Concepts for the determination of prostaglandins by tandem mass spectrometry

MISSING IMAGE

Material Information

Title:
Concepts for the determination of prostaglandins by tandem mass spectrometry
Physical Description:
viii, 176 leaves : ill. ; 28 cm.
Language:
English
Creator:
Gillespie, Todd Allen, 1962-
Publication Date:

Subjects

Subjects / Keywords:
Prostaglandins   ( lcsh )
Mass spectrometry   ( lcsh )
Chemistry thesis Ph. D
Dissertations, Academic -- Chemistry -- UF
Genre:
bibliography   ( marcgt )
non-fiction   ( marcgt )

Notes

Thesis:
Thesis (Ph. D.)--University of Florida, 1989.
Bibliography:
Includes bibliographical references (leaves 169-175).
Statement of Responsibility:
Todd Allen Gillespie.
General Note:
Typescript.
General Note:
Vita.

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 001479628
oclc - 21049910
notis - AGZ1586
sobekcm - AA00004803_00001
System ID:
AA00004803:00001

Full Text















CONCEPTS FOR THE DETERMINATION OF PROSTAGLANDINS
BY TANDEM MASS SPECTROMETRY



By

TODD ALLEN GILLESPIE


A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY





UNIVERSITY OF FLORIDA


1989




































To my loving wife, Paula















ACKNOWLEDGEMENTS


I would like to express my sincere gratitude to Dr. Richard A. Yost

for his guidance, direction, and friendship during this research. His

editorial assistance during the preparation of this dissertation and

various papers is very much appreciated.

I would like to express my gratitude to Dr. Joe Neu in the

Department of Neonatology at the University of Florida for his initiation

of this project and provision of supplies for the scintillation counting

work. Also I would like to express my sincere heartfelt thanks to Dr. Jim

Vrbanac, who, while at the Medical University of South Carolina in

Charleston, supplied samples, immunoaffinity gel, and expert advice as

well as a valued friendship during this collaborative research. In

addition, I thank Merrell Dow Research Institute for their support of this

work.

I acknowledge the members of my research committee, Drs. John G.

Dorsey, Anna Brajter-Toth, Samuel 0. Colgate and Joe Neu for their various

contributions to my thesis work and education at the University of

Florida.

This research would not have been possible or as much fun without

the support of and discussions with my friends and co-workers in the Yost

research group. I would especially like to thank Mark Hail, David

Berberich and Jodie Johnson for many helpful discussions about this work.









In addition to the people mentioned above, I would like to thank Steve

Brooks, Mark Barnes and Jim Michels for their friendship.

I would particularly like to thank my parents, during all the years

of my education whether in or out of the classroom, for their endless

support. Most of all, I thank my wonderful, caring wife Paula, for her

constant love, understanding and patience throughout my years in graduate

school. She has made this work all possible and worthwhile.
















TABLE OF CONTENTS

Page

ACKNOWLEDGEMENTS............................................... iii

ABSTRACT..................................... ................ vii

CHAPTERS

1 INTRODUCTION........................................ 1

Arachidonic Acid Metabolites
(Prostaglandins)................................... 1
Recent Analytical Advances......................... 13
Strategies for Mixture Analysis by MS/MS............ 15
Overview of Thesis Organization .................... 20

2 SAMPLE PREPARATION STUDIES......................... 22

Introduction........................................ 22
Concepts for Solid-Phase Extraction................ 23
Concepts for Derivatization........................ 33
Experimental........................................ 35
Results and Discussion............................. 43
Conclusions......................................... 52

3 OPTIMIZATION OF GC/MS AND GC/MS/MS CONDITIONS
FOR TRACE DETERMINATION OF PROSTAGLANDINS........ 54

Introduction........................................ 54
Experimental........................................ 54
Mass Spectrometry (GC/MS).......................... 56
Tandem Mass Spectrometry (GC/MS/MS)................. 64
Conclusions......................................... 80

4 DIFFERENCES IN THE COLLISIONALLY ACTIVATED
DISSOCIATION OF STRUCTURALLY SIMILAR
PROSTAGLANDINS ................................... 83

Introduction....................................... 83
Experimental........................................ 85
Efficiency Calculations............................ 88
Collision Energy Study of the [MO/TMS-PFB]"
Carboxylate Anions................................ 89









CHAPTERS Page

Collision Pressure Study of the
[MO/TMS-PFB]- Carboxylate Anions................. 93
Collision Pressure Study of the [M-PFB]'
Carboxylate Anions ............................... 95
Collision Pressure Study of the [M-H]-
Carboxylate Anions ............................... 98
Conclusions...................................... .. 100

5 EVALUATION OF SOLID-PHASE EXTRACTION GC/MS
AND GC/MS/MS FOR THE ANALYSIS OF ENDOGENOUS
PROSTAGLANDIN E2 IN URINE........................ 101

Introduction....................................... 101
Experimental ....................................... 103
Results from the Quantitation Study
of PGE2 in Urine ................................. 114
Trade-offs in the Steps of
the Analytical Procedure......................... 121
Conclusions....................................... 138

6 EVALUATION OF SOLID-PHASE EXTRACTION
PROBE/MS/MS FOR THE ANALYSIS OF ENDOGENOUS
PROSTAGLANDIN E2 IN URINE....................... 142

Introduction....................................... 142
Experimental ....................................... 143
Solids Probe Analysis ............................. 145
Direct Chemical Ionization Analysis................ 153
Conclusions......................................... 164

7 CONCLUSIONS AND SUGGESTIONS FOR FUTURE WORK........ 165

Summary............................................ 165
Future Directions.................................. 167


LITERATURE CITED................................................ 169

BIOGRAPHICAL SKETCH............................................ 176















Abstract of Dissertation Presented to the Graduate School of the
University of Florida in Partial Fulfillment of the Requirements for
the Degree of Doctor of Philosophy






CONCEPTS FOR THE DETERMINATION OF PROSTAGLANDINS
BY TANDEM MASS SPECTROMETRY


By

Todd Allen Gillespie

May, 1989


Chairman: Richard A. Yost
Major Department: Chemistry

An evaluation of the concepts for the trace determination of

prostaglandins (PGs) by tandem mass spectrometry (MS/MS) has been

achieved. Results from this work demonstrate the importance of the

optimization of various parameters in the collisionally-activated

dissociation (CAD) process before performing trace analysis. Dramatic

differences in the optimum collision gas pressure for selected-reaction

monitoring (SRM) with MS/MS for the determination of prostaglandins E2 and

F2, were observed. The differences in fragmentation behavior were examined

through the use of fragmentation, collection and overall CAD efficiency

studies. This work shows that the CAD efficiency for the derivatized and

underivatized carboxylate anions is significantly different for subtle









structural changes in PGs. A possible explanation has been proposed to

explain these dramatic differences.

The advantages and limitations of immunoaffinity purification (IA) for

sample preparation of PGs in urine have been investigated. Results show

that IA purification coupled with a short 3 m GC capillary column

utilizing electron-capture negative chemical ionization (EC-NCI) SRM can

provide a selective, sensitive and rapid method of analysis for endogenous

levels of PGE2 in urine.

A systematic study was performed demonstrating the relative trade-offs

which exist throughout the entire analytical procedure. Eleven different

analytical schemes were systematically evaluated for the trade-offs in

sensitivity, selectivity, and total time of analysis. These trade-offs

are discussed in relation to how they affect the three basic steps (sample

preparation, sample introduction and mass spectrometric detection) of PG

analysis. This study indicates that the utilization of a more selective

sample preparation method (e.g., IA) with MS/MS can reduce the

chromatographic separation time required to achieve the necessary

selectivity and sensitivity for PG analysis in urine. However, results

show that MS/MS is not necessary if IA purification and a longer

chromatographic separation (more selective) technique are employed. This

systematic study should be applicable in the evaluation of any analytical

procedure for analysis of components in a biological sample. In addition,

future work is proposed which should further enhance PG analysis by MS/MS.


viii















CHAPTER 1

INTRODUCTION



Arachidonic Acid Metabolites (Prostaglandins)


The enzymatic oxidation of arachidonic acid (AA) leads to a

multitude of biochemically important products (1). Among these substances

are prostaglandins (PGs), thromboxanes (TXs) and leukotrienes (LTs).

Collectively, these compounds are referred to as eicosanoids and

constitute what is known as the arachidonic acid cascade (Figure 1-1).

Many of these oxygen-containing metabolites have interesting and diverse

pharmacological properties and significant medicinal potential (2).

Since, the initial description of PGs in 1935, a vast body of

knowledge has accumulated on their physiology and chemistry (1,2).

Recently, attention has been focused on PGs of certain series as antitumor

agents (3). Evidence indicates that PGs such as prostaglandin E2 (PGE2),

play an important role as local mediators and modulators of renal blood

flow and excretory functions (4). It has been suggested that most of the

primary PGs found in urine are derived from renal production (5); con-

sequently, urinary levels of PGs have been applied as an index of renal

PG activity in numerous studies. Recently, PGE2 has seen application in

the induction of labor, softening of the cervix and prevention and

treatment of stress ulcers (1). While much has been learned about PGs













2



C








MO M" "^ J -- 194w"-I
n-oo ma no rc
e..Sc(~d~A~ nn-~ (~~Z~t~rra~C sS -SI

(11tP)-)l-AIr)( .O-y4a & (et 0 0or SwIi Co I


l..140taOwcieMn0fldra 4 1.4-4Mc


8Cz~Za
IiaPaO~ n.


I -


li-d....OI J.0.p ~ofP. *4-0



fd-, a.4fl4480.a. A. -- taco .
I W ..a..A.. L...
( rMreso~olp emessness E, r




co coo rxo mre emoneeA,






0
040050,054
,* 0 0 0 F,,
turr --~* r /S "--- ~^gw CO. TU ^ ~ -




coon r^Y-^-oon |
N^^^Ya^^^ TOr uwMlounfsa' c

mi, ~ ~ ~ ~ U~ o nll>,J C
.kwaun~lai .,


4,


2$ '0*c400 ISO.5~.( 2SI3 41-IOniL







44io~d*141.ti40,.
Lii.40.4..w .04umor


04SI~04

S*2T4404G *4?.*--iS
144ag4*ef.t~c a 4-an..


LT. er a 00a*A I -O sdwP tlls
/^-^J<( *CCCX) j^^^^ OM4







Om~nr~Cocoasl


L C
p wr-c-CO CCOOO
iS 20mb l~S.2S)S12-l




ccrwocw *co04
/'ess-'L^CO
i^L .*..a.l.uW Cw



cCt,-ie*coo
d-*40 1104 -Mr.C

LYC if mfn.oS C.
"w,-c-canC00oo4





CO.



+
H^d-CM-C004


i -
LT.OlwhflnMC


Figure 1-1: Arachidonic acid cascade


eD
ltt









3

biological effects and how they relate to mammalian health, difficulty in

measuring low concentrations of PGs in biological systems has hindered

the progress of research.

The analysis of PGs and other AA metabolites can be divided into

three basic steps; sample preparation, sample introduction and measurement

by such techniques as bioassay, radioimmunoassay (RIA), high-performance

liquid chromatography (HPLC), gas chromatography (GC) and gas

chromatography/mass spectrometry (GC/MS). It has been shown that with

many of the analytical methods frequently used, PG concentrations are

often overestimated by as much as a factor of ten (6). This problem can

be traced to the lack of selectivity of the entire analytical scheme used

for analysis. Therefore, the development of an analytical scheme which

provides for accurate, sensitive and selective determination of PGs is

needed.

In this chapter, a brief review of the most frequently employed

techniques for sample preparation and measurement of PGs will be

discussed. More thorough reviews of sample preparation and measurement

techniques can be found in the literature (7,8). Recently, a review on

AA metabolism with examples of various analysis methods has been published

(9). Other, more specific reviews have been written by Traitler (10) and

Kelly (11) on mass spectrometric analysis methods of eicosanoids.


Sample Preparation Techniques for Prostaglandins

Sample preparation before the measurement step is extremely

important, with the extent of extraction and purification dramatically

affecting the validity of the data. In sample preparation of biological









4

fluids, a traditional technique that is commonly utilized is liquid-

liquid solvent extraction (7). This method is time-consuming and usually

yields only 80% to 90% recovery of most PGs (12). The degree of clean-up

provided by such extractions is limited; further purification is often

necessary for body fluids such as urine.

Another popular method for separating PGs from biological matrix

components is solid-phase extraction. Three types of solid-phase

extraction are commonly employed: (1) amberlite (XAD-2) column; (2)

octadecylsilyl (ODS) column (C18); and (3) selective packing materials,

such as immunoantibodies. Bradlow (13) described the use of an XAD-2

column that is advantageous when the biological matrix contains a large

concentration of proteins. Recoveries using this method have been

reported as about 90%. Both solvent extraction and XAD-2 resin extraction

procedures are time-consuming and require evaporation of relatively large

volumes of organic solvents. Moreover, they are not very selective and

give extractions containing extraneous material that must be removed

subsequently by various chromatographic purifications. A variety of

methods using a C18 column to extract PGs from biological samples have

been developed by Powell (8,14) and other researchers (15,16). The solid-

phase extraction using a C18 column is rapid, efficient and more selective

than solvent and XAD-2 extractions. Recovery with this method has been

reported to be greater than 90% in many cases.

Specialized packing materials can provide significantly more

selective extraction of specific targeted PGs. A phenylboronic acid

column has been used to selectively isolate thromboxane B2 (TXB2) and its

metabolites (17). The recovery of radio-labeled TXB2 after extraction was











reported at 90%. Potentially, an even more selective approach to sample

preparation is to combine the extraction and purification steps into one

procedure. This has been accomplished by using an antibody-mediated

extraction procedure developed by Krause et al. (18). Basically, the

prostaglandin-antibody was coupled to cyanogen bromide-activated Sepharose

4B and used as a stationary phase for the extraction of PG from the

samples. The antibody was coupled to Sepharose and packed into a Pasteur

pipette. The plasma samples were then applied to the gel in the column.

This one-step extraction-purification method has shown improved

specificity and sensitivity. Similar methods have also been employed by

Hubbard (19) and Vrbanac (20) for the analysis of TXB2 and 6-keto-PGF1. in

urine. Another approach to exploit the high selectivity of antibody-

antigen reactions for sample extraction is double-antibody precipitation.

This technique has been used for preparation of plasma samples before HPLC

analysis for (15R)-15 methyl-PGE2 (21).

In summary, the advantages of extraction (either solvent or solid-

phase) are: (1) it eliminates some extraneous material, thereby imparting

greater specificity to the assay; and (2) it improves sensitivity of the

analysis by concentrating the material. The disadvantages are: (1) it is

time-consuming; (2) a "carry-over" of non-eluted analyte may occur (if the

same column is reused) effecting the validity of subsequent assays: and

(3) the extraction efficiency of the procedure is variable. Simple

solvent or solid-phase extraction has been shown to yield samples that do

not permit accurate validity of PG quantitation. Improved validity of

subsequent quantitation has been observed after further purification









6

steps. A more detailed explanation of solid-phase extraction and

immunoaffinity (IA) purification can be found in Chapter 2.

Three types of chromatographic purification are described below: (1)

silica acid column chromatography; (2) thin-layer chromatography (TLC);

and (3) HPLC. Group separation of PGs and related compounds is

conveniently performed by silica acid column chromatography (22,23).

Recoveries were reported for this purification method of 85% to 90%. PGs

separated by silica acid chromatography usually require further

purification by TLC or HPLC prior to quantitation by RIA or GC/MS.

Separation of PGs by TLC was first investigated by Green and

Samuelsson (24). TLC is the most commonly used method for separation of

PGs because of its efficiency, simplicity and economy compared to other

chromatographic procedures. The major groups of PGs (A, B, D, E, F and

6-keto-PGF1.) were readily separated on a silica gel plate using various

solvent mixtures (25-27). The disadvantages of TLC are its low recovery

yields (typically 80%) and the lengthy procedures required for separation

of closely related compounds. Prostaglandin-related compounds with

similar behavior are often observed to migrate in a similar way even in

different solvent systems. Such problems can be avoided by using two-

dimensional TLC. A considerable improvement of resolution is achieved by

combining two solvent systems with different chromatographic properties.

Two-dimensional TLC analysis of PGs and related compounds has been

reported from a few laboratories (8,28-30).

The conventional techniques of column chromatography and TLC usually

suffer from poor chromatographic resolution and the need to use several

solvent systems to adequately separate arachidonate metabolites. HPLC has









7

been used successfully for the separation and purification of PGs from

biological sources since 1976 (31). This technique offers several

advantages: (1) there is high resolution of closely related compounds; (2)

good reproducibility is possible; and (3) fractions containing PG peaks

can be automatically collected and later quantitated by RIA, GC/MS or

scintillation counting of radio-labeled metabolites. Both normal-phase

HPLC on silica acid (32-34) and reverse-phase HPLC on octadecylsilyl

silica (35,36) have been used to separate the cyclooxygenase products of

arachidonic acid. However, HPLC can be an extremely lengthy technique for

purification and can yield low recoveries on the order of 60%.


Techniques for Determination and Measurement of Prostaglandins

A number of analytical methods have been developed for the detection

and measurement of PGs to study their physiological and pharmacological

effects. Among those, bioassay, RIA, HPLC and GC/MS are most widely used

for the quantitation of PGs in biological fluids.

Bioassay. Biological techniques and bioassay have contributed

greatly to the development of techniques for detecting and quantitating

AA metabolites (37,38). In general, bioassay has been highly beneficial

in establishing the biological significance of the unstable products of

AA metabolism. However, it provides only approximate quantitation and

relatively low selectivity.

Radioimmunoassay (RIA). RIA of PGs was introduced in 1970 by Levine

and Van Vunakis (39) with assays developed for PGE1 and PGF2 The

literature has been expanding rapidly, and a large number of RIAs for PGs,

TXs and LTs have been reported. RIA is based on the competition between











radio-labeled and unlabeled molecules of a particular compound for binding

sites on an antibody directed against the same compound. The amount of

labeled compound is known and constant for all the tubes in an assay,

whereas the amount of unlabeled substance is either known and varied

(standard tubes) or unknown (sample tubes). A tube with no antibody

present is required as a "zero binding" tube. A tube containing no

unlabeled substance is also required as a "maximal binding" tube. When

larger amounts of unlabeled substance are present, the radioactive

molecules are displaced from the binding sites. The radioactivities of

the unbound fraction and antibody-bound fractions are usually separated

by dextran-coated charcoal or double-antibody methods, and the

radioactivity of either or both fractions is determined. The amount of

unlabeled compound in a sample tube is then obtained by comparison with

the standard tubes.

RIA has certain advantages over other quantitative methods, the most

important being its high sensitivity, with detection limits as low as a

picogram per sample. The precision and accuracy frequently compare

favorably with other methods. RIA is relatively rapid and also has high

sample capacity; for example, 100 samples can be analyzed within one or

two days, including radioactivity measurements and data processing.

RIA also has some drawbacks. First, the method is not entirely

specific under all circumstances. It is difficult to obtain a specific

antibody with a minimum of cross-reactivity and high affinity. Biological

samples, especially biological fluids (urine or plasma), usually need to

be purified through extraction, column chromatography, TLC, or even HPLC

before analysis by RIA. Appropriate purification steps are time-









9

consuming, but frequently necessary to remove most interfering compounds

and yield a specific assay with valid results. Other disadvantages of RIA

include the potential risk inherent in using radioactive materials and the

high cost of using disposable glassware, utensils, counting vials and

large volumes of scintillation fluid.

High-performance liquid chromatography (HPLC). HPLC has proven to

be useful for purification of PGs after an initial extraction procedure.

The HPLC technique is a good qualitative method; however, quantitation is

rather limited, especially for PGs. Terragno et al. (36) have found that

the highest molar extinction coefficient occurs around 192.5 nm for major

PGs, yielding detection limits in the nanogram range. Recently, a more

sensitive method using HPLC with a postcolumn derivatization and

fluorescence detection has been developed for eicosanoid quantitation.

Watkins and Peterson (41) developed a method to measure AA metabolites by

reverse-phase HPLC followed by formation of the ester derivative with P-

(9-anthroyloxy) phenacyl bromide. The disadvantages of HPLC are that this

technique can be lengthy and a relatively large volume of sample is

required for adequate detection of low levels of PGs in biological fluids.



Gas chromatography/mass spectrometry (GC/MS). GC/MS is the

analytical method of choice for the identification, characterization, and

quantitation of the products of the arachidonic acid cascade. Offering

both high sensitivity and selectivity, GC/MS has become the "gold

standard" for the analysis of PGs. Traditionally, GC/MS was used for

strictly qualitative analysis, with studies done on the determination of

the structures of several prostaglandins (42). Identification and









10

characterization of many prostaglandins and their metabolites were

performed by electron ionization/mass spectrometry through the early to

mid-1960's. In 1967, reports on eicosanoids first appeared, with limits

of detection in the low ng/mL range (42,43). The use of selected-ion

monitoring (SIM) with positive chemical ionization (PCI) and electron-

capture negative chemical ionization (EC-NCI) for quantitation

significantly improved the detection limits achieved by GC/MS.

The discussion that follows will focus on components of the

analytical technique of GC/MS for the analysis of PGs. In a typical

qualitative or quantitative analysis for PGs by GC/MS, the following steps

are performed: (1) sample preparation (extraction and purification); (2)

derivatization; (3) gas chromatographic separation; (4) ionization; and

(5) mass spectrometric detection. In the following pages, these

analytical steps will be discussed in reverse order, highlighting the mass

spectrometric component of the analysis, rather than sample preparation

which was discussed in detail earlier.

In mass spectrometric analysis, the quantitation of trace levels of

PGs is commonly performed by utilizing an isotope-labeled analog of the

compound of interest, with selective monitoring of the ions of each.

Since its introduction in 1967, stable isotope dilution (44) has been the

method of choice for quantitation. Many uses of stable isotope labeling

with SIM can be found in the prostaglandin literature (45-54). Both high

resolution mass spectrometry and low resolution mass spectrometry have

been employed for analysis of PGs. High resolution can reveal the

elemental composition of ions, which is helpful in identifying new

compounds. Low resolution is used for trace analysis despite its lower









11

selectivity. Examples of both techniques can be found in the literature

(55-59). A great deal of research has been devoted to trace analysis of

eicosanoids and their metabolites in all types of biological fluids, with

most determinations done in plasma and urine (48,49,60-63). The amounts

that have been analyzed are from the low ng to low pg/mL range, with

limits of detection as low as 50 fg reported in one study (64).

Three types of ionization are used today for most PGs analyses:

electron ionization (El), positive chemical ionization (PCI), and

electron-capture negative chemical ionization (EC-NCI). EI/MS, as

discussed earlier, is most often used for structural elucidation and

identification of new compounds. El mass spectra give structurally useful

fragmentation patterns, although the molecular ion may be weak or even

absent. For trace analysis, typical limits of detection with El are

approximately 100 pg/mL (57). PCI and EC-NCI are "gentler" ionization

techniques, generally producing less fragmentation, with a more prominent

(pseudo-) molecular ion. Thus, these techniques are useful for confirming

molecular weight, and for trace analysis by selected-ion monitoring. PCI

has been shown to be helpful in characterization of thromboxanes and

prostaglandins (65). Limits of detection vary for PCI and EC-NCI,

depending on both the compound and the reagent gas selected. Many types

of chemical ionization reagent gases have been used, but methane and

isobutane are the most common. Most trace analysis studies are now

performed with EC-NCI with methane as the reagent gas. Detection limits

are generally in the low pg/mL range, although limits as low as 50 fg/mL

have been reported (59). The three ionization techniques have been









12

compared for trace analysis of PGs, including limits of detection and

spectra obtained with each ionization technique (49,57).

Gas chromatography is generally used to separate the eicosanoids

from each other and from other potential interferents prior to their

identification or detection by mass spectrometry. The first GC/MS

analyses were accomplished with packed gas chromatography columns, which

were used extensively until the development of fused silica capillary

chromatographic columns. Until 1982, approximately equal use was made of

packed and capillary column techniques, but capillary chromatography has

led to better separation of closely related compounds. Coupled with

negative chemical ionization, it has allowed researchers to achieve limits

of detection in the low pg/mL range. These advantages have provided

higher sensitivity and selectivity in eicosanoids analysis. However,

packed column chromatography still has a role in prostaglandin analysis.

One recent study (66) showed the advantages of packed columns for highly

contaminated samples that exceeded the capacity of capillary columns.

Researchers have recently recognized the value of introducing the

capillary column directly into the ion source of the mass spectrometry

(67). This avoids problems with contamination, adsorption, and

decomposition of analytes (which can be severe with PGs) on active

surfaces in other GC/MS interfaces.

Derivatization of PGs has been important in their analysis, both to

increase their volatility for gas chromatography separation and to provide

for efficient EC-NCI to increase sensitivity of the GC/MS method. Today,

the methylester/methoxime/trimethylsilyl ether of PGs is the most

frequently cited derivative in GC/EI/MS analysis (63,64). However, it has









13

been shown that these derivatives are susceptible to hydrolysis, often

producing ions that are not optimal for selective-ion monitoring (68).

This is due to the low relative intensity of the high mass ions which are

optimal for quantitation.

The derivatization of PGs for GC/EC-NCI/MS seems to be standardizing

on the methoxime/trimethylsilyl ether/pentafluorobenzyl ester (MO/TMS/PFB)

mixed derivative (50,51,69-72). Derivatization with perfluorinated acid

anhydrides has been increasingly used for both qualitative and

quantitative work (69). These anhydrides usually incorporate a silylating

reagent such as N-(tetra-butyldimethylsilyl)/N-(methyltrifluoroacetamide).

This gives hydrolytic stability and increases high mass ion intensity for

optimal use of selective-ion monitoring. The use of such derivatives also

eliminates detection of many nonprostaglandin carboxylic acids, due to

their ability to derivatize with the carbonyl, rather than, or in addition

to, the carboxyl group. This makes these derivatives highly attractive

for detecting trace quantities of prostaglandins in biological matrices

(73).


Recent Analytical Advances


GC/MS remains the workhorse technique of PG research; however,

tandem mass spectrometry (MS/MS) and soft ionization techniques such as

fast atom bombardment (FAB) or liquid secondary ion mass spectrometry

(LSIMS) and liquid chromatography/mass spectrometry (LC/MS) are being

effectively employed. The sensitivity and selectivity of GC/MS/MS

compared to GC/MS has been studied in reports (74-76) utilizing both El

and EC-NCI. The advantages of GC/MS/MS have recently been exploited for









14

the trace analysis of PGs in biological samples (20,77,78). These studies

have been performed on both sector and quadrupole instruments. The high

selectivity of MS/MS makes it possible to perform analyses with minimal

sample preparation. MS/MS also minimizes or eliminates the need for

chromatographic separation in many cases, making the analysis extremely

rapid. MS/MS experiments have recently been reported in the literature

for analysis of underivatized prostaglandins (79,80).

In addition, with improved instrumentation has come the technique

of FAB or LSIMS (81-83). This method has aided structural elucidation,

as well as characterization of many PGs. LC/MS has become increasingly

popular in the analysis of PGs (84,85), as in all areas of chemistry.

LC/MS has the ability to analyze polar, thermally labile, and high

molecular weight eiconsanoids, and it saves time in sample preparation.

LC/MS with thermospray ionization (TSP-LC/MS) has been used by several

researchers to detect PGs and TXB2 at limits of detection as low as 10-

300 pg (on column), after derivatization with (diethylamino)ethyl chloride

(86). A series of PG standards were analyzed and investigated to show the

increase in sensitivity resulting from a post-column derivatization which

formed the methyl ester (87). The sensitivity is still not equal to the

GC/MS methods commonly employed. This is the limiting factor of LC/MS

for the analysis of PGs; however, there is reason to believe that the

necessary improvements in sensitivity can eventually be obtained. LC/MS

is a good qualitative technique which is still in its infancy. The

advantages to be gained in simplified sample preparation and the ability

to directly analyze the more polar eiconsanoids will stimulate further

improvements.









15

Another recent MS/MS technique which is promising is ion trap

(IT)MS/MS. The ITMS offers the potential for very selective and sensitive

GC/MS/MS analysis. In the ion trap, ion formation and mass analysis occur

in the same region (tandem-in-time), whereas, in tandem mass spectrometry

these two processes occur in different regions (tandem-in-space). The

analysis of PGs by this method has been reported by Strife (88,89). This

work shows the unique advantages of high sensitivity MS/MS,with nearly

100% conversion efficiencies of parent to daughter ion in MS/MS experi-

ments.

This section of Chapter 1 has shown that much progress has been made

in the area of PG sample preparation and quantitation. Many limitations

remain, especially when the sample size is limited. In the chapters to

follow some of these limitations will be addressed and new analytical

schemes will be evaluated.


Strategies for Mixture Analysis by MS/MS


Since the development of tandem mass spectrometry (MS/MS) in the

1970's, it has recently gained rapid acceptance as an exceptional

analytical tool for mixture analysis (90-93). MS/MS has the ability to

provide rapid, sensitive and selective analysis of complex biological

samples, often with minimal sample clean-up (94,95).

The MS/MS scan modes utilized in these studies are depicted in

Figure 1-2. In mixture analysis, chemical ionization of a mixture is

often utilized in the ion source of the mass spectrometer to produce ions

characteristic of the components in the mixture and to achieve a spectrum

with few fragments. Separation of the analyte from the matrix components









16

ro ro
C C

cr





C 4-
0 oa 0




ICfo O
C0 c" o ".11.14

*o
cn L





11 111,M 1 111

a a





%MOO
*4




.C
C C










cc .....









17

is achieved by the mass selection of a characteristic ion of the analyte

by the first mass analyzer (Ql). The selected parent ion undergoes

collisionally activated dissociation (CAD) through collisions with neutral

gas molecules in the fragmentation region (Q2) to yield various fragment

or daughter ions. Subsequent mass analysis of the daughter ions by the

second mass analyzer (Q3) results in the analytical signal. This method

of MS/MS analysis corresponds to a daughter scan (Figure l-2b).

Although this operational mode is highly selective, this full-scan

daughter mass spectrum usually does not exhibit sufficient sensitivity for

trace analysis of an analyte in a complex matrix. Therefore, the scan

mode of selected-reaction monitoring (SRM) is commonly employed (Figure

l-2d). A characteristic daughter ion, typically the most abundant,

resulting from the fragmentation of the selected parent ion of the

analyte, is selected by the second mass analyzer for monitoring. SRM is

analogous to the selected-ion monitoring (SIM) (Figure 1-2c) commonly used

to obtain maximum sensitivity in conventional GC/MS. Thus, an enhancement

in sensitivity is obtained at the expense of a gain in selectivity. In

addition to these MS/MS modes, the tandem mass spectrometer can be

operated as a normal MS by allowing all ions to pass through one mass

analyzer (Ql or Q3) and the collision cell (Q2), then scan the other mass

analyzer (Q3 or Ql) to produce a normal mass spectrum (Figure 2-la).

Optimization of many of these operational modes have been evaluated

throughout these studies and will be discussed in further detail as to

their significance in the trace determination of PGs.









18

Important Parameters for Trace Analysis

In order to perform trace analyses successfully, it is necessary to

think in terms of the four "S's" of analysis: (1) sensitivity; (2)

selectivity; (3) speed or analysis time; and (4) 1 or cost. In the

determination of pure analytes, sensitivity can be a very useful

criterion; however, when required to determine an analyte in a complex

matrix, sensitivity alone may become meaningless. This is due to the fact

that chemical interferents in the matrix may themselves produce a response

or interfere with the signal of the analyte. Therefore, the factor which

may determine the smallest amount of analyte which can be determined

accurately is the second "S", selectivity. The selectivity can be

described as the ability of the method to distinguish the signal of the

analyte from that of the chemical interferents (so-called chemical noise).

The limit of detection (LOD), which depends upon both the selectivity and

sensitivity, is defined as the smallest amount of analyte which can be

detected.

The LOD required in trace analyses can be achieved by improving the

selectivity of the analytical scheme. Normally, this is accomplished

through the use of extensive sample clean-up and separation to enhance the

analyte signal with respect to the matrix components signal. These

extractions and purifications increase the possibility of sample

contamination and sample loss. Additionally, the methods necessary to

increase selectivity may become time-consuming and expensive, thus the

final two "S's", speed of the analysis and cost effectiveness may not be

optimum.









19

The Four Steps Involved in Trace Mixture Analysis

The analytical scheme for trace determination of an analyte in a

biological sample by MS/MS involves four basic steps: (1) sample

preparation; (2) sample separation/introduction; (3) ionization; and (4)

detection. When developing an accurate, reliable and specific method for

mixture analysis, a range of selectivity, sensitivity, time and cost are

observed for the four steps.

In sample preparation, a rapid, low cost and selective procedure is

desired. This can be achieved through the proper choice of extraction,

purification or derivatization methods which satisfy any or all of the

four "S's". The second step involves separation of the analyte of

interest from any matrix components which have not been eliminated by the

sample preparation methods. Typically, in MS/MS, gas chromatography is

employed for separation, if the analyte exhibits sufficient volatility.

Separation of components can be accomplished on short capillary GC columns

(3 m or less), when the sample has been adequately cleaned-up (96). Short

GC columns can only be utilized for separation of complex samples if the

sample preparation methods have the necessary selectivity. The choice of

an ionization method is based on the type of analysis required and the

analyte which is to be analyzed. In the low level trace determination of

analytes in biological samples, a "soft" ionization method (e.g., chemical

ionization) is usually selected which yields an intense molecular ion with

few fragments. Furthermore, for analytes which are highly electron-

capturing (or can be derivatized), electron-capture negative chemical

ionization (EC-NCI) may be chosen in order to achieve the highest

sensitivity. Finally, the detection by MS/MS involves the optimization









20

of the parameters which constitute the operational modes which were

discussed above.


Overview of Thesis Organization


This thesis is divided into seven chapters. Chapter 2 describes

the sample preparation concepts and methods employed for this work.

Results from recovery studies on various types of extraction columns and

their characteristics are discussed in detail. The concept of

immunoaffinity purification is introduced and investigated.

The third chapter emphasizes the importance of optimizing MS/MS

parameters for trace determination of PGs. Optimization studies for

selected-ion monitoring (SIM) and selected-reaction monitoring (SRM) for

PGEZ and PGF. are described and the results discussed.

Chapter 4 presents a study of the differences in the CAD efficiency

of two structurally similar PGs (PGE2 and PGF2 ). Collision energy and

collision gas pressure studies of the carboxylate anions of four PGs are

evaluated and hypotheses for the differences noted are put forth.

The results of the quantitation study of endogenous PGE2 in urine can

be found in Chapter 5. The advantages and disadvantages of various

analytical schemes are pointed out. These schemes are systematically

evaluated for the trade-offs in sensitivity, selectivity and time of

analysis. The trade-offs are discussed in relation to how they are

affected by the three basic steps (sample preparation, sample introduction

and detection) of PG analysis.

Chapter 6 includes an evaluation of the rapid analysis techniques

of direct solids probe/MS/MS and direct chemical ionization (DCI)/MS/MS









21

utilizing an abbreviated derivatization procedure. The advantages and

limitations are discussed and the results of a quantitation study of

endogenous PGE2 in urine are presented.

The final chapter reviews the conclusions which were drawn from this

work. This chapter points out the potential of GC/MS/MS with selective

sample preparation and short GC capillary columns to determine endogenous

levels of PGE2 in urine. The importance of a systematic study of the

entire analytical scheme is finalized. In addition, future work is

proposed which should further enhance PG analysis by MS/MS.















CHAPTER 2

SAMPLE PREPARATION STUDIES



Introduction


Sample preparation is an important step in any analytical

methodology. This step prepares the sample for the detection method and

can dramatically affect the validity of the data obtained. The two main

parts of sample preparation for gas chromatography/mass spectrometry

(GC/MS) are sample purification and derivatization. When considering

sample purification, selectivity and speed of the method are of vital

importance. A method which is extremely selective can eliminate matrix

interference and reduce the separation needed in GC. Sample throughput

is always an important factor in any analytical method. A rapid sample

purification step can greatly reduce the total time of analysis.

The other main part of sample preparation is derivatization. Many

compounds are not directly amenable to GC. The thermal liability of

prostaglandins (PGs) makes it impossible for them to pass through a GC

column intact without first undergoing derivatization. This

derivatization increases the volatility of the compound and reduces the

interaction of the polar substituents on the compound with the stationary

phase of the GC column. In addition, derivatization can add sensitivity

and/or selectivity for detection of a compound. Many organic

derivatization reactions with PGs enhance the efficiency of electron-

22









23

capture negative chemical ionization (EC-NCI) mass spectrometry (51,

70-72). Electron-capture NCI with derivatization produces much simpler

mass spectra and the major fragment ions occur at the high mass range

(50,69). Thus, these two features combined with the higher ionization

efficiency of EC-NCI provide added sensitivity and selectivity needed in

trace determination of PGs.


Concepts for Solid-Phase Extraction


Solid-phase extraction (SPE) has emerged, in the last ten years, as

the method of choice for isolation and purification of arachidonic acid

metabolites (8,14-16). SPE has the advantage of using low volumes of

solvents and high recoveries of 90% to 100% for most PGs. Rapid

extractions are usually possible with simple procedures. This results in

a rapid inexpensive extraction technique. The concept of SPE is based on

the selective retention of the analyte by a sorbent bed as a solvent in

which the analyte is dissolved is passed through the column. This idea

is displayed graphically in Figure 2-1. A sample containing analytes (A)

and interference (I & M) is passed through the sorbent. The sorbent

selectively retains analytes (A) and some interference (I). However, at

the same time, many interference (M) pass unretained through the sorbent.

Appropriate solvents are then used to wash the sorbent, selectively

eluting previously retained interference (I), while the analytes (A)

remain on the sorbent bed. Purified, concentrated analytes (A) are then

eluted from the sorbent.











I I
A
A M

IM
A
A
A


Figure 2-1:


Concept of solid-phase extraction:
A analyte; I & M interference.


A











Sorbent/Analyte Interactions

Three types of chemical interactions are commonly employed in solid-

phase extractions (97). The first is the non-polar interaction which

occurs between the carbon-hydrogen bonds of the analyte and that of the

sorbent. Virtually all organic compounds have some non-polar character,

thus these types of interactions are the most commonly used to retain

analytes on sorbents. The forces which are involved in such non-polar

interactions are "van der Waals" or dispersion forces (97,98). The most

widely used sorbent in non-polar interactions is octadecyl silane bonded

to a silica substrate which is called C18. Many compounds can be retained

by a C18 sorbent, thus it is a very non-selective sorbent. In general,

non-polar solid-phase extraction is the least selective extraction

procedure. The concept of non-polar interaction is comparable to that of

reverse-phase chromatography. Retention of the analyte by non-polar

interaction is facilitated by a solvent more polar than that of the

non-polar sorbent. Elution is then accomplished by utilizing a solvent

with sufficient non-polar character to release the retained analyte from

its interaction with the sorbent.

Other interactions which are common for SPE are polar interactions.

These interactions are exhibited between many sorbents and functional

groups on analytes. All bonded silica exhibits polar interaction due to

the polar nature of the silica substrate to which the sorbent is bound

(97,98). Polar interactions include hydrogen bonding, dipole/dipole, pi-

pi and many more interactions in which the distribution of electrons is

unequal in the atoms of the functional groups. This property of polar

sorbents allows an analyte which contains a polar functional group to









26

interact with a polar group on the sorbent. Groups that exhibit these

types of interactions include hydroxyls, amines, carbonyls and other

groups containing hetero-atoms such as oxygen, nitrogen, sulfur and

phosphorous. The most common polar sorbents are silica, diol, aminopropyl

and cyanopropyl. Polar sorbents function similarly to the interactions

found in normal-phase chromatography. Non-polar solvents are used to

promote retention of the analyte on the polar sorbent. Then a solvent,

more polar than the sorbent, is utilized to elute the analyte.

The third type of interaction is ionic. This occurs when an analyte

carrying a charge (either positive or negative) interacts with a sorbent

carrying a charge opposite to that of the analyte. Ionic interactions are

more selective than non-polar and polar interactions and can be controlled

by adjusting the pH of the sample solution. It is essential to know about

the functional groups on the sorbent and the analyte because both of these

need to be charged to facilitate ionic interaction. Two classes of ion-

exchange interaction exist, cationic (positively charged) and anionic

(negatively charged). Examples of cationic interactions include the

interaction of amines and certain inorganic cations with carboxymethyl,

sulfonylpropyl and benzenesulfonylpropyl sorbents. Anionic interactions

occur when sorbents containing primary, secondary, tertiary and quaternary

amines interact with carboxylic and sulfonic acids, phosphates and similar

groups on an analyte.

Recently, covalent interactions have been exploited for extraction

of specific types of compounds (99). Covalent chromatography is highly

chemically selective, involving an interaction of greater energy than is

employed in the other extraction methods. Retention of the analyte occurs









27

when a covalent bond can form between it and the sorbent. A change in the

solvent environment facilitates elution of the analyte. This is commonly

accomplished through the use of solvents with various pH's. One example

is phenyl boronic acid (PBA) which has been immobilized for the selective

retention of compounds with 1,2- or 1,3-diols such as catecholamines and

thromboxanes (17,100).

Many of the sorbents discussed above may exhibit more than one

interaction. Both polar and ionic interactions due to the silica

substrate used can occur in all sorbents. In the case of the PBA, non-

polar, polar and ionic interactions can occur as secondary interactions

within the sorbent. The interactions which occur with a particular

sorbent are a function of the sample matrix and the solvent used for

washes and elution.


Sorbent Selection

The problem encountered in this analysis was to develop a solid-

phase extraction to selectively isolate PGs from interference in urine.

Evaluation of different sorbents followed two fundamental steps. First,

sorbents were selected which in theory have the capability to retain PGs

from urine. Next, the different sorbents chosen were tested to evaluate

their actual ability to selectively retain the PGs of interest (97).

The sorbents which were chosen for the study were determined by

examining properties of the analyte (PGs) and the matrix (urine). First,

the determination of the interactions which PGs could undergo was

examined. Areas of carbon/hydrogen content with alkyl chains suggested

that non-polar retention was probable. The presence of such polar groups









28

as hydroxyls (OH) and carbonyls (-0) indicated a potential for retention

by polar interactions. Ionic interaction was indicated by the presence

of the carboxylic acid moiety. However, this method was not evaluated for

the analysis of PGs in urine due to the excessive quantities of compounds

in urine which would undergo anionic and cationic interactions.

Considering that PBA has been used for separation of the arachidonic acid

metabolite, thromboxane B2 (TXB2), covalent interaction with some PGs

appears possible due to the 1,3-diol present on the cyclopentane ring.

Next, the properties of the matrix and the potential interfering

components which are contained in urine were considered. Urine is an

aqueous media which contains many proteins, salts and solids. Components

with polar and non-polar functionalities can be found throughout urine

samples. This suggested that the interference would undergo the same

interactions with the sorbents as the PGs. Therefore, to determine which

interactions would work most effectively, the sorbents required testing.

A sorbent testing scheme is shown in Figure 2-2 (97). First, each

sorbent needed to be prepared. This was accomplished by washing the

sorbent bed first and allowing the functional groups on the sorbent to

interact with the solvent. The next step was to remove the wash solvent

and create an environment that facilitated the analytes (PGs) retention.

After this process, the testing procedure began and involved five steps.

Standards were prepared identical to a "real" sample and applied to

the column sorbentt). The standards were then washed with the same

solvent in which they were dissolved, and the eluent collected. The

eluent was then checked for the presence of analyte, indicating sorbents

which did not provide adequate retention of the analyte. Next, strong











Optimize Retention of Standards



Optimize Elution of Standards

Identify Wash Solvents




Test Blank Matrix

Use Wash Solvents



Test Spiked Matrix



Troubleshoot if Necessary


Figure 2-2: Sorbent testing scheme.









30

elution solvents were chosen of which small volumes can be utilized to

completely elute the retained analyte. During this process, solvents

which would not elute the analyte were identified for use as wash

solvents. These were tested next with a blank matrix (urine) to determine

the solvent(s) which produced the cleanest extract. Clearly, this was far

more difficult to evaluate than the determination of analyte recovery.

After developing a procedure which provided sufficient analyte retention

and elution, as well as adequate clean-up of the matrix, the method was

tested with a sample (urine) spiked with analyte. Recoveries found in

this step were similar to those obtained with the standards. However, if

problems had been encountered, either the sorbent, wash solvents or

elution solvents may have been changed to provide for adequate retention

and elution of the analyte in the matrix.


Antibody Affinity Extraction

Extraction methods for PGs based on liquid-liquid or solid-phase

extraction are relatively nonselective and the final extracts are

frequently unsuitable for direct analysis, even by highly specific GC/MS

quantitation methods (7,101). The necessity of further purification of

the extracts before chromatographic analysis makes the analytical

procedures more complex, laborious and time-consuming to develop. This

problem has been avoided in many cases by taking advantage of immuno-

adsorption techniques to simplify extraction and clean-up procedures for

GC/MS analysis (18-21,102). Reports have shown that the selectivity of

the immunoadsorption procedures may permit the direct analysis of extracts

and eliminate the need for intermediate chromatographic clean-up.









31

Antibodies have been used for many years for the analysis of PGs by

radio-immunoassay (RIA). Unfortunately, the presence of substances within

the sample matrix which exhibit cross-reactivity with the polyclonal

antibody can be considerable (39,103). For example, antibodies for 20

carbon PGs and their metabolites may also bind the corresponding dinor

metabolites present in the matrix (19). Thus, HPLC is frequently employed

as a separation technique prior to RIA to avoid cross-reactivity. Reports

have shown that without separation of cross-reacting components by HPLC,

PG levels have been found 20 times higher than the actual levels present

(7,8). Immunoadsorption purification has been utilized as well prior to

PG analysis by RIA (104,105). However, this method has the disadvantage

of combining a purification procedure based on immunoaffinity with a

measurement procedure based upon the same principle. The advantage of

utilizing immunoadsorption for purification before GC/MS analysis is that

the highly specific antibody will enhance the selectivity by providing

discrimination which is unrelated and complements the characteristics of

GC/MS. This results in an analysis method for PGs which has a higher

degree of specificity than RIA.

These ideas have been incorporated in the sample preparation of PGs.

The inherent selectivity of the antibody-antigen interactions has been

exploited for PG analysis by Knapp and Vrbanac to obtain relatively pure

sample extracts (20,78). The basic principle of antibody affinity

extraction is displayed in Figure 2-3. In a typical affinity

chromatographic separation, the antibody is coupled to a stationary phase

(the most popular is agarose gel). The selectivity of affinity

separations is based on the principle of "lock and key" binding which





















Adsorb









Regenerate







Elute


0'


Figure 2-3:


Basic principle of antibody affinity extraction:
A analyte; I & M interference.


Wash









33

occurs in biological systems. Extraction of the sample is accomplished

by passing the solution (containing analytes and interference) through

the sorbent bed; the PGs which have affinity for the antibody are adsorbed

while other components pass through unretained. The retained or adsorbed

PGs are then eluted by changing the solvent.

Additional, secondary interactions are possible with immunoaffinity

chromatography. One important interaction discussed earlier is due to the

cross-reactivity of the polyclonal antibodies. Furthermore, non-specific

binding of interfering components may occur during the immunoadsorption

procedure. The bulk protein carrying the antibody has the potential for

interaction of components in the sample matrix. In addition, polar

interactions are possible between the silica stationary phase and any

polar functionalities found in the sample matrix.

In general, antibody affinity purification can decrease significant

loss of sample which can occur in TLC and HPLC. This method of sample

preparation is relatively rapid and requires no additional purification

of biological samples to obtain an adequate interference free GC trace.

The greatest advantage is the significant selectivity of the separation

process for antibody affinity purification compared to other conventional

chromatographic methods.


Concepts for Derivatization


Derivatization of PGs has been important in their analysis by GC/MS,

both to increase their volatility for GC separation and to provide for

efficient EC-NCI to increase sensitivity of the GC/MS method. Many

different derivatives have been used in the analysis for PGs (10,11). As

reported earlier in Chapter 1, the most commonly used derivative for









34

quantitative analysis by GC/MS is the methoxime/pentafluorobenzyl/tri-

methylsilyl (MO/PFB/TMS) derivative.

The keto group on PGE2 is converted to the methoxime derivative to

prevent silylation which can interfere with quantitation by producing

additional derivatives. Pentafluorobenzyl (PFB) esters are created to

enhance the efficiency of ionization by EC-NCI in order to achieve low

level determinations of PGs. These PFB esters have been found to give

about twice the sensitivity of the methyl ester derivative (106). Reaction

times are fast (-20 min) and quantitative (-100%) for this derivatization

procedure. The hydroxyl groups are converted to trimethylsilyl (TMS)

ethers using O-bis(trimethylsilyl)-trifluoroacetamide (BSTFA). This TMS

donor has the additional advantage of creating extremely volatile reaction

by-products which usually elute with the solvent front in the GC trace.

Even though the derivatization for quantitative analysis of PGs by

GC/MS has been thoroughly documented, there are many variations in the

literature. It has been reported that by performing the methoximation

before the esterfication a fivefold increase in the derivative yield can

be obtained (47). However, many researchers still perform the ester-

fication step first in the derivatization procedure (10,11,74). Reaction

times for the methoximation step vary in the literature ranging from one

hour at 60 to 24 hours at room temperature. These differences, in

addition to the comparison of techniques for the removal of excess

derivative reagents by liquid-liquid extraction and nitrogen evaporation

were explored in this study.









35

Experimental


Prostaglandins and Reagents

All solvents were reagent or HPLC grade. Prostaglandin E2 (PGE2) was

purchased from Sigma Chemical Co. (St. Louis, MO). [5,6,8,11,12,14,15 -

3H2]-PGE2 and Riafluor liquid scintillator were from New England Nuclear

(Boston, MA) and were a gift from Dr. J. Neu of the Department of

Pediatrics, University of Florida (Gainesville, FL). The solid-phase

extraction columns were purchased from Analytichem International, Inc.

(Harbor City, CA) and Waters Assoc. (Sep-Pak columns; Milford, MA).

3,3',4,4'-(2H4) PGE2 and the antibody affinity sorbent were gifts from

Drs. J.J. Vrbanac and D.R. Knapp of the Department of Pharmacology,

Medical University of South Carolina (Charleston, SC). The derivatization

reagents and solvents pyridine, O-methylhydroxylamine hydrochloride,

acetonitrile, and N,N-diisopropylethyl amine for GC/MS percent recovery

studies were all purchased from Sigma Chemical Co.. Pentafluorobenzyl-

bromide (PFBBr) and bis(trimethylsilyl)-trifluoroacetamide (BSTFA) were

purchased from Pierce Chemical Co. (Rockford, IL). Urine was obtained

from the author. All glassware was silanized with a solution of 5%

dimethyldichlorosilane in toluene. These two chemicals were both

purchased from Sigma Chemical Co.. Helium used as GC carrier gas and

methane (>99%) used as the chemical ionization reagent gas were from

Matheson Gas Products, Inc. (Orlando, FL).


Sample Preparation

The sorbents for the percent recovery studies were chosen and tested

according to the procedures discussed earlier in this chapter. Extraction









36

procedures were determined for the non-polar, polar and phenyl boronic

acid columns by detection of the tritium-labeled PGE2 by scintillation

counting. The sorbents chosen are listed with the final extraction

procedure used for the percent recovery studies for both scintillation

counting and GC/MS.

Non-polar columns: octyl (C8), octadecyl (C18) and phenyl (PH)

(1) Conditioned the column with 10 mL of HPLC water and 10 mL of

methanol.

(2) Passed solution of PGE2 (acidified to pH 3.5 with formic acid)

through the column.

(3) Washed the column with 10 mL of HPLC water and 10 mL petroleum

ether.

(4) Eluted PGE2 with 10 mL of ethyl acetate.

Polar columns: silica (SI), cyanopropyl (CN), aminopropyl (NH2) and

diol (20H)

(1) Conditioned the column with 10 mL of benzene:ethyl acetate

(80:20 volume:volume).

(2) Passed solution of PGE2 (acidified to pH 3.5 with formic acid)

through the column.

(3) Washed the column with 10 mL benzene:ethyl acetate (60:40 v:v).

(4) Eluted PGE2 with 10 mL benzene:ethyl acetate:methanol

(60:40:30 v:v:v).

Phenyl boronic acid column (PBA)

(1) Conditioned the column with 5 mL of 0.1 M hydrochloric acid and

5 mL of 0.1 M sodium hydroxide.









37

(2) Passed sample of PGE2 (adjusted to pH 8.5 with 0.1 M phosphate

buffer (PBS)) through the column.

(3) Washed the column with 5 mL of methanol and 5 mL of HPLC water.

(4) Eluted PGE2 with 5 mL of 0.1 M PBS (pH 6.5).

The antibody affinity columns were tested and percent recovery data

calculated only with GC/MS.

Antibody affinity column [Immunoaffinity (IA)]

(1) Conditioned the column with 20 mL of PBS (pH 7.4).

(2) Passed solution of PGE2 (acidified to pH 3.5 with formic acid)

through the column.

(3) Allowed the sample to settle into the sorbent bed for 15 min at

room temperature.

(4) Washed the column with 25 mL of PBS (pH 7.4) and 10 mL HPLC

water. Removed all remaining water in the column.

(5) Eluted PGE2 with 15 mL of 95% acetonitrile solution (v:v).

(6) Washed column with an additional 10 mL of 95X acetonitrile to

assure removal of all the PGE2.

(7) Immediately rinsed the column with 10 mL of HPLC water and

15 mL of PBS (pH 7.4).


Scintillation Counting

A stock solution of 3H-PGE2 was used for the percent recovery

studies. This solution was 0.09375 microcuries (pCi)/microliter (pL) and

had a specific activity of 169.5 pCi/millimole. Six microliters of this

original solution was diluted with 100 pL of absolute ethyl alcohol

creating a solution of 5.625 x 10-3 pCi/pL. Ten microliters of this









38

standard solution, corresponding to 3.319 x 10-4 mmoles or 0.1218 mg was

passed through each column tested. Following the extraction procedures

the eluent was collected and the solvent evaporated with nitrogen. The

3H-PGE2 was then diluted with 100 pL of PBS (pH 7.4). Additionally, 3.5

mL of Riafluor liquid scintillator were added to the 9.375 x 10'3 pCi

solution of 3H-PGE2 before the counting process. Each extraction was

performed three times with three individual columns.

A calibration curve was prepared in the same manner with the

exception of the actual extraction step (Figure 2-4a). Aliquots of 4,

5.5, 8, 10.5, 12 and 13.5 microliters of the 5.625 x 10'3 pCi/pL solution

were added to separate vials and each diluted with 100 pL of the PBS (pH

7.4). In addition, a blank containing only 100 pL of the PBS (pH 7.4) was

prepared. The Riafluor liquid scintillator was added and the standards

counted and used to calculate the percent recovery values for the

different columns tested.


Gas Chromatography/Mass Spectrometry (GC/MS)

Ten nanograms (ng) of PGE2 were passed through each column for the

percent recovery studies of standards. Following the extraction procedure

for the columns tested, 10 ng of 2H4-PGE2 were added and the solutions were

evaporated to dryness with nitrogen. The same procedure was followed for

extraction of PGE2 in urine for percent recovery studies except that the

10 ng of PGE2 added to the urine is in addition to the endogenous levels

present.

Calibration curves for the GC/MS analysis were prepared by adding

a constant amount of 2H4-PGE2 (25 ng) and increasing amounts of PGE2 in the













2000-



o o






3.0
>1500 .






C












u3.0



) 2.5 (
Q,_







0 .0 2



I 1.5


LLJ
0 ,
















Figure 2-4: Calibration curves: (a) Scintillation counting
(b) Gas chromatography/mass spectrometry (GC/MS)
with selected-ion monitoring (SIM).












40




O ^

t b 0 0 0 0 0 0 0 0


r. N ,-4 iC L r- 0
0 0

0
oa
U 4-1
0








o0 b

S, 0 0 0 0 0 0 0

S LA in tn LA Ln L LAn Ln



0
4.4
0








h)
<:s


4. o
o 0 0 0 0 0 0 0 0




0 0
O4









41

solution (Figure 2-4b). Table 2-1 lists the amounts of 2H4-PGE2 and

standard PGE2 added to each vial and the final concentrations after

dilution with 50 pL of silanizing reagent.


Derivatization for GC/MS

The methoxime/pentafluorobenzyl ester/trimethylsilyl (MO/PFB/TMS)

derivatives were formed for the GC/MS percent recovery and derivatization

studies. The method used was similar to the derivatization of H. L.

Hubbard et al. (19). The standards and samples of 2H4-PGE2 and PGE2 after

evaporation were treated with 100-200 CpL of methylhydroxylamine HC1 in dry

pyridine (4 mg/mL), allowed to stand overnight at room temperature, then

evaporated under nitrogen until dry. Each sample was acidified by adding

200 pL of IN formic acid, extracted with two 1 mL aliquots of ethyl

acetate, and the extract dried under nitrogen. Then 50 pL of

acetonitrile, 30 pL of 30% PFBBr in acetonitrile, and 15 pL of 10% N,N-

diisopropylethylamine in acetonitrile were added to the dried methoxime

derivative. Each solution was allowed to stand for 30 minutes at room

temperature before the reagents were evaporated with nitrogen. Excess

derivatizing reagent was removed by dissolving the sample in 200 pL of

distilled water and extracting with two 1 mL aliquots of a methylene

chloride:hexane (50:50 v:v) solution; the extract was then dried under

nitrogen. The trimethylsilyl derivative then was formed by adding 50 pL

of BSTFA to the standards for the calibration curve and 20 iL to the

extraction samples and allowing the solutions to stand overnight at room

temperature. One-microliter injections containing 500 pg of H4-PGE2 were

made of each standard and sample.











Instrumentation

A Beckman LS 3800 liquid scintillation counter and a Finnigan MAT

triple stage quadrupole (TSQ45) gas chromatograph/mass spectrometer were

used in these studies. Gas chromatography was carried out on a

conventional J&W Scientific (Folsom, CA) DB-1 (30 m long, 0.25 mm i.d.,

0.25 pm film thickness) capillary column in the splitless mode with helium

carrier gas at a flow rate of 41 cm/s (inlet pressure 18-20 psi). The

initial temperature of 250*C was held for 30 s, then increased at 20'C/min

to 310*C for the calibration curve and percent recovery studies of

standards. Urine percent recovery data were obtained with an initial

temperature of 100*C held for 30 s, increased at 25*C/min to 250C, then

increased again at 50C/min to 310*C.

Mass spectrometry conditions were: interface and transfer line

temperature 300C, ionizer temperature 190*C, electron energy 100 eV and

emission current 0.3 mA. Electron-capture negative chemical ionization

(EC-NCI) was carried out with methane at an ionizer pressure of 0.45 torr.

In the GC/MS percent recovery and derivatization studies, a specific

ion for PGE2 (524-; [MO/TMS-PFB]') and for 2H4-PGE2 (528; [MO/TMS-PFB]')

were selected and monitored throughout these studies. The selected ion

monitoring mode (SIM) with quadrupole one was used on the mass

spectrometer. A baseline was chosen visually on the GC trace and the

areas for PGE2 and 2H4-PGE2 calculated by the INCOS computer system for the

calibration curve and percent recovery samples. The area of PGE2 divided

by the area of 2H4-PGE, in the standards gives a ratio which is used in the

calibration curve. The amount of PGE2 recovered through each column was









43

calculated by comparing the ratio of these ions after the extraction

procedure to that of the calibration curve.


Results and Discussion


Percent Recovery Studies (Scintillation Counting)

Table 2-2 shows the percent recovery results for the different

sorbents tested. These recoveries were determined by scintillation

counting of 3H-PGE2 standards. Three samples were extracted for each

sorbent and the average and percent relative standard deviation (%RSD)

calculated. Examination of Table 2-2 indicates that a wide range of

recoveries was found for the different sorbents tested.

The non-polar sorbents investigated were octyl (C8), octadecyl (C18)

and phenyl (PH). The C18 columns had the highest percent recoveries

(97.2%) with the C8 (81.8%) and the PH (8.8%) columns having lower

recoveries. These low recovery values for the C8 and PH columns signify

either: (1) PGE2 was unretained as the initial standard solution was passed

through the column; (2) PGE2 was eluted during the wash procedure; or (3)

PGE2 was irreversibly bound to the column or not effectively eluted. The

reasons for poor recovery were investigated by examining all the eluents

which were passed through the respective columns to determine the presence

of PGE2. It was discovered that the C8 and PH columns did not initially

retain PGE2. Octadecyl columns exhibited the smallest variation from

column to column with a %RSD of 0.6.

Polar sorbents tested were silica (SI), cyanopropyl (CN) aminopropyl

(NH2) and diol (20H). As with the non-polar columns, a wide range of

recoveries were discovered for the different polar interactions tested.



































bO)

g 0


00

0 O
oo
,-.


coI


00
4U





o co
a)



> 0
o
0 0O





.-4
0
00


S -4

S "

o 0
0 0
o
O v
0,


0
o







00



0 I 4
-,41







0

--






0



>1 I *r
0





0 ) u





. 0 o 4 Q
4 PO 3



,-4

caa
A4 30
*o a)
2 CM


r *h mOE

,-4

0
-4 CU o1
re4 ( d

C -I


A~ o
a M z
fr< (f .,









45

The two sorbents that exhibited greater than 90% recovery were the SI

(100.4%) and 20H (93.7%). The recoveries for these columns were found to

be much higher than those found for CN (24.9%) and NH2 (12.6%) columns.

This difference in recoveries may be attributed to the interaction of the

polar groups on PGE2 (particularly the hydroxyls) with the hydroxyl groups

on the SI and 20H, rather than with the carbon/nitrogen interaction with

CN or the amine group with the NH2. Investigation of the eluents showed

that PGE2 was not effectively retained initially for the CN and NH2

columns. The variation from column to column (%RSD) for the SI (2.5%) and

20H (5.0%) were less than that found for CN (10.7%) and NH2 (11.3%).

In addition, phenyl boronic acid columns were tested. The percent

recovery found using this type of interaction was 40.0%, well below the

recovery values found for C18, SI and 20H columns. This extraction

technique is based on the premise that the tetrahedral anionic form of

boronates condense with 1,2- or 1,3-diols to form five- or six- membered

covalent complexes (107). The low recovery of PGE2 observed for the PBA

column can possibly be explained by the inability of the boronate to

condense with the diols on the cyclopentane ring of PGE2 to form a stable

complex. An explanation for the inability of PGs to condense with the PBA

column was reported by Lawson et al. (17). They believe that the tendency

of the planar phenyl groups to orient so that their pi (w) orbitals align

or are stacked, thereby forcing the boronic acid groups to be too close

together, not allowing the sterically fixed cyclic 1,3-diols on PGs free

access.

Data obtained from testing these sorbents suggests that PGE2 has

preference for retention on specific non-polar (C18) and polar (SI and









46

20H) sorbents. Even sorbents with the same type of interactions

demonstrate varied retention for PGE2. Recoveries for the C18, SI and 20H

columns are similar and demonstrate adequate retention (>90%) of PGE2 to

justify further investigation.


Percent Recovery Studies (GC/MS)

Percent recovery data for standard PGEZ by GC/MS is listed in Table

2-3. The columns tested in this study were those that had been found to

provide adequate retention (>90%) for PGE2 in the previous scintillation

counting experiments. In addition, this study shows recovery data for

another brand of octadecyl sorbent (Sep-Pak) and a very selective sorbent

using antibody-antigen interaction (immunoaffinity). As in the initial

recovery study (Table 2-2), three individual columns were each used to

extract three samples of standard PGE2. Three injections of each sample

were made into the GC/MS. The average of the three injections and the

three samples, in addition to the %RSD is listed in Table 2-3.

Comparing the recovery data in Table 2-2 to the data in Table 2-3,

the average recovery values for the C18, SI and 20H are similar. The

values determined by GC/MS are consistently 3-10% lower than those

determined by scintillation counting. However, this slight difference

could be attributed to the basic difference in calculating counting data

and areas of GC/MS. The variation between columns is again small (<5.0%)

for all sorbents tested. The immunoaffinity column demonstrated a percent

recovery (93.1%) quite adequate for retention of PGE2. Octadecyl columns

from two different suppliers were compared to examine differences in

retention and selectivity. The recovery data for the Sep-Pak (Waters)
















QM
0c 0









bo o

o
wt 0



















bo 4)
S> e e


4




r-4
Q-4
0'


4 l


0 4


en




CO




03

03
0'







0' Q~CA \


CMo 00 r-4 r-. CM


0~ 0t




S-4 I





o C C !





0-4 r-4 r- CM4 r-


mesQ


CM
C'!,
C"'.
-4
0'




LA
rl
40

-4


CM CA














S--
0













4




4v 41

0


-4 CM4 C


















0
0 U
4O

0
v


CM 0o
0% 0' r- cn 03 0 -0 '0
0 "%J O--*,
SC4 0 J 0 0 0 0







0 .- O M 000 % om O '
0 Or 0 0' 00 ON 00 c'








V 0\ 4 CV 0o 0o 4 1















>n0 oM c0 o n
' C0303 0 03 030 0'







0
r-4 04 Cn i-4 C4 r i-4 C4 fl CU

















0
a
'- *
Sig








M rd
^- I









48

columns were slightly lower than that found for the C18 (Analytichem)

columns. This may be attributed to experimental error in the extraction

procedure; however, reports suggest that Sep-Paks have considerable faults

compared to octadecyl sorbents from other manufacturers (108).

Table 2-4 contains the percent recovery results of standard PGE2

spiked into urine by GC/MS analysis. The columns tested in this study

were C18, SI, 20H and a combination of C18 plus immunoaffinity (IA).

Biological samples can be directly applied to the IA column; however, for

PG analysis Knapp and Vrbanac (78) have found an advantage in proceeding

the IA purification procedure with a C18 column extraction. The advantage

is that employing the C18 extraction first removes large concentrations

of extremely polar impurities found in urine which can non-specifically

bind to the IA column. Averages and %RSD are listed in the table for

three injections of each sample and the three extractions which were

performed on individual columns. The data indicate that the columns (SI

and 20H) which utilize polar interactions are not effective for retention

of PGE2 in urine, even though they were successful for PGE2 standards.

This is presumably due to competition for binding sites on the sorbent

between matrix components in the urine and PGE2. The non-polar C18 column

and the C18 column coupled with the IA column provided similar recoveries

for PGE2 in urine as in standards. The variation from column to column is

low for all cases (<7%) including the C18 and IA samples. In the case of

the IA column data, the same sorbent bed (or column) was used for all

three samples. The %RSD for the three samples, in addition to the three

average recovery values, demonstrate the reusability of the IA sorbent.

The slight decrease in the recovery values between samples 1-3 indicates

















0



w 3




S0 0
'>











0 1

W 0 0







(d


0\

CO


'0








en
00
('4

-:1





cn


00
cI c


(i N c T-4 CN A ,-4 C1M cn


14


00 0



0
1


cl
0

*r-4
*4


^-4
o0
o ..


cO
oN


r--





























o
,- O-4 0
C~ -*









i--0
tn cc 0'










CM NO f~-
00 00 00
CM

















-4














'Q 0
0
n o 0 r c u




























,-4
4 )














0
n-4 o' C-4











+ -


4
4U


-1
.
-0 < 4-
< > 4-
-- rc re
4J *


In gn ON 1 C4N Ln 0 cn -4

) 0 o so CM o r- cn
r-









50

that no carry-over of PGE2 occurred from sample to sample. Thus, the IA

sorbent can be reused for many urine samples in combination with a C18

column without loss of affinity for PGE2. The ability of IA to effectively

separate urine matrix components from PGE2 will be discussed in a later

chapter.


Derivatization Studies

Table 2-5 lists the results of the study of different derivatization

procedures. The GC/MS peak areas for three samples are listed along with

their average and XRSD. Each sample injected onto the GC column contained

500 pg of PGE2. Comparing the different results, the most effective method

of derivatization can be determined.

The first method listed followed the derivatization procedure

discussed in the experimental section. Comparing that method to a second

method, in which a more rapid methoximation at an elevated temperature (1

hr at 60C) was used, the peak area of method one was 1.4 times greater

when the 24 hour methoximation was employed. This suggests that at longer

reaction times more complete methoximation occurs. Recently, a study of

the methoximation of various PGs was reported in the literature (109).

These results showed that efficient methoximation of PGE2 by a procedure

similar to method one was 1.1 times greater than method two with a %RSD

of 11.8%. This corresponds to the values which were obtained in this

study. Another question addressed by this study is whether to perform the

methoximation step before or after the PFB esterification. Examining the

peak areas obtained for method three and comparing them to method one,

similar areas were calculated for 500 pg. The data suggest that either























3- 0 0 -


-4 r-4 r-







10 10 0 10 10
0 0 0 0 0






-4 -4 -4 -4 r-4










0 0 0 0 0

4 r-4 r-4 r r Y-4

xxx x x

1- 40 %0 CM4
0 -41 0 r-4 r-4

M ,-l CM .-


(N
C4

P4


0

0





0
0
}-A


NS








44















0
1J4
4)















4.-


0
-. .0 ou "0
"j r-4 CM f C d4
0


0
0


I
x
r4
X





-4








o
0 I







0
X





r-I

x


C4

iln
t













I'.
U, *


I
I
I
I
I
I
I
I
I
I

I

I
I

I
I
I
I
I
I
1
I
I
I
I
I
I
I
1
I
I




I
I
I


le 10 10 10
0 C0 0 0
0000
r-4 r-4 r-4 r-4

xx9x x

-4 W 4z 1
CN U 1 r- i







10 1O 1>0 10
0 0 0 0
r-4 r-4 r4 r-4




'0O C4 r-4 -4

-4 1-I CMN r-


a

4u



0
-4
41 In
Cd
0 0O















o 4


4 44
CO U0
0 0













0 0
4-4

0 "0

o a
-rI V C4





0.

C CG
0 0 Cu

4- -
u o u7












U J0

Z s-
4-) 4


0




5-I


4-1




0



-4 4 C. sa I




0
W 40







cc u



CSa < o
U 0 c0
m o 4.)








c0 0 -A

> 0






4J a 44
G-10 0 p

C4 0 4 0
p 0 0 0-l
a4 -4 4Z )

t> a0 u 0









52

step (methoximation or the PFB esterification) can be performed first.

The other question proposed in this study was the use of liquid-liquid

extraction to remove excess derivatizing reagents or the simpler, more

rapid method of only nitrogen evaporation. Two methods were studied, one

in which only nitrogen evaporation was performed after the methoximation

step, then liquid-liquid extraction after the esterification (method 4)

and the other in which only nitrogen evaporation was performed after all

derivatization steps (method 5). Comparing these two methods with method

one, the areas for 500 pg for both method four and five were approximately

40% less than method one. This suggests that removal of the excess

derivatizing reagents by liquid-liquid extraction is essential prior to

GC/MS, despite the increase in sample preparation time. The concentration

of PGE2 in urine is 100 to 400 pg/mL, thus having an effective

derivatization procedure which enhances the sensitivity of PGE2 is vital,

even at the expense of additional analysis time.


Conclusions


The optimum sample preparation for PGE2 in urine has been determined

in this chapter. Extraction and purification procedures as well as

derivatization steps have been investigated. The results from recovery

studies show that the use of either a C18 column or a combination of C18

and IA columns achieve adequate quantitative recoveries for PGE2 in urine.

Derivatization study results indicate that the time and temperature of the

methoximation reaction are important and appears to be optimum at longer

reaction times with lower temperatures. Performing either methoximation

or PFB esterification first in the derivatization procedure has little









53

effect on the area calculated for PGE2. Results from this study

demonstrate the advantage of using liquid-liquid extraction methods to

remove the excess derivatization reagents after each step in the

derivatization procedure.
















CHAPTER 3

OPTIMIZATION OF GC/MS AND GC/MS/MS CONDITIONS
FOR TRACE DETERMINATION OF PROSTAGLANDINS



Introduction


Many reports of GC/MS analysis of prostaglandins (PGs) can be found

in the literature (2,17,48,62,101). The conditions employed in each case

vary depending on the type of analysis (qualitative or quantitative),

sample matrix (urine, serum, etc.), and targeted concentration. In the

trace determination of PGs, optimization of conditions is critical.

Detection of low levels of PGs (100-400 pg/mL) in urine requires a

technique that is both sensitive and selective. The many parameters which

exist in GC/MS and GC/MS/MS can be varied according to the analysis to

enhance either sensitivity or selectivity. Thus, to achieve the proper

conditions for trace determination of PGs the various parameters must be

characterized and optimized.


Experimental


Prostaglandins and Reagents

All solvents were reagent or HPLC grade. Prostaglandin E2 (PGE2) and

prostaglandin F2a (PGF2a) was purchased from Sigma Chemical Co. (St. Louis,

MO). The derivatization reagents pyridine, 0-methylhydroxylamine

hydrochloride, acetonitrile and N,N-diisopropylethyl amine were all









55

purchased from Sigma Chemical Co. Pentafluorobenzylbromide (PFBBr) and

bis(trimethylsilyl)trifluoroacetamide (BSTFA) were purchased from Pierce

Chemical Co. (Rockford, IL). All glassware was silanized with a solution

of 5% dimethyldichlorosilane in toluene. These two chemicals were both

purchased from Sigma Chemical Co. Helium used as GC carrier gas, methane

(>99%) as chemical ionization reagent gas and nitrogen, argon and xenon

used as collision gases were from Matheson Gas Products, Inc. (Orlando,

FL).


Derivatization

The MO/PFB/TMS derivative of PGE2 and the PFB/TMS derivative of PGFZa

were prepared by the same procedure as in Chapter 2.


Instrumental Conditions

A Finnigan MAT triple stage quadrupole (TSQ45) gas chromatograph/

mass spectrometer was employed. Mass spectrometry conditions were:

interface and transfer line temperature 300*C, ionizer temperature 190C,

electron energy 100 eV and emission current 0.3 mA. GC was carried out

on a short J&W Scientific (Folsom, CA) DB-1 (3 m long, 0.25 mm i.d., 0.25

pm film thickness) capillary column in the splitless mode with helium

carrier gas at an inlet pressure of 4-6 psi. The initial temperature of

200*C was held for 30 s, then increased at 20C/min to 260*C. The

injector temperature was 300*C.

Both full scan mass spectra and selected-ion monitoring (SIM) were

used in the GC/MS studies. An electron multiplier (EM) setting of 900 V

was used for the full scan spectra and a preamp gain of 10-8 A/V. A

baseline was chosen visually on the GC trace and the area for PGE2 (524')









56

and PGF2a (569") calculated by the INCOS computer system in the SIM mode

of operation. In GC/MS/MS optimization studies full daughter spectra and

selected-reaction monitoring (SRM) were employed and areas calculated by

the same method as described for GC/MS. The EM was set at 1500 V for the

full daughter spectra obtained and a preamp gain of 108 V/A. In the

GC/MS/MS optimization the [M-PFB]" carboxylate anions of PGE2 and PGF2a

were selected in the first quadrupole (Ql) region and passed into the

collision cell (Q2). In this region these ions underwent CAD to form

characteristic fragments which were then mass analyzed in the third

quadrupole (Q3). A full daughter spectrum was acquired over the mass

range of 100-600 amu.

Calibration curves were prepared for both GC/MS and GC/MS/MS after

optimization of the various parameters. Selected-ion monitoring and

selected-reaction monitoring were used to determine linearity, precision

and limits of detection for standard PGE2 utilizing GC/MS and GC/MS/MS.

The EM was set at 1700 V for both the SIM and SRM calibration curve data

with a preamp gain of 108 V/A.


Mass Spectrometry (GC/MS)


Choice of the appropriate ionization method is essential for trace

determination of PGs by GC/MS. Electron ionization (El) has been reported

in the determination and identification of various PGs (42,43,57).

Structural information is obtained by this technique due to the abundance

of fragment ions which are produced. However, in trace analysis of PGs

the creation of a single ion with a maximized intensity is preferred.









57

Chemical ionization (CI) has the advantage of usually producing few

fragment ions and a very intense molecular ion. Many reports of chemical

ionization GC/MS for PG analysis appear in the literature (65-72). Both

positive and negative CI have been incorporated for PG analysis. The

literature reports that appropriate derivatization of PGs coupled with

electron-capture negative chemical ionization (EC-NCI) results in the

detection of low levels of PGs (69-72). These reports generally employ

methane (CH4) as a reagent gas for its ability to thermalize electrons.

Thus, both the GC/MS and GC/MS/MS optimization studies have utilized EC-

NCI with methane as the reagent gas.


Ionizer Pressure Study

Although the literature includes numerous examples of methane as a

reagent gas for EC-NCI, many dramatically different ion source pressures

have been employed. This study was performed to determine the optimum

ionizer pressure at which the [M-PFB]" ions of PGE2 and PGF2a are produced

in the ion source. Figure 3-1 shows the average areas determined at

different ionizer pressures of methane for the 524" ion (PGE2) and 569" ion

(PGF2a). The average of three one-microliter injections of a 100 pg/pL

solution of both PGE2 and PGF2a have been plotted on the graph. The

optimum ion source pressure for PGE2 and PGF2a occurs at 0.50 Torr of

methane.

At ionizer pressures lower than 0.50 Torr the [M-PFB]" ion has a

lower percent relative intensity compared to the reconstructed ion current

(RIC) for both PGE2 and PGF2,. As the ion source pressure is gradually

increased above 0.50 Torr, fragment ions begin to increase in relative
































(9OO
OLn








O O,
S01
0 D|


S0 0 0
0 3 O )



(s4unoo) 0L


I71


m
I

m


m-
\]


''LI Ijililil liii Ill


NI III l l II I l l I
0 .
C O


0
-0






0







0




0






ad
0












O
S0
N







o 00

d d


a6DJaAV


L_

0





-O



0






L
0


r-


0


0
0





0
j0
















a)
:C

































0




I.,
rLu




















a-4
:o





f.Is
34

.0)
;C




'3


01


a



o

i-j



3
03


m


m


* DaJV









59

intensity and contribute more to the RIG, thus decreasing the relative

intensity of the [M-PFB]" ion compared to the RIC.


Ionizer Temperature Study

Reports in the literature have cited ion source temperatures for

EC-NCI/GC/MS and EC-NCI/GC/MS/MS in the range of 110*C to 200C (69-78).

Figure 3-2 indicates the optimum ion source temperature observed in this

study for the analysis of PGE2 and PGF2a. Selected-ion monitoring (SIM)

of the 524' (PGE2) and 569" ion (PGF2a) was used over a range of ion source

temperatures from 100C to 190C. Three one-microliter injections of a

500 pg/pL solution of PGE2 and PGF2a were performed at ten different ion

source temperatures. The average of the three injections is plotted on

the graph. Both PGE2 and PGF2a have an optimum ion source temperature at

190*C. Thus, the [M-PFB]'carboxylate anion of PGEZ and PGF2a optimize at

the maximum ion source temperature of the instrument.

The percent relative intensity of the [M-PFB]" ion compared to the

RIC increases with an increase in the ion source temperature and reaches

a maximum at 190C. In addition fragment ions increase, as the

temperature is elevated to about 140C to 1500C, then these fragments

gradually decrease as the ion source temperature is raised above 150*C.

These two observations lead to the optimum ion source temperature of 190C

for the [M-PFB]" ion.


Electron-Capture Negative Chemical Ionization Mass Spectra

The mass spectra of standard PGE2 and PGF,, obtained at the optimum

ion source pressure and temperature, are shown in Figure 3-3a and Figure

3-3b. Both spectra demonstrate the advantage of employing EC-NCI for



























O0I
aT)o








0 0El
OO
0 0


0N C'


It i t I S .... ...ll ll i l lil i ll S lll li ll i ll il i |


III II I t I 1 illl Ill lIII
o 0 0
t 0


(stunoo)


D 0 0 0 0
0 LO 0 LO
N -

* D@Jv oBDJGAV


E

-E-q

0
0


O



a)
L



iF-
0
!-e-






CL
F--


0
a-


0
O0

c
0


.1~


























03




LL










-0
0- 0










o
I-,








2 o
0






o
--





-0










v N


eouepunqV eA!ilele %


4U
o


0
u

r4








*-4


0
1-4
--4
(d
U
-4
20


u

o






















b
.1.
aX-






U^
s-I




bU

-E4c
0) /^ /
*-i d
M 'w/'^

nCC
(^iP

P>hi
b0cr









62

prostaglandin analysis. One intense peak, the [M-PFB]" ion, dominates

each mass spectrum. This ion, PGE2 (524') and PGF2a (569"), can be utilized

for SIM. Other low intensity fragment ions can be seen in the mass

spectrum of PGE2, corresponding to the loss of derivatives attached to

PGE2. In addition, no fragments of greater than 1% relative abundance are

observed in the mass spectrum of PGF2..

The M" ion for both PGE2 (m/z 705) and PGFZ~ (m/z 750) is rarely

present in the EC-NCI mass spectra, thus it must be less than 0.1%

relative abundance. In addition, the PFB' ion (m/z 181) occurs in the EC-

NCI mass spectra of both PGs at less than 0.5% abundance.


Selected-Ion Monitoring Calibration Curve

A calibration curve for PGE2 (524") is shown in Figure 3-4. This

curve indicates the linearity and limit of detection for PGE2 with SIM.

Three one-microliter injections at nine different concentrations were

performed. The limit of detection was calculated from the calibration

data and corresponded to the amount of PGE2 which could give a GC peak area

three times greater than the average area obtained with a derivatized

blank. The use of SIM with standard PGE2 produced a limit of detection of

approximately 94 fg (femtograms) and is indicated on the curve. The

calibration curve showed good linearity above the limit of detection in

the range of concentrations expected for endogenous PGE2 in urine (100 to

400 pg/mL) (71,78). The linear dynamic range of the curve is from 500 fg

to 1 ng (solid line) and the slope of the linear regression best fit line

is 1.266 with a correlation coefficient of 0.9925. The non-linearity at

the low end of the calibration curve may be due to adsorption on the































































LIII' I I I I 1111111 I I


I"1 1 I 1 I i f 1 1 I I l l I I

00 O
0 B


I 1111111r

0


(slunoo) ~3Od ;o


I




O0

ro
















-
-o



.T



~ C


1' ro



0 0 -


DoJV


c-

0
(-D




.D
L




0



CQ
O
O





C-


O
o 5



t 0
41
"(-





0 4
00






"C-
o0
rc

o.





ij

Lw
r
*i-


C:
4- 0
0..- 4m
-r-
r\









64

column, septum or injection port and subsequent adsorption by the next

injection. Precision of the GC/MS method utilizing SIM was determined by

performing ten one-microliter injections of a 50 pg/pL solution of PGE2.

The percent relative standard deviation (%RSD) of the ten injections was

5.5%. Calibration curves for PGF2a, PGD2 and DHKF2a were similar, with

varying limits of detection, in the range of 50 to 200 fg.

The results for this study agree well with the literature. Reports

have shown LODs using EI/MS at about 20 pg for the M+* (6,17,48,62,101) and

utilizing EC-NCI/MS about 100 fg (19,20,76,77) for the [M-PFB]" ion.


Tandem Mass Spectrometry (GC/MS/MS)


Monitoring the efficiencies of the collisionally activated

dissociation processes can help in determining optimum MS/MS conditions

for trace determination of PGs. These efficiencies are affected by

collision energy and collision gas pressure. Either parameter can be

varied to maximize the CAD efficiency for a particular parent ion.

Increasing the collision energy allows for more energetic collisions,

while increasing the collision gas pressure increases the number of

collisions each ion experiences.


Collision Gas Pressure Studies

The collision gas pressure for three different gases (N2, Ar, and Xe)

was optimized to determine which collision gas and pressure were the most

efficient for selected-reaction monitoring at the maximum available

collision energy of 30 eV. Pressure-resolved breakdown curves for

selected ions of the MO/PFB/TMS derivative of PGE2 are shown in Figures












Nz Collision Gas
Collision Energy 30 eV


(a) l
o
















(b)
-S
+
















Q.
+
5-
*B


Xe Collision Gas
Collision Energy 30 eV

ooooo 524--> 434-
oooo_ 524--- 344-
A-Aa 524--> 313-
QO...0 524--* 268-


Figure 3-5:


Pressure-resolved breakdown curve of the carboxylate
anion of the MO/PFB/TMS derivative of PGEz with
collision gas: (a) Nitrogen (b) Argon (c) Xenon


Ar Collision Gas
Collision Energy 30 eV










30

25

20
(a)
.
+ it

10.,
*Q


Optimum -


ot 3 i..7~


0.0 1.0
Collision Gas


(b) s

+

















0-
(








+


Figure 3-6:


Optimum


N2 Collision Gas
Collision Energy 30 eV



Q0o0 569"-* 389-
[aQnm 569--, 317-
AAAAA 569--* 299-
^ 569"-- 273-


2.0 3.0 .4.0
Pressure (mTorr)


Ar Collision Gas

Collision Energy 30 eV



Qoo o 569--* 389-
Qrano 569--* 317-
-AAAA 569--* 299-
4 0AA0 569-- 273-


0.0 1.0 2.0 3.0 4.0
Collision Gas Pressure (mTorr)


Xe Collision Gas

Collision Energy 30 eV

oQQ.O_ 569--* 389-
ooooo 569-- 317-
A&-a 569--+ 299-
.0O.0 569-- 273-


Pressure-resolved breakdown curve of the carboxylate
anion of the PFB/TMS derivative of PGF, with
collision gas: (a) Nitrogen (b) Argon (c) Xenon









67

3-5a, 3-5b and 3-5c and for the PFB/TMS derivative of PGF2a in Figures 3-

6a, 3-6b and 3-6c. The optimum collision gas pressure is indicated on

each curve. This type of curve can be calculated by dividing the area of

a selected daughter ion by the area of all the ions in the daughter

spectrum (Di/[ZDi + P]) at each pressure. The point which is chosen as the

optimum is the pressure where one can obtain a qualitative daughter

spectrum which is "rich" in structural information with a number of

reasonably abundant daughter ions.

The optima indicated on Figures 3-5a, 3-5b and 3-5c occur at a

collision pressure of 0.5, 1.0 and 0.5 mTorr for N2, Ar, and Xe

respectively. However, in the case of PGF2a, the optima occur at

significantly higher pressures, 2.5, 1.5, and 1.0 mTorr for N2, Ar, and Xe

(Figures 3-6a, 3-6b and 3-6c). Comparing Figure 3-5a to 3-6a, the optimal

use of nitrogen as a collision gas requires a pressure five times higher

for PGF2, than for PGE2. This dramatic difference in optimum collision

pressure exists between two structurally similar PGs.

The relative differences in the optimum collision gas pressure for

PGF2a with the various collision gases can be explained by the relative

mass of the three different collision gases. The greater size of the

argon and xenon gas molecules increases the energy deposited into the

parent ion, therefore increasing the fragmentation efficiency. Therefore,

the optimum collision pressure decreases with an increase in the mass of

the gas molecules, because the abundance of the most prominent fragment

ions occur at lower pressures.

Notice the significant differences in the maximum relative intensity

of the daughter ions for the three collision gases. The curves with









68

nitrogen (Figures 3-5a and 3-6a) show a higher maximum relative intensity

for the selected reactions listed. Figure 3-6c for PGF2a, utilizing xenon

as the collision gas, is particularly interesting. The relative intensity

of the daughter ions selected are 12 times lower than the intensity of the

same daughter ions displayed in Figure 3-6a for nitrogen. In addition,

the relative intensity of the daughter ions approach zero at higher

collision gas pressures (> 1.0 mTorr). Thus, either the parent ion

carboxylatee anion) is increasing at higher pressures or other daughter

ions are more abundant with argon and xenon at these collision gas

pressures. Clearly, for the CAD process, the parent ion decreases as the

collision gas pressure is increased. Therefore, various daughter ions not

listed in these figures must be prominent with argon and xenon at higher

collision gas pressures.

This is apparent from examining Figure 3-7, which shows the

pressure-resolved breakdown curves for different selected ions with the

PFB/TMS derivative of PGF2a with argon and xenon as collision gases.

Comparing Figure 3-7a and Figure 3-7b to Figure 3-6a the relative

intensity of the daughter ions for argon and xenon in this case are

similar to that found for nitrogen. The most intense reaction in both

cases in Figure 3-7 is the selected-reaction of 569 -* 89", corresponding

to a back-bone fragmentation.

A general trend appears in all the figures for both PGE2 and PGF2,.

The loss of one and two HOTMS groups from the [M-PFB]" ion tend to maximize

together at low collision gas pressures for the three collision gases.

Subsequently, these selected-reactions gradually decrease towards zero at

higher collision gas pressures. The four selected-reactions for PGE2 in











Optimum-










J^T


0.0 1.0
Collision Ga


(a)



















(b)


Ar Collision Gas
Collision Energy 30 eV



ooooo 569-- 255-
aoooo 569--+ 215-
AAAAA 569--+ 161
040.O0 569 -- 89


2.0 3.0 4.0
is Pressure (mTorr)





Optimum


Xe Collision Gas
Collision Energy 30 eV


oooo 569-- 255-
o0ooo 569 215-
AnaA 569-4 161-
0_0_0 569-* 89-


0.0 1.0 2.0 3.0 4.0
Collision Gas Pressure (mTorr)


Figure 3-7: Pressure-resolved breakdown curve of the carboxylate
anion of the PFB/TMS derivative of PGF, with
collision gas: (a) Nitrogen (b) Argon


04-
10-


5-


II \


30-




, 20-

+1
10


"-


L









70

Figure 3-5 all increase at low collision gas pressures and then gradually

decrease as higher collision gas pressures are employed. In the case of

PGF the additional loss of the third HOTMS group (569" 299-) gradually

increases when nitrogen is utilized (Figure 3-6a) or levels off when argon

(Figure 3-6b) is employed, as the collision gas pressure is continually

increased. In Figure 3-7, the selected-reactions have relatively low

intensities at low collision gas pressures (< 1.0 mTorr), but increase

gradually and level off as the collision gas pressure is increased (> 1.0

mTorr).

Trace analysis by selected-reaction monitoring with MS/MS requires

optimization of the absolute intensity of a single daughter ion of the

selected parent ion. The curves in Figures 3-8 and 3-9 give an indication

of the optimum reactions and collision gas pressures which should be

selected for maximum SRM sensitivity for PGE2 and PGF2Z with three

different collision gases at maximum collision energy (30 eV). This type

of curve is calculated by dividing the area of selected daughter ions, Di,

by the area of the incident parent ion, Po (measured in a daughter spectrum

without collision gas). The reaction with the highest CAD efficiency

should be selected to yield the highest sensitivity for selected reaction

monitoring (SRM) trace determination of PGs. For example, in the case of

PGFZ,, choice of the 569- 299" selected reaction with argon (Figure 3-

9b) would be the optimum (overall efficiency of -2%) at a collision

pressure of 1.5 mtorr and collision energy of 30 eV. This reaction

corresponds to the [M-PFB]- [(M-PFB) 3(HOTMS)]'for the derivatized

carboxylate anion of PGF2a. Note that this overall CAD efficiency (-2%)

is obtained at the optimum pressure for any of the three gases. The more

















(a) 0
















(b)


(C)


Figure 3-8:


Ar Collision Gas
Collision Energy 30 eV
ooooo 524--* 4341
oo ao 524-- 344 -
AAA 524--" 313
60060 '94.--+ 9fiFR


Optimum


Xe Collision Gas
Collision Energy 30 eV
o0000 524-- 434-
0ooo. 524--, 344-
4AA-a 524--* 313-
00.00 524--* 268-


0.0 1.0 2.0 3o.
Collision Gas Pressure (mTorr)


Overall CAD efficiency for the selected-reaction
monitoring of the.carboxylate anion of the MO/PFB/TMS
derivative of PGEz with collision gas:
(a) Nitrogen (b) Argon (c) Xenon













Collision Energy 30 eV


(a) *
a-

















(b) .
a-


Gas Pressure (mTorr)


- Opti


mum


.Qoooo 569--* 389-
ooooo 569--* 317-
&AAaa 569--+ 299-
0.0 569--* 273-

Xe Collision Gas
Collision Energy 30 eV


2.0 3.0 4.0
Pressure (mTorr)


Figure 3-9:


Overall CAD efficiency for the selected-reaction
monitoring of the carboxylate anion of the PFB/TMS
derivative of PGF. with collision gas:
(a) Nitrogen (b) Argon (c) Xenon


Collision Energy 30 eV

QoQOo 569--p 389-
ooooo 569-- 317-
aaA 569-- 299-
000.0 569-- 273-


(c) o
a.
*-


1.5


1.0


0.5


0.0
0.1
C


0 1.0
collision Gas









73

massive the collision gas, the lower the optimum pressure. PGF2a exhibits

a slightly higher overall CAD efficiency (-2%) with xenon for the selected

reaction of 569" 317" (Figure 3-9c). This reaction corresponds to the

[M-PFB]" [(M-PFB) 2(HOTMS) (CH3)2Si-CH2]" for the derivatized

carboxylate anion of PGF2,. This suggests that xenon would be the optimum

CAD gas. However, xenon is quite expensive ($650/50 L of gas) and the

gain in CAD efficiency is slight; thus, argon would be a more practical

choice.

For PGE2, the [M-PFB]" [(M-PFB) 2(HOTMS) CO2 HOCH3]" reaction

with argon (Figure 3-8b) is optimum at a pressure 2 times lower than for

PGF2a. Even more notable is that the optimum CAD efficiency (Di/Po) for

PGE2 (-10%) is significantly higher than that for PGF2, (-2%) with all

three gases employed. Note that on Figures 3-8b and 3-8c the lowest

collision gas pressure been plotted is 0.2 mTorr. This is due to the fact

that when the zero collision gas pressure data was obtained, the actual

pressure of the collision cell was 0.2 mTorr. This indicates that

residual collision gas was present in both these cases, thus allowing

residual CAD to occur.


Collision Energy Study

Argon was chosen as the collision gas for the optimization of

collision energy. The collision energy study for selected-reaction

monitoring is shown in Figure 3-10. Data for SRM with the reactions and

Argon pressure chosen as optimum in the collision gas pressure studies at

30 eV for PGE2 and PGF2, are plotted on the same graph. Three one-

microliter injections of a 500 pg/pL solution of PGE2 and PGF2, were
















0 I


(NC OO. C O
O C' C- u





0O aO
0 <
E 0 0
LOO E O


E, E







oo C',
4,-4
CLL 4






00


O ].
0


0 0 E
0 c} 0 E 10


















(siunoo) )wI
0 0*



r.o


r >
00 0 3

( u0 0









75

performed at collision energies of 7 to 28 eV. Optima for the selected

reactions of 524 -, 268' (PGE2) and 569 -* 299" (PGF2,) are indicated on the

graph. The optimum collision energy for PGE2 (25 eV) is slightly higher

than for PGF2. (22 eV), probably due to the difference in the collision

pressures employed for each selected reaction. These plots suggest that,

once a particular collision gas, pressure, and selected reaction are

chosen, variation of the collision energy has little effect.


Daughter Spectra of Standards

Figures 3-11a and 3-11b show daughter ion spectra for the

carboxylate anions of the MO/PFB/TMS derivative of PGE2 and the PFB/TMS

derivative of PGF2a. The fragment ions labeled in the spectra are

tabulated in Table 3-1 with possible assignments of the ion's identity.

Most of the fragment ions observed in both daughter spectra are

derivative-specific. These ions occur at m/z 434, 344 and 313 for PGE2 and

at m/z 479, 389, 317 and 299 for PGF2a. The most intense daughter ion from

the fragmentation of the [M-PFB]" ion -of PGE2 is m/z 268 ion and

corresponds to the loss of (2*HOTMS-CO2-CH3OH) from the parent ion of 524'

In the daughter spectrum of PGF2a the m/z 299 ion is the most intense

and corresponds to the loss of three HOTMS groups from the parent ion of

569". The fragment ions which occur at lower masses are ions that

correspond to backbone-specific fragments. This means these ions

correspond to fragmentation of the carbon-hydrogen skeleton in both PGE2

and PGF2a. These ions include m/z 240 and 214 for PGE2 and m/z 255, 215,

201 and 161 for PGF2a.
























SCO\
I~ w- 0


C O





(0

I J


eouepunqv eGAIeleB %


| I I l a s


0




-CM


0
-0
So


C



0
-0








-o
. ,









0

-0
CO






















N


44
O C)






U' o
0 $3



r.E CE $4


0o 0
c)






o;









4-4

0 >

EA4
Cu




















0
41
-4-
44
5-S

-4S







-4m
3 ^ -'
a ( ,
o ^^ ^








l-i
L1
*i-
tn












Table 3-1: Daughter Ions of [M-PFB]- (P') of
PGE2 MO/PFB/TMS and PGF2a PFB/TMS


PGE MO/PFB/TMSa

Ion Assignment m/z %RAc

P- 524 5
[P-HOTMS]' 434 2
[P-2HOTMS]' 344 21
[P-2HOTMS-OCH3]" 313 32
[P-2HOTMS-CO2] 300 4
[P-2HOTMS-CO2-CH220] 270 11
[P-2HOTMS-CO2-CH30H]" 268 100
[P-2HOTMS-CO2-CH3OH-C2H4 ] 240 5
[P-2HOTMS-CO2-CH3OH-C4H6 -] 214 12


PGF2, PFB/TMSb
Ion Assignment m/z %RA

P- 569 36
[P-HOTMS]" 479 4
[P-2HOTMS]' 389 18
[P-2HOTMS- (CH3)2SiCH2]" 317 24
[P-3HOTMS]- 299 100
[P-2HOTMS-CO2- (CH3)2Si-CH2- ] 273 56
[P-3HOTMS-CO2]" 255 36
[P-3HOTMS-CO2-C3H4]' 215 21
[P-3HOTMS-C02-C4H6 ] 201 12
[P-3HOTMS-C02-O-CyH] 161 18

a At a collision gas pressure of 1.0 mTorr argon
and collision energy of 28.0 eV.

b At a collision gas pressure of 1.5 mTorr argon
and collision energy of 28.0 eV.


c % Relative Abundance









78

Considerations for choice of a particular selected reaction for

monitoring has been discussed recently by Strife (77). This report along

with other studies have shown that backbone-specific fragmentation confers

superior selectivity over derivative-specific fragmentation in the

analysis of biological samples. For example, when SRM is based on a

derivative-specific fragmentation, any component in a SIM chromatogram

that is derivatized has an enhanced probability of appearing in a SRM

chromatogram. However, the backbone-specific fragmentation has a lower

relative daughter ion intensity in EC-NCI/GC/MS/MS than for derivative-

specific fragmentation. Therefore, if the backbone-specific fragmentation

was chosen for SRM analysis the sensitivity would be lower than that for

the derivative-specific fragmentation found in Figure 3-8 and Figure 3-9

for PGE2 and PGF2,.


Selected-Reaction Monitoring Calibration Curve

A selected-reaction monitoring (SRM) calibration curve for PGE2

(524" 268") is shown in Figure 3-12. This curve indicates the linearity

and limit of detection for PGE2 with SRM. Three one-microliter injections

at seven different concentrations were performed. The limit of detection

was calculated from the calibration data, the corresponding amount of PGE2

which gives a GC peak area three times greater than the average area

obtained with a derivatized blank. The use of SRM with standard PGE2

produced a limit of detection of approximately 14 pg, as indicated on the

curve. The calibration curve showed good linearity above the limit of

detection in the range of concentrations expected for endogenous PGE2 in

urine (100 to 400 pg/mL) (71,78). The linear dynamic range of the curve






















































Cc
--0
- 0 .i

E I


III! I I


111i11 I I I
no


o 0


(slunoo) QOd jo


O0







-C

F 0


0
-0


CLt


60


t

c'.J
C4
Lf,
0)


0
4-




c4
0
0
"4


C.aC
2 >







0 0
kW
0



4c1






S-4
41)
(D

w 4


DGoajV









80

is from 20 pg to 5 ng (solid line) and the slope of the linear regression

best fit line is 1.007 with a correlation coefficient of 0.9989. Precision

of the GC/MS/MS method utilizing SRM was determined by performing ten one

microliter injections of a 500 pg/pL solution of PGE2. The percent

relative standard deviation (%RSD) of the ten injections was 3.9%.

Calibration curves for PGF2g, PGD2 and DHKF,2 were similar, with varying

limits of detection in the range of 5 to 30 pg. The results for this

study agree well with literature reports, which have shown LODs utilizing

EC-NCI/MS/MS of about 1 to 20 pg (19,20,76,77).


Conclusions


The optimum conditions for GC/MS electron-capture negative chemical

ionization (EC-NCI) with SIM and GC/MS/MS with SRM are summarized in Table

3-2. The optimum collision gas pressure for both qualitative and

quantitative (SRM) analysis of PGE2 are lower than the optima found for

PGF2a. The dramatically lower CAD efficiency for the carboxylate anion

of PGF2a (Figure 3-8b) compared to that of PGE2 (Figure 3-7b) clearly

indicates its greater stability under CAD conditions.

This study demonstrates the need for evaluating the CAD efficiency

in the trace analysis of PGs. Optimization of both collision energy and

collision gas pressure is essential in obtaining an accurate qualitative

daughter spectrum "rich" in structural information. The CAD reaction with

the highest CAD efficiency should be selected to yield the sensitivity for

SRM determination of PGs.

Examining the calibration curves (Figures 3-4 and 3-11) differences

between SIM and SRM are noted. Sensitivity is greater with SIM than with





















Table 3-2: Otimum Conditions for Electron-Capture Negative
Chemical Ionization Mass Spectrometry
and Tandem Mass Spectrometry



Electron-Capture Negative Chemcial Ionization
Pararmeter PGE2 PGF2a
Ion Source Pressure 0.50 Torr 0.50 Torr
Ion Source Temperature 1900C 190C


Tandem Mass Spectrometrv


Qualitative Daughter Spectrum
Parameter PGE2
Collision Gas Argon
Collision Gas Pressure 1.0 mTorr
Collision Energy 28.1 eV


PGF2a
Argon
1.5 mTorr
28.1 eV


Par
Colli
Selecte
Collision
Collis


Quantitative Selected-Reaction Monitor
ameter PGE2
sion Gas Argon
d Reaction 524' 268' 5
Gas Pressure 1.0 mTorr 1.
ion Energy 25.0 eV


ing
PGF2
Argon
69' 299"

.5 mTorr
22.2 eV









82

SRM. The limit of detection for SIM (94 fg) is slightly more than 2

orders of magnitude lower than for SRM (14 pg). Comparing the relative

peak areas of SIM and SRM at 20 pg the SIM peak area is approximately 10

times higher than the peak area of 20 pg with SRM. In addition, at higher

levels of PGE2, 500 pg, the SIM peak area is approximately 40 times higher

than the SRM peak area. The lower sensitivity of SRM is expected due to

the limited efficiency of the CAD conversion of the parent ion to the

daughter ion (approximately 12% for 524' 268-) of interest, as well as

transmission losses inherent in adding a second stage of mass analysis

(typically 10 times). However, the selectivity gained by the parent-

daughter reaction should reduce the chemical noise, in a sample matrix,

to a greater extent than the analytical signal, thus, compensating for the

lost sensitivity.















CHAPTER 4

DIFFERENCES IN THE COLLISIONALLY ACTIVATED DISSOCIATION OF
STRUCTURALLY SIMILAR PROSTAGLANDINS



Introduction


The ions formed by electron ionization (El) of the methyl ester/

methoxime/trimethyl silyl ether derivatives of prostaglandins (PGs) show

considerable fragmentation in the collisionally activated dissociation

(CAD) process (74,75). However, the carboxylate anions of certain PGs

produced by EC-NCI have been reported to be extremely stable when

subjected to CAD (76).

It has been observed that the carboxylate anions of certain PGs

exhibit little fragmentation even at high collision energies (>20 eV) and

pressures (1.5 mTorr N2). Subtle differences among the structures of

prostaglandins E2 (PGE2), F2. (PGF2.), D2 (PGD2) and 13,14-dihydro-15-keto

F2a (DHKF2.) (Figure 4-1) yield enormous differences in CAD efficiency.

The CAD efficiency for the [MO/TMS-PFB]', [M-PFB]" and [M-H]" carboxylate

anions is significantly different for closely related PGs. The low

fragmentation and CAD efficiencies of the carboxylate anions of PGF,2 and

DHKF2 compared to those of PGE2 and PGD2 clearly indicate the greater

stability of these species. In this chapter these differences are

evaluated and explained in relation to the structural differences between

the carboxylate anions for the PGs.











(a)


(b)


(c)


0 O
OH


(d)


Figure 4-1:


Structures of the four prostaglandins studied:
(a) Prostaglandin F2a (PGF2.) (b) Prostaglandin E2 (PGE2)
(c) Prostaglandin D2 (PGD2) (d) 13,14-dihydro-15-keto F2,
(DHKF2,)


COOH








COOH







COOH








COOH









85

Experimental


Prostaglandins and Reagents

The prostaglandins E2, F2a, D2 and 13,14-dihydro-15-keto F2Z, as well

as 0-methylhydroxylamine hydrochloride, N,N-diisopropylethyl-amine,

pyridine, and acetonitrile (analytical grade) were all purchased from

Sigma Chemical Co.. Pentafluorobenzylbromide (PFBBr) and bis(trimethyl-

silyl)trifluoroacetamide (BSTFA) were purchased from Pierce Chemical Co..

The methane (>99%) used as the chemical ionization reagent gas was

purchased from Matheson Gas Products, Inc.. Helium used as GC carrier

gas and nitrogen used as CAD collision gas were commercial grade, with

their purity checked by mass spectrometry.


Derivatization

The methoxime/pentafluorobenzyl ester/trimethylsilyl (MO/PFB/TMS)

derivatives (Figure 4-2) formed for the GC/MS/MS studies were prepared

according to the method in chapter 2. The trimethylsilyl derivative was

formed by adding 100 pL of BSTFA and allowed to stand overnight at room

temperature. Dilutions were made from this solution so that a 500 pg/pL

solution of each PG was used for injections. The solids probe/MS/MS

studies were performed either by analyzing the standards without

derivatization or as the PFB derivative, using only the PFBBr

esterification step above.










TMSO


(a)


OTMS


CH3O-N


(b)


TMSO




TMSO


OTMS


(c)


CH30-N


TMSO


COO-PFB




co -w



COO-PFB










COO-PFB


OTMS


(d)


TMSO


Figure 4-2:


Structures of the methoxime-pentafluorobenzyl-
trimethylsilyl (MO/PFB/TMS) derivatives of the four
prostaglandins:
(a) PGF2. (b) PGE2 (c) PGD2 (d) DHKF2a











Instrumental Conditions

GC was carried out on a short J&W Scientific (Folsom, CA)

DB-1 (3 m long, 0.25 mm i.d., 0.25 pm film thickness) capillary column in

the splitless mode with helium carrier gas at an inlet pressure of 4-6

psi. The initial temperature of 200*C was held for 30 s, then increased

at 20*C/min to 260C. The injector temperature was 300'C. One-microliter

injections of a 500 pg/pL solution of each PG were made in triplicate at

each condition for the GC/MS/MS studies.

The solids probe was used as the means for sample introduction to

study the PFB ester derivatives and the free (underivatized) PG standards.

The initial temperature was 50C and increased at 20C/min to 300.

Triplicate samples were analyzed for each derivatization procedure at each

condition for the MS/MS studies. Sample size was one microgram of the

underivatized PGs or 1 ng of the PFB ester derivatives.

A Finnigan MAT TSQ45 gas chromatograph/triple quadrupole mass

spectrometer was employed. Mass spectrometry conditions were: interface

and transfer line temperature 300C, ionizer temperature 190*C, electron

energy 100 eV and emission current 0.3 mA. Electron-capture negative

chemical ionization (EC-NCI) was carried out with methane at an ionizer

pressure of 0.45 torr.

In the MS/MS experiments, nitrogen collision gas pressure and

collision energy were varied depending on each experiment. The [MO/TMS-

PFB]-, [M-PFB]- carboxylate anions were selected in the first quadrupole

(Ql) region and passed into the collision cell (Q2). In this region these

ions underwent CAD to form characteristic fragments which were then mass

analyzed in the third quadrupole (Q3) region. A full daughter spectrum









88

was acquired over the mass range of 55-600 amu. The maximum collision

energy possible on the TSQ 45 is 30 eV.

The peak areas in the daughter spectra of selected daughter ions and

the parent ion remaining after CAD were calculated by the INCOS computer

system for each GC and solids probe sample. A baseline was chosen

visually and the calculated areas were used for determining CAD

efficiencies.


Efficiency Calculations


The abundance of the daughter ions relative to that of the remaining

parent carboxylate anion in the daughter spectrum can be controlled by

varying the CAD energy or pressure; these parameters also affect

sensitivity due to scattering losses. The processes of fragmentation and

scattering can be monitored by evaluating the fragmentation (EF),

collection (EC), and overall CAD (ECAD) efficiencies given by the

following equations (110):


EF D- fraction of ions present
i following CAD which are
daughter ions


EC P + i-- fraction of initial parent
O ions that is collected
following CAD as either
parent or daughter ions


E i-- fraction of initial parent
CAD P ion that is converted to
collectable daughter ions









89

where Po, P, and Di are the intensities of the parent ion prior to CAD, the

parent ion remaining after CAD, and a daughter ion resulting from CAD,

respectively. Note that ECAD EF x EC'

As was stated earlier in chapter 3, the above efficiencies are

affected by collision energy and collision gas pressure. Either parameter

can be varied to maximize the CAD efficiency for a particular parent ion.

Increasing the collision energy allows for more energetic collisions,

while increasing the collision gas pressure increases the number of

collisions each ion experiences. Either approach increases the amount of

energy deposited into the parent ion, and thereby increases the

fragmentation efficiency. However, an increase in collision energy or

pressure will produce an increase in scattering losses (or possibly other

loss mechanisms such as neutralization by charge exchange) and thereby

decrease collection efficiency. The overall CAD efficiency, as the

product of fragmentation and collection efficiency, will typically first

increase then level off and even decrease as the collision energy or

pressure are increased. Systematic variation of each parameter would

provide a three-dimensional plot of efficiency vs. energy vs. pressure.

Practically, such studies involve two-dimensional slices through this

three-dimensional surface, varying one parameter while keeping the other

constant.


Collision Energy Study of the [MO/TMS-PFB1- Carboxylate Anions


In light of the dramatic differences observed in Chapter 3 of the

CAD efficiencies of the carboxylate anions of the fully derivatized PGF2a

and PGE2 (differing structurally only in the presence of a carbonyl (-0)









90

group at C-9 derivatized to a methoximine, in PGE2, rather than a hydroxyl

(OH) group at the C-9 position, derivatized to a trimethylsilyl group, in

PGF2.) two PGs similar to these were studied. The PGD2 and PGE2 isomers

vary only by the interchange of the hydroxyl (OH) and carbonyl (-0) groups

on C-9 and C-ll. DHKF2. differs from PGF2a by exchange of a carbonyl (-0)

group (derivatized to a methoximine) for the hydroxyl (OH) group

(derivatized to a trimethylsilyl) at C-15.

Figure 4-3 and Figure 4-4 presents curves for fragmentation,

collection and overall CAD efficiencies versus collision energy for the

carboxylate anions of PGE2, PGF2a, PGD2 and DHKF2G. These curves show the

effects of varying the collision energy at two different collision gas

pressures. In Figures 4-3a, 4-3b and 4-3c, the collision pressure has

been established at 1.2 mTorr N2, a value which is typically optimum for

many MS/MS analyses. The fragmentation efficiency curve (Figure 4-3a)

indicates the dramatic differences in stability of the carboxylate anions

of the four PGs. At a collision energy of 30 eV the fragmentation

efficiencies ranges from a typical 80% down to only 2%. The collision

pressure must be increased to produce more efficient fragmentation. The

collection efficiency (Figure 4-3b) for the four PGs are similar. The

notable exception is PGD2, which has an unusually high collection

efficiency at collision energies of 10-25 eV. This explains the

difference noted between the fragmentation efficiency of PGE2 and PGD2

compared to their overall CAD efficiency.

Figures 4-4a, 4-4b and 4-4c show the overall CAD, collection and

fragmentation efficiencies at a collision gas pressure 2.5 times higher.

The fragmentation efficiency (Figure 4-4a) as well as the overall CAD















(a)
















(b)
















(c)










Figure 4-3:


-o 80 Collision Pressure
5 1.2 mTorr N2
e.,
+ /QQQ PGF22a
S40 oooo PGEz
-- AAAAA PGD2
4.- 50J00 DHKF2z
W 20


0.0 10.0 20.0 30.0

600


80


so

a-
+ 40








Collision Pressure
2 1.2 mTorr N2














0.0 10.0 20.0 30.0
Collision Energy (eV)

CAD efficiency of the [MO/TMS PFB]" carboxylate anions
versus collision energy at 1.2 mTorr N2:
(a) Fragmentation
(b) Collection
(c) Overall CAD
















(a)


(b)












(c)








Figure 4-4:


Collision Pressure
3.0 mTorr N2


Collision Pressure
3.0 mTorr N2


Collision Energy (eV)


CAD efficiency of the [MO/TMS PFB]' carboxylate anions
versus collision energy at 3.0 mTorr N2:
(a) Fragmentation
(b) Collection
(c) Overall CAD