czn‘o,
Molecular Weightt‘l‘
Retention Index: rm

Acid: 0 hydroxyisobutyrh
Chemical Formula: (“11.0,
Molecular Weightnu
Retention Index: 1068

Acid: arhydmxyisovalcrlc
Chemical Formula: c 1100
Molecular Weight: 11:5 l 3
Retention Index: 1171

Acid: Maleic
Chemical Formula: Q1140
Molecular Weight: 116

Retention Index: 121a

Acid:Mal’n

Chemical Formula: c‘ﬂ‘o,
Molecular Weight: 134
Retention Index: 1501

Acid: meth Imalonic
Chemical ormula: C‘H‘O
Molecular Weight: 118 ‘
Retention Index: 1212

Acid: Oxal'n

Chemical Formula: czﬂzoe
Molecular Weight: 90
Retention Index: 1119

Acid: Phen lacetb

Chemical ormula: c.a.O2
Molecular Weight: 136
Retention Index: 1280

Acid: m-hydroxybenzoic
Chemical Formula: (2111‘03
Molecular Weight: 13!
Retention Index: 1570

Acid: p-hydroxybcnzoic
Chemical Formula: (:71‘1‘03
Molecular Weight: 13!
Retention Index: 1622

4

:3 911mm 0
',
(TMS)HO.-C-CH- crz-c-ortcms;

9 9
mmm-c— (EH-c-oncms)

0':
: :
mm-c-c- ourms)
o

mz-g-OHM
@ 9
$/C"OH(TMS)
OH

:3 M

152

MS. The mass spectra obtained of glycolic and 2-hydroxyisobutyric acids by
ammonia-CI and El are compared in Figure 58a and 58b, respectively. Designate
ions in the E1 mass spectra of 2-hydroxybutyric acid are mlz = 233, 205, and 131 and
for glycolic acid, mlz = 205, 177, and161, as shown in their individual mass spectra.
The isobaric ion for these two species is mlz =205, corresponding to [M-CO]+ and
[M—CHa]+ for 2-hydroxybutyric and glycolic acids, respectively. When determining
the relative concentrations of analytes in a mixture the integrated areas under the
peaks are typically measured and compared with the integrated area under the peak
for the internal standard.

However, when two or more species are contained within one chromatographic
(i.e., reconstructed TIC) peak, then relative concentrations are determined based on
the integrated areas of the designate ions for each species. In the case of the two
components shown, the isobaric ion of mlz =205 makes determination of the relative
concentrations difﬁcult. If fact, calculations of the relative concentrations based on
these designate ions yield very poor quantitative results. When the same species are
analyzed by ammonia-CI, the problem of isobaric ions is removed. Only molecular
ions are observed in the NHs-CI mass spectra of these compounds. No interfents in
the mass spectra are present to make calculations of the relative concentrations
complicated. The results comparing the quantitative strenghts of these two techniques
for this simple mixture are summarized in Table 6 (see mix 1). In addition, several
Other simple mixtures were prepared and the results of calculated concentrations
based on El and NH3-CI mass spectral data are also summarized in Table 6 and
compared with the known starting concentrations. The comparison between actual
concentrations of the organic acids and the"counts" obtained upon integration of the

TIC chromatograms is made on a relative scale. For Mix 6, as an example, the

153

 

l‘ a" fit an
2" 2335
i

 

 

 

 

 

 

 

 

 

 

 

 

 

‘ i 1778
161
4W-rf4r—HJ ‘ I 3
ac ~ are t W21 a a a a
, Mout‘i;
O
‘ (M‘m‘)'
g d
3
E 4
O
i ., . mime
‘ (Mil-i);
.r.4 w. n“. inn. 1.
It. 8L IL ll. SL .8. 1.
N1

Figure 58. Comparison of quantitative strengths of E1 and NH3-Cl for the
analysis of 2-hydroxyisobutyric and glycolic acids. (a) Reconstructed
mass chromatograms showing isobaric ion interference for E1, and (b)
comparison of NH3-CI and 70 eV EI mass spectra

i54

1700 MZOS'W'
soc MZOS'W

g:
is

11001005)

§ll§i £3
"3' -i-ii

§§§§3§§§ rat H H as.

3
aegiséﬁig Mi: :2 ii
i

 

43333233133533 ii .

componentorganicacidmixnuecwithintegmedpeakareasobtainedfrem’
ysee.

Comparisonbetweenactualeoneenn'ationaofseveraltwoandthree
GC~(NH3-Ci)-MS and GC-BI-MS anal

Table 6:

155

lmown, relative concentration ratio of p-hydroxybenzoic : m-hydroxybenzoic acid is

approximately 4:1. A comparison of the A by NHa-CI confirms the 4:1

TIC
relationship, whereas the results from El suggest that the relative concentration

relationship is approximately 6:1. In many cases, the results from El and N113
parallel each other quite closely. The results on the analysis of these simple mixtures
suggest the potential of ammonia-CI as providing very accurate quantitative
information on simple mixtures. Even in those cases for which one component is
found in large excess relative to the other components of a mixture, such as in
mixtures 3 and 4, an accurate determination of the relative concentrations can be
made.

The capabilities of ammonia-CI relative to 131 are further tested by comparing the
results on a larger synthetic organic acid mix (the structures of the components
contained within this mixture are found in Table 5). The TIC plots obtained using
N H3-CI and E] are shown in Figure 59 below. The relative responses of the
components of the mixture are quite similar for the two techniques. Summarized in
Table 7 are the calculated integrated peak areas from the experimental results which
are compared with the "true" values of those compounds . In general, the two
techniques accurately reﬂect the concentrations of the components of the mixture. In
addition, an increase in the TIC is observed for the ammonia-CI analysis. In general,
the TIC "counts" for El and NHs-CI analyses are very comparable for the analysis of
the trimethylsilyl derivatives of organic acids. It must be mentioned, though, that
much of the TIC generated in an El analysis is in the form of structurally useless ion
current, where as the majority of ion current afforded in a NH3-CI analysis is

associated with the protonated molecule and molecular adduct ion species (usually 2

90 per cent).

156

 

   
  

 

  

 

 

 

 

 

3 P-lylronybonloio
1008- 9744
‘rl'lfm'
01100010
n—lylrnﬂuulo
‘ rm":-
..'. .' . Plonyiocotlo
d
loll‘l
a an”. Min 0
I‘ I U."..
443,10
jjjjlrII‘IITlTIIIIUIITTU‘IIITerfT‘lII

 

 

 

 

 

 

 

 

0 100 200 300 400 500 600 700
J p-Iidroryloaaolo
1003- 19472.
2-Iylroxy- Plouyloeotlo
loobutyrlo
ICIIIOI;:."'..{- rlylror’DOIIO'O
loovotor o
4 OII".
. \ mm.
-1 ::::1.
_4 - LL. g” L .. L A. i-
0 100 200 300 400 500 600 700 800
Soon

Figure 59. TIC proﬁles of the trimethylsilyl derivatives of an organic acid
synthetic mixture obtained by (a) GC—EI-MS and (b) GC-(NHa-CI)

MS

157

Table 7. Comparison of integrated areas from El and NH3-CI analyses on an
eleven component organic acid mixture. Actual concentrations are

 

 

 

 

 

 

shown in the lower half of this table
99131110111!!! AJED W Amsﬂ W
Oxallc 3000 0.07 16000 0.12
tat-hydroxyisobutyric ~ 16000 0.3 29700 0.22
glycolic 14000 0.3 22200 0.17
at-hydrortylsovaleric 11000 0.2 23200 0.17
methylmalonlc 370 0.08 13900 0.11
phenylacetlc 10300 0.2 35100 0.26
Unknown (MW-292) 5800 0.1 13900 0.10
Mallc
1600 0.1 11500 0.08
Adiplc
m-hydroxybenzoic 22 000 0 .5 63100 0 .47
p-hydroxybenzolc 46000 1.0 134600 1.0
ACTUAL CONCENTRATIONS AND REL. ABUNDANCES
118/ 1!! In]. 1201. Cone.

Oxallc 80 0.1

or hydroxylsobutyrlc— 310 0.3

glycolic 280 o 3

a-hydroxyisovaleric— 200 0. 2

methylmalonlc -——- 95 0.1

phenylacetlc —————- 220 02

Unknown MW=292)— - _.

Malia 75 0.5

Adlplc 45 0.8

m-hydroxybenzolc —— 460 05

p-hydroxybenzolc —- 920 1.0

158

The capacity for obtaining quantitative information on these mixtures depends on
the NH3 ion source pressure utilized in the CI experiment. Maleic acid is unique to
the list of organic acids summarized in Table 6 due to its unusually high proton
affinity. Maleic acid (di-TMS) readily abstracts a proton from NH“+ under
conditions of low ion source pressure (S 300 mtorr). When operating pressures are
such that proton transfer is favored for this organic acid, poor quantitative
representations of the components of a mixture containing maleic acid will result.
However, if the operating ion source pressure is such that stabilizion of electrophilic

NH + attachment to the organic acids is favored, accurate measures of the relative

4
concentrations of the components of mixtures containing maleic acid can be obtained.

Figure 60 compares the response of maleic acid (di-TMS) relative to phenylacetic
acid (di-TMS) by NH3-CI at (a) low, and (b) high ien source pressure. The analysis

performed at PNH = 200 mtorr showed a marked deviation in relative response of
3

maleic acid when compared with results available from EI-MS (which were taken as
the "true" relative concentrations of the two species). However, when the ion source

pressure was increased to PNH = 330 mtorr, the relative response of maleic acid more
3

closely paralleled the results from the E1 analysis. These results further support the
idea that the initial reaction between NH 4+ and organic acids is in the formation of
the [M-~H---NH3]+* complex. When Operating pressures are low (and hence the
time between collisions with neutral reagent molecules is .greater) proton transfer can
occur to species such as maleic acid. At higher operating source pressures, the
frequency of collisions is such that the [M-"Hn-NH31” complex is sufficiently
stabilized to yield the adduct ion, [M+NH 41 +.

The results on these synthetic mixtures of organic acids suggest that the types of

responses obtained for BI and NH3-Cl are quite similar for these compound classes.

159

 

806*. 4,864.

 

Maleic

 

 

 

 

 

 

 

 

 

 

 

 

 

86 .ﬂ

‘
_‘ :oox- sane. 233- 2;!
d
1
‘
‘
, L-
r"'Ifﬁ'lnf'l“'*I“T'T""l"*'If"
"aoo 2395:. / c
- Maleic
‘
ﬂ
1
1 @—
—‘“' ‘ﬁ"[e"'f"" "f'l"" 'f" 'fr
:00 use 406 «so see also see one no

Figure 60. Comparison of TIC proﬁles of a mixture containing maleic acid (di-
TMS) using NH3-CI at (a) PNH = 200, and (b) PNH = 320 mtorr
3

3

160

The next step was to determine whether the proﬁles obtained by ammonia-CI would
parallel those obtained by conventional El for mixtures of organic acids isolated from
urine. Shown in Figure 61a-b below are the reconstructed TIC proﬁles of urinary
organic acids isolated from a diabetic patient obtained by El and NHa-CI. Inspection
of the two profiles does reveal variations in the relative peak intensities, but these
differences are slight. This situation was of some concern. When repeat analyses of
this urine mixture were made in the BI mode, the proﬁles also showed slight
differences, just as was observed when comparing the E1 and NH3-CI proﬁles. Repeat
ammonia-CI analyses on the same mixture were quite reproducible, as illustrated in
Figure 62. The reasons for the slight variations in reproducibility in the TIC proﬁles
from one run to the next might be attributed to poor equilibration of the gas
chromatograph prior to analysis, resulting in slightly different rates of heating, and
poor reproducibility in injection technique.

Quantitation by NHa-CI is not limited to the analysis of urinary organic acids.
NHS-CI has been found to very accurately represent the concentrations of synthetic
mixtures of TMS derivatives of steroids. The relative responses obtained by NHa-CI
and El are further compared for the analysis of plasma amino acids, as shown in
Figure 63. The trifluoroacetate derivatives of the amino acids were formed and the
resultant derivatized mixture was analyzed by El, NH3-CI, and negative ion chemical
ionization (NICI), and compared. In the cases of NICI, relative responses differ
signiﬁcantly for the components of the mixture. However, the TIC proﬁles obtained
by EI and NH3-CI are more closely related

As stated earlier, the treatment of ammonia-CI here as a quantitative tool has not
been as rigorous as the ones summarized earlier in the work of Munson and

coworkers in which relative rate constants for reaction of homologous series of

161

.gasacaeggg
ugqusaeaasgsgaﬁg $85..

‘8

08¢ 00. 00. 008 00. 00. 00‘ 00. 00. 8d

 

Huh

 

U—rIjTT—

I

 

 

0.. 0.. 0.5 0.. 80 8' 0a 0.. 00d 0.

 

. O

NH...

 

 

 

 

162

dsﬂﬁaxeadbisudiiaaﬂusg
eggéééxﬁeﬁong

 

 

 

 

 

 

1I‘I'l'

 

 

Figure 63:

. rusm mm the;

163

 

 

 

 

 

 

 

 

 

 

   
   

 

 

 

 

emin-
. (a1
"‘4
C ' na-
3‘ ammu-
g, ”W" - y m an:
r em mu: m
. 1 ALL L ...L
I. m an m . m u. no
(b)
4
g: 4
g .
§
8 J
‘ 1 ‘ I 1
u m an m a u. we
Senna
4 1".“
/
T
i - Maﬁa ‘1?!”
J ‘41-le
L‘s“
1 Volta. (
I # ._, «L
200' 3‘0] 100‘ $.00] «0] reel #10.

 

TIC proﬁle for naturally occuring amino acids in a diabetic dog plasma
obtained by (a) NHa-CI, (b) BI, and (c) electroncapture negative Ion .

(ECNI) mass spectrometry.

164

compounds have been determined. What the results presented here do show is that
ammonia-CI can, under rigorously controlled conditions, yield data which compare
quite favorably to the data generated from the accepted, conventional El ionization
approach. Reconstructed TIC chromatograms are found to be quite accurately
represented, whether El or NH3-CI is employed. Furthermore, integration of
individual peaks in the generated reconstructed TIC chromatograms represent
accurately the actual concentrations of the components of the synthetic mixtures. In
some cases, the strengths of NH3-C1 surpass those of El for quantitating on coeluting
species.

To this point ammonia-CI has been shown provide an accurate representation of
the components of complex mixtures. However, the real strengths of the technique
lie in the fact that the ions generated in an NH3-CI analysis almost exclusively
represent the analyte (i.e., very little or no ion current is associated with the
derivatizin g agent). This is an attractive feature of the CI method since much lower
limits of detection can be attained when monitoring for a known compound in a
complex mixture. The presence of only molecular-type ions in the ammonia-CI mass
spectra can be exploited. For metabolic proﬁling analyses stable-isotopes are
frequently used to investigate metabolic pathways of either previously known
disease-conditions or to investigate unknown conditions. The use of chemical
ionization methods for these types of experiments is the focus of the following

chapter.

165

Chapter 6: Utilizing Ammonia-CI for Stable-Isotope Tracer Studies

Introduction

Modern methods for research in biochemistry, physiology, and nutrition typically
have involved the use of either radiolabelled or stable-isotopically labelled substrates
to determine metabolic pathways of signiﬁcance. Early studies exclusively used
radiolabelled species due to the fact that these species were relatively easy to obtain
and methods for detecting the presence of radioactive materials were relatively
sensitive as well as inexpensive. For analyses involving radioactive labels,
techniques such as radioirnrrlunoassay110 have found a great deal of success. The
use of these radiolabelled species is prevalent in the area of 111m metabolism . “l
The inherent dangers associated with the use of these substrates, however, make them
often undesirable for use in chemical analyses, especially when those analyses
involve investigation of human metabolism. 1 12

With the advent of mass spectrometry, it has now been possible to incorporate
stable-isowpically labelled substrates to detenrline an enormous range of information
about analyte(s) under investigation. Stable-isotopically labelled substrates (or
tracers) are an attractive class of compounds due to the fact that they are readily
detected using mass spectrometry, and are safe to use. The major drawbacks
associated with the use of stable isotope species is that their cost and preparation is
quite high, often precluding their attainability for routine chemical analyses. A
second drawback, as mentioned earlier, is that these stable-isotope species can often
cause severe isotope effects. Perhaps the most sever isotope effects have been

observed with deuterium labelled substrates. Chromatographic retention times, for

example, can be quite different comparing the non-labelled and isotopically labelled

166

substrates. In addition, ionization efﬁciencies can differ for the labelled and
unlabelled analogs. These different ionization efﬁciencies affect rate constants for
reactions, and these effects must be evaluated before they can be used properly for
obtaining both qualitative and quantitative information from mass spectral analysis.

Stable-isotopes have found applications, nevertheless, especially for kinetic

113-114 S 115

studies, use as internal standards for quantitative analysis by M as well

as several for other purposes.116 Perhaps the greatest number of applications of
stable-isotope experiments to mass spectrometry today is in the investigation of
biochemical pathways. Insights into various metabolic processes may be determined
through the use of these labelled precursors and the results are often useful for
assisting in the identiﬁcation of speciﬁc metabolic errors, from which a proper

regiment for treatment of disease may be made. A number of stable isotopes have

been used to study various aspects of carbohydrate metabolism,“6'118 amino acid

“8'1“ and the Krebs (T CA) cycle. In The applications of stable-

isotOpically labelled analogs in evaluating various aspects of drug metabolism123 is

metabolism,

unsurpassed.

The types of stable-isotopes incorporated have been nearly as varied as the
numbers of stable-isotope labelled tracer experiments performed. Much attention has
been placed on the use of 2H, 13C, 18O, and 15N labelled substrates for these tracer-
type experiments. Studies involving deuterium have been.found to be hindered by the
fact that the deuterium labelled species can, in many cases, readily undergo exchange
reactions. This can present severe problems in determining accurate levels of label
enrichment. However, due to the relatively low cost associated with obtaining
deuterium-labelled substrates, this labelled species continues to be the most widely
employed. l3C—labelled substrates are playing an increasingly important role in

metabolism studies. The preparation of uniformly labelled l3C—precursors however,

167

is still not trivial, thus making the cost associated with these types of labelled species
quite high.

A number of questions are addressed, in general, when incorporating labelled
substrates for these types of studies. They are: (a) what metabolic products result
from the metabolism of the labelled precursor molecule?, (b) how much of the label is
incorporated into the metabolite product(s)?, and (c) at what position in the metabolite
product(s) is the label incorporated? The proper choice of mass spectrometric method
will depend on the type of information sought in the analysis. Typically, the greatest
source of information to the analyst is the determination of the metabolite products of
the ingested labelled precursor. Determinations of the metabolites arising from the
labelled precursor by mass spectrometry requires that an appreciable enrichment in
isotope labelled ions be observed. The ﬂux of 13c from a labelled precursor, such as
[U-13C1-glucose, into various metabolite pools requires that the extent of isotope
enrichment be discernible from natural isotopic contributions.

Most metabolism studies utilize only very low concentrations of fully labelled
precursors introduced to the system. As a result, the ﬂux of labelled substrate into
various metabolite pools will, in general, be quite low. This especially becomes a
concern when there are already large natural isotopic contributions to contend with.
In the analysis of the trimethylsilyl derivatives of organic acids, for example, at
minimum, a 5.1 % natural isotope is observed at the (A+1)+ ion and 3.4 %
enrichment at the (A+2)+ ion (corresponding to the natural isotopes of silicon). In
addition, contributions from naturally occurring 13(2 and ‘3 o isotopes further add to
often very large natural isotopic levels. Only when the naiural isotopic contribution is
known accurately can the measurement of label incorporation from a stable-isotope
precursor be deduced. This requires that measurements of isotopic enrichment be

measured with a high degree of precision and accuracy.

168

Isotope-ratio and isotope-dilution mass spectrometry

A technique which has found the greatest success in determining very trace levels
of isotope enrichment (less than 0.1 % enrichment) is that of isotope-ratio mass
spectrometry.”4'127 This technique is based on the simultaneous detection
measurements of natural isotopic levels (of a reference species) with the analyte
under investigation using null point methods. Isotope-ratio mass spectrometry may
involve the analysis of permanent gases, such as C02, N2, and H20 for determining
very precisely and accurately low levels of isotope enrichment, or inorganic transition
metals for determining accurately the relative amounts of their natural isotopes. One
of the most commonly measured isotopes in isotope-ratio mass spectrometry is 13C,
which is determined from carbon dioxide in respired air. The air is collected in an
evacuated chamber and the CO2 is puriﬁed and isolated by distillation.128 Analysis
is typically carried out along with reference CO2 and mass spectral analyses are
carried out simultaneously, yielding very precise and accurate measurement of
isotopic enrichment of the test sample. Hachey, ELﬂL reviews this technique for
stable-isotope studies in the nutritional sciences. 129 The limitations of isotope-ratio
mass spectrometric analyses are obvious. The sample being analyzed must be
presented to the mass spectrometer in its most elemental form. The technique is not
capable of the analysis of intact organic molecules, since combustion of the organic
compound is required by the isotope-ratio method.

Isotope-dilution mass spectrometry (IDMS) on the otherhand, is a technique
which has found a place in analytical chemistry for problems requiring the very
precise and accurate quantitation of analytes from complex media, such as plasma,
serum, and/or urine. As early as 1970, stable-isotope dilution methods were being

incorporated to make quantitative determinations of biologically important metabolic

169

115 possible. A number of 1’31””s are ”Med yearly

Species, such as prostaglandins,
using IDMS for determining the amounts of organic constituents of serum ( and other
physiological ﬂuids).130'l32 Typically, CC or HPLC methods133 are used in
conjunction with mass spectrometric analyses by this isotope-dilution method.

In such experiments, a labelled analog of the analyte to be measured is
incorporated at some point prior to extensive work-up of the physiological ﬂuid. This
allows for factors such as loss due to derivatization and isolation to be effectively
eliminated. Prior to introduction, however, solutions containing labelled analyte and
unlabelled species are prepared in varying concentration. Typically, these standard
calibrant solutions can range from 1:1 mixtures of unlabelled to labelled species to
ratios as low as 1:1000, depending on the species being sought and the medium from
which the analyte is measured. Once a linear calibration curve has been obtained
based on measurements of the integrated areas of unlabelled and labelled species,
analyses on the physiological mixture can then be conducted in order to determine the
amount of analyte in that medium. The accuracy associated with this technique for
determining concentrations of unknowns contained within complex physiological
solutions is great. In addition, isotope-dilution methods have been used in the study
of kinetics of bile acid metabolism, and others. 134436

The majority of stable-isotope experiments performed today involve the study of
metabolic pathways, such as differences in the metabolism of healthy and disease-
state conditions. Stable isotope tracers have been used predominantly for measuring
the ﬂux of an ingested stable-isotopically labelled precursor into various metabolic
pools for metabolic proﬁling analyses. The labelled substrate is typically
administered (either orally or parenterally) into both a "control" and ”disease-state”

subject, in order to determine subtle and/or significant differences in the metabolism

of the precursor molecule. These experiments require the ingestion of a known

170

labelled precursor for a pre—determined period (frequently referred to as pulse-
labelling period) followed by periodic collections of physiological fluids, (such as
urine, plasma, etc.) during a ”chase" period from which the compounds of interest can
be isolated and analyzed for label enrichment as well as the position to which the
precursor molecule may transfer label.

The number of papers in which labelled precursors are utilized for determining

various metabolic pathways involving intermediary organic acids and amino acids is

137438 The technique best—suited for these analyses is gas

growing rapidly.
chromatography combined with mass spectrometry. As mentioned earlier, the
ionization mode employed will be determined by the type of questions to be
addressed. The chromatographic step is obviously needed so that the components of
the complex medium can be sufﬁciently separated and mass analyzed.
Determinations of the presence of precursor label into various metabolite pools
requires the presence of high mass ions in the mass spectra of organic compounds so
that accurate isotope enrichment measurements may be made. Similar to procedures
incorporated for isotope-ratio mass spectrometry, the measurement of label
enrichment in these experiments is determined by comparison to reference values of
the natural isotopic abundance. Typically, then, a control experiment (free from
labelled substrate) is performed coincident to the metabolic proﬁling analysis and the
control values for natural isotopic abundance of speciﬁc metabolites are determined.
As a result, accurate measurements of the extent of label enrichment into these
metabolite pools can be made.

Electron ionization often falls short from meeting this criterion, especially in the
analysis of organic acids from urine, as was shown earlier. The appropriate choice of
ionization method can also lead to an increased sensitivity for analysis. Methods such

as ammonia-CI concentrate the majority of ion current ( > 95 % ) in useful ions for

171

organic acid analyses. Recall that the majority of the ion current from El analyses of
derivatized organic acids resides in fragment ions indicative of the derivatizing agent.
Accurate isotope measurements require selected ion monitoring (SIM) analyses on the
enriched metabolites for the the best ion statistics to be achieved. Total ion current
measurements obtained in the full-scanning mode of the mass spectrometer
comparing EI and NH3-CI have been shown to be nearly equivalent in the analysis of
synthetic mixtures of organic acids. As a result, analyses by SIM should lead to much
lower limits of detection by NH3-CI since the total ion current resides in one or two
mlz values, which are utilized in SIM analyses.

There are cases, however, when not only are the levels of isotope enrichment
desired but, also the position of the label in the metabolite under study. Answers to
these questions require ionization methods which generate a sufﬁcient number of
fragment ions. Ionization methods such as NH3-CI would be unsuited towards
answering these types of questions. On the otherhand, structural information is
readily attained using EI-MS and these structurally useful ions provide insight as to
location of the stable-isotope in the molecule. A suitable example describing these
behaviors are found in an early application paper written, and published in Anal.
Chem ..entitled," Utility and Limitations of Electron Ionization and Chemical
Ionization Mass Spectrometry for Determining the Postion and Extent of Labelling
in Permethylated Sugars." This paper may be found in Appendix I. In this paper,
ammonia-CI is found to be the preferred chemical ionization method for determining
the extent of label enrichment where as methane-CI and El can be used to determine
the position of label enrichment. For the class of compounds studied in this paper,
electron ionization would be a very poor choice if used exclusively for making
accurate determinations of label enrichment. The types of ions observed in the EI

mass spectra of these permethylated alditols are subject to exchange and

172

rearrangement reactions, and EI results are found to yield artiﬁcially low values of
enrichment due to the prevalence of these processes. These initial experiments
demonstrated the capabilities of ammonia-CI for providing abundant molecular ions
in the mass spectra of derivatized sugars, amino acids, organic acids, and steroids.
The potential utility of this ionization method in investigating metabolic pathways
associated with the class of disease conditions termed the organic acidurias was
evident.

Initial experiments using ammonia-Cl for measuring the flux of labelled
precursors into various organic acid metabolic pools stemmed from a collaborative
project with Sweeley and Martin (Department of Biochemistry, Michigan State
University), and Schall (Veterinary School of Medicine, Michigan State University)
investigating various aspects of Juvenile Onset (Type 1) diabetes. In conditions of
diabetes it is not uncommon to observe elevated levels of ketone bodies (e.g., B-
hydroxybutyrate, acetoacetate) in urine of the uncontrolled condition, since their
release from adipose tissue is a result of the need for mobilization of free fatty acids
to supply necessary energy (ATP) to the system. A major area of research in this
laboratory has been to evaluate organic acid metabolism of diabetic patients under
insulin-control therapy. A study by Shigematsu @139 revealed signiﬁcant
differences in the amounts, and types of organic acids excreted by diabetics under
insulin-control therapy and those of age-matched controls. It was found that diabetics
consistently excrete elevated levels of 2-hydroxybutyric acid, 4-deoxythreonic acid,
and others.

An investigation as to the origin of 2-hydroxybutyric and‘4-deoxythreonic acid
was made, since the levels of these metabolites in several diabetic patients were found

to be abnormally high. It had been postulated that these two organic acid metabolites

173

might arise from metabolism of L-threonine. An elegant method for determining
whether these species do indeed arise from threonine metabolism is to perform stable-
isotope tracer experiments incorporating uniformly labelled threonine, and monitoring .
for label enrichment into these metabolite pools. The results of this study, which
were published in Blamed. Environ. Mass Spectrometry, entitled, "Urinary
Metabolites of L-Threonine in Type I Diabetes Determined by Combined Gas
Chromatography/Chemical Ionization Mass Spectrometry” may be found in
Appendix 11. This study revealed that threonine is the precursor to both of these two
metabolites. In addition, it was determined that 4-deoxyerythronic acid also was a
metabolite of threonine. Evaluation of the isotopic enrichment using EI-MS was
extremely difﬁcult in this experiment since only very low levels of 13C-enrichment
were observed (c.a. 1-3 per cent, corrected for natural isotopic contribution). Only
when NH3-CI was employed were these levels of 13C-label enrichment observed and
accurately measured.

One of the most signiﬁcant studies presently being pursued involves determining
the feasibility of such stable-isotope tracer experiments for differentiating between
enzyme deﬁciencies associated with the lactic acidemias. These conditions are often
difﬁcult to differentiate based on (a) clinical observations, and (b) urinary organic
acid metabolic proﬁles, due to the similarities in the types and quantities of
metabolites excreted in urine. Recall, lactic acidemia is a condition that can result
from a number of partial enzyme deﬁciencies and lactic acid is the principal
metabolite excreted in urine for all of these conditions. For those situations in which
the patient is suffering from partial pyruvate dehydrogenase deﬁciency, for example,
a partial enzyme block has been found to occur in the conversion of pyruvic acid to
acetyl-CoA, which is the ﬁrst intermediate in the Krebs pathway. Deﬁciency in this
enzyme typically causes a large build up of pyruvic acid (pyr) in serum. In addition, a

174

concomitant increase in lactate serum levels is observed. For some disease
conditions, it is possible to measure Pyr/Lactate ratios from plasma samples to
determine the metabolic state of the individual. For this class of lactic acidemias,
however, the ratio of pyruvate to lactate in blood often is found in ”normal” or control
ranges, thus making diagnosis of this condition difﬁcult by conventional clinical
measurements. The following section summarizes my results on this feasibility study
for the differential diagnosis of the lactic acidemias. The results of this study were
presented at the International Symposium om Mass Spectrometry in the Health
Sciences, Barcelona, Spain. A manuscript is in preparation and is the focus of the

following chapter.

175

Chapter 7: Investigation of Cell Metabolism Using Labelled Glucose
Introduction

Studies of metabolic disorders collectively known as the organic acidemias have
revealed the molecular mechanisms of more than 50 heritable disorders and an

17 Differential diagnoses have been made,

explanation for accumulated metabolites.
generally, by detecting abnormal concentrations of speciﬁc organic acids in either
plasma or urine. The most widely used and accepted approach for identifying these
metabolic disorders has been to combine gas chromatography with mass spectrometry
(GC-MS).4’43

Lactic acidemia is a syndrome of several of the most frequently diagnosed
metabolic disorders. The identiﬁcation of these disorders by GC-MS has been
achieved by detecting excessive levels of lactic, and in some cases, 2-hydroxybutyric
acids in urine. 14’1” The lactic acidemias have been a particularly intriguing class
of genetic diseases due to the diverse enzyme defects that are responsible for the
excessive buildup of lactate in physiological ﬂuids. This condition has been reviewed
extensively by Matsumoto M15 who suggest that more than 15 enzyme
deﬁciencies may lead to congenital lactic acidosis.

Differential diagnosis of the lactic acidemias has been especially difﬁcult since
several enzyme defects may give rise to clinically indistinguishable urinary metabolic
proﬁles. Chalmers presented data which suggest that congenital lactic acidoses may
be differentiated upon evaluation of the respective urinary organic acid profiles.l4
However, conditions in which uninherited and/or other acquired causes of lactic

acidosis could occur were excluded from these studies; inclusion of these factors

which could lead to increased excretion of lactate would make differential diagnoses

176

more difﬁcult. Therefore, a more conclusive method for differentiating these enzyme
deﬁciencies, which produce nearly identical clinical states, is required.

Two enzyme defects associated with lactic acidemia are pyruvate dehydrogenase
and cytochrome oxidase deﬁciencies. Pyruvate dehydrogenase (PDH) deﬁciency has
been the subject of intensive study.141 In general, lactic acid accumulation is a
consequence of total pyruvate accumulation. In conditions in which there is a partial
PDH deﬁciency, the conversion of pyruvate to acetyl-CoA is blocked and a
concomitant increase in both lactate and pyruvate is observed. However, blood
lactate/pyruvate ratios remain nearly equivalent to those for age-matched controls,
making clinical recognition of the enzyme defect difﬁcult. Cytochrome oxidase
deﬁciency has been described in patients with a wide variety of symptoms, the
majority of cases being reported with neurodegenerative disease with a pathology
typical of Heights disease. Like pyruvate dehydrogenase deﬁciency, cytochrome
oxidase deﬁciency also is characterized by grossly elevated levels of lactate in urine.
As a consequence, differential diagnosis of these enzyme deﬁciencies has been
difﬁcult when based solely on the comparison of urinary organic acid proﬁles. A
general scheme for deﬁnitive diagnoses of the congenital lactic acidoses is shown in
Figure 64. The protocol for the identiﬁcation and differentiation of the lactic
acidemias is quite extensive. As mentioned, the presentation of the disease-state can
often be masked by what appears to be "normal" blood pyruvate/lactate
concentrations. Since differential diagnosis is difﬁcult to make based on the
concentrations of urinary organic acids, enzymological approaches are needed for the
deﬁnitive diagnosis.

Chalmers has suggested that in vivo experiments involving labelled substrates
may be well-suited for the differential diagnosis of the congenital lactic acidoses.

Both cytochrome oxidase and PDH deﬁciencies involve intermediates of glucose

177

 

Clinical and Biochemical Presentation
Mood Pyr:1.no,)ﬂ,¢1ucou,mtno tells

 

 

 

 

 

 

 

 

 

vim-armature» Gamma: ~ 4 !rclude'nou—molbol’ ‘
ﬂ organic climb

———+L Dmennwbmmtj :— a Bbmommmmm (__J
__ rheology as appropm

 

 

 

 

 

 

 

m Enzymoloo —— Cultlmd Immune
Cytxhromes

 

reproduced from J. label: ”ebb. 013. Suppl. 7 Vol. 1 1984
(with permission from author)

Figure 64. Scheme for the clinical and biochemical investigation of patients with
suspected congenital lactic acidoses

178

metabolism, with channeling of either pyruvate or an intermediate of the electron
transport system through the energy-generating pathways. Applications of GC—MS
for stable isotopically labelled tracer experiments include investigations of the
intermediates of the TCA cycle, and others. "2'1“

However, few studies involving labelled tracers have been performed for
distinguishing between enzyme defects. Most recently, Lapidot M146 used [U-
13C]-glucose for studying glucose-6-phosphatase deﬁciency in patients suffering
from glycogen storage disease (Type 1). The scope of these studies has been limited,
however, due perhaps, to the high cost associated with performing such an experiment
5111119.- This report describes the feasibility of stable isotope tracer experiments for
differentiating between several of the more commonly occurring enzyme deﬁciencies
associated with the lactic acidemias. The utility of [U-‘3c1-glucose metabolism in
cultured ﬁbroblasts for distinguishing between cytochrome oxidase and PDH
deﬁciencies is the focus of the investigation. Differences in l3C-tumover in various

metabolite pools are presented for the two enzyme defects; the results suggest that

these two deﬁciencies may be readily differentiated.
Experimental

Fibroblasts of one infant with PDH deﬁciency and one infant with cytochrome
oxidase deﬁciency and those of age-matched controls were isolated following

al.145 Fibroblasts were grown to a

procedures outlined by Robinson et
concentration of 105 cells / 25 cm2 petri dish in culture medium (Or-MEM). The
medium was then removed and replaced by a medium (a-MEM with no added

glucose) to which [U-13C]-glucose (concentration = 1.0mM) was added for a

179

"pulsing” period of three hours. Following the ”pulsing” period, medium containing
labelled glucose was removed and replaced with fresh medium containing unlabelled
glucose (1mM) and incubated for a ”chase" period of 12 to 15h.

Cells were extracted with Triton X—100 and collections of extracts were made
during the [U- 13C]-glucose pulse period (t-- 0.5, l, 2, 3h for PDH and control lines; t
= 3h for cytochrome oxidase line) and the "chase" period (t = 0, 3, 6, 9, 12, 15h for
cytochrome oxidase and control lines; t = 0, l, 3, 7, 15h for PDH line). In addition,
collections of cultured medium were made at times identical to those for the cell
extracts.

Cell extracts and cultured medium were lyophilized and cell metabolites were
solubilized in pyridine and extracted from excess detergent. Subsequent
derivatization of the components and cell medium was made using
bis(trimethylsilyl)triﬂuoroacetamide (BSTFA) and pyridine (4:1) and solutions were
heated at T = 80 ’C for t = 1h. Resultant derivatized organic acids and amino acids
were sealed in capillary tubes and kept refrigerated until the analyses were performed.

All analyses were performed using an HPS985 GC/MS/DS. A 25m DB-l
megabore column (0.52 mm i.d., purchased from 1&W Sci., Inc.) equipped with on-
column injection was temperature programmed from 55 - 75 'C at 5 'C/min then from
75 - 255 'C at 10 'C/min for all analyses. Mass spectra were acquired by electron
ionization (ED-MS (electron energy = 70eV, emission current = 300 mA, and ion
source temperature, T = 150 'C), and chemical ionization (CD-MS (ammonia pressure
= 300-400 mtorr, electron energy = llSeV, emission current = 325mA, and T = 150
'C). Selected ion monitoring (SIM) of MH+ and MNH“+ ions was performed for

determining the extent of label enrichment into various metabolite pools.

180

Results and Discussion

Shown in Figure 65 a-c are the reconstructed total ion current (TIC) proﬁles for
the metabolites found in control, cytochrome oxidase, and PDH deﬁcient cells, taken
at the third hour of l3C-glucose "pulsing" period. Evaluation of the metabolic
proﬁles of the cell extracts reveal only subtle differences. Although slight differences
in relative amounts of metabolites are observed, differential diagnosis based on the
respective organic acid patterns would be difﬁcult. Table 8 summarizes the identiﬁed
organic acids and amino acids from experimentally obtained EI and NHa-CI data, and
from available literature EI mass spectra.148 Glycine (tri-TMS) is relatively
abundant in the PDH ﬁbroblasts, but is relatively less abundant in the cytochrome
oxidase and control ﬁbroblasts; however, at later times throughout the "chase" period,
glycine abundances vary slightly. The signiﬁcance, though, of these results cannot be
evaluated until additional cell lines have been examined to determine whether this
behavior is characteristic and unique to the PDH deﬁciency.

Cultured medium of the respective cell lines were also analyzed by GC-MS.
Shown in Figure 66 a-c are the reconstructed TIC chromatograms of the metabolites
secreted into control, cytochrome oxidase, and PDH cultured medium. The amino
acid and organic acid proﬁles shown correspond to the onset of the "pulsing" period.
Striking similarities are observed between the amounts, and number of metabolites
secreted into the control and PDH medium. A slightly greater number of metabolites
are secreted into the cytochrome oxidase medium at early onset of the "pulsing"
period, but this result tends not to be a consistent one throughout the entire collection

periods. The metabolites identiﬁed in the cultured medium also are summarized in

181

IICOISIIUSYID 110 ¢NIOUA7008AI 0' GILL (XIIACI [1800-8

 

o ".10.
. 18 3‘o” rle
.. so
1
to."
j u
u 37..“ so

 

 

 

 

       
  

 

_:~ __m 200 800 800 000 1000 1100

 

RECONSIRUCYED 710 CHROHATOGRAI 0' (XIIACT [1000-J

 

 

 

 

 

 

 

 

 

30.30
1 ‘. 71¢
It :00
31
1 50
‘2 80
20.21
‘ ‘ ‘7
25
20
r , t3
4
‘5 :4 27
1“‘]Y“‘]"Y‘l"‘1l111‘ ‘j‘ l1“‘]“'1l"“ ‘I‘Illiiil‘
11 300 000 000 000 1000 1800

IECONSTRUCIIO 71¢ CHROMATOGRAI 0' EXTRACI .1831-0

 

 

 

 

 

 

 

 

36.38

i ‘ as
J
< 18.10

17 30.3. 3..‘o 81
‘ 20.21 3'

t3 /2s "“ u
1 3. l/ 50
‘UU‘III‘UIV"UIVIVUII‘I'l‘I‘VliI‘I]I‘VUI'U"l""l"“l""l"‘
U 200 000 000 000 1000 1200

Figure 65. GC—EI-MS TIC proﬁles of metabolites found in (a) control, (b)
cytochrome oxidase deﬁcient, and (c) pyruvate dehydrogenase
deﬁcient ﬁbroblast cells

18 2
Table 8. Identiﬁed metabolites of both cell ﬁbroblasts and cell medium Peak

number 43-glutamic acid. Peak number 47-unknown (MWﬂ37),
believed to be a ﬁve-carbon aminoribose sugar

Beaummaormnmarhllr‘l WWI). murmur

I Acetic Acid 133 28 Methyladipic 221
2 Unknown #1 122 29 Unknown #6 278
3 Acetylglycine 190 30 Homoserine 336
4 Lactic acid 235 31 Aspartic acid 350
5 Glycolic 221 32 Unknown #7 306
6 Indole 190 33 Pipecolic acid 202
7 Alanine 234 34 Pyroglutamic 274
8 Glycine (di-TMS) 220 35 Unknown #8 350
9 Propionyl glycine 204 36 Glutamic (di-TMS) 292
10 Unknown #2 218 37 Malonylglycine 320
l l Propionyl glycine 204 38 2-aminoadipic 378
12 3-indoleacrylic 188 39 Threonic acid 425
13 Valine 262 40 N-valerylglycine 304
14 Unknown #3 336 41 2-hydroxycaproic 277
15 Urea 205 42 N—acetylaspartic 392
16 Serine (di-TMS) 250 43 Glutamic (tri-TMS) 364
17 Lcucine 276 44 Phenylalanine 310
18 Phosphoric acid 315 45 Mannitol-tetraacetate 380
19 3-(4-phenyl)propenoic 309 46 Oxaloacetate 349
20 Isoleucine 276 47 AMINOPENTOSE 438
21 Proline 260 48 Unknown #9 452
22 Unknown #4 264 49 Lysine 363
23 Glycine (hi-TMS) 292 50 B—Galactofuranoside 540
24 Glyceric acid 323 51 a—D—glucose 540
25 Serine (tri-TMS) 322 52 Tyrosine 398
26 Threonine 336 53 8-D-glucose 540

27 Unknown #5 278 54 Inositol (hexa-TMS) 613

183

88001487800788 71¢ 01180114700848 0' "8811’“ [1888-1

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

( 88.88 718
I
‘ 8
«l
18.18 so. ,“ u
t n so." 3. 3' 4.1.“
‘ \ "as , u ‘1
+ 1’ '3. | ..l
Uth “l
‘I‘Ul ‘‘‘‘‘‘‘‘ I U‘UUIIUI‘IIU"
11 888 888 888 888 1888 1 188
necoustnuCtto tlc canonarocnau or cert uroluu [1808-8
‘ 34.5: tle
. as
, 18.18 3°-
81
< " so
‘ 1 21.28 ” so 81
l i 2527. 3 3.“
7 11,12 .
- IUW[?.11.42
s
'I 9 13 24" “i italia. ‘l
0 I1'88 I 1188
ntcousrnuctto tlc eunouatocnau or utotuu [teat-t
' 30.38 '.c
d
’. 88
s
30 81
40
2.4:
44,.
"I
l
888 I 1188 1888

 

 

 

Figure 66. GC—EI—MS TIC proﬁles of metabolites excreted into (a) control, (b)
cytochrome oxidase (CytOx) deﬁcient, and (c) pyruvate
dehydrogenase (PDH) deﬁcient cell medium

184

Table 8.

The principal organic acid of ﬁbroblasts and in the cultured medium has been
identiﬁed as lactic acid. Other low molecular weight organic acids are observed, but
in only very low abundance. The majority of metabolites observed are amino acids
and glycine conjugates. Pyroglutamic acid (peak #34), a major cell and medium
metabolite is believed not to be an artifact of cyclization of glutamic acid, which
frequently results during gas chromatographic analysis. 149

The results from the reconstructed TIC chromatograms suggest that differential
diagnosis of these enzyme defects by measurement of pool sizes is difﬁcult, and only
very slight differences in the metabolic proﬁles are observed when comparing
patients to age-matched controls. A more useful approach for differentiating these
enzyme deﬁciencies from one another is based on monitoring individual metabolite
pools for 13C label enrichment and/or turnover rates after uptake of [U- 13C]-glucose.

The pertinent pathways involving glucose consumption are shown in Scheme 5a
and 5b. [ll-13C]-Glucose can enter the glycolytic pathway, thereby generating 13C-
enrichment into glycolytic intermediates, transamination products, etc. (e.g., lactate,
alanine, serine). Since only partial PDH and cytochrome oxidase deﬁciencies are
typically observed, a portion of glucose continues to be channeled through the TCA
cycle (scheme 4b). Metabolites such as glutamic acid, for example, should experience
two 13C atoms from acetyl-CoA. Shown in Figure 67 is the amount of labelled
glucose (enriched at all six carbon atoms) observed at the onset of the pulsing period.
Only 80 per cent of fully labelled glucose is observed, suggesting that more than one
source of glucose (i.e., from gluconeogenesis and breakdown of glucose-containing
carbohydrates) is available to the cells. Enrichment of l3C-label into various

metabolites may be diluted by the presence of these unlabelled sources of glucose.

185

Glucose [ 13c 6]

l

3-phosphoglycerau [13031 ——> Serine [130,]

l

Lactic Acid (“cal e— Pymvae (“cal ——->- Alanine [nos]

Amino Acid: 1 Fatty acids

\ Acetyl-CoA [ﬂea] /

of) '
Oxaloacetatt cmk

/ cis—Aconitate
Melee \

r Isocitrae
FT” 2-ltetoglutamt
8mm" smmyl'c“ alumnae (”c l
2
Scheme 5a and 5b

Summarized in Table 9 are those cell metabolites (and metabolites from cultured

medium) that experience l3C—enrichment (+) and those for which no enrichment (o)

186

be; .8 as a 8:88 .88..er 893. 8:33.88
.«o .88 06 8 3:130 Agtsgna 0333 .«o 8.689 «38 Uta—AZ "he ”Suﬁ

 

 

 

 

N\Z
smut - 8m - admit -amml .. awm - -ﬂtﬁ- - -37 - awe awe
t—— —t.: «4 1.. in; —l-_—L tart» t JAG
..GN
. a
l h
J g
4
l .3
J v
a r
ma
3d

 

 

 

 

ca-«av-«5u-cnacoacuu

187

.23 .888. one... 3 5.5283... .8 22.8 .88.. .33
089.35.... 823.32. a on 8 92.5533 .83 3.. .5655 25. n
83 3.0.2.3.... .
8.85.2909 o... 8.3 £323.83... .1.

+ + + awn \ 3v. 0805.0

e e a 8.. a 5. «snout—5 z:

+ e , + 3n \ 3... 33:20.
I. + + «an \ mun atom

+ + .. nan \ can breeze

+ + e «an \ nan Osman—0

+ + + van \ ban 05:3

+ + + mnN \ an" 30¢ 03.91—

3333 g

as. 328 E o5 ensue—ease egg 3 .368 88.8.0 3 8..
Saba 8258 Egegaooa £33585» 95 8532855232 no 33

 

188

is observed. Compounds such as lactic acid would be expected to incorporate three
13C atoms from fully labelled glucose. Shown in Figure 68 is the NHB-CI mass
spectrum of lactic acid. Approximately 40 per cent enrichment is observed at the
(A+3)+ peak for the MH+ and min; ions. These results suggest that 13c in lactate
is derived from the labelled precursor ( [U-13C1-glucose ).

Table 9 reveals, however, that 13C enrichment is observed for lactic acid in all
three cell lines. A plot of 13C-enrichment vs. collection time for lactic acid is shown
in Figure 69. The maximum in 13C-enrichment occurs at t = 3 hours during the
"pulse" period for the PDH and control cell lines. Collections were not made for the
cytochrome oxidase line at times earlier than t = 3h during the "pulsing" period, yet
the level of l3C-enlichment observed for this line suggests there is very little
difference in the extent of label enrichment and turnover times for the three cell lines.
Therefore, lactic acid cannot be considered useful as a target compound for
differentiating these enzyme defects. Similarly, alanine and glycine demonstrate
similar behaviors, with nearly equivalent levels of enrichment when comparing the
three cell lines.

The pattern of 13C-enrichment of these metabolites is. no more revealing than that
obtained when the medium is analyzed by GC—MS. In fact, only alanine and lactic
acid (in addition to glucose) show any enrichment at all. For lactate secreted into the
cell medium, slightly higher levels of enrichment are observed in the PDH line (50 %
+/- 3 %) relative to the control and cytochrome oxidase lines (c.a. 44%). Maximum
label enrichment occurs at t = 3h during the "pulsing" period for all three cell lines.
In the case of alanine, higher levels of enrichment are observed for all three cell lines,
with 13C-enrichment at the (A+3)+ peak exceeding 16 % +/. 2 % for the cytochrome

oxidase line and 9% for the control and PDH lines, the former showing nearly a two-

189

detox. 2.3.5.2.... o... 2.5 Rosco
goose .938 89a 65...... so. one: as 8.58... 3.8 Benz "8 2.5...

 

 

 

"\E
ouu on« can own our 05.. on" our 0: on
_nbhb_ nth:pP—blbihtb_hnDb—bb-bhEbt—Wpch 1
a . . j t. as ._
r
e
+75%... r
r
. .
ocio:cu\! .
«as:
e to...»

 

 

 

 

Amzpl.ov o.oo o.ooo. .o Eoeuoooo ..oe _otnxz

190

gdagauuoooﬂeaoao . ..as 53......
.gsgaaon.zax «can

 

 

 

   
 

«830508....
33"” O—Oo
a “f URL ...O—e: U—oonoﬂru
.lvp up or m i.
....m w m will... .. o
clelel ”'1’
row V
..T
tam
V.
x .2 m
3.2.6 .SOOIIIIO . tilt
69.2300 can cilia. W
.23000'6 (PM
w.
I
z .8
,e

 

_Oon 02.3305. 20¢ 0:03

l9!

fold increase over the levels obtained from the cell extract.

The results shown in Table 9 suggest that uC-enrichment is observed in the
metabolite pool of glutamic acid for the control and cytochrome oxidase lines, but is
not observed for the PDH line. Enrichment is found at the (A+2)+ peak for the MH+
ion, as expected. The NH3-CI mass spectra of glutamate in the cytochrome oxidase
and PDH lines, respectively, are compared in Figure 70 a-b. Clearly, enrichment is
observed for the cytochrome oxidase line but not for the PDH line when comparing
the contribution of 13C-enrichment at the (A+2)+ peaks for the two species. A plot of
13C-enrichment vs. collection time for the glutamic acid metabolite pool is shown in
Figure 71 to further demonstrate this behavior. At t = 3h during the ”pulse" period,
only a one per cent enrichment (corrected for natural isotopic contribution) is seen in
the PDH glutamic acid pool. At the same collection time, though, much higher levels
of l3C-enrichment are observed in the control and cytochrome oxidase lines.

Results from Table 9 suggest that an unknown (derivatized MW =437) found in
the cell extracts may also be used to differentiate these two enzyme defects. This cell
and medium metabolite has been tentatively identified as a S—carbon aminopentose
sugar based on mass spectral interpretations. A literature mass spectrum of the TMS
derivative of this compound has not been available. A structural assignment has been
proposed, therefore, on the basis of experimentally obtained NH3-CI and EI mass
spectra for the TMS [Si(CH3)3] and d9-TMS [Si(CD3)3] derivatives. Listed in Table
10 are the observed ions from both ammonia-CI and El analyses on the TMS and d9-
TMS derivatives of this compound. From the d9-TMS experiment, it is possible to
determine the number of TMS groups associated with each fragment ion, as shown in

the third column of Table 10. Furthermore, these experiments have revealed that the

RELATIVE ABUNDANCE

RELATIVE ABUNDANCE

I92

 

 

 

 

 

 

 

 

18

[MB - (cup-es: (c113), l’ L
an L
MI!+ :
at ,
"A92“ Contribution .
4, i
I
at L
a t

290 250 ﬂl ' 350
H12

_ 9
[MR (enzysr (CH3)2 1 g
i L
‘ “Aer Contribution T
- + L
p / \Mn ,
he-e-leer--- A. mu Lb-..e!-..-L.----JLI

 

Figure 70. NHs-CI mass spectrum of glutamic acid from (a) cytochrome oxidase,
and (b)pyruvate dehydrogenase ﬁbroblasts

 

I93

38 .33.... man. as .55

49580 3 Bo- gus...» 5m 08.: .3838 as 308550.02 .«o .03 "2. 95am

.23... as.»

 

. .. 32.0.. 0.6.8N L sense: 0.0.8..” P.
eruwopme.v~onaeo

\\

 

\\ u

$elcrqu r-lﬁ Idl-

< — —
e'o'lp‘n I. I;

.liiatiy.

    

032.6 .So Orlltn
69.2800 can. <l.li_<
.9300 To

 

_oom ouzonauos. Eo< 28920

N

to v
oor x ([(z+v)l+(v)Il/(z+v)l)

I94

...p

c.a.nxo..m.on_m.

«1.5..

e$139.30... . 8» :5.

+3355 « . a.
eBro-«.5 . 8n 5..

c.a.noa..m:z:oo.a.
+.ma..o«:o.a.
a
e. 5.2.
3:
gal—a.—

e... «n . .3...
+2.3...

5. e....“cozméuemm.«an...
spa ....
o2 +.n.noo._wo: . 5. as.
«2 L355 a. . 3.
are. eSec-«.5 . S» as.
can ....
o3 c.a.noa..mzz:oo.a.
can c.a..»ouroé.
«3 Locos.
8. .ea
«anon 3323.42
ﬁg
3» e... 8 . .3...
on... . L52.
“anon Engaging
“axe—4%

A5332. 536...: ..o «9335.. m2...
.ov v5 matmu 2.. .«o 8.53% «was cougars. 8&8. 5.30.0 o... 88.. 8252..

..03 5.? 684.083 3.5..» $535083 .0 89:5: 2:. @2530 2.8. Scamp—m

«or
«an

QQN
..mu

moo
bra
NV»

won

mac
«5*

man
#51

.0. 03a...

I95

derivatized molecular weight of the unknown is 437u. In addition, the parent
structure is found to consist of four TMS groups, as the molecular weight of the
deuterated analog was found to be 473 u, suggesting the replacement of four Si(CH3)3
groups with four Si(CD 3)3 groups. Schoots and coworkers recently reported on the
trimethylsilyl derivatives of carbohydrates and organic acids retained in uremic
serum,lso however, no amine-containing carbohydrate (MWﬂ37) was observed. A
possible organic acid analog of this unknown, 2—deoxyerythropentonic acid (tetra-
TMS, MW = 438), was reported, however, and chromatographic elution times
between this compound and the unknown were similar, lending support to the
unknown being an aminopentose sugar derivative.

Figure 72 gives two proposed structures for this compound and the origins of the
fragment ions. Structure 1, ribosylamine, must be considered since a number of
fragment ions can be rationalized from common ring-cleavages. This structure is
consistent with known five-carbon ring structures, yet no literature mass spectrum for
this compound has been available. However, a number of fragment ions are difﬁcult
to explain fully from analysis of this ring structure. An important ion observed in the
EI mass spectrum is found at mlz 27 8, and the d9-TMS experiment has revealed that
the fragment contains three timethylsilyl groups. The only plausible explanation for
such an ion is a result of migration of a TMS group in its formation. Migration of
trimethylsilyl groups in carbohydrate molecules have been reported previously.151
This ion might arise through elimination of ketene (CH2=C=O) from the [M-ll7]+
ion, as shown in the proposed mechanism of Figure 72 (structure 2). Structure one is
not readily able to undergo such an elimination reaction, whereas structure 11 appears
to be capable of such a rearrangement due to the presence of the unhindered 04
position. GC—MS-MS analyses are currently being evaluated to verify this precursor-
daughter relationship.

196

Proposed structures and fragment
ions for Unknown (MW=437)

Structure 1:

 

 

 

 

 

 

 

 

 

 

m/z320
TMSO-9H2 0
m 334/ m/2233¥ NHTMS
\ mlz 117
_1{‘
OTMS OTMS
Structure 2:
WS-O
TMSO-CH I
2 .5
334 (EH ‘0 §T
-—-——> 1 0
TMS‘O TMS-O c CH2
R
230 TMS-0
mac 1
l
('3Hz— cH + CH2: c=o
TMS'-O—TMS
+
mlz 278

 

 

 

Figure 72. Proposed structure and fragment ions for unknown (MW =437)
presumed to be an aminoribose sugar metabolite

197

Regardless of the identity of this compound at present, it is of special interest
because, unlike the case for glutamic acid, l3C-enrichment is observed in the PDH
and control lines, but not in the cytochrome oxidase line. Enrichment at the (A+3)+
peak for the MH+ species is observed (c.a. 15 % above the natural isotopic
contribution for similar compounds ciontaining four TMS groups). A plot of 13C-
enrichment vs. collection time for this compound shown in Figure 73 suggests that
these two enzyme defects may be differentiated from measurements of 13C-flux into
this metabolite pool.

Of interest, too, are the metabolites serine and glyceric acid. Glyceric acid is an
intermediate in the interconversion of serine and 2—phosphoglyceric acid. The levels
of 13C-enrichment incorporated into serine are nearly equivalent for the two enzyme
deﬁcient lines (c.a. 8%), though a two-fold increase in label incorporation is observed
for the control line. The higher levels of label observed in the control line for the
serine metabolite pool is based on the fact that in the two deﬁcient lines, there is more
heavy reliance on the glycolysis pathway to provide ATP. Because of this, the
concentration of 2-phosphoglyceric acid is lower in the deficient lines and as a result,
is not available for shunting into the serine metabolite pool. For all three cell lines,
maximum label enrichment is observed well into the "pulse" labelling period ( t = 3h).
The results on glyceric acid indicate that this metabolite, like glutamic acid and the
aminopentose sugar metabolite, may be used to differentiate the enzyme defects.
Interestingly, the 13C-turnover rates into the glyceric acid pool differ significantly
from the serine metabolite pool. Unlike serine, glyceric acid experiences maximum
label enrichment at the onset of "pulse" labelling (collection times, t = 0.5h and 1h).
After two hours, nearly all of the 13C-label has been consumed by the cell.

Explanations for this behavior are under investigation. Nearly ﬁve % enrichment is

198

«=8 .3330 :9. 3. .55 422.8

a. E13§§8€98380808=8§§5§02388 "ﬂaw...

A9505 05:

 

 

 

 

 

    

:32? 298$. ..o._:a..o_o..00 F.
q } A 1 > a
.. o.m
? \rl
J \o I.
\.\.\
cl 1 I l. v .. 0.3
0002.5 4.6 Olllo 1
69.9800 can allid
L ed p
.2200 OIIIO
no.8

 

acoEcotcm 03300. .2302 .2 0302.325.
.00.“. 3:09.302 0582.608”.

(con x ([(e+v)t+(v)Il/(e+v)t) .

199

observed for the (ms)+ peaks of the MH+ and MNH: ions in both the control and
PDH lines. However, results for the cytochrome oxidase line are unavailable at this
time since collections were only made at times, t = 3b dtn'ing the ”pulse" period.
Thus, determining whether this rapid turnover of glyceric acid occurs in the

cytochrome oxidase deﬁcient line is not presently possible.
Conclusions

This feasibility study suggests that turnover rates are superior to pool size
assessment by conventional GC-MS proﬁling for distinguishing between some of the
lactic acidemias. For some metabolites, differences in 13C-tumover rates and levels
of enrichment are only slight. However, we have shown that some important
metabolites may be used as target compounds for such feasibility studies. The results
for glutamic acid are not surprising. Partial PDH deﬁciency results in diminished
channeling of glucose through the TCA cycle. Since induction of glycolysis was the
principal pathway through which glucose was metabolized, this PDH deficiency
results in the buildup of the intermediates pyruvate, lactate, and amino acids, such as
alanine in ﬁbroblasts. Since the enzyme block occurs in the conversion of pyruvate to
acetyl-CoA, the amount of labelled glucose available to enter the TCA cycle will be
severely diminished, therefore resulting in low levels of label enrichment in
metabolites such as glutamic acid.

Results on the tentatively assigned amino pentose sugar metabolite are not nearly
as easily interpreted and explanations for the observed 13C-enrichment can not be
made until unequivocable identiﬁcation of the unknown is made. This five-carbon

sugar shows label enrichment into three carbon atoms, for both the control and

200

cytochrome oxidase lines, but l3C-enrichment is absent in the pyruvate
dehydrogenase line. The results suggest that this five carbon carbohydrate may be
derived exlusively from a 3-carbon metabolite of glucose.

The differential diagnosis of cytochrome oxidase and PDH deﬁciencies is
therefore possible based on the ﬂux of 13C from glucose through these speciﬁc
metabolites. The explanation for the difference in the ﬁve carbon carbohydrate
enrichment may lie in the different redox state of the cells. In cytochrome oxidase
deﬁciency there is a more reduced NADH/NAD+ state and also probably a more
reduce NADPH/NADP+ state. This would restrict entry of glucose-6-phosphate into
the hexose monophosphate shunt pathway in the cytochrome oxidase deﬁciency but
not in the control or PDH deﬁciency.

The goals of this feasibility test were to determine whether enzyme deﬁciencies
could be differentiated based on the use of labelled substrates for m studies.

Antoshechkin “L152

in a similar investigation, concluded that metabolites
excreted by ﬁbroblasts may be reﬂected by the activity of various metabolic processes
i_n_yit_m. In hereditary organic acidurias, those pathological metabolites arising from
enzyme defects are typically excreted by cells from many tissues, including
ﬁbroblasts.

As a feasibility study, there appears to befthe potential for extending this method
to the differential diagnosis of other heritable enzyme deﬁciencies associated with
congenital lactic acidosis. Experimental parameters such as the amount of labelled
precursor, length of ”pulse" time, and periodicity of collection time will be evaluated
to optimize the conditions for future experiments. Theoretically, experiments of this
kind may be employed using other fully-labelled precursors. However, the

availability and cost of other labelled precursors may preclude their routine use. The

low cost associated with the use of uniformly-labelled glucose (nearly 100 per cent

201

enriched) makes these mm studies attractive for future investigations. The
preliminary results obtained from m analyses of cultured cells suggest that
stable-isotopes might also be used mm to differentiate between various enzyme
defects that are clinically indistinguishable.

As mentioned, the future of this study is to extend this stable-isotope approach to
the investigation of several enzyme disorders, in addition to those associated with the
lactic acidemias. The goal is to determine whether metabolites, either from cell
extracts or those excreted into the cell medium may be associated with individual
metabolic conditions. l3C-label enrichment into the metabolite pools such as lactate,
alanine, serine, and others are not useful for differential diagnoses since they are not
unique to one particular condition. The goal, then is to determine whether specific
metabolites may be targeted for in selected ion monitoring experiments in order to
determine whether label enrichment is experienced in their metabolite pools. From
these results on a wide number of enzyme disorders it may be possible to correlate
speciﬁc metabolites with a particular enzyme disorder.

The long-term goal of this feasibility test, as stated, is to apply these tracers for in
m analysis. To this point, the cost associated with purchasing uniformly labelled
glucose far exceeds the economic feasibility of such experiments. For the m
studies, only 10-20 mg of uniformly labelled glucose were required per cell line (a
releatively low cost). However, to attain the same levels of enrichment into various
metabolite pools from £11119 analysis would require in excess of lg [U- 13C]-glucose
to be ingested by the subject. The cost per experiment would thus not be practical for
the differentiation of these organic acidurias. Differential diagnosis could be attained
through approaches utilizing molecular biology at a much reduced cost, but at the
expense of time and training of qualiﬁed personnel.

Lapidot and coworkers have done some elegant work in the area of m

202

analysis using labelled substrates. In their most recent work, they developed an in
35119 assay for various metabolites using [U - l3CJ-glucose on patients suffering from
Glycogen Storage Disease (GSD). Not only were they interested in the flux of carbon
into various metabolite pools, such as glutamic acid, but were also interested in the
carbon recycling nature of these metabolites. They showed that for glutamic acid,
only speciﬁc position in this ﬁve-carbon amino acid metabolite are directly involved
in glucose metabolism. In order to determine the position of labelling in these
metabolites, it was necessary to use exclusively EI ionization for mass spectrometric
analysis.

The initial experiments conducted in our laboratory for the differential diagnosis
of lactic acidemias were geared solely to targeting those metabolite pools which did
or djinm experience 13C-enrichment. A great wealth of information may become
apparent, though, if we are not only to look at the extent of label incorporation, but as
well, the posiition of label incorporation into these various metabolite pools. These
kinds of results may uncover various subtleties in the operative metabolic pathways of

these enzyme deﬁcient conditions.

203

Chapter 8: ECNI-MS w/ CCl 4 "pretreatment"
Introduction

Negative chemical ionization (NCI) is a recent addition to the plethora of

ionization methods that now exist in the area of mass spectrometry. This ionization
method is analagous to positive chemical ionization for gas phase ion molecule
reactions. Negative reagent ions are produced upon bombardment of reagent gas
contained at high pressure in the ion source by high energy electrons. Ionization to
form negative ion reagent ions typically proceeds through dissociative electron

attachment reactions, as shown in eq. 18 below:

e" + MX -------------- > M' + X' (18)
Stable reagent ions for negative ions reactions include the superoxide, 02', ion,153
methanolate (OCH3') ion,154 and hydroxyl ion.lss'156 In addition, the chloride

157 Under conditions of negative chemical

ion has found use as a reagent ion.
ionization, one of two reactions dominate, those being (a) association reactions and
. (b) proton transfer reactions. For the negative ion molecule reaction to proceed

exothennically, as shown in eq. 19 below,

+ Y ------------- > XH + ( Y-H )- (19)

the gas-phase acidity of YH must be greater than the gas-phase acidity of XH. This is
analagous to what was described earlier for positive ion-molecule reactions.

Attachment reactions in the negative ion mode, on the otherhand, are analogous to

204

those reactions most commonly observed under positive ammonia chemical
ionization conditions (i.e., the formation of the [M+reagent ion]+ adduct. Negative
ion adduct ion formation requires a third-body for stabilization. Most association
reactions observed in NCI-MS have been found to involve the chloride anion.
Reactions of Cl- with polychlorinated molecules typically result in [M+ Cl'] adduct
ion formationlss'159 Under conditions such as those described above, negative ion
detection offers no dramatic enhancement in the selectivity or sensitivity that may
already be achieved using positive ion detection.

The real strengths for negative ion mass spectrometry are realized when the mass
spectrometer is operated under conditions for electron capture negative ionization
(ECNI). A high concentration of near thermal energy electrons are generated under
high pressure conditions into the source of the mass spectrometer. These near-
thermal electrons are produced when methane, for example, is introduced (c.a. l torr)
to the mass spectrometer ion source and subjected to electron ionization. Actually,
these low energy electrons may be produced regardless of whether methane, or some
other reagent gas (which does not particpate in subsequent ion-molecule reactions) is
used.

Methane, as mentioned, has been chosen most frequently as this necessary high
pressure stabilizing gas for electron capture negative ion mass spectrometry. Recall,
as these near thermal energy electrons are produced, so to are a high flux of positve
ions (see eqn.'s 5-8). Recently, Hunt showed the strengths of a method called pulsed
positive ion —negative ion (PPINI)-MS. 160 This technique has shown enormous
potential and is available on many commercial instruments today. The instrument is
capable of operating in both the positive and negative ionization modes. In some

instances, ammonia has been used as the reagent gas to provide a greater amount of

205

selectivity in positive ion analysis.

The bombardment of methane (or other non-reactive gas, such as N 2, N113) by a
high energy electron generates an excited methane molecule. The energy of the
primary electron is reduced approximately 40 per cent with each ionization event?9 .
Secondary ions of much lower energy are formed as a consequence of this ionization
event. The "slowed" primary electrons are typically of sufficient energy to participate
in a second ionization event. These electrons, and the secondary electrons generated
are further stabilized through subsequent collisions with reagent gas until a high flux
of near-thermal electrons is produced.

The analytical utility of such a situation is that reactions involving these near-
thermal electrons and analyte molecules proceed at relatively high rates. When the
analyte molecule exhibits a suffﬁciently positive electron afﬁnity and high cross-
section for electron capture very low detection limits can be observed. It is the
electrophilicity of the analyte molecule that determines to what extent electron

capture ionization occurs. The processes which the initailly formed anion competes

with is autodetachment, as shown in eq. 20, below.
MX + e' ----------->[MX"'] ------.-—> MX + e' (20)

The molecular anions formed by electron attachment will undergo autodetachment
unless stabilize through collisional deactivation. For signiﬁcant collisional
stabilization at C1 operating pressures, the lifetime of the' [MX'*] species'should be
greater than 1 u see. Those species with positive electron afﬁnites tyically exhibit
sufﬁciently long lifetimes to be collisionally stabilized to the molecular anio 78 .

The sensitivity of this method lies in the fact that rate constants for electron

capture can be approximately 400 times larger than corresponding ion-molecule

206

reactions (in both the positive and negative ionization modes). These rate constants
vary to a much greater extent than the positive ion counterparts, though. This is due
to the fact that electron capture cross-sections are strongly structure-dependent. As a
consequence, ECNI—MS is often found to be a very selective and sensitive technique.
In fact, detection limits as low as 25 femtograms (fg) have been reported161 under
conditions of ECNI-MS. This is the case since ECNI mass spectra typically contain
very little fragment ion information and is a result of the low-energy process

associated with electron capture ionization.
A novel ionization method for EC-NI mass spectrometry

The manuscript on threonine metabolism in conditions of Type 1 diabetes
discusses the use of ECNI for the analysis of plasma amino acids. It was of interest to
determine the amount of uniformly labelled threonine that was actually present in the
plasma of the animal under study. This information was of importance because the
rate of infusion of the labelled precursor did not occur over a "pulsed" period, as
desired, but was administered as a bolus injection, at the onset of the experiment. The
result was lower levels of 13C—enrichment into various organic acid metabolite pools
than originally anticipated and there was a desire to correlate these levels of label
enrichment with the the amount of labelled precursor present in plasma.

Procedures outlined by Dufﬁeld and coworkers160 for the isolation and
derivatization of amino acids from plasma were employed. Literature mass spectra of
these types of amino acid derivatives were typically devoid of molecular weight
information. ECNI-MS appeared to be a reasonable alternative to El since these

derivativized amino acids contained highly electrophilic moieties.

207

Experimental

Initial analyses on the isolated and derivatized amino acids by mass spectrometry
were uninformative since only very small ion signals were achieved for the
components of the mixture. Gas chromatographic analysis using electron capture
detection (ECD) revealed that the amino acids were present, and of relatively high
abundance in the mixture. The poor sensitivity of the mass spectrometer was
determined to be a result of poor tuning and calibration of the instrument. Typically,
perﬂuorotributylamine (PFTBA) is used as the calibrant in both the positive and
negative ionization modes. Under negative ion conditions, though, PFI‘BA yields
only high mass ions (e.g., mlz = 633, 595, 452) under electron capture conditions.
The ECNI mass spectrum of PFI'BA is devoid of intense ions below mlz =414.

Tuning by PFTBA on these high mass ions yielded what appeared to be
"sensitive" tuning ﬁles. However, subsequent analyses of the derivatized amino
acids showed very poor sensitivities. The possibility that the quadrupole mass
spectrometer was being tuned for preferential high mass ion transmission was
investigated. In order to test this hypothesis, a calibrant mixture containing
equimolar PFI‘BA and CC14, the latter being used to ensure the production of low
molecular weight ions (i.e., Cl'), was utilized. Initial results indeed suggested
preferential transmission of high mass ions through the quadrupoles and subsequent
tuning adjustments allowed for generation of not only PFI'BA ions but also Cl',
002', and C03" ions. Following these tuning "adjustments", the amino acid
mixture was once again analyzed by ECNI mass spectrometry and this time, the
compounds were easily detected. However, close inspection of the ECNI mass
spectra of these amino acids revealed some very unique ions.

This chapter describes a novel ionization method based on carbon tetrachloride

208

ion source pretreatment for electron capture negative ion (ECNI) mass spectrometry.
The utility of this method resides in improved detection capabilities for selected
classes of compounds (e.g., amino acids) over conventional electron capture
detection while providing molecular weight information which is generated
independent of useful structural information available on the compounds by
convnetional ECNI-MS. Adduct ions of the form [M+ Cl']'are formed from reactions
of analyte molecules on ion source surfaces that have been "pre-conditioned" with
chlorine-containing compounds such as methylene chloride, chloroform, or carbon
tetrachloride. An [M+ Cl'] radical intermediate appears to exhibit an enormous cross-
section for electron capture which facilitates in abundant ion formation. Analyte
molecules not participating in these surface reactions undergo conventional electron
capture ionization, yielding the types of ions observed under conditions in which the
mass spectrometer is not pretreated. A manuscript entitled "Utility of Ion Source

Pretreatment with Chlorine-Containing Compounds for Enhanced Performance in

Gas Chromatography/Negative Ionization Mass Spectrometry " has recently been
accepted for publication in Analytical Chemistry . The manuscript, in its entirety

may be found in Appendix III.

209

Results and discussion on simple mixtures

Compounds not exhibiting high cross-sections for electron capture under
conventional ECNI conditions may be (as described in this paper) capable of forming
a radical intermediate that readily undergoes electron attachment to form the stable
molecular adduct ion, of the type [M+Cl]'. This technique may find use in that
compounds previously considered unlikely to yield stable molecular anions or
pseudomolecular anions by electron capture, can, following CCl4 pretreatment of the
ion source, produce intense negative ion mass spectra. The results of such an analysis
are shown in Figure 74. The reconstructed TIC chromatogram of an equirnolar
mixture of TMS derivatives of organic acids obtained under electron ionization
(positive ion mode) conditions are shown in Figure 74a. Shown in Figure 74b is the
convenetional ECNI mass spectrum of this same mixture. Not surprising is that the
majority of derivatized organic acids do not form detectable anions since the
trimethylsilyl groups imparts no additional electrophilicity to organic acids upon
derivatization.. Surprisingly, maleic acid exhibits a relatively high cross-section for
electron attachment, due to its highly conjugated system. When the ion source of the
mass spectrometer was pretreated with CCl4 and the organic acid mixture reanalyzed
by ECNI, the reconstructed TIC chromatogram in Figure 74c is observed. In this
case, several of the organic acids now exhibited sufﬁciently high cross-sections for
electron capture that molecular anions are observed. In fact, the TIC ”counts” for
many of these compounds in the negative ion mode equal those obtained using El.
These results are an excellent example of the strengths of this pretreatment method.
Compounds not previously considered amenable to negative ion analysis might now

be detected under the appropriate conditions.

.. 210

__m 383*_
100 X I- 10300

 

 

/ meets-ma

 

 

 

 

 

 

 

 

 

 

'Uf‘IYIj"011"‘l"‘7l"'rl‘I'1II‘V‘IV‘YYI'T
m Ec-u
o
. 100%.28600 I C C '
‘ ms-o’ ROM
micron-ma

 

 

 

 

1‘Y'ilﬁ‘VIIW‘IVII—I'V'T'T‘I'

(c) EC-Nt v! CCl‘tunln.

 

‘ 100% - 45200

 

 

 

 

 

t'mr “riff"r"“T"**rﬁﬁrf'"r
so rs. as. as. «so sso es. 150

SCAN NWBER

Figure 74. TIC proﬁle of the trimethylsilyl derivatives of a synthetic mixture of
organic acids obtained using (a) EI-MS, (b) conventional ECNI-MS,
and (c) carbon tetrachloride "pretreatment" prior to ECNI analysis

211

Another example demonstrating the strengths of this pretreatment procedure is
illustrated in Figure 75 for the analysis of a mixture of fatty acid methyl esters. Under
conventional ECNI conditions, the signals for these compounds are almost
indistinguishable from background (see Figure 75a). Following CCl4 pretreatment,
however, these species are found to yield ion signals that are easily discerned from
the background noise (Figure 75b). These results suggest that surface reactions
within the ion source of a mass spectrometer can occur readily and alter the ionization
efﬁciency of certain molecules to such an extent that unique chemistry can be
observed. Formation of these radical intermediates through ion source surface
reactions can be considered as derivatization reactions, ip_s_im. Future work in this
area might involve the desorption of other "reactive" gases onto the ion source walls
in order to investigate any ”unique” chemical behavior and selectivity that might be

exploited.

212

 

 

 

1101‘1111‘l0'01li'b‘Il'tTlﬁt'l‘UIUI'I‘Ul'Iﬁlj'jj‘ljj"l'qui'I'lt'ivl
J #
q p

d W

‘ ﬁ11l111111TtIII111 11111rrrr 1111lrrri TTI‘I‘IIVTITI Iﬁrlrl
ED. 0 ’1. 41. BL 1

. Scan Number

 

 

 

 

 

 

 

 

 

 

Figure 75. Comparison of TIC proﬁles of an equimolar mixture of three fatty acid
methyl esters, (C- 14, C-16, and C-18) obtained using (a) conventional
ECNI-MS, and (b) ECNI-MS following ion source pretreatment with
CCI
- 4

213

Final Remarks

The utility and limitations of a number of novel and existing ionization methods
for complex mixture analysis have been evaluated. This thesis demonstrated that the
type and amount of information derived in metabolic proﬁling analyses was enhanced
dramatically when existing ionization methods, such as ammonia-CI were used in
conjunction with conventional electron impact ionization mass spectrometry. The
ammonia-CI mass spectra of derivatized physiological metabolites readily provide
molecular weight information and the overall sensitivity of this technique is
comparable to El ionization. These strengths were exploited for those metabolic
proﬁling analyses in which precursor-metabolite relationships were being evaluated _
through the use of stable-isotope labelled tracers. In addition, combining the
strengths of ammonia-CI and E1 is in general, quite useful for identifying unknowns
present in complex mixtures.

The novel technique of CCl4 pretreatment for ECNI mass spectrometric analysis
was shown to be capable of lowering detection limits substantially for selected classes
of compounds. More importantly, this study has led to a better understanding of
processes important in negative ion mass spectrometry. The prevalence of surface
reactions and their effect on the formation of highly electrophilic species capable of
readily undergoing electron attachment were also studied. Much interest in the area
of reactions between species on mass spectrometer ion source surfaces has been
generated in the past few years. The unique chemistry that might occur between
analytes presented to an intentionally ”contaminated" mass spectrometer can be an
attractive and useful feature under controlled conditions.

The novel desorption/ionization method, K+IDS, has been shown to be capable of

214

yielding both quantitative and qualtitative information on mixtures of organic acids
from urine. The implementation of a new, dual-ﬁlament probe design for the K+IDS
experiment has resulted in a dramatic increase in sensitivity of the technique, making
it more competitive with other existing desorption/ionization (DI) methods in terms of
sample size requirements. The K“ ionization method is free from chemical
interference, unlike the technique of FAB, which suffers from severe matrix effects.

The quantitative strengths of the technique have been shown in the analysis of
urinary organic acids, for the purpose of providing as a mpid ”screening" technique
in the identiﬁcation of disease—states, without the need for extensive sample work-up
and sample derivatization. The results obtained from analyses of these mixtures as
well as others presented suggest that this method might be an attractive alternative to
other DI methods when quantitative information is sought. Additionally, K+IDS has
been found to be nearly universal in the types of compounds that are amenable to the
method. Analyses of steroids, carbohydrates, polymers, peptides and organic salts,
and marginally volatile compounds, such as amino acids and fatty acids, to name only
a few, have been successful by K+IDS. The technique also might ﬁnd utility as a
rapid method for determining the purity of compounds (e.g., synthetic drugs, stable-
isotOpe labelled species, etc.), since the K+DS mass spectra are, in general, free from
interferences.

The real limitation of this technique to date lies in the fact that most experiments
have been performed using a low resolution, low mass quadrupole mass spectrometer,
which does not offer features that would be desirable for K+IDS analyses of mixtures.
This technique should ﬁnd a great deal of success when utilized on higher resolution,
high mass instruments, such as the JEOL HX-l 10 or the Finnigan TQMS.
Implementation of K+IDS on either of these instruments will allow for MS/MS

studies to be perfomed, which in reality are essential for mixtures containing isobar:

215

and isomers. The TQMS will offer the advantage of speed in data acquisition, a
necessary prerequisite for K+IDS experiments in which only small sample sizes are
utilized. The high resolution capabilities of the JEOL HX—l 10, on the otherhand,
offers the advantges of both high mass and high resolution capabilities.
Implementation of the K+ IDS technique on either instrument will require
modiﬁcations to existing probes so as to maximize the sensitivity of the technique,

keeping it competitive with other DI methods.

APPENDIX I

216

Reprinted from Analytical Chemistry. 100‘. ll. "10.
Copyright 0 1986 by the American Chemical Society and reprinted by pen-laden of the copyright owner.

Utility and Limitations of Electron ionization and Chemical
ionization tor the Determination of Position and Extent of
Labeling for ‘30- and 1"‘C—Containing Permethylated Alditois by
Gas Chromatography/Mass Spectrometry

Daniel B. Kassel and John Allleon'

Department of Chemietry, Michigan State Univereity, Eoet Lansing, Michigan 48834

Theetrengthaandereakneeaeaotavartetyotlonlzatlen
methodearedecueeedhtheconlextoIOCMandyeeed
”ON'OJabeledpennemytaldtota. Thegoaiewerete
detemrlnethepcsltionolthelabelandmeextenlctlabd
hcorporaticet Electron lorriaation mace spectre wereueehl
hdetermhﬂhelocetlonoﬂhelabethowever.beratead
Webﬂuﬂonhombothhaiveecﬂheee'eymneﬂn'
molecdee appear to be ﬂared. rem h erronecua en-
tentd-labelng detemrlnetlone. leobertclonhtederenceeh
methane and lsobutane chemical Irritation (Cl) mace meo-
trareqrteﬂtalmeextedotiabehgealcdetionbepertonned
byeotvtngawedratlc equation that nae two rocta. Luau.-
geete twoveheetortabelemlchment. AmnenlaCllneee
epeetraereehovntoaccuatetyreﬂecttheextedetlabel
itcorporatlottwﬂcheaceededﬂtttorthecompomdeb-
veetlgatedhere.

Stable iaotopea. or tracern, have bequently and

eucceeafully
. been need for studying metabolic pathwaya in biological

eyetema. Typically in euch etudiee, labeled precureon (con-
taining "C. I'0, etc.) are preeented to a biological eyetem.
'I'beee labeled epeciee lead to the production of metabolitea
containing the etable iaotope. Once the label ia incorporated.
an analytical method in required to monitor one or more
metabolites and determine the extent of label incorporation
an well an where in the metabolite the label appeara.

Gan chromatography/mane apectrometry (GO/MS) h-
been extremely meful for etudying metabolic pathwaya ho
volving etable iaotopea. Ioctron ionization (El) baa beqently
been need in web etudiee due, predominantly. to the repro-
ducibility and extensive fragmentation, Le., etructural infor.
mation. aaeoeiated with the technique. However. recently a
numberdproblemaaaaociatedwiththetaehniqnehavebeen

MZIWWTOTWMM 0 1900 m W M

217

ldenthhlchhM-aheltlnadequataferenﬁ
analyeee. Sinceﬂdceaprodeecaanunaivehagmentatlnth
reaultinemamepectrahequentbcarrynomolearlarmiﬁ
information. Thiaahobecomeaapeoblem'hendeteemhh'
theenentcﬂabalhcnrpaationdncethandnlhaanentb
maynotcontainthelabeladatem.

GC/chemicalionhaticn(®-lﬁ.hovwer.haabeenhlﬂ
tobevahaableforbothmdandextent-cl-labelq
atndiea. Abundant protonated molaculaa are bequeath
peoducedbytheCImeMandthceeicnaarefwndtobe
idealfordetermlnln‘theamonntofthelabelpreaent. The
GC/Cl-MStechniqnehaabenermceafuﬂyappliedtothe
euadycfetableiaotopeehlmeanabeeeclamlnoaddalntlh
andeenrmU—SheteriodeﬂJlandfattyacidﬂﬂlntnhe
andlnmanyotherareu.

Ofpartiatlarimpoetancebthetneofetableiaotopead
eugare for undying carbohydrate metaboliam (7-10). Inva-
luable information regarding how labeled glucoee b uﬁliud
wheneubjcctedtoavarietyofmetabolicpathwayeiaattaln-
able Labeledeugarehavebeenueedforetudyingaapectad
thecitric acidcycleﬂﬂandothermetabolic pathway;

WereportberethereeultachC/MSemdieaonmrefor
thepurpceeofdeteminirxtheextentofWand'O
enrichment—not an {>de of a epec-iflc metabolic proce-
but ae producte of a new method developed for preparing
eugenthatarenearlylm‘hlabeledinthepceiﬁoncfchcia.
Documentedberearetheetren‘theandweabcaammociatad
with GC/EI-MSandeeveraltypeaofCI-MS fordeterminiu
boththepoeitionandertentcﬂabelinginthceeeynthealud
permethylatedacycﬁcmpra EI-MSiefoundtobevaluahb
fordeterminingthepoaiticoofthelabelintheaealditolabu
yieldeerroneouarceultefcr extend-labeling determinatial.
aedoeomeformeolCl. AmmoniaClieeuueetedtobethe
method ofehoice for thie claaacfcompounda for extent-d-
labelingdeterminationa.

EXPERIWALSWHON

AllexperimentewerepedcrmedcnanHP-m GC/MS/m
quadrupoleinetrument. 'I‘heGCinlet'aamcdiﬁedeothat
capmarycolumnecouldbelned. Adeectivated.nonvitreotn
capillarytranaferline(40anbn¢.0.llmmi.d.,8cientiﬁcb-
mmhdmmdmtbocmwfmvhichu
maintainedatzm'CA ' bae(55m,0.32mmi.d.)capﬂlary
GCcolumntJandWSdenh’ﬁclvnamedfcrallexperimenta.
belated. derivatized alditoh tare introduced on column; the
columnmtempenhneprcpammedhcmﬁOtoM‘Catam
clm'C/min.

Padecﬁonicnhaﬁcmtheaourcntempentmewaahelddu
'C,elect:onenee¢iea\nedm70eVandldeV. hideout!“
vaamaintainedatmaA. PerCIeaperimentaJourcet-e
perataure waa maintained at 150 ‘C. electron energy nae 150eV,
andemiaaioocurrenteraamnA. Methana(~W-Mmtar).
iobtuanet~250mtcn),andammcnia(~Mmton)weremad
.Clreagentguea Sanceprcemneaweremeanuredluplnci'
ahono'aolldaprobeattachedtoathermccouplegaugedireetb
againattheCleourceeoluma.

Selectedicnmonitceimﬁnnmlnedtointepatepeaham
ddeaignateionnfordeterminatinmcltheextentdlabaliq. PC
ahort-term replication of experimental determination den.
tentd—labenn‘therewaagoodpredaicn'lheaverageetanhd
deviationforElwaedﬁS. Theatentoflabelenrichmantb-
termlnedbyNHJIIwaaﬂtLlS.

RESULTS AND DISCUBSION

I. 104V Eloctronlonlantlon. 'I‘hefiretmethoduaedto
determinethepoeitionandertentoflabelingintheforrein
carbon permethylated alditola wan 70-eV El-MS. In thin
diacuaaiomwewillfoareoneparticularaetcﬂabeledm
the permethyl eorbitcla. Selected ion monitoring (SIM) w.
utilisedtoproduceacan-ateintepntedintenaitieaofdceknate
ionafcrthelabelenridrmentdeterminaticu. Minnie”
laiathe'IO-eVEImaaaepectl-umofunlabeledpermethyl

“www.meauoaaavuee-me

 

 

 

 

 

 

 

 

 

 

 

 

 

 

election Inﬂation «nae epoch at (a) melee.
or (er-01., rel [av-01.. an to ["ch

Pemmm sonnet.
" or ,ocn 3

rate Q :1 I
N! ”I

llC-OOI, m .
ill I"

ago—or

min I”

"Ci-00", we "I,
all C
tic-OCH,
an at
ave C

”0120013
"peel. Oeelyutelorlatnmedbyelectrcnhlutloncnpeo
mom

ecrbitcl (C“H.O.-M). Nomdeaduhnﬂdﬁlaobaerved;
thehigheetmaaeionintheepectnnn.m/e221,correaponb
tclcmcftheneutral (‘CHpCl-IJ. Anumberofpairecfioll
that appear to concepond to eimple C-C bond cleavag-
(primarybagmenta)througboutthecarbonehelctonareaho
obeerved.aaehowninﬁ¢ure1 Theaeererechceenaathe
denignateionefordeterminh‘theextantoﬂahelln.

218

 

Tlhh I. '03“. of [AN “'0. Designate l“.
”lawn”

(Men/lu)>nausrallsotopieratio tor. test test
AIM/sill x n . x
Aim/sin as I
A's-elem X

nC-IabsledStumn

conditiu

luellﬂU)>naturalbotoplcrntis la. 8c. Sud
AIM/s18 x x as
A-ln/st'n It at
Aline/rm I

labelatpositia

 

 

Deuterated analogues of these straight chain alditols have
been studied extensively by Golovkina at al. (7). Their results
suggest that both -OCH, and —ll groups can scramble, corn-
plieeting the mechanisms through which some ofthe ions in
the mass spectrum can be formed. However. they suggested
that the designate ions in Figure 2 should be formed in the
straightforward processes shown, except for m]: 45, (C,H‘O)’.

The El mass spectrum of unlabeled permethyl aorbitol
suggests that the larger primary fragment ions decompose
further by loss ofmethanol; that is, the ions cfm/r 221. I77.
and 133 eliminate methanol (32 u) to form m]: 189, 145. and
101. respectively.

Onthe basisofFigure la. El-MS appearstobewell suited
for both the determination of position and extent of labeling
in permethyl alditols. Take as an example [l-"Olper—
methylsorbitol. Golovh'na suggests that such molecules may
be regarded as symmetrical structures and that their stereo-
chemistry has no effect on their mass spectra Furthermore,
they suggest that cleavage of C-C bonds between atoms of
equal distance from both ends of the carbon skeleton can be
regarded 0 equivalent. Assuming that the compounds undc
study are close to lmﬁ labeled (as will eventually be shown).
we would expect that mlz 47 (CH,"()CH.)‘ would be as
intense as m/r 45. the 1'0 analogue (alter correcting for
naturally occurring "0). since for [l-“Olpermethylsorbitol
halfoftheseionswouldbeproducedfromthelabeledendand
halfbom the unlabeled end. For the fragmentation presented
inPigureZwewouldalsoexpecttoseem/rmequalin
intensity to In]: 221, as would be the case for the labeled]
unlabeled ion pairs 179/177. 135/133. and 91/89.

The presence/absence ofthe label in the designate ions will
depend on the location of the label. For a aorbitol molecule
labeled with “0 on C2. there should be no enhancement of
the m]: 47; Le., the label should never appear in this ion if
it were only formed as shown in Figure 2. Conversely.
there should be no m]: 221; the entire ion current for th'n
fragrnentionwwldahihtom/rmieathisionahmddalwnys
contain the label for the 2-“O-labeled sugar. The ion pairs
at]: 89/91, 133/135. and 177/179 should still be formed with
ratios of unity. Similar arguments can be made for a per-
methyl aorbitol molecule labeled with "O at the 03 position.
For any isotopic enrichment of a particular skeletal C or O.
the position of the label should be apparent from the den-
ignate ions as shown in Table I. In addition. these designate
ions should be useful to determine the extent of label en-
richment in molecules of this type. For extent of labeling
determimﬁonaSIMht-ed and the followingequationahwﬂ
be valid for any designate ion (A), of intensity I:

I(A+n)

$ label ennchment - W

x 1m (1)

 

ml. ll. Ixteat-ef-Lahellng Implication he. Designate

 

S labellq
compose-d eels at sale in eels rss
70W I
H'O-H' III in 18
0.0-“ 1! O 10
t-‘OH It. ill 1C
10 eV I
i-“O-l 181 err 137
O-“O-I 10 I 1!
13%“ I” 1“ '1
'hi - permethylaorbltd.

 

lnthisequatitmn-lfosW-labelimn-Zhr‘b-labsh
studies. I(A+n)isthelntensityoftheisotopicpeah minus
the intensity due to the natural isotopic abundance.

Showninl’igurelbisthe'IO-eVEImasaspectrumofthe
(nearly 100% labeled) [l~“0]permethylscrbitol. It i im-
portant to note that we can use the appearance of the label
in the designate fragment ions to conﬁrm its position a
suggested in Table 1. However the ratios of the labeled:
unlabeled designate ion peaks (with intensity I) are not at al
what was expected, e.g., I (223) > [(221). Table ll shows the
results of the attempts to determine the extent of labeling
basedontheintegratedpeahintensitieafortheionpainwith
m/e 135/133, 179/m, and 223/221. Peak integration based
on SIM gives values for extent of labeling greater than la)‘.
The values vary for the three sets of designate high ma.
hagment ions. Note also. that the ratio of m]: 45/47 lmpli.
less than 100% label incorporation.

Theextentofhbelingalculnﬁomutggutthahintheuas
of [l-“Olpermethylsorbitol, dissociative ionization from the
‘bottom'ofthemoleculdasshowninl’iguremisoccurriu
toagreater extentthanfrom thetop. A number ofpossible
explanations may be proposed to explain this behavior. One
possibility is that isotope eﬂects may play an important role.
However, we mid not expect isotope effects to be operative
over long ranges but would be most apparent in fragmenta-
tionsinvolVingbondstothelabeledatom. Amore reasonabh
explanation may be that the aorbitol molecule is not in fact
symmetric. as Golovkina suggests The 01 is not in exactly
the same chemical environment as 06; the same is true {a
02 and 05. etc. Therefore, the probability for ionization at
01 may be slightly different than that for 06. as well as the
unimolecular decomposition rates for processes occurring at
both ends of the molecular ion. Whatever the cause 0! the
preferential ionizaﬁon/ﬁ'agrnentation ocmrring at the bottom
of the molecule. the 70-eV mass spectra cannot be used to
determine the true extent of labeling by El.

The 70-eV mass spectrum of essentially 100% enriched
[6-"Olpermethylsorbitol is shown in Figure lc. The isotopic
distn'butionofthehigbmaasdesignate ionsisconsistentwith
thepropoaalthatproductsduetocleavege of. e.g..theC5-Q
bondareoccurringtoag'reaterextentthanthosedueto
cleavegeoftheC‘l-C‘lbond. Whatisobservedhan inveraitm
of the results found for the [l-"Olpermethylsorbitol. The
preferential fragmentation horn the 'bottom' of the molecub
in [6- “Olpermethylsorbitol produces more intense unlabeled
designate ions relative to their labeled counterpart. The
designate ions ‘and the extent of labeling results for each
(calculatedbyeq l)aresummariasd inTable II. 'I‘hercsultn
are in contrast to those for the l-“O compound, with the
exception of the ions at mlz 45 and 47. However. these ions
mpmdumdbyambudpmnotonlybythesimpla
C-C cleavage shown in ﬁgure 2.

2l9

 

m
‘ tes

s I.
1‘ us
4 D!

a:
.i. .L .L .L .L

nls

ﬂags. thlewmnrtalmmmd
[t- W

A A
i

 

 

 

T‘

 

 

Asimilartrendiaobservedforthemlabeledalditoh
Showninl'igureldhthe70-eVBlmassspsctmmofli-
“Clpermethylsorbitol. Again. the location of the label can
becmﬁrmedbasedonthepreeenceoflabeleddesignsteial.
Secondly. enrichment appears to exceed 100%. as shown in
Table Ii. indicating further that, indeed. both ends of the
molecule do not behave in a chemically identical msnnu.

Tablelmggeststhatthe‘lO-eVmassspectracouldbeussd
to distinguish between a [l-“Olsorbitol and [2- or 3-"01-
aorbitol, but not between the l- and 6-labeled spades.
However.wehsveshownherethatitispcssibletodiﬂerentists
each one. due to the asymmetry of the system. However.
determinations of extent of label enrichment cannot be per-
formed due to preferential dissociative ionisation from the
bottom of the molecule.

Amilyaccurstemeastneofthslabelsnrichmenttningthm
designate ions is further complicated by the fact that their
intensities (relative to the total ion current (T‘IC)) are so small.
However. the ion statistics in these experiments should not
be the source ofsuch large errors. continually skewed in the
same direction.

Another possible explanation for the observed behavior b
that the mass spectral fragmentation mechanisms for thus
molecules are much more complex than those previously
described. Extensive rearrangements may be occurring fol-
lowing 70-eV electron ionization and the fragment ions may
have structin'es much different than what is suggested by the
simple model in Figure 2. Thus. low-energy (18 eV) El w-
chcsen as a possible solution to the problem. Frequently, at
lower ionization energies there is less fragmentation and
scrambling; also a larger fraction of the TIC may appear a
higher mass ions. possibly improving ion statistim.

2. Low-Energy Electron Ionization. The lB-eV mam
spectrum of [l-"Olpermethylsorbitoi is shown in Figure 8.
Asexpededtherelative intensitiesofthehigbmawdesignate
ions have increased in comparison to the ‘70-eV spectrum.
Thereisadeaeaseinthesbundanceofthem/stsiomwhich
may be expected for a lower ionizing energy. The designate
ionpainobservedinthe'IO-eVEIspec-trumareprssentand
the ratios of lsbelednmlabelcd species are essentially the same
inthelowandhighenergyspectxn—ieatheloweneryspscm
still lead to calculated extent-of-labeling values greater than
lw%.whicharenottmlikethevsluesobtainedfromths70ev
s

The extent of labeling determinations from the low-energy
EI experiments for the [I-"Cl- and [l-"Ol- and [6"01per-
methyisorbitols are presented in Table II. Since they parallel
the 70-eV El results. this may suggest that scrambling 5
probably not the problem in the electron ionization spectra
and may support the proposal in which dissociative ionisatim
doesnotoccurforths‘top'snd'bottom'ofthemoleculeto
the same extent.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

F a

i‘ a”
t
t
j in ,

lllllllll'l
. - 1M“ s
4 as”
t
i
it us
:"think“
u .. Lin-.1.

min
:1
i: ...,.
in L ll nﬂrLlLJL
..LLLLLLLL

nls

Planet. mmwunasmmspsmormm
man» [1301me (quagssnssonorsnowgmn
boredcnothsspectmnhh.

Since electron ionization did not yield meaningful extent
of labeling information. a variety of chemical ionization en-
periments were performed. Proton transfer reagents should
produce (M + H)‘ ions that containtheintactmgar.thusths
extent of labeling should be apparent.

I. Methane and Isobutane Chemical Ionization.
Methane Cl-MS has been used successfully for the analyst
of a wide variety of biologically related compounds (It-M).
It has been successfully used for extent-cf-lsbsling studies.
For polar compounds such as permethylated sugars. proton-
ation and fragmentation are expected.

The methane Cl mass spectrum ofthe nonlabeled par-
methylsorbitol is shown in ﬁgure 4a. The base peak is the
protonated molecule (M + H)’ at m]: ”7. The loss d
methanol following protonation is a prominent procem leadim
tothe(M+ H-Cleﬂ‘ionatm/sﬂh. Theseintensehﬁh
miom shaildbeusefulfordeterminingtheextentcflabsl
incorporation. The methane CI mass spectrum of [l-"Olo
permethyisorbitol is shown in Figure 4b. SIM results of in-
tegrated peak intensitiesfortheionsm/rmandWm
low levels of enrichment. approximately 70%. Note that in
Figurets.thereisspeakatm/sZSS.(M+H-l~ld‘. Thh
fragment ion is an isobaric interferent in the molecular ion
region of “as mass spectrum of the l-"O-isbeled compound.
The ion at m/s ”7 not only consists of the nonlabeled pso-
tonsted parent but is also the hbeied (M + H - Hg)‘. It 5
not possible to deconvoluts the information in this region d
the spectrum for the labeled compound based on relative ion
intensities in the nonlabeled spectrum since the ratio of the
ions (M + H)‘ and ((M + H - H)‘ is pressure dependent.
Therefore. in the protonated molecular region ofthe spectrtms
ofthelsbeledcompmnd.twoionssrsprssentthatyieldthres
peaks(l.labeled;nl.nonlahelsd): m/s265.[M(nD+H-H.r;

220

 

r (marsh-If
1 «Harper
0
9

 

 

 

OJ‘
° \e—e
I i 5 T i 1
lacettonellabet

Holst. ueu-am‘mmusuoesmw
W

mgrmmm+n-nd'+m¢nn+m°:m/sw.mm
+

Attempts to determine the extent of labeling by deconvo-
lutingtheintensitieeoftheeethreeioneinvolvethesolutia
ofsquadrsticequation thathastworoota lnll'igureebthe
relative intensities for the three ions are l(265):l(287):l(m
I 0.86:2131zlm. The extent of labeling that would yield thﬁ
results could be 96.16% or 85.65%. Therefore. due to the (N
+ H - H)’ interferent. this portion of the methane Cl
spectrum cannot be used for determining the extent of label
incorporation. (The ion labeled (M + H - l'lJ‘ may also be
written as (M - l-l)‘. which could be formed directly in e
hydride transfer reaction with the ethyl cation.)

The methane Cl spectra of permethylated alditols also
contain a number offragment ions. The designate ionsatm/s
133, 177, and 221 are present in these spectra. These are
showninmoredetailinl’iguretc. Heretheseionsareformsd
following protonation. not as the result ofeiectron ionisation
it is interesting to note that the ion pairs 133/ 135. 177/17’.
and 221/223 appear (for the “(l-labeled species) with ratios
very close to unity. suggesting close to 1m% label incorpo
ration. Unfortunately. these ions are only minor Cl products.
and the ion statistics are insufficient to be relied upon for
accurate determinations of extent of labeling. although SIM
data on these ion pairs suggests so 1 5% label incorporation.
Apparently. on protonation. the asymmetry of the system b
reduced and the simple assumptions in the El discussion are
accurate (for these designate ions)—i.e., these results imply
thattheprocessesleadingtotheaeionsfonowingprotonatim
occmatsitesonthebottomandthetopofthemoleculeto
the same extent (or, that decomposition processes are simply
less complex in Cl than in 81').

Since proton transfer from CH.‘ to such molecules is an
exothermic procem in which bagmentation frequently follows.
other Cl reagents for which proton transfer would be lem
exothermic were investigated isobutane was chosen as the
next Cl reagent for study. The isobutane Cl mass spectra (1
permethylated alditols contain less fragment ions but contain
boththe(M+H)‘and('M+ H-HJ‘ions—thusthesame
problems exist for isobutane Cl as for methane Cl in that
regionofthespectrum. ltshouldalsobenotedthatthe
sensitivity of isobutane Cl appeared to be less than that for
methane Cl in the analysis ofthese compounﬁ.

A third Cl reagent with a higher proton afﬁnity than
methaneandisobutenewasnextchoeemammonia. Hm.
before theremltsofammoniaClarediscusssdJnintereetiq
feature of the methane and isobutane Cl experiments merits
discussion. lnadditiontothem+lﬂfandCM+H-HJ°
ions formed. there is also loss of methanol following proton-
ation of permethylsorbitol forming (M + H - CH,OH)’.
Prmumablymethanoleliminationoccursbyinductivecleavqe
following protonation. The interior oxygens (positions H)
areattachedtomorehighlyhmrchedcarbornwhichcanmore
readily stablize the positive charge than positions 1 and 0.
Thus. the interior oxygen may be expected to exhibit slightb
more basic character and hence be protonated more exten-

 

4 A A A A 4 -
I
I
:

 

 

A A A A A A A

 

 

Drum lady

 

 

 

 

..' ".L 'L' .L 5"]. "L '1."'.L'"LL"'.L"£"1.
nIs

Mae. WWWMmepectedmm
sol-d. (b) [130}. “it! li—‘OIW

sivelythantheterminaletheroxygem. OurClresultaforthe
[1- through 6-"Olpermethylaorbitoh suggest that the oxygen
at the 2 position is preferentially protonated. This conclueim
is based on the data presented in Figure 5. The [2-“01-
permethylsorbitol shows an enhanced loss cflabeled methand
relative to the other "O-labeled compounds. The isomer
permethylmannitolbehavminasimilarway. Wearemrrsntly
investigating possible explanations for this behavia.

L Ammonia Chemical ionisation. A great deal of with
has appeared in the literature in which ammonia Cl-MS h-
been used for the analysis ofth underivatized and (Megsi)
derivatized sugars (rs-29). The predominant ions formed are
typicallythe sdductioasau + m‘and (M+ May. Since
less energy is involved in the proton transfer from ammonia
than from methane. lem fragmentation is observed. No
thermore. ammonia Cl has been used for determining the
extentoflabelincorporatiaiinsmdieeoforganicaddsintype
1 diabetic patients m.

The ammonia Cl man of permethyl alditoh show
theexpectedm+m‘and(M+NHJ‘ionsasthepre-
dominant species. ﬁgure 6 shows the ammonia or men
spectra of nonlabeled. [l-'C]-. and [l-“Clpermethylsorbitol.
Note thatinthe nonlabeledcomposmd.thereiano(M+H
- HJ’ interferent. only the protonated molecule at In]: ”7
andtheammoniumadduaatm/rm Sinoethespectraars
simple and the sugar molecules remain intact. the determi-
nation oftheextentdhhelingisetraightforward. An
ofthedauinﬂgunﬁpamhandqrevealthatthepercsnt
label enrichment is 97 a 1% for both molecules. con
the assumption made at the beginning of the text that these
permethyl alditols were essentially lN$ labeled from the new
synthetic route that was used to make them. The extent of
labelingdeterminsdfromtheM+l0‘and(Mt-NHJ‘
peaks is the same.

22l

'l‘herefore.amn'ioriiaClbthernethotlofchoieefosdstab
miningtheextentoflabelingofcompoundsofthlstype. It
blmportanttonotethattheprohlemsassociatedwithuirg
SI for determining isotopic enrichment were obvious since
essentially pure labeled compounds were used. However. I
this were a metabolism study in which the extent of label
enrichment was small (typically <20”. the El results would
have suggested an extent of labeling that was too high. and
itwmnldbedifﬁculttorsalisethatthiavsluewasincorred.
Therefore. El data oflabeled compounds must be interpreted
with great care in such quantitative determination.

CONCLUSIONS

The above discussion focused on the mass spectrometric
behavior of labeled permethylsorbitola Iectron ionization
was a useful tool for structure determination; however. the
rateeofdimociativs ionizab'onbombothendsofthemolearb
appear to be different. which leads to diﬂicultiee in deter
mining the extent of label incorporation based on simple
assumptions

Both El and NH, Cl have been used for similar analysu
of permethylribitol. arabinitiol. erythritol. and mannitol.
Ammonia Cl confirmed that the label incorporation exceeded
97 '5 for all of these synthesized sugars.

Again. it must be emphasized that great care must be ex-
ercised in using El mass spectral data for extent-of-label
incorporation calculations on molecules such as these. De-
pending on the position of the label. results could consistently
be high or low due the unique chemical environments at the
two ends of acyclic sugars.

Ammonia Cl is a useful method for label-enrichment de-
terminations since >957s of the total ion current is carried
in the protonated molecule and the ammonium adduct.
However. Cl alone is insufﬁcient to determine the position
of the label. While ammonia Cl was useful for such studies
of "C. and “O-labeled compounds. this may not be true for
deuterated sugars. since li/D scrambling between the analyte
and Cl reagent may occur.

mowumonrrr

Theeuthorswhhtothsnhlsianlluseelmanfmhelpﬂ
dbcussions. and I. (last. Jr. hr making available the labehd
compounds.

Registry lie. '0. ism-11a "C. “703-? permethyl-
eorhitol. sous-ass; ll-“Olpermethyhorbitol. l‘t-lism 1; [O-
nOlpermethyleor'hltoL ”1971M [l-‘Clpermethyborbitd.
lOlSlGﬁl-C.

surname CITED

(time.uam.aa.-naaaum. It.
Ween“).

WLka nkm.o....sna.m tees.

Kilt" 0W” . mars
llL'Seﬂlt .e.a.m.m.mrsrs.m.
.I

. .MMWMM 1.“.
(it) mun-o1.- has. Q; it. FJ. an. arena. t0“. M.

(12) Mt”! “lilo 0.th L; WOLJLM. m.“

(13) WamnTakMuYaheQY. W 0mm. 1.78. 41.573.

(")th eats..-P «Mlle; “*7“;th
Scm‘trom ”£10.46.

"sumo-to. WW.J.D.W.~ “H."

(10) MAKNW .Ll’etmtel'. ”73.47. 4021.

(in 1.70034: W. C

manager-rain; see-aw seed-y.c.c owing

a.

RECEIVED for review December 4, 1985. Accepted February
14.1986. This work was made possible through the ﬁnancial
support of the National institutes of Health (NIH Grant Nn
BROOM 16).

222

 

Urinary

Metabolites of L—Threonine in Type 1

Diabetes Determined by Combined Gas
Chromatography/Chemical Ionization Mass

Spectrometry

 

D.B.Kasad

Departmcll or cum. Michigan State University. an basing. Michigan cam-rm. use

M. Marth

Departmcd of Biochemistry. Michigan State University. Baa Lansing, Michigan dud-l3”. UM

W.Schall

Department of Veterinary Medicine. Michigan State University. East Lansing. Michigan 488264319. USA

c. c. Sweeleyt .

Department of Biochemistry. Michigan State University. East Lansing. Michigan Cult-l3”. USA

 

Metabolic proﬁling of urinary organic acids from patients with juvenile-oust (Type 1) diabetes mellitns have
rescaled signiﬁcantly elevated levels of Z-hydroxybutyric and Heoxythreoaic adds. To test the hypothesis thd
these metabolites. as well as d—dconerythronic acid. are derived from L-threonine. stable botopblabeied threonine
was infused into as insulinaieiicient dog and the incorporation of "C into these metabolites was monitored by g-
chromatography/mass spectrometry. Electron ionization was relatively insesitive. but positive chemical ionlntlen
with ammonia as the reactant gas gave both protonated molecules and lMeNHJ’ ions. which could be analysed
by selected ion monitoring. The isotope-labeled species of Z-hydroxybntyric. 4-deoxyerythronic and d—deoxythreonle
acids were observed. bat ”C was not incorporated into other organic acids. Than. It Is proposed that Mhreonlne

is a precursor of these metabolitm.

 

lNl'RODUCTION

 

Juvenile-onset (Type 1) diabetes mellitus is caused by
the persistent deﬁciency of insulin. accompanied by a
large excess of glucegon. Several metabolic disorders
are associated with the disease, including increased
hepatic gluconeogenesis. elevated levels of blood and
urinary glucose. and excessive mobilization of free fatty
acids (derived frm adipose triglycerides) with a con-
comitant susceptibility to kctoaddosis. Electrolytic
imbalances frequently accompany the kctoacidotic
state”.

One method for exploring the metabolic status of an
individual is to monitor acidic metabolites of urine. This
is the basis of 'metabolic proﬁling' and numerous papers
have been published in which combined gas
chromatography/mass spectrometry (GC/MS) is used
to identify and quantitate the urinary organic acids.“

Diabetics who are under poor insulin control may
have increased levels of urinary 3-hydroxybutyric acid,
derived from increased oxidation of free fatty acids.
Lactic acid is also increased in the urine under some
circumstances. We have monitored approximately zoo
organic acids of morning-fasting urine specimens of 20
patients with Type 1 diabetes and of 17 age-matched
control subjects by computer-assisted GC/MS.' Of par-
ticular interest was the ﬁnding that the diabetic patients
excreted consistently higher amounts of 2-hydroxy-
butyric acid (than age-matched controls) as well as

lAuthor to whom correspondence should be addressed.

0887-03086] 1005334 SOS.”
0 ”Idby lohaneylSons Ltd

enhanced levels of Meoxythrconic acid and several
other metabolites The levels of urinary 3.hydroxy-
butyric and lactic acids were elevated in some but not
all patients.

The metabolic origins of 4—deoxythreonic. 4-
dcoxyerythronic and 2-hydroxybutyric acids have not
been previously studied. it has been reported. however.
that 2-hydroxybutyrate may be found in urine when
there is a large excess of excreted lactic acid. a con-
sequence perhaps of an elevated NADH/NAD‘ ratio.’

it can be postulated that Zohydroxybutyric acid may
be derived from threonine by threonine dehydratase
(Scheme 1). The four-carbon metabolites, 4-deoxyery-
thronic and deoxythreonic acids. may also be postu-
lated to be products of threonine reactions involving
both threonine deaminase and reductase (Scheme l).
The ramiﬁcations of this dependency would be sig-
niﬁcant. if true. since these metabolites might then be
important measures of protein catabolism in skeletal
muscle and other insulin target cells. when deprived of
glucose. To test this hypothesis. we have determined the
stable isotopic abundance of 2-hydroxybutyric. d-
deoxyerythronic, 4—dcoxythreonic and other organic
acids folloning the intravenous administration of [U-
"C]-thrconine to an insulin-deﬁcient dog with clinical
symptoms similar to human Type 1 diabetes.

Experimental
Reagents. Highly enriched [U-"C]-threonine (>993;
labeled species) was provided by the Los Alamos Stable

swear-resume
wrest-ms l”

isotope Resource. bis(trimethylsilyl)-triﬁuoroacetamide
(BSTFA) was purchased from Regis Chemical Co., and
triﬂuoroacetic anhydride was purchased from Supelco.
inc. HPLC-grade n-butanol was utilized for derivatiza-
tion.

Equipment. A HP-5985 quadrupole gas chromgto.
graph/mass spectrometer was operated in both the
positive and negative chemical ionization (Cl) modes.
A modiﬁcation of the instrument was made for fused.
silica capillary GC experiments with a 60-m DB-l wide-
bore capillary column. purchased from 1&W Sci., Inc.

CC conditions. All injections were made onocolumn. The
oven was temperature-programmed from 55 ‘C to 300“:
at a rate of S'Cmin". Carrier gas was helium (flow
rate = 2.5 ml min"). An open-split interface consisting
of deactivated. non-vitreous capillary tubing (pu rchascd
from 8.0.8.) was maintained at 280 'C. and the split
was determined to be approximately 3: 1.

MS conditions. Cl was employed for the analysis of
urinary organic acids. using ammonia as the reactant
gas in the positive ion chemical ionization (PIC!) mode.
Source pressure was held constant at 0.3-0.4 Torr to
ensure maximum sensitivity. Source temperature was
maintained at lSO'C for all analyses. Filament electron
energy was set at l75 eV. PIC l and negative ion chemical
ionization (NICI) experiments were performed on the
plasma amino acid samples. Methane was used as the
reactant gas (pressure = l90 mTorr) in the NiCI ana-
lyses. Source temperature was held constant at l35 'C.
Filament electron energy was maintained at 150 eV.
Selected ion monitoring (SIM) experiments were simul-
taneously performed on the protonated molecule and
molecular adduct ion species.

infusion of labeled threonine. Hyperglycemia was induced
in the diabetic dog by hypoinsuiinization one day prior
to infusion of labeled threonine. Just prior to infusion
of the ("Cd-threonine. insulin was administered to the
dog in order to restore normal blood glucose levels. A
control urine was collected at this time. Infusion
occurred over a period of 20min. Due to the rapid
delivery of ("Cd-threonine into the animal. urine speci-
mens were collected hourly at times just following
infusion. After 5 h collections were made at longer inter-
vals for the remainder of the 48-h collection period. A

blood sample was drawn just after completion of the
infusion.

Formation of derivatives. Organic acids were isolated from
urine specimens as described in an earlier paper by
Sweeley and Gates." Trimethylsilyl (TMS) derivatives
of the organic acids were made by reaction with BSTFA
in pyridine (4:1) at 80 ’C for l h. The resultant mixture
of derivatized organic acids was sealed in capillary tubes
and stored in a refrigerator until introduction into the
gas chromatographic/mass spectrometric system.
Plasma samples were deproteinized and the amino acids
were derivatized following the procedures outlined by
Kingston and Dufﬁeid." Amino acids in a protein-free
ﬁltrate of plasma were separated from neutral substances
on a Dowex ion-exchange column. The N-
triﬁuoroacetyl-n-butyl esters were then formed and the
resultant derivatives 'were sealed into capillary tubes.

 

RESULTS

Organic acids in nrine

Attempts were made to determine ”C isotopic abun-
dance in the organic acids of interest by conventional
electron impact mass spectrometry (EIMS) of the TMS
derivatives. To obtain acairste determinations of ”C
enrichment. it was necessary to have relatively abundant
molecular ions or high'-molewlar-weight fragment ions.
The predominant ions observed in the analyses of the
derivatized organic acids by EIMS were related to the
fragmentation of the TMS derivative and provided no
information about the relative enrichment of "C in the
species being analysed (see Fig l(a)). Furthermore. in
general. very little or no molecular weight information
was available from the EI mass spectra of these organic
acids. In addition. the results of the analyses by EIMS
showed signiﬁcant variations from one run to the next.
due to the low intensities of the highomass ions
monitored. To overcome the problems associated with
EIMS. alternative ionization methods were employed
for determining "C enrichment in urinary organic acids.
Ammonia Cl has previously been successfully used to
determine the isotOpie enrichment of I'0 and "C in
labeled permethylated sugars,n since only molecular
ions were observed in the mass spectrum. Similarly.

 

224

 

I47
tie-crept ,

 

 

 

no 60 ISO 230
ml:

 

 

l.) he”

lMti‘

l ‘
A A A l A L a
w

{—760 n'oﬁzio‘aéo '30
Iii/I
Figure l. (a) El mass spectrum at 2-hydroxybutyrie acid di-TMS
from a urine specimen collected at the 2nd h after onset of labeled-
threonine infusionfb) Ammonia Cl mass spectrum of 2-hydrorry-
butyric acid (di~TMS) from the same urine specimen. Source
pressure-”mTorr.

5

Relative obtndnnce Pl.)
8
8 a
Total run intensity at

 

 

 

organic acids produce intense molecular ions by
ammonia CI. It was concluded that this ionization
method would be suitable for the enrichment studies.

The ammonia CI mass spectrum of the bis(trimethyl-
silyl) derivative of 2-hydroxybutyric acid is shown in
Fig. l(b). Molecular weight information is apparent in
the form of the protonated molecule and molecular
adduct ion species. The [M+NHJ’ ion comprised
approximately 60% of the total ion intensity (Til). and
was ideal for the ”C isotope enrichment studies.
Enhancement of the (A+4) peaks. at rn/s 253 and 270.
was observed for both the [M+H]‘ and [M+NH.]‘
ions. respectively.

SIM was performed on both molecular species and
their isotope-labeled species (+4 u) from 2-hydroxy-
butyric acid in order to obtain the best ion statistics.
Similar experiments were performed for 4-deoxythre~
onic. Lhydroxybutyric and 4-deoxyerythronic acids.
The ions chosen for the analyses are summarized in
Table i. All ions were monitored simultaneously. and
the most abundant ion [M + NHJ‘ was given the short-
est dwell time (100 ms). in performing the SIM analyses.
it was noted that urinary 3-hydroxybutyric acid could

 

Table l. Molecular adduct ion and protonated molecules of 3-
and 3—hydroxybutyrie. 4—deoxyerytbronle and ‘-
deoxythreonie acids for SIM sslng ammonia CIMS

Organic sud tents) moribund
2-H roxybutyrie 240. 263. 266. 170
S-eroxybutyria 240. 263. 206. 270
toeoxyerythronlc 337. 341. 54. ﬂ
4-Deoxythrsonic 337. 3“. 354. 3.

 

 

 

 

 

Til

asosoosaoseoseossoeooem
Sodom

HarrelﬂtsuseolSlMundermiisacondltlonetodetsd
lowbvshol'kemiduMthadeld-NS)“
Shydmsbtrtyrleeetd(d-TMS);IIpsahsstacanﬂOsrshm
l-WandmosenamnumﬁmnS-WW
acid.

be used as an internal reference (or check) for the ”C
isotopic enrichment in 2-hydroxybutyric acid observed
in urine samples collected following [U-"Cl-threonine
administration. Conveniently. since the two acids are
isomers and eluted from the GC column within a narrow
retention time span. the same four ions were monitored
in the same gas chromatographic run. The small ion
intensity observed in the (A+4) peaks of 3-hydroxyo
butyric acid (at ml: 253 and 270) was due solely to
naturally occurring isotopic contributions. primarily due
to silicon (Si) isotopes (Le. no "C enrichment was
observed). This [A + 41’ ion intensity was used to correct
for ‘bachground‘ isotopic content of 2-hydroxybutyric
acid. True label enrichment for 2-hydroxybutyric acid
was thus obtained. and provided accurate results since
run-to-run variations were assumed to be the same for
both ada.

The SIM results for the ion pair at In]: 266 and 270
for 2- and 3-hydrox utyric acids are shown in Fig. 2.
An enrichment of C for Z-hydroxybutyric acid was
clearly observed. relative to that of 3-hydroxybutyric
acid. For 4—deoxythreonic and 4-deoxyerythronic acids.
no internal reference wa available; hence. isotopic
enrichment was obtained by subtraction of a control
urine mixture (collected jun prior to labeled-threonine
infusion).

SIM analyses were performed on the control urine
and the samples that were collected serially for 48h
following the onset of infusion. Figure 3 shows the per
cent of labeled spcdes observed for 2-hydroxybutyrie.
4-deoxyerythronic and 4—denxythreonic acids. as a func»
tion of urine collection time (after corrections were made
for the naturally occurring isotopic contributions. as
described earlier). For 2-hydroxybutyric and 4-
deoxyerythronic acids. maximum ’C incorporation
from threonine occurred between the 2nd and 6thh
following infusion. The peak maximum for 4-deoxy-
threonic acid was observed at later times. following
infusion (between the 6th and mm h). However. it was
determined that when the per cent labeled species of
4-deoxythreonic acid was normalized for the diﬂerem
pool sizes (from the urine specimens collected serially).
a shift in the 4-deoxythreonic curve resulted. so that h
was coincident with both the 2-hydroxybutyric and 4-
deoxyerythronic acid curves.

sis 225

 

 

figural. ExternallabelanrichmentoU-hydroaybutyrieM-M).
d—deoxyerythronie and d-deoxythreonlc acids (vi-TMS) determined
overazd-hcoﬂectionperioduslngammoniaCIMSnnunoniawaa
maintained at constant praaawa (Pr-38 mTorr) lor sl axperlmerwa.

Label enrichment was observed to be 2.83 s0.4l% at
the peak maximum for 2-hydroxybutyrate. For 4-
deoxythreonic acid. maximum ’C enrichment was Li a:
0.l$'/o. Interestingly. maximum "C enrichment
observed for 4—deoxyerythronic acid was 3.25:0“.
Runs showed excellent reproducibility over a wide range
of source pressures. with standard deviations frequently
calculated to be as small as mosses No "c enrichment
was observed for any of the other organic acids. Figure
4 is the reconstructed TTl plot of the organic acids found
in the urine sample collected just following infusion.
Peaks labeled I-23 are identiﬁed. and per cent label
enrichment for the acids. prior to and just following
(Uo”C)threonine infusion. are summarized in Table 2.
ion intensity was observed for most organic acids at the
(Air 4]’ peaks. However. this intensity was not due to

’C enrichment. but resulted. for the most part. from the
natural isotopic contribution of silicon. from the TMS
derivatives of the organic acids. As expected. I’C
enrichment was observed only for 2-hydroxybutyric.
edeoxyerythronic and Ldeoxythreonic acids.

Threonine in plasma

The somewhat low enrichment of "C observed for the
urinary organic acids led us to suspect that the per cent
of labeled threonine in plasma might have been some-

 

 

 

 

——f v 1 V I v V ij 1 v v

60 260 460 660
Sean Miner
Figured. Reconstructadﬂlplototorganicaeidsaxcretedlnthe
diabeticdog urinecoilectedbetweanthazndanthlollowing
(U-"Ci-thraonine infusion. Ammonia Cl reagent gas pressure-
300 mTorr.

 

is and IM Unﬁn
ﬂ “l‘ '- I-Iw
‘ case '
has Org-h cars as. tat- a
t Unknown . 0.10 til
2 2W am I.
3 1W are are
a GM as as
I Duals us a;
O 1W an as
7 2-Aminobutyrts .1 u
. 3-WM 1‘ 0..
I Mule calms u u
t. Diabetic It .5 u
it Mathyt malonls m u
t! 2 Ketobutyrle oxima 0.0 u
'3 2 Mow-1mm are an
‘4 are an
is tDeoxyer-ythronie 0.0 1‘
t. C Deoxythreonle 0.10 2.3
W 3-Daoxytetronis us ass
i. 2 Deoxytetronie 0.0 0.01
'9 Moi: on as
as Unknown (moi. vet-12!) an 012
21 Erythronle o.“ 0”
21 Threonlc 0.” 0’
23 Tropic (internal standard) on 0."

 

what lower than originally anticipated. Analyses of the
isotopic enrichment of "C-labeled species of threonine
in plasma were therefore carried out on the No
trilluoracetyl-n-butyl esters of the plasma amino acids.
Literature El mass spectra of these derivatives showed
that relatively little molecular weight information is
available by this method."

Since the derivative is strongly electrophilic. it seemed ‘
reasonable that Cl would yield molecularotype ions by
electron capture NICI mass spectrometry
EC/ NlCl/ MS. EC/NICI/MS has been found to pro-
duce abundant molecular ions through the capture of
near-thermal-energy electrons." The MC! mass spec-
trum in the molecular ion region for the threonine deriva-
tive is shown in Fig. 5. The predominant ion in the

 

lit-HT

 

Reiotive abundance nu

 

zoo 30 sec
ml:

mtECINiCt/htsumemoiecuiarionragionotthaﬂo
triﬂuoroacetyt-n-butyiesterotthreoninairornapiasmaspeciman
colactediusttolowinoirmsionotiabelsdmraoninaintothedog
(reactant gaa.methana: pressure. In mTorr).

 

225 ._

 

 

 

 

 

8mm

Flaunt. Reconstructedlllplototthamaioraminoacidsloundh
thediabeticdogplnsme.byamilaalls;presaura-mm1‘ere.

spectrum resulted from protonation of the threonine
molecule. followed by elimination of HF. Very little
fragmentation was observed for the amino acid. Ion
pairs used for determining the I’C enrichment of the
plasma threonine pool were the [M]' and [M—HFT
species. occurring at ml: 367 and 371 and at m/z 347
and 351. respectively. Another adduct ion. at m/z 480.
corresponding to [M + CO,CF,]‘. also showed a labeled
species at run/1484. SIM analyses were performed on
all three sets of ions; the plasma threonine pool con-
tained 25.0:05‘5 of the ' .-labeled species (average
enrichment in the ,three ions).

To confirm the results obtained by EC/NlCI/MS, the
plasma amino acids were also analysed by ammonia Cl.
The reconstructed mass chromatogram for the major
amino acids found in the dog plasma is shown in Fig.
6. Ammonia CI on the amino acids produced atypical-
looking mass spectra. Due to the low proton alﬁnities
of the derivatized amino acids with respect to the
ammonia reactant ions. no protonated molecules were
observed. For threonine. the two ions dominating the
mass spectrum were the molecular adduct ions. in which
one and two ammonium ions were clectrophically
attached to the molecular species (Fig. 7). These ions,
and their isotopes at +4 u. were monitored by SIM. The
results of the ammonia CI experiments on threonine
were in agreement with the results obtained by
EC/NICI/MS. The amount of labeled threonine in
plasma was determined to be 24.6: 2.3%.

 

DISCUSSION

 

There are several reports concerning the excretion of
2-hydroxybutyric acid in the urine. In addition to those
conditions where large amounts of lactic acid are
excreted.’ 2-hydroxybutyric acid has been associated
with a rare methionine malabsorption syndrome called
Oasthouse disease" and has been induced in a 65-year-
old man by 3 days of total fasting (C. C. Sweeley and
S. Fajans. unpublished results).

The excretion of 4-deoxythreonic and 4—deoxyery-
threonic acids has not previously been ascribed to any
particular disorder. An obvious approach to determine
whether these metabolites are products of threonine
metabolism would involve the use of a stable isotope
tracer experiment in a Type 1 diabetic patient. However.

 

Nonstandard-team

 

 

all

ngrntAmmonlsamaasapeetnnneIthaN-Wn-
Weaterolthreenlnn.

due to the pool size ofthreonine in humans. the mom
of labeled threonine needed for veriﬁcation would far
surpass the economic feasibility of such an experiment.
and analysis of low levels of enrichment might be
dilﬁcult. Fortunately. we found that 4—deoxythreonic.
4-deoxyerythronic and 2-hydroxybutyric acids are
excreted by Golden Retriever dogs. In normal dogs.
their levels are low. but evidence suggests that the levels
of these metabolite are signiﬁcantly elevated in urine of
dogs with a disease that is clinically indistinguishable
from human Type i diabetes.“

As a result. we were able to introduce (”Cd-threonine
in viva and monitor the stable iosotpic enrichment of
the three organic acids. and others. Since it was postu-
lated that threonine derived from skeletal muscle protein
might be the primary source of these metabolites. promo-
tion of protein anabolism was made during the onset of
infusion. subjecting the dog to deliberate blood glucose
elevation by hypoinsulinization for 24h directly prior
to the experiment. Insulin injections were made immedi-
ately prior to the threonine infusion. so that blood
glucose levels could be restored to normal. and protein
anabolism would parallel the time period of maximum
levels of labeled threonine in the plasma. Finally. it was
hoped that insulin deprivation after the infusion might
induce protein catabolism and the excretion of labeled
metabolites from threonine.

Our results indicated enrichment of "C in 2-hydroxy-
butyric. 4~deoxythreonic and 4—deoxyerythronic acids.
but only in the initial few hours following infusion of
the labeled threonine. Thus. it cannot be claimed that
the threonine was derived from a protein pool. in which
a peak of "C incorporation should have resulted at later
times. However. the experimental evidence certainly
supports our hypothesis about the origin of these meta.
bolites. This is a signiﬁcant ﬁnding. and suggests the
need for further studies of the metabolism of threonine
in Type I diabetic patients.

Acknowledgements

ThistesearchwassupportedinpartbylesearchGrantRR-NMIo

theNIH/MSU MassSpearometryI-‘adlitbieauthorsaregratefnl
tothe Les AlamosStnble La forthegillef

borntory
["CJvthreonine; support by NIH(RR-022Jl) h acknowledged.

227

nunmmums

 

CMmIUﬂ4t
Mmuangsz
Owﬁhﬂﬂd

c

.munah
cunt

caMaa
ccﬁnx
an

an

Inn

vvmmnaawmmaus
rmnnn

' nuannn

«ﬁrmed-n

m Mm mm Mme“
m m m. m...
“a mum
m we .mmemm
. m. mama...
m we me..am.
. a“ mmmmwww
n ow at tn 3
s cm sm.m.4.-
wmmmmmm
umemmnuummuu

ppmmmﬂ

7

”are m
Hakka"
can

"‘

tlll‘ ‘

 

228

Revised MS. acarosaaz

U ULITY OF ION SOURCE PRET REA TMENT WITH
CHI. ORINE-CONTAINING COMPOUNDS FOR ENHANCED PERFORMANCE

IN GAS CHROMA TOGRAPHY/NEGA TIVE IONIZA TION MASS SPECTROMETRY

Daniel B. Kassel. Kathleen A. Kayganich. J. Throck Watson‘ . and John Allison.
Department ot Chemistry
Michigan State University

East Lansing. MI. 48824

'Also Department at Biochemistry

229

Abstract:

Dramatic variations in the electron capture negative ion (ECNI) mass spectra oi several classes at
compounds can occur 8 the ion source is exposed to CCI. prior to their analysis. This ion source pretreatmers
method results in enhancements in detectabmy tor several compounds over that achieved by conventional
ECNI analysts. Abundant pseudomolecular ions ol the type [M+Ctr are observed which are usetui as
molecular weight indicators tor unknowns. The [M+Ctl‘ ions termed in the ECNl analysis tollowing ion source
pretreatment with CCI. are not a result ol direct Cf attachment in the gas phase. but are presumed to occur
through surtace reactions. investigations into the ‘universaity‘ ol this method tor enriching ECNI mass spectra

are presented and possmle mechanisms lorlM+Cl1° ion tormation are discussed.

230

This article describes an unconventional approach to etlecllng electron capture negative ionization
(ECNI) mass spectrometry (MS) which has the potential tor enhanced perlormance in the analysis cl selected
classes ol compounds. For ECNI analysis. the analyte should have sufficiently high electrophiidty to capture
neanthermal electrons and produce intense negative ion currents(t.2). The extent to which a molecule wil
produce an appreciable negative ion signal depends on its electron capture cross-sectional). Electron capture
cross sections lor inherent molecules vary by many orders ol magnitude. in contrast to the cross-sections lor
positive ion tormation by electron impact ionization (El) which dtter little(4.5). and as a resul. extraordnary. but
variable. detection Emits olten are realized among these conpounds under condtions tor negative ion

lormation.

Derivatization ot an analyte may be pertormed to increase the cross section lor electron capture. Highly
ttuorinated derivatives. such as pentalluorobenzyi (PFB) and lrilluoroacetyl (TFA) derivatives are most usetui
tor improving the response or a compound in ECNI-MS. Dependng on the derivative chosen and the structure
ot the analyte. the types ol negative ions lormed may dller considerably. The PF 8 derivatives ol most organic
acids. tor example. typically produce the carboxylate anion as the major negative ion(6). in contrast. TFA
derivatives have been lound to produce. in addtion to molecular weight intormation (in the term ot M'. [M-Hl'.
or [M-HFT). abundant lragment ions tor several classes at compounds. Thus. the extent to which negative ion
mass spectra can be used tor proMng rnolecutar weight and structural intormation can vary signiﬁcantly

between compound classes and with the choice oi derivative.

in contrast to the ECNI experiment in which a gas such as methane is introduced into the ion source to
participate in the generation ot thermal electrons tor negative ion tormation. there is the technique or negative
chemical ionization (NCI)-MS in which the analyte is ionized by interaction with an anion(7.8). Methylene
chloride NCl-MS(9.10). lor example. has been shown to produce abundant molecular adduct ions. [M+Ct1'.
However. lew lragment ions are observed. limiting the utiity ot this method tor analyses requiring structural

inlormation.

We report here an approach to negative ion MS which yields substantial beneﬁts in the analysis at
certain classes at compounds. When conventional ECNI-MS is perlormed tollowing pretreatment oi the ion

23!

source with a chlorine-containing species such as crrzcr, or ccr.. substantial ditterences in the spectra oi
analytes are observed compared to those obtained with an untreated source. This method irequently yields an
abundant [M+Cil' ion. Furthermore. and more importantly. structurally-significau ions such as those seen in
ECNI are not lost at the expense oi this [their ion iormation. which contrasts with NCl mass spectra in tha
iragrnent ions are typically not observed. in addition. substantial enhancements in the response to ECNI-MS is
trequently observed lor selected compound classes lollowing this pretreatment procedrre. Examples oi
analyses that were signiﬁcantly improved by ion source pretreatment. and possible mechanisms ior iormation

oi the observed ions are presented here.

EXPERIMENTAL SECTION
Pretreatment Procedure: cm. was introduced into the mass spectrometer source at a pressure oi
approximately 0.1 torr tor 4-5 minutes. Methylene chloride and chloroiorm also were used successluly.
During treatment. the electron filament need not be 'on.‘ The terrperature oi the ion source was maintained

between 100‘ and 130°C.

Ms Conditions: All experiments were performed on an HP 598§ GC/MSIDS which was modified tor
the detection oi negative ions. The ion source temperature was maintained at two. ECNI experiments were
pertormed using methane at pressures on the order oi 0.4-0.5 torr. as measured at the ion source using a
Hastings gauge and a thermocouple gauge. A constant pressure oi methane was maintained throughout the
experiments using a Granville-Philips 216 Pressure/Flow controller and servo driven valve assembly.
Electron energies were 110 W. with emission currents oi 300M Tuning ol the instrumerl in the negative ion

mode was achieved using an equirnolar mixture oi pertiuorotributylamine and 00..

GO Condlrlons: An HP-S (cross-linked 5% phenyl methyl siﬁcone) 30m megabore capillary cohmn
(0.53mm id. x .88u.m ﬁlm thickness. Hewlett-Packard Co., Palo Alto. CA) was temperature programmed irom
70-300'6 at 15'C per minute tor all analyses. Helium ﬂow rates were approximately 5 miimin. A 40-cm piece

at deactivated vitreous siﬂca capillary tubing. maintained at 28$‘C. was used as the GCIMS interlace.

232

Derivatization Procedures: The N-trtiluoroacetyl n-butyl ester (TAB) derivatives oi amino acids were
prepared lollowing procedures outined by Dultield et attl 1). in which trilluoroacetic anhydride (TFAA) and 125
N HCt in nobutanoi were used. The trimethylsilyl (TMS) derivatives at organic acids were prepared lollowing
the procedures at Sweeley er aI.(12) Pentailuorobenzyt (Pi-’8) esters oi organic acids and TFA derivatives oi
alcohols were pmpared'according to procedures described in the 1986-87 Pierce Hanaooir and General
Catalog (p. 198401).

Reagents: Common reagents. such as organic acid. iatty add. and amino and standards were
obtained irom either Nutritional Biochemical Corporation. Cleveland. OH or sigma Cherrical Company. St.
Louis. MO. Carbon tetrachloride was purchased irom the Aldrich Chemical Company. Mﬂvraukee. WI. Ultra
high purity methane was purchased irom Matheson Gas Products. Chicago. L The derivatizing reagents.
trittuoroaceliC'anhydride and N.Gbis(trimeihyisilyi)iriliuoroacelamide (BSTFA) were obtained irom the Pierce

Chemical Company. Roclrlord. it. and Regis Chemical Company. Morton Grove. 1. respectively.

RESULTS AND DiSCUSSlON

The eilects 01 source pretreatment with CCI. were discovered alter investigating the utiity oi a
PFTBA/CCI‘ calibrant tor the production oi high- and low-mass negative ions (e.g.. mlz sea irom PFTBA and
W2 35 irom CCU. Results irom a nurrber oi analyses by conventional ECNI-MS lollowing exposure oi the ion
source to 004 were quite surprising. Mass spectra oi the trifluoroacetyl-n-butyt ester (TAB) oi aspartic acid
obtained under ECM (CH‘) condtions. without and with C01. source pretreatment. are shown in Figures 1a
and 1b. respectively. The only diterence in the two analyses was that the ion source had been exposed to
CCI. prior to obtaining the spectrum in Figure 1b. The negative ion mass spectrum in Figure 1a shows an [M-
H]' peak at M 340 oi relatively low intensity. comprising less than 4.0% at the total ion current (TIC - 8500
counts). Several other tragment ion peaks are observed. The base peak in the mass spectrum represents the
[M-NHZCOCFJ ion at mlz 22a.

The mass spectrum in Figure 1b was obtained irom the same quantity oi aspartic acidJAB as was used
tor Figure 1a; it shows some stnTringly dtterent leatures. The base peak in the mass spectrum is no longer at
mlz 228. but at nyz 376 which represents the [M+Cil' ion. This ion proves usetui as a molecular weight

 

f”

233

indicator. with the incorporation oi chlorine being apparerl due to the dstlnct Isot0pic peak intensity pattern. A
significant increase in Tic (to 80.000 counts irom 18.200 counts) is observed upon going irom Figure la to 1b.
There is an approximate S-told increase in the integrated area oi the TTC lor aspartic-TAB lollowing a briei
exposure ol the source to 001.. in Figure 1a. the ion which is most representative oi the intact compound is
the [M-Hl' ion. while in'F’rgure 1o. r is the [wort ion. Selected ion monitoring (SIM) oi°these two -
pseudomolecular ions shows an enhancement in detectabiity by a lactor oi 40 alter pretreatment oi the
source. Also note that the lragment ions and their relative abundances. observed in Figure 1a. are essentially
unchanged in Figure 1b. Modest changes in peak intensity ratios may be attributed to small ﬂuctuations in
source pressure irom one run to the next. Note also that absolute lragment ion abundances lollowing CCi.

pretreatment increase 1-5 told over those irom conventional ECNI. The results are summarized in Table i.

Another unique leature oi the spectrum in Figure 1b is that only a very smal Cl' peak is observed. in
tact. it the ion source is brieﬂy exposed to CCi.. and then allowed to stand tree oi iurther exposure lor one
hour. the ECNI-MS analysis at aspartic-TAB still shows a very intense [M+Ct)‘ peak. These results are also
summarized in Table 1. Alter this one-hour period. onty minor amounts at Ci‘ are observed. comprising
approximately 0.1% oi the Tic. yet the [M+Cl1' is still the most intense peak in the mass spectrum. in direct 01'
negative ionization (NCI). the base peak always resuls irom Ci' (at mlz 35). in addition. depletion in the Cf ion
abundance is observed as the analyte elutes irom the 60 column. indicating that the Cl’ ions in the gas phase
are directly involved in the iormation ol the [M+0i)‘ ion during NCt. In the data observed alter 001.

pretreatment. no depletion in the Cl“ ion abundance is observed as the analyte elutes irom the 60 column.

The results above suggest that the [M+Cli' ion and the other ions observed alter source pretreatmeri
(Figure 1b) are not lormed by direct or attachmentito.13). To investigate this iurther. the reagent gas was
changed irom methane to 00i.. creating conditions lor NCI with 01' as the chemical ionization reagent ion.
The NCl mass spectmm oi aspartic-TAB is shown in Figure 1c. There are obvious diilerences between the
spectra in Figures 1b and 1c. in the latter spectrum. no lragment ions are observed. Low-mass ions such as
Cl.’ and CCI3‘ are primary products oi electron attachment to 00i.. tn the Cf-NCI analysis. SiM shows that Ci’

ions are consumed as the derivatized amino acid elutes irom the column. This observation is in contrast to that -

234

described above tor the experiment lollowing source pretreatment; liars. lonnation oi (Meetr lollowing source
pretreatment apparently does not involve ioeroiecuie interactions in the ga's phase.

it was oi interest to determine whether the unusual leatures observed in the CCi.-source pretreatmera
ECNI mass spectra oi aspartic add~TAB would also be observed lor other compounds. The triiluoroacetyl
derivative oi myristyi alcohol. CH3(CH¢),2CHzOCOCF,. was analyzed by both conventional ECNI-MS and
ECNI-MS lollowing CCL-pretreatmeru oi the ion source. Although the TFA group in’parts some electrophilc
character to the compound. it does not signiﬁcantly increase the cross section tor electron attachmers to
produce a signal that is easily dscernible irom background under conventional ECNI condtions. In tact. the
only peak observed wel above backger was lor the OCOCFa' ion at mlz 113. This behavior is not unique;
trequently. negative ion mass spectra provide no molecular weight or structural iniormation. producing only
ions indicative oi the derivatizing agert. When this compound is analyzed lollowing the CChpretreatmerI
procedure. a substantial response is observed. The [M+0l)’ ion is represented by the base peak in the mass
spectrum. Flam 2. and tragment ions are still presera. Thus. tor some compounds thatrespond poorly in
ECNI-MS analyses. CCI.-pretreatment oi the source may lead to lower detection mils. and at least provide

molecular weight iniormation ior unknowns.

Pretreatmert oi the ion source with 001. does not sham attect the response oi a compound in ECNl-
MS. Figure 3 shows the negative ion mass spectrum oi maleic add (di-TMS ester) obtained lollowing the
CCh-pretreatment method in the conventional ECNI mass spectrum. the major peak represents the M' ion.
comprising over 95% oi the TlC. in the mass spectrum obtained lollowing pretreatment. the same situation is
observed. No [Mr-Cir ion is observed. prompting the question oi why this compound is unailected by the
source pretreatment that can lead to such dramatic changes lor other compounds. Maleic acid (di-TMS ester)
diiters irom the compounds that show good response to the pretreatment method in that it show an intense M'
peak in its ECNl spectnrm. while aspartic-TAB and the TFA derivative oi myristyi alcohol dqnot. These latter
compounds exhibit an [M-H)’ ion instead. This observation suggested a possible correlation between the
stabrTity oi the molecule as an intact anion (under conventional ECNI conditions). and the lack oi response to

source pretreatment (i.e., the lack oi a tendency to term [M+Cil‘ in a 00l.-pretreated ion source).

235

The TAB derivative ot the amino acid L«prolne (L-Pro-TAB) ls irieresting in that. under conversional
ECNI conrﬁtions. the conpound iorms both M’ and (MRI ions. Figures 4a and 4b show the ECNI mass
spectra 01 L-Pro-TAB under conventional conditions and lollowing 001. source pretreatment. respectively. The
response observed upon source pretreatment is intermedate to that observed in the extreme cases at
aspartic-TAB and maleic acid (ti-TMS ester). L-Pro-TAB shows approximately a Hold enhancement at ion
promotion under ECNl condtions lollowing source pretreatment (Figure 4p). in this case. the [M+Cll' ion is not
the base peak in the mass spectrum. but carries a smaller portion oi the T10 relative to the response lor
compounds such as aspartic acid~TAB.

Table 11 summarizes the responses at several derivatized compounds under condtions in which the
pretreatment procedure was employed. in comparison to responses observed in conventional ECNI analyses.
Compounds that respond simrTarty in both ECNI methods. i.e.. compounds that do not yield dilierent mass
spectra lollowing source pretreatmert. are those that produce abundant M‘ ions in their negative ion mass
spectra. in contrast. compounds that typically exhibit a small (or no) M’ peak also exhibit enhanced responses

lollowing source pretreatment with 001..

Table 111 summarizes the responses oi a number oi compounds to ECNI with CCI.-pretreatment. While
a tavorable response does not require that the compounds be as highly lunctionalized as those presented. it
appears that at least one site oi unsaturation is required. Compounds such as dodecane are essentialy
unresponsive under ECNI-MS with or without CCI. pretreatment. Ketones. even as simple as acetone. do

produce abundant [M+Cll' ions. with inproved detection Emits lollowing the pretreatment procedure.

Duration of “pretreatment“ enact: The resuhs presented above were obtained irom the analysis oi
pure oonpounds. This prompts the mestion 01 whether the CCI.-pretreatment method is suited tor complex
mixture analysis. Experience with analyses oi sinple mixtures and tor sequential iniections 01 simple mixtures.
shows that the signal enhancements and the iormation 01 [M +01) ions described here wil occur lor at least one
hour lollowing source pretreatment. without completely depicting the source oi chlorine atoms. Figure 5 shows
the results ol such a study. The ion source was pretreated. and thecompound aspartic-TAB was iniected

several times over a one-hour period. The experiment was repeated at various ion source temperatures. The

236

data show that the chlorine in the source persists lor over an hour. being lost more rapidly at higher source
temperatures. Figure 5 shows that the response will decrease by a lactor at approximately two over a one
hour time period. Thus. I this technique is to be used tor lengthy GCIMS mixture analyses. the ﬂute
dependence must be wel understood tor the particular instnrment being used. One option that could be used
to maintain the response throughout a mixture analysis is to constantly provide a smal amount at 001. to the
ion source (lollowing pretreatment. as the analysis is being penonned). This could be done by using a
CCI./CH. nix. instead oi pure 0H. as the reagent gas. However. care must be taken to supply a constant.
but exceecﬁngly small amount oi 001.. it this is not done. the concentration oi anions such as 0013' may

dominate the chemistry in the ion source. instead oi the chemistry described in this work.

Mechanism oi [Moot] ’ iormation: Having iuled out a conventional chemical ionization (N01)
mechanism lor the iormation ol [M+Cl)‘ irom a 001.- pretreated ion source. i.e.. a gas phase ion/molecule

mechanism ol the type
Cf+M+A-+(M+CII+A

(where Asthird body). mechanisms in which the incogpgratigg ol Cl and the 'Qnization process occur in two
separate steps must be considered. Reactions such as

M' + Cim.” —s [M+Cl‘l'

where C'iﬂ-S) represents some torm oi chlorine either in the gas phase or on a surlace. can be dsmissed since
there is no evidence tor a sufficiently longived M' species lor most ol the molecules investigated. The most
reasonable mechanisms involve incorporation oi 01 into the molecule lolloww by electron attachmera.

presumably to the resulting radical. The incorporation oi 01 could occur on a suriace or in the gas phase.

Any proposed mechanism must be consistent with the experimental observations. Those considered to
be relevant here include: 1. [M+Cl)‘ is observed when the Cl' ion abundance is very small; 2. The [M+CI)‘

ions are observed tor more than an hour alter source pretreatment; 3. There is a temperature dependence to

 

 

 

237

this behavior. with the chlorine being depleted irom the ion source more rapidly at higher temperatures; 4. An
increase in the 01’ peak intensity is actually observed when some compounds elute lrom the 60 into the ion
source; 5. At least one site 01 unsaturation is required ior [M+01)’ ion iormation to occur. 0. The eilect is most
dramatic lor compounds that do not exhibit an M‘ ion under conventional ECNI conditions.

The experimental observations are consistent with storage oi chlorine in some lorm on the inner walls at
the ion source. The binding oi halogen-containing compounds to metal surlaces has been reported in the
literature. tor iron and aluminum surlacestte.15). in these studies. the suriace population oi halogen atoms
can be decreased by heating. The tact that the chlorine remains tor long periods 01 time. and the apparent
(ﬁspiacement 01 01 lrom the suriace by some compounds such as benzene is clear evidence that the halogen
atoms are stored on the suriace. The question remains whether an analyte molecule collides with the suriace.
incorporates a 01 and desorbs; or whether a gas phase 01 atom bonds to the analyte. in either case. a gas

phase [M+Cll- radcai would be lorrned. which then presumably captures an electron.

Both mechanisms may be operative. Since Cl atoms appear to be stored on the surlace. molecules that
colide with the source walls could react. On the other hand. the 01 present in the source does disappear with
time. and. thus. must go into the gas phase in an atomic or molecular lorm at some point Could the source
have a substantial concentration 01 01 atoms present in the gas phase lollowing pretreatment? One might
assume that this would 92) be the case due to the negligible signal due to Cl' at mlz 35. However. the cross
section lor electron capture tor an atomic species should be low since collisional stabilization would be

required; this phenomenon would explain the low signal at miz 35.

A stronger argument against the involvement oi tree 01(9) atoms is supported by the tact that the gas

phase reaction (1)
Cl- + CH4 —) HCI + CH3- (1)

has been observed to occur with a rate constant 01 k - 3 x 10"8 cm’ molecule" sec"(16). Thus. at the
pressures oi CH. used in the ion source. 01 atoms (rt any were present) would be consumed rapidly. The 01

then would be converted into H01. which does not capture thermal electrons(17). in addition to these

238

considerations. Grimsrud e1 al.(18) have developed a detailed kinetic model lor the high pressure electron
capture ion source. with an extensive dscussion on the possible origin oi unusual and unexpected ions such
as those which we observe here. Their work strongly suggests that suriace-assisted reactions may support It.
behaviorthatwe reportwhileagas phase radcal mechanismmaynot.

Further insights into the mechanism may be gained by considering the thermodynamics ol the iormation
or an [M+011- radcal in the gas phase and irom a surlace. For the reaction (2)

"(91 + C'tst -* Malta) 121

to occur in which the Cl is available on the suriace. there are two thermochemical considerations: the energy
required to cleave the Cisuriace bond. and the energy involved in lorming the [M+Ctj- radical (a process which

could be endothermic or exothermic). in contrast. it the reaction occurs in the gas phase. reaction (3).

“(9) + 0.to) -r Mint (3)

the breaking oi the Cl-suriace bond need not be involved in the overall reaction; i.e.. the iormation oi 0119)
must occur spontaneously beiore the reaction occurs. Consider a simple molecule with one site oi

unsaturation. lormaldehyde (Clip). 11 a 01- is added to this molecule. a 0-01 or an 001 bond may be termed.
CI- + crrzo —rCI-CH2-O° (4)

-i CHrO-Cl (5)

Are reactions (4) and (5) endothennic or exothermic? The Al-ifs oi 01(9) and Cit-10(9) are 29.1 kcanol
artd -28 kcal/moi. respectiyely. Using group equivalence tables(19). the radical product oi reaction (4) has a
AH, oi approximately ~63 kcal/mot. The same approach gives a value oi ~32 kcaiimol tor the AH. oi the
radical product oi reaction (5). Using AHKCHZClOI-t) - -49.4 kcanoi. and assuming that the HO bond energy
is 100 kcanol. AH. 01 the product oi reaction (4) would be kcalimoi. Such numbers suggest that both
reactions (4) and (5) would be exothermic. but by less than 10 kcaiimol. Thus. they could occur spontaneously

239

in the gas phase. However. ler these reactions to occur on a suriace in an exothemic process. the Cl-surtace
interaction would have to be less than 10 kcal/moi. which is rather weak. (However. I Cl atoms can desorb
irom the suriace thermaly. this may relied a weak Cl-surlace bond.) Therelere. one cannot rule out either
mechanism based on these considerations tor lormaldehyde. 11 similar calculations are pertormed tor acetone.

the situation becomes more complex. We estimate that reactions (6) and (7)
Ci° + (CH3); 0:0 -a (CHgﬁCDC-O (6)
-e C(CHﬂg '0-0 m

are both endgthermig by approximately +40 kcal/moi. Thus. it M energy is required in a suriace reaction to
cleave a suriace—0t bond. reactions (6) and (7) should not occur. These calculations led to two additional
possrble explanations lor the observed results. The ﬁrst is that energy may be releasﬁ when a 01- is
abstracted irom the suriace. For example. it Sis a metal atom or atoms on the surlace. and spades ol the

type 5001. are on the suriace. the suriace chlorination reaction may be written as

soot. i M —-e 5:001. + M0l- (3)

Reaction (8) may be exothermic tor cases such as when M is acetone. due to the strength 01 the 8-0 bond
that is lorrned. it has been reported that the triple bond in acetylene can be cleaved when this alkyne bonds to
nickel and iron surlaces at temperatures as low as 400 K(20.21). This suggests that the metal suriace - 0
bond energies can be substantial. Thus. reactions such as (8) could be 931% due to the iormation on
carbene- or carbyne-lilre species. it reaction (8) represents the pathway by which chlorination oi the analyte
occurs. the reaction may lead to various S-CCl. species on the suriace. I may be expected that such a
reaction could resul in a modlied surlace. However. it has been reported that when 00l. is adsorbed on an
iron suriace and the suriace chlorine coverage is depleted. the carbon that remains appears to ditluse into the

bulk metal at temperatures as low as 444 K. leaving the suriace clean(14).

240

A second possible explanation tor the observed towns is that suriace chlorination does not lead to a
radical at all. but to a dchlorlnated product. We estimate that reaction (9). in which two chbrlne atoms are
added to acetone is exothermic.

zei- +ICHalzC-0-iiCHslth'lC'0-Cl (91

We note that the ECNI mass spectra of many chlorinated compounds show an [M-Cl)’ ion as the peak at
highest miz. thus the neutral that is lorrned on the source walls (M013) may not actually be observed-onty the
the fragment corresponding to loss et 01. However. it seems unikety. tor the variety 01 compounds studied in
this work. that if an M01; species were being formed. it would never be observed. Also. no such species are
seen when the source is chlorinated and positive E1 spectra are obtained. The important conclusion is that the
Cl is stored on the source walls. and is available for addition to unsaturated compounds tor a period ot time that

is potentially useful in negative ionization mass spectrometry.

Addtional experimental results support the suriace mechanism. Figure 6a shows the reconstructed TIC
profile for serine-TAB with data obtained with an ion source that had been cleaned. and was free trem any
chlorine contamination. Representation ol the chromatography is typical and acceptable. Following source
exposure to 001.. reconstruction ot the profile from newly acquired data shows a dramatically ditterent peak
shape (Figure 6b). Alter baking the ion source for one hour. the sample was reanalyzed; reconstmction of the
original peak shape from newly acquired data (Figure 6a) was achieved. Thus. degradation of the peak profile
shown in Figure 6b is not attributable to any chromatographic ettect. but may reflect an intermediate chem'cat
step. i.e.. conversion of M to [MsClia The taiing may occur due to the reaction of. and subsequera slow
desorption of. species from the source was. Similar behavior has been observed in thernionic 60 detectors.

in which degradation of peak profiles can occur due to surface reactions within the detecter(22.23).

Once the [M+Cl)- radical is termed. electron capture occurs. if the source is pretreated with 001.. and
positive ion (El) mass spectra are obtained from these compounds. no evidence for such a species is
observed. The implication is that only a small fraction of the analyte uhdergees Cl-attachment. Thus. the

resulting radical must have a very large electron capture cross section. For molecules such as maleic acid (d-

 

241

TMS ester). the Cl attachment may still occur. however. upon electron attachrrlera. the 011s lost to term the
stable M' spades. reaction (10).

[M+Cil- + e‘ —e {Meal} -9 M’ 4» Cl- (10)

For molecules that have oonpgated double bonds such as 2-buten-3one. the lonnation of the stable [M-Hl' ion

dominates ever the [M+01)' ion.

Since only a smal traction ol the analyte is converted into the (Micq- radcai species. the unreacted
molecules M can still react by electron capture to term the lragment ions observed in conventional ECNI. For
some molecules. the ion current corresponding to the fragment ions B enhanced with their relative abundance
distribution being retained. This may suggest a second pathway tor formation of lragment ions via the [M+Ci)"
intermedate and the stable [M-H)‘ ion as lollows:

[Ms-CW +0. _g [MM

 

 

_, tutor)“ —-) [M-H)’+ HCl (1 l)

——) fragment ions

Thereiore. based on these pretminary results. when an ion source is exposed to chlorine-containing
compounds. such as 001.. 0H013. or CHchz. the source walls become covered with 01 in some form The
chlorine slowly deserbs. On the suriace. unsaturated molecules can react to lorm (M+0l)- radcais. These
radcals capture thermal electrons to form predominantly the [Mr-011' anion. The exceptions are molecules
whose molecular anions are stable. Further experiments are underway to provide other insights into the
mechanism. it should be noted that. in ECNI mass spectrometric studes. there have been other reports :
suggesting that a variety ot species can be stored on ion source walls for reaction with analyte molecules. The
capture of H(18.24) and C(25) atoms by molecules has been implied by ECNI experiments. in which suriace
chemistry has been suggested prior to the electron-capture step. Thus. such studies may lead to a variety of

methods in which species with high electron capture cross sections may be generated in an ion source.

242

CONCLUSIONS

An understanding oi the mechanism through which [Mi-01f ions are termed in ECNI studes in an Ion 1
source pretreated with 001. is imported. since it may lead to the developmerl ot new methodology.
Meanwhile. empirical analytical advantages can be reaized from this techrique. For many molearles ECNI-
MS with a pretreated ion source may give improved detection Imits. Also. when unknowns are detected in 00-
ECNl-MS. it can be difficul to derive molecular weight lnfenmtion because the peak a the highest miz could
represent (MHFT. [M+H)‘. or some other lragment anion. Such questions frequently can be resolved by the
chlorine pretreatment method because the [M+0l)' ion. 11 termed. ls easyte recognize from the isotope peak

intensity pattern.

with the large number oi ionization methods available for the iormation of positive and negative ions
from analytes. one can. in many cases. provide the necessary qualitative and quantitative information for
selected components oi a mixture by combining the resuus of a few approaches (such as E1 and N01). The
method presented here makes the ECNI methodology responsive to a larger number oi compound classes.
which may be advantageous in the analysis of mixtures when conventional ECNI is the ideal method for only

some 01 the components of interest.

10.
11.
12.
13.
14.
15.
16.

17.
18.
19.

20.
21 .
22.
23.

24.

243

literature Clled

Hunt. 0. P.; Crow. F. W. Anal. Chem. 1978. 50. 1781.

Jennings. K. R. 'Mass Spectrometry: Vol. 4. RAW. Johnstone. Ed The Chenical Society. Durington
House. London. Chapter 9. 1977.

Petlzzarl. E. D. .L‘Chromalogr. 1974. 98. 329.

Harrison. A. 6. 'Chemical ionization Mass Spectrometry: CRC Press. inc. Boca Raton. Florida. Chm.
2. 1988.

Christophorou. L. 6. Chem. Rev. 1976. 76. 409.

F. K. Kawahara. Anal. Chem. 1968. 40. 1009 and 2073.

Roy. T. A.; Field. F. H.; Lin. Y. Y.; Smith. 1.. L. Anal. Chem 1979. 51. 272.
Schlunegger. U. P.; Lauber. R. Org. Mass Spectrom. 1986. 21. 401.
Stemmler. E. A.; Hires. a. A Anal. Chem. 1985.57.684.

Tannenbaum. H. P.; Roberts. J. D.; Daugherty. R. C. Anal. Chem 1975. 47. 49.
Kingston. E. 8.; Dultield. A. M. Biomed. Environ. Mass Spectrom. 1982. 9. 459.
Gales. S. C.; Oendramis. N.; Sweeley. c. C. Clin. Chem. 1978. 24. 1674.
Daugherty. n. 0. Anal. Chem. 1981. 53. 626A.

Jones. R. 6. Surface Science 1979. 88. 387.

Dowben. P. A.; Jones. R. 6. Surface Science 1979. 89. 114.

Kondratiev. v. N. 'Rate Constants oi Gas Phase Reactions.‘ US. Department oi Commerce. NBS.
Oitice oi Std. Rel. Data. 1972.

Adams. N. 6.; Snilh. D. J. J. Chem Phys 1986. 84. 6728.
Sears, L. J.; Campbel. J. A: Grimsmd. E. P. Blamed. Environ. Mass Spectrom. 1987. (in press).

Franklin. J. L; Dillard. J. 6.; Rosenstock. H. M.; Draxi. K.; Field. F. H. Nat. Stand. Rel. Data Sen. Nat.
Bur. Stand. 1969. 26.

Lehwald. 3.: lbach. H. Surface Science 1979. 89. 425.

Erley. w.; Baro. A. M; lbach. H. Surface Science 1982. 120. 273.
Bombick. D. 0. Ph. D. Thesis. Michigan State Uriversity 1986.

Burger. B. V.; Prelorius. P. J. J. Chromatogr. Sci. 1987. 25. 118.
McEwen. c. N.; and Rudat. M. A. J. Amer. Chem. Soc. 1979. _101. 6470.

244

25. Buchanan. l4. V. and Rubin. l. 8. 350: Am Conlerence on Mass Spectrometry and Allied Topics.
Denver. CO 1987. Presentation moerooo.

m

This work was pertormed with instruments in the MSU Mass Spectrometry Faciity supported by a, gram
(5P4t-DRR-00480-19) irom the Division oi Research Resources oi MR and various units oi Michigan State
University.

245

Acknowledgements:

The authors are grateiul to Proi. Eric Grlmsrud tor his suggestions regarding this work and tor making
ret. 18 available to us during the review oi. this manuscrbt.

_'_

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

246

Legends

e) Conventional ECNI mass spectmm 200 ng oi aspartic acid-TAB (methane
pressure equal to 0.4 ton). Base peak at mlz 228 corresponds to (ll-H-
NHCOCFJ. integrated peak area (TIC) - 18.200.

6) ECNI mass spectrum at the same conpound lollowing ion source
preatreatment with CCI.. Base peak at m 376 corresponds to (Mr-Ctr.
integrated peak area (TIC) - 80.000.

c) NCt mass spectrum oi the same compound (CCt‘ reagent gas. pressure - 0.4
torn. (M+Ci)‘ base peak at mlz 376. integrated peak area (110) - 18.300.

Conventional ECNI mass spectmm oi the tn‘iluoroacetyl derivative oi myristyi
alcohol (MW . 310) lollowing ion source pretreatment with CCt.. The base pad:
is (Mr-Ctr. observed at mlz 345; rnlz 213 corresponds to (M-COCFJ; mlz 113
corresponds to (COzCFg'.

ECNI mass spectmm oi the maleic acid (di-TMS ester) lollowing CCI.
pretreatment (methane pressure - 0.4 tort). No (Mi-Ctr ion is observed;an 260
corresponds to the stable molecular anion. (M)'.

a) Conventional ECNI mass spectmm oi L-Pro-TAB under conditions in which
the ion source was tree irom CCI. exposure. Both (M)' and (M-H)’ are observed
at mlz 267 and 266. respectively. The lollowing ions have been identified: mlz
211. (M-C.H.)'; mlz 191. [M-C.H.-HF]; mlz 1&3. (M-C.HgCO;H-HJ; mlz 143.
[M'CeHeiozﬂ‘Hz‘HFn-

b) ECNI mass spectmm oi the same compound lollowing ion source
preatreatment with CCI.. The only new ion observed in the mass spectrum is
lound at M 302. correspondng to the (M+Ci)’ ion. An approximate 4-iold
enhancement in detectability is observed

A measure oi chlorine depletion over time tor various ion source terrperatures is
shown As the peak elutes. the integrated areas oi the (Mi-Ctr ion. Amer.
relative to the integrated total ion current. Am. is shown as a tunction oi time.
intensities oi the (M+Cl)’ ion tor aspartic acid-TAB decrease over time. and the
intensity oi signal decreases more rapidly at high source temperatures.

a) Reconstructed TlC prolile tor serine-TAB generated irom conventional ECNI
analysis. The prolile represents acceptable chromatographic behavior.

b) Reconstructed TlC proﬁle tor serine-TAB. same sample size. lollowing ion
source preatreatment with CCl‘. Surtace interactions result in severe tailing and
poor chromatographic peak shape.

 

247

TABLE 1: ECNI results lor aspartic acid-TAB obtained prior to and alter CCi4 source
pretreatment.

(a) Can ventionai EC-Nl (i.e.. without CCl. exposure)

W amaze: mm W
133 (M-cozc.H.-c.ri.r 55.4 1100 2350
22s (M-H-COCFaNH)’ 100.0 . 2000 5270
239 (M-H-COzC‘i-igl' 76.3 1520 3450 -
267 (M-H-OC4H9)‘ 23.3 470 1000
321 (M-HF‘)’ 3.7 170 470
340 (M 411'. 183 370 800
NC 8500 18200

(b) EC-Nl results directly lollowing CCI. exposure

mlz (Assg’ nmem 25 2! mi: - 228 (en Abundance inimram Am
35 (00’ 36.1 1650 -
183 (M - CO;C4H. - C4Hg)‘ 90.3 3750 8550
228 (M - H - COCFaNHr 100.0 4160 8950
239 (M - H - C0204Hgl' 89.1 3700 8450
267 (M ~ H - cotter 26.2 1100 2550
321 (M - HF)‘ 6.2 260 700
340 (M - H" 18.0 750 1580
376 (M + Ci)’ 400. 16600 41600
no 34300 80000

(c) EC-Nl results lollowing a period Of) 1 hr lollowing Ion source exposed to CCI.

mlz (Assg' nmem 96 Q! mlz . 228 ion Abundag intmrated Area

35 (Cir 0.6 25 -
183 (M - CO;C.H9 - C‘Hg)’ 38.2 1420 3000
228 (M - H - COCFgNl-O' 100.0 3920 8990
239 (M - H - CO;C4H,)‘ 70.4 2780 6980
267 (M - H - OC4H.)' 39.3 1540 3650
321 (M - HF)‘ 8.0 310 790
340 (M o i-i)’ 19.8 780 1470
376 (M + Cl] 158. 6180 12500

TIC 21500 49100

248

TABLE 2: Comparison ol response lor several analytes by El. ECM (conventional).
and ECNI (source pretreatment).
mm ’W W W
Valne - TAB 0.18 1.9 10.9
Asparllc . TAB 0.34 3.3 0.0
Myristyl - TFA 0.002 0.1 see
Glutanic - TAB 0.94 4.0 4.2
L-Proine - TAB 0.03 0.1 4.2
Cystine - TAB 19.5 25.4 1.3
Serine - TAB 9.1 12.2 1.3
Maleic (di-TMS) 7.6 8.4 1.1
Laurie - PFB 16.1 16.4 1.0

t A 3 integrated areas

249

Table 3: Survey 0! the ellecl ion source pretreatment has on various compound
classes.
929mm mm 'Emaeeem
Organic Acids (TMS):
- 2-hydroxybutyrtc [ M - TMS l' 4
- Oxalic (M | ° 0
. - Maleic (M )' 0
~ Male [M - TMS] ' 4
- Adipic [M 4 Ci] ' 4
- M-hydroxybenzoic (M 4 Cl] ' 4
- P-hydroxybenzoic [M . TMS] ‘ 4
Organic Acids (PFB esters) ( M - PFB 1' 4
Fatty Acids (methyl esters) (M 4 Ct) ' 4
Abohols (TFA ethers) i M 4 or) - 4
Anino Adds (TAB esters) Varies (see Table 2)
Heptane nitrite (M 4 CI] ' 4
Acetone [ M 4 Ci] ' 4
Pyridine [M 4 Ct] ‘ 4
3-butene-2-one ’ i M . H] ' o
Anisoie N008 0
p—Xyiene (1.1 . H] . o
Dodecane None 0
Benzene ( M - H] ’ 4

i (+) indicates an enhancement in detectability using our method over conventional ECNI-MS. (0) indicates
no enhancement in detectability using the CCI. source pretreatment ECNI-MS method. ‘

250

"Gill! la-lc

 

228
A 239

A

A

A

RﬂATiVE ABWAHCE

 

0 O
. 13'?" «34.-k
H8c4° "Ngc'. “.04.
O
’67 'ill-lil’

 

 

 

)-
F
hp
i
r;
.i
Ii.—
2
l-
at
ti
d
. —-i
j?—

 

 

 

 

 

 

 

 

 

 

200 240 230 320 360 400
ml!
8 ("WW
w
”-1
z
<
O 4
g
a
<‘
:5: ‘ 223
2 239,
‘5 ‘ 267 tn-rtl'
A111 LTﬁﬁﬁ IT‘TlITYI Ia
Tj i i i
200 240 230 320 360 400
ml:
i C (HOCID‘
'13
37
O 4
g
0
<4
5 .
3
a " \
1
FT111113‘AI ##7## 11%er
200 240 280 .120 360 400
ml:

 

 

 

251

RELATIVE ABUNDANCE

 

 

~40ch N

 

 

 

 

 

 

 

.. 540:1
J Anownmu v1
l 0 “U
1 A31. :1
)1 «i id 1? «tribe rﬁ i «i .1- . u q 4 d? a Lu - P- T
_ _ _ a _
5o .8 mac woo woo

FIGURE 3

252

 

(M)-

 

 

 

 

F

r I i I l

33NVON09V JAIIVH

I

260

220

i80

140

mlz

253

FIGURE 4

 

 

163

 

(ii-H)"

 

139

A

 

 

 

‘1‘

n&

muz<ozao< m>C<da

60 100 140 180 220 260 300 340

20

mlz

 

(H+C|)'

 

163

 

 

 

 

139

B

 

 

J _ q q q - 1 a

uuz<ozao< m>_._.<._um

60 too 140 180 220 260 300 340

20

mlz

 

 
  

 

om __ om o... 0.4. _ o
.e .= 4
4 ..
.2... V
. v
0.08 a. . uv w
0.3 o _m
4 r- _
a 4 0.83. x
.. w.
ins»
rom

m. 55lo

 

255

FIGURE 6

. A i no

 

 

 

 

 

 

I I ‘I I ‘l I 1 U
60 80 100 120 I40 160

 

11C

 

 

 

 

1 —T I 1 Y 1 r l r V U T I U T '7 U U I t I

60 80 ICC 120 140 i60
SCAN

APPENDIX II

256

This appendix (Appendix II) gives molecular weight information on some organic
acids observed in the methylmalonic acidemia disease condition for which no
molecular weight information has previously been reported. Analyses by GC-only
have veriﬁed the presence of many of these unknowns and their retention times and
indices are documented. However. these species have yet to be identiﬁed from E1
and/or NH3-CI analysis.

Some of the unknowns whose molecular weights have been determined may be
useful for disease-state identiﬁcation since some are observed at concentrations
signiﬁcantly elevated to many of the metabolites identiﬁed for this disease-state. The
presence of some of these unknowns at such elevated levels may suggest certain
metabolic errors and perhaps with identiﬁcation of these unknowns, a better
understanding of the disease condition will be obtained. In addition, the
determination of the molecular weights for these unknowns might also be useful for
obtaining "true" simulated molecular weight proﬁles of complex mixtures of organic
acids which will be useful for comparison to those molecular weight proﬁles that are
obtained experimentally utilizing the K+IDS technique.

The derivatized and underivatized molecular weights of a number of unknown
organic acids (whose concentrations are such that they gave rise to ion signals by GC-
MS analysis) and identiﬁed organic acids isolated from a methylmalonic acidemia
urine sample are reported in Table 1. Retention indices and times for both identiﬁed
organic acids and the unknowns have been obtained from triplicate GC analyses using
a DB-l 50 meter narrow bore capillary column. Concentrations of unknowns and
identiﬁed organic acids have been obtained based on calculation of creatinine levels

in the subject patient and use of the internal standard, trOpic acid.

257

Table 1. Molecular weight determinations on a number of unknowns
whose concentrations are above the detection limits of the
HPS985 quadrupole GC-MS system.

96mm WWmm‘WW

Unk 924.9 936.76 9.458 15.53 331 115
Lactic 1012.60 10.218 236.12 234 90
Glycolic 1055.00 10.642 125.35 ' 220 76
Unk 1094.0 1091.50 11.007 146.70 232 88
Glyoxylic oxime 1217. 19 12.264 28.70 233 89
2-hydroxybutyric 1217.19 12.2641 28.70 248 104
Oxalic 1231.69 12.409 114.22 234 90
3-hydroxypropionic 1261.90 12.712 49.79 234 90
3-hydroxyisobuty1‘ic 1268. 18 12.775 52.41 138 104
Pyruvic oxime 1288.91 12.983 22.06 185 ..
3-hydroxybuty1'ic 1313.14 13.226 34.76 248 104
Sulfuric 1320.02 13.295 20.02 242 98
3-OH-2-Me-butyric 1424.30 14.341 15.56 262 1 18
3-hdyroxyisovaleric 1457.40 14.673 46.92 262 l 18
Mcthylmalonic 1478.61 14.886 920.69 262 1 l8
Unk 1568.4 1569.93 15,802 81.05 278 134

Phosphoric 1660.56 16.707 278.59 314 98

Succinic

L-Glyceric
4-deoxyerythronic
Unk 1881.9
4-deoxythreonic
2-OH-2-Me-malonic

Glutaric

3-deoxytetronic
2-deoxytetronic
Unk 2168.4

2-ketoocatanoic

Malic

Unk 2256.1

Adipic

Salicylic
Pyroglutamic
Methyladipic
Unk 2384.0
Erythronic
Threonic

2-hydroxyg1utaric

Tropic:

1757.90

1827.54
1868.50

1882.82

1896.00
1944.68
2008.77
2051 .89
2105.29

2169.69

2221.69
2248. 10

2255.30

2277.30
2293.40
2327.30
2354.80

2386.00

2415.30
2458.89
2471.80
2501.39

3-OH-phenylacetic 2537.87

4-OH-phenylacetic 2610.23

258

17.679
18.376
18.786
18.924
19.051
19.541
20.186
20.620
21.154
2 l .798
22.318
22.582
22.654
22.874
23.035
23.374
23.649
23.961
24.254
24.690
24.820
25.1 18
25.482
26.204

101.30
300.46
32.08
10.04
38.60
23.15
15.30
26.55
161.43
200.50
20.49
72.73
50.94
34.16
18.99
21.44
21.15
26.66
284.53
441.67
32.47
38.46
40.83
36.18

262 118
322 106
336 120
350 134
336 120
350 134
276 132
336 120
336 120
364 148
302 158
350 134
260 116
290 146
282 138
273 119
304 160

did not observe

424 136
424 136
364 148
310 166
296 152
296 152

 

ﬂ

Unk X 2615 2614.64
2-deoxyribonic 2689.90
Unk 2707.60 2708.23
Vanillic 2904.60
Gentisic

2945.61
m-methoxycinnamic
Ribonic 2956.14
Arabinonic 2990.43
Hippuric
Citric 3060.61

Isocitric

259

26.248
26.999
27.183
29.154

29.563

29.668
30.010

30.710

35.18
69.21
41.82
110.76

124.44

91.84
66.90

551.87

378
438
311

312
370

220
526

526
323
480
480

162
150
167

168
154

148
166

166
179
192
192

10.

11.

12.

13.

14.

260

REFERENCES

Williams, RJ. Biochemical Institute Studies IV. Individual Metabolic Patterns
and Human Disease, Univ. of Texas Publication No.5109, Univ.of Texas, Austin,
TX 1951, Ch.1.

Holmes, LC. and Morrel, F.A. Applied Spectrosc. 11. 86 (1957).

Homing, E.C.and Homing, M.G. Clin. Chem. 11, 802 (1971).

Holland, J.F., Leary, 1.]. and Sweeley, C.C. J. Chromatogr. Suppl. Vol. 52 3- 26
(1986).

Homing, EC. and Homing, M.G. Clin. Chem. 11, 802 (1971).

Tuchman, M., Bowers, L.D., Fregien, K.D., Crippin, P.J., and Krivit, W. J.
Chromatogr. Sci. 22, 198 (1984).

Sweeley, C.C.; Holland, J.F.; Towson, D.S. and Chamberlin, B.A. J.
Chromatogr . 322, 173 (1987).

Gates, S.C., Smisko, M.L., Ashendel, C.L., Young, N.D., Holland, J.F., and
Sweeley, C.C. Anal. Chem . 59, 433 (1978).

Hagenfeldt, L. and Hagenfeldt, K. Clin. Chim. Acta _4_2_, 219 (1972).

Nicholls, T. Hahnel, R., Wilkinson, S. and Wysoclci, S. Clin. Chim. Acta 62
(1976).

Waterbury, L.D. and Pearce, L.A. Clin. Chem. 18, 259 (1972).

Ward, M.E., Politzer, I.R., Laseter, LL. and Alam, S.Q. Biomed. Mass
Spectrom . 3, 77 (1976).

Brooks, J.B., Kellogg, D.S., Alley, C.C., Short, H.B., Handsﬁeld, H.H. and Huff,
B. J. Infect. Dis. _1__2_9, 660 (1974).

Chalmers, R.A. J. Inher. Metab. Dis. Vol, 1 Suppl. 7 , (1984).

15.

16.

17.

18.

19.

20.

21.

22.

23.

24.

25.

26.

27.

28.

29.

30.

31.

32.

261

Matsumoto,1. Kuhara, T. Mass Spectrom. Rev . 6, 77 (1987).
Tuchman, M. and Ulstrom, R.A. Adv. Pediatr . 32, 469 (1985).

Chalmers, R.A. and Lawson, AM. "Organic Acids in Man," University Press,
Inc., Cambridge, @ 1982.

Harrison, A.G.; Jones, E.G.; Gupta, SK. and Nagy, G.P. Can. J. Chem . 431,
1967 (1966).

Munson, M.S.B. and Field, F.H. J. Amer. Chem. Soc. 38, 2621 (1966).
Levsen, K. "Selected Topics in Mass Spectrometry" 45, 897 (1983).
Beckey, H.D. Int. J. Mass Spectrom. and Jon Processes 2, 883 (1969).

Beuhler, R. J., Flanigan, E., Greene, L.J., and Friedman, L. J. Amer. Chem.
Soc . 26, 3990 (1976).

Barber,M.; Bordoli, R.S.; Sedgwick, RD. and Tyler, A.N. J. Chem Soc. Chem.
Commun .323 (1981).

Roepstorff, P.; erup, P. and Moller, J. Biomed. Mass Spectrom.12, 181 (1985).

Boyle, P.D.; Johnson, BI. and Alexander, B.D. J. Amer. Chem. Soc . 26. 1346
(1987).

Rollgen, F.W. and Schulten, H.-R. Z. Natwforsch 3Q, 1685 (1975).

Rollgen, F.W. and Schulten, H.-R. Org. Mass Spectrom . J_Q, 660(1976).
Giessmann, U. and Rollgen, F.W. Org. Mass Spectrom . 12, 539 (1977).

Stoll, R. and Rollgen, F.W. Org. Mass Spectrom . _l_6, 72 (1981).

Ackermann, B.L.; Watson, J.T. and Holland, 1.1:. Anal. Chem . 51, 2656 (1985).
Caprioli, R.M.; Fan, T. and Cotrell, J.S. Anal. Chem . 38, 2949 (1986).

Tsarbopoulos, A.; Liberato, DJ. and Jardine, 1. 35th ASMS Conference on
Mass Spectrom. and Allied Topics, Denver, CO pp.80—81.

33.

34.

35.

36.

37.

38.

39.

41.

42.

43.

45.

46.

47.

48.

49.

50.

262
McNeal, CJ. and Macfarlane, R.D. J. Amer. Chem. Soc . 103. 1609 (1981).
Blewett, LP. and Jones, 13.1. J. Phys. Rev. 5_Q, 464 (1936).

Melton, C. B. "Principles of Mass Spectrometry and Negative Ions", Marcel
Dekker, NY, pp. 68 (1970).

Kelsoo, J.F., Pantano, CG. and Garofalini, S.H. Surf. Sci . 1%, L543 (1983).
Bombick, D.D. Doctoral Dissertation, Michigan State University (1986).
Bombick, D.D., Pinkston, JD. and Allison, J. Anal. Chem . 5A, 396 (1984).
Day, R.J.; Unger, SE. and Cooks, R.G. Anal. Chem . 32, 458 (1987).
Bombick, DD. and Allison, J. Anal. Chem . 32, 458 (1987).

Nakata, H., Hoshino, Y., Takeda, N. and Tatematsu, A. Organic Mass
Spectrom. 20 (7), 467 (1985).

Schmelzeisen-Redeker, G.; Geismann, U. and Rollgen, F.W. Org. Mass
Spectrom . 29, 303 (1985).

Goodman, 8.1. and Markey, S.P., "Diagnosis of the Organic Acidemias by Gas
Chromatography-Mass Spectrometry" Liss, Inc. NY (1981).

Van Biervliet, J.P.G.M.; Bruinvis, L., VanderHeiden, C; Ketting, D.;Wadman,
S.K., Willems, IL. and Monnens, L.A.H., Dev. Med. Child Neurol. 12, 392
(1977).

Page, T. and Nyhan, W.L., Int. J. Biochem. 12, 713 (1981).

Rhead, W. and Tanaka, K., Proc. Nat'l. Acad. Sci. 71, 480 (1980).

Dancis, J. and Levitz, M. "The Metabolic Basis of Inherited Disease" McGraw-
Hill, Co., NY pp. 397-410 (1978).

Gates, SC. and Sweeley, C.C. Clin. Chem. 10, 1663 (1978).

Bencsath, F.A. and Freld, F.H., 35th ASMS Conference on Mass Spectrom. and
Allied Topics , Denver, CO pp. 29-30 (1987).

Isacchar, D. and Yinon, J. Biomed. Mass Spectrom. 6, 47 (1979).

51.

52.

53.

54.

55.

56.

57.

58.

59.

61.

62.

63.

65.

66.

67.

68.

263

Iles, R.A.; Hind, A.J. and Chalmers, R.A., Clin. Chem. 22, 1795 (1985).
Shackleton, G.J.L. and Mitchell, F.L., Steroids m, 359 (1967).

Hunt, D.F.; Giordani, A.B.; Rhodes, G. and herold, D.A. Clin. Chem. 28, 2387
(1982).

Gates, S.C., Dendramis, N. and Sweeley, C.C., Clin. Chem. 26. 1674 (1978).
Chamberlin, B.A. and Sweeley, C.C. Clin. Chem. 3314,, 572 (1987).
Lin, S-T and Klabunde, KJ. Langmuir 1, 600 (1985).

Munster, H., Budzikiewicz, H. and Schroder, E. Org. Mass Spectrom . 22, 384
(1987).

Light, K.L. unpublished results.

Jocelyn-Pare, J.R.; Lafontaine, P.; Belanger, J.; St, W—W.; Jordan, N. and Loo,
J.C. J. Pharm. Biomed. Anal. 5,, 131 (1987).

Komori, T.; Kawasaki, T. and Schulten, H-R. Mass Spectrom. Rev . ;4_, 255
(1985).

Schulten, H—R.; Komori, T. and Kawasaki, T. Tetrahedron 33, 2595 (1977).

Games, D.E.; McDowall, M.A.; Levson, K.; Schafer, K.H.; Dobberstein, P. and
Gower, J .L. Biomed. Mass Spectrom. 11, 87 (1984).

Shomo, R.E.; Chandrasekaran, A.; Marshall, A.G.; Reuning, RH. and
Robertson, L.W.Biomed. Environ. Mass Spectrom. ﬂ, 0 (1987).

Costello, C. 35th ASMS Conference on Mass Spectrom. and Allied Topics ,
Denver, CO (1987), pp.xxx.

Light. K.L.; Kassel, DB. and Allison, J. in preparation.
Martin, S.A. and Biemann, K. Mass SpectromReviews 6, 1(1987).
Rose, K.; Baroch, A. and Offord, R.E. J. Chromatogr . 268, 197 (1983).

McLafferty, F.W. "MassSpectrometry-MassSpectrometry”, Elseveir

69.

70.

71.

72.

73.

74.

75.

76.

77.

78.

79.

80.

81.

82.

83.

84.

85.

86.

264
Publishing Co., NY e 1983.

Roepstorff, P. and Fohlman, J. Biomed. Mass Spectrom. 11, 601 (1984).

Stults, J.T.; Martin, 8.; Biemann, K. and Watson, J.T. Biomed. Environ. Mass
Spectrom. (in press).

Munster, H.; Theobald, F.; Budzikiewicz, H. and Schroder, EJnt'l. J. Mass
Spectrom. and Ion Processes 12, 73 (1987).

Musselman, B.D.; Cheng, CJ. and Watson, J.T. Anal. Chem . 6, 1277 (1985).

Dempster. A.J. Phil Mag. 31, 438 (1916); Smyth, H.D. Phys. Rev. _2_5_, 452
(1925).

Munson, M.S.B. and Field, F.H. J. Amer. Chem. Soc. 88, 2621 (1966).

Aue, DH. and Bowers, M.T. "Gas Phase Ion Chemistry", M.T. Bowers, Ed.,
Academic Press, New York, 1979.

Bohme, D.K., Mackay, 6.1., and Schiff, H.1. J. Chem. Phys. 23, 4976 (1980).
McMahon, TB. and Kebarle, P. Can. J. Chem . 63 3160 (1985).

Harrison, A.G. in "Chemical Ionization Mass Spectrometry," CRC Press, Inc. ©
1982.

Klotz, C.E. "Fundamental Processes in Radiation Chemistry," P. Ausloos, Ed.,
John Wiley and Sons, Inc. New York, pp. 40 (1968).

McLafferty, F.W. in ”Interpretation of Mass Spectrometry," 3rd Edition, ©
1983, pp. xxx. , .

Clow, RP. and Futrell, J .H. J. Amer. Chem. Soc . 24, 3748 (1972).
Munson, M.S.B. and Field, F.H. J. Amer. Chem. Soc. 82, 1047 (I967).
lchikawa, H. and Harrison, A.G. Org. Mass Spectrom. _13, 389 (197 8).
Clemens, D. and Munson, B. Org. Mass Spectrom . 20, 368 (1985).
Cairns, T. and Siegmund, E. Org. Mass Spectrom . 2Q, 58 (1985).

Budzikiewicz, H.; Laufenberg, G.; and Brauner, A. Org. Mass Spectrom . 2Q, 65

 

87.

88.

89.

91.

92.

93.

94.

95.

96.

97.

98.

99.

100.

101.

102.

103.

104.

105.

265
(1985).
Hunt, D.F. "Finnigan Spectra " 6, (1) (1976).
Houle, F.A. and Beauchamp. J.L. J. Am. Chem. Soc. 101 4067 (1979).
Mcloughlin, R.G. and Traeger, J .C. J. Am. Chem. Soc. 101 5791 (1979).
Bohme, D.K. and Fehsenﬁeld, F.C. Can. J. Chem . ﬂ, 2715 (1969).
Keough, T. and DeStefano, A.J. Org. Mass Spectrom . 16, 527 (1981).
Westmore, J.B. and Alauddin, M.M. Mass Spectrom. Rev . 3, 381 (1986).
Lin, L-Y and Smith Biomed. Environ. Mass Spectrom. 3, 604 (1978).
Nakata, 1. Org. Mass Spectrom .20, 467 (1985).
Kebarle, P. Ann. Rev. Phys. Chem . 28 445, (1978).
Hatch, F. and Munson, B. Anal. Chem . 42, 731 (197.7).

Hatch, F. and Munson, B. J. Phys. Chem . _82, 2362 (1978).

Carroll, T.X., Smith, SR. and Thomas, T.D. J. Amer. Chem. Soc . 28, 659

(1975).

Mills, B.E., Martin, R.L., and Shirley, D.A. J. Amer. Chem. Soc . 23, 2380

(1976).
Benoit, EM. and Harrison, A.G. J. Amer. Chem. Soc. 22, 3980 (1977).

Iivistendahl, G. and Uggerud, E. Org. Mass Spectrom. 20, 541 (1985).

Karabatsos, G.J.; Sonnichsen, G.C>; Papaioannou, C.G.; Scheppele, SE. and

Shone, R.L. J. Amer. Chem. Soc. 32, 463 (1967).

Millard, B.J. "Quantitative Mass Spectrometry, " Heyden and Son, London

(1978).
Munson, B. Anal. Chem. 32 772A (1977).

Langevin, P.M. Ann. Chim. Phys . 3, 245 (1905).

106.

107.

108.

109.

110.

111.

112.

113.

114.

115.

116.

117.

118.

119.

1 20.

121.

266

Henglein, A. "Molecular Beams and Reaction Kinetics," Schlier, C. ed.,
Academic Press, New York, 1970.

Moran, T.F. and Hamill, W.H. J. Chem. Phys . 32, 1413 (1963).

Su, T. and Bowers, M.T. Int. J. Mass Spectrom. and Ion Processes 12, 347
(1973).

Field, F.H. ; Franklin, J.L.; and Lampe, F.W. J. Amer. Chem. Soc. 12, 2419
(1957).

Miles, J.M.; Haymone, M.; Nissen, s. and Gerich, '1. J. Clin. Invest. 11_, 1554
(1983).

Mayes, P. and Felts, J. Biochem. J . 162, 230 (1967).
Page, T. and Nyhan, W.L. Int. J. Biochem. 12, 713 (1987).

Johansson, 15.; Ohlsson, A.; Lindgren, J.B.; Agurell, S.; Gillespie, H. and
Hollister, L.E. Biomed. Environ. Mass Spectrom. 14, 495 (1987).

Young, J.W.; Veenhuizen, 1.1.; Russell, R.W. Fed. Proceed. 46, 295 (1987).

Samuelsson, B.; Hamberg, M. and Sweeley, C.C. Anal. Biochem. 33, 301
(1970).

Mamer, D.A.; Laschic, NS. and Scriver, C.R. Biomed. Environ. Mass
Spectrom . L3, 553 (1986).

Clutter, W.E., Bier, D.M., Shah, SD, and Cryer, P.E. J. Clin. Invest. 66, 94
(1980). "

Wolfe, R.R., Wolfe, M.H. and Shulman, G. In "Stable Isotopes in Nutrition",
Tumlund, JR. and Johnson, P.E., Eds. ACS Symposium Series pp.17S (1984).

Wong, W.W., Cabrera, M.P., and Klein, P.D. Anal. Chem . 36, 1852 (1984).
Mathews, D.E., Ben-Galim, E. and Bier, D.M. Anal. Chem . 5_1, 80 (1979).

Irving, C.S., Thomas, M.R., Malphus, R.W., Marks, 1... Wong, W.W., Boutton,
T.W. and Klein, P.D. J. Clin. Invest . 21, 1321 (1986).

122.

123.

124.

125.

l 26.

127.

128.

129.

130.

131.

132.

133.

134.

135.

136.

137.

138.

267

Weinhouse, 8.; Medes, G. and Floyd, N.F. J. Biol. Chem. 166, 691 (1946).

Murrat, S.; Waddell, K.A. and Davies, D.S. Biomed. Environ. Mass Spectrom.
8 (10), 500 (1981).

Keller, U., Cherrington, A. and Liljenquist, J. Amer. J. Physiol. B, 238 (1978).
Balasse, E. Metabolism 28, 41 (1979).

Miles, J.M., Haymond, M., Nissen, S., and Gerich, J. J. Clin. Invest. 11, 1554
(1983).

Penman, AD. and Wright, LA. Biomed. Environ. Mass Spectrom . 1A, 339
(1987).

Schoeller, DA. and Klein, P.D. Biomed. Environ. Mass Spectrom. 3, 29 (1979).

Hachey, D.L., Wong, W.W., Boutton, T.W. and Klein, P.D. Mass Spectrom.
Rev. 6, 289 (1987).

White, E.V., Welch, M.J. and Sun, T. Biomed. Mass Spectrom. 2, 395 (1982).
Siekmann, L. J. Clin. Chem. Clin. Biochem . 23, 129 (1985).

Patterson, D.G., Patterson, M.B., and Culbreth, P.H. Clin.Chem. 30, 619 (1984).
Takatsu, A. and Nishi, S. Clin. Chem. 33, 1113 (1987).

Stellaard, F., Sackmann, M., Berr, F. and Paumgartner, G. Biomed. Environ.
Mass Spectrom. 1A, 609 (1987).

Hofmann, A.F. and Hoffman, N.E. Gastroenterology 61, 314 (1974).

Hofmann, A.F. and Cummings, S.A. in "Bile Acids in Gastroenterology", ed.
Barbara, L., Dowling, R.H., Hofmann, A.F. and Roda, E. pp.75 MTP Press,
Lancaster, En gland( 1983).

Boulton, A.A. and Davis, B.A. Biomed. Environ. Mass Spectrom . 11, 207
(1987).

Noren, B., Stromberg, S., Ericsson, 0., Olsson L-I. and Moses, P. Biomed. Mass
Spectrom . 12, 367 (1985).

139.

140.

141.

142.

143.

144.

145.

146.

147.

148.

149.

150.

151.

152.

153.

268
Shigematsu, Y.; Sweeley, C.C; Schall, W.D. and Gossain, V. (submitted).

Robinson, B.H. and Sherwood, W.G. J. Inher. Metab. Dis. 1, Suppl. 1, 69
(1984).

Blass, J.D. "Inherited Disorders of Carbohydrate Metabolism," Burman, D.;
Holton, J.B.; Pennock, C.A., Eds. MTP Press Ltd. Lancaster, pp. 238-68 (1980).

Kapetanovic, I.M.; Yonekawa, W.D. and Kupferberg, H]. J. Chromatogr . 11g,
265 (1987).

Kalhan, S.C.; Ricanati, E.S.; Tsemg, K-Y and Savin, S.M. Metabolism 32, 1155
(1983).

Yudkoff, M.; Nissim, I.M.; Schneider, A. and Segal, S. Metabolism 3Q, 1096
(1981).

Ford, G.C.; Cheng, K.N. and Halliday, D. Biomed. Environ. Mass Spectrom . 12
(8), 432 (1985).

Lapidot, A.; Kalderon, B.; Korrnan, S.; and Gutman, A. Bookof Abstracts, Int'l.
Sympos. Appl. Mass Spectrom, Health Sci., Barcelona, Spain p. 151 (1987).

Robinson, B.H., MacMillam, H., Petroru-Benedict, R. and Sherwood, W.G. J.
PM. 111, 525 (1987).

Markey, S.P.; Urban, W.G. and Levine, S.P. "Mass Spectra of Compounds of
Biological Interest," US. Atomic Energy Commission, Vol. 2 Parts 1 and 2, ©
1982.

Oberholzer, V.G.; Wood, C.B.S.; Palmer, T. and Harrison, B.M. Clin. Chim.
Acta 62, 299 (1975). °

Schoots, AC and Leclercq Biomed. Mass Spectrom. 6, 210 (1979).

DeJongh, D.C.; Radford, T.; Hribar, J.D.; Hanessiam, S.; Bieber, M.; Dawson,
G. and Sweeley, C.C J. Amer. Chem Soc . 21, 1728 (1969).

Antoshechkin, A.G.; Tatur, V.Yu; Perevezentseva, OM. and Maximova, L.A.
Anal. Biochem . (in press), 1987.

Dzidic, 1., Caroll, D.l., Stillwell, RN. and Homing, E.C. Anal. Chem . 41, 49
(1975).

154.

155.

156.

157.

158.

159.

160.

161.

162.

269

Hunt, D.F.; Stafford, G.C.; Crow, F.W. and Russell, J.W. Anal. Chem. 18, 2098
(1976).

Smith, A.L.C. and Field, F.H. J. Amer. Chem. Soc . 22, 6471 (1977).

Berger, B.A. and Dougherty, R.C. Biomed. Mass Spectrom. 3, 238 (1981).
Dougherty, R.C.; Dalton, J. and Biros, F.J. Org. Mass Spectrom . 6, 1171 (1972).
Dougherty, R.C. Anal. Chem . 33, 625A (1981).

Stemmler, B.A. and Hites, R.A. Anal. Chem . 31, 684 (1985).

Hunt, D.F.; Stafford, CC. and Crow, F.W. Anal. Chem. 62, (1977).

Hunt, DP. and Crow, F.W. Anal. Chem . 3Q, 1781 (1978).

Kingston, Eli. and Dufﬁeld, A.M. Biomed. Mass Spectrom.2 (11), 459 (1982).