I" . III.
L252.

 

 

 

 

J
“”03

55555556

 

 

LIBRARY

MIChIf‘

University

3| I State

 

This is to certify that the

 

dissertation entitled

CHEMICAL APPROACHES TO THE STABILIZATION OF
NON-COVALENT COMPLEXES IN MATRIX-ASSISTED LASER
DESORPTION/IONIZATION MASS SPECTROMETRY

presented by

Anne M. Distler

has been accepted towards fulfillment
of the requirements for the

PhD. degree in Chemistry

 

 

654 M

 

Major Professor’ 5 Signature
5—. K. 2 w3

Date

 

MSU is an Affirmative Action/Equal Opportunity Institution

-- ---.-.- -o---.-o--- -~- -~- n -

—- -.- -.-V-.-.-.--.-.—.

-.- -.-.--—

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

 

INOil 710422W55

 

“191053 230%

 

 

 

 

 

 

 

 

 

 

 

 

 

6/01 cJCIRC/Dateouepss-p. 15

CHEMICAL APPROACHES TO THE STABILIZATION OF
NON-COVALENT COMPLEXES IN MATRIX-ASSISTED LASER
DESORPTION/IONIZATION MASS SPECTROMETRY
By

Anne M. Distler

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

DOCTOR OF PHILOSOPHY
Department of Chemistry

2003

ABSTRACT
CHEMICAL APPROACHES TO THE STABILIZATION OF NON-COVALENT
COMPLEXES IN MATRIX-ASSISTED LASER DESORPTION/IONIZATION MASS
SPECTROMETRY
By

Anne M. Distler

While non-covalent complexes have been detected using matrix-assisted laser
desorption/ionization mass spectrometry (MALDI MS), often these weakly bound
complexes dissociate in the MALDI experiment. In our studies, double-stranded
oligonucleotides serve as a model for all types of non-covalent complexes. Since the
non-covalent interactions that hold the complexes together are disrupted during the
MALDI experiment, it is important to examine each phase of the MALDI process. The
various steps of the experiment will be presented and considered as points in the
experiment where the duplex is disrupted. Methods for the stabilizatiOn of non-covalent
complexes in MALDI MS will also be discussed. It is widely known that duplex
oligonucleotides interact with a variety of stabilizing species in solution. While these
compounds are often used to visualize oligonucleotides in gels or assist in X-ray
crystallographic analysis, the presence of these compounds can also stabilize duplex
oligonucleotides in the MALDI process.

The methods developed from the oligonucleotide studies were then applied to
other molecular complexes. The cytochrome c oxidase enzyme from the Rhodobacter
sphaeroides bacteria exists as a complex of four peptide subunits, two hemes, and a

variety of lipids and metal ions. Although our goal is the stabilization of the intact

complex, information can be determined from the detection of the components of this
complex. Molecular weight information was determined for all components of the
enzyme and structural information was determined for the lipids. The partial dissociation
of the cytochrome c oxidase enzyme allowed for the determination of subunit-subunit and
subunit-lipid interactions. The unique connectivity information from the partial
dissociation of the enzyme in mass spectrometry added a new dimension to the

understanding of the lipid/protein complex.

ACKNOWLEDGMENTS

My graduate studies at Michigan State University have been very challenging and
rewarding. Many people have contributed to my success along the way. I would like to
thank my family for all their support. My parents and my brothers, Mike and Brian, have
been both a source for encouragement and a source of advice. A Special thanks to Mark
for always listening and understanding.

Thank you also to my advisor John Allison who provided me with many
challenges along the way. I would also like to thank the people at the Mass Spectrometry
Facility especially Susanne Hoffmann-Benning and Rhonda Husain and the people at the
Macromolecular Structure, Sequencing, and Synthesis Facility especially Colleen Curry,
Stacy Trzos, and Joe Leykam. You have been a source of knowledge during my studies
here.

I would also like to thank Shelagh Ferguson-Miller and her group members
especially Carrie Hiser, Ling Qin, and Yasmin Hilrni. I have enjoyed working with all
of you over the past year and have benefited greatly from your expertise in biochemistry.

Thanks to the many friends I have met here including Jamie, Qian, Anne, Doug,

and Eric. I appreciate all you have done for me.

iv

TABLE OF CONTENTS

LIST OF TABLES ......................................................................................................... vii
LIST OF FIGURES ...................................................................................................... viii
LIST OF ABBREVIATIONS ........................................................................................ xii

CHAPTER 1 Introduction

Non-covalent complexes in mass spectrometry ................................................................ 1
Instrumentation ............................................................................................................... 13

CHAPTER 2 Stabilization of Double-stranded DNA in MALDI MS

Structure of DNA ............................................................................................................ 21
Oligonucleotides in MALDI MS .................................................................................... 33 '
MALDI MS ..................................................................................................................... 45
Liquid Phase: Evaporation ............................................................................................. 60
Solid Phase: Excitation .................................................................................................. 92
Solid Phase: Desorption/ionization .............................................................................. 132
Conclusions ................................................................................................................... 141

CHAPTER 3 Analysis of Cytochrome c Oxidase enzyme using MALDI MS

Introduction ................................................................................................................... 142
Structure of Cytochrome c Oxidase .............................................................................. 145
Experimental ................................................................................................................. 153
Results and Discussion ................................................................................................. 159
Conclusions ................................................................................................................... 188
Appendix A

"5-Methoxysalicylic Acid and Spermine: A New Matrix for the Matrix-Assisted Laser
Desorption/Ionization Mass Spectrometry Analysis of Oligonucleotides", Distler, A.M.;

Allison, J. J. Am. Soc. Mass Spectrom. 2001, 12, 456-462. ....................................... 189
Appendix B

"Improved MALDI-MS Analysis of Oligonucleotides through the Use of Fucose as a
Matrix Additive", Distler, A.M.; Allison, J. Anal. Chem. 2001, 73, 5000-5003. ....... 197

Appendix C
"Additives for the Stabilization of Double-stranded DNA in UV-MALDI MS", Distler,
A.M.; Allison, J. J. Am. Soc. Mass Spectrom. 2002, I 3, 1129-1137. ........................ 20]

vi

Table 2.1

Table 2.2

Table 2.3

Table 2.4

Table 2.5

Table 2.6

Table 2.7

Table 3.1

Table 3.2

Table 3.3

LIST OF TABLES

Duplexes Used in This Work

Matrices and their Applications

Matrix Additives

Proposed Steps in the MALDI Experiment

Tested Matrices and Results

Organic Dyes Tested as Oligonucleotide Matrices

Peptide Additives and their Influence on UV MALDI Spectra of Duplex 1
Sequence of R. Sphaeroides Cytochrome c Oxidase Subunits

Cytochrome c Oxidase Samples Used in this Work

Measured and Expected Masses of Cytochrome c Oxidase Subunits

vii

Figure 1.1
Figure 1.2
Figure 1.3
Figure 1.4

Figure 1.5

Figure 2.1
Figure 2.2
Figure 2.3
Figure 2.4
Figure 2.5

Figure 2.6

Figure 2.7

Figure 2.8

Figure 2.9

Figure 2.10

Figure 2.11

Figure 2.12

Figure 2.13

LIST OF FIGURES

Electrospray ionization mass spectrum of the ras-GDP protein.
MALDI mass spectrum of a PNA duplex.

Schematic of a Voyager-DE mass spectrometer.

Schematic of a Voyager-DE STR mass spectrometer.

(A) UV MALDI-TOF mass spectrum Of a protein with a molecular weight
around 5270 Da. (B) Deconvoluted ESI-FTICR mass spectrum of the
same protein. (C) Theoretical isotope distribution of the protein.

(A) Double-stranded DNA. (B) Structure of the nitrogen bases.
Double-helical structure of DNA.

X-ray crystal structure Of a d(CGCGAATTCGCG)2.

Melting temperature plot of a double-stranded Oligonucleotide.

UV melting temperature study of a duplex oligonucleotide in solution.

Negative-ion MALDI mass spectrum of the Dickerson dodecamer,
d(CGCGAATTCGCG), using HPA as the matrix.

MALDI mass spectrum of (A) an oligonucleotide taken in negative-ion
mode and (B) a peptide taken in positive-ion mode.

Schematic of the MALDI experiment.

Negative-ion MALDI mass spectrum of duplex 1 using (A) ATT as the
matrix and (B) HPA as the matrix.

The structures of (A) 2-mercaptobenzothiazole and (B) 5-chloro-2-
mercaptobenzothiazole.

Structure of ferrocene.

Structure of (A) Alizarin Red S, (B) Chlorophenol Red, and (C) Methyl
Violet.

The structures of (A) ATT, (B) MSA, and (C) HPA.

viii

Figure 2.14

Figure 2.15

Figure 2.16

Figure 2.17

Figure 2.18

Figure 2.19

Figure 2.20

Figure 2.21

Figure 2.22

Figure 2.23

Figure 2.24

Figure 2.25

Figure 2.26

Figure 2.27

Figure 2.28

UV analysis of duplex oligonucleotide incubated with ATT for 1 week at
room temperature.

UV melting study using the acetonitrile/water solvent system.

Negative-ion MALDI mass spectrum of duplex 1 using HPA as the
matrix. Only water was used to dissolve the matrix.

Four overlapped UV spectra of double-stranded DNA (1) before contact
with the gold surface, (2) immediately after the addition of the gold
surface, (3) 10 minutes after the addition of the gold surface, and (4) 30
minutes after addition of the gold surface.

Structure of Nafron®.

(A) The crystal lattice is the network formed by the matrix molecules. (B)
The absorbed energy breaks the matrix-matrix bonds. (C) The flow of
energy into the lattice breaks the bonds that hold the complex together.
(A) A crystal lattice containing the matrix molecules and analyte. (B) A
crystal lattice containing the analyte surrounded by both the matrix and a
non-absorbing component.

The structure of fucose.

A portion of the negative ion MALDI mass spectrum of duplex l with A)
ATT/sp/fucose and B) ATT/Sp as the matrix.

Figure 2.23. Bind modes of DNA. (A) External electrostatic, (B) Groove
Binding, (C) Intercalation.

Interaction of sperrnine with an oligonucleotide.

The negative-ion UV MALDI mass spectra of duplex 2 using (A)
ATT/spermine as the matrix and (B) ATT alone as the matrix.

Negative—ion UV MALDI mass spectrum of duplex 1 using ATT/Spermine
as the matrix. The DSRR value is 0.15, the WCSF is 1.00, and the SSD is
1.00.

Negative-ion UV MALDI mass spectrum of duplex 1 using ATT as the
matrix and cobalt(III) hexammine as an additive.

Detailed view of the duplex region of the spectrum shown in Figure 2.27.

ix

Figure 2.29

Figure 2.30
Figure 2.31
Figure 2.32
Figure 2.33

Figure 3.1
Figure 3.2
Figure 3.3
Figure 3.4
Figure 3.5
Figure 3.6

Figure 3.7
Figure 3.8
Figure 3.9

Figure 3.10

Figure 3.11

Negative-ion MALDI mass spectra of duplex 1 with A) no peptide present
and B) with B-MSH present. ATT:sp was used as the matrix for both
experiments.

Negative-ion MALDI mass spectrum of duplex 1 with 100 frnol of [3-
MSH and ATT:sp as the matrix.

The negative-ion MALDI mass spectrum of duplex l with methylene blue
as an additive. ATT was used as the matrix.

A portion of the negative—ion MALDI mass spectrum of duplex l with
methylene blue (MB) as an additive. ATT:sp was used as the matrix.

The negative-ion MALDI mass spectrum Of duplex 1. HPA was used as
the matrix with ethidium (Et) bromide as an additive.

Schematic of a cell membrane.

Structure of cytochrome c oxidase from R. sphaeroides.

The X-ray crystal structure of cytochrome c oxidase.
Structure of phospholipids found in R. sphaeroides bacteria.
Structure of the glycolipids found in R. sphaeroides bacteria.
Structure of cardiolipin molecule.

Positive-ion linear MALDI-TOF spectrum of cytochrome c oxidase
sample CcOl using SA as the matrix.

Lipid region of the postive-ion reflectron MALDI-TOP spectrum Of
cytochrome c oxidase sample CcOl using DHB as the matrix.

Positive-ion PSD MALDI-TOF Spectra of the m/z 786 peak of (A) PC
(18:1,18: 1) standard and (B) cytochrome c oxidase sample CcOl.

Lipid region of positive-ion reflectron MALDI-TOE Spectra for different
stages Of cytochrome c oxidase (A) unpurified, (B) Ni-column purified,
and (C) Mono-Q purified.

PSD MALDI-TOP Spectrum of the 852 peak of cytochrome c oxidase
sample CcOl. DHB was used as the matrix. The structure of heme a is
Shown as well.

Figure 3.12

Figure 3.13

Figure 3.14

Figure 3.15

Figure 3.16

Figure 3.17

A portion of the positive-ion linear MALDI—TOF spectra of cytochrome c
oxidase sample Cc03 (A), sample CcOI (B), and sample C002 (C),
showing subunits IV' and IVA'. DHB was used as the matrix for each
experiment.

(A) Positive-ion MALDI-TOF spectrum Of Subunit IV A' using DHB as the
matrix. (B) Deconvoluted ESI-FTICR spectrum of subunit IVA'. (C)
Theoretical isotope distribution for subunit IV A'.

A portion Of the positive ion linear MALDI-TOF spectrum of cytochrome
c oxidase sample CcOI. DHAP was used as the matrix.

Positive-ion linear MALDI-TOF spectrum of cytochrome c oxidase
sample CcOl. SA was used as the matrix with sucrose as a matrix
additive.

Positive-ion linear MALDI-TOF spectrum of cytochrome c oxidase.
SA/sucrose was used as the matrix.

Positive-ion reflectron MALDI-TOP spectra of cytochrome c oxidase after

(A) before and (B) after crystal growth process. DHB was used as the
matrix.

xi

LIST OF ABBREVIATIONS

ACN ................................................................. Acetonitrile

ATT .................................................................. 6-Aza-2-thiOthymine

CcO .................................................................. Cytochrome c oxidase

CHCA .............................................................. a-Cyano-4-hydroxycinnamic acid
CL .................................................................... Cardiolipin

Da ..................................................................... Dalton, mass unit

DAHC .............................................................. Diammonium hydrogen citrate
DGD ................................................................. Digalatosyldiacylglycerol
DHAP ............................................................... 2,4-Dihydroxyacetophenone
DHB ................................................................. 2,5-Dihydroxybenzoic acid
D/I .................................................................... Desorption/ionization

DNA ................................................................. Deoxyribonucleic acid '

ESI .................................................................... Electrospray ionization

Et ...................................................................... Ethidium bromide

FTMS ............................................................... Fourier transform mass spectrometry
GDP .................................................................. Guanosine diphosphate

GTP .................................................................. Guanosine triphosphate

His .................................................................... Histidine

HPA .................................................................. 3-Hydroxypicolinic acid

Ht ...................................................................... Histidine Tag

ICR ................................................................... Ion cyclotron resonance
IR ...................................................................... Infrared

xii

LC .................................................................... Liquid chromatography

[M-HT ............................................................. Deprotonated analyte

[M+H]+ ............................................................. Protonated analyte

Mle ................................................................ Double-stranded Oligonucleotide
MALDI ............................................................ Matrix-assisted laser desorption/ionization
MB ................................................................... Methylene blue

Met ................................................................... Methionine

MGD ................................................................ Monogalatosyldiacyl glycerol
MSA ................................................................. S-Methoxysalicylic acid

MS .................................................................... Mass spectrometry

MW .................................................................. Molecular weight

m/z .................................................................... Mass-to-charge ratio

NMR ................................................................ Nuclear magnetic resonance
PAGE ............................................................... Polyacrylamide gel electrophoresis
PC ..................................................................... Phosphatidyl choline

PE ..................................................................... Phosphatidyl ethanolamine

PG .................................................................... Phosphatidyl glycerol

PNA .................................................................. Peptide nucleic acid

ppm .................................................................. Parts per million

PS ..................................................................... Phosphatidyl serine

PSD .................................................................. Post-source decay

RNA ................................................................. Ribonucleic acid

SA .................................................................... Sinapinic acid

xiii

SDS .................................................................. Sodium doceyl sulfate

SL ..................................................................... Sulfolipid

sp ...................................................................... Spermine

T ....................................................................... Matrix

TF A .................................................................. Trifluoroacetic acid

THAP ............................................................... 2,4,6—Trihydroxyacetophenone

Tm ..................................................................... Melting temperature

TOF .................................................................. Time-Of-flight

TRIS ................................................................. tris[hydroxymethyl]-aminomethane
UV .................................................................... Ultraviolet

xiv

CHAPTER 1. INTRODUCTION

1. Non-covalent Complexes in Mass Spectrometry

In organic synthesis, the main focus is the making and breaking of covalent
bonds. In the field of biochemistry, many processes involve non—covalently bound
complexes. These types of complexes have great significance since cellular function is
often triggered by weak non-covalent interactions between enzyme and substrate, protein
and drug, and antibody and antigen [1]. These complexes are under intensive
investigation in order to achieve a better understanding of how the human body functions
or malfunctions. Often drug discovery efforts involve the development of new
compounds that non-covalently bind to proteins in order to prevent a disease [2]. The
detection of these complexes is a crucial part of drug development.

Analyses of non-covalent complexes have traditionally been performed using
techniques such as gel permeation chromatography [3], centrifirgation [4], and sodium
dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) [5]. While frequently
used to study biomolecular complexes, these techniques provide little or no information
about the molecular mass and binding stoichiometry of the complexes. The more
advanced techniques of x-ray crystallography and nuclear magnetic resonance (N MR)
can provide detailed information about the structure of the complex. However, these
techniques can be difficult and time consuming. For example, when using x-ray
crystallography, homogenous crystals are needed for the analysis. The growth process
for the crystals is lengthy, tedious, and Often unsuccessful. For this reason, mass

spectrometry would be a useful technique for analyzing non-covalent complexes to

determine molecular weight information rapidly and without extensive sample
preparation.

In order to detect non-covalent complexes using mass spectrometry, the
desorption/ionization technique must meet several conditions. Most importantly the
ionization method must transfer enough energy to the molecule to allow for its ionization
and desorption, but not enough energy to cause the dissociation of the complex. Also, a
mass spectrometry technique that allows the complex to be analyzed while in a buffered
solution would also be useful. Non-covalent complexes are often sensitive to changes in
temperature or pH of the solution. If the analysis could be performed in a solution
buffered to biological pH, the non-covalent complex would be more stable during the
course of the experiment.

Two desorption/ionization techniques potentially meet the requirements for
analyzing non-covalent complexes: matrix-assisted laser desorption/ionization (MALDI)
and electrospray ionization (ESI). The developments of MALDI [6] and E81 [7] pushed
mass spectrometry into the realm of biomolecules. Recently, mass spectrometry has
become an important technique in the analysis of proteins, DNA, carbohydrates, and
lipids [1, 8-10]. With the use of MALDI or ESI, the mass range for mass spectrometric
analysis increased to include proteins and other biomolecules weighing over 100
kilodaltons (kDa). The amount Of sample needed for analysis also decreased to the
picomole (pmol) level.

The technique of electrospray ionization has had success when analyzing a
variety of non-covalent complexes [1 1-22]. In specific, protein-DNA complexes [23-24],

DNA-drug [25], and catechin-peptide [26] complexes have also been detected using ESI-

MS. ESI—MS is advantageous because it allows the sample to be introduced in a buffered
solution without requiring the formation of a solid phase target. In ESI-MS, the
experiment yields a distribution of peaks arising from differing charge states of the
analyte. The charge state of the analyte is dependent on the analyte size, as well as the
pH of the starting solution with a lower pH providing a more highly charged analyte ion.
The mass analyzer in ESI-MS is often a quadrupole or ion trap with a maximum
detectable mass-to-charge ratio of 2000. Due to the limited mass range, the analyte needs
to exist in a highly charged state. For example, the protein bovine serum albumin has a
molecular weight of around 66,000 g/mol. In order to be detected using a quadrupole ion
trap, the protein must accumulate greater than 33 charges. This can lead to complications
since the acidic conditions necessary for E81 analysis can denature non-covalent
complexes [1,2].

While ESI-MS has had many documented successes in the analysis of non-
covalent complexes, there are complications and this type of analysis is still not routine.
One protein of particular interest is the ras protein. The ras protein is involved in the
formation of a wide range of human tumors [27] and plays a central role in the signaling
of cell growth [28]. This protein exists in either an inactive form with a guanosine
diphosphate (GDP) non-covalently bound to the protein or an active form with a
guanosine triphosphate (GTP) non-covalently bound to the protein [29]. In oncogenic
ras, the mutated ras protein becomes locked in the active form, resulting in unregulated
cell growth and tumor formation. The Ganguly group designed and synthesized a series
of drugs to bind, non-covalently, to the mutated ras protein to inhibit the conversion Of

the inactive form of the ras protein to the active form [1,30]. The Ganguly group was

interested in studying these complexes using ESI-MS. Before the ras-GDP-drug
complexes could be studied, it was important to stabilize the ras-GDP complex in the
experiment. When analyzing the ras-GDP complex, the majority of the peaks present in
the spectrum represented the dissociated complex, Figure 1.1. When looking at the
deconvoluted ESI spectrum shown in the inset of Figure 1.1, the peak representing the

dissociated complex is more intense than the peak representing the intact complex.

a:

 

[ =1:

 

 

 

 

 

 

 

.é‘
U) h
C
9.1
C
_t
(D
.2
10-!
5..“
a)
a:
I III
‘.Q.AA.JLA.4 1 ‘. ‘ . . u

 

m/z

Figure 1.1. Electrospray ionization mass Spectrum of the ras-GDP protein. The inset
Spectrum is the deconvoluted mass spectrum. The asterisks denote peaks representing the
intact complex. The other peaks in the spectrum represent the components of the
complex. This figure was adapted from reference 1.

While the detection of non-covalent complexes in ESI-MS has been documented, '
there are disadvantages for the technique. As shown in Figure 1.1, there is a distribution
of peaks in the electrospray spectrum that represents each Species present in solution.

There are over 12 peaks representing the ras protein and 3 peaks representing the ras-

GDP complex detected in the experiment. As a result of this charging, spectra for

mixtures can be very complex with multiple peaks for each component. Samples for
ESI-MS analysis must also be highly purified in order to prevent clogging of the capillary
tubing by non—volatile salts. While the presence of salts is'not necessary for the
stabilization of all analytes, cations are known to stabilize biomolecules such as duplex
oligonucleotides. The removal of these salts may destabilize the complex. Also, when
hydrophobic forces are involved in the formation of non-covalent complexes, the solvents
used in mass spectrometric analysis are very important. Under the conditions needed for
electrospray ionization, the non-covalent interactions are often destroyed [31]. Although
ESI-MS has its advantages, the complicated spectra, organic solvents, and extensive
sample purification necessary for E81 analysis make MALDI an attractive choice for
analysis of biomolecules.

While the desorption/ionization technique of MALDI should allow for the study
of non-covalent complexes, often such complexes are not detected on a routine basis. For
example, a peptide nucleic acid (PNA) is structurally similar to proteins and
deoxyribonucleic acid (DNA). PNA has shown promise for the development of gene
therapeutic agents, diagnostic devices for genetic analysis, and as molecular tools for
nucleic acid manipulations [32]. When duplex PNA is analyzed using MALDI MS,
interesting results are seen [33], Figure 1.2. The actual sequence of the PNA duplex is
not important. It is only important to note that it consists of PNA strand 1 non-covalently
bound to PNA strand 2. The peak representing the non-covalent complex, labeled (1+2),
is detected, but it is of low intensity when compared to the peaks representing the single

strands, labeled (1) and (2). Also, peaks representing the homodimers, (1+1) and (2+2),

are detected as well. When ions such as (1+1) are formed, and 1 is not known to bind to

itself, this may suggest that the non-covalent complex

 

 

1
2
.E
(D
C
.9
E
Q)
.2
E
0
DC
1 +2
h 1+1 2+2

 

 

m/z

Figure 1.2. MALDI mass spectrum of a PNA duplex. This figure was adapted from
reference 33.
completely dissociates early in the experiment. The single strands then randomly
associate and co-precipitate during growth of the MALDI target crystals. The
dissociation of the complexes in MALDI has been seen with other types of non-covalent
complexes including those involving peptides bound to DNA [34]. RNA-peptide
complexes have been detected using MALDI MS although the peaks representing the
components of the complex are more intense than the peak representing the intact
complex [3 5].

Since MALDI MS does not allow for the routine detection of non-covalent
complexes, the MALDI process must be closely examined to find the cause of the

dissociation. In order to explore MALDI analysis of non-covalent complexes, a model

analyte for all types of complexes can be employed. For these studies, DNA has been
selected. Double-stranded DNA is an example of a non-covalent complex in which the
Single strands of DNA are held together by hydrogen bonds and are stabilized through
other non-covalent interactions. Using a technique such as MALDI MS, a double-
stranded oligonucleotide can be desorbed and ionized as a singly charged, gas phase ion
although only a few examples have been reported [36-3 8]. Often, there is no direct
evidence that Watson-Crick interactions remain intact throughout the MALDI experiment
and in the gas phase. When a double-stranded oligonucleotide, MiMz, is analyzed using
various desorption/ionization techniques, ions representing the two single strands are
detected [39]. Since MiMz is not detected as (MiMz-H)‘ or (M1M2+H)+ in MALDI MS,
the duplex must dissociate at some point before acceleration and detection of the ions.
Since molecular complexes are difficult to maintain in the MALDI experiment, it
is important to more closely study the behavior of these complexes. The various steps of
the MALDI process will be examined in order to define the point in the MALDI
experiment where the complex dissociates. In these studies, double-stranded
oligonucleotides will serve as a model for all types of non-covalent complexes. Since the
non—covalent interactions that hold the complexes together are disrupted during the
MALDI experiment, each phase of the MALDI process will be examined. The various
steps of the experiment will be presented and considered as points in the experiment
where the duplex is disrupted. Methods for the stabilization of non-covalent complexes
in MALDI MS will also be discussed. It is widely known that duplex oligonucleotides
interact with a variety of stabilizing species in solution. While these compounds are

often used to visualize Oligonucleotides in gels or assist in X-ray crystallographic

analysis, the presence of these compounds could also stabilize duplex oligonucleotides in
the MALDI process.

The methods developed from the oligonucleotide studies were then applied to
other molecular complexes. The cytochrome c oxidase enzyme from the Rhodobacter
sphaeroides bacterium exists as a complex of four peptide subunits, two hemes, and a
variety of lipids and metal ions. Although our goal is the stabilization of the intact
complex, information can be determined from the detection of the components of this
complex. Molecular weight information was determined for all components of the
enzyme and structural information was determined for the lipids. The lipid content of the
enzyme solution was also examined after various purification steps in order to determine
which lipids were removed after the purification processes. The partial dissociation of the
cytochrome c oxidase enzyme allowed for the determination of subunit-subunit and
subunit-lipid interactions. The unique connectivity information from the partial
dissociation of the enzyme in mass spectrometry added a new dimension to the

understanding of the lipid/protein complex.

REFERENCES

10.

11.

Pramanik, B.N.; Bartner, P.L.; Mirza, U.A.; Liu, Y.H.; Ganguly, A.K.
Electrospray ionization mass spectrometry for the study of non-covalent
complexes: an emerging technology. J. Mass Spectrbm. 1998, 33, 911-920.

Ganguly, A.K.; Pramanik, B.N.; Huang, E.C.; Liberles, S.; Heimark, L.; Liu,
Y.H.; Tsarbopoulos, A.; Doll, R.J.; Taveras, A.G.; Remiszewski, 8.; Snow, M.E.;
Wang, Y.S.; Vibulbhan, B.; Cesarz, D.; Brown, J.E.; del Rosario, J .; James, L.;
Kirschmeier, P.; Girijavallabhan, V. Detection and Structural Characterization of
Ras Oncoprotein-Inhibitors Complexes by Electrospray Mass Spectrometry.
Bioorg. Med. Chem. 1997, 5, 817-820.

Compagnini, A.; Fisichella, S.; Foti, S.; Maccarrone, G.; Saletti, R. Isolation by
gel-perrneation chromatography of a non-covalent complex of Cibacron Blue
F3G-A with human serum albumin. J. Chromatography A. 1996, 736, 115-123.

Chen, S.Y.; Matsuoka, Y.; Compans, R.W. Assembly of G1 and G2 glycoprotein
oligomers in Punta Toro virus-infected cells. Virus Res. 1992, 22, 215-225.

Robyt, J .F .; White, 3.]. Biochemical techniques: Theory and Practice;
Waveland Press, Inc.: Prospect Heights, Illinois, 1987.

Karas, M.; Bachmann, D.; Bahr, U.; Hillenkamp, F. Matrix-assisted ultraviolet-
laser desorption of nonvolatile compounds. Int. J. Mass Spectrom. Ion Processes.
1987, 78, 53-68.

Fenn, J .B.; Mann, M.; Meng, C.K.; Wong, S.F.; Whitehouse, C.M. Electrospray
ionization for the mass spectrometry of large biomolecules. Science. 1989, 246,
64-71.

Nordhoff, E.; Kirpekar, F.; Roepstorff, P. Mass spectrometry of nucleic acids.
Mass Spectrom. Rev. 1996, 15, 67-138.

Guo, B. Mass spectrometry in DNA analysis. Anal. Chem. 1999, 71, 333-337.

Griffiths, W.J.; Jonsson, A.P.; Liu, S.; Rai, D.K.; Wang, Y. Electrospray and
tandem mass spectrometry in biochemistry. Biochem. J. 2001, 355, 545-561.

Bayer, E.; Bauer, T.; Schmeer, K.; Bleicher, K.; Maier, M.; Gaus, H. Analysis of
double-stranded oligonucleotides by electrospray mass spectrometry. Anal.
Chem. 1994, 66, 3858-3863.

12.

l3.

14.

15.

16.

17.

18.

19.

20.

21.

22.

23.

Ding, J .; Anderegg, R.J. Specific and nonspecific dimer formation in the
electrospray ionization mass spectrometry of oligonucleotides. J. Am. Soc. Mass
Spectrom. 1995, 6, 159-164.

Cheng, X.; Gao, Q.; Smith, R.D.; Jung, K.; Switzer, C. Comparison of 3',5'- and
2',5'- linked DNA duplex stabilities by electrospray ionization mass spectrometry.
Chem. Commun. 1996, 747-748.

Smith, R.D.; Light-Wahl, K]. The Observation of noncovalent interactions in
solution by electrospray-ionization mass spectrometry — promise, pitfalls, and
prognosis. Biol. Mass Spectom. 1993, 22, 493-501.

Smith, D.L.; Zhang, Z. Probing noncovalent structural features of proteins by
mass-spectrometry. Mass Spectrom. Rev. 1994, 13, 411-429.

Przybylski, M.; Glocker, M.O. Electrospray mass spectrometry of
biomacromolecular complexes with noncovalent interactions - new analytical
perspectives for supramolecular chemistry and molecular recognition processes
Angew. Chem. Int. Ed. Engl. 1996, 35, 806-826.

Smith, R.D.; Bruce, J .E.; Wu, Q.; Lei, Q.P. New mass spectrometric methods for
the study of noncovalent associations of biopolyrners. Chem. Soc. Rev. 1997, 26,
191-202.

Loo, J.A. Studying non-covalent protein complexes by electrospray ionization
mass Spectrometry. Mass Spectrom. Rev. 1997, 16, 1-23.

Winston, R.L.; Fitzgerald, M.C. Mass spectrometry as a readout of protein
structure and function. Mass Spectrom. Rev. 1997, 16, 165-179.

Veenstra, T.D. Electrospray ionization mass spectrometry in the study of
biomolecular non-covalent interactions. Biophys. Chem. 1999, 79, 63-79.

Schalley, C.A. Molecular recognition and supramolecular chemistry in the gas
phase. Mass Spectrom. Rev. 2001, 20, 253-309.

Daniel, J .M.; Friess, S.D.; Rajagopalan, S.; Wendt, S.; Zenobi, R. Quantitative
determination Of noncovalent binding interactions using soft ionization mass
spectrometry. Int. J. Mass Spectrom. 2002, 216, 1-27.

Gabelica, V.; Vreuls, C.; Filee, P.; Duval, V.; Joris, B.; De Pauw, E. Advantages

and drawbacks of nanospray for studying noncovalent protein-DNA complexes
by mass spectrometry. Rapid Commun. Mass Spectrom. 2002, 16, 1723-1728.

10

24.

25.

26.

27.

28.

29.

30.

31.

32.

33.

Donald, L.J.; Hosfield, D.J.; Cuvelier, S.L.; Ens, W.; Standing, K.G.; Duckworth,
D.C. Mass spectrometric study of the Escherichia coli repressor proteins, IcIR
and GcIR, and their complexes with DNA. Protein Sci. 2001, 10, 1370-1380.

Triolo, A.; Arcamore, F .M.; Raffaelli, A.; Salvadori, P. Non-covalent complexes
between DNA-binding drugs and double-stranded deryoligonucleotides: a study
by ionspray mass spectrometry. J. Mass Spectrom. 1997, 32, 1186-1194.

Sami-Manchado, P.; Cheynier, V. Study of non-covalent complexation between
catechin derivatives and peptides by electrospray ionization mass spectrometry.
J. Mass Spectrom. 2002, 37, 609-616.

Nishimura, S.; Sekiaya, T. Human cancer and cellular oncogenes. Biochem. J.
1987, 243, 313-327.

Hall, A. The cellular functions of small GTP-binding proteins. Science. 1990,
249, 635-640.

De Vos, A.M.; Tong, L.; Milburn, M.V.; Matias, P.M.; Jancarik, J.; Noguchi, S.;
Nishimura, S.; Miura, K.; Ohtsuka, E.; Kim, SH. Three-dimensional structure of
an oncogene protein: catalytic domain of human c-H-ras p21. Science 1988, 239,
888-893.

Taveras, A.G.; Remiszewski, S.W.; Doll, R.J.; Cesarz, D.; Huang, E.C.;
Kirschmeier, P.; Pramanik, B.N.; Snow, M.E.; Wang, Y.S.; delRosario, J.D.;
Vibulbhan, B.; Bauer, B.B.; Brown, J.E.; Carr, D.; Catino, J .; Evans, C.A.;
Girijavallabhan, V.; Heimark, L.; James, L.; Liberles, S.; Nash, C.; Perkins, L.;
Senior, M.M.; Tsarbopoulos, A.;, Ganguly, A.K.;, Aust, R.;, Brown, E.;, Delisle,
D.; Fuhnnan, S.; Hendrickson, T.; Kissinger, G; Love, R.; Sisson, W.;
Villafranca, E.; Webber S.E. Pas oncoprotein inhibitors: the discovery of potent,
ras nucleotide exchange inhibitors and the structural determination of a drug-
protein complex. Bioorg. Med. Chem. 1997, 5, 125-133.

Robinson, C.V.; Chung, E.W.; Kragelund, B.B.; Knudsen, J ..; Aplin, R.T.;
Poulsen, F.M.; Dobson, C.M. Probing the nature of noncovalent interactions by

mass spectrometry: a study of protein-COA ligand binding and assembly. J. Am.
Chem. Soc. 1996, 118, 8646-8653.

Ray, A.; Nordén, B. Peptide nucleic acid (PNA): its medical and biotechnical
applications and promise for the future. FASEB J. 2000, 14, 1041-1060.

Butler, J .M.; Jiang-Baucom, P.; Huang, M.; Belgrader, P.; Girard, J. Peptide

nucleic acid characterization by MALDI-TOF mass spectrometry. Anal. Chem.
1996, 68, 3283-3287.

11

34.

35.

36.

37.

38.

39.

Lin, 8.; Cotter, R.J.; Woods, A.S. Detection of non-covalent interaction of single
and double stranded DNA with peptides by MALDI-TOF. Proteins 1998, 33,
12-21. Supplement 2.

Thiede, 3.; von J anta-Lipinski, M. Noncovalent RNA-peptide complexes
detected by matrix-assisted laser desorption/ionization mass spectrometry. Rapid
Commun. Mass Spectrom. 1998, 12, 1889-1894.

Lecchi, P.; Panell, L.K. The detection of intact double-stranded DNA by
MALDI. J. Am. Soc. Mass Spectrom. 1995, 6, 972-975.

Kirpekar, F.; Berkenkamp, S.; Hillenkamp, F. Detection of double-stranded DNA
by IR- and UV-MALDI mass spectrometry. Anal. Chem. 1999, 71 , 2334-2339.

Little, D.P.; Jacob, A.; Becker, T.; Braun, A.; Darnhofer-Demar, B.; Jurinke, C.;
van den Boom, D.; Kester, H. Direct detection of synthetic and biologically
generated double-stranded DNA by MALDI-TOF MS. Int. J. Mass Spectrom. Ion
Processes. 1997,169/1 70, 323-330.

Tang, K.; Taranenko, N.I.; Allman, S.L.; Chen, C.H.; Chang, L.Y.; Jacobson,

K.B. Picolinic-acid as a matrix for laser mass-spectrometry of nucleic-acids and
proteins. Rapid Commun. Mass Spectrom. 1994, 8, 673-677.

12

II. Instrumentation
Time-of—F light Mass Spectrometry

Matrix-assisted laser desorption/ionization mass spectrometry is a pulsed
technique. For this reason, MALDI is often coupled to a time-of-flight (TOF) mass
analyzer. In TOF-MS, the ions formed in the MALDI process are accelerated out of the
ion source to a fixed kinetic energy and the velocity of the ions is dependent on their
mass-tO-charge (m/z) ratio. Ions Of different m/z values will then separate from each
other and strike the detector at varying times. The m/z values of the ions can then be
determined by measuring the flight time. In most TOF instruments, the accelerating
voltage is between 10-30 kV. The path lengths are typically 0.5-3 In.

Time-of-flight mass analyzers have several advantages when compared to other
types of mass analyzers. First, TOF mass analyzers allow for the analysis of ions with
large m/z values. Time-of-flight instruments also have good mass accuracy with
deviations of 1 Da for m/z 5000 and is the only mass analyzer to measure within 100 Da
for m/z 50,000 [1]. The TOF instruments also exhibit great sensitivity and scanning
Speed. Many recent advancements in TOF technology have contributed to the success of
the technique [2].

However, TOF instruments have several disadvantages. Time-of-flight mass
analyzers have a poor dynamic range. There is also a no potential for tandem mass
spectrometry (MS/MS) experiments on a linear TOF and only a limited with potential
reflectron TOF instruments. While deviations of 1 Da can be seen with ions around m/z
5000, this still does not allow for exact mass determination. Time-of—flight is also limited

by the resolution. In both linear and reflectron TOF instruments, the ions formed from

13

MALDI often have a large energy distribution. The initial energy spread of the ions leads
to peak broadening at the detector leading to poor mass resolution.

Two time-of-flight mass spectrometers were used in this work. The first is the
PerSeptive Biosystems (Framingham, MA) Voyager DE, delayed-extraction time-of-
flight linear (TOF) mass spectrometer equipped with a nitrogen laser (337 nm, 3 ns
pulse), Figure 1.3. This mass spectrometer was a linear instrument that did not have

reflectron or PSD capabilities. The length of the flight tube was 1.2 meters.

 

 

 

 

 

 

 

 

 

i
l Linear
‘ detector
I I
‘ l I - — ->
I “1”“
. I ------ D
I I Ll”! path
I
l
l .
Flight
Beam l
guide ' tube
wire I
l
l
y I
i ' I
Laser i i | Video ’
1 camera
anthsueartor—i-l- I : ‘ Aperture (grounded)
' I . ' Ground grid
Prismvu.
. ‘I
Variable-voltage ' -. I," ‘ Sample ‘
grid ~ A loading
plate Main chamber
' I source chambe

 

 

 

Figure 1.3. Schematic of a Voyager-DE mass spectrometer. This figure was reproduced
from reference 3.

While TOF instruments have very limited capabilities for MS/MS experiments,

many reflectron instruments are capable of performing post-source decay (PSD)

l4

experiments. PSD analysis allows for the detection of fragment ions formed in the field-
free region of the mass spectrometer [4]. These fragment ions are structurally significant
and can often provide clues as to the structure of the precursor ion. For those PSD spectra
shown in this work, an instrument in the Michigan State University Mass Spectrometry
Facility (East Lansing, MI) was used. This instrument was the PerSeptive Biosystems
(Framingham, MA) Voyager STR, delayed-extraction time-of-flight reflectron mass
spectrometer, Figure 1.4. This instrument is also equipped with a nitrogen laser (337 nm,

3 ns pulse), but has a longer flight tube of 2.0 meters.

 

  
 
 

    
 
   
 

Variable-voltage Laser
grid attenuator
Laser Reflector Reflector
prism I detector (electrostatic
Sample 4 ----- mirror)
plate I - Beam
guide

   
 
 
  

  
 

‘- -. ~-_
I __ .— v’l
— - — — — — — — - ~u — —- ——

1"y inmhi
Wdeo tube

camera

wire

  

Main
source
chamber

   
  
  
  
 

 
   

I
'llllllllli

   

Linear

Timed lon detector

Aperture (grounded) Selector
Collision cell (optional)

 

- — ~> lon path in reflector mode
. ----- > Laser path

 

 

 

 

 

 

Figure 1.4. Schematic of a Voyager-DE STR mass spectrometer. This figure was
reproduced from reference 3.

15

FourierLralsform Mass Spectromegy

The technique of Fourier transform mass spectrometry (FTMS) is based on a
phenomenon called ion cyclotron resonance (ICR). In a uniform magnetic field, an ion
will move in a circular orbit around the center of the magnetic field in a periodic motion.
This motion is known as cyclotron motion and the frequency of this motion is known as
the cyclotron frequency. In this technique, the resolution seen in the resulting spectra is
much higher than in TOF experiments. In TOF, the resolution is lowered by the
distribution of kinetic energies present for ions Of having the same rn/z value. In FTMS,
there is no dependence on kinetic energy, only a dependence on the cyclotron frequency
of the ions. Since all ions of a given m/z value will have the same cyclotron frequency,
the resolving power of this technique is very high. A decrease in resolution during an
FTMS experiment can be caused by a space-charge effect [5]. If too many ions are
stored in the cell, ions of the same m/z value, traveling in a packet, will begin to repel
each other. This repulsion will increase the distribution of the ions in the cell decreasing
resolution and also affected the mass accuracy [6]. A number of reviews have been
written on the technique [7-8]. '

In a typical FTMS experiment, resolution of 30000 can be attained. FTMS
exhibits high mass accuracy as well with errors less than 5 parts per million (ppm). In
addition to the high resolution and mass accuracy of the technique, FTMS has the ability
to perform MS/MS experiments. Ions trapped within the FTMS cell collide with neutral
gas molecules and induce fragmentation of the analyte ions. This allows structural
information to be determined using FTMS. While the MS/MS experiments in FTMS are

similar to PSD experiments, the fragmentation patterns differ in the two techniques.

16

There are usually more fragment ions formed in the F TMS experiment and FTMS also

allows for higher order experiments (MS/MS/MS. . ..MS“).

17

Time-of-Flight versrg Fourier Trmform Mg Spectrometry

In a typical F TMS experiment, resolution of 30000 can be attained while in a
TOF experiment resolution of less than 1000 is typically seen. In order to demonstrate
the change in resolution that can be expected for F TMS, it is important to see spectra
from both techniques. In Figure 1.5(A), this is a mass spectrum from a MALDI-TOF
experiment for a protein around 5200 Da. In Figure 1.5(B), this is a deconvoluted ESI
spectrum for an FTMS experiment for the same protein. A drastic increase in resolution
is seen when compared to the resolution achieved in the TOF experiment. For
comparison, in figure 1.5(C), there is a theoretical isotope distribution for the same
protein. The FTMS technique allows comparisons to be drawn between the expected

isotope pattern and that recorded in the mass spectrum.

I (A)

 

 

(3)

Relative Intensity

A AVA‘MA

(C)

 

 

AA

r— fi A ‘v— —-» — __2

5260 5265 5270 5275 5280 5285 5290 5295 5300
m/z

Figure 1.5. (A) UV MALDI-TOF mass spectrum of a protein with a molecular weight
around 5270 Da. (B) Deconvoluted ESI-FTICR mass spectrum of the same protein. (C)
Theoretical isotope distribution of the protein.

Also, it is important to note the increased mass accuracy. In the TOF experiment,
the mass for the peak could be determined within :t 1 mass unit. In the FTMS
experiment, the mass error of each of the peaks was less than 5 ppm different than the

mass of the peaks shown in the theoretical distribution.

19

References

l. Jennings, K.R.; Dolnikowski, G.G. Mass analyzers. In Methods in Enzymology,
McCloskey, J .A., Ed.; Academic Press, Inc.: San Diego, CA, 1990; p. 55.

2. Weickhardt, C.; Moritz, F.; Grotemeyer, J. Time-of—flight mass spectrometry:
state-of-the-art in chemical analysis and molecular science. Mass Spectrom.
Reviews, 1996, 15, 139-162.

3. PerSeptive Biosystems, Inc. Voyager Biospectrometry Workstation with Delayed
Extractionm Technology: User’s Guide. Framingham, MA: PE Biosystems,
1999.

4. Spengler, B. Post-source decay analysis in matrix-assisted laser

desorption/ionization mass spectrometry of biomolecules. J. Mass Spectrom.
1997, 32, 1019-1036.

5. Taylor, P. K.; Amster, I. J. Space charge effects on mass accuracy for multiply
charged ions in ESI-FTICR. Int. J. Mass Spectrom. 2003, 222, 351-361.

6. Easterling, M. L.; Mize, T. H.; Amster, I. J. Routine ppm mass accuracy for high
mass ions: space-charge effects in MALDI FT-ICR. Anal. Chem. 1999, 71 , 624-
632.

7. Amster, I.J. Fourier transform mass spectrometry. J. Mass Spectrom. 1996, 31,
1325-1337.

8. Dienes, T.; Pastor, S.J.; Schfirch, 8.; Scott, J .R.; Yao, J .; Cui, S.; Wilkins, C.L.

Fourier transform mass spectrometry-advancing years (1992-Mid. 1996). Mass
Spectrom. Reviews, 1996, 15, 163-211.

20

CHAPTER 2. STABILIZATION OF DOUBLE-STRANDED DNA IN MALDI MS

1. Structure of DNA

In order to understand the behavior of an analyte in the MALDI experiment, the
structure of the analyte must be taken into account. Deoxyribonucleic acid (DNA) is a
biopolyrner consisting of subunits called nucleotides, Figure 2.1 (A). A nucleotide is
composed of a nitrogen containing base: adenine (A), thymine (T), cytosine (C), or
guanine (G), Figure 2.1 (B). These nitrogen bases are covalently bound to a deoxyribose
sugar that has a phosphate group bound to the 5' carbon. Nucleotides are joined together
by a phosphodiester linkage from the 5'-hydroxyl group of one nucleotide attached to the
3'-hydroxyl group of a neighboring nucleotide. This is demonstrated in the schematic Of
DNA structure shown in Figure 2.1 (A). Several nucleotides bound together are referred
to as an oligonucleotide. The uniqueness of oligonucleotide strands resides in the
sequence of the bases. A

There are many interactions present in double-stranded DNA (dsDNA), but a
major focus in simple discussions is the base-specific pairing via hydrogen bonding.
Opposing strands are held together by two hydrogen bonds between an adenine-thyrnine
base pair and three hydrogen bonds between a cytosine-guanine base pair. When adenine
binds to thymine and cytosine binds to guanine, it is referred to as Watson-Crick base
pairing. To form double-stranded DNA, or duplex DNA, two complementary
oligonucleotide strands are non-covalently bound together by Watson-Crick base pairing.

When the two strands are bound together to form the duplex, the strands bind antiparallel

21

to each other. In other words, the 5' end of one strand will bind to the 3' end of the other

strand, Figure 2.1 (A).

Sugar Phosphate

/ Backbone \

/N NH
NH N//

Guanine

NH2

N \N
/

<NH| N/J

Adenine

 

 

Nucleotide:

3' end (A) 5' end

NH2

I\N

NHO

Cytosine

O

I NH
NH’KO

Thymine

H3C

(B)

Figure 2.1. (A) Double-stranded DNA. The oligonucleotide strands are composed of
nucleotide repeating units. The 3' end of one strand bind to the 5' end of its complement.
(B) Structure of the nitrogen bases. There are 3 hydrogen bonds formed in G-C base
pairs and two hydrogen bonds formed in A-T base pairs. S represents the sugar,
deoxyribose, and P represents the phosphate group bound to the 5' end of the nucleoside.

This figure was adapted from reference 1.

When two oligonucleotides are bound together, the two chains twist together to

form a right-handed double-helix. In this helical structure, the hydrophobic nitrogen

22

bases from each chain are stacked in pairs. These base pairs are in close proximity to
each other and are situated perpendicular to the long axis of the double helix. The
hydrophilic backbone, consisting of alternating sugars and negatively-charged phosphate
groups, runs along the outside of the helix. This backbone is exposed to that water that
surrounds the DNA. The double helix is held together entirely by non-covalent forces.
First, there is hydrogen bonding that holds together the complementary bases of Opposing
strands. Then, there are the hydrophobic interactions. These non-covalent interactions
are what cause the bases to tucked inside the double helix, shielded from water with the
highly polar sugar-phosphate backbone exposed, Figure 2.2. The stability of the helical

structure is largely due to the hydrophobic interactions.

Sugar Phosphate Backbone

Base Pairs

Nitrogen Base

 

Figure 2.2. Double-helical structure of DNA. The hydrophobic bases are stacked within
the double-helix while the sugar-phosphate backbone is exposed. This figure was
adapted from reference 1.

In addition to the base pairing and hydrophobic interactions that occurs in DNA,

there are many other interactions that take place. In cells, DNA is negatively charged due

23

to the ionization of hydroxyl groups in the phosphodiester backbone. The phosphate
groups are affiliated with counter ions such as sodium, calcium, magnesium, and
protonated amines [2,3]. These various cations are responsible for stabilizing the DNA
by reducing the repulsive forces between the phosphate backbones of the complementary
strands. The cations also contribute to the compact tertiarystructure of the DNA. An X-
ray crystal structure of the duplex, d(CGCGAATTCGCG)2, is shown in Figure 2.3. In
the crystal structure, there are several examples of species known to affiliate with DNA
including magnesium, Spermine, and water molecules. Even during the crystal-growth

process, the water remains bound the oligonucleotide.

 

Figure 2.3. X-ray crystal structure of a d(CGCGAATTCGCG)2. This figure was adapted
from reference 2.

24

For an oligonucleotide consisting of 12 bases, the 11 phosphate groups can
accumulate a charge as great as —l 1. The neutral oligonucleotide with firlly protonated
phosphate groups, (0"ll + 11 HI), is referred to as the fiee acid form, or M.
Oligonucleotide ions are detected in a protonated, (M + H)+, or deprotonated, (M - H)‘,
form. Although both the protonated and deprotonated species will be studied, here we
focus in this discussion on the negative ions, Since they are generated in higher
abundance and exhibit greater resolution in MALDI MS than the positive ion forms [4].

Oligonucleotides and DNA have many interesting and unique properties. The UV
spectroscopic properties of oligonucleotides are particularly relevant to this work.
Oligonucleotides exhibit strong absorption maxima around 259 nm. This absorption is
due to the electronic transitions in the purine and pyrimidine bases. The molar extinction
coefficient of an oligonucleotide is dependent on the state of the base-pairing
interactions. When a double-stranded oligonucleotide is formed, there is a decrease in
the molar extinction coefficient, a hypochromic effect. When heated, a Solution of
double-stranded DNA should yield a 10% increase in absorbance due to this effect [5].
When the UV absorption of a nucleic acid sample is measured as a function of
temperature, the resulting plot is known as a melting curve, Figure 2.4. At low
temperatures, the oligonucleotide solution has a low absorbance. As the temperature
increase, the double helix begins to denature. As the duplex denatures, the absorbance of
the solution increases until duplex completely dissociates. The midpoint in the increase
of absorption versus temperature is known as the melting temperature, Tm. This point is

marked in Figure 2.4.

25

     
  
  
  

 

 

Transition
range
.-...- Completely
unwound
Unwindlng of double helix
“30'9““ double helix
2 m
3 (transition
§ temperature)
I I I
70 80 90 100
Temperature (‘C)

Figure 2.4. Melting temperature plot of a double-stranded oligonucleotide. Absorbance
is plotted as a function of temperature. This figure was taken from reference 6.

The melting temperature is important because it is an indication of how stable a
duplex will be at room temperature. Much research has been done on the melting
temperature of oligonucleotides [7-10]. Generally, Tm increases with the length of the
strands, increases with the salt concentration of the solution, and is dependent on the
sequence of the bases and the types of counter-ions present [1 1]. The DNA structure is
known to be sensitive to changes in ionic strength [12]. Melting point studies carry
important information on duplex stability that should be considered in the analysis of the
solution phase aspects of the experiment.

The melting temperature studies shown in this work will be displayed as a plot of
the absorbance of the oligonucleotide solution as a function of wavelength. An example

of a melting temperature study is shown in Figure 2.5. The UV spectrum in black is the

26

spectrum of a duplex oligonucleotide at room temperature and the spectrum shown in
grey represents the spectrum of the same duplex after heating the solution to 80°C. The
absorbance at 259 nm increases by 10 percent when the solution is heated. This means
the duplex is intact at room temperature and dissociates when heated. As the solution is
cooled, the duplex reforms and the absorbance decreases. By performing UV analysis,

the dissociation and formation of the double helix in solution can be monitored.

l"'-25Ci‘l
I—BOC

I
I
I
I I‘ ”'T“ _ T
I
I
I I_______

Absorbance

 

I”— ”r T ““1 v "'1 r ‘ _ “F" , “Am“—

200 250 300 350 400
Wavelength (nm)

Figure 2.5. UV melting temperature study of a duplex oligonucleotide in solution.

For the majority of the work shown here, two duplex Oligonucleotides were used.
Several important aspects of the experiment were considered when selecting
Oligonucleotides to analyze using MALDI MS. First, the mass of the Oligonucleotide is

important. As the molecular weight increases, the obtainable resolution decreases for a

27

time-of—flight experiment. In order to study Oligonucleotides bound to a small drug or a
metal such as platinum, the resolution must be sufficiently high. For this reason, double-
stranded species with molecular weights lower than 10,000 g/mol are desired. The size is
also important regarding the stability of the duplex. Short sequences can have melting
temperatures below room temperature, but duplexes of complementary strands, each with
at least 10 bases, have melting points greater than 30° C [13]. It is also important to
understand the behavior of the single-stranded species in the MALDI experiment. Are
both strands detected in the experiment and how does this correlate to the appearance of
the double-stranded ion? In order to study differences in ionization between the two
strands, there must be a Significant mass difference between them.

The duplex oligonucleotides used in these experiments are shown in Table 2.1.
The table lists the three duplexes used in this work. The duplex sequences are written in
the 5’ to 3’ direction for the top sequence and the 3’ to 5’ direction for the bottom
sequence. This demonstrates the base pairing that takes place in the duplex. The average
molecular weights for the single-stranded Oligonucleotides are written in the column next
to the sequence. The abbreviation M1 will be used to denote the strand of lower
molecular weight and M2 will be used to denote the strand of higher molecular weight.
The column labeled Duplex MW lists the molecular weight information for the duplex
with specific base-pairing (Watson-Crick base pairing) listed as MiMz as well as the
molecular weight information for the non-specific duplexes, MiMi and MzMz. The last
column provides the melting temperature for the duplex. This provides insight into the

stability of the duplexes at room temperature.

28

Table 2.1. Duplexes Used in This Work

 

 

Name Sequence Oligonucleotide Duplex Tm (°C)
MW(avg) in MW (avg) in
g/mol g/mol
Duplex 1 CCGGMTTGGCC 3646 (M1) 7292 (MIMI) 4O
GGCCTTAACCGGTT 4254 (M2) 7900 (M.M2)
8508 (M2M2)
Duplex 2 TTTTTGGTTTTT 3638 (M1) 7276 (MiMi) 28
meow 3648 (M2) 7286 (Mer)
7296 (M2M2)
Duplex 3 ACCCACCCACCC 3480 (M1) 6960 (MIMI) 42
TGGGTGGGTGGG 3813 (M2) 7293 (M.M2)
7626 (M2M2)

 

Duplex 1 will be most Often used in the duplex DNA experiments. This duplex
has a melting point of 40 °C. The single strands of duplex l differ by 608 g/mol due to
the two thymine extension on the one strand. They also contain the CG and AT binding
motifs on the single strand. These motifs are necessary for the binding Of many drugs.
Lastly, a desirable oligonucleotide would not be self-complementary in order to prove
that Watson-Crick pairing has been preserved. For example, suppose two strands, M1
and M2, are complementary. These strands are annealed together to form a duplex.
When analyzed using MALDI, there are many possible ions that may be Observed
including (MlMi-H)‘, (MiM2-H)', and (M2M2-H)'. If all three species are detected in a
1:2:1 ratio of intensities, the complexes are produced at random and we would assume
that the Watson-Crick base pairing has not been maintained. If the strand is self-
complementary, the spectrum provides no information on non-specific complexation. In
the case of duplex l, the specific and non-specific complexes would have mass

differences of 608 Da making the complexes easy to distinguish.

29

While the Dickerson dodecamer, d(CGCGAATTCGCG), does not meet all of the
criteria, it will also be used in many experiments. This oligonucleotide has been the
subject of many studies and much is known about its interactions with salts and
polyamines [2]. This nucleic acid also contains the adjacent CG and AT rich region
needed to bind drugs. There is much data on the binding of the Dickerson dodecamer to
chemotherapeutic drugs such as Hoechst 33258 [14]. For this reason, it will be a useful

tool for analysis.

30

References

10.

ll.

12.

Leja, Darryl. N.D. Jan. 6, 2003. DNA-A More Detailed Description
http://www.accessexcellence.org/AB/GG/dna2.html.

Shui, X.; McFail-Isom, L.; Hu, G.G.; Williams, L.D. The B-DNA dodecamer at
high resolution reveals a spine of water on sodium. Biochemistry. 1998, 3 7,
8341-8355.

Halle, B.; Denisov, V.P. Water and monovalent ions in the minor groove Of B-
DNA oligonucleotides as seen by NMR. Biopolymers. 1998, 48, 210-233.

Nordhoff, E.; Cramer, R.; Karas, M.; Hillenkamp, F.; Kirpekar, F.; Kristiansen,
K.; Roepstorff, P. Ion stability of nucleic acids in infrared matrix-assisted laser
desorption/ionization mass spectrometry. Nucleic Acids Res. 1993, 21, 3347-
3357.

Norman, D. Techniques applied to nucleic acids. In Nucleic Acids in Chemistry
and Biology (eds. Blackburn, G.M.; Gait, M.J.), Oxford University Press, 1996,
445-499.

Campbell, M.K. Biochemistry, 3rd ed.; Harcourt Brace and Company: Orlando,
FL, 1999, Text Figure 07.13.

Marky, L.A.; Breslauer, K.J. Calculating thermodynamic data for transitions of
any molecularity from equilibrium melting curves. Biopolymers, 1987, 26, 1601-
1620.

Blake, R.D. Cooperative lengths of DNA during melting. Biopolymers, 1987, 26,
1063-1074.

Gotoh, O.; Tagashira, Y. Stabilities of nearest-neighbor doublets in double-
helical DNA determined by fitting calculated melting profiles to Observed
profiles. Biopolymers, 1981, 20, 1033-1042.

Korolev, N.I.; Vlasov, A.P.; Kuznetsov, I.A. Thermal denaturation of Na- and Li-
DNA in salt-free solutions. Biopolymers, 1994, 34, 1275-1290.

Schildkraut, C.; Lifson, S. Dependence of the melting temperature of DNA on
salt concentration. Biopolymers. 1965, 3, 195-208.

McConnell, K.J.; Beveridge, D.L. DNA structure: what's in charge? J. Mol.
Biol. 2000, 304, 803-820.

31

13.

14.

Tong,B.Y.; Leung, M.L.C. Correlation between percentage guanine-cytosine
content and melting temperature of deoxyribonucleic acid. Biopolymers. 1977,
16,1223-1231.

Quintana, J .R.; Lipanov, A.A.; Dickerson, R.E. Low-temperature

crystallographic analyses of the binding of Hoechst 3325 8 to the double-helical
DNA dodecamer CGCGAATTCGCG. Biochemistry. 1991, 30, 10294-10306.

32

II. Oligonucleotides and Mass Spectrometry

Matrix-assisted laser desorption/ionization mass spectrometry (MALDI MS),
developed by Karas et al. [ l ], has been extensively used to analyze biological compounds
at the sub-picomole level. To date, most of the published MALDI MS work has involved
the analysis of peptides and proteins, but some work has been done using
Oligonucleotides [2,3]. The development of MALDI matrices Optimized for the analysis
of Oligonucleotides has lagged behind that for peptides [4], but is currently receiving
considerable attention.

Oligonucleotides and DNA behave very differently than peptides and proteins in
the MALDI experiment. Differences of note include the sensitivity, resolution, extent of
prompt fragmentation, and the need for cleanup prior to analysis. For all classes of
biomolecules studied by MALDI, as sample handling methods and TOF instruments have
been developed, typical sample sizes used for analysis have decreased. Peptide analysis
at the subpicomole level was rapidly achieved, while it has taken longer to realize similar
MALDI detection limits for Oligonucleotides. There have been significant advances from
earlier reports of MALDI analyses of oligonucleotides, in which 10-100 pmol of material
were required, to recent reports that Show oligonucleotide detection at the femtomole
level [5,6].

When a pure peptide is analyzed by MALDI-TOF MS, it usually yields a single
mass spectrometric peak representing the intact molecule in either a protonated or a
deprotonated form. Matrices and sample conditions have been reported to increase the
extent of prompt fragmentation of peptides in MALDI [7], but generally there is little

predictable control over the process. In contrast, Oligonucleotides frequently fragment

33

promptly and extensively in linear time-of-flight mass spectrometry [8]. While
fragmentation is useful for structure determination, a decrease in sensitivity can also be
seen.

In terms Of the spectra that result, salts have a very different impact when
analyzing peptides compared to Oligonucleotides. The analysis of pure samples certainly
makes mass spectral interpretation easier for all biomolecules. Depending on whether
compounds are isolated from biological sources, from electrophoresis experiments (gels
or membranes), or are synthesized, peptides and Oligonucleotides may be presented for
MS analysis in the presence of compounds such as salts, buffers, and glycerol, frequently
in higher relative amounts than the analytes. In all cases, high amounts of salts can
sufficiently change the MALDI target material to lower analyte signals. The presence of
compounds, such as glycerol, can lead to problems during the crystal growth step of the
MALDI experiment is to grow crystals. Salts in the analyte or matrix solutions can also
be very detrimental. If NaCl is present, two peaks may appear representing a single
peptide in positive ion mode, most frequently representing the [M+H]+ and [M+Na]'r
ions, separated by 22 Da. As the salt contribution increases, sodium adducts become
more intense. The presence of salts has a much more dramatic influence on the MALDI
spectra of oligonucleotides, Figure 2.6. With an ionic phosphodiester backbone,
oligonucleotides can contain a large number of anionic sites that must combine with an
assortment of cationic species such as H+, Na+ and K+ to desorb in a singly-charged form.
Combinations of cations to provide a charge balance can lead to a substantial number of
peaks representing the intact molecule. Whenever ions for a single species are generated

with a range of m/z values, detection limits are lowered. Also, it becomes much more

34

difficult to detect multiple components of similar mass. The ideal Situation is to generate
one mass spectral peak per component and not several peaks. In Figure 2.6, there are up
to four sodium ions bound to the phosphate groups of the Oligonucleotide. As a result,

multiple peaks in the spectrum represent the intact Oligonucleotide.

 

 

 

 

 

 

I
I -
I (M-H)
I
‘ I
g i
w I
P I
'9'" I
5 I
S * .+~
E
c‘lr’ : +Na
I +Na
I +Na
I
I
3600 3650 3700 3750 3800
mlz

Figure 2.6. Negative-ion MALDI mass spectrum of the Dickerson dodecamer,
d(CGCGAATTCGCG), using HPA as the matrix. There are four sodium ions bound to
the Oligonucleotide.

In the analysis of oligonucleotides by MALDI TOF MS, the resolution achieved
is Often less than what would be realized for peptides of similar size. Adduct formation
and nucleotide base loss have frequently been cited as causes of the reduced resolution

[9]. For larger Oligonucleotides, this is clearly the case. However, even for smaller

sequences with resolved sodium and potassium adducts, resolution for an Oligonucleotide

35

peak is less than that obtained for a peptide peak of similar mass on the same instrument.
The use of an ion reflector does not dramatically improve the resolution of
oligonucleotides in MALDI-MS [10]. Often, we were unable to detect oligonucleotides
using the reflectron mode of the mass spectrometer. One of the few instances in which
relatively small oligonucleotides were analyzed by MALDI and spectra containing
isotopic resolution were obtained was recently published, using a MALDI TOF
instrument with an extended flight tube [1 l].

The differences in MALDI analysis of oligonucleotides and peptides are evident
when examining Figure 2.7. In Figure 2.7 (A), an oligonucleotide with a molecular
weight around 3600 g/mol has been analyzed in negative-ion mode using ATT as the
matrix. In Figure 2.7 (B), a peptide of similar mass to the oligonucleotide has been
analyzed in positive-ion mode using the same matrix. There are some notable
differences. First, when looking at the peaks representing the molecular ion in each
spectrum, the peptide peak has much higher resolution than that of the Oligonucleotide
peak. In Figure 2.7 (B), there is only one peak detected that represents the intact peptide
ion. No fragmentation has occurred and no alkali ions are seen bound to the peptide.
When examining figure 2.7 (A), there is extensive fragmentation. Peaks representing
fragment ions from the oligonucleotide are more intense than the peak representing the
intact, deprotonated Oligonucleotide. Also, there are two resolved peaks that have higher
m/z values than that of the oligonucleotide. The first peak represents a sodium ion
binding to a phosphate and the second peak represents a potassium ion binding to a

phosphate. Since there are three species detected that all represent the intact

36

Oligonucleotide, there is a decrease in sensitivity of the analyte. The sensitivity is also

decreased by the fragmentation of the analyte.

 

 

 

 

 

 

 

 

I
i (A)
(NI-H)-
'3 II
(D
C
.93
C +
; (mm
.2
E ,
0 I
m I
I
I
I
L- 1 LL_ ,2, 2, ___,__g M_
P.“ J! T_ _ '—'7 7'— _' —_—___ _—T—— ._. _'_ 7—‘1 —_h_~—'j
1 500 2000 2500 3000 3500 4000

mlz

Figure 2.7. MALDI mass spectrum of (A) an oligonucleotide taken in negative-ion mode
and (B) a peptide taken in positive-ion mode. AT'T was the matrix for both experiments.

While peptides and oligonucleotides have many differences in terms of their
MALDI MS analysis, they also have some notable similarities in terms of the
development of the analytical utility of MS for their analysis. Early in the history of
MALDI, demonstrations of the ability to generate signals from large, intact, ionized
proteins maintained excitement for this new method, allowing it to be further developed.
While positive results have been reported for proteins with molecular weights of several
hundred thousand, the mass spectral peaks were very broad, and it was difficult to extract
what would be considered an accurate mass. The early results showed that, in most

cases, the molecular weight information determined from MALDI MS was more useful

37

than that derived from gel electrophoresis. Today, the real strength of MALDI in protein
analysis does not lie in its ability to generate ions of intact proteins, but in the
subpicomolar sensitivity obtainable for smaller peptides. The analysis of enzymatic and
chemical digestion products is the application contributing greatly to protein analysis.

Scientists have certainly worked to demonstrate that large oligonucleotides can be
characterized by MALDI as well. While a few reports have been published
demonstrating the detection of oligonucleotides containing 50 or greater bases, the
MALDI response is much more sharply a firnction of molecular weight than is Observed
for peptides, making the detection of large oligonucleotides far from routine. Since a
potential application was the use of mass spectrometry to replace gel electrophoresis
when sequencing DNA using the Sanger method, the requirement of a mass
spectrometric method was different than when sequencing peptides. Peaks with good
resolution must be detectable not only at low rn/z values, but over a very wide mass
range. The DNA sequencing approach is based on a ladder sequencing method. While
mass spectrometry techniques may not, in the near future, replace gels for DNA
sequencing, other powerful applications are still emerging, notably the analysis of single
nucleotide polymorphisms (SNPS) [12,13]. In this area, mixtures of small
oligonucleotides, usually with molecular weights less than 6,000 Da, are generated which
are indicative of errors in genetic code. This is an application to which MALDI MS is
well suited.

The need continues to develop methods for improved analysis of oligonucleotides
of all sizes. The SNP application is clearly driving method development for analyses of

oligonucleotides of sizes for which the challenges are not the creation Of new

38

instruments, but improved matrix chemistry. In order to successfully analyze SNPs using
mass spectrometry, several goals must be achieved. The first of these is lower detection
limits. Often, when analyzing SNPS, there is a limited amount of sample available. The
ability to perform analyses with less or no cleanup is also an important. If contaminants
in a sample interfere with the analysis, the impurities must be removed. The purification
steps needed to remove such contaminants can lead to sample loss. Experimental
conditions in which fragmentation is decreased or eliminated must be defined as well. In
the analysis of mixtures, if one could rely on generating only one peak per component, as
is common for peptides, then the analysis of oligonucleotide mixtures would be much
easier. Lastly, MALDI MS is being automated for the analysis of multiple samples. If
automation is to be successful, a homogeneous MALDI target must be grown in the
initial step. If the MALDI crystals are not homogeneous, the automated system may not
find an ideal sample spot and the data from the experiments would be unreliable.

Many matrices have been developed for peptide analysis [14,15] by MALDI such
as nicotinic acid, ferulic acid, Sinapinic acid, and 2,5-dihydroxybenzoic acid. Since,
generally, these are not as useful for Oligonucleotide analysis, alternate matrices have
been identified [16,17]. Several matrices were found to be compatible for the analysis of
oligonucleotides using UV lasers, including 3-hydroxypicolinic acid (HPA) [18] and 6-
aza-2-thiothymine (ATT) [19]. Since the development of these matrices, HPA has
become the standard matrix for the analysis of larger nucleic acids while ATT is used for
sequences containing less than 25 bases [2]. In addition to UV-MALDI, work has been
done using infrared (IR) lasers [20, 21]. Successful matrices for IR-MALDI include

succinic acid and urea [22].

39

Since matrix selection alone does not overcome some Of the problems associated
with the MALDI MS analysis of oligonucleotides, a number of matrix additives have
been developed. The utility of ammonium salts has been demonstrated to reduce the
formation of alkali ion adducts. Ammonium acetate was the first to be used in the
MALDI experiment [8]. Since then, other ammonium salts have appeared in the
literature with the most successful being diammonium hydrogen citrate [23, 24]. Thus, if
an Oligonucleotide exists in solution in polyanionic form (0‘ ") containing an 11 number of
charges, it forms a singly-charged anion by adding combinations of Hi, Na+, and K+ ions.
If N114+ ions are present, these compete effectively with alkali ions in complexing with
negatively charged phosphates. During the desorption/ionization process, ammonia is
lost, leaving protons behind. In this paper, the fully protonated form of an
oligonucleotide, [0"n + n H+], or the free acid form, will be referred to as M. If no alkali
ions are involved and all of the phosphates are neutralized by protons, negative ion
MALDI yields deprotonated molecules, designated as [M-H]". In our lab, the role of
polyamines as matrix additives has been explored and sperrnine was found to improve
MALDI spectra for single-stranded oligonucleotides, eliminating both the need for
ammonium citrate as well as desalting [5, 25, 26].

In order to study non-covalent complexes in MALDI, DNA was selected as a
model for all types of non-covalent complexes. Oligonucleotides are easy to synthesize.
Specific sequences can be designed and synthesized in order to explore the effects of
composition and molecular weight of the analyte. DNA is also of biological interest to

further understand the role of DNA binding molecules such as chemotherapy agents.

40

Double-stranded DNA is also a useful model because the interaction of the analyte with

other molecules can be monitored by other methods such as UV spectrophotometry.

41

References

10.

Karas, M.; Bachmann, D.; Bahr, U.; Hillenkamp, F. Matrix-assisted ultraviolet-
laser desorption of nonvolatile compounds. Int. J. Mass Spectrom. Ion Processes.
1987, 78, 53-68.

Nordhoff, E.; Kirpekar, F.; Roepstorff, P. Mass spectrometry Of nucleic acids.
Mass Spectrom. Rev. 1996, 15, 67-138.

Limbach, P.A. Indirect mass spectrometric methods for characterizing and
sequencing oligonucleotides. Mass Spectom. Rev. 1996, 15, 297-336.

Zhang,L.-K.;Gross,M.L. Matrix-assisted laser desorption/ionization mass
spectrometry methods for oligodeoxynucleotides: improvements in matrix,
detection limits, quantification and sequencing. J. Amer. Soc. Mass Spectrom.
2000, 11, 854-865.

Asara, J .; Allison, J. Enhanced detection of oligonucleotides in UV MALDI MS
using the tetraamine Spermine as a matrix additive. Anal. Chem. 1999, 71 , 2866-
2870.

Roskey, M.T.; Juhasz, P.; Smimov, I.P.; Takach, E.J.; Martin, S.A.; Haff, L.A.
DNA sequencing by delayed extraction-matrix-assisted laser
desorption/ionization time of flight mass spectrometry. Proc. Natl. Acad. Sci.
1996, 93, 4724-4729.

Lavine, G.; Allison, J. Evaluation of bumetanide as a matrix for prompt
fragmentation matrix-assisted laser desorption/ionization and demonstration of
prompt fragmentation/post-source decay matrix-assisted laser
desorption/ionization mass spectrometry. J. Mass Spectrom. 1999, 34, 741-748.

Currie, G.; Yates, J. 111. Analysis of oligodeoxynucleotides by negative-ion
matrix-assisted laser desorption mass spectrometry. J. Am. Soc. Mass Spectrom.
1993, 4, 955-963.

Nordhoff, E.; Cramer, R.; Karas, M.; Hillenkamp, F .; Kirpekar, F.; Kristiansen,
K.; Roepstorff, P. Ion stability of nucleic-acids in infrared matrix-assisted laser

desorption/ionization mass spectrometry. Nucleic Acids Res. 1993, 21 , 3347-
3357.

Juhasz, P.; Roskey, M.T.; Smimov, I.P.; Haff, L.A.; Vestal, M.L.; Martin, S.A.
Applications of delayed extraction matrix-assisted laser desorption ionization

time-Of-flight mass spectrometry to oligonucleotide analysis. Anal. Chem.
1996, 68, 941-946.

42

ll.

12.

l3.

14.

15.

l6.

17.

18.

19.

20.

21.

Koomen, J .M.; Russell, D.H. Ultraviolet matrix/assisted laser desorption/
ionization mass spectrometric characterization of 2,5-dihydroxybenzoic acid-
induced reductive hydrogenation of oligonucleotides on cytosine residues. .1.
Mass Spectrom. 2000, 35 , 1025-1034.

Griffin, T.J.; Smith, L.M. Genetic identification by‘mass spectrometric analysis
of single-nucleotide polymorphisms: ternary encoding of genotypes. Anal.
Chem. 2000, 72, 3298-3302.

Fei, 2.; Smith, L.M. Analysis of single nucletide polymorphisms by primer
extension and matrix-assisted laser desorption/ionization time-of-flight mass
spectrometry. Rapid Commun. Mass Spectrom. 2000, 14, 950-959.

Bdrnsen, K.O.; Sch'ar, M.; Widmer, H.M. Matrix-assisted laser desorption
ionization mass spectrometry and its applications in chemistry. Chimica. 1990,
44, 412-416.

Hillenkamp, F .; Karas, M.; Ingendoh,A.; Stahl, B. Matrix assisted UV-laser
desorption/ionization: a new approach to mass spectrometry of large biomolcules.
In Biological Mass Spectrometry, Burlingame, A.; McCloskey, J .A., Eds.;
Elsevier Scientific: Amsterdam, 1990; p.49

Parr, G.R. Fitzgerald, M.C.; Smith, L.M. Matrix-assisted laser
desorption/ionization mass spectrometry of synthetic oligodeoxyribonucleotides.
Rapid Comm. Mass Spectrom. 1992, 6, 369-372.

Spengler,B.; Pan, Y.; Cotter, R.J.; Kan, L.S. Molecular weight determination of
underivatized oligodeoxyribonucleotides by positive-ion matrix-assisted
ultraviolet laser-desorption mass spectrometry. Rapid Commun. Mass Spectrom.
1990, 4, 99-102.

Wu, K.J.; Steding, A.; Becker, C.H. Matrix-assisted laser desorption time-of-
flight mass spectrometry of oligonucleotides using 3-hydroxypicolinic acid as an

ultraviolet-sensitive matrix. Rapid Comm. Mass Spectrom. 1993, 7, 142-146.

Lecchi, P.; Le, H.M.T.; Pannell, L.K. 6-aza-2-thiothymine: a matrix for MALDI
spectra of oligonucleotides. Nucleic Acids Res. 1995, 23, 1276-1277.

Kirpekar, F .; Berkenkamp, S.; Hillenkamp, F. Detection of double-stranded DNA
by IR- and UV-MALDI mass spectrometry. Anal. Chem. 1999, 71, 2334-2339.

Berkenkamp, S.; Kirpekar, F .; Hillenkamp, F. Infrared MALDI mass
spectrometry of large nucleic acids. Science. 1998, 281, 260-262.

43

22.

23.

24.

25.

26.

Nordhoff, E.; Ingendoh, A.; Cramer, R.; Overberg, A.; Stahl, B.; Karas, M.;
Hillenkamp, F.; Crain, P.F. Matrix-assisted laser desorption/ionization mass
spectrometry of nucleic acids with wavelengths in the ultraviolet and infra-red.
Rapid Commun. Mass Spectrom. 1992, 6, 771-776.

Li, Y. C. 1.; Cheng, S.W.; Chan, T.W.D. Evaluation of ammonium salts as co-
matrices for matrix-assisted laser desorption ionization mass spectrometry Of
Oligonucleotides. Rapid Commun. Mass Spectrom. 1998, 12, 993-998.

Pieles, U.; Ziircher, W.; Schar, M.; Mdser, H.E. Matrix-assisted laser-desorption
ionization time-Of-flight mass-spectrometry — A powerful tool for the mass and
sequence-analysis of natural and modified oligonucleotides. Nucleic Acid Res.

1993,21, 3191-3196.

Asara, J .M.; Hess, J.S.; Lozada, E.; Dunbar, K.R.; Allison, J. Evidence for
binding of dirhodium bis-acetate units to adjacent GG and AA sites on single-
stranded DNA. J. Am. Chem. Soc. 2000, 122, 8-13.

Distler, A.M.; Allison, J. S-Methoxysalicylic acid and Spermine: a new matrix
for the matrix-assisted laser desorption/ionization mass spectrometry analysis of
oligonucleotides. J. Am. Soc. Mass Spectrom. 2001, 12, 456-462.

III. Matrix-assisted laser desorption/ionization

Matrix-assisted laser desorption/ionization mass spectrometry (MALDI MS) has
become an important technique for the analysis of biomoleCules such as peptides and
oligonucleotides. In a typical MALDI experiment, an excess of an organic matrix is
combined with an analyte. The concentration of the matrix solution is typically three
orders of magnitude greater than the concentration of the analyte solution. For example,
one may mix 1 uL of a 1 11M solution of a peptide with 1 [IL Of a near-saturated solution
of a matrix such as or-cyano-4-hydroxycinnamic acid, with a concentration of
approximately 25 mM. If crystals representative of the solution are formed, the relative
molar amounts of matrixzanalyte present will be 25,000: 1. In the UV MALDI process,
the analyte is co-crystallized with a matrix. The matrix is usually a small organic
molecule that absorbs light at the same wavelength as a laser. While visible and infrared
lasers can be used in the MALDI experiment, most often a nitrogen laSer is used. A
nitrogen laser emits pulses of light at a wavelength of 337 nm. In the MALDI process,
the pulses of the laser light irradiate the sample and desorb the analyte from
matrix/analyte crystals to yield gas phase ions, Figure 2.8. The matrix also minimizes the

sample degradation due to the laser irradiation [1].

45

hv,'v,337 nm
[M+H]+

[m+H]+

organic solid

rmM ~10,000:1 ~ 5 mm2

m - matrix
M - analyte

 

 

Matrix:analyte crystals

Figure 2.8. Schematic of the MALDI experiment.

While the mechanism of MALDI is not fully understood, there are several
functions the matrix must serve. First, the matrix must absorb the energy from the laser
light and transfer it into the excitation energy of the solid system. A small volume of the
matrix/analyte crystals undergo a near instantaneous phase transition to a gaseous state
[2]. During this process, the analyte molecules and matrix molecules are desorbed

simultaneously with only limited internal excitation.

The matrix must also serve as a solid state "solvent" for the analyte molecules [3].
A small volume of solution containing a micromolar concentration of the analyte is
deposited with an equal volume of a millimolar matrix solution. The analyte molecules
are incorporated in a large excess of matrix molecules. After the solvent evaporates,
analyte molecules can be isolated from each other in a solid solution of analyte molecules
in the matrix. It is also believed that the formation of the crystals provides an in-situ
cleaning of the sample, leading to the high tolerance against contaminants [2]. While it is

believed the analyte must co-crystallize with the matrix, this is again an area of debate.

46

Experiments have led to the conclusion that protein incorporation into the crystals of the

solid MALDI matrices is not a prerequisite for MALDI [4].

The matrix may also play a role in the ionization of the analyte. The matrix
molecules may be photoexcited or photoionized. These excited matrix molecules on ions
may then transfer a proton to the analyte molecules. Two UV photons provide the
necessary ionization energy by the peeling of excitation energy from two excited
molecules or by the absorption of a second photon from an already-excited molecule [3].
Chemical reactions can lead to the formation of the quasimolecular ion as either [M + H]+

or [M — H]'. The effect of the gas-phase proton affinities of the matrices has been

discussed [5-6].

Lastly, it is believed that the matrix must be easy to sublime [7]. Often matrices
like Sinapinic acid and dihydroxybenzoic acid have enhanced volatility with the
photochemical or pyrolytic cleavage Of functional groups such as carboxylic acids or
hydroxyl groups upon irradiation. The thermodynamic properties of the matrix
compounds are also proposed to be important in the MALDI process [8]. It is proposed
that there is a bottleneck of energy transfer to the analyte molecules that is caused by the
high rate of the sublimation of the excited matrix molecules. With all of the requirements
for a matrix, only a small set of organic molecules has been successful in the MALDI

experiment.

47

Sampfile Preparation

While there is no universal sample preparation method allowing for the analysis
of many analytes, there is a standard method of sample preparation. The analyte is
dissolved at a concentration of around 1 pmol/uL. The matrix is dissolved at a
concentration of around 25 mM. A variety of solvents and solvent systems have been
used to dissolve the matrix. It is helpful if the solvents used for the matrix are miscible
with the solvent used for the analyte. Most commonly, the analyte is dissolved in water
and the matrix is dissolved in a mixture of acetonitrile and water. The presence of
acetonitrile increases the volatility of the matrix solution and assists in the formation of
the mixed crystals, or crystals containing the analyte surrounded by the matrix. The
matrix is selected based on the type of analyte. Several matrices and their applications

are listed in the Table 2.2.

Two types of additional components may be present in the target that is grown
from solution. One class of components is impurities including compounds such as salts,
buffers, detergents, and/or acids and bases. For many analytes, additional components
have been identified that have a positive influence on the experiment. These are referred
to as additives.

Several additives have been used with MALDI samples to enhance the quality of
the mass spectra. Additives can serve several different purposes. Additives have been
found to increase the homogeneity of the MALDI crystals and increase sample-tO-sample
reproducibility. These effects can be helpful when using automated systems or during
quantitation experiments. Also, additives can decrease or increase the amount of analyte

fragmentation. An increase in fragmentation would allow for more structural information

48

to be determined and a decrease in fragmentation would allow for an increase in the
sensitivity of the analyte. Additives can also decrease the levels of cationization,
increase ion abundance, and increase resolution.

Table 2.2. Matrices and their Applications

 

 

Matrix Application
Sinapinic Acid Peptides, Proteins

(Molecular Weight > 10,000 g/mol)
a-Cyano-4-hydroxycinnamic acid Peptides, Proteins

(Molecular Weight < 10,000 g/mol)
2,5-Dihydroxybenzoic acid Peptides, Proteins, Carbohydrates, Lipids,

Polymers, Small molecules

2-(4-Hydroxy-phenylazo)-benzoic acid Proteins, Polymers, Lipopolysaccharides

3-Hydroxypicolinic acid Oligonucleotides, DNA
Dithranol Nonpolar polymers
2,4,6-Trihydroxyacetophenone Small Oligonucleotides
Trans-3-indoleacrylic acid Nonpolar polymers

Picolinic acid Nucleic acids, Proteins
6-Aza-2-thiothymine Small oligonucleotides, Proteins

 

In MALDI analysis of peptides and proteins, one of the first additives used was
trifluoroacetic acid (TF A). TF A was needed to enhance the solubility of proteins in
water. For analysis of polymers, compounds such as polystyrene are only ionized and
detected in the MALDI experiment if copper or silver ions are present to bind to the
double bonds in the compound [9]. Mixing additives such as 2-hydroxy-5-
methoxybenzoic acid with DHB (10%) have improved sensitivity 2-3 fold. This mixture
also known as "super DHB" is now widely used in MALDI when studying compounds

such as glycoproteins.

49

In oligonucleotide MALDI MS, the first additives developed were ammonium
salts such as ammonium hydrogen citrate [10] and ammonium fluoride [1 1-12]. When
used in oligonucleotide analysis, the molar amount of ammonium salts introduced into
the MALDI target is similar to the amount of matrix. Both the matrix and additive are
typically present in the initial solution at concentrations of approximately 25 mM. The
ammonium ions in the target preparation have a significant impact on the species
desorbed and ionized in the MALDI experiment. It is assumed that the additive alters the
state in which the analyte exists in the target crystals. In addition to the elimination Of
multiple alkali-ion adduction, the ammonium salt seems to play a significant role in
enhancing both the desorption and the ionization of intact oligonucleotides [11]. Since
the discovery Of ammonium salts as additives, several other additives have been
developed for the MALDI analysis of oligonucleotides, notably the tetrammine Spermine
[13-14], related amines [15], and fucose [l6].

Additives have been used in the MALDI experiment with a variety of analytes in
order to improve MALDI analysis. In Table 2.3, several matrix additives are listed along

with the analyte and the effect of the additive on the resulting MALDI spectra.

50

Table 2.3. Matrix Additives

 

 

Additive Sample Type Effect
Ammonium salts Oligonucleotides Decrease Na+ adducts,
improves sensitivity
Ammonium salts Phosphopeptides Aid in detection
Cesium Iodide Polymers Aids in ionization
Copper salts Polymers Aids in ionization
Fucose Oligonucleotides Improves sample spot
homogeneity
Nitrocellulose (substrate) Peptides Decrease Na adducts,
Proteins improves sensitivity
Silver Trifluoroacetate Polymers Aids in ionization
Sodium Iodide Polymers Aids in ionization
Spermine Phosphopeptides Decreases Na adducts,
DNA improves sensitivity
Sucrose Peptides Improves sample spot
Oligonucleotides homogeneity
Trifluoroacetic acid Peptides Increases protonation and
(TFA) Proteins solubiligI of analyte

 

51

MALDI Mechanism

When analyzing non-covalent complexes in MALDI, it is important to consider
the different steps of the experiment. In order to determine, the point in the MALDI
experiment at which double-stranded oligonucleotides dissociate, a model or mechanism
for the experiment must be used. Many aspects of the MALDI mechanism have been
proposed, refuted, and discussed. New models continue to appear. Each of the processes
proposed for investigation here may or may not be an important step in the MALDI
experiment. A number of reviews and summaries have been recently published that
guided the formation Of this model [5, 17-20].

Since M.M2 is not detected as (M1M2-H)' or (M 1M2+H)+ in the MALDI
experiment, it must dissociate at some point before the acceleration and detection of the
ions. There are many steps in the MALDI experiment where the double-stranded species
could denature. Shown in Table 2.4 are the proposed steps in the MALDI experiment
leading to both the preservation of the double-strand as well as the steps leading to
dissociation. Table 2.4 is divided into the three main processes involved in a MALDI
experiment: 1) Liquid PhasezEvaporation, 2) Solid Phase: Excitation, and 3) Solid Phase:
Desorption/Ionization. In order to discuss these steps, the nomenclature recently

introduced by Karas [17] will be used.

52

26.3
$8
8.9
3.8

5%

CV

2. 211N229. fig .1 $2255
N2.22 t ~2.2¢.8 t $2.352
3.211229. fig T $222.52

«2 +2.31% .1 arfk2.2v

$2+.2FH8_ t E e + $2.25;

.693 5.22
$2+.2Es_ t as; + :3 N22

 

«:0. 2.25 29.5 on 9:23.. macaw

69
an
an
39
as

E

3

22:12.3 t 322.255
222255 .I $2.2:L9_
2.912% T $222.5.
2.3: .I $2.358.

2:. t $2.223.
cozmwiEoEozmamom ”omega. Eow

 

$2,221,292 TE e + $2.25.;

 

85.2w “826 2.8

698 5.22
$2253 .I as; + 35. N22
cozmco m>w ”omega E: :

 

 

28. 685.43.23.30 8 5.68.. m 3m

2:65:83 5.22 65 5 83m 3835 .3 29¢

 

53

In the liquid phase/evaporation process, a solution of the double-stranded
oligonucleotide, M1M2, is combined with a saturated matrix (T) solution. If the
experiment proceeds without dissociation of the strands, the MALDI target will contain
intact analyte molecules surrounded by matrix molecules as shown in equation (1).
However, in the solvent evaporation step, the strands may dissociate, equation (1').

If the duplex is not denatured during the liquid phase/evaporation step, matrix
molecules in the crystal surround it. The next part of the MALDI experiment involves
the excitation of the solid phase, equations (2) and (2'). In the excitation process, UV or
IR light irradiates the solid phase and the matrix absorbs energy. The asterisks in the
table denote an energetically excited species. If sufficient energy flows from matrix
molecules to the lattice and into the analyte, the double-stranded species may denature,
equation (2'). '

If the double-stranded species is maintained throughout the first two steps, the
dissociation must occur during the solid phase desorption/ionization process, equations
(3'b-3'e). After excitation, various species are desorbed from the MALDI target as shown
in equation (3a—3e). Included in Table 2.4 is the direct generation of ions from the
excited target, as well as the evolution of gas phase ions through intermediate gas phase
clusters, as has recently been proposed by Karas et al. [17]. Brackets indicate the
condensed phase, braces indicate desorbed gas phase clusters, and parentheses are used to
indicate gas phase ions. The double-stranded analyte may be desorbed in this part of the
experiment, only to dissociate once in the gas phase, equation (3'b). It is also possible

that only ions representing the single strands are desorbed, equation (3'c) and (3'c).

54

Lastly, the double-stranded species may desorb from the solid phase with no charge,

equation (3'd). Only charged species will be detected in the MALDI experiment.

55

Quantifiers

In order to measure the success of an experiment, several criteria will be used.
When analyzing the annealed double-stranded species, M.Mz, the Single strands (Mi-H)
and (M2-H)’ are detected. The strand with the lower molecular weight will be denoted as
M I while the strand with a greater molecular weight will be denoted as M2. It is of
interest to monitor the ratio of the intensity (I) of the (Mi-H)“ peak to the intensity of the
(M2-H)’ peak. This ratio is referred to as the single strand distribution (SSD), equation
(4)-

_ IIMII
" 1(M2)

SSD (4)

To the extent that the single strand ions evolve from the duplex, it is also important to
determine how the SSD value correlates with other variables in the experiment, and to
determine if the SSD correlates with single strand composition. If the single strand ions
are fragments of the singly charged gas phase duplex, then one may eXpect that the
decision of which strand retains the charge would be determined by the base composition
of the individual strands. To determine if, in fact, the SSD is in some way related to the
presence of the duplex, it is important to perform all of the experiments proposed in this
work both on annealed double strands as well as on 1:1 mixtures of the single strands.
The goal of this work is to optimize the ratio of the intensity of the double strand peak
when compared to the intensities of the single strand peaks. This is called the double
strand retention ratio (DSRR), equation (5).

__ [(MIM2)
‘ I<M.1+ 1(M2)

 

(5)

56

While larger DSRR value is desirable, the peaks detected in the duplex region must be
the result of specific pairing and not random dimerization. The goal of this work is to
retain the maximum extent of Watson-Crick base pairing when detecting a double-
stranded species. If the M1M2 duplex dissociates during crystal formation, some fraction
of M1 and M2 may recombine. However, during crystal formation, the single strands may
also group together in non-Watson-Crick pairing. As mentioned previously, M1 and M2
can form three different duplexes: MIMI, M1M2, and M2M2. In order to quantify the
amount of Watson-Crick pairing involved, we will calculate a Watson-Crick Selectivity
Factor (W CSF ), equation (6).

1(MIM2)

WCSF:
[(MIMI)+I(MIM2)+I(M2M2)

 

(6)

If we begin with equimolar amounts of M1 and M2 and the WCSF value Of a given
spectrum is 0.5, this suggests that the distribution of duplex peaks was dictated by
probability and not by the Watson-Crick base pairing. As the value of the WCSF
approaches 1, Watson-Crick pairing is preserved and the non-covalent forces were
maintained throughout the MALDI experiment. As we change the variables in different
experiments, the value of the WCSF will provide information about the nature of the
species detected and give direction to subsequent experiments.

The resolution will also be monitored to aid in spectral analysis. If the resolution
is poor, the peaks representing the duplex in the spectra may be several hundred mass
units wide. If this is the case, a small molecule, such as a drug, could not be detected
bound to the oligonucleotide. Thus, the resolution must be optimized in the experiments
as well. Since the single-stranded Oligonucleotides used in these experiments are Of

similar molecular weight, the resolution for the peaks representing each of the single

57

strands is comparable. For this reason, the resolution will be determined for the M1 peak
in all the experiments. The resolution of the peak representing the double-stranded

oligonucleotide, M1M2, will also be determined.

58

Exploring Non-covalent Complexes in the MALDI Experiment

In the following sections, the MALDI process will be presented in the three parts
denoted in Table 2.4, 1) the Liquid Phase: Evaporation, 2) the Solid Phase: Excitation,
and 3) the Solid Phase: Desorption/Ionization. In each section, the possible problems that
could lead to the dissociation of the non-covalent complex will be presented. The
research performed for this work will be presented by posing a series of questions
regarding the steps of the MALDI experiment. In each section, the question will be
presented and experiments designed to prove or disprove the hypothesis will be
introduced and the results of these experiments will be discussed.

Since the MALDI process has been artificially divided into three portions, it is
still important to note each step is highly dependent on the other steps. For example,
changing a condition in the liquid phase of the MALDI experiment may affect the
outcome of the desorption/ionization step. Similarly, changing a variable to affect the
outcome of the excitation step, may also affect the outcome of the crystal growth step.

Also, when using double-stranded DNA as model for all types of non-covalent
complexes, it is important to note the nature of the Oligonucleotide. With the negatively
charged phosphodiester backbone, Oligonucleotide MALDI spectra exhibit improved
resolution, increased sensitivity, and increased signal-to-noise ratios when analyzed in
negative-ion mode. For this reason, all MALDI spectra shown in the DNA experiments

were performed in negative-ion mode.

59

References

10.

ll.

Bakhtiar, R.; Tse, P.L.S. Biological mass spectrometry: a primer. Mutagenesis.
2000, 15, 415-430. ‘

Bahr, U.; Karas, M.; Hillenkamp, F. Analysis of biopolymers by matrix-assisted
laser desorption/ionization (MALDI) mass spectrometry. Fres. J. Anal. Chem.
1994, 348, 783-791.

Karas, M.; Bahr, U.; Giebmann, U. Matrix-assisted laser desorption ionization
mass spectrometry. Mass Spectrom. Rev. 1991, 10, 335-357.

Homeffer, V.; Dreiseward, K.; Ludemann, H.-C. Hillenkamp, F .; Lage, M.;
Strupat, K. Is the incorporation of analytes into matrix crystals a prerequisite for
matrix-assisted laser desorption/ionization mass spectrometry? A study of five
positional isomers of dihydroxybenzoic acid. Inter. J. Mass Spectrom. 1999,

185/186/18 7, 859-870.

Breuker, K.; Knochenmuss, R.; Zenobi, R. Gas-phase basicities Of deprotonated
matrix-assisted laser desorption/ionization matrix molecules. Inter. J. Mass
Spectrom. 1999, 184, 25-38.

Jorgensen, T.J.D.; Bojesen, G.; Rahbek-Nielsen, HR. The proton affinities Of
seven matrix-assisted laser desorption/ionization matrices correlated with the
formation of multiply charged ions. Eur. Mass Spectrom. 1998, 4, 39-45.

Beavis, R.C.; Chait, B.T. Factors affecting the ultraviolet laser desorption of
proteins. Rapid Commun. Mass Spectrom. 1989, 3, 233-237.

Vertes, A.; Bijbels, R.; Levine, R.D. Homogeneous bottleneck model of matrix
assisted UV laser desorption of large molecules. Rapid Commun. Mass Spectrom.
1990, 4, 228-233.

Mowat, I.A.; Donovan, R.J.; Maier, R.R.J. Enhanced cationization of polymers
using delayed ion extraction with matrix-assisted laser desorption/ionization.
Rapid Commun. Mass Spectrom. 1997, 1 I, 89-98.

Pieles, U.; ZI’Ircher, W.; Schiir, M.; Mdser, H.E. Matrix-assisted laser-desorption
ionization time-of—flight mass-spectrometry — A powerful tool for the mass and
sequence-analysis of natural and modified Oligonucleotides. Nucleic Acid Res.

1993, 21, 3191-3196.
Li, Y. C. 1.; Cheng, S.W.; Chan, T.W.D. Evaluation of ammonium salts as co-

matrices for matrix-assisted laser desorption ionization mass spectrometry of
oligonucleotides. Rapid Commun. Mass Spectrom. 1998, 12, 993-998.

60

12.

13.

14.

15.

16.

17.

18.

19.

20.

Cheng, S-W.; Chan, T-WD. Use of ammonium halides as co-matrices for matrix-
assisted laser desorption/ionization studies of Oligonucleotides. Rapid Commun.
Mass Spectrom. 1996, 10, 907-910.

Asara, J.; Allison, J. Enhanced detection of oligonucleOtides in UV MALDI MS
using the tetraamine Spermine as a matrix additive. Anal. Chem. 1999, 71, 2866-
2870.

Distler, A.M.; Allison, J. 5-Methoxysalicylic acid and Spermine: a new matrix
for the matrix-assisted laser desorption/ionization mass Spectrometry analysis of
Oligonucleotides. J. Am. Soc. Mass Spectrom. 2001, 12, 456-462.

Chou, C-W.; Williams, P.; Limbach, P.A. Matrix influence on the formation of
positively charged oligonucleotides in matrix-assisted laser desorption/ionization
mass spectrometry. Int. J. Mass Spectrom. 1999, 193, 15-27.

Distler, A.M.; Allison, J. Improved MALDI-MS analysis of oligonucleotides
through the use of fucose as a matrix additive. Anal. Chem. 2001, 73, 5000-
5003.

Karas, M.; Gliickmann, M.; Schafer, J. Ionization in matrix-assisted laser
desorption/ionization: singly charged molecular ions are the lucky survivors. J.
Mass Spectrom. 2000, 35, 1-12.

Zenobi, R.; Knochenmuss, R. Ion formation in MALDI mass spectrometry.
Mass Spectrom. Rev. 1998, I 7, 337-366.

Knochenmuss, R.; Stortelder, A.; Breuker, K.; Zenobi, R. Secondary ion-
molecule reactions in matrix-assisted laser desorption/ionization. J. Mass
Spectrom. 2000, 35, 1237-1245.

Liao, P.C.; Allison, J. Ionization processes in matrix-assisted laser-desorption
ionization mass-spectrometry — matrix-dependent formation of [M+H]+ vs
[M+Na]+ ions of small peptides and some mechanistic comments. J. Mass
Spectrom. 1995, 30, 408-423.

61

IV. The Liquid Phase: Evaporation

The creation of the target is an important part of a successfirl MALDI experiment.
A micromolar solution of the analyte is combined with an equal volume of a saturated or
near-saturated matrix solution. In order to facilitate the dissolution of the organic matrix,
a mixture of acetonitrile and water is often used as the solvent of choice. The presence of
the acetonitrile in the matrix solution also aids in the rapid formation of crystals due to its
volatility. As the solvents evaporate, the volume decreases, the concentrations rise, and
the solubility limits of the components are reached [1]. When the precipitate begins to
form with rapid evaporation, the matrix crystals grow under kinetic control [2] and can
trap the analyte molecules within, equation (1) and (1') in Table 2.4.

By the time the crystals are formed, are there any double-stranded Oligonucleotide
molecules remaining? During the formation of the MALDI target, there are many points
at which a double-stranded Oligonucleotide could dissociate. Several variables will be
investigated including pH, choice of matrix, solvent, crystal growth rates,temperature,

and the presence of other solution components.

62

Do matrix molecules in the initial solution induce the fragmentation at the duplex?

Several matrices are commonly used for the analysis of oligonucleotides using
UV light at 337 nm. These include 3-hydroxypicolinic acid (HPA) [3] and 6-aza-2-
thiothymine (ATT) [4]. The structures of these matrices are very similar to those of
species known to stabilize DNA such as Hoechst 3325 8, which binds in the minor groove
of the duplex [5]. However, the solution used to create the MALDI target has 1,000-
100,000 times more matrix molecules than analyte molecules. With such an excess, the
aromatic matrix molecules could interfere with the hydrogen bonding in the DNA by
penetrating between the two strands. If matrix molecules insert between the two strands,
they may competitively hydrogen bond to each strand, forcing the duplex to dissociate,
reaction (1') in Table 2.4.

Typical MALDI spectra of duplex l are shown in Figure 2.9. In Figure 2.9(A), 6-
aza-2-thiothymine (ATT) was used as the matrix. In Figure 2.9(B), 3-hydroxypicolinic
acid (HPA) was used as the matrix. The peaks representing the single-stranded species
(Mi-H)‘ and (M2-H)‘ are the most intense peaks in the spectra. The duplex regions of
the spectra have been magnified and clearly no peaks representing the duplex have been
detected. This means, at some point during the MALDI experiment, all of the duplex has

dissociated.

63

 

 

 

 

 

 

 

(A) ATT
_ (Mz-HI'
(Mi'H) x 10
.é‘ F (M1M2'H)
(D
C
.5 I
E A -—-A——--+ 4:52...“ ~ A - A
.fg’ (B) HPA
2
or
CE
x 10
3000 4000 5000 6000 7000 8000 9000

m/z

Figure 2.9. Negative-ion MALDI mass spectrum of duplex 1 using (A) ATT as the
matrix and (B) HPA as the matrix.

64

Survey of MALDI Mgtrices

If no duplex ions are detected when ATT or HPA is used as the matrix, perhaps
these choices of matrices do not allow for the detection of non-covalent complexes. In
order to investigate the effect of the matrix on the double-stranded DNA, a series of
MALDI experiments were performed. A variety of compounds were tested as matrices in
the MALDI experiment including 3-aminobenzoic acid, 5-aminosalicylic acid, and 3-
hydroxyanthranilic acid, with no success. These matrices are listed in Table 2.5 with the
relative signal intensities recorded for the single-stranded Oligonucleotide. NO peaks
representing the double-stranded species were detected in any of the experiments using
the compounds in Table 2.5 as matrices.

While no compounds listed in Table 2.5 allowed for the detection of double-
stranded DNA, we have discovered a new matrix, 5 -methoxysalicylic acid, for analysis of
single-stranded Oligonucleotides with limited fragmentation, high resolution, and good
signal-to-noise ratios [6]. A reprint of the article is included in Appendix A.

A traditional MALDI matrix has an aromatic ring and can have conjugated, acidic
functional groups such as a carboxylic acid. Many of the compounds tested as matrices
shown in Table 2.5 possess functional groups such as carboxylic acids and hydroxyl
groups. These functional groups on the compounds may be responsible for the
denaturation of the complex. Matrices with more unusual structures were tested as well
including the mercaptobenzothiazoles, Figure 2.10. 2-Mercaptobenzothiazole (MBT)
and 5-chloro-2-mercaptobenzothiazole (CMBT) are effective for analyzing peptides,

glycolipids, oligosaccharides, and synthetic polymers [7]. However,

65

 

CE

Illi-

3112

Table 2.5. Tested Matrices and Results

 

Matrix

Intensity of (Mi-H)“

 

 

Aminobenzamide
5-Aminobenzoic acid
2-Amino-4-chloro benzoic acid
Aminoisoquinoline
4-Aminosalicy1ic acid
3-Aminoquinoline
Anthranilonitrile
or-CyanO-hydroxycinnamic acid
1 ,8-Diamino-4,5-Dihydroxyanthraquinone
1 ,S-Diaminonapthalene
Dihydroxyacetophenone
2,5-Dihydroxybenzoic acid
Diphenylbutadiene
2,3-Diphenyl maleic anhydride
Dithranol

3-Hydroxyanthranilic acid
N-Hydroxynapthalimide
2-(4-Hydroxyphenylazo)benzoic acid
5-Methoxysalicylic acid
Phenazine

Sinapinic acid

Thymidine
Trihydroxyacetophenone

 

*><><><><><><><><><><

*-
*

*xx *X «It-

*-
Ir

><

{-

><

 

X-no signal, *-weak signal intensity, **-strong signal intensity

66

mercaptobenzothiazoles did not allow for the detection of intact double-stranded
oligonucleotides. The spectra of the single-stranded oligonucleotides had much lower
signal-to-noise ratios and poorer resolution when compared to HPA, ATT, and MSA.
The mercaptobenzothiazoles in Figure 2.10 do not possess hydroxyl group or
carboxylic acid groups, but they do possess thiol groups. These thiol groups are capable
of proton donation as well. Oligonucleotides are charged in solution and those charges
may be retained when trapped in the MALDI crystals. When dealing with a charged

analyte, the matrix would not need to participate in the ionization of such species. Also,

 

(>1, (>55,

(A) (3)
Figure 2.10. The structures of (A) 2-Mercaptobenzothiazole and (B) 5-chloro-2-

mercaptobenzothiazole.

when using a matrix that has acidic groups, a saturated solution of that matrix may have a
pH value of less than 3. Such a pH value will denature the DNA duplex as well. For this
reason, simple organic molecules with no functional groups may be ideal for the MALDI
analysis of DNA. In order to test this theory, aromatic molecules such as anthracene with
no acidic groups were tested as matrices for oligonucleotides.

Several problems arise when such organic compounds are used. First, organic
compounds with no functional groups tend to be insoluble in water. Since
oligonucleotides are soluble in water, this poses a problem when forming the MALDI
target. The organic solvents used to dissolve compounds such as anthracene are
immiscible with water. The matrix would have to be spotted and dried before adding the
aqueous solution of the analyte. Anthracene/Oligonucleotide crystals could only be
formed when anthracene was dissolved in ethanol. While ethanol and water are miscible,
the samples were still difficult to spot due to the volatility of the alcohol, making it
difficult to pipette accurately. After formation of the MALDI target with the
anthracene/alcohol solution, only ions of anthracene were detected. No analyte

molecules were detected in the experiment.

67

I.

Similar to anthracene, the compound di(cyclopentadieny1)iron was used. This
compound is commonly referred to as ferrocene. The structure of ferrocene is shown in
Figure 2.11. This compound absorbs light at 337 nm and can successfully be desorbed
and ionized without the presence of a matrix. When used as a matrix, ion from the
ferrocene were detected in the experiment. No peaks were detected that represented the
single or double-stranded oligonucleotides. SO, while the ferrocene molecule is an

efficient absorber at 337 nm, the molecule must not allow for the desorption or ionization

@

€93

Figure 2.11. Structure of ferrocene.

of the analyte.

 

Lastly, another class of organic compounds was tested for their utility in
analyzing oligonucleotides. Dyes are commonly used in a variety of fields including
microscopy, biochemistry, and analytical chemistry. Often, these molecules are used to
visualize materials such as membranes, DNA, organelles, or proteins [8]. These dyes can
be organic molecules that are neutral, cationic, or anionic [9]. Examples of each type of
dye are shown in Figure 2.12. In Figure 2.12, Alizarin Red S is a negatively charged dye
in solution while Chlorophenol red is a neutral dye. Methyl violet is an example of a

cationic dye with a fixed charge on the nitrogen group.

68

 

+NCH 5 CI"

(3)

(C)

Figure 2.12. Structure of (A) Alizarin Red S, (B) Chlorophenol Red, and (C) Methyl
Violet.

Dyes absorb in the ultraviolet/visible region of the spectrum and ionize/desorb
after laser irradiation without the presence of a matrix. Due to their ability to harvest the
light from the laser, dyes may be suitable matrices. The dyes used as matrices in the
MALDI experiment are listed in Table 2.6. In each of these experiments, ions
representing the intact dye molecules were seen in the resulting spectra. NO ions from the

oligonucleotides were detected in any of the experiments.

69

Table 2.6. Organic Dyes Tested as Oligonucleotide Matrices

 

Alizarin Red S Clayton Yellow
Alizarin Yellow G Cresol Purple ,
Bromochlorophenol Blue Cresol Red
Bromocresol Green Metacresol Purple
Bromophenol Blue Methyl Orange
Bromophenol Red Methyl Violet
Chlorophenol Blue Phenol Red
Chlorophenol Red Thymol Blue

 

 

 

I tested a variety of compounds as matrices in the MALDI experiment. These
compounds were structurally diverse. None of these compounds allowed for the
detection of the double-stranded oligonucleotides. While many of the compounds used in
this study had structures similar to that of known matrix molecules, Often these
compounds did not allowed for the detection of single-stranded oligonucleotide ions or
peptide ions. Only very few compounds are successful as matrices in the MALDI

experiment.

70

Meltingj‘emperflrre Studies

Of the over 100 compounds that I have tested as matrices in the MALDI
experiment, only three compounds allowed for the detection of single-stranded
oligonucleotides: HPA, ATT, and MSA. The structures of these three matrices are

shown in Figure 2.13.

O
C H 3
”N I 6-Aza -2-thiothymine O
ATT
S A N /N ( ) I N
H O \
H l
/
O o H H 0
H O 3-Hydroxypicolinic acid
5-Methoxysalicylic acid (HPA)
(MSA)
OCH 3

Figure 2.13. The structures of ATT, MSA, and HPA.

These matrices allowed for the detection of oligonucleotides in the MALDI experiment
and were the most promising choice for analyzing duplex Oligonucleotides. The matrices
ATT and HPA will be of particular interest because their use as oligonucleotide matrices
has been highly documented. Since these matrices did not allow for the detection of
duplex DNA, could these compounds be dissociating the duplex in solution? In order to
explore the effect of the matrix molecule in solution with DNA, melting temperature

studies were performed.

71

The UV spectrophotometer used in this study did not allow for analysis of
saturated matrix solutions due to the high molar absorptivity of the matrix compounds
from 250-400 nm. The first matrix compound studied was ATT. Several solutions of
ATT were created in order to find the optimal concentration for the UV study. For the
purpose of this experiment, a concentration of 50 micromolar yielded good results. A UV
spectrum of ATT was recorded at room temperature. The ATT solution was then heated
to 80°C and another UV spectrum was recorded. There was no change in the absorbance
spectrum. This proved the molar absorptivity of ATT does not depend on the
temperature of the solution, so the presence of ATT would not interfere with the melting
temperature study.

A solution of ATT was added to a solution of the Dickerson dodecamer duplex so
the final concentration of the ATT was 50 micromolar and the final concentration of the
duplex was one micromolar. The absorbance at 259 nm was monitored for 30 minutes
and no increase in absorbance was detected. At a micromolar concentration, the addition
of the ATT molecules did not denature the double stranded species even when the duplex
oligonucleotide was in solution with the ATT for a time exceeding that of the typical
MALDI experiment.

The oligonucleotide solution was then stored for one week at room temperature
and another melting temperature study was performed, Figure 2.14. For this study, a UV
spectrum of the incubated solution was acquired. This solution was heated and a 10%
increase in absorbance was detected. These data Show the oligonucleotide duplex had not
been denatured after 1 week in a 50 micromolar solution of matrix. Another study was

performed using duplex 1 as the analyte. Similar results were found. ATT does not

72

cause the dissociation of the duplex even after a one week incubation period. While the
concentration of the ATT solution was lower than that for a typical MALDI experiment,

the matrix molecules were in contact with the duplex for an extended period of time.

 

 

 

 

§ {:25 c"
l“ ‘—80 c
.9 I_#_.
O
(D
.0
<
200 250 300 350 400

Wavelength (nm)

Figure 2.14. UV analysis of duplex oligonucleotide incubated with ATT for 1 week at
room temperature.
Is this experiment really kinetically relevant? The following reaction (7) occurs in
solution at a specific rate:

M1M2+T—5MI+M2+T (7)
Suppose the reaction proceeds under simple kinetics, equation (8).

"dIMIMzi

d1 = k[M1M2][T] (8)

Here the concentration of the matrix, [T], is less than 1/100th of the concentration used in

a typical MALDI experiment. However, the length of time, or dt, is over 1000 times

73

greater than a typical MALDI experiment. SO, the same effect should be seen in the
melting temperature study when compared to the MALDI experiment.

Melting temperature studies were also performed using HPA and MSA with
similar results. SO, it was concluded that the matrix molecules in solution with the

duplex DNA do not appear to cause the dissociation of the non-covalent complex.

74

Could the use at organic solvents disrupt the non-covalent (areas in double-stranded
DNA ?

While organic solvents are typically used in a MALDI experiment to dissolve the
organic matrix, the common crystallization process must Often be re-evaluated. For
example, many kinds of samples have been analyzed by MALDI using the
water/acetonitrile/trifluoroacetic acid (TFA) solvent system and is a recommended
solvent system for variety of peptides and proteins [10]. Originally this solvent system
was designed specifically for the analysis of peptides, although it has now been applied
for MALDI analysis in general. Since analysis of oligonucleotides is enhanced when
using negative-ion mode and the oligonucleotides are readily soluble in water, TFA is not
needed in these experiments to enhance protonation or solubility of the analyte. For this
reason, TF A will not be used in any of the experiments discussed here.

While TFA is not needed for oligonucleotide analysis, the use of acetonitrile is
still necessary to facilitate the dissolution of the matrix. The presence of organic co-
solvents, such as acetonitrile, may induce the dissociation of the double-stranded
oligonucleotides. It is possible for an organic solvent to induce separation of the duplex
Since several common organic solvents are known to denature DNA including alcohols
such as methanol, ethanol, and n-propanol, and amides such as forrnamide, N,N-
dimethylformamide, and propionamide [11].

In order to determine if acetonitrile denatures DNA duplexes with fewer than 15
bases per strand, a series of melting temperature studies were performed. A one
micromolar solution of duplex 1 was formed using a 1:1 acetonitrilezwater solution. The

absorbance of the solution was scanned from 200-400 nm. The solution was then heated

75

and another UV spectrum was acquired. There was a ten percent increase in absorbance
at 259 run when compared to that for the sample at room temperature, Figure 2.15. This
indicates that the duplex remains intact when in the presence of acetonitrile. It was
concluded that the acetonitrile in the starting solution does not denature the duplex. This
experiment was performed at several concentrations of acetonitrile. Even with a 3:1
mixture of acetonitrilezwater, the duplex remained intact. Since acetonitrile evaporates

from the MALDI target at a higher rate than the water, the MALDI target will never

contain greater than 50% acetonitrile.

 

Absorbance

  

 

I
L__E__ _fi# _

215 235 255 275 295 315 335
Wavelength (nm)

Figure 2.15. UV melting study using the acetonitrile/water solvent system.

It is also important to consider the timescale of the two experiments. In the
melting temperature studies, the duplex remained in contact with the acetonitrile between

15-60 minutes. In a typical MALDI experiment, the duplex is in contact with the

76

acetonitrile for less than 2 minutes. Since the duplex is in contact with the acetonitrile for
a relatively short period of time during the MALDI experiment, there are fewer
opportunities for dissociation of the duplex in the MALDI experiment when compared to
that of a melting temperature experiment.

While acetonitrile does not dissociate the duplex in the UV experiment, the
organic co-solvent is still an important variable in the MALDI experiment. A solvent is
used not only to dissolve the matrix, but to assist in the rapid formation of the crystals as
well. If the solvent affects the crystal growth, using solvents with varying volatility could
alter the result of experiments. I have tested many solvent combinations in the MALDI
experiment using ATT and HPA as the matrices. Solvent combinations used include only
water, water/acetonitrile combinations, water/methanol, acetone, water/alcohols, and
water/fluorinated alcohols. Methanol and ethanol used alone were difficult to Spot on the
MALDI target due to sample spreading. Combinations of water with these alcohols
proved unsuccessful as well. When alcohols were used in solution preparation, no peaks
representing the duplex or any dimers were detected, giving a DSRR value of zero.

Other organic solvents and solvent systems were examined for dissolution of the matrix
including pyridine, pyridine/water, and acetonitrile. Again, Spectra from these
experiments contain no duplex or dimer peaks.

While the presence of acetonitrile does not disrupt the non-covalent forces in
double-stranded DNA, we have seen interesting results when only water is used to
dissolve the matrix. In this experiment, a saturated aqueous solution of HPA was used.
The HPA was mixed with the double stranded analyte and spotted on the sample plate.

After spotting, the solution was dried with a heat gun to facilitate crystal formation.

77

Heating was used in an attempt to offset the fact that, without acetonitrile present, it
would take much longer for the 1 uL droplet to evaporate. A microliter of a 1:1
ACN/H2O solution evaporates in approximately 30 seconds, while 1 pL of H20
evaporates in roughly 5 minutes. While heating can eliminate time as a variable for the
comparison, it introduces temperature as a new variable. The effect of temperature is
discussed further in another section of this chapter. The results of the water only
experiment are shown in Figure 2.16. Compared to Figure 2.9, the peaks representing the
single-stranded species are still the most intense peaks in the spectrum, with a SSD of

1.0; the peak for the smaller strand has increased in relative intensity.

 

 

 

 

 

(MVHI' lle-H)'
e
(I)
C
9
E
0
>
1:
E
C)
II
(M1M2'H1-
‘M‘Ml "H"! (MzMz-H)‘
3000 4000 5000 6000 7000 8000 9000

mlz

Figure 2.16. Negative-ion MALDI mass spectrum of duplex 1 using HPA as the matrix.
Only water was used to dissolve the matrix.

In the spectrum, relatively intense peaks were observed in the duplex region.
Using this method, the value of the DSRR increases from 0 to 0.1. From one standpoint,

conditions have been found that allow for duplex molecules to be detected. However, the

78

WCSF is 0.5 , suggesting that the initial duplex has not been retained. One interpretation
is that the duplex completely dissociates during the crystal growth step, and when water
alone is used as the solvent, single strands are more likely to dimerize during crystal
growth.

Similar results are seen when aqueous solutions are used without any heating.
That is, the appearance of the dimers is not related to heating or length of time needed for
crystal growth, but rather to the use of only water in the preparation of the MALDI target.
It is interesting to note that higher laser powers were needed when the crystal growth
time was increased. As the length of time needed for crystal growth increases, the
analyte may not be trapped as efficiently in the matrix crystals. If the analyte is not fully
incorporated into the matrix, more laser power would be needed to generate ions.

In order to explore the effects of different solvents, several experiments were
performed using a variety of solvents. With HPA as the matrix, water was the only
solvent that led to changes in the values of the spectral quantifiers. When these
experiments were repeated using ATT as the matrix, this effect was not seen. The DSRR
value for ATT in water alone was zero and the intensity of the peaks representing the

single-stranded species were of low intensity.

79

Does the 2H ot the matrixmnalge solution cause the dissociation at the non-covalent

complex?

Often in MALDI MS of peptides and proteins, matrices such as Sinapinic acid and
a-cyanO-4-hydroxycinnamic acid are used. When these matrices are dissolved in an
acetonitrile/water solution, the pH of the matrix solution can be lower than 3. These
highly acidic conditions can denature non-covalent complexes. Previously, it has been
shown that using buffered solutions, near physiological pH, can lead to enhanced
detection of non-covalent complexes when using acidic matrices [12-13]. Even for basic
matrices with a solution pH of 6-7, the detection of non-covalent complexes is enhanced
by using buffers [14-15].

Could the addition of a biological buffer maintain the pH throughout the MALDI
experiment? In order to explore the effect of pH on the DSRR values, a series of MALDI
experiments was performed. Previously, buffers using compounds such as ammonium
citrate [14-15] and ammonium bicarbonate [12] had been used to stabilize protein-protein
and duplex DNA complexes in MALDI MS. These buffers were used first in this work.
Ammonium citrate and ammonium bicarbonate solutions were made at concentrations of
10 mM, 25 mM, 50 mM, and 100 mM. Matrix solutions of HPA and ATT were made by
dissolving the matrix in a 1:1 acetonitrilezbuffer solution. While SSD values changed
slightly in the experiments, there was no increase in the DSRR. NO duplex peaks were
detected in any of the experiments.

Other buffer solutions were used in the MALDI experiment. These buffer
solutions included N-2-hydroxyethylpiperazine-N'-2-ethanesulfonic acid (HEPES),

piperazine-N,N'-bis(ethanesulfonic acid) (PIPES), tris-(hydroxymethyl)aminomethane

80

(TRIS), imidazole, tricine, and several phosphate buffers. These buffers, when used in
the MALDI experiment, did not allow for the detection of the duplex oligonucleotides.
Also, the HEPES, PIPES, TRIS, and imidazole buffers suppressed the signal for the
single-stranded oligonucleotides as well. The phosphate buffers were formed by
dissolving phosphate salts with sodium and potassium as the counter ions. While the
phosphate buffers used did not suppress the signal for the single-stranded
Oligonucleotide, the presence of the sodium and potassium ions caused a decrease in

resolution, sensitivity, and an increase in the alkali ion adduction seen in the spectra.

81

Does the [en gh ot time associated with crystal gowth aflect the dissociation at the

duplex?

As mentioned previously, the matrix and/or solvents may play a role in the

dissociation of the duplex, shown below, equation 10.

M1M2 + nX —> MIXm + M2Xp (X = solvent or matrix) (10)
This reaction, if it occurs, does so at some rate. While it has been demonstrated that the
matrices and solvents do not denature the duplex in solution, the duplex could still be
dissociating during the crystal growth step. Since the matrix solution is saturated or
nearly saturated, the matrix will be the first component in solution to precipitate. As the
matrix precipitates, the duplex may interact with the growing crystal surface and
dissociate. If the crystal growth process is slow, the duplex will have more opportunities
to interact with the growing crystal surface. By decreasing the time needed for crystal
growth, the double-stranded molecules may be trapped in the MALDI target before
denaturation occurs. A variety of variables can affect the length of time needed for
crystal growth including the solvents, temperature, volume of solution deposited, and
atmospheric conditions as well.

Solvent selection determines the rate of crystal growth with the addition of
volatile organic solvents decreasing the time necessary for formation of crystals
compared to that for water alone. The use of volatile solvents may also facilitate the
incorporation of the double-stranded oligonucleotides into the matrix. If the solvents
quickly evaporate, the analyte molecules may become trapped among the matrix

molecules more quickly than they are separated, by some mechanism, in solution. In

82

addition to using volatile solvents, the sample plate may be heated in order to increase the
rate of crystal growth, or cooled to decrease evaporation rates.

In order to increase the rate of precipitation, volatile solvents were used including
long chain alcohols. Long chain alcohols serve as an excellent model because while
there is structural similarity, the evaporation rates (vapor pressures) decrease as alkyl
chain length increases. For these experiments, four alcohols were selected, methanol,
ethanol, propane], and butanol. These solvents were used alone or in a 1:1 solution with
water to dissolve the matrix. In each of these experiments, no change in the DSRR was
seen.

In order to further explore the effect of crystal growth time, the plate was heated
and cooled during sample preparation to adjust evaporation times. While the heating of
the sample plate had little effect on the spectral quantifiers, cooling the sample plate
seemed to have a great effect. When a cold plate was used, the time of sample spot
evaporation increased from 1-2 minutes to 5-10 minutes depending on the solvent system
used. When these sample spots were analyzed, it was difficult to detect even the single-
stranded oligonucleotides. Most likely, given a longer period of time, the MALDI
crystals containing analyte embedded in the matrix were not formed. These mixed
MALDI crystals are necessary for a successful MALDI experiment. Instead, the
Oligonucleotides likely crystallized separately from the matrix (thermodynamic limit) and
could not be desorbed and ionized.

The amount of sample deposited on the sample plate was also varied. Instead of
depositing 1 1.1L of a solution on a MALDI plate and waiting for it to evaporate, the target

was constructed in several steps. For example, a fifth of a microliter of matrix was

83

deposited with an equal volume of matrix. Since the evaporation time is proportional to
volume, reducing the volume allows the crystals to grow more quickly without
introducing any new variables. A variety of experiments were performed with differing
volume deposited in differing orders. Despite the changes, the experiments yielded the
same DSRR as the traditional one-microliter approach.

The atmospheric conditions were also altered. While MALDI crystals are
normally grown under atmospheric pressure, the solutions could be added to the MALDI
target and then placed in a vacuum. Again, this allows for all the solution variables to
remain constant. Similarly, the MALDI solutions were deposited on the target and dried
under a stream of nitrogen. This increased the rate of evaporation. When the crystals
were grown under vacuum or a stream of nitrogen gas, no changes in the spectral
quantifiers were seen, even though droplet evaporation time changed substantially.

Since no duplex ions were detected in this experiment, it was concluded that the length of

time associated with the crystal growth does not affect the dissociation Of the duplex.

84

Does contact with a surtace denature the double-stranded oligonucleotides?

During crystal growth formation, the duplex oligonucleotides are in contact with
the matrix molecules, solvents, and also the sample plate. We have discussed the
interaction of the DNA with the matrix molecules, the solvents, and the growing matrix
crystals. The non-covalent complex also comes in contact with the sample plate during
the course of a MALDI experiment. Often in MALDI MS, sample plates can be stainless
steel, gold, or Teflon-coated. The surface may be completely flat or have depressions
known as wells. These sample wells can vary in size. A mass spectrometrist would not
consider the sample plate to be a variable although there are many types of sample plates
available.

Biological materials such as proteins and oligonucleotides are known to interact
with a variety of surfaces including membranes [16], mica [17], and gold [18]. Enzymes
are often complexes that consist of multiple protein subunits held together by non-
covalent forces. When an enzyme comes in contact with a metal surface, a decrease in
the activity of the enzyme is often seen. A decrease in the activity is often caused by the
denaturation of the complex. Similarly, when the duplex is spotted onto the sample plate,
the interaction with the metal surface may cause dissociation. In order to explore this
theory, another melting temperature study was performed. A solution of duplex 1 was
placed in a cuvette. A scan was taken to form the baseline for the experiment. Then, a
solid gold washer was placed in the cuvette to simulate the surface of gold plate. Scans
were taken immediately after the addition of the gold and then every ten minutes for an
hour. No increase in absorbance was seen at 259 nm, Figure 2.17. The cuvette

containing the gold and duplex 1 was sealed and stored at room temperature for 24 hours.

85

Another spectrum was obtained. The solution was then heated to 80°C and another UV
spectrum was acquired. A 10% increase in the absorbance of the solution was detected at
259 nm. When the gold is in contact with duplex l for 24 hours, the duplex still remains
intact.

An identical experiment was performed using a stainless steel washer. Similar
results were seen. It was then concluded that contact with the metal surface of the

MALDI target does not cause the dissociation of double-stranded DNA in solution.

 

 

Absorbance
/

 

200 250 300 350 400
Wavelength (nm)

Figure 2.17. Four overlapped UV spectra of double-stranded DNA (1) before contact
with the gold surface, (2) immediately after the addition of the gold surface, (3) 10
minutes after the addition of the gold surface, and (4) 30 minutes after addition of the

gold surface.

86

To further explore the effect of the surface of the sample plate, a Nafion® coating
was placed on the stainless steel plate. A small amount of Nafion® was spread on a
traditional MALDI sample plate and dried with a heat gun. The structure of the Nafion®

is shown below in Figure 2.18.

[(—CF2 CI:2 }7CF CI:2 —]
x I y 0

ll
(oer, <|:F TOCF, CF, -|SI—OH
CF3 z 0

Figure 2.18. Structure of Nafion®.

The layer of Nafion® on the sample plate did not improve the DSRR and WCSF values
for the experiments. However, the Nafion® surface is hydrophobic and allowed for
larger sample volumes to be deposited on the gold or stainless steel MALDI target. This
was helpful when dealing with a dilute solution of analyte. Teflon coated plates were
also used to analyze double-stranded DNA, but again no increase in the DSRR value was
seen.

In order to explore a different type Of desorption/ionization experiment, a DIOS
plate was purchased fi'om Mass Consortium (San Diego, CA). DIOS is the
desorption/ionization on silicon and has been recently used to desorb/ionize peptides and
proteins without the use of a matrix [19-21]. The porous silicon surface of the DIOS
plate has a strong ultraviolet absorption that enhances the laser desorption/ionization

process [19]. In this technique, the analyte is deposited on the surface and can be

87

desorbed/ionized without a matrix. A sample Of duplex 1 was deposited on the DIOS
sample plate. When analyzing duplex 1 using DIOS, no peaks representing the single or
double-stranded oligonucleotides were detected in either negative or positive ion mode.
Since no oligonucleotide peaks were detected, the DIOS technique may not allow for the
desorption of negatively-charged species such as an oligonucleotide. However, this
surface could still be used to perform a MALDI experiment. A solution of duplex l was
combined with either a HPA or ATT matrix solution. These solutions were then
deposited on the DIOS plate. Again, no peaks representing the intact duplex were
detected. In this experiment, duplex is never exposed to any metal surface.

When examining the data from the melting temperature studies, combined with
the data from the MALDI experiments, it does not appear to be the surface of the MALDI

target that causes the duplex dissociation.

88

Conclusions

For the variables in the liquid phase portion of the experiment, the solution to the
problem of duplex dissociation could involve multiple changes to the experiment. For
example, the most success may come from using water/methanol solution to dissolve the
matrix, ATT as the matrix, and depositing the matrix/analyte solution onto a heated plate
to reduce the crystal growth time. A variety of combinations of the different variables
were used in the MALDI experiments. No increases in the DSRR values were seen with
any of the experiments.

While the variables in the liquid phase have been examined, the complex could
still dissociate during this portion Of the experiment. Dissociation during the liquid phase
would explain the non-specific dimerization seen in Figure 2.16. It is possible the
complex could dissociate when the duplex contacts the growing matrix crystals. Since a
reduction in crystal growth time has no effect on the outcome of the experiment, another
approach must be taken to stabilize the duplex. The variables that will be discussed in the
solid phase portions of the MALDI experiment may also affect the liquid phase portion of

the experiment.

89

References

l.

10.

11.

Asara, J .M. Enhancing spectra in desorption/ioniziation mass spectrometry
through the use of chemical additives. Michigan State University: East Lansing,
MI, 1999.

Nishioka, K., Thermodynamic vs. kinetic critical nucleus and the reversible work
of nucleus formation. In Advances in the understanding of crystal growth
mechanisms, Nishinaga, T.; Nishioka, K.; Harada, J .; Sasaki, A.; Takei, H., Eds.;
Elsevier Science, St. Louis, MO, 1997; p. 3.

Wu, K.J.; Steding, A.; Becker, C.H. Matrix-assisted laser desorption time-of-
flight mass spectrometry of oligonucleotides using 3-hydroxypicolinic acid as an
ultraviolet-sensitive matrix. Rapid Comm. Mass Spectrom. 1993, 7, 142-146.

Lecchi, P.; Le, H.M.T.; Pannell, L.K. 6-aza-2-thiothymine: a matrix for MALDI
spectra of oligonucleotides. Nucleic Acids Res. 1995, 23, 1276-1277.

Quintana, J .R.; Lipanov, A.A.; Dickerson, R.E. Low-temperature
crystallographic analyses of the binding of Hoechst 3325 8 to the double-helical
DNA dodecamer CGCGAATTCGCG. Biochem. 1991, 30, 10294-10306.

Distler, A.M.; Allison, J. 5-Methoxysalicylic acid and Spermine: a new matrix
for the matrix-assisted laser desorption/ionization mass spectrometry analysis of
oligonucleotides. J. Am. Soc. Mass Spectrom. 2001, 12, 456-462.

Xu, N.X.; Huang, Z.H.; Watson, J .T.; Gage, D.A. Mercaptobenzothiazoles: a
new class of matrices for laser desorption ionization mass spectrometry. J. Am.
Soc. Mass Spectrom. 1997, 8, 116-124.

Kieman, J .A. Classification and naming of dyes, stains, and fluorochromes.
Biotech. Histochem. 2001, 5/6, 261-277.

Green, F .J . The Sigma-Aldrich Handbook of Stains, Dyes, and Indicators.
Milwaukee, WI: Aldrich Chemical Co, 1990.

PerSeptive Biosystems, Inc. Voyager Biospectrometry Workstation with Delayed
Extractionm Technology: User's Guide. Framingham, MA: PE Biosystems,
1999.

Levine, L.; Gordon, J.; Jencks, W.P. The relationship of structure to the

effectiveness of denaturing agents for deoxyribonucleic acid. Biochemistry. 1963,
2, 168-173.

90

12.

13.

14.

15.

16.

17.

18.

19.

20.

2].

Woods, A.S.; Buchsbaum, J .C.; Worrall, T.A.; Berg, J .M.; Cotter, R.J. Matrix-
assisted laser desorption/ionization of noncovalently bound complexes. Anal.
Chem. 1995, 67, 4462-4465.

Woods, A.S.; Huestis, MA. A study of peptide-peptide interaction by matrix-
assisted laser desorption/ionization. J. Am. Soc. MaSs Spectrom. 2001, 12, 88-96.

Lecchi, P.; Panell, L.K. The detection of intact double-stranded DNA by
MALDI. J. Am. Soc. Mass Spectrom. 1995, 6, 972-975.

Lin, 8.; Cotter, R.J.; Woods, A.S. Detection of non-covalent interaction of single
and double stranded DNA with peptides by MALDI-TOF. Proteins Suppl. 1998,
2, 12-21.

Vestling, M.M.; Fenselau, C. Poly(vinylidene difluoride) membranes as the
interface between laser desorption mass spectrometry, gel electrophoresis, and in
situ proteolysis. Anal. Chem. 1994, 66, 471-477.

Fang, Y.; Hoh, J .H. Early intermediates in sperrnidine-induced DNA
condensation on the surface of mica. J. Am. Chem. Soc. 1998, 120, 8903-8909.

Gearheart, L.A.; Ploehn, H.J.; Murphy, C.J. Oligonucleotides adsorption to gold
nanoparticles: a surface-enhanced Raman spectroscopy study of intrinsically bent
DNA. J. Phys. Chem. B. 2001, 105, 12609-12615.

Wei, J.; Buriak, J .; Siuzdak, G. Desorption/ionization mass spectrometry on
porous silicon. Nature 1999, 399, 243-246.

Shen, Z.; Thomas, J .J .; Averbuj, C.; Broo, K.M.; Engelhard, M.; Crowell, J.E.;
Finn, M.G.; Siuzdak, G. Porous silicon as a versatile platform for laser
desorption/ionization mass mpectrometry. Anal. Chem. 2001, 33, 179-187.

Thomas, J .J .; Shen, Z.; Crowell, J .E.; Finn, M.G.; Siuzdak, G.

Desorptin/ionization on silicon (DIOS): a diverse mass spectrometry platform for
protein characterization. Proc. Nat. Acad. Sci. USA. 2001, 98, 4932-4937.

91

V. Solid Phase: Excitation

The next portion of the MALDI experiment involves the excitation of the
crystalline MALDI target. After the solvent evaporates, a crystal of matrix molecules is
formed with analyte molecules trapped within the matrix network, Figure 2.19 (A). The
solid target is then irradiated with either IR or UV light. In the case of UV MALDI, the
matrix molecules absorb the light to form electronic excited states, equation (2) and (2')
from Table 2.4. Intersystem crossings yield highly vibrationally excited matrix
molecules. Vibrational energy is transferred into the host lattice, and the vibrational
temperature of some portion of the crystal rises to the point where a phase transition
occurs, leading to desorption. Vertes has proposed a bottleneck model, in which energy
from the host lattice, or matrix, moves only very slowly into the analyte molecules, due to
the vibrational energy frequency mismatch between weak hydrogen bonds of the lattice
and strong covalent bonds in the analyte [1-3].

While models are still being developed to describe energy flow and content
following excitation, it is clear that the host absorbs the energy, and it is transferred to the
analyte thereafter, as time allows. In an ideal situation, the energy distributed in the
lattice would be large enough to dissociate the non-covalent bonds holding the lattice
together, but not large enough to dissociate the bonds holding the complex together,
Figure 2.19 (B). If large amounts of energy are available, and energy flow is rapid
relative to desorption rates, then energy could move into a DNA duplex that exceeds the
bond dissociation energy. In this case, the analyte could dissociate into the two single
strands, Figure 2.19 (C). It is important to note that the Vertes bottleneck model was

developed at a point in time when scientists were trying to understand why, in a MALDI-

92

linear TOF MS experiment, peptides would yield a single peak with no fragmentation.
This could be rationalized by considering that energy flow into the analyte in a very short
time-frame could be minimal. It has more recently been found that some energy does
flow since analyte ions do fragment. The situation could be very different when the
analyte is an oligonucleotide duplex. Because the duplex is held together by hydrogen
bonds, energy does not need to flow from the lattice into the covalent bonds of the
analyte to induce fragmentation. The energy must only flow into the hydrogen bonds that
hold the strands together. This may occur much more efficiently. Since a lattice of non-
covalently bound matrix molecules is broken in the MALDI process, it is likely that the

non-covalent bonds between the duplex oligonucleotides could be broken as well.

1—1

1ft”?

 

 

 

lie

:1 E1 E] D (A) I: [:1 C] D

E] :1 CI :1 I: [I 1: 1:1

. e—n . . E n .

1:1 1:1 I: CI CI [:1 1:1 El

1:1 [II D E] :1 1:1 (:1 :1
(B) (C)

Figure 2.19. (A) The crystal lattice is composed of analyte (cylinders) trapped within the
network formed between the matrix molecules (squares). Non-covalent bonds are
represented by the lines connecting the boxes and cylinders. (B) The distribution of
energy into the lattice can break the non-covalent interactions that hold the matrix
molecules together in the lattice, (C) if the flow of energy into the lattice breaks the non-
covalent bonds between the matrix molecules, it could also break the non-covalent bonds
that hold the complexes together.

93

Also, it is important to note that, while peptides do not fragment in linear
MALDI—TOF MS, oligonucleotides do. This is undoubtedly due to a combination of
effects. Low energy fragmentation pathways may be available to Oligonucleotides that
are not options for peptides. However, clearly all fiagrnentation Of analytes requires

energy transfer into these analytes following irradiation.

94

I too much ene is bein de osited into the anal e molecule ollowin irradiation
how cgn the amount at enemy depositeflnto the ppphrte be decre_a_se_d?_

The energy deposited during the excitation process allows the analyte and matrix
to desorb from the target. When the energy is transferred into the lattice, the energy
breaks the non-covalent bonds that form the lattice. If the energy left after the excitation
process is sufficiently large, the strands may dissociate, Table 2.4 equation (2'). In order
to maintain the duplex, the amount of energy transferred to the lattice can be decreased.
This could be achieved in several ways.

First, the laser irradiation can be decreased. The influence of laser power is
complex. The available energy increases, at least over some energy range, as more
power is provided. As the crystals absorb the energy during irradiation, two scenarios
may result. In the first example, more energy flows into the analyte, yielding fewer
duplex ions. In contrast, the desorption of the analyte ions could occur more quickly,
with less time for energy to flow into the analyte, yielding more abundant duplex ions.

A detailed study of how the proposed quantifiers vary with laser power was
performed. The MALDI mass spectrometer used in this study was equipped with a N2
laser, operated at constant power. The sample irradiation was controlled by an attenuator.
This attenuator was assigned arbitrary values that ranged from 0, for full attenuation, to
4600 for unattenuated laser irradiation. These experiments were started at the threshold
energy, or the lowest laser power that yielded ions. The laser power was increased in
increments of 50 arbitrary units. After each increase, the SSD, DSRR, and WCSF values
were calculated and recorded. At high laser powers, the ions representing the single-

strands became saturated. NO laser power allowed for an increase in the DSRR.

95

In addition to the laser power, the choice of matrix will affect the flow of energy
into the lattice. Since the molar absorptivity of the matrix as well as the heat of
sublimation and vaporization must be considered in the MALDI experiment, a variety of
matrices were used in the laser power study. Matrices such as a—cyano-hydroxycinnarnic
acid and Sinapinic acid are considered to be “hot” matrices, or matrices that tend to
induce fragmentation of the analyte [4,5]. Other matrices such as HPA and ATT are
considered to be “cool” matrices, or matrices that tend to reduce the amount of
fragmentation of the analyte [4,5]. One proposed explanation for the hot and cold nature
of matrices involves the temperature at which they sublime, and hence, desorb the analyte
[6]. These four matrices were selected for the laser power study.

Another option is to decrease the energy absorption of the target crystal. A
typical crystal lattice is shown in Figure 2.20 (A). The non-covalent complex is entirely

surrounded by matrix molecules. Each matrix molecule has the potential to absorb the

 

 

 

 

 

 

 

 

 

 

 

 

 

 

(A) (B)

Figure 2.20. (A) A crystal lattice containing the matrix molecules and analyte. The
analyte is entirely surround by absorbing compounds. (B) A crystal lattice containing the
analyte surrounded by both the matrix (white boxes) and a non-absorbing component
(grey boxes).

96

energy from the laser and transfer this energy into the analyte, dissociating the complex.
If the energy flowing into the crystal lattice exceeds that of the bond dissociation energy
of the duplex, the duplex would dissociate. What if the lattice was composed of a non-
absorbing compound as well, Figure 2.20 (B)? In Figure 2.20 (B), the matrix molecules
are shown in the white boxes and the non-absorbing component is depicted by the gray
boxes. The analyte is now surrounded by components that do not absorb all energy from
the laser. This could decrease the amount of energy transferred to the lattice, and
eventually the analyte, in the MALDI excitation process.

In order to explore the effect of matrix diluents, compounds were needed that
would not absorb light from the laser. One such compound was the monosaccharide

fucose. The structure of fucose, also known as deoxygalactose, is shown in Figure 2.21.

CH3

    

HO OH

 

HO "OH
Figure 2.21. The structure Of fucose.

F ucose has been used in previous experiments and has been shown to increase resolution
and sample homogeneity in the analysis of peptides [7-8]. Since fucose does not absorb
light at 337 run, it decreases the amount of energy absorbed by the crystal lattice at a
given spot on the target. The amount of fucose used in the experiments was varied in
order to find the optimal concentration. Fucose was found to be most effective at

concentration similar to that of the matrix and has been found to improve the quality of

97

the spectra of oligonucleotides, as it does for peptides [9]. A number of similar sugars
have been investigated. Compared to fructose, sucrose, glucose, galactose, and maltose,
the influence of fucose is, in fact, unique. The role of fucose in the MALDI experiment
has been fully investigated and the results published [9]. A reprint can be found in
Appendix B. .

In order to explore the role of fucose in the MALDI experiment, several
experiments were performed. Several duplexes were used as the analyte with HPA or
ATT as the matrix. The concentration of fucose was also varied. The effect of fucose is
most apparent when used at a concentration of 25 mM, a concentration similar to that of
the matrix. When fucose is used as an additive, the analyte is first spotted on the plate.
Then, equal amounts of fucose and matrix are added to the MALDI target and allowed to
dry. Figure 2.22 shows when fircose is used with duplex l and ATT, the DSRR value
increases from 0.05 to 0.15. However, a WCSF value of 0.6 suggests that the Watson-
Crick base pairing is not preserved in this MALDI experiment. This experiment suggests
that the duplex may be dissociating in the liquid phase and fucose facilitates the random
dimerization process.

When the matrix, analyte, and fucose are in solution together, it is not known if
the resulting crystals retain this bulk composition. Due to the differing solubilities of the
matrix, fucose, and the analyte, it is unknown if the crystal composition mirrors that of
the solution composition. As the matrix/analyte/fucose solution used to make the
MALDI target begins to evaporate, the first solubility limit encountered is that of the
matrix, so matrix crystals begin to grow. At some point, the analyte and fucose must be

incorporated into the matrix crystals because ions of both fucose and the analyte are

98

detected in the MALDI experiment. At some later point in the evaporation process, the

solubility limit of fucose is encountered, and the composition of the crystals may change.

A) ATT + sp + fucose

 

 

(MtMZ'Hi—

{I

1
e (M.MI-HI' I I
:C; a I I! (MzMz'H)-
:6 I \ I In III
a) A I A
.?. ~r~“‘.u" AM" ‘3. i “J .v
E WW fl \~/\\ \W‘Ammw MA w ~mNH-r WF “MM“ WNW
0
I:

B) ATT + sp
, ”WM/w iikwlxgk _,_ A
7000 7500 8000 8500 9000 9500 1 0000

m/z

Figure 2.22. A portion of the negative ion MALDI mass spectrum of duplex 1 with A)
ATT/sp/fucose and B) ATT/Sp as the matrix. While the WCSF is 0.5 for both
experiments, the DSRR is 0.05 without the fucose and 0.15 with the addition of fucose to

the MALDI target.

While there are certainly a number of possible explanations for the mechanism
through which fucose acts, the effect is real. If crystals are grown from a matrix/analyte

solution containing fucose, the absorber has been diluted, and this should allow for some

intervention in terms of energy deposition and redistribution.

99

Can additives stabilize the double-stranded DNA during the excitation process?

Many compounds exist that will bind to the nitrogen bases or to the phosphate
groups in an oligonucleotide, either alone or in a duplex. Nucleic acids interact
reversibly with a variety. of chemical species including water, metal ions such as Na+ and
Mg”, and small organic (usually cationic) molecules. The interaction of the DNA with
such species can be classified as electrostatic, groove-binding, or intercalating [10]. All
species binding to DNA will bind to the DNA by one of these binding modes. A

schematic of the binding modes of DNA is shown in Figure 2.23.

 

 

 

 

Figure 2.23. Bind modes of DNA. (A) External electrostatic, (B) Groove Binding, (C)
Intercalation. The figure was adapted from reference 10.

100

Electrostatic interactions occur along the exterior of the helix through non-specific
interactions such as those between alkali metal ions and the negatively charged phosphate
groups. In Figure 2.23 (A), this has been demonstrated with a positively-charged
polyamine. This positively charged amine can actually decrease the effective charge on
the DNA by neutralizing two phosphate groups. There are a variety of ions that interact
with DNA and a number of these species carry multiple positive charges. In contrast,
groove-binding interactions, Figure 2.23 (B), involve the direct interactions of the bound
molecule with the edges of the base pairs in the major or minor groove. Compounds can
bind in either the major or minor groove of the DNA double-helix. Groove binding
molecules, since they interact with the base pairs of DNA, can often be sequence specific
binding in regions rich in A-T or G-C base pairing. Chromomycin, distamycin, and
Hoechst 33258 bind in the minor groove of DNA and stabilize the double—stranded
species [10]. Intercalation of a planar aromatic ring system between the base pairs can
occur as well, Figure 2.23 (C). Intercalators include ethidium, daunomycin, and
adriamycin [10].

Compounds binding to DNA through any of these binding modes may help
maintain the duplex throughout the experiment. If these molecules were added to the
initial matrix solution, the oligonucleotides may be trapped in the matrix crystals with
these compounds attached. The presence of intercalators, groove-binders, and cationic
species may stabilize the double-stranded species during the formation of the MALDI
target. The compounds may also stabilize the duplex during the excitation phase, so the
distribution of energy in the crystal lattice does not cause the dissociation of the complex.

Lastly, the presence of stabilizing additives in the MALDI complex may stabilize DNA

101

duplex throughout the desorption/ionization process. The solid phase processes are
closely linked. For this reason, it would be difficult to discern if the addition of an
additive stabilizes the duplex through either step or both steps. The effect of stabilizing
additives will be presented here, but discussed further in the Solid Phase:

Desorption/Ionization section.

102

Electrostatic interactions
AIMEE

As mentioned previously, oligonucleotides have a negatively charged phosphate
backbone when in solution. In order to form a singly charged anion, combinations of
hydrogen, potassium, and sodium ions form adducts with the phosphate groups. In mass
spectrometry, the desalting of samples leads to simplified spectra containing fewer alkali
ion adducts and better resolution. Desalting/cleanup is commonly performed in MS
laboratories prior to analysis using desorption/ionization methods. While the spectra
have better resolution and peaks have fewer cation adducts, desalting the analyte solution
can decrease the stability of the double stranded species due to the repulsion of the
negatively charged phosphate groups [1 1].

The MALDI process typically yields only singly charged ions. However,
oligonucleotides can be trapped in the MALDI crystals with a higher charge state. If
desolvation energies increase as the number of charges increase [12-13], MALDI
sensitivity for oligonucleotides should be low. Additives were developed to not only
eliminate the alkali ion problem, but to provide protons to allow oligonucleotide
desorption as a singly charged ion. These additives improve spectra for oligonucleotides
without desalting the samples. When ammonium ions are present in solution, they
compete effectively with the alkali ions in complexing with the negatively-charged
phosphate groups. Adding ammonium salts such as ammonium acetate and diammonium
hydrogen citrate reduces the formation of alkali ion adducts [14-16]. If multiple charges
on a single strand prohibit desorption, multiple charges on a duplex create more problems

due to interchain repulsions. Thus, additives such as ammonium salts and Spermine

103

become more important when attempting to analyze double-stranded DNA. Their
effective use reinforces the importance of ionic forms of oligonucleotides during crystal
growth.

While diammonium hydrogen citrate (DAHC) is very useful in the analysis of
oligonucleotides, its does not influence the formation/preservation of double-stranded
oligonucleotide complexes. When DAHC is used as an additive in the MALDI
experiment, the DSRR values for the spectra do not increase. The resolution of the peaks
representing the single-stranded species increases. Also, there are fewer peaks
representing the binding of alkali ions to the phosphate groups. So, DAHC is helpful
when analyzing the single-stranded components of the duplex DNA, but does not help

maintain the non-covalent interactions between the two strands.

Polyamines

Previously, the role of polyamines as matrix additives has been explored and
Spermine was found to improve MALDI spectra for single-stranded oligonucleotides,
eliminating the need for ammonium citrate as well as desalting [17-19]. While the
presence of sperrnine enhances the detection of the single-stranded oligonucleotides,
could sperrnine stabilize the duplex in the MALDI experiment? Spermine and other
polycations increase the stability of duplexes in solution. This fact is demonstrated by
the increase in the melting temperatures of the double-stranded species when in the
presence of polyamines. If Spermine stabilized duplex DNA in solution, it is possible for

Spermine to stabilize DNA in the MALDI experiment.

104

Spermine is a polyamine that is of particular interest. Spermine is known to assist
in the analysis of oligonucleotides in MALDI [17-19]. Spermine also serves a key role in
the nucleus of cells. In the nucleus, sperrnine efficiently counteracts the repulsive
electrostatic interactions that occur between negatively charged strands and allows the
DNA to condense into a more compact structure [20]. The binding of sperrnine to an

oligonucleotide is demonstrated in Figure 2.24.

H 2 "'"”-~._.~,_‘~. 0_ :0 N

H 2 -

H 2 {CD’K/ ‘ N
Spermine (1H 2 g V" DNA

H 2 ...-- '0— :0

Figure 2.24. Interaction of sperrnine with an oligonucleotide. This figure was
reproduced from reference 12.

Spermine is also used by crystallographers when growing crystals for X-ray
crystallographic analysis. When Spermine is present at high concentration compared to

that of the oligonucleotide, homogenous crystals of the DNA will grow from the solution.

105

When these crystals are analyzed using X-ray crystallography, spermine is typically not

resolved in the final crystal structure.

Additives such as sperrnine could aid in the stabilization of the double stranded
species in the crystal formation process and the subsequent steps of excitation and

desorption/ionization. Figure 2.25 shows results of an experiment in which a double-

stranded oligonucleotide was analyzed using ATT as a matrix.

 

 

 

 

 

 

A) ATT:sp (M 1 M2.-H)-
+Sp‘

E

in

C

Q3

E W

E B) ATT (M1M2'H'G’

E

0

n:

6060 65b0 7600 7500 8600
mlz

Figure 2.25. The negative-ion UV MALDI mass spectra of duplex 2 using (A)
ATT/Spermine as the matrix and (B) ATT alone as the matrix.

It is important to note that previously duplex 1 was analyzed using ATT as the matrix and
no ions representing the duplex were detected. However, when duplex 2 is analyzed

using ATT:sp or ATT as the matrix, Figure 2.25, the single strands are detected in

106

addition to the double-stranded oligonucleotide with the loss of the guanine (G) base,
(Mle-G-H)’. When Spermine was added, the intact duplex, (Mle-HY, was detected
as well as its Spermine adduct. While the DSRR values of 0.05 for the duplex peaks were
relatively small, the WCSF value was 1. These results are significant and suggest that
additives known to complex with the duplex in solution may play an important role in
stabilizing the duplex in the MALDI experiment as well. Also, the Spermine stabilizes
the duplex in the experiment and also remains bound to the duplex throughout the
excitation and desorption/ionization steps. It is also important to note the dramatic
increase in resolution when examining the (Mle-G—H)‘ peak in both spectra.

The influence of Spermine has been firrther investigated. In the development of
sperrnine as an effective additive for single strand oligonucleotides, studies of the
additive influence over a broad concentration range led to the identification of an
optimum amount to use. Similarly, a set of experiments was performed with the duplex
as the analyte. The optimal concentration of sperrnine as a co-matrix was 12.5 mM.

Often in MALDI analysis of oligonucleotides, the resulting spectra are dependent
on the sequence of the analyte. For this reason, all experiments were performed with
duplexes having different sequences. When duplex 1 is analyzed using the
ATT/Spermine matrix, the results of experiment are different, Figure 2.26. The DSRR
value is 0.15 with the SSD and WCSF having values of 1.00. A peak representing a

Spermine bound to the duplex is also detected.

107

i (MPH)—

 

 

 

 

 

(Mz‘Hl’

E"
(D
C
9
E
0
.2
E l
‘1’ l
‘1 i

l “

i I (M1M2'Hl-

. ‘W Maul... MVWJIL, .
3000 4000 5000 6000 7000 8000 9000

mlz

Figure 2.26. Negative-ion UV MALDI mass spectrum of duplex 1 using ATT/Spermine
as the matrix. The DSRR value is 0.15, the WCSF is 1.00, and the SSD is 1.00.

108

Cgstallogglphic Condensing Agents

When crystals of double-stranded DNA are grown for analysis by X-ray
crystallography, there are a variety of condensing agents used to neutralize the anionic
strand-strand repulsions, allowing regular crystals to form [21]. These agents, ofien
polycationic, include the polyamines, sperrnine and spermidine, that have already been
discussed and other inorganic compounds such as ruthenium(III) hexammine chloride
[22] and cobalt (III) hexammine chloride. Similar to sperrnine, these inorganic
compounds play a key role in crystal growth and are frequently not resolved in the
resulting crystal structure. These crystallographic condensing agents may be largely
eliminated as part of the crystal growth process. For example, when growing crystals of
d(AGGBrCATGCCT) in the presence of cobalt(III) hexammine, no cobalt hexammine
ions are bound to the DNA in the resulting crystal structure [23]. Since sperrnine
improved the MALDI analysis of duplex DNA, perhaps other condensing agents would
have a similar effect. The cobalt hexammine counterion exhibits behavior similar to that
of the Spermine when in the presence of duplex oligonucleotides [24]. Cobalt (III)
hexammine also decreases inter-strand repulsions of the duplex by decreasing the net
effective charge on the phosphate groups. In a crystal structure, a hydrated
cobalt(III)hexammine is found to bind to the phosphate groups of oligonucleotides [25].

When cobalt hexammine is added to the MALDI matrix, there are significant

changes in the resulting spectrum, Figure 2.27. The peaks at m/z 8000 represent the

109

 

Relative Intensity

 

 

 

 

 

 

 

 

 

3000 4000 5000 6000 7000 8000 9000
mlz

Figure 2.27. Negative-ion UV MALDI mass spectrum of duplex 1 using ATT as the
matrix and cobalt(III) hexammine as an additive.
duplex, with approximately 50% of the complex remaining intact. From an analytical
standpoint, it is important to realize that, using MALDI MS, if a peptide P is analyzed, a
peak in the positive ion spectrum will appear corresponding to (M+H)+, and frequently as
(M-H)‘ if negative ions are being analyzed. However, if a noncovalently bound complex
such as a duplex is analyzed, only the components may be detected. To detect the intact
duplex present in a solution, Co(III) hexammine can be added to stabilize the complex.
Details of the mass spectral data reveal how Co(III) hexammine participates in
this experiment, Figure 2.28. A peak representing the intact duplex in ionic form is
observed, (Mle-H)‘, at m/z 7899. Peaks at higher m/z values represent duplexes

containing one or more cobalt ions. Cobalt addition accompanies the loss of 2 hydrogen

110

(M1 M2“5H+2CO)-

 

 

 

 

 

 

 

    

 

.53
8
9
E I
“>’ 1
:3 . (M1M2-9H+4Co)'
(or) ; (M1 M24")-

i

l

7500 7900 8300

mlz

Figure 2.28. Detailed view of the duplex region of the spectrum shown in Figure 2.27.
At m/z 7899, the deprotonated duplex is detected. For this experiment, with the
particular amount of additive used, complexes containing up to 4 Co(II) ions can be
observed. The peak at m/z 8013 corresponds to the addition of 2 Co(II) ions and the loss
of S protons. At m/z 8127, 4 Co(II) ions are bound to the duplex and 9 protons are
displaced. Complexes with more Co ions attached are observed at higher additive
concentrations, but at the expense of sensitivity.

atoms. This, in addition to the fact that the overall charge on all of the ions observed is -
1, indicates that the cobalt atoms are bound as Co”. Multiple cobalt adducts of both the
single strand and duplex ions are formed. Experiments performed over a range of
Co(III)(NH3)6 concentrations suggest that the maximum number of Co ions that can be
attached to an oligonucleotide is proportional to the number of phosphates present, with
one cobalt able to complex with 2 phosphate groups. When the cobalt is in a +3 state, the

metal:phosphate ratio of 1:3 is observed in solution complexes, as measured using

circular dichroism. This ratio of complexation appears to be a prerequisite to crystal

1]]

formation [26]. Since multiple metal complexes are observed for the single strands, it is
assumed that a single cobalt ion interacts with two adjacent phosphate groups of a single
chain-intrachain interactions need not be considered. It should be noted that experiments
have been reported in which FeC13 was used as an additive to study small, single-strand
oligonucleotides by MALDI-MS. In these experiments, iron as both Fe3+ and Fe2+ was
observed to complex with these analytes in the gas phase [27].

In MALDI mass spectrometry, the ideal situation is to have one mass spectral
peak represent one analyte. When multiple peaks are formed, distributing information
over a number of m/z values, detection limits may be lowered. However, in this case, the
reagent provides a substantial increase in sensitivity that outweighs the appearance of
multiple peaks.

The effect of enhancing MALDI signals for duplexes is not observed when Co(II)
compounds such as CoClz are added to the solution from which the MALDI matrix
crystals are grown, a surprising observation since the cobalt appears in the gas phase
ionic species as Co(II). This suggests that a redox step is key in the formation of the
observed species. In addition to the obvious utility of Co(III) hexammine as a tool in
MALDI MS, the example presented here suggests that MALDI may also prove to be a
useful analytical tool for investigating the processes through which reagents react with

duplexes and assist in the generation of condensed phase samples.

112

Metal ions

Similar to amines and polyamines, metal ions are known to stabilize duplex DNA
structures. In order to explore the effects of metals as additives, several experiments
were performed. A variety of metal ions were selected due to their documented
interactions with DNA. In a previous study, potassium, cesium, and sodium were found
to increase the melting temperature or duplex DNA in solution at concentration of up to l
M [11]. A variety of concentrations were used as well in order to record the DNA/metal
ion ratio and its effect on the DSRR and WCSF values. Salts of other metals including
copper, nickel, iron, lithium, calcium, silver, zinc, and lead were also used. With the
addition of these metal ions, the DSRR values did not increase. The resolution of the
single-stranded species did decrease significantly. As the positively charged metal ions
interacted with the negatively charged oligonucleotide, metal-DNA adducts were formed
complicating the spectra and decreasing the resolution of the peaks.

Compounds such as spermine and DAHC have been found to decrease the
number of alkali ions that bind to the phosphate groups of DNA during the MALDI
process. The experiments described above were repeated using ATT:spermine,
ATTzDAHC, or HPAzspermine as the matrix. While the number of peaks representing
the metal ions binding to the DNA decreased, there was still no increase in the DSRR

values for any of the experiments.

113

Wand Minor Groove Binders

The major and minor grooves of DNA are very different. As a result, different
types of compounds tend to bind in the types of grooves. Molecules that bind in the
minor groove of DNA are typically polycyclic compounds that have the ability to twist
their structure to fit into the helical curve of the minor groove. The major groove of
DNA is more suited to the binding of proteins and peptides. Groove-binding molecules
tend to be more sequence specific than other types of compounds that interact with DNA.

One of the first groove-binding compounds studied was chromomycin.
Chromomycin has been found to bind readily to DNA containing a —XCGX- motif in the
presence of magnesium ions [28-29]. In this motif, X is simply any nucleotide.
Chromomycin and compounds similar in structure bind to DNA not as a monomer but as
ammm.

Experiments were performed using chromomycin as an additive with the duplex
d(CCGGAATTGGCC)-d(Tl"GGCCAATI‘CCGG). This duplex was selected because it
contained the consecutive CG base pairs. The initial experiment was performed by
mixing a microliter of a 25 micromolar solution of duplex with a 25 mM solution of
chromomycin. A microliter of this solution was then mixed with a microliter of either the
ATT or HPA matrix solution. Both experiments yielded DSRR values of 0. Another set
of experiments was performed by incubating the chromomycin solution with the duplex
solution in the presence of 1) no magnesium ions, 2) 1 mM magnesium chloride, 3) 10
mM magnesium chloride, or 4) 25 mM magnesium chloride. These solutions were
analyzed using ATT and HPA after 1) 0 minutes, 2) 1 hour, 3) 12 hours, and 4) 24 hours.

The incubated solutions were held at 37°C for the allotted time period. After incubation,

114

the solutions were analyzed in MALDI with HPA or ATT as the matrix. No duplex ions
were detected.

Similar studies were performed using groove-binding compounds known to bind
in A-T sequences of the DNA minor groove. These compounds are mostly crescent
shaped and contain an amino group capable of hydrogen bonding with the A-T base
pairs. These are not capable of binding with GC pairings due to the amino group on the
guanine. Distamycin, known to bind in the AATT minor groove of the oligonucleotide
d(CGCAAATTTGCG), had no effect on the DSRR values for the MALDI experiments.
The aromatic diamidines, berenil and 4’,6-diamidino-2-phenylindole (DAPI), were also
used. Again, studies performed with DAPI and berenil also gave DSRR values of 0.

Lastly, Hoechst 33258 was used with similar results.

115

 

Histones and Peptides

If some small molecules that are known to interact with DNA can stabilize the
duplex in the MALDI experiment, larger molecules such as proteins and peptides may
also have a stabilizing effect. Interactions between DNA and proteins play important
roles in various biochemical processes. Chromosomes contain DNA and an equal mass
of histone and non-histone proteins [30]. Histones have molecular weights between
13,000 and 30,000 g/mol. Histones are proteins that are highly conserved across species
and contain many basic (Arg, Lys) residues. It is proposed that these positively charged
amino acid side chains interact with and stabilize the DNA through salt bridges. In sperm
cells, DNA is tightly packed due to the presence of arginine-rich proteins called
protamines. Protamines bind to DNA with their a-helices in the major groove of the
DNA, where they neutralize the charge of the phosphates and enable the tight packing of
the duplexes [10]. While the main driving forces for protamine-DNA binding are the
ionic interactions, there is a distinct co-operativity that cannot be explained on the basis
of electrostatic interactions alone [31]. Protamines perform functions similar to histones,
but are generally much smaller, with molecular weights between 4,000 and 10,000 g/mol
[32]. The effect of the presence of peptides on double-stranded oligonucleotides has not
been examined.

In an experiment designed in our laboratory to explore the differences in
resolution and sensitivity between a peptide and an oligonucleotide in UV-MALDI-MS, a
solution was prepared that contained equal amounts of a peptide and oligonucleotide of
similar molecular weights. In the resulting spectra, peaks were detected for the peptide,

(P+H)+, the oligonucleotide, (O+H)+, and a peptide-oligonucleotide complex, (P+O+H)+.

116

 

Similar results have been reported previously [33]. In order to examine the influence of
peptides on double-stranded oligonucleotides, several peptides were combined with
duplex oligonucleotides and analyzed by UV MALDI-MS. The peptides were used at a
level of 1 pmol or less. An additive that is most effective at a level similar to that of the
analyte, rather than that of the matrix, will be referred to as a microadditive.

The peptides, dynorphin A, cytochrome c, fibrinogen binding inhibitor peptide
(FBI peptide), B-melanocyte stimulating hormone (B-MSH), katacalcin, kemptide, B-
casein, ubiquitin, kassinin, and oxytocin were all purchased from Sigma (St. Louis, MO)
and used without further purification. Peptides were dissolved in MilliQ water at
concentrations of 1.0 pmol/pL. A microliter of a peptide stock solution was combined
with a microliter of the stock double-stranded oligonucleotide solution. When incubated,
the peptide/oligonucleotide solution was heated to 37° C for 15 minutes. A microliter of
the resulting solution was spotted onto the MALDI target with a microliter of the matrix
solution. Samples analyzed with fucose were prepared by mixing 1 pL of the
peptide/oligonucleotide solution with 1 0L of the fucose solution and l uL of the co-
matrix solution on a gold sample plate. The mixture was then allowed to air-dry.

The duplex of d(CCGGAATTGGCC) bound to d(TTGGCCAATTCCGG) was
used for the experiments designed to evaluate peptides as microadditives with
ATT:spermine as the matrix, Figure 2.29(A). The SSD is 0.7 and the DSRR is 0.006.
While the intact duplex, (Mle—H)‘ is detected in this experiment, the peak representing

the (MzMz-H)“ species is surprisingly intense, giving the spectrum a WCSF of 0.3.

117

l A (Ma-Hy

(M1‘H). If

 

i F \k (MzMz'H).
‘. A‘ ~ _“ L 4hA‘A—w’

 

 

 

 

 

 

 

Relative Intensity

(M1M2-H)'
+sp
_’
3000 4000 5000 6000 7000 8000 9000

mlz

Figure 2.29. Negative-ion MALDI mass spectra of duplex 1 with A) no peptide present
and B) with B-MSH present. ATT:sp was used as the matrix for both experiments. 25
pmol of duplex and lpmol of B-MSH were deposited onto the MALDI target. For A), the
SSD is 0.7, the DSRR is 0.006, and the WCSF of 0.3. For B), the SSD is 1.1, the DSRR
is 0.06, and the WCSF is 1.0. The peptidezduplex ratio is 1:25.

When only 1 pmol of B-MSH was added to the target (which contained 25 pmol
of duplex) used to obtain Figure 2.29(A), the spectrum shown in Figure 2.29(B) is
obtained. The SSD changes dramatically to 1.1, the DSRR is 0.06, and the WCSF is 1.0.
There are no peaks representing either (M.Ml-H)‘ or (MzMz-HT. In this experiment, [1-
MSH stabilized, to some extent, double-stranded DNA ions in UV MALDI MS, when

present at a level of one picomole on the MALDI target. Other peptides evaluated are

listed in Table 2.7. All peptides were used as additives by adding 1 pmol of peptide to 25

118

pmol of duplex 1 and analyzed using ATT:sp as the matrix. The DSRR and SSD values
are listed in the table. All WCSF values are l.

The peptides evaluated as microadditives can be grouped based both on their size
and amino acid content. It is possible that the most successful experiments would have
involved peptides that have a size similar to that of the oligonucleotide. From spermine
experiments performed previously [17-19], we have learned that additives capable of
multiple interactions with the oligonucleotides are important. For this reason, the number
of basic residues in the peptide may be important, with multiple basic residues increasing
the number of electrostatic interactions possible. Previously it was determined that
peptides containing basic residues increase the melting temperature of duplex
oligonucleotides in solution [34]. Protonated backbone nitrogens may aid in stabilization
as well.

When examining the data in Table 2.7, attempts were made to correlate the
experimental results with the structural features of the peptides. The two peptides used
with the most notable success were dynorphin A and B-MSH. Both of these larger
peptides contain basic residues with three arginines and two lysines out of the 17 amino
acids in dynorphin A and six basic residues out of the 22 amino acids residues in B-MSH
including three lysines, two arginines, and one histidine. Dynorphin A and B-MSH
presumably interact with the double-stranded oligonucleotides at multiple sites. The
backbone nitrogens may be important in the stabilization of the duplex. A positively-
charged backbone in a peptide may behave similarly to spermine in solution. The side
chains of the peptide may also play an important role. The amino acids, tyrosine,

phenylalanine, and tryptophan, contain aromatic side chains. These side chains may

119

 

 

8.2m 8.0 8... od 3 5 www 8..me
8.8m 8.o 8.0 od 3 we <<< .28 .28.: 8.88.:
8.34 8.0 8... od 3 3 <<<<<< 8.8.8me
8.2m . 2:. 8... O... 3 m... ”.255 8.8 ...z_2.8m
8.38 23 8.8 o.o 3 3 do? 5.28.5 8.83
88? 8.02 8... od 3 3 v. m....-_>.8m-wz
8.8.. 8.2: cod 3 3 3 v.5. 8.8.:

8.8: 8... 8... od 3 3 52359:: 888. .8
8. E 88 3.8 o.o 3 mo <>m<mm.. 88me

8.88 8.0 8.8 O... 3 Z z<zoa2m>zamzommdwwzo 52838.

8.88 8.9 8.8 3 3 Z oxaawoamaxmsmiomoxxm... :22-..

8.2.3 8. : 8.: 3 3 Z ozo>>5x&_mm..u.oo> < 5:885

25 9: .\. was ammo awe; am» 8533 8.8:. 838..

 

_ 5.8a 8 2.8% 5.22 >: 8 8:25.: :2: E; 83.83. 92.8.— .2 as;

120

partially intercalate between the nitrogen-containing bases of the oligonucleotide,
stabilizing the duplex. Increases in the melting temperature of duplex DNA have been
reported when in the presence of peptides containing phenylalanine residues [34]. Basic
side chains may also play an important role as discussed previously.

Since peptides containing only alanine residues were not as successful as
additives, it appears that the backbone nitrogens are not effective in stabilizing the
duplex. Small peptides containing basic residues such as Lys-Lys-Lys were found to
increase the DSRR value although these smaller peptides were less effective than
Dynorphin A and B-MSH. Similarly, peptides containing aromatic side chains were also
capable of increasing the DSRR values for the experiments.

Since the best results are seen with the peptides Dynorphin A and B-MSH, which
contain both aromatic and basic amino acids, it is not perhaps the size of the peptide that
is of primary importance, but the spacing between basic amino acids. When basic amino
acids in a larger peptide are separated by other residues, the separation between those
basic residues may be similar to the spatial separation between the phosphates of the
oligonucleotide. This may lead to greater stabilization of the duplex. Clearly, more work
is required to find the Optimal peptide, but these experiments establish structural aspects
of the additives that are important and show that peptides can, at relatively low levels,
stabilize to some extent duplexes in the UV MALDI experiment.

In order to maximize the interaction between the peptides and the double-stranded
oligonucleotides, several conditions were varied in the experiments. At first the peptide
and oligonucleotide were combined on the MALDI plate before the addition of the

matrix. To insure thorough mixing, peptide and oligonucleotide solutions were also pre-

121

mixed before deposition on the target. There was no difference between spectra acquired
using on target mixing or pre-mixed solutions. A solution of the peptide and
oligonucleotide was also pre-mixed and incubated at both room temperature and at 37°C
prior to MALDI analysis. When incubated at either temperature, samples again yielded
spectra similar to the unincubated samples.

For the data shown in Table 2.7, one picomole of each peptide was used in the
formation of the MALDI target. While 1 pmol was optimal for most peptides,
improvements were seen with as little as 50 frnol of peptides present in the target. FBI
peptide, katacalcin, and B-MSH all yielded duplex spectra with improved DSRR values
with less than 1 pmol of peptide present in the target. While addition of these peptides to
the MALDI target improved the DSRR, the B-MSH peptide gave the best results. Figure
2.30 shows the negative-ion MALDI mass spectrum of a target containing 100 fmol of [3-
MSH and 25 pmol of double-stranded oligonucleotide, using ATT:sp as the matrix. The
SSD was 0.94, the DSRR was 0.09 and the WCSF was 1.00. The presence of the peptide
still influences the resulting mass spectrum even when the relative concentration of the
oligonucleotidezpeptide is 250: 1.

Here we show that certain peptides can stabilize, to some extent, double-stranded
DNA ions in MALDI, when present at levels of less than one picomole on the target.
Since they are most effective at levels similar to that of the analyte, rather than the

matrix, we will refer to these generally as microadditives.

122

(Mi-Hr

(Mz-H)' (M M H -
~ 1 2" )
E‘
W
C
a
.5 +sp
3 x10 —'
E ,—
O
m

 

 

 

 

 

 

trawl. W...

I r p l l

3000 4000 5000 6000 7000 8000 9000
mlz

Figure 2.30. Negative-ion MALDI mass spectrum of duplex l with 100 fmol of B-MSH
and ATT:sp as the matrix. The SSD was 0.94, the DSRR was 0.09 and the WCSF was
1.00. The peptidezduplex ratio is 1:250.

123

Intercalators

Intercalators are planar, aromatic cations that insert themselves into the aromatic
ring system between the DNA base pairs. As a results of this insertion, often
intercalators cause the DNA helix to unwind, distorting the DNA backbone. Two main
intercalators were studied. These are methylene blue and ethidium bromide. While it
was believed that methylene blue could act as both a groove-binder and as an intercalator,
spectroscopic data clearly indicates the intercalation of methylene blue between two
consecutive base pair of an oligonucleotide [35-36].

Stock solutions of methylene blue and ethidium bromide were made at a
concentration of 1 mM in water. Saturated matrix solutions of ATT and HPA were made
using a 1:1 acetonitrile/water solution and a saturated ATT/spermine co-matrix was made
using a 1:1 solutions of acetonitrile and 25 mM aqueous spermine solution. For
experiments performed with ethidium bromide and methylene blue, the oligonucleotide
solution was mixed with an equal volume of the additive solution prior to spotting on the
sample plate. A microliter of matrix was then added.

After formation of the target, the oligonucleotide/methylene blue/matrix crystals
were analyzed. It was that methylene blue binds to the single-stranded oligonucleotide in
the MALDI expeirment. The resulting spectrum is shown in Figure 2.31, with the
structure of the methylene blue cation. While methylene blue (MB) is observed bound to
the single-stranded species, its presence appears to stabilize the duplex as well, Figure
2.31. In this experiment, the DSRR value was 0.03, the SSD was 0.9, and the WCSF was

1.0. With a WCSF of 1.0, it appears that a portion of the duplex was maintained

124

throughout the entire UV-MALDI experiment, due to the presence of this additive,

although the duplex represents only a small percentage of the initial complex.

 

 

 

 

 

 

 

 

(M1-Hr
_ N \
(Mz‘H) Q \
H3C\N S \ +CH3
2‘ CH3 CH3
":7:
c
m
E x10
a)
.5
8 MB _
3500 4500 5500 6500 7500 8500

Figure 2.31. The negative-ion MALDI mass spectrum of duplex 1 with methylene blue
as an additive. ATP was used as the matrix. The SSD was 0.9, the DSRR was 0.03, and
the WCSF was 1.0. The additive duplex ratio is 40:1.

As discussed previously, the presence of spermine can lead to an improvement in
resolution of oligonucleotide MALDI expeirments. The addition of spermine to the
matrix solution may also improve the results of experiments using other additives such as
methylene blue. If the methylene blue experiment from Figure 2.31 is repeated with an

ATT matrix solution containing spermine, the resolution in the duplex region is

improved, the DSRR value increases, and a methylene blue adduct is detected in the

125

duplex region, Figure 2.32. The DSRR increases significantly from 0.03 in Figure 3 to

0.07 in Figure 5. The WCSF is 1.0 and the SSD is 0.9.

 

 

 

 

(M1M2-H)'
.E‘
m
C
:3 +MB
— I:
a)
.2
E
0)
CK
7000 7400 7800 8200 8600 9000
mlz

Figure 2.32. A portion of the negative-ion MALDI mass spectrum of duplex l with
methylene blue (MB) as an additive. ATT:sp was used as the matrix. The DSRR is 0.05
and the WCSF is 1.0. The spermine:MB:duplex ratio is 500:40:l.

As mentioned previously, ethidium bromide has been found to stabilize duplexes
in solution. When used in the MALDI experiment, equal volumes of duplex l and
ethidium bromide solutions were combined. One microliter of the
oligonucleotidezethidium solution was spotted on the MALDI target with one microliter
of the HPA matrix solution. The molar amounts of components of the target are
matrixzadditivezoligo, 2000:40:1. This is typical since effective additives for MALDI MS
are usually present at molar amounts greater than the analyte. The resulting spectrum is

shown in Figure 2.33, with the structure of the ethidium ion. The ethidium cation binds

126

to both the single and double-stranded species of duplex 1. That is, in addition to analyte
ions, (A—H)‘, there are also (A-2H+Et)' ions observed. With the appearance of the

dimer peaks, the DSRR increases from 0 (in Figure 1 B) to 0.05.

 

   
    
 

 

 

 

 

<M1-Hr
M -H '
( 2 ) Et
.E‘
m
c
93
E
0
% X 10 (M1M2-H)'
Fr" ’-
+Et (M1M1'Hr
_, +Et
(M2M2'H)-
Et +Et +Et

 

 

3000 4000 5000 6000 7000 8000 9000
mlz '

Figure 2.33. The negative-ion MALDI mass spectrum of duplex 1. HPA was used as the
matrix with ethidium (Et) bromide as an additive. Ethidium adducts of both the single
and double-stranded oligonucleotides are observed. The SSD is 1.0, DSRR is 0.05 and
the WCSF is 0.5. The additive duplex ratio is 40:1.

In Figure 2.33, while there are peaks representing the (Mle-H)‘ species, there
are also peaks representing the (MIMI-H)" and (MzMz-H)' species. While the DSRR
increased from 0 to 0.05, the WCSF is 0.5, suggesting random dimerization. If random
dimerization occurs, the duplex may completely dissociate and the single strands are

likely dimerizing with no sequence specificity as the crystals are growing. Thus, while

127

ethidium addition does result in "dimer ions", we conclude that this additive does not

stabilize the initial duplex throughout the UV-MALDI experiment.

128

References

10.

11.

12.

Vertes, A.; Gijbels, R. Levine, R. D. Homogeneous bottleneck model of matrix-
assisted ultraviolet—laser desorption of large molecules. Rapid Commun. Mass
Spectrom. 1990, 4, 228- 233.

Vertes, A.; Levine, R.D. Sublimation versus fragmentation in matrix-assisted
laser desorption. Chem. Phys. Lett. 1990, 1 71, 284-290.

Bencsura, A.; Navale, V.; Sadeghi, M.; Vertes, A. Matrix-guest energy transfer in
matrix-assisted laser desorption. Rapid Commun. Mass Spectrom. 1997, I I, 679-
682.

Zenobi, R.; Knochenmuss, R. Ion formation in MALDI mass spectrometry.
Mass Spectrom. Rev. 1998, I 7, 337-366.

Karas, M.; Bahr, U.; Strupat, K.; Hillenkamp, F.; Tsarbopoulos, A.; Pramanik,
B.N. Matrix dependence of metastable fragmentation of glycoproteins in MALDI
TOF mass-spectrometry. Anal. Chem. 1995, 67, 675-679.

Karas, M.; Bahr, U.; Stah-Zeng, J .R. Steps toward a more refined picture of the
matrix function in UV MALDI. In: Large Ion: Their Vaporization, Detection and
Structural Analysis; T. Baer; C.Y. Ng; I. Powis (Eds); Wiley: London, 1996.

Billeci, T.M.; Stults, J .T. Tryptic mapping of recombinant proteins by matrix-
assisted laser desorption/ionization mass spectrometry. Anal. Chem. 1993, 65,
1709-1716.

Gusev, A.I.; Wilkinson, W.R.; Proctor, A.; Hercules, D.M. Improvement of
signal reproducibility and matrix/comatrix effects in MALDI analysis. Anal.
Chem. 1995, 67, 1034-1041.

Distler, A.M.; Allison, J. Improved MALDI-MS analysis of oligonucleotides
through the use of fucose as a matrix additive. Anal. Chem. 2001, 73, 5000—
5003.

Blackburn, G.M.; Gait, M.J. Nucleic acids in chemistry and biology, 2nd ed.;
Oxford University Press: New York, 1996.

Schildkraut, C.; Lifson, S. Dependence of the melting temperature of DNA on
salt concentration. Biopolymers. 1965, 3, 195-208.

Asara, J. Enhancing spectra in desorption/ionization mass spectrometry through
the use of chemical additives. Michigan State University, 1999.

129

13.

14.

15.

l6.

17.

18.

19.

20.

21.

22.

23.

24.

Takayama, M. Formation of M+' ions and electronic excitation under fast-atom
bombardment conditions by using a liquid matrix. J. Am. Soc. Mass Spectrom.
1995, 6, 114-119.

Currie, G.; Yates, J. 111. Analysis of oligodeoxynucleotides by negative-ion
matrix-assisted laser desorption mass spectrometry. J. Am. Soc. Mass Spectrom.
1993, 4, 955-963.

Wu, Z.; Biemann, K. The MALDI mass spectra of carbosilane-based dendrimers
containing up to eight fixed positive or 16 negative charges. Int. J. Mass
Spectrom. Ion Processes. 1997, I65-166, 349-361.

Pieles, U.; Zilrcher, W.; Schiir, M.; Méser, H.E. Matrix-assisted laser-desorption
ionization time-of-flight mass-spectrometry — A powerful tool for the mass and
sequence-analysis of natural and modified oligonucleotides. Nucleic Acid Res.

1993, 21, 3191-3196.

Asara, J .; Allison, J. Enhanced detection of oligonucleotides in UV MALDI MS
using the tetraamine spermine as a matrix additive. Anal. Chem. 1999, 71, 2866-
2870.

Asara, J .M.; Hess, J .S.; Lozada, E.; Dunbar, K.R.; Allison, J. Evidence for
binding of dirhodium bis-acetate units to adjacent GG and AA sites on single-
strandcd DNA. J. Am. Chem. Soc. 2000, 122, 8-13.

Distler, A.M.; Allison, J. 5-Methoxysalicylic acid and spermine: a new matrix
for the matrix-assisted laser desorption/ionization mass spectrometry analysis of
oligonucleotides. J. Am. Soc. Mass Spectrom. 2001, 12, 456-462.

Esposito, D.; Del Vecchio, P.; Barone, G. Interaction with natural polyamines
and thermal stability of DNA. A DSC study and a theoretical reconsideration. J.
Am. Chem. Soc. 1997, 119, 2606-2613.

Pelta, J .; Livolant, F.;1Sikorav, J .L. DNA aggregation induced by polyamines
and cobalthexamminc. J. Biol. Chem. 1996, 271, 5656-5662.

Karthe, P.; Gautham, N. Structure of d(CACGCG)-d(CGCGTG) in crystals
grown in the presence of rutheniumIII hexammine chloride. Acta Cryst. 1998,
D54, 501-509.

Nunn, C.M. Crystal structure of d(AGGBrCATGCCT): implications for cobalt
hexammine binding to DNA. J. Biomol. Struct. Dyn. 1996, 14, 49-55.

Malinina, L.V.; Makhaldiani, V.V.; Tereshko, V.A.; Zarytova, V.F.; Ivanova,

M.F. Phase diagrams for DNA crystallization systems. J. Biomol. Struct. Dyn.
1987, 5, 405-433.

130

25.

26.

27.

28.

29.

30.

31.

32.

33.

34.

35.

36.

Ramakrishnan, B.; Sekharudu, C.; Pan, B.; Sundaralingam, M. Near-atomic
resolution crystal structure of an A-DNA decamer d(CCCGATCGGG): cobalt
hexammine interaction with A-DNA. Acta Cryst. 2003, D59, 67-72.

Kankia, B.K.; Buckin,V.; Bloomfield, V.A. Hexamminecobalt(III)-induced
condensation of calf thymus DNA: circular dichroism and hydration
measurements. Nucleic Acids Res. 2001, 29, 2795-2801.

Hettich, R.L. Formation and characterization of iron-oligonucleotide complexes
with matrix-assisted laser desorption/ionization Fourier transform ion cyclotron
resonance mass spectrometry. J. Am. Soc. Mass Spectrom. 1999, 10, 941-949.

Gao, X.; Patel, DJ. Solution structure of chromomycin'DNA complex.
Biochemistry. 1989, 28, 751-762.

Aich, P.; Sen, R.; Dasgupta, D. Role of magnesium ion in the interaction between
chromomycin A3 and DNA: binding of chromomycin A3-Mg2+ complexes with
DNA. Biochemistry. 1992, 31 , 2988-2997.

Bradbury, E.M.; Moss, T.; Hayashi, H.; Hjelm, R.P.; Suau, P.; Stephens, R.M.;
Baldwin, J .P.; Cranerobinson, C.; Nucleosomes, histone interactions, and role of
histone-H3 and histone-H4. Cold Spring Harb. Sym. 1977, 42, 277-286.

Porschke, D. Nature of protamine-DNA complexes: a special type of ligand
binding co-operativity. J. Mol. Biol. 1991, 222, 423-433.

Ausio, J. Histone H1 and evolution of sperm nuclear basic proteins. J. Biol.
Chem. 1999, 274, 31115-31118.

Harrison, J .G.; Balasubramanian, S. Synthesis and hybridization analysis of a
small library of peptide-oligonucleotide conjugates. Nucl. Acids Res. 1998, 26,
3136-3145.

Robledo-Luiggi, C.; Vera, M.; Cobo, L.; Jaime, E.; Martinez, C.; Gonzalez, J.
Partial intercalation with nucleic acids of peptides containing aromatic and basic
amino acids. Biospectroscopy. 1999, 5, 313-322.

Bradley, D.F.; Stellwagen, N.C.; O’Konski, C.T.; Paulson, C.M. Electric
birefringence and dichroism of acridine orange and methylene blue complexes
with polynucleotides. Biopolymers. 1972, 11, 645-652.

OhUigin, C.; McConnell, D.J.; Kelly, J .M.; van der Putten, W.J.M. Methylene

blue photosensitised strand cleavage of DNA: effects of dye binding and oxygen.
Nucleic Acids Res. 1987, 15, 7411-7427.

131

VI. Solid Phase: Desorption/Ionization

In the last phase of the MALDI experiment, the analyte molecules finally evolve
into gas phase ions, reactions (3a-e, 3a'-e'). There may be direct desorption of matrix
molecules and ions, and analyte molecules and ions, from the excited crystals, reactions
3a-c. The possibility of desorption of larger clusters has been proposed. Desolvation
reactions, similar to those proposed in ESI, could then lead to gas phase analyte ions,
reactions 3d,e. If multiply-charged ions are generated from clusters, recombination with
electrons would yield the observed species. The formation of gas phase ions through
desorbed clusters has recently been considered by Karas [1].

If the analyte duplex is intact to this point in the experiment, what factors may
lead to its destruction or stabilization? If singly-charged duplex ions are desorbed either
directly from the condensed phase or from a large gas phase cluster, fragmentation may
occur due to a number of circumstances. First, the energy content may be sufficient for
dissociation to occur. The (Mle-H)‘ ion could be unstable in the gas phase because of
the energy transferred into the complex during excitation. It could also be unstable
because of complex charge distributions such as multiple anionic sites or multiple
protonated bases within the ion that carries an overall single charge. When the solvent is
removed, repulsive forces could be sufficient to lead to fragmentation. While Karas
proposed that multiply charged species may be initially formed in the gas phase in
MALDI, and recombination reactions occur, so that singly charged ions are the "lucky
survivors", his prime focus was on peptides, which could be formed with +n charges, and
could react with desorbed electrons [1]. Since oligonucleotides would most likely be

desorbing with excess negative charges, recombination would have to be with available

132

cations (N a+(g)) and desorbed proton donors such as (T+H)+. If charge recombination
were an important part of the gas phase aspect of MALDI, this would be detrimental to
the stabilization of DNA duplexes, since recombination reactions are exothermic. If
duplexes remain intact until the step where gas phase ions are formed, approaches may be

developed to stabilize the duplexes at this point.

133

What torm does the double-stranded species, M !M g, take in the crystals 0 er
evaporation at the solvents, and how does this influence the MALDI mass spectra?

There are those who feel that analytes are trapped in the MALDI matrix in neutral

 

form, while others find it obvious, "based on chemical common sense", to expect that the
target is a solid solution with species in forms representative of their states in solution [1].
If oligonucleotides are present in the solid matrix containing multiple charges, a
condition determined during the target formation step, then the impact of this fact may be
seen in the final step. The energy required to desorb and essentially desolvate multiply
charged ions simply does not appear to be available in MALDI. Thus, it is important to
consider this aspect, since it is an issue that is more important for oligonucleotides than
for peptides. The state of the analyte might be controllable by the pH or ionic strength.
The existence of oligonucleotides in a charged state would explain why DNA desorption
is more difficult than peptides desorption with DNA signals having greater dependence
on size of the analyte. However, the strands may also exist in a neutral state. For this
case, when the crystal forms from the solution, the ionic form present in the solution must
convert to the neutral form. If the oligonucleotide has no charges, it will undergo
desorption and ionization like a peptide or other neutral analyte.

There are several solution conditions that will play a role in the charge state of the
analyte in the MALDI target. The first of these is pH. In basic and neutral solutions, this
variable may be relatively simple to investigate. The pKa of the phosphate backbone
indicates how many phosphate groups should be ionized with a matrix such as ATT. At a
neutral pH, there could also be a distribution of positive and negative charges on a single

strand. At lower pH's, the phosphates can be protonated. With a neutral phosphate

134

backbone, the duplex would be more stable than a duplex with negatively charged
phosphate backbones. However, at a very low pH, all of the phosphates may be in
neutral form with some of the nitrogen bases protonated as well. The protonation of the
nitrogen bases eliminates their ability to hydrogen bond to opposing bases, decreasing the
stability of the duplex.

Again, for a system where conditions have been found such that MALDI analysis
of a duplex yields a spectrum where at least a portion of the duplex remains such as
Figure 2.33, it is important to map out the variation of the quantifiers with pH of the
solution. As a droplet evaporates, the concentrations change making it difficult to
interpret a set of spectra obtained by growing crystals from solutions with simply a range
of initial pH values. Adding a strong acid, HCl, or a strong base, NaOH, will not provide
much information about the pH throughout the MALDI process. For this reason, buffers
will be used to keep the pH constant during most of the evaporation process. It may be
difficult to determine, when changes are observed, if the changes are due to the stability
of the duplex in the first phase of the experiment or in the final desorption step. While
difficult, this is clearly an important question and pH is a relevant variable in this regard.

Using duplex 1 as the analyte and ATT/sp as the matrix, the influence of pH was
explored. The ATT/Sp starting solution had a pH of around 7.5. This allowed for the
detection of double-stranded DNA as shown in Figure 2.26. The starting matrix and
analyte solution were then buffered to various pH values: 5.0, 6.0, 7.0, 8.0, and 9.0. For
comparison, hydrochloric acid was added to an ATT/Sp matrix solution to yield a pH of

2.0.

135

If analyte molecules can be trapped in the matrix in an ionic form, as the duplex
ions desorb, the negative charges on the double-stranded oligonucleotides will lead to
repulsive interactions, dissociating the duplex. In order to understand the charge state of
the DNA in the MALDI target, the charge state in solution must be understood. ESI-MS
was a useful tool to study the charge states of an oligonucleotide in solution. In ESI, a
single-stranded oligonucleotide, d(le), in solution, was injected into the mass
spectrometer and a negative-ion spectrum was recorded. The pH of the starting solution
was around 7.5, similar to that of the ATT/sp matrix starting solution. The spectrum
contained two peaks representing two charge states of the oligonucleotide. The -2 and -3
charge states were seen in the ESI-MS experiment. This suggests the oligonucleotide

may only contain 2-3 charges when in solution.

136

It the duplex remains intact until the final step, and the gas phase ions promptly and

extensively dissociate, how can the species be stabilized?

The dissociation of the duplex ions may be due to energy content of the desorbed
ions. If this is the case, the addition of fucose may help at this phase of the experiment as
well. A certain amount of energy is required to desorb molecules and ions from the
MALDI target. If the amount of energy deposited into the analyte exceeds the amount of
energy needed for desorption, the analyte will be desorbed with an excess of energy.

This excess energy may be sufficient to dissociate the complex. The addition of non-
absorbing species such as fucose to the matrix may decrease the energy transferred to the
analyte and enable the molecules to desorb intact, but not dissociate once in the gas
phase.

Matrix additives used to increase the bond dissociation energy of the duplex may
also stabilize the duplex in the desorption/ionization process. If molecules known to bind
to DNA can increase the dissociation energy of the duplex, the excited gas phase ions
may not have enough energy to break the non-covalent bonds holding the two strands
together. Additives such as spermine have been found to form a stable complex.
Spermine even holds the duplex together throughout the desorption/ionization step
evident by the observation of the (M1M2+sp)+ peak in the spectrum. While the formation
of this complex is not ideal, it provides insights on how to stabilize the duplex ions in the
gas phase. The presence of the (M.M2+sp)+ peak in the spectrum also shows that the
additives discussed in the Solid Phase: Excitgtion section may also stabilize the duplex in

the desorption/ionization step.

137

The bond dissociation energy of the gas phase duplex may be small due to the
charge distribution as discussed previously. In order to change the situation and
investigate this aspect, the charges on the phosphate groups need to be eliminated. To do
so, the phosphates on the oligonucleotides could be alkylated through a tributylstannyl
phosphate intermediate [2-3]. While this synthesis has been used for small
oligonucleotides containing only 2 nucleotides, the synthesis was not successfirl for larger
oligonucleotides with 12 nucleotides. While the derivitization experiment was not
successful to neutralize the phosphates, conclusions can be drawn from the PNA duplex
spectrum shown previously in Chapter 1. In PNA, there are the same nitrogen bases
contained in DNA, but the phosphate backbone is replaced with a peptide backbone.

This peptide backbone does not carry negative charges. For this reason, comparisons can
be drawn from the PNA data and the DNA data. Both DNA and PNA duplexes
dissociate in the MALDI experiment. This means the presence of negative charges on
the phosphate backbone of DNA does not cause duplex dissociation. 1

It is possible that all desorbed double-stranded DNA ions have sufficient energy
to break the non-covalent bonds holding together the strands even if the bond dissociation
energy has been increased. If this is true, excited double-stranded ions, (Mle-H)", can
be cooled by collisions in the gas phase to prevent their dissociation. When (M.Mz-H)’*
collides with a neutral molecule, the excited ion can transfer energy before it dissociates.
It has been proposed that many collisions occur for desorbed ions, explaining the fact that
matrix and analyte ions form with similar velocity distributions [4]. Thus, collisional

processes after desorption may already occur. It has been suggested that fucose

138

influences MALDI spectra because the fucose molecule thermally degrades into small
molecules (C02, H2O), collisionally cooling the desorbed ions.

In order to cool the analyte ions, collisions must take place very early in the
process, before the accelerating field is applied and before dissociation occurs. Any
approach that may create a higher pressure just above the MALDI target was considered.
For example, NaCO3 thermally decomposes to form CO2 and Na2O. If a MALDI target
is coated with a layer of transparent material such as NaCO3, ions may experience a
larger number of collisions prior to acceleration and detection. When this experiment
was performed, there was no increase in the values of the DSRR and higher laser powers
were needed to detect the single-stranded species. Glycerol was also added to the
MALDI sample plate, in close proximity to the solid being irradiated. Glycerol has a
very low vapor pressure. The glycerol molecules constantly evaporate during the course
of the experiment and allow the analyte ions to collisionally de-excite. Again, the
presence of glycerol on the plate did not result in any increase in the DSRR value.

While this method could provide a way for molecules to collisionally de-excite,
potential problems could arise due to collision induced dissociation (CID). With CID,
the desorbed analyte ions moving at a high velocity would collide with a neutral
molecule. This collision would then create excited analyte ions that may then dissociate.
While CID is a possibility, this does not appear to be taking place. When examining the
data for the fucose experiment, Figure 2.26, there are no ions detected representing

fragments of the duplex or the single-stranded oligonucleotides.

139

References

1. Karas, M.; Glilckmann, M.; Schafer, J. Ionization in matrix-assisted laser
desorption/ionization: singly charged molecular ions are the lucky survivors. J.
Mass Spectrom. 2000, 35, 1-12.

2. Ohuchi, S.; Takashi, 1.; Tsujiaki, H. o-Alkylation of dialkyl phosphates and 0-
alkyl phosphonates via the stannyl intermediates. Nucleosides and Nucleotides.
1992, 11, 749-757.

3. Ayukawa, H.; Ohuchi, S.; Ishikawa, M.; Hata, T. Alkylation of phosphate group
of nucleotides via tributylstannyl phosphate derivatives. Chem. Lett. 1995, 81.

4. Juhasz, P.; Vestal, M.L.; Martin, S.A. On the initial velocity of ions generated by
matrix-assisted laser desorption ionization and its effect on the calibration of
delayed extraction time-of-flight mass spectra. J. Am. Soc. Mass Spectrom. 1997,
8, 209-217.

140

VII. Conclusions

Since the non-covalent interactions that hold the complexes together are disrupted
during the MALDI experiment, each phase of the MALDI process was examined. The
results of these experiments have been published and a reprint of the article is included in
Appendix C. The various steps of the experiment were presented and considered as
points in the experiment where the duplex is disrupted. While the dissociation of the
duplex does not occur due to interactions with the metal plate, organic solvents, or matrix
molecules, the separation of the duplex could occur during the liquid phase portion of the
experiment. Dissociation during the liquid phase of the experiment explains the random
dimerization seen in many of the experiment presented here. It is possible that as the
crystal surface grows, the analyte may interact with the surface and dissociate. While
additives may stabilize the duplex oligonucleotides in the MALDI experiment, the
complex still dissociates. When HPA is used as the matrix, random dimerization is
consistently seen. The double-stranded DNA may have a higher affinity for the growing
HPA crystals.

Although the analyte is not completely stabilized in the MALDI experiment,
information can be learned from the partial dissociation of non-covalent complexes. This
is particularly true when analyzing multi-subunit proteins such as cytochrome c oxidase.

The mass spectrometric analysis of cytochrome c oxidase will be discussed in Chapter 3.

141

CHAPTER 3. Mass Spectrometric Analysis of the Protein Subunit, Lipid, and
Heme Components of Cytochrome c Oxidase from R. Sphaeroides and the
Stabilization of Non-covalent Complexes from the Enzyme

I. Introduction

Information provided from the DNA studies can offer insight into the MALDI
analysis of other types of non-covalent complexes. As mentioned previously, cellular
function is often determined by processes that involve non-covalent interactions, such as
enzyme with substrate, receptor with hormone, and antibody with antigen [1]. In addition
to transitory interactions, enzymes often exist as non-covalent complexes of several
protein subunits. Cytochrome c oxidase, the focus of this work, is such an enzyme, with
numerous subunits ranging from 4 to 13 depending on the source of the protein.

Cytochrome c oxidase is a multisubunit membrane protein that acts as the
terminal complex of the respiratory chain in both mitochondria and bacteria. The enzyme
from Rhodobacter sphaeroides consists of four protein subunits. During the catalytic
cycle, electrons are transferred from cytochrome c, via a transitory interaction, through
two copper centers and two heme groups, to reduce oxygen to water. In addition to the
hemes, metals, and protein subunits, six phospholipids have been resolved in the crystal
structure of the bacterial enzyme. Based on the electron density, they were determined to
be phosphatidylethanolamines [2]. While lipids are not always included in discussions of
the enzyme’s structure and function, it is clear from the crystal structure that these
components play an integral role. Over 30% of all the proteins encoded by the human

genome are membrane proteins and many are important drug targets, yet few have been

142

crystallized. The content and influence of lipids on the properties of these proteins is
becoming recognized as an important issue [3—4].

As with all intrinsic membrane proteins, the subunits of cytochrome c oxidase
have hydrophobic domains. This enzyme presents a challenge to mass spectrometry not
only because of its hydrophobicity, but because it exists as a non-covalent complex of the
subunits associated with lipids and hemes. While positive-ion UV MALDI MS analysis
of this analyte may be expected to yield one peak representing the protonated complex at
approximately 130,000 Da, the complex dissociates in the MALDI experiment with only
some of the protein subunits detected. How and why the complex dissociates is not clear.
However, if all the components can be detected, MALDI can prove to be a useful
analytical tool. Ideally, a MALDI matrix would allow for the optimal detection of all the
components, but in this instance it is not the case. Here we demonstrate that the use of
different matrices and matrix additives can allow for MALDI detection of all components
of the complex, as well as stabilize the intact cytochrome c oxidase and limit the extent of
dissociation. The results provide new information on the composition of the enzyme and

connectivity of components parts.

143

References

l.

Pramanik, B.N.; Bartner, P.L.; Mirza, U.A.; Liu, Y.H.; Ganguly, A.K.
Electrospray ionization mass spectrometry for the study of non-covalent
complexes: an emerging technology. J. Mass Spectrom. 1998, 33, 911-920.

Svensson-Ek, M.; Abramson, J .; Larsson, G.; Témroth, S.; Brzezinki, P.; Iwata,
S. The X-ray crystal structures of wild-type and EQ(I-286) mutant cytochrome c
oxidases from Rhodobacter sphaeroides. J. Mol. Biol. 2002, 321, 329-339.

Camara-Artigas, A.; Brune, D.; Allen, J .P. Interactions between lipids and
bacterial reaction centers determined by protein crystallography. Proc. Natl.
Acad. Sci. USA. 2002, 99, 11055-11060.

Garavito, R.M.; Ferguson-Miller, S. Detergents as tools in membrane
biochemistry. J. Biol. Chem. 2001, 276, 32403-32406.

144

11. Structure of Cytochrome c Oxidase

Enzymes such as cytochrome c oxidase often exist as a non-covalent complex of
several protein subunits. Cytochrome c oxidase is the terminal complex of the respiratory
chain in both mitochondria and bacteria and plays a crucial role in the energy transfer
process. This enzyme is the last step in the electron transfer chain to reduce molecular
oxygen to water. Cytochrome c oxidase is a protein complex that is integrated into the
mitochondrial membrane of eukaryotes and into the cytoplasmic membrane of several

species of bacteria, Figure 3.1.

 

7 Outside of car
W

    

 

museums}: 1'

Figme 3.1. Schematic ofa cell membrane. This figure was reproduced from reference 1.

Cytochrome c oxidase fi'om Rhodobacter sphaeroides consists of four protein
subunits. Cytochrome c oxidase belongs to the superfamily of heme-copper oxidases.
Both mammalian and R. sphaeroides cytochrome c oxidase contain four reactive metal

centers including heme a, heme a3, Gulf, and C113 as well as a non-reactive calcium ion.

145

Electron transfer in the enzyme is initiated when cytochrome c binds to subunit II on the
external side of the membrane. During the catalytic cycle, electrons are transferred from
the protein cytochrome c through the cytochrome c oxidase via two copper ions and two
heme groups. The interaction of subunit II and the cytochrome c protein is demonstrated

in Figure 3.2.

 

Figure 3.2. Structure of cytochrome c oxidase from R. sphaeroides. This figure was
reproduced from reference 2.

The cytochrome c oxidase fi'om R. sphaeroides can serve as a model for
mammalian cytochrome c oxidase and will be used throughout this discussion. The
structure of subunit I is the largest and most conserved subunit of the cytochrome c

oxidase enzyme. Subunit I contains both hemes, the magnesium, and the calcium. The
1 46

magnesium site is located at the interface of subunit I with subunit 11. While the
magnesium may not have a specific function, it is important in stabilization of the whole
protein complex.

Subunit II is composed of two parts. On the nitrogen terminal side there is a
hydophobic domain that spans the membrane. There is also a carbon terminal side that is
hydrophillic in nature. It is this domain that possesses the ligands that bind to the two
copper ions as well as the carboxyl residues that bind to cytochrome c. Subunit H
interacts strongly with subunit 1 and is bound partly by the stabilizing magnesium ion.

Subunit III has many transmembrane helices. Again, the function of this subunit
is not fully understood, but it is believed to maintain the functional integrity of the
enzyme complex. Little is know about subunit IV as well. It was recently discovered in
a high-resolution crystal structure. Subunit IV has no homology with any of the 13
subunits of the bovine cytochrome c oxidase.

In addition to the hemes, metals, and proteins, six phospholipids have been
resolved in the crystal structure of the enzyme. Based on the electron density, four of the
lipids were determined to be phosphatidylethanolamines [3]. While hemes and lipids are
not always included in discussions of the enzyme, these components play an integral role
in establishing its structure. Membrane proteins are found in cellular membranes
composed of lipids. Lipids such as phosphatidylcholine, phosphatidylethanolamine,
phosphatidyl glycerol, and cardiolipin determine the biophysical and biochemical
properties of the membrane proteins [4]. As shown in Figure 3.3, lipids are highly

affiliated with subunit four of the cytochrome c oxidase.

147

 

Figure 3.3. The X-ray crystal structure of cytochrome c oxidase. Phospholipids are
shown in red, subunit I is shown in dark blue, subunit II is shown in light blue, subunit III
is shown in green and subunit IV is shown in black. This figure was reproduced fi'om
reference 2.

In the crystal structure of cytochrome c oxidase form R. sphaeroides, six

phospholipids have been identified. Four of these phospholipids have been identified as

 

,L [L " ’, ' " ‘ ' and the other two lipids have not been identified due to poor
resolution. Two of the phospholipids are buried in a v-shaped cleft in subunit III with
one interacting with subunit 1 as well. The rest of the lipid molecules are found at the
interface between subunit IV and subunits I/III. These lipids are responsible for

separating subunit IV almost completely fiom the rest of the enzyme.

148

The lipids believed to be associated with cytochrome c oxidase are rather diverse.
Glycerophospholipids are the most commonly found membrane lipids. Glycerol serves
as the backbone for these lipids with two acyl groups attached to the glycerol as well as
the phosphate group. The phosphate of these lipids is often linked to hydrophilic groups
such as choline (phosphatidylcholine), ethanolamine (phosphatidylethanolamine), inositol
(phosphatidylinositol), serine (phosphatidylserine), and glycerol (phosphatidylglycerol).

The structures of the most common phospholipids found in the bacteria are shown in

Figure 3.4.
o
(A) HO—CHz—(fH—CHz—O—Ll—O—CHZ ii
OH OH Hc—o—c—R1
H2C—O—fi—R2
o
(B) HzN—CH—CHz-O—llJ—O—CHZ 0
COOH 0H Hc—o—izl—R1
H2C—0—fi—R2
o
o
(C) H2N—CH2—CH2—o—iDI—o—CH2 0
OH H —o—i‘l—R1
HZC—O—fi—Rz
CH3
(D) Hac—N"—c1-12—CHz—o—P—o—CH2 ii
H3 OH Hc—o—c—R1

H2 —O—fi_R2

Figure 3.4. Structure of phospholipids found in R. sphaeroides bacteria. (A)
Phosphatidyl glycerol, (B) Phosphatidyl serine, (C) Phosphatidyl ethanolamine, (D)
Phosphatidyl choline. Rland R2 represent the carbons chains of the fatty acids.

149

The membranes of R. sphaeroides bacteria are also known to contain a variety of
glycolipids. A glycolipids contains a saccharide group in place of the phosphate group.
The glycerol backbone is still present with two fatty acid groups attached. The
glycolipids known to exist in this bacterium include monogalactosyl diacylglycerol and
digalactosyl diacylglycerol. It is also possible for the glycolipids to possess fimctional
groups stemming from the saccharide. Sulfolipids such as sulfoquinovosyl

diacyl glycerol contains a sulfate on the sugar group. The structures of three glycolipids

are shown in Figure 3.5.

 

CHZOH
CH so H
II
0H _ _ —R
HC|3 o c 1 OH
Hzc—o—le—R2 ”0 —CH2 ('3'
H
O OH HC—O—C—R1
(A) (B)
Hzc—O—fi—Rz
CHZOH
HO 0 0 CH2
- I
OH Hc—o—c—R1
0” 11.2,c—o—c—R2
(C) H ll

0

Figure 3.5. Structure of the glycolipids found in R. sphaeroides bacteria. (A)
Monogalatosyl diacylglycerol, (B) Sulfoquinovosyl diacylglycerol, (C) Digalactosyl
diacylglycerol. Rland R2 represent the carbons chains of the fatty acids.

Another lipid that is of particular interest is cardiolipin. Compared to the

structure of the lipids shown previously, cardiolipin is rather unique, Figure 3.6.

Cardiolipin is structurally similar to phosphatidyl glycerol. Figure 3.4 (D). However,

150

cardiolipin has another phosphatidyl glycerol lipid attached to the primary glycerol
backbone. This gives a cardiolipin molecule three glycerol backbones and four fatty acid ,
group. The cardiolipin molecule also has a molecular weight of greater than 1000 g/mol,

much higher than other lipids in the bacteria.

0 O

I I | I

Ra—C—O—CH OH OH OH Hc—o—c—R1
R4—fi—O—CH2 Hzc—O—fi—Rz
O 0

Figure 3.6. Structure of cardiolipin molecule. R1, R2, R3, and R4 represent the carbons
chains of the fatty acids.

As with many membrane proteins, cytochrome c oxidase has hydrophobic subunits. This
enzyme presents a challenge to mass spectrometry not only due to its hydrophobicity but
because it exists as a non-covalent complex of the four protein subunits, lipids, and
hemes. While positive-ion UV MALDI MS analysis of this analyte may be expected to
yield one peak representing the protonated complex at 130,000 Da, the complex
dissociates completely during the desorption/ionization process with only some of the
protein subunits detected. How and why the complex dissociates in the MALDI
experiment is not clear. However, if the complex does dissociate and all the components
can be detected, MALDI can prove to be a useful analytical tool for protein chemists.
While, ideally, a MALDI matrix would allow for the optimal detection of all the
components, this is not the case. Here we demonstrate that the use of different matrices
and matrix additives can allow for the detection of all components of the

complex, stabilize the intact cytochrome c oxidase, and limit the extent of the

dissociation to determine information on the connectivity of components.

151

References

1.

Cell Biology Tutorial, University of Arizona, February 4, 2002,
http://www.biology.arizona.edu/cell_bio/problem_sets/membranes/overview.htrnl

Hilmi, Yasmin. Use of mutagensis, novel detergents, and lipid analysis to
optimize condition for achieving high resolution 3-dimensional crystal structure
of Rhodobacter sphaeroides cytochrome c oxidase. Michigan State University,
2002. East Lansing, MI.

Svensson-Ek, M.; Abramson, J .; Larsson, G.; Tbrnroth, S.; Brzezinki, P.; Iwata,
S. The X-ray crystal structures of wild-type and EQ(I-286) mutant cytochrome c
oxidases from Rhodobacter sphaeroides. J. Mol. Biol. 2002, 321 , 329-339.

Camara-Artigas, A.; Brune, D.; Allen, J .P. Interactions between lipids and

bacterial reaction centers determined by protein crystallography. Proc. Natl.
Acad. Sci. USA. 2002, 99, 11055-11060.

152

III. Experimental

The compounds, 2,5-dihydroxybenzoic acid (DHB), 6-aza-2-thiothymine (ATT),
Sinapinic acid (SA), 2',4'-dihydroxyacetophenone (DHAP), fucose, and calcium chloride
were purchased from Sigma-Aldrich (St. Louis, MO). Sucrose was purchased from the
Mallinckrodt Specialty Chemical Company (Paris, KY). The standard lipids used in the
experiments were purchased in chloroform from Avanti Polar Lipids, Inc. (Alabaster,
AL). The decyl maltoside and dodecyl maltoside used as detergents to solubilize the
proteins were purchased from Anatrace, Inc. (Maumee, OH). Dodecyl maltoside is also
referred to as lauryl maltoside (LM). All solvents used were HPLC grade.

Rhodobacter sphaeroides was used as a homologous host for overproduction of
cytochrome c oxidase protein. When plasmids carrying engineered versions of the genes
for cytochrome c oxidase subunits are placed into R. sphaeroides strains that are missing
these genes [1-2], the bacteria produce large quantities of the engineered cytochrome c
oxidase proteins [3]. Addition of six histidines to the end of one of the Subunits [4-5]
allows simple and rapid purification of the protein by nickel affinity chromatography.
The sequences of the four subunits of Rhodobacter sphaeroides are shown in Table 3.1.
The molecular weights of the subunits are included in the table as well.

Rhodobacter sphaeroides strains 25-1, 157, and 167, which overexpress different
versions of cytochrome c oxidase, were grown and cell membranes were prepared as
described previously [2,4]. Cytochrome c oxidase protein from strain 25-1 (sample
C001) was isolated by nickel affinity column chromatography without further

purification as described [4].

153

Table 3.1. Sequence of R. Sphaeroides Cytochrome c Oxidase Subunits

 

Subunfi

Sequence

Molecular Weight

 

Subunit I

MADAAIHGHEHDRRGFFTRWFMSTNHKDIGVLYLF
TGGLVGLISVAFTVYMRMELMAPGVQFMCAEHLES
GLVKGFFQSLWPSAVENCTPNGHLWNVMITGHGIL
MMFFVVIPALFGGFGNYFMPLHIGAPDMAFPRMNN
LSYWLYVAGTSLAVASLFAPGGNGQLGSGIGWVLY
PPLSTSESGYSTDLAIFAVHLSGASSILGAINMIT
TFLNMRAPGMTMHKVPLFAWSIFVTAWLILLALPV
LAGAITMLLTDRNFGTTFFQPSGGGDPVLYQHILW
FFGHPEVYIIVLPAFGIVSHVIATFAKKPIFGYLP
MVYAMVAIGVLGFVVWAHHMYTAGLSLTQQSYFMM
ATMVIAVPTGIKIFSWIATMWGGSIELKTPMLWAL
GFLFLFTVGGVTGIVLSQASVDRYYHDTYYVVAHF
HYVMSLGAVFGIFAGIYFWIGKMSGRQYPEWAGKL
HFWMMFVGANLTFFPQHFLGRQGMPRRYIDYPEAF
ATWNFVSSLGAFLSFASFLFFLGVIFYTLTRGARV
TANNYWNEHADTLEWTLTSPPPEHTFEQLPKREDW
ERAPAH

63147 (I)
63016 (1')
64040 (I'+Htl)
P=I-Ma
Htl = SNH6

 

Subunit H

MRHSTTLTGCATGAAGLLAATAAAAQQQSLEIIGR
PQPGGTGFQPSASPVATQIHWLDGFILVIIAAITI
FVTLLILYAVWRFHEKRNKVPARFTHNSPLEIAWT
IVPIVILVAIGAFSLPVLFNQQEIPEADVTKVTGY
QWYWGYEYPDEEISFESYMIGSPATGGDNRMSPEV
EQQLIEAGYSRDEFLLATDTAMVVPVNKTVVVQVT
GADVIHSWTVPAFGVKQDAVPGRLAQLWFRAEREG
IFFGQCSELCGISHAYMPITVKVVSEEAYAAWLEQ
ARGGTYELSSVLPATPAGVSVE

 

 

32931 (II)

29122 (IIA)

29340 (IIB+Ht2)

30661 (IIC)

IIA = II — [1-25] -
- [288-303]

IIB = IIA -

[282-287]
nc=n-n2n
th = H5

 

Subunit HI

Subunit IV

MAHAKNHDYHILPPSIWPFMASVGAFVMLFGAVLW
MHGSGPWMGLIGLVVVLYTMFGWWSDVVTESLEGD
HTPVVRLGLRWGFILFIMSEVMFFSAWFWSFFKHA
LYPMGPESPIIDGIFPPEGIITFDPWHLPLINTLI
LLCSGCAATWAHHALVHENNRRDVAWGLALAIALG
ALFTVFQAYEYSHAAFGFAGNIYGANFFMATGFHG
FHVIVGTIFLLVCLIRVQRGHFTPEKHVGFEAAIW
YWHFVDVVWLFLFASIYIWGQ
MAETNKGTGPMADHSHPAHGHVAGSMDITQQEKTF
AGFVRMVTWAAVVIVAALIFLALANA

 

30172 (111)
30041 (111')
111' = 111 — Met

6390 (IV)

6259 (IV')

5403 (IVA)

5272 (IVA')

IV' = IV — Met
IVA = IV — [1-10]
IVA' = IVA — Met

 

154

For further purification of cytochrome c oxidase from Rhodobacter strain 157
(sample C002), the protein concentration was first determined using the Pierce
(Rockford, IL) BCA protein assay kit, modified to include 0.25% deoxycholate in the
buffers. The membrane sample was then diluted to a final protein concentration of 10
mg/mL. Lauryl maltoside was then added to the membrane sample to a final
concentration of 1%. The solution was stirred in the cold for 15-20 minutes before
ultracentrifirgation at 200,000 g for 30 minutes. The supernatant was loaded onto a home-
packed QIAGEN (Valencia, CA) Ni-NTA superflow column attached to a fast
performance liquid chromatography (F PLC) system. After washing the sample with 13
column volumes of buffer A (10 mM Tris, 40 mM KCl, 10 mM imidazole, 0.05% LM,
pH=8.0) , the enzyme was eluted with a linear gradient from 0% to 100% buffer B
(lOmM Tris, 40 mM KCl, 150 mM imidazole, 0.05% LM, pH=8.0) over 15 column
volumes. After the first column, green fractions were pooled, concentrated, and washed
using a Millipore (Billerica, MA) Amicon Centriplus YM-lOO concentrator.

For further purification, the sample was loaded onto a Phannacia (Peapack, NJ)
MonoQ column connected to an FPLC system. After loading the sample and washing
with 2 column volumes of buffer A (10 mM Tris, 0.05% LM, pH=8.0), the enzyme was
eluted with a linear gradient fi'om 0% to 100% buffer B (10 mM Tris, 0.05% LM and 500
mM KCl, pH=8.0) over 20 column volumes. Often there were two peaks in the
chromatogram as well as an elongated tail following the second peak. In this case, only
fractions from the second peak were collected. The active fractions were pooled, washed

and concentrated using Centriplus YM-100. Approximately 50 mL of 10 mM Tris, 0.24%

155

decyl maltoside at a pH of 8.0 was used to wash the enzyme to ensure the buffer was
completely exchanged.

For further purification of cytochrome c oxidase from Rhodobacter strain 167
(sample Cc03), the same purification methods were applied except that in the nickel
column purification step, buffer A contained 10mM Tris, 220mM KCl, lmM imidazole,
pH 8.0 and 0.05% dodecyl maltoside, and buffer B contained 10mM Tris, 220mM KCl,
150mM imidazole, pH 8.0 and 0.05% dodecyl maltoside .

Saturated solutions of the MALDI matrices ATT, SA, and DHAP were made
using a 1:1 acetonitrile/water solution. The DHB solution was made at a concentration of
25 mM in 1:1 acetonitrile/water solution. Stock solutions of fucose, sucrose, and calcium
chloride were made at concentrations of 25 mM in water. When additives such as
sucrose were used in the MALDI experiment, a volume of the additive solution was
added to an equal volume of the protein solution before being deposited on the MALDI
target.

The standard lipid solutions were diluted with chloroform to concentrations of 1
pmol/uL. In order to analyze the standard lipid solutions and the lipid extracts from the
enzyme, 1 uL of DHB was spotted on the plate and allowed to dry. A microliter of the
lipid solution was then deposited, since the two solvent systems are incompatible.

Linear MALDI mass spectra were recorded on a PerSeptive Biosystems
(Framingham, MA) Voyager delayed extraction (DE) time-of-flight linear (TOF) mass
spectrometer equipped with a nitrogen laser (337 nm, 3 ns pulse). For the positive ion
MALDI spectra reported here, the accelerating voltage was 20 kV, the delay time,

selected for optimum resolution, was 450 ns, the grid voltage was 94.0 % of the

156

accelerating voltage, and the magnitude of the guide wire voltage was 0.30% of the
accelerating voltage. Transients from 50 laser shots were averaged for each spectrum.
Reflectron and post source decay (PSD) MALDI mass spectra were recorded on a
PerSeptive Biosystems (Framingham, MA) Voyager-DE STR. Biospectrometry
workstation equipped with a reflectron TOF mass spectrometer. For these experiments,
the accelerating voltage was 20 kV, the grid voltage was 80% of the accelerating voltage,
and the delay time was 100 ns. The mirror ratio for the first PSD stitch spectrum was
1.00 and the mirror decrement ratio was set at 0.90. Ten stitches were acquired for every
PSD spectrum. Typically, transients from 50 laser shots were averaged for each stitch.

FTMS spectra were acquired on a Bruker Daltonics (Billerica, MA) Apex III
mass spectrometer with a 7 Tesla superconducting shielded magnet. Argon was used as
the collision gas for the experiments. Samples were introduced into the electrospray ion
source using a syringe pump at a flow rate of 1 uL/min.

Images in this dissertation are presented in color.

157

References

1. Shapleigh, J .P.; Gennis, R.B. Cloning, sequencing, and deletion from the
chromosome of the gene encoding subunit I of the aa3-type cytochrome c oxidase
of Rhodobacter sphaeroides. M01. Micro. 1992, 6, 635-642.

2. Zhen, Y.; Qian, J .; Follmann, K.; Hayward, T.; Nilsson, T.; Dahn, M; Hilmi, Y.;
Hamer, A.G.; Hosler, J .; Ferguson-Miller, S. Overexpression and purification of
cytochrome c oxidase from Rhodobacter sphaeroides. Protein Expr. Purif 1998,
13, 326-336.

3. Hosler, J .P.; Fetter, J .; Tecklenburg, M.M.J.; Espe, M.; Lerma, C.; Ferguson-
Miller, S. Cytochrome aa3 of Rhodobacter sphaeroides as a model for

mitochondrial cytochrome c oxidase. Purification, kinetics, proton pumping, and
spectral analysis. J. Biol. Chem. 1992, 267, 24264-24272.

4. Hiser, C.; Mills, D.A.; Schall, M.; Ferguson-Miller, S. C-terminal truncation and
histidine-tagging of cytochrome c oxidase subunit II reveals the native processing
site, shows involvement of the c-terminus in cytochrome c binding, and improves
the assay for proton pumping. Biochemistry 2001, 40, 1606-1615.

5. Mitchell, D.M.; Gennis, R.B. Rapid purification of wildtype and mutant

cytochrome c oxidase from Rhodobacter sphaeroides by Niz+-NTA affinity
chromatography. FEBS Lett. 1995, 368, 148-150.

158

III. Results and Discussion

Cytochrome 0 oxidase from R. sphaeroides contains four protein subunits. The
sequences of these subunits are listed in Table 3.1. These four subunits have different
sequences because of differing sites of processing and the addition of Histidine tags.
These changes in sequence are also listed in Table 3.1. The cytochrome 0 oxidase
enzyme may contain different versions of the subunits shown in Table 3.1. In this work,
three versions of cytochrome 0 oxidase were used. Sample C001 has a histidine-tagged
subunit I (I'+Ht1), a subunit II with an artificially-processed C-terrninus (IIA), a native
subunit H1 (111'), and two naturally occurring forms of subunit TV (IV' and IV A').
Sample C002 is the same as C001 but with overexpressed, artificially-shortened version
of subunit IV (IVA') and a subunit II with 25 amino acids removed from the N-terminus
(IIC). Sample C003 has a native subunit I (I'), an artificially-processed and histidine-
tagged subunit II (IIB+Ht2), a native subunit H1 (111'), and an overexpressed, engineered
longer version of subunit IV (IV'). These three versions are indicated in Table 3.2.

Table 3.2. Cytochrome c Oxidase Samples Used in this Work

 

Oxidase Subunit I Subunit H Subunit III Subunit IV

 

sample

C001 1' + Htl IIA III' IV' and IVA'
C002 1' + Htl IIC III' IV A'

C003 1' IIB + Ht2 III' IV ’

 

When analyzing large proteins in MALDI MS, Sinapinic acid (SA) is commonly
used as a matrix. When Ni-column purified C001 (from a solution which also contains a
solubilizing detergent) is analyzed using MALDI MS with SA, the complex completely

dissociates and the individual subunits I, II, and III are detected with varying efficiency,

159

Figure 3.7. Clearly, singly and doubly charged forms of a component with a molecular
weight of around 64,000 g/mol are present and two additional components with similar
molecular weights of approximately 30,000 g/mol can be resolved. The expected masses
for the subunits are shown in Table 3.1. The lipids that are assumed to be present have
molecular weights in the 700-1500 g/mol range, but are not detected due to a high
baseline in the low mass region of the spectrum. The smallest of the subunits, subunit

IV, has a molecular mass of approximately 5300 Da and is not detected in the MALDI

1 mm“

(III+H)+

 

Relative Intensity

 

 

I—l (II+H)+ (1+2H)2+

 

 

 

H 1 7 I 1 l I I I

10000 20000 30000 40000 50000 60000 70000 80000 90000 100000
mlz

Figure 3.7. Positive-ion linear MALDI-TOF spectrum of cytochrome 0 oxidase sample
C001 using SA as the matrix.

experiment. Also, no intact enzyme is detected. Reflectron capabilities were available
and reflectron mode was used to increase resolution in the low m/z regions. However,
when analyzing the region shown in Figure 3.7 using a reflectron TOF MS, the peaks

representing subunits I-III could not be detected.

160

In Figure 3.7, the large protein subunits are detected. While this is important, the
detection of the other components is essential as well. A method is needed to monitor
enzyme content during purification, to allow for the detection of all parts of the
cytochrome c oxidase complex. Ideally, all of the necessary information could be
acquired in one experiment, but this is not always possible. In these experiments, the
challenge was divided into three mass ranges: 500-2000 Da (lipid region), 5-10 kDa
(subunit IV), and 20-70 kDa (subunit I-III). Enhanced detection of the complete non-
covalent complex was also a goal (>100 kDa). Each mass range was considered and the
problems were approached in a stepwise manner in order to develop a method for the
analysis of the entire enzyme. To enhance the various m/z regions of the spectrum as
well as to preserve the intact enzyme, several matrices and additives were evaluated and

developed.

Lipid region (SOD-2.000 Da)

When analyzing membrane enzymes such as cytochrome c oxidase, the protein
subunits are often the primary focus [1-2]. However, part of a complete analysis is the
characterization of the lipids present. Lipids play a significant role in the structure of
membrane proteins, as revealed in several recently published high resolution X-ray
crystal structures [3-5]. The lipids may be critical in stabilizing a homogeneous
conformation at the molecular level [6], and thus should be formally considered as part of
the enzyme.

Many types of lipids have been previously studied using a variety of mass

spectrometry techniques [7]. Phospholipids isolated from biological membranes have

161

been studied using both ESI and MALDI MS [8]. Lipid mixtures have also been
analyzed in MALDI MS, although at levels above those commonly used for analytes [9-
10]. Also, structural information for phospholipids has been determined using MALDI or
ESI coupled to a Fourier transform ion cyclotron resonance mass spectrometer (FT-ICR
MS) [1 1].

When analyzing the components of cytochrome c oxidase with molecular masses
under 2000 Da, analysis is complicated not only by the complex mixture of lipids present,
but also by the presence of the detergent. Throughout these MS studies, the solubilizing
detergent used was lauryl maltoside. This detergent has a molecular weight of 510 g/mol
and appears in the spectra as a protonated molecule at 511 Da, a sodium ion adduct,
(M+Na)+, at 533 Da, or a potassium ion adduct, (M+K)+, at 549 Da. When the
concentration of the detergent is kept below a value of 1 mM, little interference occurs.
At higher concentrations, signal suppression results in both MALDI and ESI.
Electrospray is particularly sensitive, even to non-ionic detergents such as lauryl
maltoside. When ESI was used to desorb/ionize cytochrome c oxidase samples, detergent
molecules formed singly-charged dimer and trimer ions yielding peaks with m/z values as
high as 1531 Da. A high concentration of detergent often resulted in the suppression of
lipid peaks, making the analysis of the cytochrome c oxidase samples by ESI very
difficult.

In a MALDI experiment, no peaks representing the lipids were detected using SA
as the matrix due to the high baseline at low m/z values. This, in part, could be due to the
higher laser powers necessary to detect the large protein subunits. In order to enhance the

lipid region of the spectrum, the matrix can be changed and reflectron mode can be used.

162

When the matrix DHB is used for the samples, the matrix effects are decreased allowing
the peaks representing the lipids to be detected, Figure 3.8. The cytochrome c oxidase
used in Figure 3.8 was the same Ni-column purified sample used in Figure 3.7. The
labels shown in Figure 3.8 indicate the head group of the lipid, the total number of

carbons in the fatty acid chains, and the number of sites of unsaturation in the alkyl

 

 

 

 

 

 

   
 

groups.
I
l SL
, [3453
I PG+Na” SL
3‘ [34:2] PS[36:3] [34;01
.5 Pcrutzl
g 130
s [36:01 SL[36:0]
.g Psp2a] Hm“°a
g PC[32:2] MGD
O: [34.2] 1' [36.2]
PE *+Na* 1
[34:1] (:1 l
l l

650 700 750 800 850 900 950
mlz

Figure 3.8. Lipid region of the postive-ion reflectron MALDI-TOF spectrum of
cytochrome c oxidase sample C001 using DHB as the matrix.
MGD(monogalactosyldiacylglycerol), PG(phosphatidylcholine),
PE(phosphatidylethanolamine), PG(phosphatidylglycerol), PS(phosphatidylserine),
SL(sulfolipid).

For example, the peak in Figure 3.8 at m/z 819 Da is labeled SL[34:2]. That is, the lipid
is a sulfolipid with two fatty acid groups that contain a total of 34 carbons with 2 sites of

unsaturation. There are a variety of fatty acid chain combinations that could give this

163

very result including SL(18:0, 16:2), SL(l8z2, 16:0), and SL(18:1, 16:1). For simplicity,
the peak is labeled as the sum of the fatty acid chain carbons. Several peaks have been
labeled in Figure 3.8. At 716 Da, there is a peak representing a PE with 34 carbons and
two sites of unsaturation. There is also a peak 2 Da higher representing a PE with only
one site of unsaturation. In addition to phospholipids, there are also sugar-containing
lipids in the spectrum. At 783 Da, there is a peak representing a monogalactosyl
diacylglycerol (MGD). Also, 22 mass units higher at 805 Da, there is a peak representing
the sodiated MGD[36:2]. At 819 Da and 823 Da, there are peaks due to the presence of
sulfolipids. These two peaks represent sulfolipids with the same number of carbons in
the fatty acid. They differ only in the number of sites of unsaturation. The sulfolipid
yielding a peak at 819 Da has two sites of unsaturation while the sulfolipid that yields a
peak at 823 Da has no sites of unsaturation.

While many peaks in Figure 3.8 could be assigned to a specific type of lipid based
on m/z values alone, several peaks could not confidently be identified. The peak detected
at 851.5 Da could represent heme a with a nominal mass of 852 Da or a sulfolipid with
saturated fatty acids that contain 36 carbons that would have a nominal mass of 851 Da.
Peaks that correlated to phosphatidylcholines and phosphatidylserines were difficult to
assign due to the similarity in the mass of the head group. The peak at 730 Da could be
from a PS lipid with fatty acid combination [32:3] or the PC lipid [32:2]. At 786 Da,
there is also the possibility of PC[36:2] or PS[36:3].

While determining the masses of the lipids and the types of lipids present is
important in mass spectrometric analysis, structural information is needed as well. When

the lipid components were detected, a peak was observed at m/z 786. After performing

164

PSD analysis on this peak, a fragment ion was produced at m/z 184. This peak
corresponds to the head group of a PC lipid. If the peak at m/z 786 represented a PS
lipid, the fragment ion would have an expected mass of 186 Da. In order to further
confirm the structure of the peak at 786, a standard solution of phosphatidylcholine (18:1,
18:1) was analyzed. When post-source decay (PSD) is performed on the peak at 786, a
fragment ion at m/z 184 was detected. The PSD spectra for the m/z 786 from the

standard and the cytochrome c oxidase sample are shown in Figure 3.9.

(A) C17H33-COO-CH 786

|
C17H33'COO‘CH

l
l I 184 r—87 59
+
i CHLPOJ'CHz'CHle(CH3)3

 

 

Relative lntensuty
P
F
1
t
E
L
t I

 

 

 

l

l

100 200 300 400 500 600 700 800
mlz

Figure 3.9. Positive-ion PSD MALDI-TOF spectra of the m/z 786 peak of (A) PC
(18:1,18zl) standard and (B) cytochrome 0 oxidase sample C001. DHB was used as the
matrix.

From the mass of the head group, the mass of the precursor ion, and the information

about the types of fatty acids found in Rhodobacter sphaeroides [12], the fatty acid

composition was assumed to be 18:1, 18:1. Small additional peaks in Figure 3.9 (B) are

165

likely PSD fragments from precursor ions close to m/z 786 since the MALDI-TOF-PSD
CXperiment does not provide unit resolution for precursor selection.

While the lipids are important for stabilization of the enzyme, the purification
processes performed to isolate a membrane protein often strip away many of the lipids. It
then becomes important to understand the types of lipids present at various stages of
purification. When the protein is first solublized from a membrane preparation in these
experiments, there are 70 phosphate-containing lipids per average protein molecule and
only 16-23 and 8-11 phospholipids/oxidase molecule after Ni2+-NTA purification and
subsequent MonoQ column purification respectively [13]. These results suggest that 10-
15 lipids may be specifically affiliated with the oxidase. An analytical method that
allows for the detection of the lipid and protein components provides important
information from the purification standpoint and is a key for standardizing procedures for
crystallography [6]. Further information to assist in understanding the crystal growth
process can also be determined by dissolving and analyzing the crystals grown for X-ray
analysis. Since a variety of lipids can be analyzed in the MALDI experiment, X-ray
crystallographers have a rapid means to analyze their enzymes after the various phases of
purification. Often lipids are analyzed using thin-layer chromatography (TLC). This
provides information about the classes of lipids present, but does not provide any specific
information about the fatty acids chains. The MALDI experiment yields more molecular
information including the masses of the lipids present.

When analyzing cytochrome 0 oxidase solutions after various purification steps,
the lipid content can be determined from the MALDI mass spectra. The mass spectra of

cytochrome c oxidase solutions from the various levels of purification contain fewer lipid

166

peaks than the solutions directly isolated from the membrane without any purification
procedures, Figure 3.10. In Figure 3.10, the cytochrome c oxidase solution was analyzed
after (A) initial isolation, (B) Ni-column purification, and (C) both MonoQ and Ni-

‘ (A) MGD SL[3420]

[36:2]

+

+Na

 

 

 

3‘ SL[34:2]
'Z’ PGtNa‘ SL[36:0]
32 [3432] Heme a
0
.2
E
0
n:
(C)
PC[36:2]
PS[36:3]
l
650 700 750 800 850 900 950 1000
mlz

Figure 3.10. Lipid region of positive-ion reflectron MALDI-TOF spectra for different
stages of cytochrome c oxidase (A) unpurified, (B) Ni-column purified, and (C) Mono-Q
purified.

column purification. Many conclusions can be drawn from this experiment.

The MGD[36:2] decreases in relative intensity with every step of purification. While the
relative peak height of the peaks representing SL[34:2] and SL[34:0] changes in the first
two phases of isolation and purification, the relative peak intensity of the SL decreases

after the MonoQ purification, Figure 3.10 (C). The PG[34:2] seems to remain associated

with the enzyme through both purification processes. The peak at 852 Da, SL[36:0] or

167

heme a, also remains in solution throughout the purification steps. When MALDI
analysis was performed on redissolved crystals from X-ray analysis, the resulting

spectrum indicated that lipids are further lost in the crystal growth process as well.

Hm

The lipids are not the only cofactors present in cytochrome c oxidase. Non-
covalently bound hemes are essential in the transport of electrons and protons. In the
bacterium R. sphaeroides, the type of heme present in cytochrome c oxidase is heme a.
This type of heme contains four different side groups: vinyl, methyl, carboxylic acid, and
famesyl.

Lipid extracts from the various stages of purification of C001 oxidase were
analyzed using TLC. Bands from the TLC plates were excised and extracted with
chloroform to retrieve the lipids [13]. These extracts were again analyzed using MALDI
MS. It was found that extracts from several of the bands yielded a peak at an m/z value
of 852 Da. This peak is also seen in the partial spectrum of the isolated enzyme, Figure
3.8. The PSD spectrum of the peak at 852 Da is shown in Figure 3.11. This is the m/z
region of the experiment where the lipids are found, but the PSD spectrum is not
characteristic of a lipid. The fragment ions are consistent with those expected for heme
a. The structure of heme a is also shown in Figure 4 with the m/z values of the heme
fragments. As shown in Figure 3.8, the peak at 852 was assigned as a sulfolipid or a
heme group. With the information determined from the PSD spectrum, this peak can

now be assigned as representing heme a.

168

[ 852

578

Relative Intensity

 

 

300 400 500 600 700 800 900 1000

 

Ho—E—crgcr-t, 0H2cr-t2 —r|;|—0H

Figure 3.11. PSD MALDI-TOF spectrum of the 852 peak of cytochrome 0 oxidase
sample CcOl. DHB was used as the matrix. The structure of heme a is shown as well.
R1 represents a formyl group and R2 represents a farnesyl group.

169

Subunit IV (5-10 kDa region)

Important information can be determined about the molecular weight of subunit
IV, which was not detected in the initial analysis, Figure 3.7. The full-length form of
subunit IV has a predicted molecular weight of 6390.38 g/mol while a shorter form of the
protein has a predicted molecular weight of 5403.28 g/mol. The sequence of the longer
form of subunit IV is shown in Table 3.1. The underlined portion of the sequence may be
removed by processing, or translation may start at the methionine following the
underlined portion, to produce the shorter form. In this discussion, the shorter version of
subunit TV will be referred to as IVA and the longer version will be referred to as IV.

In Figure 3.7, subunits I-IH are detected at high laser powers with a high
background when SA is used as the matrix. At lower laser power, subunit IV is detected
with poor resolution and the lipids are not detected. When the matrix is changed to DHB,
sensitivity and resolution increase for both the lipids and subunit IV. With the enhanced
detection of subunit IV, much can be learned about the enzyme. When different versions
of cytochrome c oxidase are analyzed, mass spectral analysis can easily identify which
version of subunit IV is present, Figure 3.12. The C003 sample, Figure 3.12(A), from a
strain of R. sphaeroides engineered to produce the longer form of the subunit, contains
the expected subunit IV (6300kDa). The C001 sample analyzed in Figure 3.12(B)
contains subunit IVA and a small amount of subunit IV, which suggests that the shorter
form predominates in the wild-type bacteria which do not overexpress an engineered
form. The C002 sample from a strain engineered to produce the shorter version of the
subunit, shows the expected peak representing subunit IVA(5300 kDa), Figure 3.12(C).

From FigureS, the mass of subunit IVA was found to be 5273 Da and subunit IV was

170

(lV'+H)”

Relative Intensity

    

{(C) (IVA'+H)*

5000 5500 6000 6500 7000
mlz

Figure 3.12. A portion of the positive-ion linear MALDI-TOF spectra of cytochrome c
oxidase sample C003 (A), sample C001 (B), and sample C002 (C), showing subunits
IV' and IVA'. DHB was used as the matrix for each experiment.

found to be 6260 Da. While the sequence in Table 1 shows an N-terrninal methionine on
both subunit IV and IVA, the m/z values from Figure 5 are 131 Da lower than expected.
This is consistent with the loss of the N-terrninal methionine for each version of subunit
IV. The peaks in the spectrum actually represent subunit IV' and IV A' , using the
nomenclature shown in Table 3.1. Several experiments were performed and the average
molecular weight information from these experiments is shown in Table 3.3. N-terminal
amino acid sequencing of both forms of the subunit has also indicated removal of the
starting methionines [14]. Unexpectedly, in Figure 3.12(B) an additional peak 71 Da
lower than the peak representing IVA' is seen. This represents the subunit IV A' with the

additional loss of an alanine at the N terminus.

171

The molecular weight information determined from the MALDI TOF experiment
was confirmed for both subunit IV' and IVA' using ESI-FTICR MS. A sample of C001
cytochrome c oxidase was dissolved in an acetonitrile/formic acid solution at a
concentration of approximately 1 micromolar. This solution was then infused into the
electrospray source at a rate of one microliter/minute. While no peaks representing the
larger subunits were detected, peaks from subunit IV A' were detected in the electrospray
FTICR MS experiment. In order to demonstrate the varying resolution from the MALDI
TOF experiment to the ESI FT ICR MS experiment, spectra from each experiment are
shown in Figure 3.13. The spectrum in Figure 3.13(A) is from the MALDI TOF
experiment shown in Figure 3.13(C). The Figure shown in 3.13(B) is the deconvoluted
ESI-FTMS data and the spectrum shown in Figure 3.13(C) represents the theoretical
isotope distribution for subunit IV A' at a resolution of 30000. The most intense peak in
the theoretical spectrum is 5271.8493 Da. The most intense peak in the experimental
FT MS experiment was 5271.8390, corresponding to an error of 2 ppm. Although the
data are not shown, a peak was detected for subunit IV'. The most intense peak in its
deconvoluted spectrum was 6258.3228 Da while the theoretical mass of the most intense
peak was 6258.2984 Da. This corresponds to an error of less than 4 ppm. The
information from the FTMS spectra confirms the sequences of subunits IV' and IVA'

given in Table 3.1.

172

(IVA'+H)*

 

 

 

 

l (A)

l
g- , A
(D
1.3 (B)
E.
.“2’
E
:13

.. A_A‘j\/\A
(C)

i
l
l

 

 

A

 

j V

5260 5265 5270 5275 5280 5285 5290 5295 5300
mlz

I

Figure 3.13. (A) Positive-ion MALDI-TOF spectrum of subunit IVA' using DHB as the
matrix. (B) Deconvoluted ESI-FTICR spectrum of subunit IVA'. (C) Theoretical
isotope distribution for subunit IVA'.

Subunits I-III (20-70 kDa Region)

As shown in Figure 3.7, the larger subunits of cytochrome c oxidase are detected
with varying efficiency. This is due to, in part, the nature of the enzyme. Proteins such
as cytochrome c oxidase found in cell membranes are hydrophobic and are much more
difficult to handle. Hydrophobic proteins often have a tendency to precipitate or
aggregate in aqueous solution. These proteins are often solubilized and stabilized with a
detergent, if solution experiments on the active form are being performed. However,
such additives lead to complications during mass spectrometric analysis. Sensitivity

decreases in both MALDI MS [15] and ESI MS experiments [16] when detergents are

173

present. While the removal of detergents may improve the sensitivity, this may lead to
problems when analyzing enzymes that require detergents to maintain their higher order
structure in solution.

While the mass spectrometric analysis of membrane proteins proves to be
difficult, success has been reported with a variety of techniques. Previously, 2-(4-
hydroxyphenylazo)benzoic acid (HABA) has been used as a matrix in MALDI
experiments to detect subunit I-III of cytochrome c oxidase from R. sphaeroides [17]. In
the resulting spectrum, the peaks representing subunit H and III are of lower intensity
than the peaks representing the singly- or doubly-charged subunit 1. Sinapinic acid was
used to analyze the bovine version of the enzyme as well, although a spectrum containing
the peaks representing the largest subunits was not reported [18]. Also, partial success
has been reported using electrospray [19-22].

There has been greater success when analyzing other types of hydrophobic
peptides and proteins [23-25]. Previously, MALDI MS has been used to analyze
hydrophobic peptides with the peptide being solubilized in aqueous formic acid prior to
the addition of Sinapinic acid as the matrix [26]. Other approaches have included the use
of chloroform/methanol as the solvent for both the matrix and the peptide [24]. Analysis
of hydrophobic peptides and proteins is still difficult, possibly due to the low ionization
efficiencies in both MALDI-MS and ESI-MS [25].

While SA as a matrix allows for the detection of subunits I, II, and 111, Figure 3.7,
the matrix, 2',4'-dihydroxyacetophenone (DHAP) allows for the enhanced detection of

subunits II and 111, Figure 3.14. DHAP is a useful matrix in this application, since it

174

 

suppresses the formation of the doubly charged subunit I as well, simplifying the m/z 20-

 

 

 

30 kDa region of the spectrum.
1 (III+H)’
l
l
e l
(D
c l
2 :
E l
0 l
.3
_¢I_1
:2 l
. (Hi-H)“
l (“HY
25000 35000 45000 55000 65000 75000

m/z

Figure 3.14. A portion of the positive ion linear MALDI-TOF spectrum of cytochrome c
oxidase sample C003. DHAP was used as the matrix.

When analyzing high molecular weight compounds using MALDI-TOF MS,
often there are small shifts in the m/z value of a given peak from spectrum to spectrum.
Since molecular weight information was desired for all of the subunits of C001 and
C003 cytochrome c oxidases purified by various methods, the m/z values from several
different MALDI experiments were averaged. The results of these measurements are
listed in Table 3.3. While the standard deviations shown in Table 3.3 are large, the
deviations are still within the mass difference of only 1-2 amino acids. This insures that

there cannot be substantial errors in the accepted sequences.

175

Table 3.3. Measured and Expected Masses of Cytochrome c Oxidase Subunits

 

 

Subunit Sample Measured Standard Expected Subunit
Mass (Da) Deviation Mass (Da) Assignment
Subunit I CcOl 63939.64 1 l 1.21 64039.62 I'+Ht
Subunit I C003 62864.27 89.35 63015.59 1'
Subunit 11 C001 29167.65 15.39 29122.33 IIA
Subunit 11 C003 29359.14 62.05 29339.53 IIB+Ht
Subunit HI C001 30021.57 26.31 30041.14 III'
Subunit HI C003 30140.85 105.07 30041.14 III'
Subunit IV C001 5272.53 0.92 5272.09 IV A'
Subunit IV C001 6259.73 1.52 6259.19 IV '
Subunit IV C003 6260.05 1.87 6259.19 IV '

 

While the masses of subunits I-III cannot accurately be determined using

MALDI-TOF MS, the average molecular weight values shown in Table 3.3 can provide

information about the processing of the cytochrome c oxidase in the bacteria. For

example, the full-length amino acid sequence for subunit I corresponds to a molecular

weight of 63 146.79 g/mol or, with the N-terrninal methionine removed, 63015.59 g/mol.

As prepared for study in specific experiments, subunit I can be engineered to have a His-

tag (Htl), 64170.82 g/mol, and may also have the N-terrninal methionine removed,

64039.62 g/mol (I'+Ht1). The C-terminal His-tag used for subunit I contains 2 amino

acids in addition to six histidines: SNHHHHHH. When examining the results shown in

Table 2, the C001 protein contained the engineered His-tagged protein while the C003

sample contained the naturally processed protein without the presence of a His-tag.

For the analysis of subunit 11, similar results were found. In the C001 sample,

subunit 11 did not contain a His-tag and the portions of the sequence underlined in Table 1

were removed as expected. This was confirmed by the mass spectrometric data that

contained a peak at 29167 Da, close to the expected value of 29122 Da. Also, in the

C003 sample, subunit H was the part of the enzyme engineered with a C-terminal His-tag

176

consisting of only six histidines. For this protein, the portion of the sequence shown in
bold in Table 3.1 was cleaved from the protein, in addition to the underlined portion. The
mass of this protein, IIB+Ht2, was expected to be 29339 Da and this is close to what is
seen in the mass spectrum. 1

When examining the results from the C001 samples, it can be determined that
subunit III was processed with the loss of the N-terminal methionine. The expected mass
for the N-terminally truncated subunit III (III') is 30041 g/mol. This is in agreement with
the data shown in Table 1. Due to the large standard deviation for the results from the
C003 form, no conclusions can be drawn from this data regarding the processing of the

protein, although processing is expected to be identical to that of the C001 sample.

Non-co_vz_110nt complexes

When cytochrome 0 oxidase is purified, lipids are partially removed in the
process. The removal of lipids from the enzyme is known to weaken the complex. In
addition to lipid removal, organic solvents, matrix, and low pH can also weaken the
complex so that dissociation occurs and only the individual subunits are detected in the
MALDI experiment. If dissociation could be eliminated, the complete, intact enzyme
could be detected. An additional possibility, limited dissociation, could also be useful. If
fragments of the complete complex remained bound as non-covalent subcomplexes, the
aspects of connectivity could be realized. For example, when changing the MALDI
matrix from SA to DHAP, the region of the spectrum around the peak for subunit 1
changes, as shown in Figure 3.14. There are peaks detected at 63.7 kDa and 64.5 kDa as

well. It is possible these peaks could represent the addition of a heme or lipid to subunit

177

1. Due to the deviations in these mass measurements, it is impossible to assign these
peaks to subunit I plus a specific lipid or heme. As an example, the peak at 63.7 kDa
could represent the addition of a lipid (2:7 00 Da) to subunit 1’. The peak at 64.5 kDa
could have a number of assignments. First, the addition of two lipids (:700 Da) is likely.
The peak could also represent the addition of 1 cardiolipin molecule (231500 Da) to the
subunit 1'. The peak could represent the heme (858 Da) and a lipid ($700 Da) bound to
subunit 1' as well. In the region of the spectrum around 30 kDa, there are peaks
representing subunit IIA (29.5 kDa) and subunit III' (30 kDa). There are also four peaks
from 30.5-33 kDa. As an example, the peak denoted with an asterisk is at 31 kDa. This
peak could represent the addition of a cardiolipin (21500 Da) to subunit IIA.

In order to decrease the extent of enzyme dissociation, stabilizing compounds can
be added to the cytochrome 0 oxidase solution. The presence of compounds such as
sugars has been found to stabilize proteins in solution. While the exact mechanism of the
stabilization effect remains unknown, it has been found that both sucrose and glucose
strengthen the pairwise interaction between hydrophobic groups in model systems and
stabilize the proteins to heat denaturation through the effect on the structure of water
[27]. Sucrose has been found to decrease the denaturation of a protein when exposed to a
denaturing agent such as guanidine hydrochloride [28]. Since the organic solvents and
matrix molecules used in MALDI may denature the protein in solution, the presence of a
stabilizing compound such as sucrose could maintain the non-covalent forces between the

subunits of the enzyme.

178

When sucrose is added to the C001 cytochrome c oxidase sample and a MALDI
analysis is performed using Sinapinic acid as the matrix, several peaks are detected that

represent complexes of the cytochrome c oxidase subunits, Figure 3.15. There is a broad

 
      
 
   

 

 

 

(l+H)”

‘
l
g l .
8 1 (mm) (l+ll+lll+IV+H)*
‘1 I

l +

‘ ("4’”) (ll+lll+H) (1+1 m I I+H) (l+ll+lll+H)

(II+C+H)" DEW”) A Al
10000 30000 50000 70000 110000 130000

mlz

Figure 3.15. Positive-ion linear MALDI-TOF spectrum of cytochrome c oxidase sample
C001. The cytochrome c oxidase was the nickel column purified sample from Table 2.
In the labels, 1 represents subunit I'+Ht, H represents subunit HA, III represents subunit
HI', C represent cytochrome 0. SA was used as the matrix with sucrose as a matrix
additive.

peak hour 34-35.5 kDa representing subunit IVA' bound to subunit HA (34.4 kDa) or III'
(35.3 kDa) as well as a peak at 69 kDa representing subunit IVA' bound to subunit 1'
(expected 68287). These data are consistent, structurally, with the location of subunit
IVA' between subunits I' and III' in the crystal structure [5]. Subunit IV does not appear
to be in contact with subunit HA [5], so the 34-35.5 kDa peak most likely represents an

association of subunits III' and IV A'. At 59 kDa, there is a peak due to the binding of

subunit HA to subunit HP. The complex of [I', IIA, IVA'] or [1, III', and IV A'] is detected

179

at approximately 93 kDa. At 125 kDa, there is a broad peak representing the intact
complex with and without subunit IVA'. In these experiments, subunit IVA' may be
present at a lower concentration than subunits I-III due to lower expression (lack of
engineered overexpression in sample CcOl). Subunit IVA' may also dissociate more
easily from the complex, since subunit IVA' is usually surrounded by lipids [5] and may
be more easily stripped from the complex during purification as lipids are removed.
Therefore it may be expected that some of the intact enzyme will consist of only subunit
1, II, and III as shown at the high m/z end of the spectrum in Figure 3.15.

Several peaks in the spectrum represent complexes of subunit IV and other
subunits. Why are these partial dissociations of the enzyme not random? Subunit IV
contacts subunits I and 1H through lipids in the crystal structure [5], so it is not surprising
that combinations of subunit IV with subunits I and 111 can be detected. Subunit IV does
not appear to be in contact with subunit H, so association of these two subunits would not
be expected.

There is also a peak at approximately 46.5 kDa. This m/z value for this peak does
not match any combinations of the oxidase subunits. The peak is very broad, but similar
in mass to subunit II bound to cytochrome cy (17,681 g/mol), the hydrophobically
anchored cytochrome c believed to be the normal electron transfer partner of the R.
sphaeroides cytochrome c oxidase [29]. The complex of cytochrome c bound to subunit
IIA would have an expected mass of 46,803 g/mol. This is approximately what is seen in
the spectrum. It is also interesting to note that cytochrome cy is only found bound to
subunit H. During the electron transfer process, cytochrome c binds to subunit H on the

external side of the membrane.

180

Can the complexes detected be representative of complete dissociation of the
enzyme followed by random recombination during MALDI matrix crystal growth?
There are several reasons why we believe this is not the case. A peptide, M, can be
detected as a dimer, (2M+H)+, in the MALDI experiment, but only at high concentrations
of the peptide. The initial concentrations used in these experiments were less than 10
11M. Also, the observed peaks are not representative of random recombination. For
example, if cytochrome 0 were concentrated enough, one would expect to see subunits I
and III, as well as subunit H, bound to the cytochrome c protein. However, this is not the

case.

 

Through the use of sucrose, peaks representing the partial dissociation of
cytochrome c oxidase can also provide information on the subunit-lipid interactions in the
intact protein. While the X-ray crystal structure reveals that subunit IV is affiliated with
subunits I and III, the crystal structure shows lipids bound to subunit IV as well. When
sucrose was used as a matrix additive with Sinapinic acid, the subunit IV region of the
spectrum changed, Figure 3.16. The sample analyzed in Figure 3.16 is the same CcOl
sample analyzed in Figure 3.12(B). In addition to the peaks representing the two versions
of subunit IV, additional peaks are seen. In Figure 3.16, a PC is detected bound to

subunit IVA'.

181

 

 

 

 

 

: (IVA'+H)+
l
g |
‘0 i
g
5
8’ +K"
s
0
I! %
-Ala
l .
PC IV'+H
l 1‘ ’ e. -
i - .1 . L . -T L 1
5000 5500 6000 6500 7000 7500 8000

mlz

Figure 3.16. Positive-ion linear MALDI-TOF spectrum of cytochrome 0 oxidase.
SA/sucrose was used as the matrix.

When a highly purified solution of C003 is analyzed prior to crystal growth,
several lipids are detected including PG, MGD, SL, and PC. Crystals are then grown for
x-ray crystallographic analysis. After the x-ray analysis, the crystal was dissolved in
water and the solution was analyzed using MALDI MS. The PG, MGD, and SL did not
remain bound to the enzyme during the crystal growth process. The PC was present in
both spectra suggesting that PC most likely remains bound to subunit IVA'. This is

consistent with the results shown in Figure 3.16.

182

 

 

 

A
w 8.
PG Heme
g, MGD
(D
C
.9
E
“g: (B)
E
o
at
P10
. .. .-... . .. . .... .-... -. .._... _ , ..u .--_.. ..ALAA_..-...--... 1...--.L
750 770 790 810 830 850 870 890
m/z

Figure 3.17. Positive-ion reflectron MALDI-TOF spectra of cytochrome c oxidase after
(A) before and (B) after crystal growth process. DHB was used as the matrix.

183

References

10.

ll.

Musatov, A.; Robinson, N.C. Subunit analysis of bovine cytochrome bct by
reverse phase HPLC and determination of the molecular masses by electrospray
ionization mass spectrometry. Biochemistry 1994, 33., 10561-10567.

Liu, Y-C.; Sowdal, L.; Robinson, N.C. Separation and quantitation of
cytochrome c oxidase subunits by reversed-phase high performance liquid
chromatography. Arch. Bioch. Biop. 1995, 324, 135-142.

Tsukihara, T.; Aoyama, H.; Yamashita, E.; Tomizaki, T.; Yamaguchi, H.;
Shinzawa-ltoh, K.; Nakashima, R.; Yaono, R.; Yoshikawa, S. The whole structure
of the 13-subunit oxidized Cytochrome c Oxidase at 2.5 A. Science 1996, 272,
1136-1144.

Luecke, H.; Schobert, B.; Richter, H.T.; Cartailler, J .P.; Lanyi, J .K. Structure of
bacteriorhodopsin at 1.55 A resolution. Mol. Biol. 1999, 291, 899-911.

Svensson-Ek, M.; Abramson, J .; Larsson, G.; T6rnroth, S.; Brzezinki, P.; Iwata,
S. The X-ray crystal structures of wild-type and EQ(I-286) mutant cytochrome c
oxidases from Rhodobacter sphaeroides. J. Mol. Biol. 2002, 321 , 329-339.

Garavito, R.M.; Ferguson-Miller, S. Detergents as tools in membrane
biochemistry. J. Biol. Chem. 2001, 276, 32403-32406.

Jensen, N.I.; Gross, M.L. A comparison of mass spectrometry methods for
structural determination and analysis of phospholipids. Mass Spectrom. Rev.
1988, 7, 41-69.

Kim, B.H.; Chang, Y.S.; Lee, B.D.; Ryu, S.H.; Shin, D.H. Mass spectrometric
analysis of change in phospholipids in biological membrane by external
environmental effects. Microchem. J. 1999, 63, 3-8.

Schiller, J .; Siib, R.; Petkovic', M.; Hilbert, N.; Miiller, M.; Zsch6rnig, 0.;
Amhold, J .; Arnold, K. CsCl as an auxiliary reagent for the analysis of
phosphatidylcholine mixtures by matrix-assisted laser desorption and ionization
time-of-flight mass spectrometry (MALDI-TOF MS). Chem. Phys. Lipids. 2001,
113,123-131.

Petkovié, M.; Schiller, J .; Miiller, M.; Benard, S.; Reichl, S.; Amhold, K.; Arnold,
J. Detection of individual phospholipids in lipid mixtures by matrix-assisted laser
desorption/ionization time-of-flight mass spectrometry: phosphatidylcholine
prevents the detection of further species. Anal. Biochem. 2001, 289, 202-216.

Marto, J .A.; White, F.M.; Seldomridge, S.; Marshall, A.G. Structural
chacterization of phospholipids by matrix-assisted laser desorption/ionization

184

 

12.

13.

14.

15.

l6.

l7.

18.

19.

20.

21.

fourier transform ion cyclotron resonance mass spectrometry. Anal. Chem. 1995,
67, 3979-3984.

Irnhoff, J. Polar lipids and fatty acids in the genus Rhodobacter. System. Appl.
Microbiol. 1991, 14, 228-234.

Hilmi, Y. Use of mutagenesis, novel detergents, and lipid analysis to optimize
conditions for achieving high resolution 3-dimentional crystal structure of
Rhodobacter sphaeroides cytochrome c oxidase. Ph.D Thesis. Michigan State
University. 2002.

Hiser, C.; Ferguson-Miller, SM. 2003. Personal communication.

Jeannot, M.A.; Zheng, J .; Li, L. Observation of gel-induced protein modifications
in sodium dodecylsulfate polyacrylamide gel electrophoresis and its implications

for accurate molecular weight determination of gel-separated proteins by matrix-

assisted laser desorption ionization time-of-flight mass spectrometry. J. Am. Soc.
Mass Spectrom. 1999, 10, 512-520.

Loo, J .A. Studying noncovalent protein complexes by electrospray ionization
mass spectrometry. Mass Spectrom. Rev. 1997, 16, 1-23.

Ghaim, J .B.; Tsatsos, P.H.; Katsonouri, A.; Mitchell, D.M.; Salcedo-Hemandez,
R.; Gennis, R.B. Matrix-assisted laser desorption ionization mass spectrometry of
membrane proteins: demonstration of a simple method to determine subunit
molecular weights of hydrophobic subunits. Biochim. Biophys. Acta. 1997, 1330,
113-120. ‘

Marx, M.K.; Mayer-Posner, F .; Soulimane, T.; Buse, G. Matrix-assisted laser
desoprtion/ionization mass spectrometry analysis and thiol-group determination
of isoforrns of bovine cytochrome c oxidase, a hydrophobic multisubunit
membrance protein. Anal. Biochem. 1998, 256, 192-199.

Whitelegge, J .P.; Gundersen, C.B.; Faull, KF. Electrospray-ionization mass
spectrometry of intact intrinsic membrane proteins. Prot. Sci. 1998, 7, 1423-
1430.

Barnidge, D.R.; Dratz, E.A.; Jesaitis, A.J.; Sunner, J. Extraction method for
analysis of detergent-solubilized bacteriorhodopsin and hydrophobic peptides by
electrospray ionization mass spectrometry. Anal.Biochem. 1999, 269, 1-9.

Hess, S.; Cassels, F.J.; Pannell, L.K. Identification and characterization of

hydrophobic Escherichia coli viruelence proteins by liquid chromatography-
electrospray ionization mass spectrometry. Anal. Biochem. 2002, 123-130.

185

 

22.

23.

24.

25.

26.

27.

28.

29.

Whitelegge, J .P.; Coutre, J .L.; Lee, J .C.; Engel, C.K.; Prive, G.G.; Faull, K.F.;
Kaback, H.R. Toward the bilayer proteome, electrospray ionization-mass

spectrometry of large, intact transmembrane proteins. Proc. Natl. Acad. Sci. USA.
96, 10695-10698.

Schindler, P.A.; Van Dorsselaer, A.; Falick, A.M. Analysis of hydrophobic
proteins and peptides by electrospray ionization mass spectrometry. Anal.
Biochem. 1993, 213, 256-263.

Green-Church, K.B.; Limbach, P.A. Matrix-assisted laser desorption/ionization
mass spectrometry of hydrophobic peptides. Anal. Chem. 1998, 70, 5322-5325.

Breaux, G.A.; Green-Church, K.B.; France, A.; Limbach, P.A. Surfactant-aided,
matrix-assisted laser desorption/ionization mass spectrometry of hydrophobic and
hydrophilic peptides. Anal. Chem. 2000, 72, 1169-1174.

Schey, K.L. Hydrophobic proteins and peptides analyzed by matrix-assisted laser
desorption/ionization. Methods Mol. Biol. 1996, 61, 227-230.

Back, J .F.: Oakenfull, D.; Smith, M.B. Increased thermal stability of proteins in
the presence of sugars and polyols. Biochemistry, 1979, 18, 5191-5196.

Taylor, L.S.; York, P.; Williams, A.C.; Edwards, H.G.M.; Mehta, V.; Jackson,
G.S.; Badcoe, I.G.; Clarke, A.R. Sucrose reduces the efficiency of protein
denaturation by chaotrOpic agent. Biochim. Biophys. Acta. 1995, 1253, 39-46.

' Myllykallio, H.; Zannoni, D.; Daldal, F. The membrane-attached electron carrier

cytochrome cy from Rhodobacter sphaeroides is functional in respiratory but not
in photosynthetic electron transfer. Proc. Natl. Acad. Sci. USA 1999, 96, 4348-
4353.

186

V. Conclusions

When analyzing the cytochrome c oxidase complex, there are many components
to consider including the protein subunits, hemes, and lipids. Typically, analysis requires
the extraction of the components fi'om the intact complex. We have shown here a method
for analyzing the protein subunits, hemes, and lipids in one MALDI experiment. When
using DHB as a matrix, the protein subunits, hemes, and lipids were detected. There are
advantages to performing the analysis on a solution of the intact enzyme. We have
analyzed lipid extracts of the enzyme and found the chloroform-solubilized extracts to be
difficult to handle. The lipid extracts were also unstable in the chloroform and are easily
oxidized. When analyzed in the aqueous enzyme solution, the lipids were more stable,
easy to handle, and produced better signals in the MALDI experiment. This makes
MALDI an attractive method for lipid component analysis in membrane proteins because
the need for extraction of the lipids is eliminated. All the components of the complex can
be detected and characterized in the MALDI experiment with the selection of the correct
matrix and matrix additives.

In cytochrome c oxidase from R. sphaeroides, all the subunits are connected to
each other and all are likely associated with lipids, but these connections are not all equal
in area and strength. The unique connectivity information from the partial dissociation of
the enzyme in mass spectrometry adds a new dimension to our understanding of the
lipid/protein complex. Information from the partial dissociation may be even more
valuble for enzymes containing more subunits, such as the l3-subunit cytochrome c

oxidase from bovine heart.

187

 

APPENDIX A

188

 

 

S-Methoxysalicylic Acid and Spermine: A New
Matrix for the Matrix-Assisted Laser
Desorption/Ionization Mass Spectrometry
Analysis of Oligonucleotides

Anne M. Distler and John Allison”
Department of Chemistry, Michigan State University, East Lansing, Michigan, USA

 

S-Methoxysalicylic acid (MSA) is demonstrated to be a useful matrix for matrix-assisted laser
desorption/ ionization time-of-flight (T OF) mass spectrometry of oligonucleotides, when
desorption/ ionization without fragmentation is desired. When MSA is combined with the
additive spermine, the need for desalting is reduced. The MSA/spermine matrix yields linear
TOF mass spectra with improved resolution, less fragmentation, and less intense alkali ion
adduct peaks than those spectra obtained using 3-hydroxypicolinic acid and 6-aza-2-thiothy-
mine with spermine or diammonium hydrogen citrate as additives. Instrumental conditions
are discussed to improve the spectral resolution, specifically the use of longer delay times in

the delayed-extraction ion source.
American Society for Mass Spectrometry

(I Am Soc Mass Spectrom 2001, 12, 456-462) © 2001

 

 

atrix-assisted laser desorption/ ionization

mass spectrometry (MALDI MS), developed

by Karas et al. [1], has been extensively used
to analyze biological compounds at the subpicomole
level. To date, most of the published MALDI MS work
has involved the analysis of peptides and proteins.
MALDI MS has also been used for the analysis of
oligonucleotides [2, 3]. The development of MALDI
matrices optimized for the analysis of oligonucleotides
has lagged behind that for peptides [4], but is currently
receiving considerable attention. This report discusses
the performance of 5-methoxysalicylic acid, with
spermine as a matrix additive, for the detection of
oligonucleotides in MALDI with the suppression of
prompt fragmentation in the linear time-of-flight (T OF)
MS experiment.

Oligonucleotides and DNA behave very differently
than peptides and proteins in the MALDI experiment.
Differences of note include the sensitivity, extent of
prompt fragmentation, the need for cleanup prior to
analysis, and resolution.

Certainly, for all classes of biomolecules studied by
MALDI, as sample handling methods and TOF instru-
ments have been developed, typical sample sizes used
for analysis have decreased. Peptide analysis at the
subpicomole level was rapidly achieved, although it has
taken longer to realize similar MALDI detection limits
for oligonucleotides. There have been significant ad-

 

Addreae reprint requests to Dr. john Allison, Department of Chemistry,
Michigan State University, East Lansing, MI 48824-1322 E-mail:
allisonecemsnsuedu

O 2001 American Society for Mass Spectrometry. Published by Elsevier Science Inc.

1044-0305/01/S2000
P11 51 044-0305(00)00212-4

189

vances from earlier reports of MALDI analyses of
oligonucleotides, in which 10-100 pmol of material
were required, to recent reports that show clear detect-
ability at the femtomole level [5, 6].

When a pure peptide is analyzed by MALDI-TOF
MS, it usually yields a single mass spectrometric peak
representing the intact molecule (in protonated or dep-
rotonated form). Matrices and sample conditions have
been reported to increase the extent of prompt fragmen-
tation of peptides in MALDI [7], but experirnentalists
have little predictable control over the process, gener-
ally. In contrast, oligonucleotides frequently fragment
promptly and extensively in linear TOF MS [8].

The analysis of pure samples certainly makes mass
spectral interpretation easier for all biomolecules. De-
pending on whether compounds are isolated from
biological sources, from electrophoresis experiments
(gels or membranes), or are synthesized, peptides and
oligonucleotides may be presented for mass spectro-
metry analysis in the presence of compounds such as
salts, buffers, and glycerol, frequently in higher relative
amounts than the analytes. In all cases, high amounts of
salts can sufficiently change the MALDI target material
to lower analyte signals. Glycerol is usually a problem
as a component when the first step of the MALDI
experiment is to grow crystals. In terms of the spectra
that result, salts have a very different impact when
analyzing peptides compared to oligonucleotides. 1f
NaCl is present, two peaks may appear representing a
single peptide in positive ion mode, most frequently
representing the [M + l-I]+ and [M + Na]* ions, sepa-
rated by 22 Da. As the salt contribution increases,

Received September 7, 2000
Revised November 30, 2000
Accepted November 30, 2000

J Am Soc Mass Spectrom 2001,12, 456-462

sodium adducts become more intense. The presence of
salts has a much more dramatic influence on the
MALDI spectra of oligonucleotides. With an ionic phos-
phodiester backbone, oligonucleotides can contain a
large number of anionic sites that must combine with an
assortment of cationic species such as I-I“, Na“, and K+
to desorb in a singly charged form. Combinations of
cations to provide a charge balance can lead to a
substantial number of peaks representing the intact
molecule. Whenever ions for a single species are gen-
erated with a range of m/z values, detection limits are
lowered. Also, it becomes much more difficult to detect
multiple components of similar mass. Certainly, a sim-
ple situation is to generate one mass spectral peak per
component, not several.

In the analysis of oligonucleotides with MALDI TOF
M5, the resolution achieved is often less than what
would be realized for peptides of similar size. Adduct
formation and nucleotide base loss have frequently
been cited as causes of the reduced resolution [9]. For
larger oligonucleotides, this is clearly the case. How-
ever, even for smaller sequences with resolved sodium
and potassium adducts, resolution for an oligonucleo-
tide is less than that obtained for a peptide of similar
mass on the same instrument. The use of an ion reflector
does not dramatically improve the resolution of oligo-
nucleotides in MALDI MS [10]. One of the few instances
in which relatively small oligonucleotides were ana-
lyzed by MALDI and spectra containing isotopic reso-
lution were obtained was recently published, using a
MALDI TOF instrument with an extended flight tube
[11]. We show that, for the TOF instrument used in this
work utilizing gridded ion optical components and
delayed extraction (DE), “nontraditional" ion source
settings can improve the resolution.

Although peptides and oligonucleotides have many
differences in terms of their MALDI MS analysis, they
also have some notable similarities in terms of the
development of the analytical utility of mass spectro-
metry for their analysis. Early in the history of MALDI,
demonstrations of the ability to generate signals from
large, intact, ionized proteins maintained excitement for
this new method, allowing it to be further developed.
Whereas positive results had been reported for proteins
with molecular weights (MW) of several hundred thou-
sand, the mass spectral peaks were very broad, and it
was difficult to extract what would be considered in
mass spectrometry an accurate mass. The early results
showed that, in most cases, MW ’3 determined were
more useful than those derived from gel electrophore-
sis. Today, the real strength of MALDI in protein
analysis does not lie in its ability to generate ions of
intact proteins, but in the subpicomolar sensitivity
obtainable for smaller peptides. The analysis of enzy-
matic and chemical digestion products is the applica-
tion that is making the real contributions to protein
analysis. Scientists have certainly worked to demon-
strate that large oligonucleotides can be characterized
by MALDI as well. Whereas a few reports have been

190

A NEW MALDI MATRIX FOR oucorvucu-zonoas 457

published demonstrating the detection of SO-mers, the
MALDI response is much more sharply a function of
MW than is observed for peptides, making the detection
of large oligonucleotides far from routine. Because a
potential application was the use of mass spectrometry
to replace gel electrophoresis when sequencing DNA
using the Sanger method, the requirement of the mass
spectrometry method was different than when sequenc-
ing peptides. Peaks with good resolution must be
detectable not only at low m/z values, but over a very
wide mass range, because the approach for DNA anal-
ysis is a ladder sequencing method. Although mass
spectrometry may not in the near future replace gels for
DNA sequencing, other powerful applications are
emerging, notably the analysis of single nucleotide
polymorphisms (SNPs) [12, 13]. In this area, mixtures of
small oligonucleotides, usually with molecular weights
less than 6000 Da, are generated which are indicative of
errors in genetic code. This is an application to which
MALDI MS is well suited. Certainly, the need continues
to develop methods for improved analysis of oligonu-

cleotides of all sizes. The SNP application is clearly,

driving method development for analyses of oligonu-
cleotides of sizes for which the challenges are not the
creation of new instruments, but improved matrix
chemistry. The goals continue to be lower detection
limits, the ability to perform analyses with less or no
cleanup (recognizing the contaminants most commonly
encountered, and that purification steps always lead to
sample loss), and to identify experimental conditions in
which fragmentation is decreased or eliminated. Cer-
tainly, in the analysis of mixtures, if one could rely on
generating only one peak per component, as is common
for peptides, then the analysis of mixtures of oligonu-
cleotides would be much easier.

Many matrices have been developed for peptide
analysis [14, 15] by MALDI such as nicotinic acid,
ferulic acid, sinapinic acid, and 2,5-dihydroxybenzoic
acid. Because, generally, these are not as useful for
oligonucleotide analysis, alternate matrices have been
identified [16, 17]. Several matrices were found to be
compatible for the analysis of oligonucleotides using
UV lasers, including 3-hydroxypicolinic acid (HPA) [18]
and 6-aza-2-thiothymine (ATT) [19]. Since the develop-
ment of these matrices, HPA has become the standard
matrix for the analysis of larger nucleic acids while ATT
is used for sequences containing less than 25 bases [2].
In addition to UV-MALDI, work has been done using
infrared lasers [20, 21]. Successful matrices for IR-
MALDI include succinic acid and urea [22]. Because
matrix selection alone does not overcome some of the
problems associated with the MALDI MS analysis of
oligonucleotides, a number of matrix additives have
been developed. The utility of ammonium salts has
been demonstrated to reduce the formation of alkali ion
adducts. Ammonium acetate was the first to be used in
the MALDI experiment [8]. Since then, other ammo-
nium salts have appeared in the literature with the most
successful being diammonium hydrogen citrate [23, 24].

 

458 DISTLER AND ALLISON

Thus, if an oligonucleotide exists in solution in polyan-
ionic form as 1"", it forms a singly charged anion by
adding combinations of FF, Na“, and K‘” ions. If NH4+
ions are present, these compete effectively with alkali
ions in complexing with negatively charged phos-
phates. During the desorption/ ionization process, am-
monia is lost, leaving protons behind. In this paper, the
fully protonated form of an oligonucleotide, [P"‘ +
nH*], the ”free acid form," will be referred to as M. If
no alkali ions are involved and all of the phosphates are
neutralized by protons, negative ion MALDI yields
deprotonated molecules, designated as [M - H]'. In
our lab, the role of polyamines as matrix additives has
been explored and spermine was found to improve
MALDI spectra for single-stranded oligonucleotides,
eliminating both the need for ammonium citrate as well
as desalting [5, 25].

We demonstrate here that 5-methoxysalicylic acid
(MSA) is useful in the MALDI analysis of oligonucleo-
tides. In our experiences, it is unique in generating
intact oligonucleotide ions with very little fragmenta-
tion. MSA is not a new matrix in the field of MALDI. It
has been suggested that a mixture of MSA, 2,5-dihy-
droxybenzoic acid, and fucose is a useful matrix for the
analysis of peptides [26], whereas MSA with 9-anthra-
cenecarboxylic acid has been demonstrated as a matrix
for the analysis of polymers [27]. MSA has also been
used in a multicomponent matrix when quantitation
was being attempted [28]. We show here that MSA is
also an effective matrix for the analysis of oligonucleo-
tides. The MALDI spectra obtained using MSA are
compared here to those obtained using the traditional
HPA and ATT matrices. Best results were obtained
using spermine as an additive, rather than diammo-
nium hydrogen citrate.

Experimental

The 12-mer oligodeoxyribonucleotide d(CGCGAAT-
TCGCG) was purchased from the Michigan State Uni-
versity Macromolecular Structure, Sequencing, and
Synthesis Facility (East Lansing, MI). The stock solution
used had a concentration of 46.6 pm/pL and was used
without desalting. The matrices 6-aza-2-thiothymine
and 5-methoxysalicylic acid were purchased from Al-
drich (Milwaukee, WI). Spermine and 3—hydroxypico-
linic acid were purchased from Fluka (Milwaukee, WI).
All matrices were used without further purification.
When spermine was used as a matrix additive, it was
prepared at a concentration of 25 mM in water. When
diammonium hydrogen citrate (J .T. Baker, Phillipsburg,
NJ) was used, it was prepared at a concentration of 50
mM in water. Saturated matrix solutions were made
using a 1:1 acetonitrile / additive solution. Samples were
prepared by mixing 1 uL of analyte solution with 1 pl.
of the matrix solution on a gold sample plate and
allowing the mixture to air dry.

MALDI mass spectra were recorded using an Applied
Biosystems (Framingham, MA) Voyager delayed ex-

191

I Am Soc Mass Spectrom 2001, 12, «156—462

a,» -4] c .
/\ [,N K / /
I :4. L In
ATT MSA HPA
O

Spermine

Figure 1. Structures of the matrices and additives used in this
study.

Diammonium Hydrogen Citrate

traction linear time-of-flight mass spectrometer
equipped with a nitrogen laser (337 nm, 3 ns pulse). For
the negative ion MALDI spectra reported here, the
accelerating voltage was -15 kV, the delay time was
700 ns, the grid voltage was 94.5% of the accelerating
voltage, and magnitude of the guide wire voltage was
0.20% of the accelerating voltage. Typically, 50 laser
shots were averaged for each spectrum.

UV/ visible spectra of aqueous solutions of the ma-
trices were recorded using an ATI Unicam (Cambridge,
UK) UV2 spectrophotometer. All spectra were acquired
with a scan Speed of 120 ml min, a data interval of 0.5
nm, and a 2.0 nm bandwidth. Spectra were acquired for
three matrices, each at a concentration of 2.0 X 10‘5 M.
Quartz cuvettes were used with water in the reference
cuvette.

Results and Discussion

To demonstrate the advantages of MSA as a MALDI
matrix for oligonucleotide analysis, we selected the
Dickerson dodecamer, d(CGCGAATTCGCG), referred
to here as DD, as an example of a typical oligonucleo-
tide. We have obtained similar results for a variety of
other oligonucleotides, including d(ACCCAC-
CCACCC), d(AAAAACCAAAAA), d(‘I'ITITG-
G'ITI'IT), and d(CCGGAATTGGCC). DD was chosen
as an example, because it has been used extensively in
the literature, and contains all four bases. Furthermore,
while all of these oligonucleotides fragment in typically
used matrices such as ATT, DD fragments most exten-
sively. The nomenclature suggested by McLuckey [29]
is used to identify fragment ions.

Data are shown for the analysis of DD using the
three matrices HPA, ATT, and MSA. Their structures
are shown in Figure 1. Using the traditional additive,
diammonium hydrogen citrate (DHC), these three ma-
trices yield the spectra shown in Figure 2. The A'IT/
DHC mixture yields the complex spectrum shown in
Figure 2a. The resolution for the peak representing the
deprotonated molecule at m/z 3645 is low, and extensive

 

 

] Am Soc Mass Spectrom 2001, 12, 456-462

 

 

mo 21” 1.00 8100 3000

Figure 2. MALDI-MS negative-ion mass spectra of DD with
DHC as an additive using (a) ATT, (b) HPA, and (c) MSA.

fragmentation occurs. The ya and ws peaks, which are
more intense than that for the deprotonated molecule,
are formed following phosphodiester bond cleavage
between the C and A nucleosides. In contrast, the
HPA/DHC matrix produces a much different MALDI
spectrum, Figure 2b. The deprotonated molecule is the
dominant peak in the spectrum, with only small frag-
ment ion peaks present. In comparing Figure 2a, b, the
fragment ions do not change when the matrix is
changed, but l-IPA produces fewer fragment ions. The
resolution for the deprotonated molecule peak is clearly
higher than in the ATT/Dl-IC spectrum. At m/z values
above m/z 3645, a series of sodium ion adducts are
observed, in which an l-I+ is replaced by a sodium ion
in the mono-anion. The low mass shoulder on the m/z
3689 peak ([M - 3I-l+ + 2Na+]’) is a potassium ad-
duct, [M - 2H" + 101'.

The spectrum obtained using MSA/DHC as the
matrix, shown in Figure 2c, closely resembles that
obtained using HPA/DHC. There are fewer detectable
fragment ions in Figure 2c and the peaks are less
intense, compared to those in Figure 2a, b. Again, the
change in matrix does not yield a substantial change in
the fragmentation pattern, but changes the extent of
fragmentation. The deprotonated molecule peak is
again the most intense in the spectrum. It is interesting
to note that the potassium adduct has greater intensity
than the monosodium adduct. In Figure 2c, alkali ion
adducts contribute to the unresolved high m/z "tail.”
Although the addition of ammonium citrate reduces
cation adduction, there is still a need for desalting. The
ammonium citrate clearly works more efficiently when
used with HPA or MSA, and does provide spectra that
are superior to those obtained using the matrices alone.
Clearly, the choice of matrix (as well as the choice of
additive [30]) has a substantial impact on the extent of
fragmentation, with MSA allowing for desorption of
intact oligonucleotides with minimal formation of frag-
ment ions. A reduction of fragmentation accompanied
by formation of an intense peak representing the dou-
bly charged molecular ion has also been seen using
3-l-IPA, at an excititation wavelength of 355 nm [31].

In our experience, spermine is found to be consis-
tently superior to ammonium citrate as a matrix addi-
tive. As shown in Figure 3, the choice of matrix affects

192

A NEW MALDI MATRIX FOR OUGONUCLEOTIDES 459

(a) A"

f stimuli .. I;
serial?"

4 a A MJ‘K .
Y '

i0“ 21“ m 3100 soon

 

 

 

Figure 3. MALDI-MS negative-ion mass spectra of DD with
spermine (S) as an additive using (a) ATT, (b) HPA, and (c) MSA.

the performance of spermine as an additive. The nega-
tive ion MALDI spectrum obtained using ATT/sperm-
ine as the matrix. Figure 3a, yields a fragmentation
pattern similar to that seen using DHC as an additive,
Figure 2a. Regardless of the additive, fragment ions are
more abundant than the deprotonated molecule. With
spermine as the additive, the resolution of the [M -
H]’ peak increases, and only small sodium and potas-
sium adduct peaks are present. In addition to the alkali
ion adducts, there are two small higher-m/z peaks that
represent a matrix and a spermine adduct. This is not
the case in the HPA/spermine spectrum shown in
Figure 3b. Although the fragmentation is minimal, the
spermine adduct becomes a dominant peak in the
spectrum. Spermine is ineffective in eliminating the
formation of alkali ion adducts in HPA for this experi-
ment. The five sodium adducts in the 771/: range 3667-
3755 increased in intensity from those observed when
HPA/DHC is the matrix. Adduct ions are also formed
in which both spermine and alkali ions are attached.
When combined with spermine, MSA is an ideal
matrix for the desorption/ ionization of intact oligonu-
cleotides, as shown in Figure 3c. The spectrum obtained
using the MSA/spermine matrix contains very small
wa-wlo and ya fragment ions. The peak representing the
intact analyte is fully resolved with the sodium and
potassium adducts greatly reduced. The spermine ad-
duct is of low intensity compared to the intact, depro-
tonated molecule. Also, there is no peak representing
the doubly deprotonated molecule as seen previously in
other spectra of oligonucleotides containing negligible
fragmentation [31]. Thus, if the goal is to obtain spectra
in which one oligonucleotide yields a single peak, MSA
will limit fragmentation and spermine will limit the
formation of adducts, and this combination of MSA and
spermine retains the favorable properties of each matrix
component. The performance of spermine has been
discussed previously and it is not our intention to
reintroduce it here. We do note that the oligonucleotide
samples were used as supplied from the Michigan State
University Macromolecular Structure, Sequencing, and
Synthesis Facility. As synthesized, using standard
methods [32], these samples had a high, but poorly
defined salt content, and contained other contaminants

 

460 DISTLER AND ALLISON

”fl”

 

 

w v v v

I" a. an run an n I"
‘

Figure 4. MALDI-MS negative-ion mass spectrum of 10 fmol of
DD, using MSA/spermine as the matrix.

such as a buffer. For oligonucleotides made in this way,
desalting is unnecessary if spermine is used.

The spectra in Figures 2 and 3 were obtained using
47 pmol of oligonucleotide as the analyte. With this
amount, the spectra are strong and the differences can
be clearly documented. The MSA/spermine combina-
tion allows for spectra to be obtained with sample
amounts below 1 pmol. Figure 4 shows a portion of the
spectrum obtained using the MSA/spermine combina-
tion with a total of 10 fmol of DD in the target. A
signal-to-noise ratio of 12:1 is achieved (prior to
smoothing).

The results presented here clearly show that matrix
selection can have a dramatic influence on the extent of
fragmentation of oligonucleotides. For the three matri-
ces studied, MSA and HPA are both organic acids, and
ATT is considered as a neutral matrix. If ATT were
drawn in the enol form, all three would have aromatic
hydroxy groups that could be the source of protons for
analyte ionization in the positive ion MALDI experi-
ment. Rather than considering structural features such
as specific functional groups, there may be a correlation
between matrix performance and molar absorptivities.
The UV/ visible spectra of equimolar aqueous solutions
of the three matrices used in this work are shown in
Figure 5. As suggested by their spectra, MSA has the
greatest molar absorptivity at 337 nm with a value of
2240 M’1 cm". ATT and HPA have molar absorptivi-
ties at 337 nm of 1135 and 911 M‘1 cm”, respectively.
Although the UV experiment was conducted in solu-

 

 

 

fl

W ("I“)

Figure 5. UV-visible spectra of the matrices. For each, the
concentration (aqueous) was 2.0 X 10'5 M.

193

I Am Soc Mass Spectrom 2001, 12, 456-462

tion, we expect a similar result for the solid matrices.
This is reflected in the fact that, when MSA is used as a
matrix, significantly lower laser powers are required to
generate intense signals. If a matrix is an efficient “light
harvester," with a high molar absorptivity, absorbing
crystals may quickly achieve higher temperatures than
would crystals of molecules with lower extinction coef-
ficients. Because the matrix molecules absorb the en-
ergy, time is required for energy to flow from matrix
molecules into analyte molecules [33]. At higher tem—
peratures, desorption rates increase, decreasing the
time in which energy can flow into the analyte. The
result would be less prompt fragmentation. This is one
possible explanation for the reduced fragmentation
observed when MSA is used. With the data provided
here, our intention is not to develop predictive capabil-
ities for fragmentation of oligonucleotide ions as a
function of matrix choice, but rather to establish that
fragmentation is greatly reduced by use of the MSA/
spermine matrix combination.

Whenever new matrices are proposed, the question
of sample spot homogeneity should be addressed. For-
mation of MSA crystals is visibly different than for ATT
and HPA. The MSA solution, applied to a MALDI plate,
tends to spread more than other matrices, yielding a
thinner, more uniform bed of crystals. Spectra such as
those shown can be obtained at most points across the
target, unlike the other matrices for which the crystal
growth is more extensive around the outer edges of the
target.

All of the spectra shown were obtained with what
we consider to be ”nontraditional" tuning of the
MALDI DE-TOF instrument used for this worlc The PE
Biosystems TOF mass spectrometer uses a ”gridded”
ion source and a linear TOF mass analyzer; changes in
voltages suggested here may not have the same effect
on spectra obtained using instruments that do not
employ grids. If analyzing a peptide with a molecular
weight in the 3000-4000 range, an accelerating voltage
of 20 kV, a grid voltage of 94°/o—95% the accelerating
voltage, a guide wire voltage of approximately 0.05%,
and a delay time of 100 ns are the typical conditions
selected for optimal resolution. Using these conditions,
the resolution obtained for an oligonucleotide is much
less than that for a peptide of similar mass. The reduc-
tion in resolution for oligonucleotides could result from
a number of causes including ion fragmentation in the
ion source during acceleration, and slow desorption/
ionization-kinetics following laser irradiation. Frag-
mentation which occurs on a time scale longer than the
ion generation and shorter than the acceleration time
will be manifested in a linear TOF experiment as tails on
the fragment ion peaks towards higher m/z values [34].
However, the spectra shown here have negligible tail-
ing on the fragments peaks and more prominent tailing
on the peak correlating to the intact ion. Processes other
than fragmentation during acceleration must occur
which lead to the decreased resolution in an oligonu-
cleotide spectrum. The slow desorption/ ionization ki-

 

I Am Soc Mass Spectrom 2001, 12, 456-462

 

 

(a) iflll
M
)4
i “

fl] 1“.
”1
. A A‘ fi ‘J
I. m “I 87! ms 37‘"

*

Figure 6. MALDI-MS negative-ion mass spectra of DD with
MSA/spermine as the matrix with (a) 100 ns delay and (b) 700 ns
delay.

netics for an oligonucleotide may be another cause of
decreased resolution. Oligonucleotides are more polar
than peptides or proteins and, thus, require more en—
ergy to generate intact, gas phase ions [10]. Increasing
the delay time is one way to compensate for the
processes that may lead to low resolution. Improve-
ments in resolution as a function of tuning are not
unique to the one instrument on which these experi-
ments were performed. Improvements were also ob-
served on a PerSeptive Biosystems Voyager STR DE
instrument. A portion of the DD spectrum obtained
using MSA/ spermine and the standard tuning is shown
in Figure 6a; the same region of the spectrum, obtained
using a much longer delay time, is shown in Figure 6b.
For longer delay times, slightly higher guide wire
voltages are recommended. This combination results in
much improved resolution. The optimized conditions, a
guide wire voltage of 0.200% and a delay time of 700 ns,
provide improved resolution for all peaks in the spec-
trum in both positive and negative ion modes. This
result is not matrix specific. Although we make this
observation, and routinely use long delay times when
obtaining MALDI spectra of oligonucleotides, there is
no one process that could clearly be linked to the delay
time-resolution correlation. Low resolution for oligonu-
cleotides under standard ion source conditions does not
appear to be due to decomposition of the intact anions
during acceleration, because spectra obtained using
long delay times do not contain fragment ion peaks that
are absent or less intense when short delay times are
used. Desorption kinetics, linked to the unique variety
of charge states of the oligonucleotides that may be
trapped in the matrix, may be very different than those
for peptides that are trapped in predominantly neutral
form.

If the combination of a new matrix and modified
tuning leads to the advantage of increased resolution,
one may expect the improvement to be further realized
by using a reflectron. Again, oligonucleotides behave
very differently than do peptides in MALDI MS [10].
Resolution is much lower for oligonucleotides and the
use of a reflectron does not significantly improve the
resolution [10]. When analyzing DD with each of the

194

A NEW MALDI MATRIX FOR OLIGONUCLEUI'IDES 461

three matrices on a Voyager STR in reflectron mode,
there is no change in resolution. The [M - I-Il‘ peak is
smaller and the spectra are worse than if a linear TOF
MS is used. The intensity of the [M — I-Il‘ peak may
decrease in reflectron mode because we have delayed
all fragmentation from prompt (in source) to occurring
in the flight tube. If this is the case, a post—source decay
experiment should yield an intense spectrum. This is
not observed. It is possible an electron may be ejected
from the [M — H1" species leaving [M — H] in the
flight tube. Whatever the mechanism, best results are
achieved for oligonucleotides in negative mode in a
linear MALDI experiment.

Conclusion

Although HPA/DHC and ATT/DHC are matrix/addi-
tive combinations commonly used for oligonucleotide
analyses, MSA proves to be a superior matrix for
detection of the intact analyte in the 12-20 base pair
range when spermine is used as an additive. The
MSA/spermine matrix provides alkali ion adducts of
low abundance, little fragmentation, and better resolu-
tion when compared to ATT and I-IPA with either DHC
or spermine. Spermine, when used with MSA, reduces
the need for desalting of oligonucleotide samples.

Although analyses of double-stranded oligonucleo-
tides have been reported with several matrices [21, 22],
such noncovalently bound complexes are still difficult
to routinely detect intact. In most cases, if a double-
stranded oligonucleotide is analyzed using MALDI MS,
peaks representing only the single strands are observed
in the spectrum. Although MSA/spermine is very ef-
fective as a matrix for single strands, it does not allow
for double strands to be detected as intact ions. The
search continues for a matrix that will routinely allow
such noncovalent interactions to remain intact.

References

1. Karas, M.; Bachmann, D.; Bahr, U.; Hillenkamp, F. hit. I. Mass
Spectrom. lon Processes 1987, 78, 53-68.
2. Nordhoff, E.; Kirpekar, P.; Roepstorff, P. Mass Spectrom. Rev.
1996, 15, 67-138.

. Limbach, P. A. Mass Spectom. Rev. 1996, 15, 297-336.

. Zhang, L-K.; Gross, M. L. I. Am. Soc. Mass Spectrom. 2000, 11,
854-865.

Asara, 1.; Allison, 1. Anal. Chem. 1999, 71, 2866-2870.

. Roskey, M. T.; Juhasz, P.; Smimov, I. P.; Takach, E. J.; Martin,

S. A.; Haff, L. A. Proc. Natl. Acad. Sci. USA 1996, 93, 4724-4729.

7. Lavine, (3.; Allison, I. I. Mass Spectrom. 1999, 34, 741-748.

8. Currie, (3.; Yates, I. II]. I. Am. Soc. Mass Spectrom. 1993, 4,
955—963.

9. Nordhoff, E.; Cramer, R.; Karas, M.; Hillenkamp, F.; Kirpekar,
F.; Kristiansen, K.; Roepstorff, P. Nucleic Acids Res. 1993, 21,
3347-3357.

10. Iuhasz, P.; Roskey, M. T.; Smimov, I. P.; Haff, L. A.; Vestal,
M. L.; Martin, S. A. Anal. Chem. 1996, 68, 941-946.

11. Koomen, I. M.; Russell, D. H. I. Mass Spectrom. 2000, 35,
1025-1034.

12. Griffin, T. J.; Smith, L. M. Anal. Chem. 2000, 72, 3298-3302.

DU

cum

462

13.

14.

15.

16.

17.

18.

19.

20.

21.

23.

DISTLER AND ALLISON

Fei, Z; Smith, L. M. Rapid Commun. Mass Spectrom. 2000, 14,
950-959.

86msen, K. 0.; Schir, M.; Widmer, H. M. Chimica 1990, 44,
412-416.

Hillenkamp, F.; Karas, M.; Ingendoh, A.; Stahl, B. In Biological
Mass Spectrometry; Burlingame, A.; McCloskey, I. A., Eds.;
Elsevier Scientific: Amsterdam, 1990; p 49.

Parr, C. R. Fitzgerald, M. C .; Smith, L. M. Rapid Commun. Mass
Spectrom. 1992, 6, 369 -372.

Spengler, 8.; Pan, Y.; Cotter, R. J.; Kan, L. S. Rapid Commun.
Mass Spectrom. 1990, 4, 99-102.

Wu, K. J.; Steding, A.; Becker, C. H. Rapid Commun. Mass
Spectrom. 1993, 7, 142-146.

Lecchi, P.; Le, H. M. T.; Pannell, L. K. Nucleic Acids Res. 1995,
23, 1276-1277.

Kirpekar, F.; Berkenkamp, 5.; Hillenkamp, F. Anal. Chem. 1999,
71 , 21334—2339.

Berkenkamp, 5.; Kirpekar, F.; Hillenkamp, F. Science 1998, 281,
260—262.

. Nordhoff, E.; Ingendoh, A.; Cramer, R.; Overberg, A.; Stahl, B.;

Karas, M.; Hillenkamp, F.; Crain, P. F. Rapid Commun. Mass
Spectrom. 1992, 6, 771—776.

Li, Y. C. L; Cheng, S. W.; Chan, T. W. D. Rapid Commun. Mass
Spectrom. 1998, 12, 993-998.

195

24.

27.

8

31.

32.

I Am Soc Mass Spectrom 2001, 12, 456-462

Pieles, U.; Ziircher, W.; Schir, M.; MBser, H. E. Nucleic Acid
Res. 1993, 21, 3191-3196.

. Asara, ]. M.; Hess, I. S.; Lozada, E.; Dunbar, K. R.; Allison, I.

1. Am. Chem. Soc. 2000, 122, 8-13.

. Gusev, A. I.; Wilkinson, W. R.; Proctor, A.; Hercules, D. M.

Anal. Chem. 1995, 67, 1034-1041.
Tang, X.; Dreifuss, P.; Vertes, A. Rapid Commun. Mass Spectrom.
1995, 9, 1141-1147.

. Gusev, A. 1.; Wilkinson, W. R.; Protor, A.; Hercules, D. M.

Frescnius I. Anal. Chem. 1996, 354, 455-463.

. McLuckey, S. A.; Van Berkel, J.; Glish, G. L. I. Am. Soc. Mass

Spectrom. 1992, 3, 60-70.

. Zhu, Y. F.; Chung, C. N.; Taranenko, N. L; Allman, S. L;

Martin, S. A.; Haff, L.; Chen, C. H. Rapid. Commun. Mass
Spectrom. 1996, 10, 383—688.

Zhu, L.; Parr, G.; Fitzgerald, M. C.; Nelson, C. M.; Smith, L. M.
I. Am. Chem. Soc. 1995, 117, 6048-6056.

Applied Biosystems. Models 392 and 394 DNA/RNA Synthesiz-
ers User’s Manual. Part 902351, Document Revision 8. Norwalk,
CT, March 1994.

. Vertes, A.; Gijbels, R.; Levine, R D. Rapid Commun. Mass

Spectrom. 1990, 4, 228-233.

. Nordhoff, E.; Karas, M.; Cramer, R.; Hahner, 5.; Hillenkamp,

F.; Kirpekar, F.; Lezius, A.; Muth, J.; Meier, C.; Engels, I. W. I.
Mass Spectrom. 1993, 30, 99-112.

APPENDIX B

196

 

Anal. Chem. 2001, 73. 5000-5003

Improved MALDI-MS Analysis of Oligonucleotides
through the Use of Fucose as a Matrix Additive

AmhfistlerandJolmAliison‘

Department of Chemistry. Michigan State University, East Lansing, Michigan 48824-1m

With the development of matrix additives. the MALDI-MS
analysis of oligonucleotides has improved greatly. When
the monosaccharide fucose is combined with the matrix,
the homogeneity of the MALDI target. signal strength. and
signal duration are increased. The sensitivity of the
MALDI experiment increases. allowing for improved
detection of components in a complex mixture of oligo-
nucleotides, such as that encountered in sequencing
experiments. The addition of fucose to the matrix also
causes a reduction in the extent of fragmentation of
oligonucleotides.

Matrix-assisted laser desorption/ ionization mass spectrometry
(MALDI-MS). developed by Karas et al..I allows for the charac-
terization of many types of molecules with great speed and
efficiency. MALDI-MS has emerged as an important technique
for the analysis of biomolecules. such as peptides and oligonucie
otides. Despite the success of the technique. MALDI-timeof-ilight
(T OF) MS. has several disadvantages. including variable resolu-
tion and discrimination against components of mixtures. In
addition. simple target preparation methods typically result in poor
sample—target homogeneity.

Matrix additives have been developed in response to limitations
of the MALDI experiment in its simplest form. Matrix additives.
such as nitrocellulose. improve spectra from peptide and protein
samples contaminated with salts and synthetic polymers.2 Sper-
mine as an additive effectively desalts oligonucleotide samples.
eliminating the need for purification prior to analysis.3 Carbohy-
drates as matrix additives have also been successfully used for
MALDI analyses. Sucrose and glucose were first used as matrixes
in infrared (IR) laser desorption. multiphoton-ionization ('MUPD-
MS with a reflectron TOF mass spectrometer.‘ oGiucose. D-
ribose. and ofructose were first used in combination with the
matrix 2.5dihydroxybenzoic acid (DHB) in order to increase
resolution In ultraviolet (UV) MALDI analyses of peptides.
performed using Fourier transform mass spectrometry (FTMS).s

 

‘ Phone: 517-355-9715. ext 492. Fax: 517353-1793. E-mall: alllsonO
cem.msu.edu.

(1) Karas. M.; Bachmann. D.: Bahr. U.: Hillenkamp, F. Int. J. Mass Spectrom.
lon Processes. 1987. 78. 53-68.

(2) Kussman. M.; Nordhoff, E: Rahbek-Nielsen. H.; Haebel. S.; Rossel-Larsen.
M.; Jakobsen. L.; Cobom. J.; Mirgorodsitaya. 8.: Kroli-Kristensen. A.; Palm.
L: Roepstorff, P. 1. Mass Spectrom. 1997. 32. 593-601.

(3) Asara. J.; Allison. J. Anal. Chem. 1999. 71. 2866-2870.

(4) Beavis, R. C.: Lindner. J.; Grotemeyer, J.; Schlag. E. W. Chem. Phys Lett.
1988. 146,310-314.

(5) Raster. C.: Castoro. ]. A.; Wilkins. C. L I. Am. Chem. Soc. 1992. 114.
7572-7574.

5000 Analytical Chemistry, Vol. 73. NO. 20. October 15. 2001

197

The use of carbohydrates as matrix additives was then extended
to MALDI-TOFMS in the analysis of mixtures from peptide
digestions. with fucose being most effective.6 When equal amounts
of fucose and matrix were combined. several improvements in
the experiments resulted. including simultaneous desorption and
ionization of an increased number of components in a peptide
mixture (compared to that seen with matrix alone). reduction in
sodium adduct formation. improved sample spot homogeneity and
reproducibility. and higher resolution.6

Attempts have been made to explain the role of fucose as a
matrix additive in the MALDI experiment. It has been suggested
that a multicomponent matrix produces a more homogeneous
collection of crystals by minimizing the "variation of the analyte
to-matrix molar ratio and analyte concentration".7 It has also been
suggested that fucose may rapidly decompose to yield COM and
H203). which can collisionally cool gasphase ions during desorp
tion.5 Other effects of fucose have been linked to its ability to cool
the matrix. since it is a nonabsorbing component of the matrix
mixture.s This is an interesting idea. If MALDI target crystals that
contain both absorbing and nonabsorbing molecules can be
grown. then perhaps less energy will be deposited per unit volume.
resulting in lower temperatures being achieved upon laser
excitation. In considering a possible additive that does not absorb
at 337 nm. a sugar may be a reasonable option. Monosaccharides
are soluble in the solvents commonly used in MALDI. Although
not aromatic. they contain some of the structural features that
many matrixes have and have a size similar to that of typical matrix
molecules. However. one would not likely turn to fucose as the
first choice. When analyzing peptides by MALDI-FTMS. Billeci
et a1. evaluated the effect of incorporating glucose. fructose.
filactose. maltose. and sucrose at various concentrations. and
found fucose to be the most useful.‘ For oligonucleotide analysis
by MALDI-TOF MS. we evaluated fructose. glucose. sucrose.
galactose. and maltose as matrix additives. Fucose remains
uniquely effective for this application. as well.

When analyzing oligonucleotides using MALDI-MS. additives
have proven to be important. MALDI analysis of oligonucleotides
has been a greater challenge than peptide analysis. First. new
matrixes were needed in order to successfully detect intact
oligonucleotide ions. The search for better matrixes still contin-
ues.‘ The development of additives was an important step.

 

(6) Billeci. T. M.; Stults, J. T. Anal. Chem. 1993. 65. 1709-1716.

(7) Cusev. A. 1.: Wilkinson. W. R: Proctor. A.; Hercules. D. M. Anal. Chem.
1995. 67. 1034-1041.

(8) Distler. A M.; Allison. J. I. Am. Soc. Mass Spectrom. 2001. 12. 456-462.

10.1021/aw15550+ CCC: $20.00 0 2001 American Chemical Society
Published on Web 09/19/2001

 

Additives such as ammonium citrate9-‘° and spermine3 serve a
specific purpose. the reduction of alkali-ion adducts. Fucose as
an additive offers a different contribution to oligonucleotide
analysis. If a sample has a high sodium chloride concentration
and forms heterogeneous crystals. can two additives be used. and
are their influences additive? There is a possibility that if the
crystal composition reflects solution composition. the matrix can
be diluted to the point that a MALDI experiment is no longer
taking place if too many additives are used simultaneously. in our
experience. the use of multiple additives has had no adverse
effects on the MALDI experiment. When spemrine. fucose. and
matrix are combined. the advantages of both spermine and fucose
are realized.

Although fucose has been used to aid in the analysis of
peptides and proteins. no work has been reported on the effects
of fucose In oligonucleotide analysis. Here. we present several
benefits of using fucose in the MALDI analysis of oligonucleotides.
These benefits include a reduction in fragmentation. increased
detection of components in a complex mixture. improved sample
spot homogeneity. and increased signal duration.

EXPERIMENTAL SECTION

The oligonucleotides used in this work were purchased from
the Michigan State University Macromolecular Structure. Se
quencing. and Synthesis Facility (East Lansing. MI). The stock
solutions had concentrations of 50 pmol/pL and were used without
purification. The 6-aza-2-thiothym1ne (ATT). 5-methoxysalicylic
acid (MSA. an oligonucleotide matrix developed at Michigan State
University“). and L-fucose were purchased from Aldrich Inc
(Milwaukee. WI). Spermine (sp) and 3hydroxypicolinic acid
(HPA) were purchased from F luka (Milwaukee. WI). All matrixes
were used without further purification. The aqueous fucose
solution was made at a concentration of 50 mM. When spermine
was used as a matrix additive. it was prepared at a concentration
of 25 mM in water. Saturated matrix solutions were made using
a 1:1 acetonitrilezspermine solution or 1:1 acetonitriiezwater.
Samples were prepared by mixing 1 pL of the analyte solution
with 1 pl. of the fucose solution and l aL of the comatrix solution
on a gold sample plate. The mixture was then allowed to airdry.

A Sequazyme oligonucleotide sequencing kit containing the
enzymes and buffers necessary for the digestion of oligonucle
otides for subsequent analysis by MALDI-MS was purchased from
PE Biosystems (Framingham. MA). HPA. snake venom phos-
phodiesterase (SVP). bovine spleen phosphodiesterase (ESP). and
the standard oligonucleotide d(AGGCATGCAAGC’ITCAGTAT-
TCT AT) were diluted according to the instructions provided in
the kit. Several dilutions of the SVP and 88? were made in order
to provide a range of sequence coverage. The procedure for
exonuclease digestions has been described previously." Briefly.
ammonium citrate buffer (pH = 9.4) was used for SVP digestions.
and BSP reaction buffer was used for BSP digestions. A 1-,uL
portion of the oligonucleotide solution was diluted to 3 aL in a
microtube. The manufacturer of the sequencing kit suggests that

 

(9) Li. Y. C. 1.: Cheng. S. W.; Chan. T. W. D. Rapid Commun. Mass Spectrom.
1998. 12. 993-998.
(10) Pieles, U.; lurcher. W.; Schilr. M.; Muser. H. E. Nucleic Acids Res. 1993.
21. 3191-3196.
(1 i) Smirnov. l. P.; Roskey, M. T.; Juhasz, P.; Takach. E. J.; Martin. S. A.; Hall.
1.. A Anal. Biochem. 1996. 238. 19-25.

198

 

A) ATT/3p
lM-Hl'
II 1 _ A L

Relative Intensity

 

 

B) A'I'r/sp/tucose
We
V7"? y.
L LL: LL -'i . I; Y
1800 2300 2300 3300 3800

mlz

Figure 1. MALDI-MS negative ion mass spectra of d(CGCGAAT-
TCGCG) using (A) ATT/spermine (sp) as a matrix and (B) ATT/sp/
fucose as a matrix.

100-400 picomoles of material are needed for sequence analysis
using an exonuclease. To the microtube. 1 aL of the diluted
enzyme was added. along with 1 pL of the appropriate buffer. The
solutions were Incubated at 37 °C for 20 min before analysis by
MALDI-MS.

Linear MALDI mass spectra were recorded on a PerSeptive
Biosystems (Framingham, MA) Voyager delayedextraction time-
of-flight linear (TOF) mass spectrometer equipped with a nitrogen
laser (337 nm. tins pulse). Typically. 50 laser shots were averaged
for each spectrum. For the negative ion MALDI spectra reported
here. the accelerating voltage was -15 kV. the delay time was
700 ns.‘ the grid voltage was 94.5% of the accelerating voltage.
and the magnitude of the guide wire voltage was 0.20% of the
accelerating voltage. We have previously reported that longer
delay times improve resolution for oligonucleotides.a No changes
in the extent of fragmentation are observed when longer delays
are used.

RESULTS All) DISCUSSION

The behavior of oligonucleotides in the MALDI experiment
differs greatly from that of peptides. Although a pure sample of a
peptide usually yields a single peak representing the intact
molecule. oligonucleotides frequently undergo prompt fragmenta-
tion in linear TOF mass spectrometers.‘2 Although fragmentation
can be very useful In structure determination. it is often a
complication. especially when dealing with mixtures. The literature
showsm‘ that the Dickerson dodecamer. d(CGCGAATTCGCG),
has been extensively studied. When employing ATT. a matrix
typically used for MALDI analysis of oligonucleotides. the Dick-
erson dodecamer fragments extensively. as shown In Figure 1 A.
The peaks representing fragment ions are of greater intensity than
the peak representing the intact. molecular ion. The most intense
peak corresponds to the w. fragment at m/z 2489. When fucose
is used in combination with ATT. the extent of fragmentation is
reduced (Figure 18). The molecular ion peak is now the most
intense peak In the spectrum. Decreased fragmentation suggests

 

(12) Currie. C.: Yates 111.]. 1. Am. Soc. Mass Spectrom. 1993. 4. 955-963.
(13) Drew. H. R.; Wing. R. M.; Takano. T.; Broke. C.: Tanaka. 5.: ltakura. K.;
Dickerson. R. B. Proc. Natl. Acad. Sci. USA. 1981. 78. 2179-2183.

(14) Chiu. T. K.; KaczorCrzeskowiak. M.; Dickerson. R E. 1. Mol. Biol. 1999.
292. 589-608.

Analytical Chemistry. Vol. 73. No. 20. October 15. 2001 5001

 

ATT + fucose

ATT

Intensity

 

A

11 16

Fashion

Piglet. lntensityofthem - H]' peakfordmiaplottedagainst
position across the sample spot. The thinner line represents the
sample spot created using the matrix ATT without additives. and the
linehboldrepresentsthesamplespotwlthATTasthematrtxand
hiooseasanadditive.‘l'heratioottheareasunderthetwowrvesls
3.6.

that either less energy is deposited Into the analyte in the
desorption/ ionization process. or it is subsequently collisionally
cooled. In addition to a reduction in the extent of fragmentation.
there is an increase In resolution for the [M — Hl‘ peak from
400 (fwhm) without fucose to 600 (fwhm) with fucose. We do note
that the oligonucleotide samples were used as supplied from the
Michigan State University Macromolecular Structure. Sequencing,
and Synthesis Facility. As synthesized. these samples had a high.
but poorly defined. salt content and contained other contaminants.
When analyzed using ATT alone. the molecular Ion peak for the
oligonucleotide has a high mass tail that extends to an m/z value
of 3800 as a result of alkali ion adduction. As shown In Figure
IAB. the addition of spermine to the matrix eliminates the need
for purification of the oligonucleotide solution. The performance
of spermine has been discussed previous .33

In addition to spectral changes. fucose affects the MALDI
target homogeneity. Often. target crystals created using the dried
droplet method" are inhomogeneous. with only certain portions
of the target yielding strong signals. The homogeneity of the
MALDI target is of growing importance as the extent of automa-
tion in MALDI analysis increases. Elaborate methods are being
developed for increasing spot uniforrnlty. Including the use of
electrospray techniques to produce a uniform distribution of
crystals.“ The addition of fucose to the matrix solution provides
a simple alternative. To demonstrate this. two MALDI targets were
made by the dried droplet method. with one using ATT as the
matrix and the other using ATT/fucose. Spectra were taken from
16 equally spaced sites across the span of each well. At each
location. spectra from 50 laser shots were averaged and recorded.
All experiments were performed at the same laser power. The
oligonucleotide 00712 was used In this experiment. because only
a peak representing the Intact molecular Ion Is detected. The
intensities of the [M - Hl’ peak In each spectrum for both the
target made with ATT and that with ATT/fucose were plotted
against the position In the well. shown In Figure 2. At halfof the
IocadorrsintlreATTtargeteithernosignalorweaksignalsoould
be detected. The data shown for ATT in Figure 2 Is not

 

(15) Karas. M.; Hillenkamp. F. Anal. Chem. 1988. 60. ROI-23%.
(16) Hensel. R. R.; King. R. C.: Owens. K. C. Rapid Commun. Mass Spectrom.
1997. II. 1785-1193.

5002 Analyflcal Chenilstry, Vol. 73. NO. 20. October 15. 2001

199

ATT/fucose

Intensity

 

 

5 1 0 11 13
Accumulated spectra
Figure I. lntensityoithe[M - Hl‘ peakofdmuplottedagainst
the number of accumulated 50-shot spectra from a single location.
The unbolded line represents analysis of the target created using the
matrix ATT without additives. and the line in bold denotes the target
madewithATTasthematrixandtucoseasanaddltive.Thearea
under the curve created using ATT/tuoose is 3.5 times that for the

curve generated using ATT alone.

1 3

unexpected. indicative of the 'doughnut'-shaped ring of crystals
that typically forms. This was not the case In the ATT/fucose
target. At every location across the well. strong signals were
achieved. The strongest signal from the ATT target was of
comparable intensity to the weakest signal from the ATT/fucose
target Similar results were achieved In experiments performed
with HPA and HPA/fucose targets. With a homogeneous sample
spot. automation could more easily be used In analysis. because
a spectrum can more likely be obtained from any default starting
position. The amount of time necessary for analysis would also
be greatly decreased.

In addition to an improvement in crystal uniformity. signal
duration Is Increased with the addition of fucose to the sample
spot. Again. the oligonucleotide d(T) .2 was used to document this
aspect A location was selected in the target that gave ample signal.
The intensity of the molecular ion peak was then recorded from
the spectrum corresponding to the sum of 50 laser shots. Spectra
were taken repeatedly. with each set of 50 transients resulting In
a spectrum from the same sample location on the MALDI target
until the signal-tonoise ratio was <2. These results are shown in
Figure 3. The sample spot with ATT alone allowed for 4 spectra
to be acquired without changing location. but 10 acceptable
spectra (500 shots) were acquired from the sample spot with ATT/
fucose. An increase In signal duration in the ATT/fucose target
can explain one of the aspects of the results seen In Figure 2.
With a prolonged signal. the peaks In the ATT/fucose spectra
will be more intense than those spectra resulting from the ATT
target.

The Increase In both sample spot homogeneity and signal
duration. as well as the reduction in the extent of fragmentation.
results In an increase In sensitivity. Increased sensitivity is
Important when analyzing mixtures. such as those from exonu
clease digestions. Sequencing using exonucleases is an Important
application for the MALDI mass spectrometric analysis of oligo
nucleotides. When using the exonuclease sequencing method.
enzymes of various concentrations are used in order to maximize
sequence coverage for a range of oligonucleotide sizes. Each
digestion reaction yields a set of several oligonucleotides of

A) HPA

 

[— B) A‘l'l’lsplfucc
G

Relative Intensity
>
>

x10
I C C) MSA/splfuo

   
 

 

1100 2100 3100 4100 5100

Fig-'0 4. MALDI-MS negative ion mass spectra of the digestion
products of d(AGGCATGCAAGCTTGAGTATTCTAT) using the ma-
trixes (A) HPA. (B) ATT/sp/iucose. and (C) MSA/sp/iucose as in the
original document and shown in Fig. 5.

varying length. MALDI analyses were performed on the same
digestion products with various combinations of matrixes and
matrix additives. When HPA. MSA/spermine. and ATT/spermine
were used as matrixes without fucose. 7 digestion products were
detected. The spectrum obtained using HPA Is shown In Figure
4 A. When fucose Is added to the ATT/spermine sample. 10
digestion products are easily detected (Figure 48). When the
same digestion mixture is analyzed using MSA/spermine/fucose.
IZ digestion products are clearly seen (Figure 4C). With the
increase In number of digestion products detected. more sequence
Information is provided from each reaction. and fewer enzyme
dilutions are necessary to achieve complete sequence coverage.
Also note that although the sequencing kit recommended that
100-400 pmol be used In the analysis. only 50 pmol was used
here. This explains the low signal-tonoise ratio In Figure 4A.
Fucose offers increased sensitivity. hence. an Increased signal-
tonoise ratio.

With the addition of fucose. several spectral changes are seen
when analyzing doublestranded oligonucleotides as well. As
shown in Figure 5. the detection of the duplex differs when fucose
is included In the matrix crystal. In Figure 5A. the doublesuanded
oligonucleotide Is detected using ATT/spermine/fucose as a
matrix. but in Figure 58. the analyte Is detected using only A'I'I‘/
spermine. In Figure 5A. the addition of fucose to ATT/spermine
during sample preparation increases the intensity of the duplex
Mold relative to the Intensity of the peaks representing the single
strand for the same laser power.

The addition of fucose to the sample can increase resolution.
as well. This was realized when analyzing a doublestranded
oligonucleotide composed of single strands that differ in molecular
weight by 9 Da. When analyzing the duplex d(AAAAACCAAAAA)
annealed to d(TTI'I’I‘C-CTI'ITI'). only one peak representing one
of the singlestranded species is detected when the matrix ATT/
spemrine is used (spectrum not shown). When the duplex is
analyzed. again using ATT as the matrix with both fucose and
spermine as additives. the resolution Improves from 400 (fwhm)
to 700 (fwhm) to allow for the clear detection of both the TC-
containing strand as well as the ACcontaining strand. The peaks
representing the singlestranded species are also of equal Intensity
when fucose Is used. This experiment demonstrates an Improve-

200

A) ATT/splfucose

(Mer‘Hl

 

 

Relaflve htemIty

B) ATT/3p

OM

7000 7500 8000 8500 9000 9500 10000
mlz

Figure 5. MALDI-MS negative ion mass spectra of the double-
stranded oligonucleotide. d(CGCGAATTCGCG) bound to its comple-
ment. d(TTCGCGAATl’CGCG). In (A). when ATT/spliucose matrix
is used. the intensity of the duplex ions increased 3—fold in comparison
to (B) the spectrum taken using only ATT/3p as the matrix.

 

ment in resolution. as well as a more uniform response for all
components. with the addition of fucose to the MALDI target.

CONCLUSIONS

As the MS analysis of single nucleotide polymorphisms (SNPs)
develops. the improvement of samplespot homogeneity In MALDI-
MS will be essential. Because of the large number of samples
required. SNP detection is becoming a highly automated process.
from the creation of the sample spot to analysis of the spot by
MALDI-MS.l7 With this automation. there Is a need for an
Increased probability of successfully acquiring a spectrum from
all positions on a sample spot. With an increase In the crystal bed
homogeneity through the uses of fucose. analysis time will be
decreased. In addition. the increase In sensitivity Improves the
probability of successful analyses.

When analyzing molecules using MALDI-MS. the ability to
detect the analyte is reflected in the signal-to-noise ratio of the
experiment. We find that signals are consistently higher In the
analysis of oligonucleotides when fucose Is used as a matrix
additive. To document this. two MALDI targets were created
which contained the same amount of an oligonucleotide. The first
target was made using ATT; the second. with ATT/fucose. For
each. 10 random locations In the target were selected. From each
location. 10 spectra that were the sum of IO transients were
generated. and these 10 spectra combined to give a single
spectrum. When fucose Is added. the noise is unchanged. but the
signal for the oligonucleotide Is an order of magnitude larger.
From our experiences and the data shown In Figures 2 and 3. at
least an order of magnitude Increase in signal Is expected.

Received for review June 29. 2001. Accepted August 20.
2001.

A0015550+

 

(IT) LI. J.; Butler. J. M.; Tan. Y.; Lin. H.; Royer. S.: Ohler. L.; Shaler. T. A.;
Hunter. J. M.; Pollart. D. J.; Monforte. J. A.; Becker. C. H. Elxtrophoresrs
1999. 20. 1258—1265.

Analytical Chemistry. Vol. 73. No. 20. October 15. 2001 5003

APPENDIX C

201

 

Additives for the Stabilization of Double-
Stranded DNA in UV-MALDI MS

Anne M. Distler and John Allison

Department of Chemistry, Michigan University, East Lansing, Michigan, USA

 

Molecular complexes such as double-stranded oligonucleotides contain non-covalent bonds
that are difficult to maintain in the MALDI experiment. Quantifiers are introduced in order to
evaluate, summarize, and compare spectra from experiments in which additives are used to
stabilize duplex oligonucleotides. Compounds known to complex with and stabilize duplex
molecules can be useful as additives in MALDI. Spermine and methylene blue, present at
concentrations similar to the matrix, are detected, bound to the duplex. When peptides are
used as additives, the duplex is stabilized when the peptide is present at an amount less than

that of the duplex.
for Mass Spectrometry

(I Am Soc Mass Spectrom 2002, 13, 1129-1137) © 2002 American Society

 

 

ltraviolet matrix-assisted laser desorption / ion-
ization mass spectrometry (UV MALDI MS) [1]
has become an important technique for the
analysis of biomolecules such as peptides and oligonu-
cleotides. While analysis of oligonucleotides initially
lagged behind peptide analysis, the development of
new matrices and matrix additives has made significant
improvements in oligonucleotide characterization by
MALDI MS [2-7]. The first additives used with oligo-
nucleotides were ammonium salts such as ammonium
hydrogen citrate [8] and ammonium fluoride [9, 10]. It
is assumed that the additive alters the charge state or
environment in which the analyte exists in the target
crystals. In addition to the elimination of multiple
alkali-ion adduction. the ammonium salt seems to play
a significant role in enhancing both the desorption and
the ionization of intact oligonucleotides [9]. Recently,
other additives have been developed for the MALDI
analysis of oligonucleotides, notably the tetraamine
spermine [2, 3], related amines [11], and fucose [4, 5].
Through the use of additives, UV MALDI MS has
been used successfully to study many types of covalent
molecules. However, molecular complexes containing
non-covalent bonds are more difficult to analyze in the
UV MALDI experiment. Using a technique such as UV
MALDI MS, a double-stranded oligonucleotide can be
desorbed and ionized as a singly charged, gas phase
ion, although only a few examples have been reported
[12-14]. While oligonucleotides used in this study are
12—14 bases in length, these duplexes can serve as
models for DNA / drug binding studies. Duplexes of
low molecular weight are particularly useful because

 

Address reprint requests to Dr. J. Allison, Chemistry Department. Michigan
State University, East Lansing, MI 48824, USA. E-mail: allisonflcem.
msu.edu

© 2002 American Society for Mass Spectrometry. Published by Elsevier Science Inc.

1044-0305 / 02/ $20.00
PI] SlO44—0305(02)00430—0

the resolution in a MALDI-time-of-flight (TOF) MS
experiment is sufficient to detect small, organic mole-
cules bound to the duplex. Also, they represent an
analytical challenge of mass spectrometry, representa-
tive of non-covalent complexes, for which techniques
could be developed to stabilize them.

When a double-stranded oligonucleotide, Mle, is
analyzed in negative-ion MALDI-TOF MS from a sim-
ple matrix / analyte target, only ions representing the
two single strands, (M1 - H)’ and (M2 - H)’, are
usually detected [15]. Since M1M2 is not detected as
(MINI; - H)’ or (MIMZ + I~I)+ in UV MALDI MS, the
duplex must dissociate at some point before accelera-
tion and detection of the ions. There are many steps in
the MALDI experiment where the double-stranded
species could denature including the initial formation
of the MALDI target and the desorption/ ionization
(D/ I) process. The presence of stabilizing additives in
the experiment may maintain the non-covalent complex
throughout all phases of the experiment.

There are several compounds known to complex
with and stabilize double-stranded DNA at the cellular
level. In cells, DNA is negatively-charged due to the
ionization of the phosphate groups in the phosphodi-
ester backbone. The phosphate groups can be affiliated
with counter-ions such as sodium ions, magnesium
ions, and polyamines such as spermine [16, 17]. The
polyamines are positively charged at physiological pH
and bind to DNA, shielding the negative charges and
decreasing the repulsion between the strands. Spermine
allows double-stranded DNA to condense into a more
compact structure in cells [18]. When oligonucleotides
are crystallized for X-ray crystallographic analysis,
spermine is frequently added to facilitate crystal growth
[18, 19]. Spermine may also stabilize the duplex in the
MALDI experiment.

While not naturally found in cells, several small,

Received April 5, 2W2
Revised May 15, 2002
Accepted May 20, 2002

202

 

1130 DISTLER AND ALLISON

organic compounds are known to bind to the major and
minor grooves of double-stranded oligonucleotides or
intercalate between the bases. The presence of such
species may stabilize the duplex through the MALDI
crystal growth and desorption/ ionization processes.
One example is ethidium bromide. Ethidium (Et) is a
positively-charged species that is known to intercalate
between the bases of duplexes. Ethidium bromide has
been found to stabilize the double-stranded duplex
poly(dT) bound to poly(dA), increasing the melting
temperature by 14 °C [20, 21].

While certain small molecules are known to stabilize
double-stranded DNA, larger molecules such as pro-
teins and peptides may also have a stabilizing effect on
duplexes. Interactions between DNA and proteins play
important roles in many biochemical processes. Chro-
mosomes contain DNA and an equal mass of histone
and non-histone proteins [22]. Histories, molecular
weights between 13,000 and 30,000 g/mol, are highly
conserved across species and contain many basic resi-
dues. These positively-charged amino acid side chains
interact with and stabilize the DNA through salt
bridges.

Similar to histones, protamines are arginine-rich pro-
teins found in sperm cells. Cationic protamines bind to
DNA with their a—helices in the major groove of the
DNA, where they enable the tight packing of the
duplexes [23]. While the main driving forces for prota-
mine-DNA binding are the ionic interactions, there is a
distinct cooperativity that cannot be explained on the
basis of electrostatic interactions alone [24]. Protamines
perform functions similar to histones, but are generally
smaller, with molecular weights between 4000 and
10,000 g/mol [25]. Non-covalent complexes between
proteins and oligonucleotides have been studied in UV
MALDI previously [26]. However, the effect of the
presence of peptides on the spectra of double-stranded
oligonucleotides has not been examined.

There is no new matrix or additive that allows for the
complete conservation of the intact duplex in UV
MALDI MS. For this reason, insights must be extracted
from the smaller spectral changes that can be measured
when the experimental variables are changed. We in-
troduce here three simple quantifiers that allow for UV
MALDI spectra of double-stranded oligonucleotides to
be evaluated. summarized, and compared. Using these
quantifiers, we demonstrate that the presence of stabi-
lizing additives has a measurable effect on the duplex
region of the resulting UV MALDI mass spectra. The
quantifiers allow one to maintain a numerical focus on
the goals in these measurements, possibly leading to the
development of a blended matrix (with multiple addi-
tives), in which components each serve to chemically
address different aspects of this system.

Experimental

The double-stranded oligonucleotides used in this work
were purchased from the Michigan State University

J Am Soc Mass Spectrom 2002, 13. 1129-1137

Table 1. Oligonucleotides used in this work

 

 

Average
molecular
Name Sequence weight T...‘
Duplex 1 5‘-CCGGAA1TGGCC-3' 3646
3'-GGCCTTAACCGGTT-5' 4254 40 °C
Duplex 2 5'-ACCCACCCACCC-3' 3480
3'-TGGGTGGGTGGG-5' 3813 42 ’C
Duplex 3 5'-AAAAACCAAAAA-3' 3648
3'TlTlTGGTTI'l'T-5' 3638 28 °C

 

'Tm ie the melting temperature at which the duplex dissociates.

Macromolecular Structure, Sequencing, and Synthesis
Facility (East Lansing, MI) and are shown in Table 1,
with their molecular weights and melting temperatures
(T I,\). They are all stable in solution at room temperature
with melting temperatures ranging from 28 to 42 °C.
The presence of the duplex in the initial solutions was
confirmed by UV analysis at 259 nm.

The peptides, dynorphin A, fibrinogen binding in-
hibitor peptide (FBI peptide), B-melanocyte stimulating
hormone (B-MSH), katacalcin, kemptide, trilysine, N-e-
acetyl-lysine, leucine enkephalin, Boc-MNF-arnide,
hexaalanine, trialanine methyl ester, and triserine were
all purchased from Sigma (St. Louis, MO) and used
without further purification. Peptides were dissolved in
MilliQ water at concentrations of 1.0 pmol/51.1...

The stock duplex solutions had concentrations of 25
pmol/uL. The compounds, 6-aza-2-thiothymine (ATT),
chromomycin, daunomycin, ethidium bromide, meth-
ylene blue, distamycin, and Hoechst 33258 were pur-
chased from Aldrich Inc. (Milwaukee, WI). Spermine
(sp) and 3-hydroxypicolinic acid (HPA) were purchased
from Fluka (Milwaukee, WI). When spermine was used
as a matrix additive, it was prepared at a concentration
of 25 mM in water. The stock solutions of ethidium
bromide and methylene blue were 1 mM in water.
Saturated matrix solutions were made using a 1:1
acetonitrile/spermine solution or 1:1 acetonitrile/wa-
ter. For experiments performed with ethidium bromide
and methylene blue, the oligonucleotide solution was
mixed with an equal volume of the additive solution
prior to spotting on the sample plate. For those exper-
iments using peptides as additives, a microliter of a
peptide stock solution was combined with a microliter
of the stock double-stranded oligonucleotide solution.
When incubated, the peptide/oligonucleotide solution
was heated to 37 °C for 15 min. A microliter of the
resulting solution was spotted onto the MALDI target
with a microliter of the matrix solution.

Linear MALDI mass spectra were recorded on a
PerSeptive Biosystems (Framingham, MA) Voyager de-
layed-extraction time-of-f'light (T OF) linear mass spec-
trometer equipped with a nitrogen laser (37 nm, 3 ns
pulse). For the negative ion MALDI spectra reported
here, the accelerating voltage was -15 kV, the delay
time. selected for optimum resolution, was 700 ns [3],
the grid voltage was 94.5% of the accelerating voltage,

203

J Am Soc Mm Spectrom 21112.13, 1129—1137

 

 

 

 

 

a
. (Maw
(Mr'H) x 10
E I: (M.M,-H)'
L . a h ‘ AL A;
i b I
x 10

3000 4000 5M 0000 7000 8000 9000
ml:

Figure 1. The negative-ion UV MALDI mass spectra of duplex 1
using (a) ATT and (b) HPA as matrices. Duplex 1 consists of two
strands with molecular weights of 3646 g/mol and 4254 g/mol.
The strand with the lower molecular weight is referred to as M,
while the strand with the larger molecular weight is referred to as
M2. N0 duplex ions are detected (position indicated by arrow)
giving a DSRR of 0.

and the magnitude of the guide wire voltage was 0.20%
of the accelerating voltage. Typically, transients from 50
laser shots were averaged for each spectrum.

Melting temperature studies and UV analyses were
performed using an ATI Unicam (Cambridge, UK) UV2
spectrophotometer. Oligonucleotides have strong UV
absorption maxima at 259 nm. This absorption arises
almost entirely from the complex electronic transitions
in the purine and pyrimidine components [23]. In
double-stranded oligonucleotides, the base-base stack-
ing results in a decrease in molar absorptivity compared
to that for the two single-stranded components. This is
known as the hyperchromic effect. For melting studies,
UV spectra were taken using the oligonucleotide solu-
tions at room temperature. The analyte solutions were
then heated to 90 °C in a sand bath for 10 min in order
to denature the double-stranded oligonucleotides. If the
duplex existed at room temperature, the absorbance
should increase by 10% due to the unstacking of the
aromatic bases upon denaturation. To insure renatur-
ation of the duplex, the absorbance can again be re-
corded after the solution returns to room temperature.
Allspectrawereacquiredfrom200t0400nmwitha
scan speed of 120 nm/min, a data interval of 0.5 nm,
and a 2.0 nm bandwidth. Quartz cuvettes with a 1 cm
path length were used, with water in the reference
cuvette.

Results and Discussion

When analyzing duplex 1 using ATT and HPA as
matrices, only peaks representing the two single strands
are detected, Figure 1a and b respectively. While these
matrices have previously shown promise for the detec-
tion of non-covalent complexes in UV MALDI MS [12,
13], the duplex is not preserved in these experiments.
This may be related to size differences between the two

STABILIZATION OF DOUBLE-STRANDED DNA 1131

experiments. Similar results are achieved with duplex 2;
no intact duplex peaks are detected even though these
duplexes have melting points which indicate that they
are stable at room temperature. When analyzing duplex
3, a small peak representing the duplex with a loss of a
guanine is seen. No peak is detected representing the
intact (Mle — H)” species. All the experiments were
completed with the same level of double-stranded
oligonucleotide. We have examined the correlation be
tween the amount of duplex deposited on the target
and the double-stranded region of the spectrum and
have found no influence, when working in a normal
range for MALDI samples (1-100 pmol of analyte).
Due to the negatively charged phosphate backbone,
negative—ion mode is often used to study oligonucleo-
tides in MALDI MS and will be used in this discussion.
When analyzing the annealed double-stranded species,
M.Mz, in negative-ion mode, ions representing the
single strands, (M, — H)’ and (M2 — H)‘, dominate the
spectrum. Negative—ion spectra exhibit better resolution
and a higher signal-to-noise ratio. In order to measure
the success of an experiment, several criteria will be
used when an additive allows for the detection of
duplex ions. The strand with the lower molecular
weight will be denoted as M1 while the larger comple-
mentary strand will be denoted as M2. The ratio of the
intensity (I) of the (M1 - H)" peak to the intensity of the
(M2 - H)” peak is of interest. In previous experiments,
our laboratory has analyzed double-stranded oligonu-
cleotides in the UV-MALDI experiment where only ions
from one of the strands are formed. This result is
unexpected since the complementary strands are
present in equal amounts and areof comparable molec-
ular weights. This effect appears to depend on both the
sequence of the oligonucleotides and the matrix used in
the experiment. The ratio of the intensity of the (M1 —
H)‘ peak to the intensity of the (M2 - H)‘ peak is
referred to as the single strand distribution (SSD), eq 1.

_I_("_"1_)
SSD-[(M2) (1)

Note that the nomenclature used in eq 1 is not, tech-
nically, correct. We chose to write [(M1) rather than
1(M, - H)“, for example, to allow the quantifier to be
used in either negative or positive ion experiments.
To the extent that the single strand ions may evolve
from the duplex, we would like to monitor how the SSD
value correlates with other variables in the experiment,
and to determine if the SSD correlates with single
strand composition. If, for example, the single strand
ions are fragments of the singly charged gas phase
duplex, then one may expect that the decision of which
strand retains the charge would be determined by the
base composition of the individual strands. On the
other hand, if duplex dissociation occurs early in the
experiment, one may anticipate an SSD of 1.0. For the
spectra shown in Figure 1a and b, the SSD values are

204

1132 DISTLER AND ALLISON

 

 

 

 

 

E

 

 

3400 4400 5400 64m 7400 0400 9400

nth
Figure 2. The negative-ion MALDI mass spectrum of duplex 1.
HPA was used as the matrix with ethidium (Et) bromide as an
additive. Ethidium adducts of both the single and double-
stranded oligonucleotides are observed. The SSD is 1.0, DSRR is
0.05, and the WCSF is 0.5. The additive duplex ratio is 40.1.

both 0.9. That is, D/ I of the larger strand is slightly
favored.

The goal of this project is to optimize the ratio of the
intensity of the double strand peak to the intensities of
the single strand peaks. This double strand retention
ratio (DSRR) is defined in eq 2.

= 1(MiM2)
KM.) + 1(M1M2) + 1(Mz)

While the SSD values were 0.9 in Figure 1a and b, the
DSRR was zero for both experiments. The most intense
analyte-related peaks in the spectrum represent the
(M, — H)“ and (M2 - H)“ ions. The duplex must have
dissociated before the acceleration and detection of the
ions. The presence of stabilizing additives in the target
may hold the complex together. Several experiments
were performed using stabilizing additives in order to
study their effect on the SSD and DSRR values for each
s

 

DSRR (2)

As mentioned previously, ethidium bromide has
been found to stabilize duplexes in solution. When used
in the MALDI experiment, equal volumes of duplex 1
and ethidium bromide solutions were combined. One
microliter of the oligonucleotideethidium solution was
spotted on the MALDI target with one microliter of the
HPA matrix solution. The molar amounts of compo—
nents of the target are matrixzadditivezoligo, 2000:40:1.
This is typical since effective additives for MALDI MS
are usually present at molar amounts greater than the
analyte. The resulting spectrum is shown in Figure 2,
with the structure of the ethidium ion. The ethidium
cation binds to both the single and double—stranded
species of duplex 1. That is, in addition to analyte ions,
(A - H)“, there are also (A - 2H + Et)“ ions observed.
With the appearance of the dimer peaks, the DSRR
increases from 0 (in Figure 1b) to 0.05.

While the DSRR increases with the addition of

I Am Soc Mass Spectrom 2002,13,1129-1137

ethidium bromide, the goal of this work is to retain the
maximum extent of Watson-Crick base pairing when
detecting a double-stranded species. If the M,Mz du-
plex completely dissociates during crystal formation,
some fraction of M, and M2 may recombine in a
non-Watson-Crick pairing. The single-stranded species,
M, and M2, can, and clearly do, form three different
dimers: M,M,, M,M,, and M2M2. In order to quantify
the amount of Watson-Crick pairing involved, we will
calculate a Watson-Crick selectivity factor (WCSF), eq 3.

1(M1M2)

WCSF “ I<M.M.) + I(M.M2> + I<M2Mz)

 

(3)

If the experiment begins with pure duplex and com-
plete dissociation occurs, during crystal growth, then
there will be equimolar amounts of M, and M2 available
in the solution on the sample plate. If random dimer
formation occurs, a WCSF value of 0.5 would be ex—
pected. This suggests that the distribution of duplex
peaks was dictated by random condensation events.
Even in peptide analysis, non-specific dimerization has
been observed [27]. As the value of the WCSF ap-
proaches 1, Watson-Crick base pairing is preserved and,
presumably, the non-covalent forces were maintained
throughout the MALDI experiment. When changing the
variables in the MALDI experiment, the value of the
WCSF provides information about the nature of the
species detected and gives direction to subsequent
experiments.

In Figure 2, while there are peaks representing the
(M,Mz - H)“ species, there are also peaks representing
the (M,M, - H)“ and (MzMz — H)“ species. While the
DSRR increased from 0 to 0.05, the WCSF is 0.5,
suggesting random dimerization. If random dimeriza-
tion occurs, the duplex may completely dissociate and
the single strands are likely dimerizing with no se-
quence specificity as the crystals are growing. Thus,
while ethidium addition does result in ”dimer ions”, we
conclude that this additive does not stabilize the initial
duplex throughout the UV-MALDI experiment.

Other compounds known to bind to double-stranded
DNA [23] were evaluated as matrix additives as well
including chromomycin, distamycin, berenil, daunomy-
cin, and methylene blue. Of these compounds, distamy-
cin and berenil were found to bind to the single-
stranded oligonucleotides in the UV-MALDI
experiment. There were, however, no duplex or dimer
peaks detected, giving a DSRR of zero for both addi-
tives. The SSD was 0.7 for the berenil experiment and
1.0 for the distamycin experiment. Chromomycin and
daunomycin had no apparent effects on the experiment.

Methylene blue was also found to bind to the single—
stranded oligonucleotide. The experimental procedure
used for the ethidium bromide experiments was fol-
lowed using methylene blue as the additive. The solu-
tion was then spotted with ATT as the matrix. The
resulting spectrum is shown in Figure 3, with the

205

I Am Soc Mass Spectrom 2W2,13,1129-1137

 

 

 

 

("1'”)
(Ma-HT I N\
3 new 3 on
'5”: in
x10
0 a
3500 «no sEoo Joe '1st Son

mlz

Figure 3. The negative-ion MALDI mass spectrum of duplex 1
with methylene blue as an additive. ATT was used as the matrix.
The SSD is 0.9, the DSRR is 0.03, and the WCSF is 1.0. The additive
duplex ratio is 4&1.

structure of the methylene blue cation. While methylene
blue (MB) is observed bound to the single-stranded
species, its presence appears to stabilize the duplex as
well, Fi 3. In this experiment, the DSRR value was
0.03, the SSD was 0.9, and the WCSF was 1.0. With a
WCSF of 1.0, it appears that a portion of the duplex was
maintained throughout the entire UV-MALDI experi-
ment, due to the presence of this additive, although the
duplex represents only a small percentage of the initial
complex.

In addition to intercalating and groove-binding mol-
ecules, spermine was also found to be an effective
additive. When spermine is used with ATT, there are
significant changes in the double-stranded region of the
spectrum. Figure 4a shows the spectrum of duplex 3
when ATT and spermine are used as a matrix and
Figure 4b shows a spectrum of the same duplex with
ATT as the matrix. Without spermine, a peak represent-
ing a duplex fragment, due to the loss of a guanine, is

a ATTISD (Mi Mi”).

(M.M.«sr

i bATT

0000 6st 7000 7500 8600
M

Figure 4. Portions of the negative-ion MALDI mass spectra of
duplex 2 using (a) ATT/5p and (b) ATT as the matrix. Note the
intact duplex ion is formed when spermine is present while only
a duplex ion fragment is formed with ATT alone. In the ATT/3p
experiment, the DSRR is 0.10 and the WCSF is 1.0. The SSD is 1.0
for both experiments. The additivezduplex ratio is 50m.

 

 

STABILIZATION OF DOUBLE-STRANDED DNA 113

seen. When spermine is present, two additional peaks
are seen. The intact, deprotonated duplex, (M,Mz -
H)“ is detected as well as its spermine (sp) adduct,
(M,Mz - H + sp)“. Compared to Figure 1a, the DSRR
increases from zero to 0.1. The SSD values are 1.0 for
both experiments. The WCSF value for this spectrum is
l, with only high mass peaks relating to the M,Mz
species being detected.

While spermine may contribute to duplex stability in
the crystal growth event, the detection of (M,Mz - H +
sp)“ suggests that a spermine complex was formed
and remained intact through the D / I event. The effect
is duplex-dependent. Spermine stabilizes duplex 3,
but when ATT:sp is used in the analysis of duplex 1
or 2, ions representing the intact duplex are not exclu-
sively detected; peaks representing (M,M, — H)“ and
(MM, - H)“ are detected as well.

In Figure 4a, while most of the duplex dissociated,
some duplex ions were detected in the MALDI analysis.
The addition of spermine to the matrix has, at some
point in the experiment, allowed for the detection of
what appears to be authentic double-stranded DNA
ions with Watson-Crick base-pairing preserved. The
improvement in resolution when spermine is added
may suggest that spermine-duplex adducts are trapped
in the crystals. This may lower the average number of
negative charges on trapped oligonucleotides, simplify-
ing desorption kinetics.

While spermine stabilizes the duplex during the
MALDI experiment, it also acts to effectively displace
alkali ions, decreasing the abundance of sodium and
potassium adducts [2]. Spermine may serve an impor-
tant role in the eventual development of a mixed matrix
for duplex analysis. Addition of spermine is preferable
to sample purification. In analyzing other biopolymers
by MS, it is certainly common to purify by, for example,
ion exchange if a sample has a high salt content.
However, when the analyte is a DNA duplex, this may
not be a logical procedure to use. Ions such as Na“ and
K+ stabilize duplexes; melting points are lowered when
salts are removed [28]. Thus, desalting may well de-
crease the success of a duplex analysis by UV-MALDI
MS. Spermine allows for some salt to be present, and
displaces alkali ions as the target is formed.

In the evaluation of spermine as an additive, Figure
4, the double-stranded oligonucleotide is detected in
both experiments. In Figure 4b, the presence of salts
interferes with the analysis. The duplex oligonucleotide
may form many alkali ion adducts leading to decreased
resolution and signal—to—noise ratio. The presence of
spermine in Figure 4a leads to an improvement in the
resolution. The presence of spermine in the matrix
solution may also improve the results of experiments
using other additives such as methylene blue. If the
experiment in Figure 3 is repeated with a matrix solu-
tion containing spermine, the resolution in the duplex
region is improved, the DSRR value increases, and a
methylene blue adduct is detected in the duplex region,
Figure 5. The DSRR increases significantly from 0.03 in

206

 

1134 DISTLER AND ALLISON

(M.Mz-HY

+MB

 

WW

7000 use nos 82% ”00 0000
M

 

 

Figure 5. A portion of the negative-ion MALDI mass spectrum of
duplex 1 with methylene blue (MB) as an additive. ATT:sp was
used as the matrix. The DSRR is 0.05 and the WCSF is 1.0. The
sperrninezMBzduplex ratio is 500:4021.

Figure 3 to 0.07 in Figure 5. The WCSF is 1.0 and the
SSD is 0.9.

There have been many cases in UV-MALDI MS
where a complex is observed, possibly as an analyte
with a matrix molecule or analyte with an additive. It
may be significant that, when the additive MB appears
to allow intact duplex detection, there is direct evidence
in the spectrum for existence of the M,Mz-MB complex.
This suggests that MB forms a strong complex with the
duplex that is more stable than the duplex alone, and
that it is involved in duplex preservation in the target
preparation step.

If some small molecules that are known to interact
with DNA can stabilize the duplex in the MALDI
experiment, larger molecules such as proteins and pep-
tides may also have a stabilizing effect. In an experi-
ment designed in our laboratory to explore the differ-
ences in resolution and sensitivity between a peptide
and an oligonucleotide in UV-MALDI-MS, a solution
was prepared that contained equal amounts of a pep-
tide and oligonucleotide of similar molecular weights.
In the resulting spectra, peaks were detected for the
peptide, (P + H)", the oligonucleotide, (O + H)+, and a
peptideoligonucleotide complex, (P + O + H)+. Simi-
lar results have been reported previously [29]. In order
to examine the influence of peptides on double-
stranded oligonucleotides, several peptides were com-
bined with duplex oligonucleotides and analyzed by
UV MALDI-MS. The peptides were used at a level of 1
pmol or less. An additive that is most effective at a level
similar to that of the analyte, rather than that of the
matrix, will be referred to as a microadditive (with the
hope that there will not be a future need to separately
define nanoadditives, picoadditives, etc).

Duplex 1 was used for the experiments designed to
evaluate peptides as microadditives with ATT:sper-
mine as the matrix, Figure 6a. The SSD is 0.7 and the
DSRR is 0.006. While the intact duplex, (M,Mz — H)“ is
detected in this experiment, the peak representing the
(MzMz - H)“ species is surprisingly intense, giving the
spectrum a WCSF of 0.3.

When only 1 pmol of B-MSH was added to the target

] Am Soc Mass Spectrom 2002,13, 1129-1137

 

 

 

 

 

a (Ma-Hr
(Mmr P‘P
E l (MM-Hr
. _ L _
b . ‘—
. («A.M.-H)
+89

3000 4000 5000 6000 7000 0000 9000

mlz

Figure 6. Negative-ion MALDI mass spectra of duplex 1 with (a)
no peptide present and (b) with B-MSH present. ATT:sp was used
as the matrix for both experiments. 25 pmol of duplex and 1 pmol
of B-MSH were deposited onto the MALDI target. For (a), the SSD
is 0.7, the DSRR is 0.006, and the WCSF is 0.3. For (b). the SSD is
1.1, the DSRR is 0.06, and the WCSF is 1.0. The peptidezduplex
ratio is 1:25.

(which contained 25 pmol of duplex) used to obtain
Figure 6a, the spectrum shown in Figure 6b is obtained.
The SSD changes dramatically to 1.1, the DSRR is 0.06,
and the WCSF is 1.0. There are no peaks representing
either (M,M, — H)“ or (M,Mz - H)“. In this experi-
ment, B—MSH stabilized, to some extent, double-
stranded DNA ions in UV MALDI MS, when present at a
level of one picomole on the MALDI target. Other peptides
evaluated are listed in Table 2. All peptides were used
as additives by adding 1 pmol of peptide to 25 pmol of
duplex 1 and analyzed using ATT:sp as the matrix. The
DSRR and SSD values are listed in the table. All WCSF
values are 1.

The peptides evaluated as microadditives can be
grouped based both on their size and amino acid
content. It is possible that the most successful experi-
ments would have involved peptides that have a size
similar to that of the oligonucleotide. From spermine
experiments performed previously [2, 3], we have
learned that additives capable of multiple interactions
with the oligonucleotides are important. For this reason,
the number of basic residues in the peptide may be
important, with multiple basic residues increasing the
number of electrostatic interactions possible. Previously
it was determined that peptides containing basic resi—
dues increase the melting temperature of duplex oligo-
nucleotides in solution [30]. Protonated backbone nitro-
gens may aid in stabilization as well.

When examining the data in Table 2, attempts were
made to correlate the experimental results with the
structural features of the peptides. The two peptides
used with the most notable success were dynorphin A
and B-MSH. Both of these larger peptides contain basic
residues with three arginines and two lysines out of the
17 amino acids in dynorphin A and six basic residues
out of the 22 amino acids residues in B-MSH including
three lysines, two arginines, and one histidine. Dynor-

207

 

 

I Am Soc Mass Spectrom 2M,13,1129-1137

STABILIZATION OF DOUBLE-STRANDED DNA 1135

Table 2. Peptide additives and their influence on UV MALDI spectra of duplex 1

 

 

Peptide Peptide sequence SSD WCSF DSRR 96 Arg 96 Lys MW

Dynorphin A YGGFLRRIRPKLKWDNO 1 1 1.0 0.1 17.66 11.76 2147.50
B—MSH AEKKDEGPYRMEHFRWGSPPKD 1.1 1.0 0.1 9.09 13.64 2660.90
Katacalcin DMSSDLERDHRPHVSMPONAN 1.1 1.0 0.0 9.52 0.00 2436.60
Kemptide LRRASVA 0.9 1 0 0.0 28.67 0.00 771.90
FBI peptide HHLGGAKOAGDV 1 O 1 0 0.0 0.00 8.33 1169.30
Trilysine KKK 1 0 1 O 0.0 0.00 100.00 402.60
N-e-aeetyl-Lys K 1 0 1 0 0.0 0.00 100.00 168.23
Leucine enkephalin YGGFL 1 O 1 0 0.0 0.00 0.00 665.60
Boc-MNF amide BMNF 0 8 1 0 0 0 0.00 0 00 610.60
Hexaalanine AAAAAA 1 0 1 0 0 0 0.00 0 00 444.50
Trialanine methyl ester AAA 0 9 1 0 0 0 0.00 0 00 305.30
Triserine SSS 1 1 1 0 0 0 0.00 0 00 279.20

 

phin A and B—MSH presumably interact with the dou-
ble-stranded oligonucleotides at multiple sites. The
backbone nitrogens may be important in the stabiliza-
tion of the duplex. A positively-charged backbone in a
peptide may behave similarly to spermine in solution.
The side chains of the peptide may also play an impor-
tant role. The amino acids, tyrosine, phenylalanine, and
tryptophan, contain aromatic side chains. These side
chains may partially intercalate between the nitrogen-
containing bases of the oligonucleotide, stabilizing the
duplex. Increases in the melting temperature of duplex
DNA have been reported when in the presence of
peptides containing phenylalanine residues [31]. Basic
side chains may also play an important role as dis-
cussed previously.

Since peptides containing only alanine residues were
not as successful as additives, it appears that the back—
bone nitrogens are not effective in stabilizing the du-
plex. Small peptides containing basic residues such as
LysLys-Lys were found to increase the DSRR value
although these smaller peptides were less effective than
Dynorphin A and B—MSH. Similarly, peptides contain-
ing aromatic side chains were also capable of increasing
the DSRR values for the experiments.

Since the best results are seen with the peptides
Dynorphin A and B-MSH, which contain both aromatic
and basic amino acids, it is not perhaps the size of the
peptide that is of primary importance, but the spacing
between basic amino acids. When basic amino acids in
a larger peptide are separated by other residues, the
separation between those basic residues may be similar
to the spatial separation between the phosphates of the
oligonucleotide. This may lead to greater stabilization
of the duplex. Clearly, more work is required to find the
optimal peptide, but these experiments establish struc-
tural aspects of the additives that are important and
show that peptides can, at relatively low levels, stabilize
to some extent duplexes in the UV MALDI experiment.

In order to maximize the interaction between the
peptides and the double-stranded oligonucleotides,
several conditions were varied in the experiments. At
first the peptide and oligonucleotide were combined on
the MALDI plate before the addition of the matrix. To

insure thorough mixing, peptide and oligonucleotide
solutions were also premixed before deposition on the
target. There was no difference between spectra ac-
quired using on target mixing or premixed solutions. A
solution of the peptide and oligonucleotide was also
pre-mixed and incubated at both room temperature and
at 37 °C prior to MALDI analysis. When incubated at
either temperature, samples again yielded spectra sim-
ilar to the unincubated samples.

For the data shown in Table 2, one picomole of each
peptide was used in the formation of the MALDI target.
While 1 pmol was optimal for most peptides, improve-
ments were seen with as little as 50 frnol of peptides
present in the target. FBI peptide, katacalcin, and
B-MSH all yielded duplex spectra with improved DSRR
values with less than 1 pmol of peptide present in the
target. While addition of these peptides to the MALDI
target improved the DSRR, the B-MSH peptide gave the
best results. Figure 7 shows the negative-ion MALDI
mass spectrum of a target containing 100 frnol of
B-MSH and 25 pmol of double-stranded oligonucleo-
tide, using AT'Tzsp as the matrix. The SSD was 0.94, the
DSRR was 0.09 and the WCSF was 1.00. The presence of

(Mi-HT

(M.-H)'

(”Ma-HT

 

 

 

 

3000 4W 5M 0000 7000 soon 9000
mlz

Figure 7. Negative-ion MALDI mass spectrum of duplex 1 with

100 frnol of B-MSH and ATT:sp as the matrix. The SSD is 0.94, the

DSRR is 0.09, and the WCSF is LIX). The peptidezduplex ratio is
1:250.

 

208

1136 DISTLER AND ALLISON

the peptide still influences the resulting mass spectrum
even when the relative concentration of the oligonucle-
otidezpeptide is 250:1.

All of the experiments reported here were performed
on a single ”batch" of a given duplex. We have pur-
chased synthetic, annealed duplexes from both an on-
campus facility and from commercial suppliers. In the
case of the Michigan State University Facility, oligonu-
cleotides are usually synthesized as primers and we are
one of a few groups on campus that request synthesized
strands and their annealed duplexes. We have found
that some sources occasionally provide a duplex that is
more stable than the same duplex in another batch. Salts
and buffers that are commonly provided in such sam~
ples, while not specified, could alter resulting MALDI
spectra. All of the spectra presented here have been
duplicated on multiple dates from the same sample,
and from other batches of the same sample, to confirm
that no result was influenced by an impurity or con-
taminant in a single batch. From multiple experiments
on identical samples, we believe that the quantities
reported here are reflections of the relative influence of
the additives discussed. Spectra were obtained from
several locations across the target. Those shown repre-
sent an average for the target. In identical experiments,
the WCSF and SSD quantifiers are reproducible to 10%.
For example, in Figure 7, the WCSF value was 1.0 1' 0.1
and the SSD has a value of 0.94 t 0.90. This was not the
case for the DSRR values. For the experiments shown
here, the DSRR values reported ranged from 0 to 0.1.
The experiments that yielded DSRR values of zero did
so consistently. When duplex peaks were detected in an
experiment, the DSRR values were reproducible to 25%.
For the additives shown, typically 5—10% of the duplex
molecules were stabilized in the experiment, and can be
viewed as having similar stabilizing capabilities.

Conclusion

Methylene blue, spermine, and B-MSH as additives
allow for the detection of small intact DNA duplexes.
Each of these compounds must be capable of stabilizing
double-stranded DNA in at least the crystal growth
process, and possibly in the desorption/ ionization pro-
cess as well. While these additives all increase the DSRR
values, they may participate in a different way through-
out the MALDI process. For example, spermine aids in
the crystallization of the duplex, and complexes with
the oligonucleotide during crystal growth. During the
desorption/ ionization of the oligonucleotide, the
spermine may dissociate from the duplex, resulting in
the detection of the intact duplex, (M,Mz - H)“. The
spermine may also stabilize the duplex through the
desorption/ ionization process with the (M,Mz + sp -
H)“ species being detected. In order for the spermine to
bind to the DNA and stabilize it, the spermine must be
present at a concentration equal to, or greater than, the
oligonucleotide concentration. In contrast, peptides are
capable of stabilizing the duplex oligonucleotides even

I Am Soc Mass Spectrom 2002, 13,1129-1137

when present at concentrations less than that of the
oligonucleotide. The peptides may stabilize the duplex
through another mechanism. If the peptide can stabilize
the duplex whether it is present at a level of 1 pmol or
50 frnol (always present at a level less than the 25 pmol
of duplex used for each experiment presented here),
each peptide may be capable of interacting with more
than one duplex throughout the experiment. Since the
peptide is needed in only a catalytic amount, it may be
acting as a chaperone for the doubleostranded oligonu-
cleotides. The peptide may bind and stabilize the du-
plex during the crystal growth step. After depositing
the duplex on the growing crystal surface, the peptide
may dissociate from the duplex, allowing it to interact
with another duplex still in solution.

Alternatively, the equilibrium constant for a duplex-
spermine complex may be much smaller than that for a
duplex-peptide complex. Thus, to achieve the same
effect, a greater concentration of spermine is needed,
although the mechanism is the same. Even speculating
on this possibility is difficult because it is not known
how many additive molecules interact with a single
duplex. In order to fully understand this process, the
stoichiometry of the critical step must be known.

Since peptide-duplex complexes are not detected in
these MALDI experiments, a chaperone mechanism is
more likely. While it may be different from the mecha-
nism for spermine, the peptide-oligonucleotide interac-
tion, in some ways, parallels the behavior of spermine.
When crystals of oligonucleotides are required for X-ray
crystallographic studies, spermine is frequently added
to facilitate crystal growth. However, spermine is usu-
ally not found in the crystals that are formed. Multiply
protonated, positively-charged spermines presumably
interact with the oligonucleotides, but the spermine
molecules are efficiently eliminated as the crystals
grow. In a similar way, peptides such as B-MSH may
bind with duplexes only until they become incorpo-
rated into the matrix crystal, and become released at
that time.

Why do DNA duplexes dissociate in the UV MALDI
experiment? Results presented here suggest that exten-
sive, if not complete, dissociation occurs in the first step
during growth of the crystal target. We have investi
gated the possibility that the duplexes dissociate in the
initial solution due to the organic solvents and excess
matrix molecules, but these do not appear to be suffi-
cient for inducing dissociation. One possibility is the
crystals themselves. When a duplex lands on a growing
crystal in the initial solution, multiple possibilities fol-
low. It may desorb from the surface in duplex form, it
may incorporate into the crystal in duplex form, or it
may dissociate, leaving one strand bound to the surface.
If the matrix crystals are responsible for dissociation of
non-covalent complexes that are analyte in a MALDI
experiment, methods for more rapid crystal growth
could result in more efficient trapping of such com-
plexes for subsequent analysis.

209

 

I Am Soc Mass Spectrom 2002, 13, 1129-1137

References

1.

10.

ll.

12.

13.

14.

Karas, M.; Bachmann, D.; Bahr, U.; Hillenkamp, F. Matrix-
Assisted Ultraviolet-Laser Desorption of Non-Volatile Com-
pounds. lnt. I. Mass Spectrom. 1987, 78, 53—68.

. Asara, I. M.; Allison, I. Enhanced Detection of Oligonucleo-

tides in UV MALDI MS Using Tetraamine Spermine as a
Matrix Additive. Anal. Chem. 1999, 71, 2866-2870.

. Distler, A. M.; Allison, I. S-Methoxysalicylic Acid and Sperm-

ine: A New Matrix for the Matrix-Assisted Laser Desorption/
Ionization Mass Spectrometry Analysis of Oligonucleotides.
I. Am. Soc. Mass Spectrom. 2001, 12, 456-462.

. Distler, A. M.; Allison, I. Improved MALDI-MS Analysis of

Oligonucleotides through the Use of Fucose as a Matrix
Additive. Anal. Chem. 2001, 73, 5000—5003.

. Shahgholi, M.; Garcia, B. A.; Chiu, N. H. L.; Heaney, P. I.;

Tang, K. Sugar Additives for MALDI Matrices Improve Signal
Allowing the Smallest Nucleotide Change (A21) in a DNA
Sequence to be Resolved. Nucleic Acids Res. 2001, 29, e91/1-
e91 / 10.

. Koomen, I. M.; Russell, W. K.; Hettick, I. M.; Russell, D. H.

Improvement of Resolution, Mass Accuracy, and Reproduc-
ibility in Reflected Mode DE-MALDI-TOF Analysis of DNA
Using Fast Evaporation-Overlayer Sample Preparations. Anal.
Chem. 2000, 72, 3860-3866.

. Zhang, L. K.; Gross, M. L. Matrix-Assisted Laser Desorption/

Ioniation Mass Spectrometry Methods for Oligodeoxynucle—
otides: Improvements in Matrix, Detection Limits, Quantifica-
tion, and Sequencing. I. Am. Soc. Mass Spectrom. 2000, 11,
854-865.

. Pieles, U.; Ziircher, W.; Schar, M.; Meser, H. E. Matrix-

Assisted Laser Desorption Ionization Time-of-Flight Mass
Spectrometry: A Powerful Tool for the Mass and Sequence
Analysis of Natural and Modified Oligonucleotides. Nucl.
Acids Res. 1993, 21, 3191-3196.

. Yi, Y. C. L.; Cheng, 5.; Chan, T. W. D. Evaluation of Ammo-

nium Salts as Co-matrices for Matrix-Assisted Laser Desorp-
tion/Ionization Mass Spectrometry of Oligonucleotides.
Rapid. Commun. Mass Spectrom. 1996, 12, 993-998.

Cheng, S. W.; Chan, T. W. D. Use of Ammonium Halides as
Co-matrices for Matrix-Assisted Laser Desorption/ Ionization
Studies of Oligonucleotides. Rapid Commun. Mass Spectrom.
1996, 10, 907-910.

Vandell, V. E.; Limbach, P. A. Polyamine Co-matrices for
Matrix-Assisted Laser Desorption/ Ionization Mass Spectro-
metry of Oligonucleotides. Rapid Commun. Mass Spectrom.
1999, 13, 2014-2021.

Lecchi, P.; Panell, L. K. The Detection of Intact Double-
Stranded DNA by MALDI. I. Am. Soc. Mass Spectrom. 1995, 6,
972-975.

Kirpekar, F.; Berkenkamp, 5.; Hillenkamp, F. Detection of
Double-Stranded DNA by IR- and UV-MALDI Mass Spectro-
metry. Anal. Chem. 1999, 71, 2334-2339.

Little, D. P.; Iacob, A.; Becker, T.; Braun, A.; Darnhofer-Demar,
B.; Iurinke, C.; van den Boom, D.; Koster, H. Direct Detection
of Synthetic and Biologically Generated Double-Stranded

15.

16.

17.

I8.

19.

20.

21.

27.

31.

210

STABILIZATION OF DOUBLE-STRANDED DNA 1137

DNA by MALDI-TOF MS. Int. I. Mass Spectrom. lon Processes
1997, 169/170, 323-330.

Tang, K.; Taranenko, N. 1.; Allman, S. L; Chen, C. H.; Chang,
L. Y.; Iacobson, K. B. Picolinic-Acid as a Matrix for Laser
Mass-Spectrometry of Nucleic-Acids and Proteins. Rapid Com-
mun. Mass Spectrom. 1994, 8, 673—677.

Shui, X.; McFail-Isom, L; Hu, G. G.; William, L. D. The B-DNA
Dodecamer at High Resolution Reveals a Spine of Water on
Sodium. Biochemistry 1998, 37, 8341-8355.

Halle, B.,' Denisov, V. P. Water and Monovalent Ions in the
Minor Groove of B-DNA Oligonucleotides as Seen by NMR.
Biopolymers 1998, 48, 210-233.

Tippin, D. 8.; Sundaralingam, M. Nine Polymorphic Crystal
Structures of d(CCGGGCCGG), d(CCGGGCCm(5)CGG),
d(Cm(5(CGGGCCm(5)CGG), and d(CCGGGCC(Br)(5)CGG)
in Three Different Conformations: Effects of Spermine Binding
and Methylation on the Bending and Condensation of A-
DNA. I. Mol. Biol. 1997, 267, 1171-1185.

Egli, M.; Williams, L. D.; Gao, 0.; Rich, A. Structure of the
Pure-Spermine Form of Z-DNA (Magnesium Free) at l-A
Resolution. Biochemistry 1991, 30, 11388-11402.

Tishchenko, E. 1.; Karapetyan, A. T.; Borisova, O. F. Different
Complexes of Ethidium Bromide and Their Role in Stabilizing
the (dA),,-(dT')n Structure. Mol. Biol. 1996, 30, 826-829.
Vardevanyan, P. 0.; Antonyan, A. P.; Manukyan, G. A.;
Karapetyan, A. T.; Shchyolkina, A. K.; Borisova, O. F.
Ethidium Bromide Binding to Native and Denatured Poly-
(dA)Poly(dT). Mol. Biol. 2000, 34, 272-276.

. Bradbury, E. M. Nucleosome and Chromatin Structures and

Functions. I. Cell. Biochem. 1998, 30/31, 177-184.

. Blackburn, G. M.; Gait, M. I. Nucleic Acids in Chemistry and

Biology; 2nd ed. Oxford University Press: New York, 1996.

. Porschke, D. Nature of Protamine-DNA Complexes: A Special

Type of Ligand Binding Cooperativity. I. Mol. Biol. 1991, 222,
423-433.

. Ausio, I. Histone H1 and Evolution of Sperm Nuclear Basic

Proteins. I. Biol. Chem. 1999, 274, 31115-31118.

. Tang, X.; Callahan, I. H.; Zhou, P.; Vertes, A. Noncovalent

Portein-Oligonucleotide Interactions Monitored by Matrix-
Assisted Laser Desorption/ Ionization Mass Spectrometry.
Anal. Chem. 1995, 67, 4542-4548.
Karas, M.; Hillenkamp, F. laser Desorption Ionization of
Proteins with Molecular Masses Exceeding 10,000 Daltons.
Anal. Chem. 1988, 60, 2299-2301.

. Schildkraut, C.; Lifson, 5. Dependence of the Melting Temper-

ature of DNA on Salt Concentration. Biopolymers 1965, 3,
195—208.

. Lin, 5.; Cotter, R. 1.; Woods, A. 5. Detection of Non-Covalent

Interaction of Single and Double Stranded DNA with Peptides
by MALDI-TOF. Proteins 1996, 52, 12-21.

. Harrison, I. G.; Balasubramanian, 5. Synthesis and Hybridiza-

tion Analysis of a Small library of Peptide-Oligonucleotide
Complexes. Nucleic Acids Res. 1996, 26, 3136-3145.
Robledo-Luiggi, C.; Vera, M.; Cobo, L.; Iaime, E.; Martinez, C.:
Gonzalez, I. Partial Intercalation with Nucleic Acids of Pep-
tides Containing Aromatic and Basic Amino Acids. Biospec-
troscopy 1999, 5, 313—322.

 

   

IIIIIIIIIIIIIIIIIIIIIIIIIIIIIII

'Hllllllllllllllll".|ll|~|l‘|l|‘|ill
93 02470 8715

312