uvwmu-A yr” Q-Ow-onnwv-mv a“

n. \IflIlF-‘uuqvuu - .u

.hncne 'IA‘4-

.‘..’..n ”q“.

«a

. ;‘_..—)... ,.

.\.|

.,.. .(...V..

vh-un u?"

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 
 

 

   
 
 

 

 

 
 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 
 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

‘ M313

1
(W ’3

lGANST Ll RARI

mm ”willMillllwillIlmmlilmlml

3 1293 01801 7040
Michigan State
University

 

 

 

 

This is to certify that the

dissertation entitled

An Exploration of the MALDI Experiment by
Power Titrations, Mixture Analysis, and Prompt
Fragmentation Analysis

presented by

Gary Nelson Lavine

has been accepted towards fulfillment
of the requirements for

Ph . D . degree in mm

 

90L fl/fl»;
/

Major professor

Date /2./0. 78

 

MS U is an Affirmative Action/Equal Opportunity Institution 0-12771

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

 

DATE DUE

DATE DUE

DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

mo WM“

An Exploration of the MALDI Experiment by Power Titrations, Mixture Analysis,
and Prompt Fragmentation Analysis

BY

Gary Nelson Lavine

A DISSERTATION

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

DOCTOR OF PHILOSOPHY

Department of Chemistry

1 998

ABSTRACT

An Explorationof the MALDI Experiment by Power Titrations, Mixture Analysis,
and Prompt Fragmentation Analysis

By

Gary Nelson Lavine

The technique of power titrations was used to study the matrix assisted
laser desorptionflonization (MALDI) experiment. Power titrations are performed
by irradiating a single spot on a MALDI sample and increasing the laser power at
regular intervals. Component spectra are downloaded to disk so that sets of
spectra with the same laser power are saved together. The resulting spectra are
then displayed in the order that they were collected. By examining the results
from power titrations the onset of different peaks can be compared. The onset Of
all of the peaks produced are very highly correlated with each other.

Power titrations have been found to be the most useful for the evaluation
of MALDI samples prepared by different methods. Even small differences in the
sample preparation method used can cause significant changes in the spectra
that are produced. The dried droplet method, the Vorm method, the crushed
crystal method, and a “combination” method were all evaluated with power
titrations. Of these methods, the crushed crystal method was found to produce
the best analyte signals for a wide variety of peptides. Use of the power titration

procedure also highlighted the difficulty of producing reproducible samples.

Another aspect of the MALDI experiment that was evaluated was the
occurrence of suppression effects in peptide mixtures. The most typical
suppression occurs when an analyte that is easily detected, such as bradykinin,
interferes with the detection of another analyte. In addition it is possible for an
analyte that is not easily detected, such as bucalin, to interfere with the detection
of other analytes. In the mixtures studied the magnitude of the suppression
effect that occurred was dependent on the ratio of the analyte concentrations
rather than the total amount of analytes.

Another aspect of the MALDI experiment that was studied was the
formation of prompt fragment ions. The formation of prompt fragment ions was
found to be matrix dependent. Bumetanide was identified as a promising matrix
for prompt fragmentation analysis. The results from prompt fragmentation using
bumetanide as the matrix are compared with prompt fragmentation using a-
cyano-4-hydroxycinnamic acid and 2,5-dihydroxybenzoic acid. In addition a 4:1
mixture of bumetanide and a-cyano-4-hydroxycinnamic acid was evaluated. This

matrix mixture produces abundant fragmentation at low laser power. With the
use of this matrix mixture a very intense y" prompt fragment ion of the b-chain of
insulin is produced. The y" prompt fragment ion was then selected for post-
source decay (PSD) analysis. This is the first time that a prompt fragment ion

has been successfully selected for MS" analysis.

I would like to dedicate this dissertation to my parents Robert and Bylaina. My
father was responsible for sparking my interest in science at a very early age.
My mother has been a constant source of support ever since the first grade,
when I had trouble learning to read. In return I promise to get a “real” job
someday.

ACKNOWLEDGMENTS

I am deeply grateful to all of the people who in one way or the other
helped me throughout my graduate career. One group of friends who provided a
lot of support were my dinner group. At various times this group included Joy
Heising, Sue Pasco, Shannon Majors, Tina Erickson, and Kevin White.
Throughout the first four years we managed to get together to cook each other
dinner and to socialize. These get togethers were a much needed outlet. From
this group the last remaining person Joy Heising deserves a special thanks as
my best friend during this crazy time in my life.

I was also fortunate enough to receive some financial support through the
MSU-NIH Mass Spectrometry Facility. I would like to thank both Doug Gage and
Dr. Jack Watson. In addition I would like to thank Mel Micke and Nat Riddle for
technical assistance over the course of my career. I also received a dissertation
completion fellowship from the College of Natural Science. Finally I was also
supported by Science Theatre for a semester as I worked on a book of science
demonstrations. I also would like to thank my father and sister who both
provided some last minute emergency proofreading.

Of course the most important person during my graduate career was my
advisor John Allison. It is not an easy thing to have me as a student and JA was
more than equal to the challenge. I am only now starting to realize how much I

have Ieamed under JA’s guidance.

TABLE OF CONTENTS

List of Tables. ...................................................................................................... ix
List of Figures ....................................................................................................... x
List of Abbreviations ............................................................................................ xiii
Chapter one. Introduction .................................................................................... 1
1. Historical Context ............................................................................................ 1
2. Research Objectives-a Technique to Probe the MALDI System ..................... 7
3. References ................................................................................................... 12

Chapter two. The Development of Power Trtrations as 3 Technique for Studying

the MALDI Process ...................................................................... 14

1. Introduction ................................................................................................... 14
2. Experimental ................................................................................................. 17
3. The Usefulness of Power Titrations .............................................................. 32
4. References ................................................................................................... 33
Chapter three. Evaluation of Sample Preparation Techniques ........................... 34
1. Introduction ................................................................................................... 34
2. The Standard Method of Sample Preparation ............................................... 36
3. Alternative Sample Preparation Procedures ................................................. 43
4. References ................................................................................................... 60

Chapter four. Suppression Effects of Peptides and Proteins Using a-cyano-4-

hydroxycinnamic Acid as a matrix for MALDI .............................. 61
1. Introduction ................................................................................................... 61
2. Results and Discussion ................................................................................. 65
3. References ................................................................................................... 81

vi

Chapter five. Prompt Fragmentation (PF-MALDI) ............................................... 83

1. Introduction ................................................................................................... 83
2. Results and Discussion ................................................................................. 85
3. Chemistry of Fragmentation............................................................. ............. 99
4. Combining Matrices .................................................................................... 106
5. Prompt Fragmentation-Post Source Decay ................................................ 107
6. Conclusion .................................................................................................. 1 11
7. References ................................................................................................. 1 12
Appendix ........................................................................................................... 114

vii

LIST OF TABLES

Table 4.1 Peak heights of bradykinin under different conditions
Table 4.2 Peak heights of the b-chain of insulin under different conditions

Table 5.1 Effect of type of residues on fragmentation

viii

Figure 1.1
Figure 1.2
Figure 2.1
Figure 2.2
Figure 2.3
Figure 2.4

Figure 2.5

Figure 2.6

Figure 2.7

Figure 3.1

Figure 3.2

Figure 3.3

Figure 3.4

Figure 3.5

Figure 3.6

Figure 3.7

Figure 3.8

LIST OF FIGURES

Three dimensional representation of a typical MALDI experiment
Three dimensional representation of a typical power titration
Diagram of the Vestec 2000, linear MALDI-TOF instrument
Power titration of angiotensin in a-cyano-4-hydroxycinnamic acid
Matrix region of the power titration spectrum of angiotensin
Theoretical spectrum of C4H3N“

Region of the protonated analyte peak in the power titration of
angiotensin

Region of the Na+ peak in the power titration of angiotensin

Three component spectra of the power titration of angiotensin in
a-cyano-4-hydroxycinnamic acid

Power titration spectrum of the seven component mixture, prepared
by the analyte first dried droplet procedure

Power titration spectrum of the seven component mixture, prepared
by the matrix first dried droplet procedure

Power titration spectrum of the two component mixture, prepared
by the dried droplet procedure

Matrix region of the power titration spectrum of the two component
mixture, prepared by the dried droplet procedure

Power titration spectrum of the two component mixture, prepared
by the Vonn procedure

Matrix region of the power titration spectrum of the two component
mixture, prepared by the Vorm procedure

Power titration spectrum of the two component mixture, prepared
by the crushed crystal procedure

Matrix region of the power titration spectrum of the two component
mixture prepared by the crushed crystal procedure

Figure 3.9
Figure 3.10
Figure 3.11

Figure 3.12

Figure 3.13
Figure 3.14
Figure 3.15

Figure 4.1
Figure 4.2
Figure 4.3
Figure 4.4
Figure 4.5

Figure 4.6
Figure 4.7

Figure 5.1
Figure 5.2
Figure 5.3

Power titration spectrum of the two component mixture, prepared
combo procedure

Matrix region of the power titration spectrum of the two component
mixture, prepared by the combo procedure

Comparison of the component spectra taken at a laser power of
1450 for the different preparation procedures

Expanded view of bradykinin in the comparision of the component
spectrum taken at a laser power of 1450 for the different
preparation procedures

Comparision of the component spectra taken at a laser power of
2500 for the different preparation procedures

‘ Power titration spectrum of the two component mixture, prepared

by the pure crushed crystal method

Matrix region of the power titration spectrum of the two component
mixture, prepared by the pure crushed crystal method

High power spectrum of six component mixture

MALDI spectrum of 5 pm cytochrome C

Low power spectrum of six component mixture

Spectrum of 1:1 mixture of bradykinin and b-chain of insulin
Spectrum of 1:9 mixture of bradykinin and b—chain of insulin

3 consecutive spectra of a 1:1 :1 mixture of bradykinin, b-chain of
insulin and insulin

3 consecutive spectra of a 3:1 :1 :1 mixture of bucalin, bradykinin, b-
chain of insulin, and insulin

Structure and ultraviolet-visible absorption spectrum of bumetanide
MALDI mass spectrum of bumetanide as a matrix

Prompt fragmentation spectra of dynorphin with a) DHB, b) HCCN
and c) BMN

Figure 5.4

Figure 5.5

Figure 5.6

Figure 5.7

Prompt fragmentation spectra of MSH with a) DHB, b) HCCN,
c) BMN (karat symbol above two peaks represents pairs of
an and an — NH3 ions)

Prompt fragmentation spectrum of the b-chain of insulin in
a) HCCN, b) DHB (square brackets represent pairs of yn and
Z”,1 ions)

Prompt fragmentation spectra with a) BMN, b) HCCN and
b) 4:1 BMN:HCCN

Post source decay of the y" prompt fragment ion of the b-chain
of insulin

xi

LIST OF ABBREVIATIONS

A/D .............................................................................. Analog-to-digital conversion
A ................................................................................................................. Alanine
AH+ .................................................................................... Protonated angiotensin
B ....................................................................................... Bradykinin, Cysteic acid
b-chain .......................................................................... Oxidized b-chain of insulin
BH’ ...................................................................................... Protonated bradykinin
BMN ..................................................................................................... Bumetanide
CAD .................................................................. Collisionally—activated dissociation
C ............................................................................................................... Cysteine
CH+ ................................................................................ Protonated cytochrome C
CH2” ................................................................... Doubly protonated cytochrome C
CH33+ ..................................................................... Triply protonated cytochrome C
Cu+ ........................................................................................................ Copper ion
D ....................................................................................................... Aspartic acid.
Da ................................................................................................................. Dalton
DHB ............................................................................... 2,5-dihydroxybenzoic acid
Dyn ......................................................................................................... Dynorphin
E ....................................................................................................... Glutamic acid
F ....................................................................................................... Phenylalanine
H ............................................................................................................... Histidine
H’ ................................................................................................................ Proton
HCCN ................................................................. a-cyano-4-hydroxycinnamic acid
I .................................................................................................. Insulin, lsoleucine
IH+ .............................................................................................. Protonated insulin
IbH“ ............................................................................ Protonated b-chain of insulin
J ..................................................................................................................... Joule
K .................................................................................................. . ............. Leucine
K” ....................................................................................................... Potasium ion
L ....................................................................................................................... Liter
M ................................................................................................ Matrix, Methionine
M/z ......................................................................................... Mass-to-charge ratio
MALDI .................................................. Matrix-assisted laser desorption/ionization
MeH+ .................................................................................................................................... Protonated mellitin
MS ............................................................................................ Mass spectrometry
MS" ............................................................... Mass spectrometry to the nth power
MSH ................................................................. a-melanocyte stimulating hormone
MW .............................................................................................. Molecular weight
MyH+ ................................................................................... Protonated myoglobin
MyH-z2+ ..................................................................... Doubly protonated myoglobin
Na+ ....... ' ................................................................................................ Sodium ion
ns ........................................................................................................ Nanosecond
O .............................................................................................................. Omithine
Q ............................................................................................................. Glutamine

xii

P .................................................................................................................. Proline

PF ....................................................................................... Prompt Fragmentation
PSD .......................................................................................... Post source decay
R ................................................................................................................ Arginine
S ................................................................................................................... Serine
t ........................................................................................................................ time
TOF .................................................................................................... Time-of-flight
UV ............................................................................................ Ultraviolet radiation
V ....................................................................................................................... Volt
W .......................................................................................................... Tryptophan
Y ............................................................................................................... Tyrosine

xiii

Chapter one.
Introduction

1. Historical Context

Lasstgescrptim

Laser desorption mass spectrometry (MS) was one of the earliest
methods used to analyze thermally labile molecules. Laser desorption has been

performed with a large number of analytes. These analytes include alkali salts,

oligosaccharides, oligionucleOtides, peptides and polymers1. In this work I will

be concentrating on the analysis of peptides. For peptides the upper mass limit

of laser desorption is approximately 2000 Da2.

Laser desorption MS is performed by dissolving an analyte in a volatile
solvent such as methanol or ethanol. A drop of the sample solution is then
placed on a stainless steel surface and the solvent is allowed to evaporate. A

thin layer of the analyte remains on the surface after the solvent has evaporated.

This layer is typically too thin to be seen by the unaided eyez. The sample is
then placed in the mass spectrometer. A variety of types of mass spectrometers
have been used for laser desorption; these instruments include dual sector mass
spectrometers, time-of-flight mass spectrometers, and Fourier transform mass
spectrometers. The sample is then irradiated by the laser and the analysis
begins. The most common laser that is used with peptides is a C02 laser. This

type of laser produces photons at a wavelength of 10.6 pm and a pulse width of

40 ns.

Few results have been reported using this technique for the analysis of

peptides. The first spectrum of a peptide was reported in 19783. This experiment
was performed with a double-focusing mass spectrometer. In this experiment

the laser was fired to produce the desorption, then there was a delay of 500 us

before an extraction pulse was applied to accelerate the ions. A custom made
array detector detected a range of mlz values in a single pulsed acquisition. An
octapeptide with a molecular weight of 718 Da was analyzed with this instrument.
The resulting spectrum consisted of Na” and K” adducts of the intact peptide,
and a single fragment that was also a metal adduct.

Time-of-fiight mass spectrometry has also been used for laser desorption

of peptides4v5. In this experiment the sample was irradiated with light from a
CO; laser. There is then a delay time of 10-20 us before the extraction voltage is
applied. In this work a short delay produced poor resolution and multiple
cleavages of the peptide backbone, while longer delays produced higher

resolution and little fragmentation. The peaks corresponding to the analyte were

the sodium and potassium adduc$4:5.

Another type of mass spectrometry that has been used in the analysis of
laser desorption products is Fourier transform mass spectrometry. In this
technique the analyte is desorbed and ionized by the laser. The products are
accelerated into a sample cell. The cell is contained in a homogenous magnetic
field and an electrostatic potential is applied to the end plates of the cell so that
the ions are forced into circular trajectories based on their mass-to-charge ratio.

An excitation pulse is then used to force all of the ions of a given mlz value to

bunch together in space and orbit together. A pair of receiver plates detects the
current from these orbits. The time data is converted into data in the frequency

domain. The frequency of an ion packet in a circular orbit is inversely

proportional to the mlz of the ions in the packets. After the analyte is desorbed
the analytes are allowed to stay in the cell for 3-10 seconds before the detection
is done. Using this method, a protonated analyte is typically detected along with
sodium and potassium adducts. In addition there are very intense peaks from
fragmentation products. The fragments tend to occur from breakage of the
peptide bonds7.

There are two main theories about the mechanism of laser desorption.
When energy is introduced into a solid analyte two processes it can undergo are
desorption and decomposition. Desorption is a high energy fast process, while
decomposition is a low energy slow process. In general the desorption of
thermally labile molecules is thought to occur when enough energy is put into the
molecule so that the rate of desorption exceeds the rate of decomposition. The
laser provides a very efficent means to heat the sample quickly. A wide array of

lasers are used in this work with laser powers that from 50 kW to 100 MW and
wavelength that range from 151 pm to10.6 nm3. These lasers include both
continous wave and pulsed lasers. The laser raises the temperature of the
stainless steel surface at a rate of 10° — 10‘2 Kls 3. Energy then flows through

the surface adsorbate bond to heat the analyte. The energy from the laser is

transferred to the peptide at a relatively slow rate, from the substrate, so the

bonds between the substrate surface and the peptide break before the bonds in

the analyte break3.

Matrix Assisted Laser IZLesorption/lonization (MALDI)

In 1987 Hillenkamp modified the typical laser desorption experiment by

adding a matrix of nicotinic acid in a large excess to the analyte9. This technique
greatly increased the mass range of analytes that could by detected. This first
work was done with a NdIYAG laser at a wavelength of 266 nm. Nicotinic acid
has a high molar absorbtivity at this wavelength. The technique was soon
adapted to UV lasers by the use of different matrices. UV lasers have the
advantage of being inexpensive and compact. Much of the work that is being

done currently in this field uses nitrogen lasers, which emit light at a wavelength

of 337 nmlo. This technique is referred to as matrix assisted laser
deorptionfionization (MALDI).

The MALDI experiment is typically performed by mixing together a matrix
solution and an analyte solution and letting the combined sample dry. Typical

matrices include cinnamic acid derivatives such as caffeic acid, sinnapinic acid,
and a-cyano-4-hydroxycinnamic acid1 1. In addition 2,5—dihydroxybenzoic acid
has also found use as a MALDI matrix12.

Matrix assisted laser desorption/ionization has become a very powerful

technique for the analysis of peptides and proteins. Using this technique the

mass range is well in excess of 100,000 Da and attomolar amounts can be

detected13v14. The predominant analyte related peak is from the protonated

analyte. However there are also minor peaks representing metal ion adducts.
The majority of work with this technique is performed using time-of-fiight mass
spectrometry. MALDI is typically considered a “cool” desorption technique since

peaks from fragmentation products are typically not observed. However

metastable decay does occur after the acceleration out of the ion source15.

Current Models of the MALDI Process

The mechanisms that have been proposed for the desorption process in
MALDI are similar to the mechanisms proposed for Laser desorption. The
matrices used in this technique are always strong absorbers in the wavelength
region in which they are used. The matrix absorbs energy from the laser as a
first step. The most prevalent model for the desorption process is that the matrix
heats up very quickly without heating the analyte. The high temperature causes
the matrix to quickly sublime. As the matrix quickly sublimes the analyte is
entrained into the gas phase. The analyte does not have a lot of internal energy

because energy is slow to transfer from the matrix to the analyte. This model is

called the bottleneck model16. Another possible model for the desorption
process is that as energy is deposited into the matrix a pressure gradient is set

up in the matrix. Once this pressure gradient exceeds a threshold value a

pressure pulse is created which causes the matrix to suddenly expand”. The
rapidly expanding matrix would have analyte entrained within it.
The ionization in the MALDI process could be occurring by a number of

means. One of the processes that could be involved is excited state acid-base

chemistry. In this model, laser irradiation of the matrix produces excited state

acids, which could then react with the analyte or other matrix molecules13.
According to this mechanism the proton transfer could occur in the solid state or
in the gas phase. If the protonation occurred in the gas phase it would be similar

to chemical ionization. The. ionization process could also proceed through the

steps of photoionization of the matrix and proton transfer to the analyte19.
The majority of current theories on the mechanism of ionization rely on

either photoionization or photoexcitation of a single matrix molecule to give rise

to the proton transferzo. However other possibilities are also being proposed.

Multiply excited matrix aggregates, formed by photoexcitation, may also serve as

the source of protons for protonated analytes 21.22. One other possibility is that
matrix excitation occurs through fast heating of the matrix. It has been shown

that for peptides under 3000 Da 3 protonated analyte peak can be detected by

quick heating of a model sample on a filament 23. The relatively low mass limit
of this experiment could be explained by the fact that the filament heat the matrix
and sample at a much slower rate than laser irradiation. The one area of general

agreement in the field is that a variety of ionization processes are occurring in the

MALDI experiment19:20.

2. Research Objectives-a Technique to Probe the MALDI System

The variables that can be controlled; during a MALDI experiment include
the number of acquisitions, the laser power, and the position of the laser spot.
These parameters can be used to define a three dimensional space. We define
the acquisition number as our abscissa. Component spectra are acquired at a
constant rate throughout the experiment, so the total number of component
spectra is directly proportional to the length of the experiment. The ordinate
corresponds to the position of the laser spot on the surface. The laser power is
defined as the 2 value in this representation of the MALDI experiment. Finally, in
our instrument the sample is placed on a round sample pin that has a radius of 1
mm, these pins are inserted into a carousel which is inserted into the instrument.
The laser irradiates the sample with a laser spot that is much smaller than the
surface of the sample pin. Turning the carousel moves the sample, as the
carousel turns the individual pins spin in place. The laser spot position is fixed at
an edge of the pin and the pins rotate to allow different parts of the sample to be
irradiated. The position value of the laser spot actually corresponds to the
amount that the sample has been rotated. The actual amount that the sample
pin rotates varies from experiment to experiment, for this reason no attempt is
made to quantitate how much the pins rotates.

In a typical acquisition in the MALDI experiment the sample position and
laser power are varied with the number of acquisitions, and are averaged

together. This type of experiment is represented on the axes system in Figure

1.1. In this example the experiment is started at a laser power of 200 and then
raised to a laser power of 250 over the course of 5 acquisitions. The laser power
is then increased to 275 over the next 15 acquisitions. During the first 20
acquisitions the same spot on the sample is irradiated. Over the course of the
next 40 acquisitions laser spot position and the laser power are both varied. As
illustrated in Figure 1.1 the typical MALDI experiment does not represent an
orderly trip through the range of possibilities that we have defined. The only goal
of the typical MALDI experiment is to produce a high resolution, high signal to
noise spectrum.

In order to reproducibly study the effects of different variables on the
MALDI experiment a reproducible spectrum acquisition protocol must be
developed. Power tirations were developed in order to fill this need. A power
titration is simply a method to systematically explore the variable space that we
have defined. In order to study the effects on a given spot the laser position is
always held constant. During this experiment, the laser power is increased at
regular intervals. Typically the experiment is continued for a large number of
total acquisitions so a wide range of laser powers can be explored. The
representation of this type of experiment in the previously defined variable space
is shown in Figure 1.2. In order to observe the changes that occur during the
experiment a spectrum is always saved to disk just before the laser power is

increased. A new spectrum is then started at a higher laser power.

EoEcoaxo 5.222 .325 m .0 22353.32 .mcoficoEfi moi... 2 2:9“.

0 ON

o...

         

COS

. . @035”
«a @
md

v.9 .
od
com

CNN
owm

cow

O
on
N
JeMod Jasel

com

can

ova

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Jemod Jesej

10

'88
L/ l / // ' g 60¢
H in i / / / // // // / / / 0"“ s f
Ll HUN/[W
Ll /////////// "OW".
L1 ll/[i/l/I/HH/ "NH"
//// 90.00""
"I. <\ ( ( ( (< < << glitzy,“ -. :2.
l \M \\ \\ \\ \l \ \ \{ xix...” S
\\\l \l \\ T'HT¥\\ \\ \\ .00.... 0!
l\l\\\ ’9’"
\\l\\\\\\\ ”a $9
\\\\'\\\\ \i 8
\.\,\.\,\.\.\\\\. cs £3?
:53: § § § 3 .55 :‘2’ 3 \

Figure 1.2 Three dimensional representation of a typical power titration

By performing a series of power titrations under slightly different conditions,
much can be Ieamed about the MALDI experiment. Power titrations can be used
to study the effect that repeated laser shots have on the sample surface. Since
the laser spot is held constant, one spot is subjected to a large number of laser
shots. In addition the power titration technique can be used as a standardized
acquisition method to study different types of MALDI samples. One example of
different samples that can be compared is MALDI samples that are prepared by
different techniques. The effects of power titrations on mixtures of different
analytes has also be studied.

By using power titrations with mixtures prepared by different sample
preparation techniques the best methods for analyzing mixtures were identified.
MALDI is a very quick and easy method to analyze mixtures of peptides and
proteins. By optimizing mixture analysis I hope to make the technique even
easier. In addition I will evaluate the likelihood of selected components of a
mixture not being detected.

At very high laser power in power titrations peaks sometimes appear
corresponding to peptide fragments. These fragments are forming in the ion
source and are not a result of metastable decay. These fragments could be
useful for obtaining structural information about peptides. In addition these
prompt fragments provide an opportunity to study the fragmentation of peptides

under a different set of conditions.

11

3. References
1)Conzemius, R. J.; Capellen, J. M. International Journal of Mass Spectrometry and
Ion Physics 1980, 34, 197-271.
2)Cotter, R. J. Analytica Chimica Acta1987, 195, 45-59.

3)Posthumus, M. A.; Kistemaker, P. G.; Meuzelaar, H. L. C.; Brauw, M. C. T.
Analytical Chemistry 1978, 50, 985-991.

4)Tabet, J. C.; Cotter, R. J. Analytical Chemistry 1984, 56, 1662-1667.
5)Stoll, R.; Rollgen, F. W. Organic Mass Spectrometry 1979, 14, 642-645.
6)Buchanan, M. V.; Hettich, R. L. Analytical Chemistry 1993, 65, 245A-258A.

7)\Mlkins, C.; Weil, D.; Yang, C. L. C.; James, C. F. Analytical Chemistry 1985, 57,
520-524.

8)Levis, R. J. Annual Reviews in Physical Chemistry 1994, 45, 483-518.
9)Karas, M.; Hillenkamp, F. Analytical Chemistry 1988, 60, 2299-2305.

10)Beavis, R. C.; Chait, B. T. Rapid Communications in Mass Spectrometry 1989,
3, 342.

11)Beavis, R. C.; Chaudhary, T.; Chait, B. T. Organic Mass Spectrometry 1992, 27,
156-158.

12)Karas, M.; Ehring, H.; Nordhoff, E.; Stahl, B.; Strupat, K.; Hillenkamp, F.; Grehl,
M.; Krebs, 8. Organic Mass Spectrometry 1993, 28, 1476-1481.

13)Doktycz, S. J.; Savickas, P. J.; .,Krueger D. A. Rapid Communications In Mass
Spectrometry 1991, 5, 145-148.

14)Vonn, O.; Roepstorff, P.; Mann, M. Analytical Chemistry 1994, 66, 3281-3287.

15)Spengler, B.; Kirsch, D.; Kaufmann, R. Rapid Communications in Mass
Spectrometry 1991, 5, 198-202.

16)Vertes, A.; Gijbels, R. Rapid Communications in Mass Spectrometry 1990, 4,
228-232.

17)Johnson, R. E.; Sundqvist, B. U. R. Rapid Communications in Mass
Spectrometry 1991, 5, 574-578.

12

18)Dikler, S.; Kelly, J. W.; Russell, D. H. Journal of Mass Spectrometry 1997, 32,
1337.

19)Ehring, H.; Karas, M.; Hillenkamp, F. Organic Mass Spectrometry 1992, 27.
20)Liao, P.-C.; Allison, J. Journal of Mass Spectrometry 1995, 30, 408-423.

21)Knochenmuss, R.; Dubois, F.; Dale, M. J.; Zenobi, R. Rapid Communications in
Mass Spectrometry 1996, 10, 871-877.

22)Knochenmuss, R.; Karbach, V.; Wresli, U.; Breuker, K.; Zenobi, R. Rapid
Communications in Mass Spectrometry 1998, 12, 529-534. -

23)Karbach, V.; Knochenmuss, R.; Zenobi, R. Journal of the American Society for
Mass Spectrometry 1 998, 9, 1226-1228.

13

Chapter two.
The Development of Power Titrations as a Technique for Studying the
MALDI Process

1. Introduction

Goals
The technique of matrix-assisted laser desorption/Ionization has greatly

 

expanded the range of analytes that can be detected with mass spectrometry.
However the mechanism of this technique is poorly understood. At this time
even the effects that different parameters have on the experiment are poorly
mapped out. Power titrations were developed in an attempt to better understand
the multidimensional nature of data acquisition in the MALDI experiment.

Once reliable data are obtained the data can be used to help answer
questions that are fundamental to the MALDI experiment. Power titrations
provide a means to study the effect of continued irradiation of the sample. The
development of some guidelines could be useful for comparing the results
between different laboratories and provide rules for obtaining a standard MALDI
spectrum.

The results from power titrations can also be used to help probe
mechanistic aspects of the MALDI experiment. The most obvious experiment
would be to determine which matrix peaks correlate the most highly with the
formation of the protonated analyte peak. Any peaks that have a strong
correlation with the protonated analyte could be incorporated into proposed
mechanisms of the desorptionfionization process. In addition by observing the
laser power onset of the various matrix peaks, reactions in the matrix could be

potentially identified.

14

The Power Titration Procedure

There are a number of variables that are carefully controlled in the power
titration experiment. The entire experiment is performed at a single spot on the
sample. Our instrument has been modified so that the focus of the laser spot
can be controlled by changing the distance between the focussing lens and the

sample. Although the effect of spot size is poorly understood, the size of the

laser spot has been found to affect the MALDI experiment1. As the lens is
moved further away from the sample a larger area of the sample is irradiated.
The position of the lens can be characterized by a vemier scale on the stage on
which the lens is mounted. Unfortunately the numbers on the scale have not
been calibrated to actual laser spot sizes. The difficulty lies in the measurement
of the laser spot size on the sample surface. We simply use the scale values as
a relative indication of the laser spot size. A value of 65 corresponds to the lens
being positioned up against the source housing to produce the smallest spot
possible. This is the standard position for the lens and produces a spot with a
radius on the order of .1 mm. A position of 40 is the largest spot that we found to
be practical for MALDI analysis. Vlfith this lens position the laser spot has a

radius of about .4 mm. As the laser spot gets bigger, higher laser powers are

necessary to achieve the fluence 2 to perform the MALDI experiment.
The next component of the power titration experiment is to select an initial
laser power. The laser power is best discussed relative to the threshold for a

given sample. I define a sample’s threshold as the laser power at which the

15

protonated analyte peaks have a signal-to-noise ratio of 3:1. The initial laser
power can either be above or below the threshold laser power. Using an initial
laser power that is above the threshold laser power produces a power titration,
which contains signal throughout the experiment. By using an initial laser power
that is below the threshold laser power, the appearance laser powers of matrix
and analytes can be detected and compared.

In the MALDI experiment the number of spectra that can be recorded at a

given spot with a single laser power is limited. The higher the laser power that is

used the more spectra that can be recorded before the analyte signal decays 3.

The amount of material (matrix and analyte) that is ablated from the sample, per

laser shot, is also increased “at higher laser powers4. The amount of material

that is ablated in a typical MALDI experiment is 1 x 1013 - 1 x 10“ particles per
laser shot 4. When the signal is depleted at any laser power the power can be

increased to produce more signal5. These trends indicate that photochemistry
is occuring on the matrix at laser powers used in the MALDI experiment. The
types of photochemistry that could be occurring are photodissociation and solid
state dimerization. Products that could be from photodecomposition and dimers

are detected in the matrix spectrum. Solid state dimerization of cinnamic acid

derivatives has also been reported5.

In order to determine if any chemistry is occurring below the laser power
necessary to produce desorptionfionization, a sample can be irradiated below the
threshold laser power prior to the power titration. The results from a pre-

irradiated sample can then be compared with the results from a standard power

16

titration. If the MALDI threshold values are shifted then the matrix is undergoing
photochemistry at laser powers that are lower than required for the MALDI
experiment.

Finally the number of laser shots that are acquired in a single spectrum
needs to be determined. The advantage of averaging together a small number of
laser shots is that subtle changes in intensities can be detected. In contrast

averaging together a large number of spectra allows for spectra with less noise.

2. Experimental

Instrumental Setup

The instrument used in this work was a Vestec 2000, illustrated in Figure
2.1, linear time-of-flight instrument with an adjustable focussing lens for the laser.
The Vestec 2000 has a single stage source that has a maximum acceleration
voltage of 30 W. In addition this instrument also has a guide wire in the flight
tube to improve the transmission of ions. This instrument used stainless steel
sample pins with a radius of 1 mm. These pins were mounted in a 24 pin
carousel that could be rotated to select different samples. The movement of the
carousel is controlled by pressing a joystick left and right. By moving the joystick
up and down the laser power can also be controlled. In addition one of the
joystick buttons is used to start the acquisition of a spectrum while the other
button is used to download a spectrum to disk.

The laser used with this instrument is a VSL-337ND nitrogen laser (Laser

Science Incorporated). This pulsed laser produces a 3 ns pulse of radiation at a

17

', | ground
I grid
I I
I
. I
I
\ laser
20kV
gnd

detector /

 

 

 

Figure 2.1 Diagram of the Vestec 2000, linear MALDI-TOF instrument

18

 

 

wavelength of 337.1 nm. The pulse energy of this laser varies from 150 to 250
uJ depending on the age of the nitrogen tube. The laser beam is passed through

a 935-5-opt (Newport, Irvine, CA) laser atenuator, which controls how much
radiation reaches the sample. The atenuator has been calibrated with the use of
an 818j—09/DC (Newport, Irvine, CA) energy detector. A low laser power setting
of 200 represents a laser power of 24 uJ. While a high laser setting of 5000
represents a laser power of 68 uJ. In this range the laser power varies linearly
with the laser setting. For all of the results reported in this work only the laser
setting will be reported, since the actual laser power was not measured for every
experiment. An aproximation of the actual laser power could be calculated by
interpolation. However the accuracy of these calculated values would be suspect
since different laser cartridges were used over the course of this work and
slippage was observed in the laser attenuation mechanism.

The standard sample preparation was done by placing 1 pL of an analyte
solution of 1:1 acetonitrile: 0.1% aqueous trifluoroacetic acid (T FA) on the
sample pin. 1 pL of 1:1 acetonitrile: 0.1% TFA saturated in matrix solution was
than added to the sample pin. The solution was then allowed to air dry to form
crystals.

The first step in a power titration is to pick a part of the sample to irradiate.
Typically I selected either the center or the left edge of the sample, since this
laser position produces the greatest sensitivity, due to geometry of the ion
source. Next a starting laser power is selected. The threshold of five samples is

measured and the initial laser power is selected relative to the measured

19

threshold. Depending on how large of a range of laser powers are to be
explored, a laser power increment is also determined. The number of
acquisitions to be averaged together to produce a spectrum is then entered into
the acquisition control software.

At this point the actual power titration experiment can be performed. The
first spectrum is acquired by pressing the start button on the joystick. When the
acquisition is completed the component spectrum is immediately downloaded to
disk. While the spectrum is being transferred, the laser power is incremented.
As soon as the spectrum is saved, another acquisition is started. This cycle is
repeated 30 to 40 times to produce a complete power titration.

Once the acquisition Is complete the separate data files are made into a
multifile for display purposes. A multifile is a single file that contains a series of
spectra that are sorted in the order they were acquired. Using the mfutil
program, which is included with the acquisition software, creates the multifile.
The resulting file is very hard to examine due to noise being present in a very
complicated display. Smoothing all of the data can alleviate this problem. The
entire multifile is then smoothed with the Savitsky-Golay smoothing algorithm. A
smooth of 25 points is done on each trace, of 50,000 points, in the multifile. The
smoothed multifile can now be displayed in a number of different ways. Typically
a hidden 3-D format was used to make the data easy to interpret. This format
stacks the spectra along a Z-axis, with the early spectra obscuring the later

spectra.

20

A typical power titration

An example of a typical power titration is shown in Figure 2.2. This figure
shows results of a power titration of angiotensin (M.W. = 1296.5) in a-cyano-4-

hydroxycinnamic acid. The sample was prepared by the standard method. For
this sample the focusing lens was put in a position of 50 to provide a large laser
spot. An initial power setting below threshold of 750 and a step size of 50 were
selected. The number of acquisitions was set to 7, to provide a sensitive
technique with a good signal-to-noise ratio.

This set of parameters led to a fairly typical power titration as shown in
Figure 2.2. Power titrations are performed with time-of-fiight spectra, so the x-
axis is the flight time in nanoseconds. The y-axis is relative signal intensity
based on the detectors current at a given time. The z-axis is the number of the
spectrum as it is collected in the power titration experiment. As the spectrum
number goes up, the laser power at which a given spectrum was acquired also
rises. The laser power of any individual spectrum can be calculated by
multiplying the spectrum number by the step size and adding the initial laser
power. In this case spectrum number 20 would be [20 x 50] + 750 = 1750. In
addition the entire power titration is displayed with both a vertical and a horizontal
offset. These offsets are used to make it easier to see as many of the peaks as
possible.

The general trend that is illustrated by Figure 2.2 is that the matrix ions

start forming at a lower power than the protonated analyte peak. At higher laser

21

 

Eon o_Em::_o>x9u>c-Tocm>o-d E Emcoaoacm Co Saab: .0301 NM 939...
me £95905:

        

 

 

 

 

 

 

 

 

 

 

 

   
  

 

 

 

 

 

 

   

Seen 8%. 88
%u o N _ >
e \ .
as. .. . .
0.. L
W at i 1 , 14%
g in 41‘) 34).,
ow iii}... . fl
., ,t t . 3335...).11 . . _ , 3.4.
.1. _§§§%3£ .1 . . . , . .,
.33 3.8.5....5331 ._._§_. . . _ ., . . a.
:3... a! 1‘ usfibiw‘ag JJKYfJJz .. .. . _ . . . . . .. c
.< .4 ’ .flgrénussygrgi : in J ; ..; . .. . H H H . .
on i Jail??? . L. , . _. . , , g . 1 -88.
.J. 2.1. . ._ . . . , .. ..
_ .. . . .x. e a a . .. _ . l . .. _
2. I ., .. .H .. ,,. . m
, ., ... , W. A . M
. . . . _ N.
._ . .. .. N. a
.. : 5‘ “a. s,“ M
.. ... B
. _ u
.. a .. -oooow S.
.... ... m
._
loooon
. fl.

 

22

powers all of the peaks get more intense. The first 4 spectra that were collected
in this power titration show virtually no signal. At a laser power of 900 the first
few matrix peaks become visible, shown at point B in component spectrum
number 6. The peaks which start at points A and B, correspond to mlz’s of 172
and 189. The mlz 189 peak represents the charged intact matrix, while the mlz
172 peak represents the result of loss of H20 from the protonated matrix. At a
laser power of 1300 the protonated angiotensin produces a peak, shown at point
C of Figure 2.2. From experience with this type of sample, we would expect
angiotensin to have a threshold of about 1000. The laser power of 1300 is
significantly higher than the typical threshold value from the standard MALDI data
acquisition. Apparently extended irradiation of the sample below the threshold
value causes the threshold value to rise. This trend is an indication that the
surface of the sample is affected well before peaks are detected in the power
titration experiment.

As laser power is increased, additional peaks become apparent. Shown
at point D in Figure 2.2 at high laser power, 1750, an adduct begins to form with
angiotensin. The time-of -flight of this adduct corresponds to an mlz that is 65
mass units higher than the protonated analyte peak. The matrix region of this
power titration spectrum is displayed in Figure 2.3. Typically the adduct peaks
are most often due to attachment of a metal ion. One possibility is that this peak
is the result of a zinc adduct. However the isotopic distribution does not appear
to be distribution that would be expected for Zn. Alternatively the peak at mlz 65

could be due to 04*H3N. This could be a fragment from the matrix or an impurity

23

Ausuatu! artnejeu
O

 

     

   
       
  
 

 

 

      

 

 

 

 

 

 

 

 

 

 

 

 

 

 

o o o
s s s s
T "i ‘1‘ ‘T c’
d‘
77-» .‘Li
.. x 9.31;;
_ l_ _ fii_*;;:¥:‘fl:‘§}
@flk‘i‘ g
'\\‘;\-g‘fi
:& ~ I Ԥ
:fi-fia 3“
fl] < 2
11 , I
e: c
II .9
'0'- .9
ca
.. E s.
l -§ i= “6
E
‘3
n.
in
r:
.9
'5
5
.g g.
T o
a
Q)
5
‘5
I:
.9
5’
a _x
thLB .5
>’ N W
o E
on
N
9
:3
65 .9
WM” LL

24

in the matrix. The isotopic distribution seems to be similar to the distribution that
would be expected for C4H3N“. The calculated isotopic distribution of CngN+ is
shown in Figure 2.4. At a laser power of 1950 a peak at an mlz value of 65
becomes apparent in the power titration. This peak does not display the isotopic
distribution of Zn. The signal at mlz 65 occurs at a significantly higher power
than the corresponding analyte adduct. In addition to an adduct being formed
with the analyte there also seems to be an adduct formed with the matrix itself.
At mlz 254 [189 + 65]“ a large peak is detected in the power titration. This peak
first appears at a laser power of 1250. It is interesting to note that the matrix
adduct appears at a laser power that is 350 higher than the matrix peak while the
analyte adduct appears at a laser power that is 450 higher than the analyte peak.
It seems reasonable that the appearance of the adduct peak occurs at a
significantly higher laser power than the cooresponding peak from protonation.
The formation of the adduct ion seems to be a higher energy process than
protonation. The adduct peak likely represents [angiotensin + CngNj’fi At a
laser power of about 1750 many of the matrix peaks become saturated and some
new peaks start to form in the matrix region.

Additional information can be obtained from power titrations by zooming in
on single peaks. Figure 2.5 shows the protonated analyte peak in detail. This
shows the component spectra presented with no horizontal offset. In order to
make the individual peaks easier to see the power titration is viewed with a large
vertical offset, so the spectrum is viewed from a perspective that is well above

the spectrum. The peak at the lowest laser power has the best resolution of the

25

 

1003

75%
.3:
m
C
2.
.E
m.
.5 so;
9
2
25%

 

 

 

 

 

 

Figure 24 Theoretical spectrum of C4H3N+

26

   

relative Intensity

TIme-of-Flight, ns

Figure 2.5 Region of the protonated analyte peak in the power titration of
Angiotensin

27

spectrum #

series. In fact the low power peak has a resolution of 600, while the highest

power peak has a resolution of 300 (FWHM). It has previously been observed

that the best resolution in MALDI is obtained at low laser power 2. In addition to
the change in resolution the center of the peak is also moving to longer flight
times at higher laser powers. There is a 45 ns difference between the center of

the lowest power peak and the center of the highest power peak. The main

q

reason that the center of the peak is moving to longer flight times seems to be

the right edge of the peak is moving to longer delay times, while the left edge

9m _
I

seems stable. The asymmetric peak that is formed at high laser powers has a
center that is shifted to higher flight times.

In order to determine the cause of the spreading and shifting of the
protonated analyte peak, the Na+ peak was examined. This peak is displayed in
Figure 2.6, under the same conditions as were used to display the protonated
analyte peak. Although the peak shapes show quite a bit of variation, the center
of the peaks remains fairly stable. The difference in time-of-flight between the
lowest and highest power Na" peaks is about 1 ns. This shift is well within the
2ns resolution of the digital oscilloscope. The peak for the sodium ion seems to
get broader in a symmetric way. Other matrix peaks also move to higher flight
times at higher laser powers. A possible explanation for the difference between
the Na“ peak and other peaks in the spectrum is that a reaction is necessary to
form all of the other peaks besides the metal ions. Apparently the chemistry can
take place over a longer period of time at higher laser powers. The delay in ion

formation indicates that the ionization is occurring in the gas phase.

28

 

  

relatlve Intensity

'I'lme-of-Flight, ns
Figure 2.6 Region of the Na+ peak in the power titration of angiotensin

29

spectrum ll

 

As another way to display the results of this power titration, Figure 2.7
displays some of the results from the power titration by an alternative method. In
this figure the calibrated spectra acquired at three different laser powers are
displayed on the same scale. For this figure each spectrum was calibrated, by
time-of-flight to mlz values, individually using the Na+ and either the matrix peak
or the [M+H]+ peak. In this series of spectra it is obvious how much of a
difference laser power can make in the resulting mass spectra. The shape of the
peaks is also strongly affected by the laser power used for a spectrum. At higher
laser powers the major matrix peaks have a squared off appearance due to
saturation of the 8 bit a/d converter of the instrument. Some of the major peaks
that are seen in these spectra are the matrix peak at 189 Da. and the protonated
dimer at 379 Da. In addition the peak resulting from the loss of water from the
protonated matrix molecule is at 172 Da. At the laser power of 1650 the [Mi-65]+
peak is shown. In addition this same spectrum contains a peak from the matrix

adduct at a mlz of 254.

30

Eon o_Emcc_o>xo._o>;-v.ocm>o.o 5 52.2065 ho scams: 538 9: .6 86on «cocanoo no.5. Nu 2:9“.
a):

ll ilrlll.ltli\ltJiJ\Iril.<xt5 QW<7i .573. ‘u . 3
. _

‘3

:.

[cocoa

 

 

comm u 332. comm.

234...... 3.. 3; 3.1.8..

 

 

 

 

 

tlrtliillrili

 

 

 

 

82 u 838 .58.

hjsuezuj wine's:

 

1083

r882.
08. u 830g .03.

 

 

3i

3. The Usefulness of Power Titrations

The utility of power titrations can be extended in several directions. By
changing the laser powers that are used in the experiment a wide variety of
conditions can be explored. Another parameter that can be varied is the
increment in the laser power between the different acquisitions. The experiment
can also be altered by irradiating the sample for a set period of time at a laser
power that is well below the laser power at which the power titration is started.
This type of experiment could determine if there are effects on the sample at low
irradiation.

Unfortunately there are a couple of limitations on the use of power
titrations. First power titrations can be difficult to interpret due to the large .
amount of data that are present. The low resolution and irregular peak shapes
further complicate the interpretation since they make calibration more difficult.
Finally the interpretation of power titrations is perhaps most difficult because of
the irreproducibility of the experiment. Significant differences appear in power
titrations of similar samples. These differences could be due to the fact that
there is a significant amount of variation between samples prepared by the
standard sample preparation method. For this reason power titrations are not
useful for elucidating mechanistic details of the MALDI experiment or a standard
method for comparison between different laboratories. In the next chapter, I will
investigate the crystallization process, in an attempt to make the experiment

more reproducible.

32

 

4. References

1)Dreisewerd, K.; Schurenberg, M.; Karas, M.; Hillenkamp, F. IntemationalJoumal
of Mass Spectrometry and Ion Processes 1995, 141, 127-148.

2)lngendoh, A.; Karas, M.; Hillenkamp, F.; Giessmann, U. lntemational Journal of
Mass Spectrometry and Ion Processes 1994, 131 , 345-354.

3)Westman, A.; Demirev, P.; Huth-Fehre, T.; Bielawski, J.; Sundqvist, B. U. R.
lntemational Journal of Mass Spectrometry and Ion Processes 1994, 130, 107-
1 15.

4)Westman, A.; Huth-Fehre, T.; Demirev, P.; Bielawski, J.; Medina, N.; Sundqvist,
B. U. R. Rapid Communications in Mass Spectrometry 1994, 8, 388.

5)Westman, A.; Huth-Fehre, T.; Demirev, P.; Sundqvist, B. U. R. Journal of Mass
Spectrometry 1 995, 30, 206-21 1.

6)Kaupp. G. Angewandte Chemie. lntemational Edition 1992, 31, 592-595.

33

Chapter three.
Evaluation of Sample Preparation Techniques

1. Introduction
The standard sample preparation technique for MALDI samples is the
dried droplet methodi. This technique simply involves mixing a drop of a matrix
solution with a drop of an analyte solution and allowing the solvent to evaporate.
This method tends to produce irregular samples. The same sample may have

many different shapes of crystals in it. In addition, samples can vary in the

distribution of different size crystals and the overall density of crystals2. The
crystal surfaces also vary from highly reflective to dull.

The standard preparation procedure has many weaknesses to it.
Samples that have been prepared by the dried droplet method typically have only
a couple of “sweet spots”. These few selected areas of the sample produce
much better signal at lower laser power than the rest of the sample. The

standard method has also been found to be sensitive to some contaminants such

as detergents3. In addition the dried droplet method displays different sensitivity
for different analytes. All of these aspects of the experiment and the sensitivity
for all analytes could potentially be improved with a more carefully controlled
preparation procedure.

Before it is possible to improve upon the sample preparation procedure it
is important to understand how the samples form. The first requirement is that
both the matrix and analyte must be soluble in the solvent system that is used.
The matrices are all small organic molecules which tend to be soluble in organic

solvents. The analytes in this work are peptides and proteins, Which are soluble

34

 

in aqueous solvents. The solubility of these analytes is typically enhanced in
acidic solutions. For this reason the solvents that are typically used are a
mixture of organic solvents and aqueous 0.1% trifluroacetic acid. Another factor
that is important in the preparation of samples is that there must be a large

excess of matrix relative to the amount of analyte. A typical ratio of matrix-to-

analyte is 1000:14.5, The resolution of the experiment can be adversely affected
by having too much analyte present in the sample. In addition, some samples
can be difficult to detect when the matrix-to-analyte ratio is lower than 1000:1

By careful observation of the crystallization process, using a video
microscope, it is observed that sample crystallization typically occurs in the
following manner. After the sample is completely mixed and placed on the
sample surface, the solvent begins to evaporate. The height of the drop first
decreases as the solvent evaporates. The more volatile component of the
solvent system evaporates first so that the composition of the solvent system is
constantly changing. Over the course of 30 seconds or so the droplet is gone and

all that is left are wet crystals. Once the crystals have started to form the majority

of the analyte attaches to hydrophobic crystal faces5. Due to the inclusion
process. the hydrophobicity of the analyte has been found to be a major
contributor in the response factor for analytes. In addition the pH of the solution

can affect this process, and pH has also been found to be important in the

preparation of MALDI samples7. The last step that occurs is that the individual
crystals dry out and once the crystals are completely dry they can be inserted

into the instrument and the analysis can be performed.

35

 

By careful control of the crystallization process, it should be possible to
develop an ideal sample preparation procedure. The crystals formed by this
method should bevery uniform, to produce a uniform response across the entire
sample. This method should also be very reproducible so that the results from
different experiments can be compared. In addition this technique should have
high sensitivity and allow the detection of femtomoles of analyte. The ideal
procedure would also allow for the experiment to be performed in the presence of

high concentrations of contaminants typically present in the analysis of biological

molecules. These contaminants include detergents and buffers such as sodium
dodecylsulfate and urea.

Many alternative methods of sample preparation have been developed by
myself and others. One technique that has been used is to prepare the sample
by using two layers. The first layer provides seed crystals for the second layer

which contains the analyte. The first layer can be a layer of crushed crystals or a
layer formed by rapid evaporation of the matrix solution3v3. Spin coating has

also been used in an attempt to produce a uniform sample9. Additives such as

fucose and nitrocellulose have also been evaluated for use in improving MALDI

samples1 0:1 1.

2. The Standard Method of Sample Preparation

As a way of evaluating the performance of the standard sample preparation
technique a seven component mixture was analyzed. The analytes in this

mixture were bradykinin (M.W. = 1060), angiotensin (MW. = 1296), melittin

36

(M.W. = 2847), oxidized b-chain of insulin (M.W. = 3496), insulin (M.W. = 5734),
cytochrome C (M.W. = 12, 384) and myoglobin (M.W. = 17,400) in a equimolar
seven component mixture. This mixture represents a wide mass range of
peptides and proteins. The experiment was performed two ways to determine if
there was any difference between two similar methods. The first method was to

put down a 1 uL drop of the mixture and then to place a 1 uL drop of matrix on

top of the first drop. The results from this experiment are shown in Figure 3.1.
While the signals from bradykinin, melittin and insulin are very strong, the signals
from angiotensin, oxidized b-chain of insulin and myoglobin are weak. The

second technique was to reverse this process by first placing a 1 uL drop of
matrix solution followed by the 1 uL drop of the mixture. The result of this

procedure is shown in Figure 3.2. By reversing the order the droplets are placed
the signals for bradykinin and angiotensin are less Intense. The signals from
cytochrome C and myogloblin are at least 50% more intense than with the first
technique. This simple change in technique can make a major difference in the
spectra that are produced for a mixture. When the matrix solution is placed down
first, the solvent begins to evaporate immediately. As the solvent evaporates the
matrix can begin to start the crystallization process, so that when the analyte
solution is added the analytes can more easily be incorporated onto the faces of
the crystals that have already begun to form. This process can occur even
though no crystallization Is visible before the addition of the analyte solution.

In order to further study the effects of different sample preparation

techniques, a two component mixture of 2 picomoles of bradykinin and

37

 

9:383 .0306 not“. .2: 923m 9: 3 3535 .oSEE Ewcanoo 553 on. B 826on coast: oozed mm 959“.

a. ..__u__.._.._e.e=._._.
Sees e88 .83. -.

c Villrll.llll-l.ill-lll

 

so

.mp0 \\ llll .ll.
w% or

I

 

 

 

 

\

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

  

_
.122 “m w

 

 

 

oooow

oocom

oooon

Mjsuatuj omelet

38

9389,. 8.8.6 no.5 .2: xtfiE o... 3 ooaaoa .oSSE Eocanoo :98 oz. .0 E383 5:8... 838 Nd 2:9“.

2. J=u=queEfi

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

coco —

 

 

 

 

cooou

 

Ausuejuj angels:

.Im 88.”

 

39

2 picomoles of insulin was used. The samples were prepared by the matrix first
dried droplet procedure. This mixture represents a best case mixture of two
components that both have high response factors in the MALDI experiment. In
addition, the structure of these two analytes is very different. Bradykinin is a

single peptide chain while insulin is composed of two peptide chains that are

connected by two disulfide bonds”. As shown in Figure 3.3, the power titration
of the two component mixture with the standard MALDI experiment produces
strong signals for both bradykinin and insulin. However this technique does
show some weaknesses. The most serious weakness is the intense copper ion
adduct that forms with bradykinin. The copper that forms the adduct with
bradykinin is most likely from trace amounts in the analyte and matrix. In
addition it would be desirable to have equal intensity peaks representing
bradykinin and insulin. With this simple mixture it is easy to observe that the
insulin not only produces doubly protonated analyte ions but that there is also a
small peak for the singly-charged insulin dimer. Most metal ions do not seem to
be a problem with this sample preparation. The signals from Na” and K+ are
weak and the corresponding adducts are also weak. As shown in Figure 3.4 the
metal ions are also weak in the matrix region. In addition, the region from 4000
to 6000 ns (~m/z 10- 100) has very little signal until high powers are used. The
results from the power titration of the binary mixture prepared by the dried droplet
method can now be compared with the results of other sample preparation

techniques.

40

{Wan-xi

9.389.“. .235 no.5 o... .3 coconoa .9525 «5:888 03. oz. .o 8.58% 22.9... 526m was. 9:9“.

 

 

 

 

 

 

 

2. £25.35:

.88.

. oooou
N+—IN+= . M. l.
. . W
_ B
__ M. ,_ m.
.. 9
88” M
B
. U
. .308. m

.51. . ,_ .88.. .
, .....+m. ._

 

41

Kijsuetuj eAjtejeJ

40000
_30000
#20000
-1 0000

v

 

\ {A
iiiiii
‘3 9T,“ i

 

 
  
  

.._.__.l\ a..- - _..-« - _........ ..
A
A

 

““41 ,,___._,_...-A...-.~.._.w-——-—‘

\-_A_.~..JK_

. .JL..- - -J\-~ -—-
A.

 

 

 

 

 

a :
fl . 4a
fig. 7- fl
- ::“‘ ‘.
-dmr‘ ‘

 

.-~-.—~

v
M '

MAvu~I-~—M

...~. . .—_... ...~.

__,.__~. _ -..-e..

.._ shy...

 

A

15 I00

10000

 

 

 

 

 

 

 

 

 

TIme-of-Fllght. ns

Figure 3.4 Matrix region of the power titration spectrum of the two component mixture. prepared by the dried droplet

procedure

. 3. Alternative Sample Preparation Procedures

One altemative technique that was proposed by Vorm uses a two step

sample preparation process3. In this technique a layer of matrix is first placed
on the sample surface by placing a drop of the acetone saturated in matrix. The
acetone dries quickly leaving a thin layer of matrix. A small amount of analyte
solution with no matrix is then placed on top of the matrix layer. The idea is to
only dissolve some of the first layer and use the remaining matrix as seed
crystals. This technique does produce a homogeneous layer of crystals. This

technique has also been reported to produce peaks that have a good resolution

and sensitivity. In addition this type of sample is durable to being rinsed3.

The results for the analysis of the binary mixture are shown in Figure 3.5.
The Vorm method produces a strong signal for bradykinin while producing a very
‘ weak, broad signal for insulin. Bradykinin forms strong adducts with both Na“
and K*. The Cu“ adduct does not seem to be present. The relatively crowded
matrix region is shown in Figure 3.6. The metal ions of Na+ and K+ produce very
strong signals throughout the power titation. In addition the [M+Na]* peak
becomes intense earIy in the power titration. There are three major peaks in the
region of 4000-6000 ns. The most intense ion in this region is Ni“, while the
other two ions are matrix fragments. No peak is present at the predicted flight
time for the fragment at m/z = 65, from C..H3N+ . The major features of the Vorm
sample preparation procedure are that bradykinin produces a much stronger
signal than insulin and that metal ions play a very prominent role with this

technique. The intensity of metal ions could have been reduced by washing the

43

 

oSuoooa 88> oz. .3 uohonoa .oSBE 2.2858 95. on. .o 82.8% cogs... hosed mam 9:9“.

we acetates:

 

 

 

 

 

 

 

 

 

 

 

 

 

 

+5.5.

 

loooow

toooow

.oocon

loocov

 

M81103!" angels:

Ausuezuj “new

 

45

357% 1,
— i? I

«-

 

 

 

 

 

 

 

 

 

TIme-of-Fllght, ns

Figure 3.6 Matrix region of the power titration spectrum of the two component mixture, prepared by the Vorm procedure

sample, however this was not done so that affect of metal ions on the samples
could be studied.

Another two step sample preparation procedure is the crushed crystal
method. In this method a droplet of saturated matrix solution of a mixture of
acetonitrile and water is placed on the sample surface. This droplet is allowed to
dry completely to leave a collection of pure matrix crystals. The matrix crystals
are crushed under a microscope slide and all loose crystals are wiped off of the
sample surface. The crushing leaves behind a polycrystalline film. A droplet that
is saturated in the matrix and also contains the analyte is then placed on the thin

layer and allowed to air dry. The crystal formation typically start 15 seconds after

the analyte droplet is deposited3. The main advantages of this technique, as
reported by the developers of this technique, are that the sensitivity is increased
and this sample can also be washed to remove salts. 4

Performing a power titration on the binary mixture using the crushed
crystal procedure shows good response for both bradykinin and insulin. The
power titration is shown in Figure 3.7. Using this sample preparation method
bradykinin has a Na“ adduct that appears at low laser powers. In addition, at
high laser powers bradykinin also forms an adduct with Cu”. The matrix region
of this power titration, shown in Figure 3.8, is distinguished mostly by the fact that
there are relatively few peaks in the matrix region. The metal ions do not appear
until relatively high laser power. However the [M+Na]+ adduct is formed at low
power. Using the crushed crystal sample preparation procedure a simple power

titration is produced that has good response for both bradykinin and insulin.

46

 

2:802... .895 cognac on. .3 oouoqoa .2328 «cocoa—=8 92 05 .0 82828 .8on. Logan. Em 9:9“.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

47

a: Escapes:
oooom oowov omega oomow
a o , l 1 l | l- . o
so \
0% \i ll. ..
OJ 9 |\ Nasal: r...
e x l m...
cm rig-I‘ll... J... ... ... Irina}; .ooooe
.rIIIIalrnIIIIIIHIIIIIII-Ilaill._allil.i J. ., ..
.Elllllrliilhjllémrillilllilllllllili...
on ., ..E‘gj44 ..nliiLJaR... .
;: -oooow

.,
: mi...
. s... .x.
. ,_ , m
V. ._ 8o8 e
... U
f 9.
... _ .w

_ . .. ...

. ...+_ ‘ 88..

 

 

AJISUBJUI angels:

40000
r30000
~20000

10000

 

it? .31}

{5, fi-ail

iii

 

 

 

1 .

1.

g:

.A

1

 

 

 

 

 

 

 

 

1

I'
,v

A
A

10000

 

 

 

 

 

 

 

 

 

 

 

 

 

- a? -
. 3)}
flit} .41 r
i ./
m” /r" o
ll O 9
s ' N NM”

12000

Tlme-of-Fllght, ns

Figure 3.8 Matrix region of the power titration spectrum of the two component mixture prepared by the crushed crystal

procedure

Careful observation of the crystallization, by light microscope, that
occurred with the crushed crystal sample preparation method indicated that there
was actually two separate types of crystallization occurring. In order to
understand how these two different crystallization processes were occurring at
the same time, it is important to understand the sample surface that was used.
All samples were placed on top of a round sample pin that had a diameter of ~
1mm. The pin had a series of concentric circular grooves in the top surface.
Some of the crystallization was occurring at the top of these grooves where the
thin film had been crushed. However, much of the crystallization occurred inside
the circular grooves where larger crystals of matrix were trapped. For
comparison another sample preparation procedure was developed that would
isolate the bigger crystals. To meet this goal and still try to have a reproducible
surface the Vorrn and crushed crystal methods were combined. The first layer
used is the saturated acetone droplet of the Vorm procedure. On top of this first
layer analyte solution that is saturated in matrix is added. This sample
preparation method was named “the combo method.” When a power titration is
performed on the binary mixture prepared by the combo method the results are
similar to the cmshed crystal results with good response for both bradykinin and
insulin, similar to the results from the crushed crystal method. These results are
shown in Figure 3.9. The biggest difference between the results from these two
sample preparation techniques is that the Cu“ adduct of bradykinin is very
intense. Bradykinin also seems to form a second adduct with two copper ions.

In order for the charge on bradykinin to be +1 a hydrogen ion must be lost from

49

053003 388 05 E 3.39:. .0538 20:889. 02 as he ...—.50on 5:85 bosom ad 059“.

2. 2953...:

 

 

 

 

 

 

 

 

i5
5! ll.ll..li...5

:,
...:
533.: a; 1.5543,:
.5 4 . ..fl‘éiflez.
'1‘
7.1.5
55,...

£32445. 5 5.5,, ,..W

5555 55

 

 

 

L5. I"
I"!
i

:1
5 ,7
5 :G

.55.,
...

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

:5} ..

555 gjfi. 5.: 5,,
3 0? +1- m. , _,_555 688

55 555 .898 W

_ 55. W.
_ 0+m_ 5 .,_.

.z. . 5 _
.xm _,,,

 

50

the bradykin molecule. This type of hydrogen loss is rarely observed. The matrix
region of the power titration, shown in Figure 3.10, is similar to the matrix region
produced by the crushed crystal procedure. One difference between these two
procedures is that the signal for [M+Na]+ is more intense for the crushed crystal
sample preparation procedure than the combo method. Also, the peaks in the
high mass matrix region are much more intense with the combo method than the
crushed crystal method.

In order to directly compare the results from the different sample
preparation procedures, calibrated spectra were compared at a low and a high
laser power. The component spectra acquired at a laser power of 1450 are
shown in Figure 3.11. The laser power of 1450 was chosen because it is near
the threshold for the insulin peak. This laser power is a typical condition that may
be used to detect both of the analytes. At this laser power the peak for bradykinin
is saturated with both the standard and combo procedures. The cmsh procedure
provides the most even response with the bradykinin/insulin mixture. For the
three methods that produced a significant signal for insulin, the resolution is
directly proportional to the height of the peak: the higher the peak the lower the

resolution. This observation follows a general trend that'has been observed for

MALDI13. An expanded view of the region around the bradykinin peak is shown
in Figure 3.12. For every method except the combo method a Na” adduct is
present. Both the standard method and the combo method have a large Cu”

adduct. The Vonn method is the only technique that produces a K“ adduct.

51

l"*"‘-"'“1

Kusuotui annals:

40000

ii?-

 

12000

A
.A

      

i
E m h .~

A
I.

10600

,i __ .'_ -_¥,_%
1 m‘ -7

<
fl ,5 ,
, ::::fi.f‘1{};» {
, a ,

 

 

 

 

 

9
AA

 

____ ' 5“ ‘_

A
L

    

4 r

:
'.~.
‘4'

 

,é___~_ :——:———j

_<]‘ /

A
A
A

If

 
 

 

 

 

 

 

 

 

 

 

i“ Na' ll K’

 

 

 

 

 

 

 

Tlme-of-Fllght, ns

Figure 3.10 Matrix region of the power titration spectrum of the two component mixture. prepared by the combo procedure -

 

140000{

 

 

 

 

 

 

 

 

 

[B++u+
i
1200007 ;
DIL [I+2H]+2 [I+'H]+ I
: ‘_1_¢,'___L_ A A LA “—L“
1ooooo+ I
I
;; aooooi E
(I)
C
‘9 AAAAA A
.5 '" ____ -——
m i f
g |
-% eooooi i
I ‘
400001 '
if .- _- A A“ _ _ new
20000T
: s
o 1 ‘ “my "‘Hfivfifi'fl“ A ‘ i A “I
2000 4000 6000

Figure 3.11 Comparison of the component spectra taken at a laser
power of 1450 for the different preparation procedures

53

 

 

 

 

* dried droplet
i
I
600007
1 Vorm
E ;
«n . .
c I
g !
.5 400001
9 i
g i crushed crystal
0
1
20000“ i
combination '
1
j
0 l i i i i
1060 1080 1100 1120 1140
mlz

Figure 3.12 Expanded view of bradykinin peak in the comparison of the
component spectra taken at a laser power of 1450 for the
different preparation procedures

54

The spectra that were taken at the very high laser power of 2500 have two
additional features. These spectra, shown in Figure 3.13, have intense peaks
representing the doubly protonated insulin molecule. The intensity of the doubly
protonated molecule scales well with the intensity of the singly protonated
molecule. In addition a contaminant can be detected at m/z 3500. This is most
likely some b-chain of insulin from the insulin sample. Based on the expected
response factor of the b-chain of insulin, the amount of this contaminant must be
well below 2 pm. This contaminant can provide some measure of the relative
sensitivity of the preparation procedures for a minor component. Both the
crushed crystal method and the combo method produce a detectable peak for
the contaminant. At this high laser power all of the sample preparation methods
except the Vorrn method give a strong signal for insulin. However the resolution
of these peaks is low. Adducts due to Na+ and K”'fall within the mass range of
the large insulin peaks. However the Cu“ adduct be seen as a high mass
shoulder to the insulin peak, since the resolution of the spectrum is just high
enough to allow the two peaks to be differentiated from each other. In all of
these sample preparation procedures insulin also has some high mass adducts.
These adducts are formed by components of the matrix, such as the matrix itself,
and the product of water loss from the matrix, complexing with an analyte
molecule.

As mentioned previously the crushed crystal method that was evaluated
included some large seed crystals in the circular grooves of the sample pin. In

order to produce a pure crushed crystal preparation the metal surface that was

55

 

1400001‘
lB+Hl+

[|+H]+

1zooooi

El u+2lu+2
ll- .0.

1000001

 

 

l

} .
800001Ab_ ;
f - J\g

 

 

f _‘“AAL‘L ‘LL AA A
|

600001

relative intensity

 

 

 

 

 

 

 

20000 :
. l
l l ll
2000 4000 6000
rn/z

Figure 3.13 Comparison of the component spectra taken at a laser
power of 2500 for the different preparation procedures

56

used for the sample preparation was smoothed so that no ridges were visible.
When the procedure is done on this surface only one type of crystal was formed
and the crystallization started very quickly. In order to further control the crystal
formation the sample was heated with a hotplate to accelerate the evaporation of
the solvents and make the sample preparation more uniform. The temperature
of the sample block that the was used to hold the sample pins was monitored
with a thermocouple (Omega Engineering, Stamford, CT). When the
temperature equilibrated at the desired value, the droplet was placed on the
prepared sample pin. One of the most outstanding characteristics of this type of
sample preparation method is that two separate threshold fluences can be
observed. To easily observe the two thresholds the power titration was started
just above the first threshold value. The mixture used in this experiment is 2
picomoles each of angiotensin and insulin. The result of this experiment is
shown in Figure 3.14. Evidence of the first threshold can be seen in the first
spectrum of the power titration. This first spectrum contains both of the major
analytes. Only a few laser shots can be acquired at these low powers before the
products at the first threshold are depleted. The matrix region for this technique
is shown in Figure 3.15. The most outstanding aspect of the matrix region is that
there are few metal ions and virtually no signal from 4000-6000 ns. The
relatively few peaks in the matrix region indicates that few reactions are ‘
occurring. However the reactions that are necessary for the protonation of
analytes are still occurring. For the analysis of mixtures, the crushed crystal

method produces the most even response for a mixture of peptides and proteins.

57

359: Euro 8520 use 05 3 8.305 .oSBE «cocanoo 92 05 he Eszooam coach: 5260 3d 0590

m: .Egtcoéez

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

oo on oo ow cocoa as on
Us o No a , c o
O
.0
Q% 0.. l
8.,
-ooooF
oooom
1
P
B
n»
A
a
,oooon mm
a
u
w.
hm
.oooov
+E+m=

 

58

MjSUOIUI annals:

40000
~30000
20000
10000

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

_ l;
‘1‘, VIVA-Jill 2
l WM!

1 l 1W" “9321 fl
4 Ml ,1 , and”
“35! #JJ

‘ '95
3!“
Jun:
d‘

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Tlmo-of-Fllght. no

crystal method

Figure 3.15 Matrix region of the power titration spectrum of the two component mixture, prepared by the pure crushed

4. References

1)Hillenkamp, F.; Karas, M.; Beavis, R. C.; Chait, B. T. Analytical
Chemistry 1991, 63, 1 193a.

2)Westman, A.; Huth-Fehre, T.; Demirev, P.; Sundqvist, B. U. R. Journal
of Mass Spectrometry 1995, 30, 206-211.

3)Xiang, F.; Beavis, R. C. Rapid Communications in Mass Spectrometry
1994, 8, 199-204.

4)Medina, N.; Huth-Fehre, T.; Westman, A.; Sundqvist, B. U. R. Organic
Mass Spectrometry 1994, 29, 207-209.

5)Knochenmuss, R.; Dubois, F.; Dale, M. J.; Zenobi, R. Rapid
Communications in Mass Spectrometry 1996, 10, 871-877.

6)Beavis, R. C.; Bridson, J. N. Journal of Physical Chemistry 1993, 442-
447.

7)Cohen, S. L.; Chait, B. T. Analytical Chemistry 1996, 68, 31-37.

8)Vorrn, O.; Roepstorff, P.; Mann, M. Analytical Chemistry 1994, 66, 3281-
3287.

9)Perera, l. K.; Perkins, J.; Kantartzoglou, S. Rapid Communications in
Mass Spectrometry 1995, 9, 180-187.

10)Gusev, A.; Wilkinson, W. R.; Proctor, A.; Hercules, D. M. Analytical
Chemistry 1995, 67, 1034-1041.

11)Kussmann, M.; Nordhoff, E.; Rahbek-Nielsen, H.; Haebel, S.; Larsen,
M. R.; Jakobsen, L.; Gobom, J.; Mirgorodskaya, E.; Kroll-Kristensen,
A.; Palm, L.; Roepstorff, P. Journal of Mass Spectrometry 1997, 32,
593-601.

12)Voet, J.; Voet, D. Biochemistry, John Wiley & Sons: New York, 1990.
13)Dreisewerd, K.; Schurenberg, M.; Karas, M.; Hillenkamp, F.

lntemational Joumal of Mass Spectrometry and Ian Processes 1995,
141, 127-148.

60

 

Chapter four.
Suppression Effects of Peptides and Proteins Using a-cyano-4-
Hydroxycinnamic Acid as a Matrix for MALDI

1. Introduction

The technique of matrix-assisted laser desorption/Ionization time-of-flight
analysis has been vastly improved by the introduction of instrumental
innovations. The rest of the studies in this work were performed on a Voyager
Elite mass spectrometer with delayed extraction. The source regions of the
Voyager elite is very different from the ion source on the Vestec 2000. One
improvement ion source over the instrument that was initially used is that the
sample is no longer placed on 24 sample pins in a carousel. Instead the samples
are all placed on a single plate that has 100 spots on it. This plate not only
allows many more samples to be run at onetime, it also provides a much more
uniform surface for the samples to be deposited on. The uniform surface
increases the reproducibility between spectra acquired from different sample
positions and makes external calibration much more reliable.

Another improvement to the technique of MALDI is the introduction of a

reflectron. The reflectron was first introduced by Mamyrin1. A reflectron is
simply a device that has a voltage slightly greater than accelerating voltage. The
gradient of the field is in the oppisite direction as the accelerating voltage
gradient. The reflectron increases the resolution of the time-of-flight technique.
The cause of poor resolution in time-of-flight analysis is the kinetic energy spread
of the analyte ions. A reflectron provides a method to correct for this energy

spread. The reflectron is placed at the end of the field free region. When the

61

packet of ions reaches this region the ions with high kinetic energies will reach
the reflectron before ions of low kinetic energy. The high energy ions also
penetrate farther into the reflectron field than the low energy ions. The high
energy ions end up spending more time in the reflectron than the low energy
ions. The ions are then detected at a plane in space, where the low energy ions
and the high energy ions of the same mlz will arrive at the same time.

Recently the technique of MALDI-TOF has been further improved by the
introduction of delayed extraction analysis. Delayed extraction is in part based

on time-lag focussing, which was developed by Wiley and McLaren in the

1950’s2. This technique introduced a delay between ion production and ion
extraction. During this delay the ions distribute in the ion source. When the
accelerating voltage is applied, the different ions are at slightly different
potentials. The ions are then detected at a plane in space where the ions are
focussed together. An additional effect of the delay time is that neutrals from the

desorption process diffuse, so that the ions encounter fewer collisions on their

way out of the ion source. This technique was applied to MALDI by Brown et al3.

Delayed extraction has been shown to significantly improve the resolution of the

MALDI-TOF process4. In addition, delayed extraction can be used with a
reflectron to give even higher resolution.
The technique of matrix-assisted laser desorptionfronization is well suited

to the analysis of mixtures. MALDI has been used to analyze a wide variety of
mixtures. These analytes include low molecular weight compounds5,

oligosaccharides6 and mixtures of peptides and proteins7o8. This work involves

62

the analysis of mixtures of peptides and proteins. A challenge in the analysis of
mixtures by MALDI-MS is to produce uniform response for a wide range of
analytes. Two different approaches have been recommended to optimize the

analysis of mixtures of peptides and proteins. One approach involves
modification of the sample preparation process9v10. Another approach is to add

a non-absorbing matrix additive to the sample1 1. One type of mixture where a
uniform response is very important is the mixture of an analyte and an internal

standard for quantitation of the analyte. Two different methods, using additives, 2

 

have been proposed for the quantitative analysis of peptides and proteinsav1 1.
One matrix for the MALDI mixture analysis of peptides and proteins is a-

cyano-4-hydroxycinnamic acid. This matrix has been found to produce intense

signals for a wide variety of peptides with M.W.‘s of less than 6000 Da. For

proteins, this matrix produces a high degree of multiple charging, indicating that it

is very efficient at protonation”.

A wide variety of analyte characteristics have been reported to influence
the signal intensity in the MALDI experiment. In general, smaller analytes are
more easily detected than large analytes13. Using a-cyano-4-hydroxycinnamic
acid as the matrix it has been found that the degree of mass discrimination that is
observed is dependent on the pH of the solution and the rate of crystal growth“.
In addition the various analytes in a mixture often have different response
factors. The sensitivity for a given analyte in the MALDI experiment is dependent

on the hydrophobicity of the analyte as well as the basicity of the analyte.

63

Peptides that are both hydrophobic and have many basic sidechains are easily
detected in the MALDI experiment.

In addition to the factors discussed above analyte suppression can also
alter the signal intensity of an analyte in the MALDI experiment. Both matrix
suppression and analyte suppression have been discussed in the MALDI
literature. Matrix suppression describes when an analyte peak is observed with

no matrix peaks in the low mass region. This effect occurs at low matrixzanalyte . h

ratios of 10:1 to 2000:1 15:15. One explanation for this effect is that there is a

 

competition for available protons. When the matrixzanalyte ratio is low enough
the analyte(s) may be able to successfully scavenge all of the available protons.
Another type of suppression that has been reported is analyte suppression. This

effect occurs when the signal from a concentrated component actually

suppresses the signal from a minor component”. Competition for available
protons from the matrix, can occur between different analytes. A large excess of
matrix should prevent this competition from occurring. Preparing samples with a

matrix-to-analyte ratio greater than 5000 has been reported to reduce analyte

suppression". For the purpose of this work I define analyte suppression as the
ability of one analyte or group of analytes to lower the signal intensity produced
by a given amount of analyte from the signal intensity that is observed when the
analyte is analyzed by the same preparation procedure with only matrix. In this
work we examine some analyte suppression effects that occur at a high matrix-

to-analyte ratio.

2. Results and Discussion

Six Component Mixture

In order to evaluate how frequently analyte suppression occurs, a mixture
analysis on a six component mixture of peptides and proteins that ranged in
molecular weight from 1060 to 12600 Da. The analytes included bradykinin
(M.W = 1060.8) , angiotensin (M.W. = 1296,A), melittin (M.W. = 2848M), b-chain
of insulin (M.W. = 3495,lb), insulin (M.W. = 5735,l), and cytochrome C (M.W. =
12600,C). The results of this analysis at a low and a high laser power are shown
in Figure 4.1. When a high laser power was used all the analytes could be
detected. However the peaks for bradykinin and angiotensin are saturated while
the peaks from the singly and doubly protonated cytochrome C are very weak.
Figure 4.2 is a spectrum of 5 pm of cytochrome C with no other analytes. The
peaks from cytochrome C are 3 times more intense when pure cytochrome C is
analyzed. In addition analysis of the pure cytochrome C also produces peaks
from the triply charged analyte. By using a much lower laser power none of the
peaks are saturated, however cytochrome C can not be detected at all. The
remaining analytes seem to fall into three classes. Bradykinin and angiotensin
both produce intense peaks with the bradykinin peak being almost twice as
intense as the peak from angiotensin. The peaks from melittin and insulin are
about the same intensity. While the peak from the b-chain of insulin is about half
of the intensity of the melittin and insulin peaks. These variations in the
intensities of the peaks of these standards can not be fully explained by

differences in hydrophobicity and amount of basic residues.

65

 

9:38 E2888 x6 Co 5.58% .638 :9: 3. 2:9“.

 

I.

 

 

 

 

 

u oooos

.- ooooo

 

Ausuatur eArtelaJ

ooowv

 

 

 

 

0000 r

as

0 069525 Ea m .0 E2603 5.722 N6 2:9“.

ooow oooo ooow
O

 

... Doom

 

 

 

H 0000 _.

Mysuawg angels:

67

Hydrophobicities can be compared by comparing analytes on the unitless

Bull and Breese scale”. The larger the Bull and Breese value the more
hydrophillic a peptide. Peptides with positive Bull and Breese values are
hydrophilic while peptides with negative values are hydrophobic. All of these
analytes have large negative Bull and Breese values and contain similar
numbers of basic residues. In addition the subtle differences between these
analytes in hydrophobicity and number of basic residues do not correlate to the
different responses of the analytes. For instance melittin has the most basic
residues and the second most negative Bull and Breese value of the six
components in the mixture. In contrast angiontensin contains only a single basic
residue and a greater Bull and Breese value than melittin. Yet angiotensin
produces a much more intense peak than the peak that is produced from the
same amount of melittin.

The conditions used for this work were optimized to study the possiblity of
suppression in the absence of matrix saturation. In an attempt to eliminate the

possibility of matrix saturation, matrixzanalyte ratios of 15,000 to 60,000 were

used. These ratios are well above the recommended minimum of 500017. The
other sample preparation parameters were chosen to mimic a routine sample
analysis. A fast drying sample preparation method was used to produce
reproducible results. A typical solvent system of 1:1 0.1% TFAzacetonitrile was
used to for all of these experiments. The same experiment was also performed
at a much lower laser power to eliminate saturated peaks. The most obvious

results, shown in Figure 4.3, is the suppression of the signal from the b-chain of

68

40000 *
BH

relative intensity

AH
20000 I

 

 

 

 

 

 

5000

1 0000

m/z

Figure 4.3 Low power spectrum of six component mixture

69

insulin. The peak for the b-chain of insulin should be easy to detect due to the
analyte's hydrophobicity and number of basic residues. The high mlz of the b-
chain of insulin compared to bradykinin and angiotensin could partially account
for the lower single intensity. However, while the signal strength goes down

significantly for peaks greater than mlz 6000, in the range of 1000-6000 mlz the

response of analytes has been found to vary “file”.

For this mixture of six commonly used calibration compounds, there is
significant variation in the responses of the analytes. Some of the discrepancies
seem to be due to analyte suppression. In particular the signal from Cytochrome
C has been shown to be suppressed by other components of the mixture. In
addition it appears the signal from the b-chain of insulin is also being
suppressed. The analyte suppression that I have observed could be due to only
a limited number of protons because of saturation of the matrix. It is reasonable
to expect that when the MALDI process is limited by the number of protons
available, very slight differences in the analytes could be critical in the sensitivity
of the various analytes. In addition, the fact that MALDI on cytochrome C as a
single component produces a triply protonated molecule, could be an indication
that more protons are available for protonation of cytochrome C when it is pure
then when it is in a mixture. The variations of the analyte responses are such

that minor components of a mixture could not easily be detected.

70

W

In order to determine the extent of the suppression of the b-chain of insulin
signal by bradykinin, I examined a mixture of bradykinin and the oxidized b-chain
of insulin. These two analytes have similar basicity since in each case about
twenty percent of the ratios are basic. In addition these analytes are also both
hydrophobic. The b-chain of insulin is a little more hydrophillic since the cystine
residues are oxidized. Bradykinin has a Bull and Breese value of —940.
Bradykinin and the b-chain of insulin are also similar in size in comparision to the
large range of molecular weight values that are analyzed by MALDI.

This work was done at a low laser power that did not cause any of the
analyte peaks to saturate. This experiment was performed with equimolar
amounts of bradykinin and the oxidized b-chain of insulin. The experiment was
done with total analyte concentrations of 5 pm and 20 pm. Under both of these
conditions the spectrum is dominated by the bradykinin peak as shown in Figure
4.4. When the ratio between the analytes was adjusted to 1:9 bradykininzb-chain
of insulin the spectrum was dominated by the peak from the b-chain of insulin.
This experiment, Figure 4.5, was also done with total analyte concentrations of
5pm and 20 pm. At the higher total concentration the peak due to the b-chain of
insulin increased by 30% while the peak from bradykinin remained at about the
same intensity. In addition the doubly protonated b-chain of insulin, detected at a
mlz of 1740, also is more intense at the higher concentration. By comparing the

peaks that are produced in these four experiments we can begin to draw some

71

 

 

+ 5 pm total concentration
BH
60000 "

.3:
2 IbH+
2 H
.E
g 4000
:3 20pm total concentration
2

20000+

 

 

 

0
1 000 1 500 2000 2500 3000 3500
mlz

Figure 4.4 Spectrum of 1:1 mixture of bradykinin and b-chain of insulin

72

 

 

 

40000 f
5 pm total concentration
30000-1 +
le
+
BH
92‘
en
E
2
.E
0)
E- 20000
% 20pm total concentration
10000-l
0
1 000 1 500 2000 2500 3000 3500

mlz

Figure 4.5 Spectrum of 1:9 mixture of bradykinin and b-chain of insulin

73

conclusions. The peak heights for bradykinin and the b-chain of insulin are listed

in Tables 4.1 and 4.2.

 

 

 

 

 

 

 

ratio [bradykinin] Peak height
bradykinin: pmols arb. units
b-chain

Pure 0.5 11405
bradykinin

1:1 2.5 29629

1:1 10 35409

1:9 0.5 3263

1:9 2 3181

6 comp. mix 5 34780

 

 

 

 

 

Table 4.1 Peak heights of bradykinin under different conditions

 

 

 

 

 

 

 

 

 

 

 

 

Ratio [b-chain] Peak height
bradykinin: pmols arb. units
g-chain
Pure b-chain 2.5 9260
1 :1 2.5 1534
1:1 10 2576
1:9 4.5 6462
1 :9 18 10957
6 comp. mix 5 2816
Table 4.2 Peak heights of the b-chain of insulin under different conditions

74

The peak that resulted from 4.5 pm of b-chain of insulin in the 1:9 mixture
is twice as intense as the peak from 10 pm of b-chain of insulin in a ratio of 1:1.
The signal from the b-chain of insulin seems to be suppressed by bradykinin.
The signal from 2.5 pm of bradykinin in the 1:1 mixture is almost ten times as
intense as the signal from 2 pm of bradykinin in the 1:9 mixture. The b—chain of
insulin seems to be able to suppress signal from bradykinin when it is in excess.
In addition the peak intensity of both of the analytes in the 6 component mixture
seem to be similar to the peak intensities observed in the 1:1 mixtures. It
appears the most intense analyte peak in the spectrum seems to be responsible
for the majority of the suppression effects. Overall the results of these
experiments seem to be very similar when the amount of total analyte is changed
by a factor of 45 When the experiment is done with larger quantities of analytes
both signals simply get a little more intense but the general trend remains the
same, indicating that the suppression of the b-chain of insulin is not due to
competition for protons. The large ratio of matrix:analyte seems to have been
effective in preventing matrix suppression. It seems that the suppression that
we are observing is occurring through some other mechanism than a competition
for available protons. We do not seem to be at the proton supply limit for the

experiments shown in Figures 4.4 and 4.5.

Analfle Suppression by an Undetected Analfle

It is commonly accepted that large peaks suppress small peaks in the

MALDI experiment. A reasonable explanation for this trend is that there are a

75

limited number of protons available for the protonation of analytes. However it is
possible to observe suppression effects from an analyte that is not even
protonated. Some analytes are difficult to detect by MALDI. One of these
analytes is bucalin (M.W. = 1053 Da). Bucalin is a peptide that has three acidic
residues. However despite these acidic residues, bucalin is a hydrophobic
peptide. This combination of properties is fairly rare in a peptide since acidic
residues are very hydrophillic.

Spectra vary in the MALDI experiment so absolute intensity comparisons
can be difficult. These comparisons are made even more difficult when there is
no dominant analyte peak for reference. In this work I have chosen to look at
series of multiple spectra for trends. In order to isolate significant trends from
different sample it is important to have an even more homogenous sample
surface than used in the previous work. A three layer sample preparation
technique was used to produce reproducible samples. The first layer was 0.5 uL
of acetone saturated with a-cyano-4-hydroxycinnamic acid. This forms a very
thin, wide layer of matrix crystals. A 1.0 uL drop of the analyte in distilled water
was placed on top of the first layer. This droplet would dissolve some of the
matrix to create a well In the first layer. A second 0.5 pL drop of acetone
saturated with a-cyano-4-hydroxycinnamic acid was then added to the analyte

solution. This technique gave the most reproducible results as shown in the
results.
The three component mixture of bradydinin, b—chain of insulin, and insulin

was used to demonstrate this effect. Spectra from three consecutive spots are

76

shown in Figure 4.6. In these spectra bradykinin is always the most intense
peak, while insulin is just slightly less intense. These spectra also have a large
peak for the doubly protonated insulin molecule. In addition all of the analytes
produce a significant amount of adduct peaks. The high incidence of adducts is
most likely due to the type of sample preparation that was used.

When three equivalents of bucalin are added to the mixture the spectra
change, shown in Figure 4.7. While the other analytes produce large peaks,
bucalin is not detected at all. The bucalin (M.W. = 1053) peak would appear next

to the bradykinin peak. It has been observed that peptides with acidic residues

are harder to detect than peptide with basic residues19. The other three
components in the mixture all contain basic residues. In the spectra of the four
component mixture bradykinin is still the most intense peak. However the
intensity of the insulin peak is much reduced. The peak from the b-chain of
insulin seems to be only slightly suppressed by the bucalin. It is interesting to
note that the peak from the doubly protonated insulin is also suppressed to a
significant degree. The protonation of insulin seems to occur in a sequential
fashion. Apparently a critical number of protonated insulin must be formed to
increase the rate of formation of the doubly charged insulin fast enough to make
the doubly protonated insulin form in the time scale of the experiment. The
bucalin is most likely competing with the b-chain of insulin and insulin for the

hydrophobic crystal faces that are produced on the a-cyano-4-hydroxycinnamic

acid crystals.

77

 

 

 

 

 

 

 

 

 

BH+ |H+
IbH+
1500001
IH22+
l..- ,. .

.é‘
(I)
grooooo
E.
m
.2
:5
9

50000

 

 

 

 

 

 

 

1 000 2000 3000 4000 5000 6000
mlz

Figure 4.6 3 consecutive spectra of a 1:1 :1 mixture of bradykinin, b-chain
of insulin, and insulin

78

BH
1 50000!

 

 

1 00000

relative intensity

 

 

 

 

 

 

 

 

O W
1 000 2000 3000 ITI/Z 4000 5000 6000

Figure 4.7 3 consecutive spectra of a 3:1 :1 :1 mixture of bucalin,
bradykinin, b-chain of insulin, and insulin

79

The reason that a peak from bucalin does not appear in the spectrum may
be that the peptide is too acidic to accept another proton at the energies
achieved in this experiment. The effect that bucalin has on this mixture indicates

a possible downfall of mixture analysis by MALDI.

80

3. References

1)Mamyrin, B. A.; Karataev, V. l.; Shmikk, D. V.; Zagulin, V. A. Sov. Phys. JETP
1973, 37. 46-48.

2)erey, W. C.; McLaren, I. H. Review of Scientific Instrumentation 1953, 26,
1 150-1 157.

3)Brown, R. S.; Lennon, J. J. Analytical Chemistry 1995, 67, 1998-2003.

4)Vestal, M. L.; Juhasz, P.; Martin, S. A. Rapid Communications in Mass
Spectrometry 1 995, 9, 1044-1050.

5)Lidgard, R.; Duncan, M. W. Rapid Communications in Mass Spectrometry
1995, 9, 128-132.

6)Harvey, D. J. Rapid Communications in Mass Spectrometry 1993, 7, 614-619.
7)Perkins, J. R.; Smith, 8.; Gallagher, R. T.; Jones, D. S.; Davis, S. C.; Hoffman,
A. D.; Tomer, K. B. Joumal of the American Society of Mass Spectrometry

1993, 4, 670-684.
8)Billeci, T. M.; Stults, J. T. Analytical Chemistry 1993, 65, 1709-1716.

9)Xiang, F.; Beavis, R. C. Rapid Communications in Mass Spectrometry 1994, 8,
1 99-204.

10)Kussmann, M.; Nordhoff, E.; Rahbek-Nielsen, H.; Haebel, S.; Larsen, M. R.;
Jakobsen, L.; Gobom, J.; Mirgorodskaya, E.; Kroll-Kristensen, A.; Palm, L.;
Roepstorff, P. Joumal of Mass Spectrometry 1 997, 32, 593-601.

11)Gusev, A.; Wilkinson, W. R.; Proctor, A.; Hercules, D. M. Analytical Chemistry
1995, 67, 1034-1041.

12)Beavis, R. C.; Chaudhary, T.; Chait, B. T. Organic Mass Spectrometry 1992,
27, 156-158.

13)Farmer, T. B.; Caprioli, R. M. Joumal of Mass Spectrometry 1 995, 30, 1245-
1254.

14)Cohen, S. L.; Chait, B. T. Analytical Chemistry 1996, 68, 31-37.

15)Chan, T.-W. D.; Colbum, A. W.; Derrick, P. J. Organic Mass Spectrometry
1991, 26, 342-344.

81

16)Knochenmuss, R.; Dubois, F.; Dale, M. J.; Zenobi, R. Rapid Communications
in Mass Spectrometry 1996, 10, 871-877.

17)Gusev, A. |.; Wilkinson, W. R.; Proctor, A.; Hercules, D. M. Fresenius Joumal
of Analytical Chemistry 1996, 354, 4550-4563.

18)Bull, H. B.; Breese, K. Archives of Biochemistry and Biophysics 1 974, 161,
665-670.

19)Dikler, 8.; Kelly, J. W.; Russell, D. H. Journal of Mass Spectrometry 1 997, 32,
1 337.

82

Chapter five.
Prompt Fragmentation (PF) MALDI

1. Introduction

Initially, MALDI was unique in that fragmentation did not appear to occur for
the analytes that have dominated biological mass spectrometry (polypeptides).
Subsequently, fragmentation products have been observed. Metastable
processes were first detected, and the post source 'decay (PSD) experiment was
developed. The PSD experiment is performed by lowering the reflectron voltage
incrementally so that the products of metastable decay are focussed on the
detector. The initiation of prompt fragmentation in peptide analysis is a new

development, although it is notable that the process does occur to a large degree

for other types of analytes such as oliglionucleotides 1. While conditions have

been described with which prompt fragment ions can be generated, the

technique of prompt fragmentation is not in widespread use 2. High laser powers
are usually employed, which can increase chemical noise and lower mass
spectral resolution in the linear time-of-flight (T OF) MS experiment. In MALDI, in-
source fragmentation must occur on a very short time scale in order to yield
peaks in the resulting spectrum for which mlz values can be correctly assigned.

If the ions slowly dissociate in the first region of a TOF-MS ion source, they will
have a wide range of flight times, and will not be detected as a single peak.
During the development of power titrations, it was observed that at very high
laser powers there was seme extra “noise” in the high mass region. This “noise"

corresponded to fragment ions. The peaks for these fragment ions were very

83

broad and fewin number. A large number of prompt fragment ions could be
detected by delayed extraction (DE) TOF-MS, which has become an important
development in the investigation of prompt fragmentation-matrix assisted laser
desorption/ionization— mass spectrometry (PF-MALDI -MS). If prompt-
fragmentation is not really “instantaneous”, occurring in a few vibrational periods,
than DE will assist in the detection of prompt fragment ions by allowing
fragmentation to occur within the first 350 ns or more of the experiment, under

field-free conditions.

Further Fragmentation pf Prompt Fragment Ions

It is expected that prompt fragment ions would fail to further dissociate to
yield abundant PSD fragment ions. If the energy available is all deposited in the
desorption/ionization (D/l) event, that used for prompt fragmentation would not be
available for PSD. Nonetheless, the combination would be a powerful one. If
peptides undergo prompt fragmentation, and these prompt fragments could be
analyzed in the PSD experiment, then MS"-type experiments could be performed
in a reflectron TOF-MS. As with other methods, collisional activated dissociation
can be used to add additional energy to ions after extraction from the source.
However collisions can alter flight times, scatter ions away from the ion-optical
axis, and lower resolution in TOF experiments.

In this chapter a new matrix, bumetanide, is evaluated for its ability to
induce the formation of prompt fragment ions of peptides in the linear MALDI-

TOF experiment. Its use is compared to two commonly used MALDI matrices.

84

In addition, the use of bumetanide in mixed matrices will be presented. Using a
mixed matrix, sufficiently intense fragment ions can be generated such that PSD
spectra can be obtained. Thus, the MALDI/PF/PSD MS experiment is evaluated
here as well. The data will be used to determine how PF and PSD spectra differ,
and will lead to considerations of what factors influence the formation of prompt

fragment ions.

2. Results and Discussion

Experimental Details

Peptide and protein samples of dynorphin, a-melanocyte stimulating
hormone, and the b-chain of insulin were obtained from Sigma Chemical Co. (St.
Louis, MO) and Bachem (T orrance, CA). All of these peptides were used without

further purification. Dihydroxybenzoic acid (DHB), a—cyano-4-hydroxycinnamic

acid (HCCN) and bumetanide (BMN) were obtained from Sigma Chemical Co.
(St. Louis, MO) and used without further purification. All mass spectra were
obtained on a Voyager Elite reflectron time-of-flight mass spectrometer
(Perseptive Biosystems, Inc, Framingham, MA) equipped with a nitrogen laser
(337 nm, 3-ns pulse). The acceleration voltage used was 20 W for all of the
experiments. Data were acquired with the data system provided and based on a
transient recorder with 2-ns resolution. This mass spectrometer is equipped with
delayed extraction. During the delay, a reverse bias of ~ 0.1 °/o of the
accelerating potential is applied in the first region of the source. Due to the

electronics used for the switching of the voltage on the acceleration grid the

85

minimum delay time is 170 ns. For all of the linear TOF experiments, the voltage
on the accelerating grid was 95 % of the total accelerating voltage. The delay
time was varied from 220-570 ns. For the experiments with the reflectron, the
accelerating grid was set at 80% of the total accelerating voltage and the delay
time was 650 ns. All experiments were done with the guide wire set at 0.250 "/0
of the total accelerating potential.

All of the peptide-containing targets were prepared by a three-step
preparation process. First 1 uL of 3:5 acetonitrile: 0.1% TFA was placed on the
sample plate. To this droplet 0.4-0.5 uL of the analyte in 1:1 acetonitrile: 0.1%
TFA was added. The droplet of analyte solution typically contained 80 pm of the
analyte. One uL of a saturated solution of the matrix in acetone was added to
the droplet. The combination matrix was prepared by making separate solutions
of the two matrices saturated in acetone. The two solutions were then combined
in a ratio 4:1 BMN:HCCN (vol/vol). The resulting mixture was then used as the

matrix solution. The cytochrome C sample was prepared by a dried droplet

procedure outlined by Lennon 3 for proteins.

Maximizing Prompt Fragment Ions

We have been exploring matrices and approaches to increasing the
amount of prompt fragment ions formed in MALDI, such that sequence
information could be more easily obtained. An ideal matrix for sequencing by
prompt fragmentation would produce the same types of fragment ions for a wide

range of peptides and proteins with predictable relative intensities. The types of

86

fragment ions produced would include a series of N-terminal fragments and a
series of C-terminal fragments. Both N-terminal and C-terminal fragment ions
are necessary to elucidate an entire structure since prompt fragment ions can not
be detected below an mlz of 700. In this best-case scenario, the fragment ions
of the different types could be differentiated from each other. The sequence
could be determined by the mass differences between the members of the
various series. Another possible approach could make use of multiple matrices,
some of which have more types of fragment ions than the others. By comparing
and contrasting the fragment peaks that appeared in the various spectra the
individual series could be identified. For instance, if one matrix produced c", y...
and zn fragment ions and another matrix produced on and yn fragment ions, the 2"
fragment ions could easily be identified by contrasting the spectra taken with the
different matrices. I

The work discussed here focuses on characterizing the different types of
fragment ions that are produced with different matrices in the PF-MALDl
experiment. We performed a survey of commonly used and new candidate
matrix molecules. This survey included many “poor" matrix compounds, since a
matrix that produces abundant prompt fragment lens may not be considered an
ideal matrix for the standard MALDI analysis of peptides. The compound that
gives sufficiently intense fragment ions such that it was selected for further
investigation as a matrix is bumetanide. Bumetenide, 3-(aminosulfonyl)-5-
(butylamino)-4-phenoxybenzoic acid (BMN), is a sulfur-containing diuretic with

the stmcture shown in Figure 5.1. This compound has a local absorbance

87

 

ooESoEan Co 5:50QO 63988 6383320392: ocm Sexism mm 659".

A85 59.2995

com ovm own oom

Iz>> m

OI

/m o:

omw

eoueqiosqe

88

maximum in the UV-Vis spectrum (aqueous) at 335 nm (Figure 5.1) At the laser
wavelength of 337 nm the molar absorptivity of bumetanide is 5745 L/cm mole.
This value is comparable to the molar absorbtivity of 2,5-dihydr0xybenzoic acid
(DHB) at the same wavelength, which is 4213 Ucm mole. However both of these

values are fairly low compared to the molar absorptivity of or-cyano-4-

hydroxycinnamic acid (HCCN) in methanol, at the same wavelength, which is
22,194 Ucm mole. Bumetanide also has a molecular weight of 364 Da, which is
somewhat higher than that for most matrix molecules. For instance, the
molecular weights of DHB and HCCN are both less than 200 Da. One
characteristic that all three of these matrices have in common is that they have
similar melting points, between 200-250° C.

The MALDI mass spectrum of BMN as a matrix has not been previously
reported. The spectrum obtained at a laser power slightly above threshold is
shown in Figure 5.2. The peak assignments were made by mixing KCI, NaCl and
bradykinin into the BMN, and using peaks from these components for internal
calibration. The m/z values measured in this experiment were then used to
calibrate the spectrum of pure matrix, shown in Figure 5.2. The most intense
peak represents the protonated molecule, [BMN+H]*, at mlz 365. The molecular
ion region of the spectrum (Figure 5.2, inset) shows the formation of a relatively
intense molecular ion, BMN‘. at m/z 364, and an m/z 363 peak, [BMN-r-H-Hzl“.
There is a typical collection of lower m/z peaks formed by elimination of small
neutrals such as H20 and 002 from the carboxylic acid group. There is a notable

cluster of relatively intense peaks at around m/z 285, which appear to be formed

89

 

 

 

 

 

 

 

relative intensity

 

 

 

 

 

200 400 600

Figure 5.2 MALDI mass spectrum of bumetanide as a matrix

 

800

by elimination of the aminosulfonyl group from the various ionic forms of the
intact molecule ([BMN+H]*, BMN*, etc.). For instance, the [BMN+H]+ ion
eliminates an aminosulfonyl group to produce an m/z 285 ion [BMN+H-NH2802]*.

A low mass peak at m/z 74 is detected with some batches of BMN. This
peak represents C4H9NH3“, which could be formed via a 1.2-elimination of
protonated butyl amine following protonation of the secondary nitrogen. The peak
at m/z 74 could also represent an impurity in bumetanide. The matrix spectrum
also exhibits peaks at m/z values greater than that for the [BMN+H]" ions. Such
higher mass peaks typically represent species such as the proton bound dimer of
the matrix, and adducts of intense fragment ions with matrix molecules. This is
not the case here. There is no peak representing [2BMN-i-H]+ (m/z 729), for
example. These higher mass peaks appear to more likely represent ionized
products of a covalent coupling of two matrix molecules, in which there are
multiple reaction products. At higher laser powers, some products appear to be
formed by the covalent coupling of three matrix molecules. Whatever their origin,
it is important to understand the m/z values and relative intensities of matrix-
related ions in this experiment, since the utility of prompt fragmentation MALDI is
in part determined by the useful low mass limit and the ability to calibrate
accurately. This is established by the high m/z matrix peaks. In order for a
matrix to be useful, it is important to understand the high m/z peaks formed from
the matrix at higher laser powers - conditions at which PF MALDI experiments
are performed. Spectra obtained at higher power exhibit the same high m/z

peaks, and some at even higher mlz values. The major difference is that their

91

relative intensities increase. However, the working low mlz limit of PF MALDI
using BMN is essentially m/z 780. Also of note is the fact that the peaks in Figure
5.2 representing the commonly observed metal ions (Na*, K”) and alkali ion

adducts of the matrix are usually small and often not observed at all.

Prompt Fragmentation of Peptides with Different Matrices

In order to establish that PF-MALDI spectra can be very different when
generated from different matrices, and to show how spectra obtained using BMN
differ from those obtained using other common MALDI matrices, the results of
PF-MALDI using BMN were compared with the results using DHB and HCCN. To
produce prompt fragment ions, we deviate from the standard MALDI experiment
in that a long delay time of 520 ns and elevated laser powers are used. One
analyte used here to compare these three matrices is dynorphin. When
dynorphin, a somewhat hydrophobic peptide, is used as an analyte, a high
degree of fragmentation is produced in the PF-MALDI experiment as shown in
Figure 5.3. The nomenclature of peptide fragment ions is illustrated in appendix
1. The prompt fragmentation of dynorphin also follows some consistent trends in
the types of fragments formed in different matrices. Regardless of the matrix
used, there seems to be more fragment ions that include the N-terminus than the
C-tenninus. All of the PF-MALDI spectra of dynorphin contain more peaks due to
N-terminal fragment ions than C-terminal fragment ions. The fact that a greater
variety of N-terrninal fragments are detected indicates that the protonation of

residues towards the N-terminus may be preferred. In addition, all of the spectra

92

161514131211109 3 7 6 5 4 3 21xyz
G—G-FnL—R—R—l-R-P-K-L-K-W-D-N-Q
abc12345678910111213141516

 

 

 

 

 

 

 

 

C15 Y15
22 \ “
8) °9
j Y14
I
38 66H
. .. 1
1
+2
b) 615
‘1 V15
Y14
3'3 a? b14 315 1‘1
b8 Y1o C11 812 C12 b C13 \ t"15 /
83 c9 d 13 914 .,
(I 1 1.1; 1:121 /‘/Y11 & K\1 ‘

relative intensity

C) i

 

Y15
as 2"15
+2 c a1\
1x7 / 69 C10 8" I" ‘12 c14 ’15 "1\
as I‘ l 1 C12 33913
l 810\d11 d2 /y1111\b141\1b15*
l t
’ I

1000 1200 1400 1600 1800
mlz

Figure 5.3 Prompt fragmentation spectra of dynorphin with a) DHB,
b) HCCN and c) BMN

93

 

 

 

 

 

 

 

 

 

have peaks from fragment ions that are due to backbone cleavage with
accompanying side chain losses, most commonly (1" ions.

There are some differences in the fragmentation that occurs for this
peptide when each of these three matrices is used. When DHB is used as the
matrix in the PF-MALDI experiment, Figure 5.3a, an intense series of on fragment
ions and many yn, 2n + 2, and a" fragment ions are observed. Previously an

fragment ions have not been observed with PF-MALDI performed with DHB and

a variety of other matrices4. The spectmm obtained using DHB also has a large
number of tin and 2.1 + 2 fragment ions. The spectra that are acquired using
HCCN and BMN as the matrices are very similar as shown in Figure 5.3 b, c. All
of the matrices studied produce a doubly protonated analyte peak. It is
interesting to note that while the +2 peak is very intense in both the spectra
obtained with DHB and HCCN, the +2 peak is the same intensity as most of the
other peaks in the case where BMN is the matrix. When PF-MALDI spectra are
obtained for melanocyte stimulating hormone (MSH) (M.W. = 2662) as the
analyte, Figure 5.4, the peak for the doubly protonated analyte is of medium
intensity when DHB is the matrix, high intensity with HCCN and low intensity
when BMN is used as the matrix. For this peptide, the prompt fragmentation of
[MSH + H]“ seems to be particularly sensitive to the type of matrix used. The
fragmentation spectrum of MSH in DHB, Figure 5.4a, contains exclusively on, yn,

and znfragment peaks, as previously reported for prompt fragmentation fragment

ions4. In contrast, MSH produces an intense peak from the doubly protonated

94

212019181716151413121110 9 8 76 5 4 3 21xyz
A-E-K—K—D—E-G—P—YaR—M-EuH-F-R-W-G-S-P-K-D
ab012 3 4 5 6 7 8 9101112131415161718192021

 

+2 c 1

 

,1! :77 2142,21
21; 011621130) 2"1>\Y13[Z"1Y41 / 11715/ V1 2713.190“ w

I

b) «*2

x11 b21 Z 21

 

8&1

relative intensrty
‘\
"5
<9:
\ 3
2,
/
<—’;:'

a22

C) zeil‘

023

313 3 a21 I

c1 c 14 c 15 c17 '

310 11:12 I c 15m a,,
K 1 \ “1” °3 1”? Will lazl’yzl’ "
, . l '
1000 1500 2000 2500

mlz
Figure 5.4 Prompt fragmentation spectra of MSH withe a)DHB, b)HCCN,
c)BMN karat symbol above two peaks represents pairs of an
and an - NH3 ions)

95

analyte molecule with very few fragments when analyzed in HCCN (Figure
5.4b.).

When BMN is the matrix for this very hydrophilic peptide a series of an
ions are apparent in addition to the on, yn and zn ions (Figure 5.4c.). Of the three
matrices that were used in this work BMN produced a PF-MALDI spectrum with
the most peaks from prompt fragment ions. In addition the signals due to prompt
fragment ions are the most intense when BMN is used as the matrix. Using this
matrix some intemal fragment ions are also formed. In addition, a series of [an -
17] ions are also observed. For each of the an fragment ions in the series a15-a18,
there is a corresponding peak that is seventeen Daltons lower. These peaks are
shown as losses from the appropriate an fragment ions in Figure 5.4c.

The PF-MALDI spectra that we have observed for MSH and Dynorphin
. differ from the expected results. The types of fragmentation ions that have been
observed are different from previously reported results. In order to compare our
results to the results obtained in other laboratories, PF MALDI was performed
with the b-chain of insulin. This analyte has been characterized by Brown at 61!.4
using both DHB and HCCN as matrices. .When DHB is used in our instrument, at
laser powers significantly above threshold, a portion of the spectrum obtained is
shown in Figure 5.5a. In this spectrum the yn and zn +2 ions are labeled as a pair
when they appear, by the use of a bracket. The spectrum, both in terms of the
m/z values and the relative intensities, is very similar to that reported by Brown.
For both our results and the spectrum reported by Brown a number of prompt

fragment ions are formed which appear to be predominantly on, y", and zn+2

96

302928272625242322212019181716151413121110 9 8 7 65 4 3 21m

F-V-N-Q-H-L-B-G-S-H-LV-Q-A-LY-LV-B—G-Q-R-G-F-F-Y-T-P-K-A
113123456 7 a 910111213141516171819 202122 2324252627282930

 

 

 

relative Intensity

 

 

 

 

1 000 1 500 2000 2500 . 3000 3500

mlz

Figure 5.5 Prompt fragmentation spectrum of the b-chain of insulin in a) HCCN,
b) DHB (square brackets represent pairs of Yn and 2"" ions)

97

fragment ions. The most intense peak in this spectrum is not a fragment ion at
all, but the doubly charged [M+2H]+2 peak. There are also two intense 6 ions.
The C29 fragment ion produces a very intense and broad peak. In addition the C13
peak is also rather intense. All of these trends in the intensities of the peaks are
very similar for the previously reported spectra and our own.

We obtained slightly different results when we used a-cyano-4-

hydroxycinnamic acid as the matrix as shown in Fig. 5.5b. In this case we
observed predominantly yr. fragment ions. This PF-MALDI spectrum included a
series of yn ions from ya - y24. The spectrum contained only peaks from yn ions
unlike the previously reported result for the PF-MALDI spectrum of the b-chain of
insulin in HCCN, which contained both a series of cn and zn+2 fragment ions as
well as a series of yn fragment ions. The most dominant feature of the PF-MALDI
spectrum of the b-chain of insulin obtained using HCCN, is a very intense peak
representing the y" fragment. The intense peak from the y“ fragment is present
in both the previously reported PF-MALDI spectnim and our results. While the
results of our experiment are similar they are not identical. A speculation about
why this may be due to differences in the hardware used. Brown’s instrument
maintains a field free region during the delay timez. While our instrument

maintains a 0.1% reverse bias during the delay time.

98

3. Chemistry of Fragmentation
Effect of Basic Residues
By examination of how different residues affect the fragmentation
processes that occur, it should be possible to determine some of the
mechanisms of fragment ion formation. One of the most important types of

residues to understand are basic residues. Basic residues have been found to

hinder the formation of fragment ions in the PSD-MALDI experiments. The
presence of basic residues does not seem to reduce the amount of fragmentation
that occurs in the PF-MALDI experiment. All of the peptides studied have basic
residues. Both dynorphin and melanocyte stimulating hormone have a high
percentage of basic residues, yet they both readily undergo prompt
fragmentation. In addition, the b-chain of insulin contains a small percentage of

basic residues, yet it also produces many fragment ions in the PF-MALDI

experiment. This result is the opposite of the trend that is observed in PSD 5.
Basic residues actually seem to increase the amount of prompt fragmentation
that is observed. Basic residues provide a preferred site for protonation on a

peptide. When the basic residue is protonated, there is a suppression of the

random backbone cleavages, that would otherwise occurs. The fact that basic
residues do not reduce the amount of fragmentation indicates that charge
initiated chemistry is not the only mechanism of prompt fragmentation. One
product of the PF-MALDI experiment that is caused by basic residues are [bn +

H20] fragment ions. These fragments are typically associated with basic

99

 

residues5. Examples of this type of fragmentation are observed in spectra of the
b-chain of insulin and MSH which both contain a lysine residue immediately

adjacent to the C-tenninus.

Elect of Acidic Re§1du_e§

Acidic residues in a peptide seem to have an effect on the type of prompt
fragment ions that are formed, due to stabilization of transition states that can
lead to additional fragment ions. One example is the prompt fragmentation that
is produced by dynorphin, Figure 5.3. This analyte produces a series of a ions,
as - am and as - am, when DHB is used as the matrix. An intense hp and C14
fragment replaces the missing a14 fragment. The odd backbone cleavage in this
series occurs when the n+1 residue is aspartic acid. This is the only acidic
residue in the peptide. The same trend can be observed for the an ion when the

matrix used is bumetanide or HCCN. This type of fragmentation has been

observed in the PSD of charged derivatives 7. The fragmentation of charged
derivatives must occur through a charge remote process. In addition, the
bulkiness of the charge derivatizing group may prevent the peptide from
condensing into a distorted secondary structure.

Acidic residues also seem to effect the prompt fragmentation that occurs
with melanocyte stimulating hormone. The effect can be seen most easily in the
spectrum using HCCN as a matrix, Figure 5.4b. Since the PF-MALDI experiment
typically produces the least fragmentation, it is reasonable to assume that HCCN

produces only the lowest energy prompt fragment ions. The two most prominent

100

cn ions that are detected are the on and c2, ions. The on fragment occurs when
the n+1 residue is glutamic acid and the 021 fragment occurs when the n+1
residue is aspartic acid. It is interesting to note in this spectrum that the two 2,,
ions (217 and 221), both are formed via a skeletal cleavage of an acidic residue. A
possible mechanism for the formation of the 217 fragment ion could involve
charge remote fragmentation with a hydrogen shift. In this mechanism the NH-
CHR bond is cleaved, with an accompanying shift of the acidic hydrogen to the
CHR group. Assuming that the initial protonation occurred at a site in between
the cleaved bond and the C-tenninus, a zn+2 ion is produced. This mechanism
only produces a product with the observed m/z of a zn+2 ion if the fragmentation
is a charge remote process. There seems to be no charge-initiated pathway
available for a bond cleavage to produce this type of fragment ion. It is
interesting to note that of the three glutamic acid residues in MSH, none of them
form z fragment ions when HCCN is the matrix used, Figure 5.4b. The additional
carbon present in the side-chain of glutamic acid produces a less stable 7
member ring transition state, compared to the six member ring that is formed with
an aspartic acid residue.

The prompt fragmentation spectrum of the b-chain of insulin in HCCN,
shown in Figure 5.5b. also is strongly affected by the presence of acidic residues.
One of the few N-terminal fragments is a particularly intense C13 ion. This
fragment ion is the result of the cleavage of the peptide bond between cysteic
acid and glycine. Cysteic acid is the oxidized form. of the amino acid cysteine

that contains the very polar sulfate group. PF-MALDI spectrum of the b-chain of

101

insulin in HCCN, also contains relatively intense ya and ya fragment ions, which
represent cleavage of skeletal bonds involving a cysteic acid residue. A possible

mechanism for this fragmentation is shown in equation 1.

 

C02 ‘1' S02 +J|
O\ 1,0 H+ Y-L-V-N
‘5 l 1 W '
W H
.. - —> (1)
Y-L-V-N (WI—Q R +
l H O H+

 

0 l

I
H
\NfiI-Q-R-G/V
1'1 0

In this mechanism a six-membered ring intermediate is formed with the sulfate
group. The acidic hydrogen then shifts to the amide nitrogen on the C-terminal
side of the side-chain and the amid bond is cleaved. The stable products of this

reaction are 802 and C02 along with the peptide.

Aname Structure

A closer look at the results of the PF-MALDI experiment with dynorphin
and melanocyte stimulating hormone reveals that not all-of the differences in the
observed fragmentation are due to the matrix used. The PF-MALDI spectra of
dynorphin with the three different matrices all have peaks from an and CI"
fragment ions as well as peaks from on, y... and zn fragment ions. The spectra of
MSH in the same matrices do not seem to have nearly as many fragment ions

and the fragment peaks are not as intense. The an and tin fragments ions are

102

 

associated with charge-initiated high energy fragmentation processesa. This
discrepancy is most likely due to the types of residues that are in these two
different peptides. MSH contains nine large polar residues (arginine, lysine,
aspartic acid, glutamic acid, glutamine, asparagine) and six small non-polar
residues (glycine, alanine, and proline) of a total of 22 residues. Dynorphin
contains eight of these large polar residues and three small non-polar residues
out of 16 total residues. As shown in Table 5.1, the large polar residues appear
to form a number of fragment ions by mechanisms similar to those already

discussed. Small non-polar residues have been shown to restrict the

fragmentation that can occur 3. Both glycine and proline have been observed to
prevent the formation of en fragment ions that would result cleavage of the N-
CHR bond. The residues with small non-polar sidechains are least likely to
stabilize the transition state of a fragmentation process. Large polar residues

have been observed to allow more fragmentation reactions to occur in the PSD

spectra of N-terminal charged derivatives7. The observation that so much of the
fragmentation that we observe occurs through the same type of process, instead
of the processes observed in PSD of standard peptides, indicates that prompt
fragmentation shares some characteristics of PSD fragmentation of charged
derivatives. One possibility is that the fragmentation in PF-MALDI is not charge

directed.

103

 

 

 

 

 

 

 

 

 

 

 

Peptide # residues % acidic % basic % polar Ions obsvd
MSH 22 23% 23% 46% c... yn, z.1
Dynorphin 16 6% 31% 37% on, yn, zn ,
b-chain of 30 6% 0% 6% 2:, y...
insuin zn(weak)

 

Table 5.1. Effect of type of residues on fragmentation

The secondary structure of the analyte is also likely to be a factor to

consider in the mechanism of prompt fragmentation. As already discussed the

charged derivatives would have little secondary structure. The fact that so much

stabilization seems to occur by neighboring sidechains, in a manner similar to

that observed for charged derivatives, suggests that the secondary structure of

the analyte during fragmentation is linear. In addition, the absence of a series of

bn ions indicates that there is no significant secondary structure. To form the

typical b ion, significant secondary structure is necessary to produce a stable

protonated oxazolone 9. Possible fragmentation mechanisms to produCe cn and

yr, prompt fragment ions are shown in equations 2 and 3 respectively.

104

 

 

R' ,, H+ —>
R r—j
H o H WV

I——_l
HZN
O
H+
R!

NH
\Rn MN 2
yHI/w O
Egg/E (3)
+

The on ion could be formed by a transfer of a hydrogen from the n+1 sidechain to
the nitrogen at the N-terminus of the same residue. The N-CHR bond would then
cleave. Although this mechanism is shown as a charge remote process it could

also occur as a charge-initiated process. This mechanism takes into account the

need for an n+1 residue that is not glycine or proline and the proposed linear

105

 

structure of the peptide at the time of fragmentation. The prompt fragmentation
to form yn ions could occur through a hydrogen shift from the alpha carbon to the
nitrogen on the C-terrninal side with an accompanying cleavage of the amide

bond.

4. Combining Matrices

Only a limited amount of mechanistic information can be determined by
analyzing the prompt fragmentation products of a few analytes with different
matrices. If a prompt fragment could be selected for post source decay, then the
structure of the prompt fragment ion could potentially be determined. Using the
conditions that have been explored already, this experiment has not been
possible. The method of combining matrices was explored as a means to make
the prompt fragmentation - post source decay (PF-PSD) experiment possible.

Both HCCN and BMN have rather extreme characteristics as prompt

fragmentation matrices. While BMN produces very good fragmentation for some
analytes, it has a higher laser threshold and tends to produce noisy spectra. In
addition BMN performs poorly for analytes that are over 3 kDaItons.
Alternatively, HCCN has a low laser threshold and has a higher mass limit but
produces relatively little fragmentation. In an attempt to gain more control of the
type of fragment ions that are formed in the PF-MALDI experiment, mixtures of
the HCCN and DHB with BMN were explored. By combining these two matrices,
unique results can be obtained. The combination that was found to be the most

useful was 80% BMN and 20% HCCN. This combination of matrices produced

106

high mass ions and a large number of fragment ions. Figure 5.6a shows the
results of using bumetanide for the analysis of the b-chain of insulin at low power.
Only a few peaks from fragment ions are observed and their intensities are very
low. Figure 5.6b shows the spectrum of b-chain of insulin in HCCN at the same
laser power. This is a high laser power for HCCN and the spectrum is very noisy
and contains relatively few peaks. The much improved results of using the
combined matrix are displayed in Figure 5.6c. All of the peaks that were
recognizable in Figures 5.6a and 5.6b are also present in this low noise

spectrum. In addition the y11 fragment peak is still very intense. The intensity of

 

this peak makes it an ideal candidate for prompt fragmentation - post-source

decay analysis.

5. Prompt Fragmentation-Post Source Decay

The combination matrix was used for the prompt fragmentation post
source decay analysis of the y11 fragment of the oxidized b-chain of insulin,
Figure 5.7 Localized positive charges have been shown to produce simplified
post source decay spectra. The spectra shown in Figure 5.7, have a great
number of fragment ion peaks. The post source decay analysis of prompt
fragment ions could be useful in determining the structure of the prompt fragment
ions. The structure of the y" prompt fragment could have two different types of

structures. The charge could be isolated on the N-tenninal nitrogen. This

structure should fragment to produce N-ten'ninal fragment ions 7. Alternatively

there could be a wide variety of structures with the proton residing on any of the

107

 

 

 

relative intensity

 

 

 

 

Figure 5.6 Prompt fragmentation spectra with BMN, HCCN and
a mixed matrix

108

 

 

b3

relative intensity

 

al4

 

 

 

111098 765 43 2 1m
G-Q-R-G-F-F-Y-T-P-K-A
aloc123456 7 991011

a

5
b4 7W3 0 NH
c . 6' 3
4 9 NH
/ 6 3 37”“3

‘ /

1273

 

 

l
500

 

 

Figure 5.7 Post source decay of the y11 prompt fragment ion of the
b-chain of insulin

109

 

amide nitrogens. In fact the proton could be mobile during the decomposition

process 6. Alternatively the charge could be isolated on the N-terrninal nitrogen.
In the prompt fragmentation post source decay spectrum shown in Figure 5.7.
almost exclusively N-terminal fragments are observed. One surprise is that there
are also many fragments that show a further loss of 17 Da corresponding to a
loss of NH3. In some cases there are 2 NH3 losses for the same fragment. The
second NH3 loss must be from an arginine sidechain. None of the fragment ions
that do not contain arginine show a loss of NH3. However the a3 fragment ion,

which is the lowest mass N-terminal fragment ion that contains an arginine

 

residue, produces fragment ions with both a loss of one and two NH3 molecules. b
The as fragment shows the same pattem of loss of NH3 except that the [as — NH3]

peak is smaller than the as fragment peak. It is interesting to note that the

intensity of the peak from the at ion is weak and there is no peak from [a4- NH3].

There are many more fragment ions near the N-terminus than the C-terrninus.

Most likely this is due to the charge being isolated on the N-terminus. The

charge is most likely not located on the arginine residue since no peaks are

present to indicate a ye or y1o fragment ion. These peaks would most likely be

intense is the charge was localized on the arginine residue. Two other peaks of

interest are the [be + H20] and [b1o + H20] fragment ions. This type of

fragmentation is rare, however it has been observed with basic residues.5

110

6. Conclusion

In this work the use of bumetanide as a matrix for PF-MALDI has been
demonstrated. Bumetanide produces a greater number of prompt fragment ions
than other matrices that have been examined for PF-MALDI. In addition,
bumetanide can be combined with HCCN to make the to make the PF-PSD

experiment possible.

 

By analyzing the PF-MALDI spectra of a three peptides it has also been 1"
possible to make some mechanistic observations. In general both basic and
acidic residues seem to increase the number of prompt fragment ions that are
observed for a given peptide. It seems that large polar molecules increase the g,

amount of prompt fragment ions while small non-polar residues hinder the
process of prompt fragmentation. The large polar side chains are most likely
stabilizing transition states that lead to peptide backbone cleavages, indicating
that the peptides have a relatively linear secondary structure during the prompt
fragmentation process. Finally the observation that all of the fragment ions
produced in the PF-PSD of the y" fragment of the b-chain of insulin contain the
N-terminus, indicates that the fragment’s charge is isolated on the N-terminus.
Apparently some prompt fragmentation can proceeds through a charge-initiated

pathway.

 

111

7. References
1)Zhu, Y. F.; Taranenko, N. l.; Allman, S. L.; Taranenko, N. V.; Martin, S. A.;
Haff, L. A.; Chen, C. H. Rapid Communications in Mass Spectrometry 1997,
11, 897-903.
2)Brown, R. S.; Lennon, J. J. Analytical Chemistry1995, 67, 3990.
3)Lennon, J. J.; Walsh, K. A. Protein Science 1997, 6, 2446-2453.

4)Brown, R. S.; Carr, B. L. ; Lennon, J. J. Joumal of the American Society of
Mass Spectrometry 1996, 7, 225-232.

5)Dikler, 8.; Kelly, J. W.; Russell, D. H. Joumal of Mass Spectrometry 1997, 32,
1337.

6)Dongre, R. A.; Jones, J. L.; Somogyi, A.; Wysocki, V. H. Journal of the
American Chemical Society 1996, 118, 8365.

7)Liao, P.-C.; Huang, Z.-H.; Allison, J. Joumal of the American Society of Mass
Spectrometry 1997, 8, 501 -509.

8)Bouse, J. 0.; Yu, W.; Martin, S. A. Joumal of the American Society of Mass
Spectrometry 1995, 6, 822-835.

9)Yalcin, T.; Khouw, C.; Csizmadia, l. G.; Peterson, M. R.; Harrison, A. G.
Journal of the American Society of Mass Spectrometry 1 995, 6, 1 164-1 174.

112

 

Appendix

113

standard backbone cleavages occur as follows

R.
H2N\l/ \T-

 

OH

 

 

 

 

 

 

 

114

R1 0 + Rn

dz

R1 0 + Rub

WNMH

H HH

R"b- partial sidechain

115

+

H

I—Z—

x2

Ru 0

Rm 0
N—HOH

Y2

R" O Rm 0
#11260”
H H

 

 

I'IICHIGRN STATE UNIV. LIBRRRIES
lllll lllllllllllllllllllll |||||l llllllllllll llllllllllllllllll
31293018017040