v“

.. “.42
awn. 4
.

 

 

 

LIBRARY
Michigan State
University

 

 

 

This is to certify that the
thesis entitled

The Study of Bovine Cytochrome c Oxidase Using Matrix-
Assisted Laser Desorption/Ionization Mass Spectrometry

presented by
Qian Li

has been accepted towards fulfillment
of the requirements for the

MS. degree in Department of Chemistry

0%— W

Major Professor’ 5 Signature

5. 6.07m

Date

MSU is an Affinnative Action/Equal Opportunity Institution

--—-— -.-.-v-.-.-.- -— ~ - -

_--.- -.------o

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

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

6/01 c:/ClRC/DateDue.p65-p. 15

THE STUDY OF BOVINE CYTOCHROME C OXIDASE USING MATRIX-
ASSISTED LASER DESORPTION/IONIZATION MASS SPECTROMETRY

By

Qian Li

A THESIS
Submitted to
Michigan State University
In partial fulfillment of the requirements
for the degree of
MASTER OF SCIENCE
Department of Chemistry

2003

ABSTRACT

THE STUDY OF BOVINE CYTOCHROME C OXIDASE USING MATRIX-
ASSISTED LASER DESORPTION/IONIZATION MASS SPECTROMETRY

By

Qian Li

Noncovalent interactions bind biological complexes such as drug-peptide complexes,
duplex DNA, and multi-subunit proteins. A recent analytical tool employed by
biochemists to study noncovalent complexes is mass spectrometry. To explore better
conditions to stabilize noncovalently bound complexes in the matrix-assisted laser
desorption/ionization (MALDI) experiment, the protein cytochrome c oxidase (CcO) was
chosen for this study. CcO is a membrane protein. It contains 13 subunits, with a
monomer molecular weight over 200,000. CcO provided two challenges in its MALDI
MS analysis. The first challenge is related to the fact that it is a membrane protein; these
tend to aggregate in water. Subunits I, II, and III are very hydrophobic and not detected
in the conventional MALDI experiment. The detergent lauryl maltoside was investigated
as an additive to solubilize these hydrophobic subunits. The effects of the concentration
of lauryl maltoside and protein on the resulting spectra were investigated. The second
challenge is related to the fact that all of intact enzyme completely dissociates in the
conventional MALDI MS experiment. Can the intact enzyme be preserved and detected?
Our experimental results suggest that with the appropriate matrix, solvents, pH and

detergent, the CcO complex can be detected in MALDI MS.

ACKNOWLEDGMENTS

Thank you Dr. John Allison for your patient guidance and strong support. You led me
into the magic world of mass spectrometry. Those complicated and abstract theories
became so simple and tangible after your explanation. I benefit a lot from all the
discussions with you. Thank you for putting up with my poor English and your precious

time reading and correcting this manuscript.
Thank you labmates Anne, Donna, Jamie, Leah and Eric. You are my best American
culture dictionary. You made the lab a really fun place to work in. Thank you for all the

discussions and laughter.

Thank you Dr. Shelagh M. Ferguson-Miller for your help in this project. You are always

enthusiastic to answer my questions.

Thank you Dr. Denise Mills. You offered great help in solving lots of the key problems

in my research.

Thank you Dr. Denis Proshlyakov for providing us plenty of protein materials to work

with. Your suggestions helped us a lot.

Thank you my parents for your support and encouragement. You always show your

pride in my endeavor. You are my great source of inspiration.

iii

TABLE OF CONTENTS

 

List of Tables .................................................................................... iv
List of Figures .................................................................................... iv
List of Abbreviations ............................................................................ iv
Chapter 1: The Study of Non-covalent Complexes Using MALDI-MS ..................... 1

Introduction ................................................................................. 1
Why do non-covalent complexes dissociate in MALDI-MS? ....................... 2
The effect of pH ........................................................................... 4
The effect of organic solvents ........................................................... 9
Other factors that disrupt non-covalent interactions ................................. 10
“First shot” phenomenon ................................................................ 12
Could the complex come from non-specific aggregation? .......................... 13
References ................................................................................. 15

Chapter 2: Introduction to Matrix-Assisted Laser Desorption/Ionization Time-of Flight

 

(MALDI-TOE @ss Spectrometry ............................................................ 17
Introduction ............................................................................... l7
Instrumentation ........................................................................... l 8
MALDI experiment ...................................................................... 21
Matrix ...................................................................................... 22
Solvent ..................................................................................... 26
PH .......................................................................................... 28
References ................................................................................. 31

C_hapter 3: Review of Study of Cytochrome c Oxidase Using MassSQectrometry 33

Introduction ........................................................................................ 33
The study of cytochrome c oxidase using mass spectrometry .............................. 35
References .......................................................................................... 46

_Ch_aDter 4: The Detection of Cytochrome c Oxidase Subunits Using Mass

Spectromem ....................................................................................... 47
Experimental section ............................................................................. 47
Results and discussion ........................................................................... 48
Conclusion ......................................................................................... 75
References ........................................................................................ 76

Chapter 5: The Study of Cytochrome c Oxidase Complexes Using MALDI-TOF-MS 78

Experimental section .............................................................................. 78
Results and discussion ............................................................................ 78
Conclusion ......................................................................................... 92
References ......................................................................................... 93

iv

Chapter 6: Conclusions and Future Work .................................................... 94

Appendix .......................................................................................... 96
Cytochrome c Oxidase Purification ........................................................... 97

Table 2.1

Table 3.1

Table 3.2

Table 4.1

Table 4.2

Table 4.3

Table 4.4

Table 5.1

LIST OF TABLES

Ionization Methods for Analysis of Proteins and Peptides by MS 18
The Relative Effects of Detergents on Signal Quality in MALDI MS 36
Molecular Masses of the Subunits of Bovine Heart and Liver Cytochrome c
Oxidase 40
Isotope Distribution Peak List for all the Peaks Observed in the Deconvoluted
ESI—FTMS Spectrum 54
Molecular Weights of Protein Subunits of Bovine Heart Cytochrome c
Oxidase 57
Sequences and Molecular Weights of Protein Subunits of Cytochrome c
Oxidase from Bovine Heart 60
Detergents Evaluated for CcO Analysis Using MALDI MS 65

The Main Species in Cytochrome C Oxidase Enzyme from Bovine Heart and
Their Approximate Molecular Weights 82

vi

Figure 1.1

Figure 1.2

Figure 1.3

Figure 1.4

Figure 1.5

Figure 1.6

Figure 1.7

Figure 2.1
Figure 2.2

Figure 2.3

Figure 2.4

Figure 2.5

Figure 3.1

Figure 3.2

Figure 3.3

LIST OF FIGURES

Non-covalent compound dissociates in MALDI 3

E81 mass spectrum of the ras-GDP complex at pH 4.7 in 2 mM ammonium
acetate buffer. Inset shows the deconvoluted mass spectrum of the ras—GDP
complex with molecular mass 19,294 Da. 5

E81 mass spectrum of the ras—GDP complex obtained at pH 3 in 2 mM
ammonium acetate buffer. Inset shows the deconvoluted mass spectrum of
the ras—GDP complex with molecular mass 19,9244 Da and apo-ras with
molecular mass 18,851.4 Da. 6

E81 mass spectrum of ras—GDP obtained at pH 2.5 in 2 mM ammonium
acetate buffer using 20 mM protein solution. Inset shows the deconvoluted
mass spectrum of the apo-ras with molecular mass 18,854 Da. 6

Non-covalent complex dissociates when incorporating into matrix crystals 10

Non-covalent complex dissociation could occur in the desorption/ionization
process. 11

a) Sum of first shots on a given spot. b) Sum of second and following shots

on a given spot 13
Voyager-DE Mass Spectrometer l9
Reflectron TOF mass spectrometer 21

Photon absorption by the matrix in MALDI, causing desorption and
ionization. 23

Structure of some most commonly used matrices 24

Distribution of sinapinic acid matrix crystals on the MALDI-MS target wells

as observed by scanning electron microscopy 28
Biological function of cytochrome c oxidase 33
l3-Subunit cytochrome c oxidase structure 34

MALDI mass spectrum of the Ni2+- NTA resin purified His-tagged aa3-type

cytochrome c oxidase complex from R. sphaeroides. a) MALDI spectrum of
subunits with less than 40,000 Da molecular weight. b) MALDI spectrum of
subunits larger than 40,000 Da molecular weight. 38

vii

Figure 3.4 Positive-ion linear MALDI-TOP spectrum of cytochrome c oxidase from R.
Sphaeroides. SA/sucrose was used as the matrix. 44

Figure 4.1 a.) Absorption spectrum of oxidized form of cytochrome c oxidase;
b.) Absorption spectrum of reduced form of cytochrome c oxidase (reduced

With N328204) 49
Figure 4.2 Positive-ion linear MALDI-TOF spectrum of cytochrome c oxidase. The ten
nuclear coded subunits were detected. SA was used as matrix. 51
Figure 4.3 HPLC chromatogram. 52

Figure 4.4 Deconvoluted ESI-FTMS spectrum of cytochrome c oxidase eluted from
HPLC at 20 minute. 53

Figure 4.5 a) Deconvoluted ESI-FTICR spectrum. B) Theoretical isotope distribution
for subunit XIII 55

Figure 4.6 Mechanism for the CcO analysis in the conventional MALDI experiment 62
Figure 4.7 The formation of detergent micelles 63
Figure 4.8 The structure of the detergents evaluated 66

Figure 4.9 Positive-ion linear MALDI-TOE spectrum of cytochrome c oxidase.
Subunits I, II and III were detected. ATT was used as matrix. Lauryl
maltoside was used as an additive. 69

Figure 4.10 The detergent micelle-subunit complexes could fall apart in MALDI MS. 1.
the complexes fall apart when they are incorporated into the matrix crystals.
2. The complexes dissociate in the desorption/ionization process. 71

Figure 4.11 Positive ion linear MALDI-TOP spectra of subunits I, II and III of
cytochrome c oxidase using different concentration of lauryl maltoside as
solvent. SA was used as the matrix. CcO concentration: luM 72

Figure 4.12 Positive ion linear MALDI-TOF spectra of subunits I, II and III of
cytochrome c oxidase in different concentration. SA was used as the
matrix. The concentration of lauryl maltoside: 8*CMC 74

Figure 5.1 Positive-ion linear MALDI-TOF spectrum of BSA. SA was used as
matrix. Concentration of BSA: 50 11M. 79

Figure 5.2 Positive-ion linear MALDI-TOF spectrum of C00 complexes. ATT (in 60
mM NILOAc) was used as matrix. Concentration of CcO: 3 uM. 83

viii

Figure 5.3

Figure 5.4

Figure 5.5

Figure 5.6

Figure 5.7

Figure 5.8

Positive-ion linear MALDI-TOF spectrum of CcO complexes. ATT (in 60
mM NH40Ac) was used as matrix. Concentration of CcO: 3 uM. 84

Positive-ion linear MALDI-TOF spectrum of CcO complexes. ATT (in 60

mM NH40Ac) was used as matrix. Concentration of C00: 3 uM. 85

Positive-ion linear MALDI-TOF spectrum of CcO complexes. ATT (in 60
mM NH40Ac) was used as matrix. Concentration of CcO: 3 uM. 86

Positive-ion linear MALDI-TOP spectrum of C00 complexes. ATT (in 60
mM NI-LOAc) was used as matrix. Concentration of CcO: 3 uM 87

Positive-ion linear MALDI-TOP spectrum of CcO complexes. ATT (in 60

mM NH40Ac) was used as matrix. Concentration of C00: 3 uM 88

Low resolution at high m/z values 90

ix

ACN

ATT

BO

BSA

CcO

CHCA

CL

CMC

DE

DHB

ESI

FTMS

GDP

GPC

GTP

HABA

HNE

HPLC

IR

LDI

LM

LIST OF ABBREVIATIONS

Acetone nitrile

6-aza-2-thiothymine

Bacterioopsin

Bovine serum albumin

Cytochrome c oxidase
a-cyano-4-hydroxycinnamic acid
Cardiolipin

Critical micelle concentration
Delayed extraction
2,5-dihydroxybenzoic acid
Electrospray ionization

Fourier transform mass spectrometry
Guanosine diphosphate

Gel permeation chromatography
Guanosine triphophate
2-(4-hydroxyphenylazo (benzoic acid)
4-hydroxy-2-nonenal

High performance liquid chromatography
Infrared

Kinetic energy

Laser desoption/ionization

Lauryl maltoside

MALDI
[M+H]

m/z
N-terminus
PC

PE

PG

PSD

SA
SDS-PAGE
TFA

TOF

UV

Matrix-assisted laser desorption/ionization
Protonated molecule

Mass-to-charge ratio value
Amino-terminus of peptide

Phosphatidyl choline

Phosphatidyl ethanolamines

Phosphatidyl glycerols

Post-source decay

Sinapinic acid

Sulfate polyacrylamide gel electrophoresis
Trifluoroacetic acid

Time-of—flight

Ultraviolet

xi

Chapter One: The Study of Non-covalent Complexes Using MALDI-MS

Introduction

Non-covalent complexes play a very important role in biological system. Proteins serve
specific functions and may be transported to their destination by associating with a
specific partner. They can interact with other proteins, peptides, oligonucleotides, metal
ions and other small molecules. They frequently bind by non-covalent forces. Non-
covalent interactions are found everywhere in life science. For example, drugs function
by binding to a specific target, most often, proteins or oligonucleotides. A better
understanding of their structure will assist researchers in the discovery of small molecules

that act as inhibitors to diseases.

Non-covalent forces include: hydrogen bonds, Van der Waals forces, ion-ion interactions
and hydrophobic interactions. They are usually much weaker than covalent bonds. A
protein structure can be broken down into four levels. The primary structure refers to the
amino acid sequence. The secondary structure involves the interaction of amino acids
within the chain to form hydrogen-bonded structures such as a-helices and B-sheets.
Tertiary structure is the “global” folding of a single polypeptide chain. The major driving
force is the hydrophobic effect. Quaternary structure refers to the association of two or
more polypeptide chains into a multi-subunit structure. All four types of non-covalent
interactions are found in the interactions between subunits in a multi-subunit protein.
Except for the primary structure, non-covalent forces are involved in all the higher order

structures. They are critical to preserve protein stability and activity. Disruption of these

forces by a change of pH, solvent composition, temperature, ionic strength and addition

of reducing reagents may cause partial or complete denaturation of the protein.

Many conventional analytical techniques have been employed to study non-covalent
compounds. Centrifugation, sodium dodecyl sulfate polyacrylamide gel electrophoresis
(SDS-PAGE) and gel permeation chromatography (GPC) have been used, but don’t
provide enough information regarding molecular weight and stoichiometry of the protein.
X-ray crystallography and NMR spectrometry have been used to solve 3-D structures of
various proteins, but they lack the direct information for the non-covalent complexes
based on molecular weight information [1]. With the development of soft ionization
techniques yielding little or no fragmentation such as electrospray ionization (E81) and
matrix-assisted laser desorption/ionization (MALDI), mass spectrometry has become a
very important method to study non-covalent complexes. They have the obvious
advantages of providing accurate molecular and stoichiometric information with great

speed and sensitivity.

Why do non-covalent complexes dissociate in MALDI-MS?

MALDI has been demonstrated to be an important method for the structural
characterization of proteins. Coupled with the time-of-flight mass analyzer, MALDI has
been used to determine molecular weights up to 500,000 Da. It is extremely sensitive.
The amount of sample required for MALDI-MS is in the sub-picomole- and quite often
even in the low femtomole- range. The sensitivity achieved in MALDI is 10 times that of

ESI [1]. Although MALDI is usually better than ESI for detecting high molecular weight

species [2], most of the non-covalent complexes analyzed using mass spectrometry were
reported by ESI. They include multi-subunit proteins, protein-nucleic acid, peptide-drug,
duplex DNA-drug, polypeptide-metal ion, and protein/peptide-peptide complexes [3, 16].
As shown in Figure 1.], most non-covalent complexes dissociate when they are analyzed

by MALDI-MS.

+130

i

q

 

In solution

Wkfi

?
I
I
I
I

 

In MALDI spectrum

Figure 1.1: Non-covalent compound dissociates in MALDI

The application of MALDI-MS to non-covalently bound compounds may be limited by
its harsh sample preparation conditions. In a conventional MALDI experiment, the
analyte is usually mixed with matrix in a solvent consisting of water and an organic co-
solvent. An acid such as trifluoroacetic acid and formic acid is often used to improve the
solubility. After that, the matrix and analyte are cocrystallized on a metal surface. To
preserve the native conformation of proteins, physiological conditions should be
approached. The low pH and organic solvents used in the MALDI experiment could

destroy the protein native structure. The protein inclusion into the matrix crystals could

also break the non-covalent bonds. What’s more, the contact with the metal surface is a

possibility suggested by some biochemists.

The effect of pH

pH is very important to control the state of multi-subunit proteins. Most proteins at the
physiological pH are above their isoelectric pH and have a net negative charge. When
the pH is adjusted to their isoelectric point, their net charge will be zero. Charge
repulsion of similar molecules will be minimized and many proteins precipitate. When
the pH is far below (above) their isoelectric point, the proteins will have a net positive
(negative) charge. The like charges repel each other and prevent the protein from
aggregating. But in some area where the density of positive (negative) charge is very
high, the repulsion forces will be great enough to unfold the protein. That can lead to

denaturation.

The pH effect was investigated by several groups. One of the important non-covalent
complexes in cancer research is the association of the ras protein with either guanosine

diphosphate (GDP) or guanosine triphosphate (GTP). Studies of these noncovalent
interactions are the basis for the development of the strategy to discover inhibitors for
relevant disease processes. Ras-GDP was used as a non-covalent model studied using
ESI by Ganguly et a1. [1]. When the pH was adjusted to 4.7, an ESI spectrum was
obtained. The deconvolution of this spectrum yielded an average molecular mass of
19,295, which is in good agreement with the molecular weight of the Ras-GDP complex.

The spectrum is shown in Figure 1.2. When the pH is decreased to 3, there are two sets

of ion distributions in the spectrum. One set corresponds to the intact complex with an
average weight of 19, 294. The other set deconvoluted to an average molecular weight of
18, 852, which represents the protonated free apo-ras protein. The spectrum is shown in
Figure 1.3. When the pH further decreased to 2.7, only the free apo-Ras protein was
observed. It indicates that, at low pH, the Ras-GDP complex unfolded and the free

protein was released. The spectrum is shown in Figure 1.4.

 

 

 

 

 

 

 

 

19,294
A?
a
Q)
E
0 1L
.2 - ..
E
SE

 

 

 

 

 

 

 

1.000 2.200
m/z

Figure 1.2: ESI mass spectrum of the ras-GDP complex at pH 4.7 in 2 mM ammonium
acetate buffer. Inset shows the deconvoluted mass spectrum of the ras—GDP complex
with molecular mass 19,294 Da.

1 8,852

 

 

 

 

 

 

 

 

i
.‘é‘ 1 9,294
g 1
H
.5
Q)
. Z
L5 .
Q)
m i
' A ....._U L.‘ ”I--- - ..- ILL‘
800 2 00
m/z '3

Figure 1.3: ESI mass spectrum of the ras—GDP complex obtained at pH 3 in 2 mM
ammonium acetate buffer. Inset shows the deconvoluted mass spectrum of the ras-GDP
complex with molecular mass 19,924.4 Da and apo-ras with molecular mass 18,851.4 Da.

 

 

18,854

_ it--_..l

 

 

 

 

 

 

Relative Intensity

 

 

 

 

 

 

 

 

 

 

 

m/z

Figure 1.4: ESI mass spectrum of ras—GDP obtained at pH 2.5 in 2 mM ammonium
acetate buffer using 20 mM protein solution. Inset shows the deconvoluted mass

spectrum of the apo-ras with molecular mass 18, 854 Da.

The effect of pH was also shown in the study of a substrate-enzyme complex by Karas et
a]. [5]. Sinapinic acid was used as matrix. When the matrix was prepared in ethanol-
water (1:1) (pH<2), ions for the aminopeptidase-bovine growth horrnone-releasing factor
(GHRF) were absent from the spectrum. When 1 M ammonium citrate was added to
control pH, ions representing the enzyme, substrate and enzyme-substrate complex were

observed.

In MALDI-MS, most oligomers dissociate. Human famesyl protein transferase is a
heterodimer consisting of one a- and one B-subunit. MALDI analysis using sinapinic
acid dissolved in ACN-0.1%TFA (70:30, vzv) gave peaks corresponding to each of these
two subunit monomers. Using sinapinic acid dissolved in ACN-0.2 M Bis-Tris (30:70)

(pH=7), a peak representing the heterodimer was detected [6].

All of the above experimental results suggest that pH does play a role in breaking non-
covalent bonds. A general way, employed in MALDI MS experiments, to maintain a
proper pH for the stabilization of the protein structure is through the use of buffers. Even
though MALDI has the advantage of a relatively higher tolerance for impurities such as
buffers, detergents and other salts than ESI [8], analysis might be impaired by the
interference of the impurities. The analyte must be prepared in a buffer that is not
detrimental to MALDI MS analysis. It was found that some buffers alter crystal growth
and the analyte signal was not observed [15]. The effect of a variety of buffer solutions
on the MALDI-MS experiment has been investigated [7, 15]. The major impact of

buffers is the protein incorporation into the matrix crystals and the ionization process. As

the solvent evaporates, there will be a competition among all the components for the
matrix crystal surface. Those that are in direct contact with matrix crystals are more
easily ionized. The volatility and hydrophobicity are important properties that influence
the degree to which the protein is incorporated into the relatively hydrophobic matrix
crystals. Ammonium salts are the most commonly used buffers, such as ammonium
acetate, ammonium citrate and ammonium bicarbonate. They can be tolerated by
MALDI MS analysis at relatively high concentrations (500 mM for ammonium acetate)
without degrading the spectrum. They are even found to enhance the signal intensity in
some cases [8]. It is not yet clear if the effect is based on the suppression of alkali-adduct
formation, the degree of analyte inclusion into the matrix, or a more effective desorption/
ionization process. F arrner and Caprioli [6] studied the specific non-covalent interactions
of RNAse S and of leucine zipper dimers. Dissolving both matrix and complex in
ammonium hydrogen carbonate, ammonium acetate, and ammonium citrate gave the

most intense signals for the complexes.

To overcome the disrupting effect of low pH, the use of a neutral or basic matrix can be
another solution. Many basic matrices have been discovered for this purpose [18, 19].
Of particular interest has been the recent observation of several types of non-covalent
complexes using 6-aza-thiothymine (ATT) as the matrix. It seems that ATT is becoming
a “magic” matrix for non-covalent complex study. The first success was reported on
double-stranded oligonucleotides [17]. It has also found success in the analysis of
proteins and peptides. Leucine zipper polypeptide dimers were observed in MALDI MS

using ATT in buffer solutions whereas the use of CHCA in ACN-0.1%TFA showed no

dimer peaks [6]. Woods et al. [9] and Glocker et al. [10] also reported the successful use
of ATT and ammonium salt buffers to detect zinc finger peptides and enzyme-substrate

complexes.

The effect of organic solvents

Solvent composition is another important factor that could lead to the dissociation of non-
covalent complexes. The presence of nonpolar organic solvents can weaken the

hydrophobic bonds of the proteins and thus the protein structure will be changed.

The disturbing effect of organic solvents was investigated by Ganguly et al. using the
Ras-GDP complex [4]. Methanol was added to a Ras-GDP solution at pH 5.8 (this pH
can preserve the native conformation) to give a 20% methanol solution. The dissociation
of the Ras-GDP complex occurred. Both the intact Ras-GDP complex and the free apo-
Ras protein were observed. When the methanol was increased to 50%, the complex

could no longer be detected.

Karas et al. determined the number of subunits of a multi-subunit complex using
MALDI-MS [5]. Streptavidin was mixed with nicotinic acid matrix in 10% ethanol in
water. The base peak is the monomer with additional peaks recorded for dimer, trimer,
and tetramer. When 50% ethanol was added, only the monomer peak showed up in the

spectrum.

Other factors that can disrupt non-covalent interactions

All of the above experimental results suggest that the organic solvent and low pH are the
main reasons that limited MALDI to non-covalent complex study, rather than the

desorption/ionization process itself. But other factors cannot be completely excluded.

The interaction between matrix and analyte molecules could disrupt the non-covalent
bonds. Figure 1.5 shows the non-covalent complex dissociating when incorporating into
matrix crystals. The mechanism of how the matrix and analyte cocrystallize is still under
investigation. Trial of a variety of matrices would be a solution if some matrices do not

disturb the non-covalent bonds.

DDDDDDDDDD

 

up

 

DDDDDDDDDD
0000000000
0000000000
DDDDDDDDDD

DDDDDUDDDD

Figure 1.5: Non-covalent complex dissociates when incorporating into matrix
crystals

Some biochemists suggested that contact with a metal surface could dissociate non-
covalent complexes. The metal surfaces used in MALDI MS are usually gold and steel.
If this is true, “sandwich” preparation methods might be useful [11]. In this experiment,
a droplet of matrix solution is applied to the sample well first. The solution is dried
immediately and a thin layer of matrix crystals is formed. Subsequently, a droplet of a

mixture of analyte and matrix is deposited on the thin layer. This avoids the direct

10

contact of analyte with the metal surface. Moniatte et al. reported that, using a sandwich
preparation, the crystallization was more homogeneous and more reproducible peaks

were obtained [12].

The dissociation of non-covalent compounds could also occur during the desorption and
ionization process as shown in Figure 1.6. The bonds may be too weak to survive the
laser irradiation. Crosslinking reagents can be used to stabilize the bonding. Farmer et
al. demonstrated that glutaraldehyde is a useful crosslinking chemical to join protein
subunits covalently [6, l3]. Glutaraldehyde reacts primarily with the e-amino group of
lysine to form cross-linking chains of differing lengths. The protein glucose-6-phosphate
dehydrognase is known to exist as a homodimer in solution. The dimer signal was

increased substantially after the incubation with glutaraldehyde.

Laser irradiation a ‘3

UDDDDDDDDD

 
   

DU DDDDDDDU D D

Figure 1.6: Non-covalent complex dissociation could occur in the
desorption/ionization process

Chemical additives can be added to stabilize the non-covalent structure. Spermine and

methylene blue are found to bind to small DNA duplexes and stabilize them [20].

Glycerol is reported to stabilize proteins in ESI MS [21].

ll

“First shot” phenomenon

A “first shot” phenomenon was observed in many cases in the study of non-covalent
complexes using MALDI-MS [6, 12, 14]. That is, the complex peak was detected by the
accumulation of the first few laser shots on a specific location. Subsequent laser shots
onto this location resulted in the failure to detect non-covalent species. Figure 1.7 a)
shows the spectrum of “first shots” of porin, consisting of three identical subunits, each
with a molecular weight of about 37 kDa. Subsequent exposure of the same sample area
yielded only ions of subunits as shown in Figure 1.7 b). Two reasons were addressed to
explain the “first shot” phenomenon. A small amount of protein may precipitate onto the
surface of the matrix crystal and then is desorbed as an intact complex by the first few
laser shots. Those that were included into the matrix crystals dissociated into their
individual parts when they interact with the matrix molecules. Another explanation is:
the first shot of the laser on the sample area will result in photodissociation of the protein.
As a consequence, subsequent laser shots onto the same spot resulted in the detection of
the dissociated parts. The latter hypothesis was ruled out by an irradiation experiment.
The sample was irradiated by a UV lamp prior to the MALDI MS analysis simulating UV
exposure in MALDI desorption. MALDI MS analysis of these irradiated samples gave
results identical to the non-irradiated samples. So they concluded that the intact complex

was only maintained at the surface.

12

J a W (111,250)

 

  

 

10,000 105,000 200,000

Relative Intensity

b SU+

 

 

 

 

 

‘Wrtrr'rrr

20,000 100,000

Figure 1.7: a) Sum of first shots on a given spot.
b) Sum of second and following shots on a given spot

Could observed complexes in MALDI MS come from non-specific aggregation?

Non-specific aggregation of binding partners in solution or in the gas phase is a
possibility we should be aware of. That is, the complex detected might come from
random association rather than specific bonding. Dilution experiments should be
performed to eliminate the possibility of non-specific condensation. Aerolysin is a

virulence factor secreted by human pathogen Ad hydrophola. After activation and

13

concentration, this protein condenses to form a transmembrane channel responsible for
target cell death. The oligomer was found to be a heptamer. MALDI-TOF-MS was used
to confirm the stoichiometry of the complex [12]. In the spectrum, except for the
heptamer, dimer, trimer and tetramer with decreasing intensities were also observed.
They might indicate a non-specific condensation formed in the sample preparation or in
the gas phase. In dilution studies, the trimer and tetramer were detected with the same
intensity ratio when the sample was diluted or concentrated five times. A non-specific
aggregation would at lease produce a small variation in the signal intensity ratio or a
small amount of pentamer and hexamer that wasn’t observed. This suggests that the

formation of trimer and tetramer was due to specific non-covalent interactions.

In this thesis, MALDI-MS will be used to study a multi-subunit protein, bovine

cytochrome c oxidase, which consists of thirteen different subunits. It will be shown that

pH and organic solvents are the primary limitations to the detection of the complex.

14

References

1.

10.

ll.

12.

l3.

14.

15.

16.

Pramanik, N. B.; Bartner, L. P.; Mirza, A. U.; Liu, H. Y.; Ganguly, K. A. J. Mass
Spectrom., 33, 911-920 (1998)

Smith, L. D.; Zhang, Z. Mass Spectrom. Rev., 133, 411-429 (1994)
Loo, A. J. Mass Spectrom. Rev., 16, 1-23 (1997)

Ganguly, A. K.; Pramanik, N. B.; Chen, G.; Tsarbopoulos, A. Practical
Spectroscopy, 32, 361-387 (2002)

Karas, M.; Bahr, U.; Ingendoh, A.; Nordhoff, E.; Stahl, B.; Strupat, K.;
Hiilenkamp, F. Analytica Chimica Acta, 241, 175-185 (1990)

Farmer, B. T.; Caprioli, M. R. J. Mass Spectrom., 33, 697-704 (1998)

Amini, A.; Dorrnady, J. S.; Riggs, L.; Regnier, E. F. J. Chromatography A, 894,
345—355 (2000)

Gross, J .; Strupat, K. Trends in Analytical Chemistry, 17, 470-484 (1998)

Woods, S. A.; Buchsbaum, C. J.; Worrall, A. T.; Berg, M. J.; Cotter, J. R. Anal.
Chem, 67, 4462-2266 (1995)

Glocker, O. M.; Bauer, H. J. S.; Kast, J.; Volz, J.; Przybylski, M. J. Mass
Spectrom., 31, 1221-1227 (1996)

Kussmann, M.; Nordhoff, E.; Rahbek-Nielsen, H.; Haebel, S.; Rossel-Larsen, M.;
Jakobsen, L.; Gobom, J.; Mirgorodskaya, E.; Kroll-Kristensen, A.; Palm, L.;
Roepstorff, P. J. Mass Spectrom., 32, 593-601 (1997)

Moniatte, M.; Lesieur, C.; Vécsey-Semjén, B.; Buckley, T. J .; Pattus, F.; Van der
Goot, G. F.; Van Dorsselaer, A. Int. J. Mass Spectrom. Ion Proc., 169/170, 179-
199 (1997)

Farmer, T. B.; Caprioli, R. M.; Fenselau, C. C.; Smith, P. B. Biol. Mass
Spectrom., 20, 796-800 (1991)

Rosinke, B.; Strupat, K.; Hiilenkamp, F.; Rosenbusch, J.; Dencher, N.; Krtiger,
U.; Galla, J. H. J. Mass Spectrom., 30, 1462-1468 (1995)

Kallweit, U.; Bomsen, O. K.; Kresbach, M. G.; Widmer, M. H. Rapid Commun.
Mass Spectrom., 10, 845-849 (1996)

Beck, L. J.; Colgrave, L. M.; Ralph, F. 8.; Shell, M. M. Mass Spectrom. Rev., 20,
61-87 (2001)

15

l7. Lecchi, P.; Pannell, K. L. .1. Am. Soc. Mass Spectrom., 6, 972-975 (1995)
18. Fitzgerald, C. M.; Parr, R. G.; Smith, M. L. Anal. Chem, 65, 3204-3211 (1993)

19. Jespersen, S.; Niessen, A. M. W.; Tjaden, R. U.; Van der Greef, J. J. Mass
Spectrom., 33, 1088-1093 (1998)

20. Distler, M. A.; Allison, J. Am. Soc. Mass Spectrom., 13, 1129-1173 (2002)

21. Grandori, R.; Matecko, I.; Mayr, P.; Miiller, N. J. Mass Spectrom., 36, 918-922
(2001)

16

Chapter Two: Introduction to Matrix-Assisted Laser
Desorption/Ionization Time-0f-F1ight(MALDI-TOF) Mass

Spectrometry

Introduction

Since its introduction in 1988, matrix-assisted laser desorption/ionization mass
spectrometry (MALDI-MS) has been established as an important analytical instrument
for biochemical and biomedical research. The successful use of a matrix extended the
application of original laser desorption MS to large molecules [1]. It has many
advantages over other ionization methods. Table 2.1 compares different ionization

methods of MS for the analysis of proteins and peptides.

It was reported that MALDI-MS can be used to determine molecular weights up to

1,500,000 [2]. It has the advantage of high accuracy and short analysis time. It has been

successfully used in the analysis of proteins, oligonucleotides, and biopolymers [3].

l7

Table 2.1: Ionization Methods for Analysis of Proteins and Peptides by MS [29]

 

 

 

Ionization method For MW Advantages Disadvantages
informatio
11 upper
mass limit
(D3)
Fast atom 20,000 Ability to analyze Matrix interference
mixtures; MW and below m/z 500; widely
bombardment limited sequence variable response;
information for pure peptide-induced
small peptides; suppression effects;
adaptable for LC/MS; limited to flow rates of
widelLavailable. 510 LL/min.
Electrospray 2150,000 Ability to analyze Multiple charging may
mixtures; limited complicate
ionization sequences interpretation of data;

information for pure glycoproteins may not
small peptides; adapts yield useful

 

easily for LC/MS; information.
most suitable for
quadruple MS.
Matrix-assisted laser >250,000 Highest achieved Low resolution; ability
mass range for to detect structural
desorption/ionization proteins and variants; not presently
glycoproteins; adapted for LC/MS;
relatively insensitive limited structural
to salts; ability to information.

analyze mixtures;
simple to operate.

 

 

 

 

 

 

 

Plasma desorption $50,000 Simple to operate; Low resolution; ability
ability to analyze to detect structural
mixtures; limited variants; not presently
sequence information adapted for LC/MS;
for pure small long acquisition time,
peptides. often >lh.

Instrumentation

The linear MALDI-TOF-MS instrument diagram is shown in Figure 2.1. Ions are formed

in the MALDI source. A N2 laser with a wavelength of 337 nm (pulsed width of 3 ns) is

18

used. ions are accelerated by an electric field in the ion source. They then drift down the

flight tube and hit the front plate of the detector.

 

   
     
 
 

 

 

 

 

 

 

 

 

 

Linear
detector
- __ _>
lon path
------ >
Laser path
Beam
guide
wire
Laser Video
.camera
Laser _ :""'" Aperture (grounded)
attenuator
I Ground grid
Prism V ~ . ‘ ‘
Variable-voltage Sample
grid loading
. chamber
Sample plate Main
source chamber

 

 

 

Figure 2.1: Voyager-DE Mass Spectrometer

l‘)

The analyzer coupled to MALDI is most often time-of—flight (TOF) MS. The separation
of different ions with different m/z values is based on their different flight times in the
field-free drift region. The basic formula for TOF mass analysis is shown using the

following equations:

Kinetic Energy: K.E. = 1/2 my2 = qu
Velocity of the ion: v = (2qu/m)/2

i=1... / v= Ld (m/q)”/(2Es)”

q: charge on the ion

E: accelerating electric field

Ld: the length of the flight region
t: flight time

s: the length of the accelerating electric field

A problem associated with the linear TOF-MS is poor resolution. Because of the kinetic
energy distribution [30], even ions with the same m/z values will reach the detector at
different times. A reflectron, shown in Figure 2.2, was designed to solve this problem.
After ions travel through the field free drift region, they are decelerated and reflected by
an electrostatic field. The slower ions will spend less time in this region and catch up
with the faster ions. This can compensate for the initial velocity differences. The
reflectron mode can also be used for post-source decay (PSD) studies to get more

structural information, by detecting fragment ions.

20

Laser beam

 

 

I Detector x ~. '
Ion mirror

 

 

 

Figure 2.2: Reflectron TOF mass spectrometer

MALDI Experiment

A routinely used MALDI sample preparation procedure is called the “dried-droplet”
method. It is performed as follows: take 1 11L of matrix solution, which is usually
dissolved in ACN/water (v:v=l :1), and 1 11L analyte solution, which is often dissolved in
water. These two solutions are either mixed sequentially on the sample well, or they are
mixed first in a vial, then the mixture is applied to the sample plate. The droplet is air-

dried and ideally, analyte-doped matrix crystals are formed.

Sample preparation is the most important step for the MALDI experiment. Many
research studies have been carried out to investigate the influence of different sample
preparation methods on the spectrum [6, 8, 13-18]. The variations such as: matrix,
solvent system, pH, and evaporation time are very important for a successful MALDI
experiment. A low pH and organic solvents are commonly used in conventional MALDI
experiments. They don’t correspond to conditions that would preserve the native

conformation of proteins. They are the main limitations to the application of MALDI-

21

MS to non-covalent complex studies. It would be helpful to get a deeper insight into

these MALDI parameters when better conditions are explored to study non-covalent

complexes.

Matrix

The use of a matrix is a breakthrough to apply mass spectrometry to intact biomolecule

analysis. It is believed to serve several functions [28].

1.

It isolates analyte molecules. The molar ratio of matrix-to-analyte is usually from
100:1 to 50,000:1 [12]. The presence of a large excess of matrix separates the
analyte molecules and thus the strong molecular forces between analytes are

reduced.

It absorbs energy from the laser and transfers it to the solid crystal lattice. There
are mainly two types of laser sources used, UV lasers and IR lasers. Photons in
the UV-wavelength range can excite molecules electronically, whereas the
irradiation in the IR region will excite them vibrationally or rotationally. Usually,
the analyte biomolecules have no absorption at the wavelength of the laser. When
the laser energy is deposited in the matrix, it causes the excitation of many matrix
molecules. This energy can flow into the crystal lattice, which is usually formed
by hydrogen bonds, and heats up a small volume. Both matrix and analyte
embedded in the matrix experience a phase transition into the gas phase in a very

short time. Figure 2.3 shows the desorption/ionization process. Since the laser is

22

not directly absorbed by the analyte, they are kept intact. That is why MALDI is

characterized by little fragmentation and is called a “soft” ionization technique.

3. It aids in ionization of analyte molecules via a series of photochemical reactions.

This will be discussed in the pH section.

[M+H]+

 
 
  

Fag?" [A+H]+ M: Matrix
J A: Analyte

 

 

Matrix
Sample embedded

Figure 2.3: Photon absorption by the matrix in MALDI, causing desorption and
ionization.

The search for a new matrix candidate is still a trial-and-error procedure. So is the choice
of a matrix for a specific analyte. So far, hundreds of matrices have been evaluated.
Figure 2.4 shows the structures of several most commonly used matrices. Most of them
contain a carboxylic acid group because it may be easy for them to give up a proton
during the ionization process. The acidic matrix could be a problem for noncovalent
complex study since most proteins tend to denature at low pH. Is the acidic functional
group required for a good matrix? Actually, the real importance of the acidic firnction

groups is not established yet. It seems that they are not required for a successful MALDI

23

analysis. Matrices without carboxylic acid groups have also been reported [25]. More
and more basic matrices have been investigated to meet different goals such as the study
of non-covalent complexes [4], because they allow the sample to be prepared under non-

denaturing conditions.

H c O O
3 \
0 \JK OH
OH
HO C
|0 HO lll
CH3 N
sinapinic acid (SA) a-cyanzgzlhyédcrlolrggnnamrc
O O

CH
HO HN 3
0H )\ll/

OH H
2, 5-dihydroxybenzoic acid 6-aza-2-thiothymine
(DHB) (ATT)

Figure 2.4: Structures of some most commonly used matrices

Different matrices have very different properties, which have a great influence on the
crystallization. For a certain analyte, some matrices yield strong signals but others don’t
work at all. It is not well understood how matrices behave and how to select a matrix.
Based on experience, some general guidelines are known. For example, sinapinic acid is
suitable for high molecular weight proteins. However, for a specific analyte, this

information is far from sufficient.

24

The most remarkable property for matrix molecules is their ability to incorporate analyte
into their crystal lattice and form analyte-doped matrix crystals. The doping level
depends on the affinity of the analyte to absorb to the matrix and the relative diffusion
rates of the matrix and analyte [5]. Different matrices have different doping levels for a
specific analyte. Solvent composition and crystallization conditions can also alter the
doping level. It was found that the doping could be blocked by the presence of
contaminants such as salts and detergents [5]. Strupat and coworkers found that the
intimate interactions between matrix and analyte are essential for analysis of large

analytes in the MALDI process [ l ].

Different morphology of matrix crystals also influences the MALDI experiment. Under
microscopic inspection, many matrices accumulate at the rim of the sample well after
crystallization. Only sinapinic acid seemed to evenly distribute around the whole well
[19]. Some matrices such as nicotinic acid or sinapinic acid form small crystallites,
which are about 10 pm in size. Others such as DHB form needle-like crystals which are
several hundred micrometers in length [19]. Sample morphology can differ dramatically
depending on what kind of matrix is used. Sample surfaces are heterogeneous,
containing two phases: the crystalline matrix phase and the isotropic phase with analyte
doped in it [20]. Usually the best result is obtained from the homogeneous area. The

heterogeneity can be a potential problem in MALDI.

Matrix has a great influence on fragmentation. Some matrices are classified as “hot”

since they are more capable to induce fragmentation. Karas and coworkers found that,

25

for protonated glycoproteins, post-source decay decreases in the order of SA>DHB>HPA
[21]. The energy released from the proton transfer reaction can induce fragmentation.
The difference in proton affinity between analyte and matrix can determine the energy
available [22]. Proteins and peptides have a higher proton affinity than most of the
matrices and thus can extract a proton from a matrix molecule. To detect the intact
molecule and prevent fragmentation, a matrix with a higher proton affinity should be
chosen. Whether or not the sublimation temperature of matrices also influences the

internal energy of the analyte is still under debate [23].

Solvent
The most commonly used solvent in the MALDI experiment is the combination of water

and an organic solvent such as acetonitrile, methanol, or ethanol.

There are mainly two reasons to use organic solvents:

1) Most matrices are organic molecules and are not very soluble in water. The addition of
organic solvents can increase the solubility of the matrix [6].

2) The use of organic solvents can facilitate rapid crystal growth. Some organic solvents
are volatile and evaporate very quickly. This can speed the crystallization process [7] and

have a significant influence on the analyte distribution in the MALDI crystals [26].

The crystal growth in MALDI experiment is quite different from that under equilibrium

conditions. Carroll and Beavis [5] addressed the relationship between the speed of

crystal growth and doping levels. In the case of mother liquor where crystals grow very

26

slowly, the rate of protein diffusion is greater than the rate that protein is removed fi'om
the solution into the growing matrix crystal. But the protein molecule has to be able to
stay long enough on the crystal surface to get trapped inside it. The doping level in this
case will be limited by the rate of adsorption of proteins to the growing crystal. Another
extreme case is, the crystals grow very fast with respect to protein diffusion, and the
doping level will only depend on the bulk concentration of the protein present. In the
MALDI experiment, we usually take the latter one, so the proteins included in the matrix
crystals is at approximately the same molar ratio of matrix to analyte as in the original
solution. Conditions such as the composition of water and organic solvents, and the
molar ratio of matrix to analyte, can be manipulated to alter the doping level and thus

obtain the best result.

A study of the effect of the different composition of water and ACN on the crystals and
mass spectra was done by Vaidyanathan et al. [8]. Different combinations of water and
ACN were tested. Figure 2.5 shows the distribution of sinapinic acid matrix crystals on
the MALDI-MS target wells as observed by scanning electron microscopy. Solvent a)
contains more organic solvent and evaporates very fast. It appears to induce smaller and
finer crystals. Solvent c) contains more water and evaporates more slowly. The resulting
crystals are larger and feather-like. Those grown in solvent b) contain both crystal types
but the smaller ones predominate. These experimental results show that smaller crystals

yield better spectra than the bigger ones.

27

 

a) b) c)

Figure 2.5: Distribution of sinapinic acid matrix crystals on the MALDI-MS target wells
as observed by scanning electron microscopy
Despite these advantages of fast crystallization, slow crystallization techniques are used

in some case to overcome suppression effects on the matrix crystallization process

causing by involatile additives [27].

Water must be present to maintain surface tension. Small organic molecules have very
small surface tension and the sample will spread. This will cause a dilution effect and

induce irregular crystal formation.

P!
In most MALDI experiments, trifluoroacetic acid or formic acid is used. The use of acid
can help the dissolution of some analytes [9]. Trifluoroacetic acid is also found to

improve the sample homogeneity in some instances.

The second reason that acid is used is to maintain a low pH environment. Most matrices

discovered so far are carboxylic acids and they become ions when the pH is higher than

28

their pK, values. Their salts crystallize quite differently from the fi'ee acid form of the

molecule [10, l 1].

Another reason is concerned about the MALDI mechanism. The MALDI mechanism is
still under debate. One approach made by Ehring et al. is the photochemical ionization
model [24]. This model assumed that the ion-generation in UV-laser
desorption/ionization (LDI) is initiated by photoionization and formation of radical ions
that then react with neutrals to form the final ions. Extending this approach to MALDI,
they proposed that the matrix has to form gas-phase acidic ions. They act as proton
donors and transfer a proton to analyte molecules [21].

MH+ + A —+ M + AH+

To fully protonate the matrix, a low pH is preferred.

But that is not the only way for the matrix to get protonated. Excited-state proton transfer
(ESPT) is also a popular model. The exited matrix molecule is more acidic than when in
its ground state. It is easier to give up a proton to the nearby analyte or matrix.

M + by -—> M*

M* + A —+ (M-H)’ + AH+

M + M* _. (M-H)‘ + MH+

MALDI is such a complex phenomenon that no unified model has yet been set up to

explain it. It comprises the formation of protonated, cationized and even radical species.

29

Karas et al. suggested that this process is relatively independent of the matrix, solvent

composition, solution pH, and analyte acid-base properties [10].

30

References

l.

10.

ll.

12.

l3.

14.

15.

l6.

17.

18.

19.

Karas, M.; Bachmann, D.; Bahr, U.; Hillenkamp, F. Int. J. Mass Spectrom, Ion
Proc., 78, 53-68 (1987)

Schriemer, C.D.; Li, L. Anal. Chem, 68, 2721-2725 (1996)

Hillencamp, F.; Karas, M.; Beavis, C. R.; Chait, T. B. Anal. Chem, 63, 1193a-
1202a(1991)

Fitzgerald, C. M.; Parr, R. G.; Smith, M. L. Anal. Chem, 65, 3204-3211 (1993)

Carroll, A. J .; Beavis, C. R. Laser Desorption and Ablation in the series: Methods
in Experimental physics Chapter 7

Vorm, O.; Roepstorff, P.; Mann, M.; Anal. Chem, 66, 3281-3287 (1994)
Busch, L. K. Spectroscopy , 14, 14-16 (1999)

Vaidyanathan, S.; Winder, L. C.; Wade, C. S.; Kell, B. D.; Goodacre, R. Rapid
Commun. Mass Spectrom., 16, 1276-1286 (2002)

Overberg, A.; Hassenbiirge, A.; Hillenkamp, F. M. L. Gross(ed.), Mass Spectrom.
Biol. Sci: Tutorial, 353, 181-197 (1992)

Karas, M.; Gliickmann, M.; Schafer, J. J. Mass Spectrom., 35, 1-12 (2000)

Jensen, C.; Haebel, S.; Andersen, O. S.; Roepstorff, P. Int. J. Mass Spectrom, Ion
Proc., 160, 339-356 (1997)

Watson, T. J. Matrix-assisted Laser Desorption/Ionization, chapter 10 in

Introduction to Mass Specrometry 278-302 (1997)

Zhang, N.; Doucette, A.; Li, L. Anal. Chem, 73, 2968-2975 (2001)
Karas, M.; Hillenkamp, F. Anal. Chem, 60, 2299-2301 (1988)
Xiang, F .; Beavis, R. C. Rapid Commun. Mass Spectrom., 8, 199-204 (1994)

Xiang, F.; Beavis, R. e. Org. Mass Spectrom., 28, 1424-1429 (1993)

Li, L.; Golding, R. 13.; Whittal, R. M. J.Am Chem. Soc., 118, 11662-11663 (1996)
Dai, Y. Q.; Whittal, R. M.; Li, L. Anal. Chem, 68, 2721—2725 (1996)

Strupat, K.; Karas, M.; Hillenkamp, F. Int. J. Mass Spectrom, Ion Proc., 111, 89-
102 (1991)

31

20.

21.

22.

23.

24.

25.

26.

27.

28.

29.

30.

Doktycz, S. J .; Savickas, P. J .; Kriiger, D. A. Rapid Commun. Mass Spectrom., 5,
145-148 (1991)

Karas, M.; Bahr, U.; Strupat, K.; Hillenkamp, F. Anal. Chem, 67, 675-679 (1995)
Zenobi, R.; Knochenmuss, R. Mass Spectrom. Rev., 17, 337-366 (1998)

Mowry, D. C.; Johnston, V. J. J. Phy. Chem, 98, 1904-1909 (1994)

Ehring, H.; Karas, M.; Hillenkamp, F. Org. Mass Spectrom., 27, 472-480 (1992)
Fitzgerald, C. M.; Parr, R. G.; Smith, M. L. Anal. Chem, 65, 3204-3211 (1993)

Homeffer, V.; Forsmann, A.; Strupat, K.; Hillenkamp, F.; Kubitscheck, U.
Anal. Chem, 73, 1016-1022 (2001)

Cohen, L. s.; Chait, r. B. Anal. Chem, 68, 31-37 (1996)

Hillenkamp, F .; Karas, M.; Ingendoh, A.; Stahl, B. Biol. Mass Spectrom.
49-60

Carr, A. S.; Hemling, E. M.; Bean, F. M.; Roberts, D. G. Anal. Chem, 63,
2802-2824 (1991)

Cotter, J. R. Analytical Chemistry News & Features, 445A-451A (1999) July

32

Chapter Three: Review of Study of Cytochrome c Oxidase Using Mass

Spectrometry

Introduction

Cytochrome c oxidase (CcO) is the terminal enzyme in the respiratory chain in both
mitochondria and bacteria. Cytochrome c oxidase contains 4 redox active metal centers
to catalyze the process. The electrons from cytochrome c are accepted by Cu,, and then
transferred to heme a, finally to heme a3 and the Cub binuclear center, where the
molecular oxygen gets reduced and water is produced. The biological process is

summarized in Figure 3.1.

 

Figure 3.1: Biological function of cytochrome c oxidase
(Picture adapted from: http://www.bch.msu.edu/research/ppg/cyox.htm)

33

The polypeptide composition of CcO is variable, depending on the evolutionary stage of
the organism. 2-4 subunits have been identified in bacteria, 9 in yeast and 13 in
mammalian tissues [1 1]. Figure 3.2 shows the ribbon structure of the l3-subunit enzyme.
Subunits I, II, and III are the three largest subunits. Surprisingly, they are similar in all
the members of this heme-copper oxidase superfamily. They are believed to be the most
important functional subunits in C00. Organism at higher evolutionary stages contains a
number of small subunits, which are thought to help stabilize and modulate the activity of

the enzyme [5].

 

Figure 3.2: 13-Subunit cytochrome c oxidase structure
(Picture adapted from: http://www.msu.edu/~hillier/MF_CcO%20pics.htm)

34

Over the last years, many analytical techniques such as EPR, and absorbance and
resonance Raman spectroscopies, have been used to improve the understanding of the
structure and function of this enzyme [7]. With the development of soft ionization
techniques yielding little or no fragmentation, such as matrix-assisted laser
desorption/ionization and electrospray ionization, mass spectrometry is becoming more
and more important in the analysis of large biomolecules including CcO. The study of
CcO has been mostly reported by MALDI, since MALDI has a higher tolerance of

buffers and salts than ESI.

The Study of Cytochrome c Oxidase Using Mass Spectrometry

CcO presents a challenge to mass spectrometry due to its hydrophobicity. CcO is a
membrane protein, usually embedded in lipid bilayers. Some of its domains are in
contact with the alkyl chains of the lipids, since they are highly hydrophobic. The other
regions that are exposed to the aqueous phase are more hydrophilic and soluble in water.
Subunits I, II, and 111 from bovine heart are mainly located in the membrane space. If
CcO falls apart in solution, hydrophobic subunits may aggregate. This could make them
difficult to detect by MS. Detergents can be used to help solubilize the hydrophobic
proteins. It was found that the detergent lauryl maltoside is a good one to fully disperse
CcO [9]. Fortunately, it is a nonionic detergent and is compatible with MALDI analysis.
Table 3.1 shows the relative effects of detergents on signal quality in MALDI [15]. The
study proved that ionic detergents are detrimental to MALDI MS analysis. They might
interfere with the crystallization process. Nonionic detergents can be tolerated at

relatively higher concentrations without degrading the spectrum.

35

Table 3.1: The Relative Effects of Detergents on Signal Quality in MALDI MS [15]

Class Effect on MALDI spectrum

1 nooeo en...a...wave...nen nan... ’
2 little effect

3 ‘ “seminary 3. signal intensity sneer

4

spectrum supressed: detergent must be removed

 

 

 

 

 

 

 

{ Detergent Class
I n-octyl-glucoside l
I n-dodecyl-glucoside 1
I octanoyl-N-methylglucamide l
I decanoyl-n-methylglucamide l
f n-dodecyl-beta-D-maltoside 2
r octylphenolpoly(ethyleneglycolether)lo 3

(Triton X-100)

 

1
octylphenolpoly(ethyleneglycolether)7

 

 

 

 

 

 

 

 

(Triton mm 3
I polyethylene glycol (PEG 2000) 3
r dodecylpoly(ethyleneglycolether)9 4
(Thesrt)
I isotridecylpoly(ethyleneglycolether)3 4
I CHAPS 4
rCHAPSO - 4
[ n-dodecyl—N,N-dimethyl- 4
3-ammonro- 1 -propanesulfonate
’ sodium dodecylsulfate (SDS) 4

36

The investigation of CcO using mass spectrometry has become even more complicated
since the purification of the oxidase by different methods (even from the same tissue)
yields enzymes that may differ in subunit composition, aggregating state, and
phospholipids content [5]. The different starting forms make it very hard to find a
consistent MALDI condition that can be applied to the detection of the enzyme that has
been purified using different methods. Studies done so far have proven the success of

different matrices for MS detection of the enzyme, purified in different ways.

Cytochrome c oxidase purified from R. sphaeroides was studied by Ghaim et al. [1].
Three subunits were detected. The enzyme sample was in 0.05% lauryl maltoside, 100
mM potassium phosphate buffer and pH 8.0. The original concentration of the enzyme
was 50 uM. In the MALDI experiment, the enzyme was diluted 50:] with water. Matrix
was prepared by dissolving 1.5 mg of HABA([2-(4-hydroxyphenylazo (benzoic acid)] in
1 mL of solvent containing 66% acetonitrile, 33% water, 0.1% TFA. The calculated
molecular weights for the 3 subunits are: subunit 1: 63,986; subunit 11: 32,940; subunit
111: 30,139. Singly-charged subunits I and III were clearly detected, as were their doubly
and triply charged species. Subunit II can be possibly assigned to a peak at m/z 29,947,
although it is less than the expected value by 2993 Da, possibly due to loss of 25 amino

acid residues [1]. Their spectra are shown in Figure 3.3.

37

8)

 

 

 

 

 

 

 

 

 

TI 1 I
' lib 1?: 15293 i
.. 9?, mm» 8188:” Sub 1.. ..
i 215986 "j 29947r 31898 «
'- SUbI c '1
SUDI“ \ 5111) III ‘
1 / - 30530 _
.l
_n I 1 J 4 ~
10000 20000 30000 40000
b) ’_I T “*1 1""— v
>4 i _ Sub 1'
.‘g' " I 6376I\ "'
c:
2
E
0
>
‘5
.e.
0
o:
l _l l I I
40000 50000 60000 70000
m/z

Figure 3.3: MALDI mass spectrum of the Nizt NTA resin purified His-tagged aa3-type
cytochrome c oxidase complex from R. sphaeroides. a) MALDI spectrum of subunits
with less than 40,000 Da molecular weight. b) MALDI spectrum of subunits larger than
40,000 Da molecular weight.

Our lab also studied CcO from R. Sphaeroides in which four subunits were observed in
MALDI. The starting enzyme solution contained 10 mM KHzPO4, 1 mM EDTA, 0.2%
lauryl maltoside, pH 7.2. The matrices used in our experiments were made using 1:1
acetonitrile/water solution. When sinapinic acid (SA) was used as matrix, only subunits
I, II, and III were detected in the MALDI mass spectrum. Subunit IV was absent. When
2,5—dihydroxybenzoic acid(DHB) was used instead, subunit IV was detected, both the

processed form and unprocessed form. But each of these two versions is 131 Da lower

38

than the expected value, which is due to the loss of the N-terminal methionine. The
spectrum from ESI-FTMS of subunit IV is in good agreement with the result from

MALDI-MS.

The mammalian enzyme, containing 13 nonidentical subunits, has been studied by high
resolution SDS-polyacrylamide gel electrophoresis. It is by far the most complex
membrane protein to have its structure solved [13]. The three largest subunits are
encode in the mitochondrial DNA and are the core of the enzyme. The smaller ten

subunits are nuclear-coded [6].

The earliest study of C00 subunit components from bovine heart in mass spectrometry
was carried out by Schindler et al. [8]. Subunits VIIIa and VIIIb were isolated from

bovine heart and successfully detected in ESI-MS.

CcO isolated from bovine heart and liver was studied by Marx and coworkers [2]. After
a series of purification procedures, the final solution contained 0.5 M NaCl, 0.05%
lauryl-B-D-maltoside or 1% octyl-B-D-glucopyranoside, 5 mM NH4HCO3 buffer, pH 7.6.
The matrix used was sinapinic acid. It was prepared as a saturated solution in 3:1
water/acetonitrile.

The results are shown in Table 3.2.

39

Table 3.2: Molecular Masses of the Subunits of Bovine Heart and Liver Cytochrome c
Oxidase

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Subunits Expected Detected Expected Detected
molecular weight (M+H)+ molecular (M+H)+
(bovine heart) weight (bovine
liver)
1 56,993 ND 56,993 ND
11 29,918 ND 29,918 ND
111 26,023 ND 26,023 ND
IV 17,153 ND 17,153 17,156
V (Va) 12,436 12,436 12,436 12,436
VI (Vb) 10,670 10,670 10,670 10,670
VH (VIb) 10,063 10,066 10,063 10,066
VIII (VIa) 9,436 ND 9,539 9,541
IX (V Ic) 8,479 8,524 8,479 8,520
X (VIIb) 6,674 6,676 6,619 6,620
XI (VIIa) 6,357 6,358 6,357 6,357
XII (V IIc) 5,441 5,442 5,441 5,440
XIII 4,962(Lys) 4,962 5,048 5,047
(VIII) 4,992(Arg) 4,992

 

 

 

 

 

40

 

Most of the subunits were detected except subunits I, II, III, IV, and VIII for the enzyme
from bovine heart and subunits I, II, and III for the enzyme from bovine liver. The
reason they suggested for the disappearance of subunits I, II, and III was, those
hydrophobic subunits tend to aggregate even in the presence of detergent and are no
longer accessible for MALDI-MS. Most of the molecular weights of the detected
subunits match the published values very well except that some corrections were made

for subunits IX and X.

Improved results were obtained by Musatov [3]. The protein (25 11M) was solubilized in
20 mM his-S04 buffer, pH 7.4 containing 2 mM dodecyl maltoside. The molecular
weight of each subunit was verified by MALDI-MS and ESI-MS. For the ESI-MS
analysis, samples that elute from HPLC were mixed with 50% acetonitrile/O.5% acetic
acid to yield 2 11M protein. 1] subunits eluted from the reversed-phase HPLC and were
detected by ESI-MS. However, the spectrum was not shown. The two most
hydrophobic subunits, I and III, precipitate on the column and thus are not detected. For
the MALDI experiment, luL of a solution containing 2 11M CcO and 20 mM dodecyl
maltoside was mixed with SA dissolved in acetonitrile/T FA (v:v=1:1) on the sample
plate. All of the 13 subunits were observed in MALDI. Compared with Marx’s result
[2], the improved MALDI spectrum might be due to the increased concentration of lauryl

maltoside.

41

The solution conditions such as the type of detergent used and its concentration, the
buffer and other components in the solvent, are all essential for the study of cytochrome c
oxidase in mass spectrometry. That could explain why different matrices are most useful
for analysis of the enzyme from the same tissue, but in solutions resulting from different

purification procedures.

Although the polypeptide composition is the main part of this enzyme, the other
components can not be ignored. The heme group, metals, and phospholipids are all
tightly associated with the protein and thus should also be considered as part of the
enzyme. Six phospholipids have been resolved in the crystal structure from R.
Sphaeroides CcO [4]. The amount of lipids found associated with the oxidase complex
depends on the purification methods. In our lab, Distler et al. [4] did a broad mass
spectrometric analysis of lipid and heme components of C00 from R. Sphaeroides. In the
study, the improved detection of lipid was achieved by using DHB as matrix dissolved
inlzl acetonitrile/water. The masses of the lipids and the type of lipids present were
determined in reflectron mode MALDI-MS. The structural information was obtained by
running post-source decay (PSD) analysis. MALDI-MS was demonstrated to be a strong
tool to monitor the lipids content after various levels of purification of the enzyme. In
addition, a peak at m/z 852 was confirmed to represent the heme group by a PSD

experiment.

Another challenge CcO presents to mass spectrometric analysis is, it is a noncovalent

complex of several different subunits. For CcO from bovine heart, the molecular weight

42

for the monomer calculated for the protein moiety is 204,005 Da and that for the other
constituents (except for the cholic acid) identified in the electron density map is 6,998
[6]. Although it was expected to see one peak representing the whole enzyme, the
complex completely dissociated in the MALDI experiment and only individual subunit
was detected. How and where the complex fell apart in the MALDI experiment is still
unclear. The interaction with matrix, organic solvent and low pH used in the MALDI
experiment could all make contributions. It was found that some organic solvents
denature proteins including cytochrome c oxidase [10]. pH also has an effect. The
optimal pH range for the enzyme to maintain its activity is 5.0 to 8.5. Evidence showed
that protein denaturation occurred at pH 4.5 [12]. Previous studies showed that some
compounds such as fucose [14] can be used in the MALDI experiment as an additive to
help stabilize the noncovalent complex in the harsh MALDI sample preparation
conditions. Distler et al. reported the detection of the intact enzyme from R. Sphaeroides
by the addition of sucrose. The spectrum is shown in Figure 3.4. Several peaks were

observed representing the complexes of CcO subunits [4].

43

(”HY

 
  
    
 

 

 

  

 

 

E (II+H)* ,
g II+II+III+IV+HI
“ * |

i (III+H)"

I ll+lll+H ’ (l+ll+|ll+H1

. ( ) (l+ll/III+H)

(1+ V+H
(lll/ll+C+H) VA ) l]
1 0000 30000 50000 70000 90000 1 1 0000 1 30000

m/z

Figure 3.4: Positive-ion linear MALDI-TOF spectrum of cytochrome c oxidase from R.
Sphaeroides. SA/sucrose was used as the matrix.

Besides the protein subunits and other components, there are also some other studies that
have been carried out on CcO. In Marx’s study, 7-diethyl-amino-3-(4’-

maleimidylphenyl)-4-methylcoumarin was used to investigate cysteine status [2].

Mass spectrometry was also used to investigate the chemical modification of the enzyme
[3]. Bovine heart enzyme was incubated in lipid peroxidation product 4-hydroxy-2-
nonenal (HNE) and then studied by MALDI-MS and ESI-MS. Both of these methods

showed that 6 subunits were detected as HNE additives, subunits 11, IV, Vb, VIIa, VIIc

44

and VIII. Identification of the HNE reaction site in subunit VIII was achieved by ESI

tandem ESI-MS analysis.

The study of cytochrome c oxidase using mass spectrometry is far from being mature.
The challenge it placed on mass spectrometry will give us a great opportunity to
investigate better conditions to apply MALDI-MS and ESI-MS to membrane protein
analysis. The use of detergents, the presence of other components in the enzyme and the

stabilization of the whole noncovalent complex remains to be further studied.

45

References

1. Ghaim, B. J .; Tsatsos, H. P.; Katsonouri, A.; Mitchell M. D.; Salcedo-hemandez, R.;
Gennis, B. R. Biochimica et Biophysica Acta 1330, 113-120 (1997)

2. Marx, K. M.; Mayer-Posner, F.; Soulimane, T.; Buse, G. Anal. Biochem. 256, 192-
199 (1998)

3. Musatov A.; Carroll, A. C.; Liu, Y. Henderson, 1. G.; Weintraub, T.S.; Robinson, C. N.
Biochem. 41, 8212-8220 (2002)

4. Distler A. M.; Qin, L.; Hilmi, Y.; Hiser, C.; Ferguson-Miller, S.; Allison, J.
unpublished.

5. Gregory, C. L. Ph.D. Dissertation, Michigan State University (1988)

6. Tsukihara, T.; Aoyama, H.; Yamashita, E.; Tomizaki, T.; Tamaguchi, H.; Shinzawa-
Itoh, K.; Nakashima, R.; Yaono, R.; Yoshikawa, S. Science 272, 1136-1144 (1996)

7. Michel, H.; Behr, J .; harrenga, A.; Kannt, A. Annu. Rev. Biophys. Biomol. Struct. 27,
329-356 (1998)

8. Schindler, A.P.; Van Dorsselaer, A.; Falick, M. A. Anal. Biochem. 213, 256-263
(1993)

9. VanAken, T.; Foxall-VanAken, S.; Castleman, S.; Ferguson-Miller, S. Methods in
Enzymology 125, 27-35 (1986)

10. Yu, C.; Yu, L.; King, E. T. J. Biol. Chem. 250, 1383-1392 (1975)

11. Kadenbach, B.; Jarausch, J .; Hartrnann, R; Merle, P. Anal. Biochem. 129, 517-521
(1983)

12. Gregory, L. Ferguson-Miller, S. Advances in membrane biochemistry and
bioenergetics 301-309 (1988a)

13. Gennis, R. B.; Ferguson-Miller, S. Current Biology 6, 36-38 (1996)
14. Distler, M. A.; Allison, J. Anal. Chem. 73, 5000-5003 (2001)

15. Ole, V.; Chait, T. B.; Roepstorff, P. 41th ASMS Conference Proceedings 621a-621b
(1994)

46

Chapter Four: The Detection of Cytochrome C Oxidase Subunits Using

Mass Spectrometry

Experimental Section

I worked with several different bovine heart cytochrome c oxidase (CcO) samples, which
contained different buffers and detergents.

Sample A was isolated from bovine heart using the Yoshikawa purification procedure
with a minor modification (See Appendix A). The enzyme is in a 0.2 M phosphate buffer
(pH=7.4), 25% (NH4)2SO4 and 0.5% sodium cholate.

Sample B was purified, using an Ultrafree-15 centrifugal filter (Millipore, Bedford MA)
from sample A in 20 mM N-2-hydroxyethylpiperazine-N'-2-ethanesulfonic acid (Hepes)-
KOH (pH 7.4), 15 mM KCl and 0.01% lauryl maltoside.

Sample C contains 10 mM phosphate buffer (pH=7.4), 1 mM ethylenediaminetetraacetic
acid (EDTA), 180 mM KCl and polyethylene glycol dodecyl ether (Brij-35)
(concentration is unknown). The concentration of CcO is 108 1.1M, which was

determined by UV-VIS spectrometry.

The compounds, 2, 5-dihydroxybenzoic acid (DHB), 6-aza-2-thiothymine (ATT),
sinapinic acid (SA), and a-cyano-4-hydroxycinnamic acid (CHCA) were purchased from
Sigrna-Aldrich (St. Louis, MO). Standard solutions of these MALDI matrices were
prepared using water/acetonitrile (v:v=1:1). Water was used to dilute the CcO samples to

5 11M. Equal volumes of matrix and protein solutions were deposited on the MALDI

47

steel sample plate. Calibration mixtures 2 and 3, purchased from PerSeptive Biosystems

(Framingham, MA), were used for calibration.

Reversed-Phase High Performance Liquid Chromatography: the HPLC analysis of CcO
subunits was performed using gradient elution from a C13 reversed phase column and a
Microm HPLC system. The gradient was made from mixtures of solvent A (5% [0.1%

formic acid] and 95% ACN) and solvent B (95% [0. 1% formic acid] and 5% ACN).

Linear MALDI mass spectra were recorded on a PerSeptive Biosystems (Framingham,
MA) Voyager delayed extraction (DE) time-of—flight (TOF) mass spectrometer with a
nitrogen laser (337 nm, 3 ns pulse length). Typically 50 laser shots were averaged for
each spectrum. The accelerating voltage was 20 kV, the delayed time was 500 ns, the
grid voltage was 94% of the accelerating voltage, and the magnitude of the guide wire

voltage was 0.1% of the accelerating voltage.

FTMS spectra were acquired on a Bruker Daltonics (Billerica, MA) Apex 111 mass

spectrometer with a 7 Tesla superconducting shielded magnet.

Results and Discussion
UV Absorption Spectra
The concentrations of cytochrome c oxidase samples were determined by UV
spectrometry. Figure 4.1 shows the UV spectra of oxidized and reduced forms of the

purified CcO enzyme sample C. The spectrum of the oxidized form (spectrum a) has an

48

absorption peak at 416 nm and the reduced form obtained with sodium dithionite has

absorption peaks at 440 nm and 602 nm (spectrum b).

0.1 '

0.05

 

 

i” ’——’“ ‘—'_“——“ ‘flr‘ ‘—_ "'

40 450 500 55? f 600 650 700

nm

Figure 4.1: a.) Absorption spectrum of oxidized form of cytochrome c oxidase;

b.) Absorption spectrum of reduced form of cytochrome c oxidase (reduced with
Na2S204)

The protein concentration can be calculated using the Lambert-Beer law:

A = ebC

A: absorption

8: extinction coefficient (37,000 L cm'lM'l for cytochrome c oxidase»

b: thickness of the cell (0.05 cm in our experiment)

C: concentration

49

The absorption at 602 nm in spectrum b is 0.07044. The spectrum of the oxidized form
(spectrum a) has no absorption at 602 nm and thus can be used as the background, which
is 0.05.

A = ebC

That is: (007044-005) = 37,000 L cm"M"* 0.05 cm * C

The concentration of cytochrome c oxidase can be determined as 108 pmol/uL.

The concentration of the other cytochrome c oxidase samples can be determined using

the same method.

The Detection of Cytochrome c Oxidase Subunits Using MALDI TOF MS

The three largest subunits of cytochrome c oxidase (CcO) are encoded in the
mitochondrial DNA and are the core of the enzyme. The smaller ten subunits (subunits
IV to XIII) are nuclear-coded. In the conventional MALDI experiment, all of the ten
nuclear coded subunits were detected using MALDI TOF MS. Figure 4.2 shows the
spectrum obtained for cytochrome c oxidase sample C using sinapinic acid (SA) as the
matrix. Among the tested matrices, SA is the best for the detection of the ten smaller
subunits in terms of the resolution and signal intensity. The spectrum is easy to interpret,
as the predominant peaks are the protonated molecules of these subunits. No
fragmentation was observed. These peaks can be used to determine the molecular weight
of each subunit. The results are summarized in Table 4.2. However, the problem with
SA is the adduct formation. Small peaks due to SA attached to subunits were observed.

For example, the inset of Figure 4.2 shows the peak representing protonated subunit X111

50

at m/z 4,964. Beside it, a peak at m/z 5,172 was detected. The mass difference between

these two peaks is 208, which corresponds to the molecular weight of sinapinic acid.

 

 

 

 

 

 

 

 

 

 

 

 

 

[X111+Hl’
x111 4,964
X11 i
1
Q1 fl
.m ll +
a . HDGIHSA]
.2 VII (15.172
1—1 ll 1
d) l i I I]
.5 x1 -
% I X [X
m V
V111
VI
1V
, 111....1 111...,11. . .,_.1.... _. ___.._
2000 20000
m/z

Figure 4.2: Positive-ion linear MALDI-TOF spectrum of cytochrome c oxidase. The ten
nuclear coded subunits were detected. SA was used as matrix.

The Detection of Cytochrome c Oxidpse Subunits UsingESI-FTMS
Compared with MALDI, ESI has a lower tolerance for impurities such as detergents,

buffers and salts. CcO sample C was analyzed using ESI-FTMS and only peaks

51

representing singly charged detergent molecules were detected. The reason might be due
to the presence of a high concentration of detergent. To eliminate the detrimental effects
of the detergent, CcO was purified using reversed-phase high performance liquid

chromatography. The HPLC chromatogram is shown in Figure 4.3.

b)

1200000

A‘S w
T) 4:»

4:4
...“‘.":.‘.
I...“ o—o—o *- .._4

1000000 1
l
l

———0—

800000 10.

'. ‘ “H ..-OO'

600000

v—o ,.__ —.. ..-o

400000

1
i
19
ll
ii

200000

 

0

   

60

~200000

 

" Time (minute)
-40oooo

Figure 4.3: HPLC chromatogram of CcO.

The solution that eluted from the HPLC was introduced into the electrospray ion source
using a syringe pump at a rate of one micromolar/minute. The spectra obtained by ESI-
FTMS were deconvoluted and many peaks were observed. Figure 4.4 shows the
deconvoluted spectrum of the fraction eluted from the HPLC at 20 minutes. However,
only a few of these peaks have isotopic distributions. For example, if we zoom in on the

peak around 4,960, a set of isotopic peaks is observed as shown in Figure 4.4. The

52

isotOpe distribution can help us determine which of these deconvoluted peaks actually

represent a true species.

 

 

 

 

Relative Intensity

 

 

l
l
i
|

 

 

 

 

l _ p i __g _ .__._ _ __lw l

.. l— ____. h. LAC. i.,_ _ _.¥ IT __ ___t___JL.ll':a g __ c, f. I“
5,000 7:000 9,000 l l ,000 13,000 1 5,000

El _-u
i

m/z

Figure 4.4: Deconvoluted ESI-FTMS spectrum of cytochrome c oxidase eluted from
HPLC at 20 minute.

Table 4.1 summarized all the peaks that have a set of isotopes in the deconvoluted ESI-

FTMS spectra of fractions collected from HPLC and their masses.

53

Table 4.1:

Isotope Distribution Peak List for All the Peaks Observed in the

Deconvoluted ESI-FTMS Spectrum

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Peak A (m/z) Peak B (m/z) Peak C (m/z) Peak D (m/z) Peak E (m/z)
4957.2045 5436.41 19 6352.6384 6668.6893 7504.8336
4958.3002 5437.4722 6353.6658 6669.7213 7505.8708
4959.2949 5438.5006 6354.6963 6670.7480 7506.9162
4960.3258 5439.5052 6355.7200 6671.7726 7507.9551
4961.3281 5440.5233 6356.7595 6672.7833 7509.0003
4962.3630 5441.5279 6357.7105 6673.7901 7510.0412
4963.4429 5442.5622 635 8.7238 751 1.0836
6359.7960
Peak F (m/z) Peak G (m/z) Peak H (tn/z) Peak 1 (m/z) Peak J (m/z)
8514.5227 9942.1964 10041.6835 10055.6054 11352.6916
8515.4653 9943.1704 10042.6751 10056.7167 11353.9618
8516.4898 9944.1430 10043.6686 10057.8310 11355.3582
8517.5481 9945.1156 10044.6627 10058.9472 11356.7659
8518.6178 9946.3154 10045.6572 10060.0595 11358.1727
8519.5989 9947.2975 10046.6521 10060.9132 11359.4670
8520.6478 9948.2701 100476480 100620241 1 1360.8719
8521.7085 9949.2441 10048.6412 10063.1392 11362.2854
8522.7783 9950.2214 10049.6340 10064.2565 l 1363.6982
9951.4239 10050.6310 10065.3694
9952.4021 10051.7079 10066.2225
10052.7054 10067.3373
Peak K (m/z) Peak L (m/z) Peak M (m/z) Peak N (m/z)
l 1392.4589 12426.6438 12290.7631 13076.0436
11393.5167 12427.6141 12292.2669 13077.4887
l 1394.3728 12428.6788 12293.4870 13078.6609
l 1395.3990 12429.7340 12294.7044 13080.1064
11396.3789 12430.7552 12295.9366 13081.2788
11397.4204 12431.7966 12297.1391 13082.7291
11398.2673 12432.8371 12298.3579
11399.2867 12433.8763 12299.8567
11400.3017 12434.9200 12301.0776
12435.9873 12302.2986

 

 

 

 

 

54

 

 

Some of the peaks listed in Table 4.1 could possibly represent cytochrome c oxidase
subunits. For example, the m/z 4960 peaks in the deconvoluted spectrum (Figure 4.5 a))
are very close to the molecular weight of subunit XIII. To determine if this peak really
represents subunit XIII, we compared this spectrum with the theoretical isotope
distribution of subunit XIII that is shown in Figure 4.5 (b). They are very similar, but the
peak with the highest relative intensity detected in the deconvoluted ESI-FTMS spectrum
is 1.3 Da off the theoretical value. This is most probably due to a systematic error.

Meanwhile two isotope peaks that have the lowest relative intensity are not detected.

Relative Intensity

 

Figure 4.5: a) Deconvoluted ESI-FTICR spectrum. B) Theoretical isotope distribution
for subunit X111

55

By comparing the deconvoluted spectrum with the theoretical isotope distribution, peak
A, B, C, D, F, I, and L in Table 4.1 can be determined to represent CcO subunits XIII,
XII, XI, X, IX, VII and V respectively. The molecular weight information determined
from ESI FTMS confirmed the results obtained from MALDI-TOF-MS. Other peaks in
Table 4.1 are not completely identified yet. Their masses don’t match the molecular
weight of any subunit. Some of them might represent some subunits with some amino
acid residues falling off. For example, the average m/z value of peak G is around 9,948.
It is 77 mass units lower than the molecular weight of subunit VII, which is 10,025
calculated from the amino acid sequence. Considering the systematic error, the mass
difference is very close to the molecular weight of the residue mass of alanine. The
alanine is the C- terminal residue of subunit V11. So peak G could represent subunit VII

with alanine residue falling off.

Table 4.2 summarized the results from MALDI MS and ESI FT MS. Most of the
molecular weights of the CcO subunits detected in MALDI-TOF-MS and ESI-FT MS
match the values calculated from the reported amino acid or DNA sequence very well
except for subunits VII, VIII, IX and X. The mass data indicates that the published
sequences are correct for subunits IV, V, VI, XI, X11, and XIII. There is no

posttranslational modification at these subunits.

56

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Table 4.2: Molecular Weights of Protein Subunits of Bovine Heart Cytochrome C

Oxidase

Subunits Expected molecular MALDI-TOF- ESI-FTMS g
weight“ MS ‘

1b 57, 032 NDg ND

11b 29,932 ND ND

111b 26,021 ND ND

1yb 17,152 17,157 ND

v b 12,436 12,436 12,432

v1 ° 10,670 10,671 ND

v11d mpg; 10,063 10,0_60

V111c 2,43_6_‘ 9,532 ND

1xb 8.519; 8,524 8,518

xb gm 6,667 6,671

x1 b 6,357 6,355 6,355

x11 b 5,441 5,440 5,440

x111 b 4,962(Lys) 4,964 4,961

 

 

 

 

 

 

Molecular masses (Da) were determined either from direct amino acid sequence
analysis or by DNA sequence analysis. b Reference 14. ° Reference 15. d Reference
16. ° Reference 17. f Experimentally determined mass (Da) by MALDI-MS.

g Experimentally determined mass (Da) by ESI-FTMS. h ND, not detected. i The
results from MALDI MS and ESI FTMS aren’t in agreement with the values calculated
from the reported amino acid or DNA sequence.

In Table 4.2, subunit X shows molecular weights of 6,667 and 6,671 Da in MALDI MS
and ESI FTMS respectively, the value calculated from the amino acid sequence reported

by Yoshikawa et al. [14] being 6,244. The mass difference between the published value

57

and the value resulting from mass spectrometry is about 423. The error is greater than
the molecular weight of an amino acid residue. The molecular weight determined from
the gene sequence is 6,674 [19], Which is much closer to our value detected by mass
spectrometry (A[real mass-expected mass]=7). Yoshikawa’s sequences resulting from
the crystal structure failed to detect some amino acid residues. The C-terrninal residues
of their protein sequence of subunit X, -H-G-H-A-S-K-K, must be corrected to —G-W-A-

S-F-P-H-K-K.

The same reason can be applied to subunit VIII. The molecular weight detected by
MALDI MS is 9,532, the value calculated from the amino acid sequence reported by
Yoshikawa et al. [14] being 9,436. The difference between these two is 96, which
corresponds to the molecular weight of residue phenylalanine (P). This is conformed by
the value determined from the gene sequence [17]. Yoshikawa’s sequences resulting
from the crystal structure failed to detect the C-terrninal residue —P, which will result in

an additional 97 Da [17].

For subunit IX, The mass difference between the m/z detected by MALDI MS and the
expected molecular weight is about 42. The reason might be that the N-terminus could
be blocked with an acetyl group, which will give a molecular weight 42 Da higher than
the molecular weight calculated from the amino acid sequence. If this is true, the
molecular weight of subunit D( will be 8,521, which is very similar to the value we

detected by MALDI MS.

58

For subunit VII, The molecular weight determined by MALDI MS and ESI FTMS
10,063 and 10,060, while the value calculated from the protein sequence is 10,025. The
additional 38 mass unit could indicate an acetylation at the N-terminus of this
polypeptide chain. But if this is true, the acetylation will result in an addition of 42 mass
units. This difference of 4 mass units could be resulted from four cysteine residues
forming two disulfide bonds. This will eliminate four hydrogen atoms, which will give a
molecular weight four mass units less than the theoretical value. This conclusion is in
agreement with the study of Marx et al.. It was found that four cysteine residues in
subunit VH are involved in the formation of disulfide bridges by the use of 7-diethyl-

amino-3-(4’-maleimidy1phenyl)-4-methylcoumarin (CPM) [21].

Based on the above discussion, the correct sequences and molecular weights (without

methionines oxidized) of the subunits of CcO from bovine heart have been shown in

Table 4.3.

59

 

 

 

 

 

 

 

 

 

Table 4.3: Sequences and Molecular Weights of Protein Subunits of Cytochrome c

Oxidase from Bovine Heart

Subunits Sequence Molecular

Weight

1a MFINRWLFST NHKDIGTLYL LFGAWAGMVG TALSLLIRAE 57,032
LGQPGTLLGD DQIYNVVVTA HAFVMIFFMV MPIMIGGFGN
WLVPLMIGAP DMAFPRMNNM SFWLLPPSFL LLLASSMVEA
GAGTGWTVYP PLAGNLAHAG ASVDLTIFSL HLAGVSSILG
AINFITTIIN MKPPAMSQYQ TPLFVWSVMI TAVLLLLSLP
VLAAGITMLL TDRNLNTTFF DPAGGGDPIL YQHLFWFFGH
PEVYILILPG FGMISHIVTY YSGKKEPFGY MGMVWAMMSI
GFLGFIVWAH HMFTVGMDVD TRAYFTSATM IIAIPTGVKV
FSWLATLHGG NIKWSPAMMW ALGFIFLFTV GGLTGIVLAN
SSLDIVLHDT YYVVAHFHYV LSMGAVFAIM GGFVHWFPLF
SGYTLNDTWA KIHFAIMFVG VNMTFFPQHF LGLSGMPRRY
SDYPDAYTMW NTISSMGSFI SLTAVMLMVF IIWEAFASKR
EVLTVDLTTT NLEWLNGCPP PYHTFEEPTY VNLK

113 MAYPMQLGFQ DATSPIMEEL LHFHDHTLMI VFLISSLVLY 26,021
IISLMLTTKL THTSTMDAQE VETIWTILPA IILILIALPS
LRILYMMDEI NNPSLTVKTM GHQWYWSYEY TDYEDLSFDS
YMIPTSELKP GELRLLEVDN RVVLPMEMTI RMLVSSEDVL
HSWAVPSLGL KTDAIPGRLN QTTLMSSRPG LYYGQCSEIC
GSNHSFMPIV LELVPLKYFE KWSASML

111a MTHQTHAYHM VNPSPWPLTG ALSALLMTSG LTMWE‘HFNSM 29,932
TLLMIGLTTN MLTMYQWWRD VIRESTFQGH HTPAVQKGLR
YGMILFIISE VLFFTGFFWA FYHSSLAPTP ELGGCWPPTG
IHPLNPLEVP LLNTSVLLAS GVSITWAHHS LMEGDRKHML
QALFITITLG VYFTLLQASE YYEAPFTISD GVYGSTFFVA
TGFHGLHVII GSTFLIVCFF RQLKFHFTSN HHFGFEAAAW
YWHFVDVVWL FLYVSIYWWG 8

1V3 AHGSVVKSED YALPSYVDRR DYPLPDVAHV KNLSASQKAL YL152
KEKEKASWSS LSIDEKVELY RLKFKESFAE MNRSTNEWKT
VVGAAMFFIG FTALLLIWEK HYVYGPIPHT FEEEWVAKQT
KRMLDMKVAP IQGFSAKWDY DKNEWKK

Va SHGSHETDEE FDARWVTYFN KPDIDAWELR KGMNTLVGYD 12,436
LVPEPKIIDA ALRACRRLND FASAVRILEV VKDKAGPHKE
IYPYVIQELR PTLNELGIST PEELGLDKV

qu ASGGGVPTDE EQATGLEREV MLAARKGQDP YNILAPKATS “L670
GTKEDPNLVP SITNKRIVGC ICEEDNSTVI WFWLHKGEAQ
RCPSCGTHYK LVPHQLAH

Vqlc AEDIQAKIKN YQTAPFDSRF PNQNQTRNCW QNYLDFHRCE juy067
KAMTAKGGDV SVCEWYRRVY KSLCPISWVS TWDDRRAEGT
FPGK I (N-terrninal is blocked with an acetyl group)

VIIId ASAAKGDHGG TGARTWRFLT FGLALPSVAL CTLNSWLHSG 9,533
HRERPAFIPY HHLRIRTKPF SWGDGNHTFF HNPRVNPLPT GYEKP

 

 

 

 

60

 

 

[Xe STALAKPQMR GLLARRLRFH IVGAFMVSLG FATFYKFAVA 8,521
EKRKKAYADF YRNYDSMKDF EEMRKAGIFQ SAK
(N-terminal is blocked with an acetyl group)

x3| IHQKRAPDFH DKYGNAVLAS GATE‘CVAVWV YMATQIGIEW 6358
NPSPVGRVTP KEWREQ

 

 

XIt FENRVAEKQK LFQEZDNGLPV HLKGGATDNI LYRVTMTLCL 6,674
GGTLYSLYCL GWASFPHKK

 

 

x11a SHYEEGPGKN IPFSVENKWR LLAMMTLFE‘G SGFAAPE‘E‘IV 5,442
RHQLLKK

x111a ITAKPAKTPT SPKBQAIGLS VTFLSFLLPA GWVLYHLDNY 4,962
KKSSAA

 

 

 

 

 

a Reference 14. b Reference 15. ° Reference 16. a Reference 17. 3 Reference 18.
f Reference 19.

Why Weren’t Subunits I, II and III Detected?

While the conventional MALDI experiment allowed the detection of the ten smaller
subunits, the intact CcO complex and the three largest subunits were not detected.
Instrument settings were adjusted to achieve better results for the higher mass range, but
this alone did not result in the detection of additional peaks. This result was
disappointing since subunits I, II and III are the core subunits of cytochrome c oxidase.
They are believed to be the most important firnctional subunits to catalyze the biological

process. The question is, why weren’t the three core subunits of CcO detected?

A possible mechanism is proposed as shown in Figure 4.6. Under the traditional MALDI
conditions, the CcO complex dissociates. Cytochrome c oxidase is a membrane protein
and contains hydrophobic segments. Subunits I and III are located in the membrane

space and are more hydrophobic than the other smaller ones. To minimize the contact

61

with water and thus increase the enthalpy, these hydrophobic subunits tend to aggregate

in solution. Their aggregation makes them no longer accessible to the MALDI analysis.

Denaturing
factors 0
denature

  
  

protein Hydrophobic
I subunits aggregate
O o Q Q (1
O

O O

Figure 4.6: Mechanism for the CcO analysis in the conventional MALDI experiment

Some approaches have been developed for the analysis of hydrophobic proteins and
peptides by MALDI-MS or ESI-MS. Different solvent compositions have been
successfully used to prevent hydrophobic peptides or proteins from aggregating.
Schindler and coworkers reported that chloroform/methanol/water mixtures worked well
with bacterioopsin in ESI-MS [l]. Neat formic acid was found to give the best result for
bacterioopsin (BO) from halobacterium halobitun and chloroform/methanol/water
(2:5:2,v/v/v) containing 2% acetic acid was found to be the best for analyzing the BO
fragments by ESI-MS [2]. The acidic conditions could result in the esterification of the
threonine and/or arginine amino acid residues and cleavage of acid-labile peptide bonds
[10]. Green-Church and Limbach used nonaqueous solvents such as chloroform and
chloroform/methanol solutions to prepare the matrix that they used. Hydrophobic

peptides were dissolved in chloroform [3]. This approach is excellent to detect

62

hydrophobic analytes but not good for the characterization of hydrophilic peptides, thus
not one is ideal for hydrophobic/hydrophilic mixture analysis. Recently, Bird et al. [4]
used a new method, temperature-leap tactic, to maintain the solubility of hydrophobic
proteins. All of these solutions don’t preserve the protein native conformation and thus

are not ideal for our analysis of CcO.

Detergents have also been used to solubilize hydrophobic proteins [5]. They are used to
extract membrane proteins. Detergent molecules are amphiphilic molecules. Above their
critical micelle concentration (CMC), detergent micelles are formed (Figure 4.7). Their
nonpolar tails are hidden in the interior region through hydrophobic interactions and their
polar heads are exposed to the water environment. Due to this unique property,

detergents can solubilize hydrophobic proteins by forming micelles with them.

2 Critical Micelle fig}.
Hydrophilic head Concentration (CMC) K /"

. it?

A detergent-micelle in water

 

 

 

Hydrophobic tail

Figure 4.7: The formation of detergent micelles

Detergents can be classified as three types: ionic, non-ionic, and zwitterionic, based on
the nature of their hydrophilic head. Ionic detergents contain a head group with a net
charge. Non-ionic detergents contain uncharged, hydrophilic head groups. They are

better suited for breaking lipid-lipid and lipid-protein interactions than protein-protein

63

interactions. They are considered non-denaturing. Another advantage of non-ionic
detergents is that they can be tolerated at relatively high concentrations in the MALDI
MS analysis without degrading the mass spectra. These two advantages make it a

possible solution to our problem.

The Modified MALDI TOF MS Experiment for the Detection of Subunits I, II and
m

Experimental Section

To investigate the effects of different detergents on the detection of the three hydrophobic
subunits of C00 in MALDI-TOF-MS, the following detergents were evaluated: sodium
cholate, lauryl maltoside, decyl maltoside, polyoxyethylene 23 lauryl ether (brij-35),
triton-lOO, and polyethylene glycol sorbitan monolaurate (tween-20). The structures and

physical properties of these detergents are shown in Table 4.4 and Figure 4.8.

64

Table 4.4: Detergents Evaluated for CcO Analysis Using MALDI MS

 

 

 

 

 

 

 

 

 

 

Detergent Detergent Type M.W. CMC (mM)

Sodium cholate Anionic Bile acid 414.6 9-15

Triton-X100 Non-ionic 650 0.25
Polyoxyethylene

Tween-20 Non-ionic 1228 0.059
Polyoxyethylene

Brij-35 Non-ionic 1200 (avg.) 0.09

Lauryl Maltoside Non-ionic 5 10. 1 l .8

 

 

n~9.5

Sodium cholate

Triton-X 1 00

65

 

HO(CH2CH20)“’ (OCHZCHZY‘OH

0 CH(OCH2CH2)yOH

CH20(CH2CH20)""CH2CHZO

        

c57‘3‘CICHziocHe.

Sum of w+x+y+z=20 Tween-20
c,,H,5(OCH,CH,)n0H

n~23

Brij-35

HOCH, HorzH2

0
HOIIII llllOllll OCH2(CH2)1UCH3

H0 70H HO 0H
Lauryl Maltoside

Figure 4.8: The structure of the detergents evaluated.

66

The matrices SA, ATT, DHB were used by dissolving them in ACN/water (v:v=1:1)
solution. Equal volumes of matrix and protein solutions were deposited on the MALDI
sample well. Bovine serum albumin (BSA), purchased from Sigma-Aldrich (St. Louis,

MO), was used for calibration.

Linear MALDI mass spectra were recorded on a PerSeptive Biosystems (Framingham,
MA) Voyager delayed extraction (DE) time-of—flight (TOF) mass spectrometer with a
nitrogen laser (337 nm, 3 ns pulse length). Typically 50 laser shots were averaged for
each spectrum. The accelerating voltage was 25 kV, the delayed time was 600 ns, the
grid voltage was 90% of the accelerating voltage, and the magnitude of the guide wire
voltage was 0.3% of the accelerating voltage. All spectra were obtained in positive

mode.

Results and discussion

Different detergents show very different properties that have a very important influence
on the protein solubilization and MALDI analysis. The type of detergent used for a
specific protein and the amounts are critical for a successful solubilization of the
membrane protein [6]. The selection of a proper detergent for a target protein is a trial-
and-error procedure. Some studies found that triton X—100 [7] and sodium cholate [8] are
effective at solubilizing CcO. But they actually don’t influence the detection of subunits

I, II and III in MALDI-MS.

67

Lauryl maltoside (LM) turned out to be the best for the analysis of the hydrophobic
subunits of CcO in MALDI-MS. This result is in good agreement with the study of
Ferguson-Miller, which showed that lauryl maltoside is the best detergent to fully
disperse CcO [9]. Using lauryl maltoside as an additive to the MALDI matrix, subunits I,
II and III were successfully detected. Subunit II is easier to be detected since it is

relatively hydrophilic. The spectrum is shown in Figure 4.8.

However, a new question was raised: since detergent molecules separate these
hydrophobic subunits by forming micelles with them, we would expect to see a peak in
the MALDI spectrum representing the detergent micelle bound to each subunit. A typical
lauryl maltoside micelle consists of about 140 monomers and weighs around 76,000 Da if
there is no protein present in the solution. If a lauryl maltoside micelle is formed around
the protein, a larger micelle would be expected to accommodate the protein molecule.
The bigger a protein molecule, the larger a detergent micelle would be expected. It was
found that the lauryl maltoside micelle around a CcO monomer weighs about 106,000 Da
[20]. We can estimate that the lauryl maltoside micelle around each hydrophobic subunit
will weight between 76,000 and 106,000 Da. For example, we would expect to see a
peak representing subunit III bound to lauryl maltoside (LM), (subunit III) (LM),40-210 in
the MALDI spectrum (See Figure 4.9). In fact, it was not observed. Only individual

subunits were detected. Why wasn’t the subunit-detergent micelle complex detected?

68

II

(Subunit III)(LM),,,0_210 ?

 

III

Relative Intensity

‘ i A A

  

 

 

20,000 m/z 100,000

Figure 4.9: Positive-ion linear MALDI-TOF spectrum of cytochrome c oxidase.
Subunits I, II and III were detected. ATT was used as matrix. Lauryl maltoside was used
as an additive.

Actually, this is another example of the dissociation of a non-covalent complex in
MALDI-MS. To illustrate this issue, it needs more detailed study on the mechanism of
how the protein is incorporated into the matrix crystal lattice. Beavis and Bridson did a
study using trans-sinapic acid [11]. Crystals of sinapic acid were produced in a solution
containing a specific dye (Coomassie Brilliant Blue G250) to determine which of the
crystal faces were responsible for the protein inclusion. The crystals were grown by slow
evaporation. X-ray crystallography suggested that the face that is responsible for the
interaction with an analyte protein shows great non-polar properties. It lacks free

hydroxyl groups for hydrogen bonding. It was concluded that hydrophobic interactions

are the primary interactions.

69

If this mechanism is true, we can explain the dissociation of the detergent-subunit
complex in the following way: the hydrophobic tails of detergent molecules bind to the
hydrophobic portions of the protein. The resulting micelle no longer has the amphiphilic
character. When the micelle comes into contact with the growing crystal surface, the
competition between matrix and detergent for the hydrophobic portion of the protein
results in the dissociation of the micelle, releasing the protein to the crystal surface. This

process is shown in Figure 4.10 1).

Some researchers have questioned Beavis’s study. Firstly, their crystals were grown
under an equilibrium state, which is quite different from the real MALDI fast growth of
crystals [12]. Secondly, the presence of the dye could have an effect on the
crystallization. The dye might be specifically binding to hydrophobic faces. As a result,
a question was raised: is the incorporation of analytes into matrix crystals a prerequisite
for MALDI? The concept of analyte incorporation as a prerequisite was ruled out by the
study of Homeffer et al. [12] and Gliickmann et al. [13]; however their work is not

widely accepted.

The dissociation of detergent-subunit complexes could also have occurred in the
desorption/ionization process since the interaction between them is very weak. The
detergent micelle-subunit complexes didn’t dissociate when the crystals were formed.
The complexes were trapped inside the matrix crystal lattice. Upon the laser irradiation,
the detergent molecules fall off from the complex. This process is shown in Figure 4.10

2).

70

DECIDE]

1 L313<>lj 13:] D

DC] C1 DC] [:1

O DECIDE DECIDE

16?
Q

DDDDDDD

I3DI3131313C1 DUDE“: (3%;
I3131313C11313 2 gfig l]::>
DDDDDDD 1313:11313 O

D Matrix 0 Hydrophobic subunit 2 Detergent

Figure 4.10: The detergent micelle-subunit complexes could fall apart in MALDI MS. 1.
the complexes fall apart when they are incorporated into the matrix crystals. 2. The
complexes dissociate in the desorption/ionization process.

There are some other factors observed that influence the detection of subunits I, II and

111, such as the concentration of detergent and the concentration of the protein.

The spectra obtained using different concentrations of lauryl maltoside solutions are
shown in Figure 4.11. The original CcO sample was in a low concentration of detergent
solution. When water was used to dilute the protein, it also diluted the detergent and
hindered the micelle formation. Theoretically, only above the critical micelle
concentration (CMC), can detergent molecules form micelles and solubilize membrane
proteins by surrounding them with a hydrophobic cavity. However, it was found in our
experiments that even when lauryl maltoside is used, when used at concentrations slightly
above CMC, subunits I, II, and III are still not detected. Using SA as the matrix, when

the concentration of lauryl maltoside was increased to 10*CMC, subunits 1 and II were

71

observed. Subunit III is extremely hydrophobic. It makes up the one of major portions

of the enzyme that is embedded in the membrane [20]. It was not detected in the low

concentration of detergent. When the detergent concentration was increased to 100*

CMC, subunits I, II and III were detected.

Relative Intensity

Mag-w water

 

 

 

 

 

 

 

 

kw 5*CMC
Li I 10*CMC
l
1“ 100*CMC
20,000 m/z 100,000

Figure 4.11: Positive ion linear MALDI-TOF spectra of subunits I, II and III of
cytochrome c oxidase using different concentration of lauryl maltoside as solvent. SA
was used as the matrix. CcO concentration: 111M

72

Similar results were obtained when ATT was used as the matrix. The difference is,
subunits I, II, and III were detected when the lauryl maltoside concentration is as low as
the CMC. Thus, ATT is a better matrix for the detection of the three hydrophobic

subunits.

The effect of the concentration of CcO was also investigated. The spectra are shown in
Figure 4.12. A solution of 8*CMC lauryl maltoside was used to dilute CcO in all the
following experiments. When concentrations of CcO were above 2 11M, subunits I, II,
and III were not detected. Instead, some unknown peaks were observed in that mass

range. These peaks couldn’t be assigned to any of the subunits. They might be resulting
from the aggregation of some smaller subunits. When the CcO concentration was about
1 11M, these unknown peaks were not observed and subunits I and II were observed.
When the CcO concentration was decreased to 0.5 11M, all subunits I, II and III were
detected. It seems that lower CcO concentrations can help detect these hydrophobic
subunits at certain detergent concentrations. The reason might be that at lower

concentrations the aggregation process occurs more slowly.

Similar results were obtained when ATT was used as the matrix. Those three subunits

can be detected when the concentration of CcO was between 5 11M and 0.1 11M.

73

II
CcO concentration 2 1.1M

 

 

 

 

II
a?
g CcO concentration luM
11>) I
02 II "AT—

CcO concentration 0.51.1M

 

111"] III
I

 

CcO concentration 0.111M

 

M‘...A‘h

100,000

 

20,000 m/z

Figure 4.12: Positive ion linear MALDI-TOF spectra of subunits I, II and III of
cytochrome c oxidase in different concentration. SA was used as the matrix. The

concentration of lauryl maltoside: 8*CMC

74

Our experiments proved that the detergent concentration and protein concentration are
critical variables for the detection of the hydrophobic subunits of CcO using MALDI-
TOF-MS. Consider that detergent molecules are present in the solution to help prevent
the hydrophobic subunits from aggregating. What really matters is not the absolute
concentration of CcO but the ratio of the concentration of detergent to protein. At a
certain ratio of [CcO]:[lauryl maltoside], the detergent can fully disperse the hydrophobic
subunits without imposing detrimental effects on the MALDI analysis. When the
[CcO]:[lauryl maltoside] ratio is too high, hydrophobic subunits are not fully solubilized
and may aggregate and not be detected. When the [CcO]:[lauryl maltoside] ratio is too
low, too many detergent molecules might interfere the crystallization process and degrade
the spectrum. If the absolute concentration of CcO is beyond the sensitivity, those

subunits were also not detected.

Conclusion

Cytochrome c oxidase does dissociate in the conventional MALDI experiment. Only the
ten nuclear coded subunits could be detected when conventional MALDI conditions were
used. Improved experiments were achieved by the use of non-ionic detergents as matrix
additives. Among all the detergents evaluated in our experiments, lauryl maltoside is the
best one to fully disperse this membrane protein and make all of its subunits accessible to
MALDI detection. Some small subunits were detected using ESI-FTMS. They

confirmed the results from MALDI-TOF-MS.

75

in

References

1. Schindler, P. A.; Van Dorsselaer, A.; Falick, M. A. Anal. Biochem. 213, 256-
263 (1993)

2. Schaller, J .; Pellascio, C. B.; Schlunegger, P. U. Rapid Commun. Mass Spectrom.
11, 418-426 (1997)

3. Green-Church, B. K.; Limbach, A. P. Anal. Chem. 70, 5322-5325 (1998)
4. Bird, H. G.; Lajmi, R. A.; Shin, A. J. Anal. Chem. 74, 219-225 (2002)

5. Breaux, A. G.; Green-Church, B. K.; France, A.; Limbach, A. P. Anal. Chem.
72, 1169-1174 (2000)

6. Maire, L. M.; Champeil, P.; Méller, V. J. Biochimica Biophysica Acta,
1508, 86-111 (2000)

7. Musatov, A.; Ortega-Lopez, J.; Robinson, C. N. Biochem. 39, 12996-13004
(2000)

8. Musatov, A.; Robinson, C. N. Biochem. 41, 4371-4376 (2002)

9. VanAken, T.; Foxall-VanAken, S.; Castleman, S.; Ferguson-Miller, 8. Methods
Enzymologz 125, 27-3 (1986)

10. Tummala, R.; Ballard, M. L.; Breaux, A. G.; Green-Church, B. K.; Limbach, A.
P. In Advances in Nucleic Acid and Protien Analyses, Manipulation, and
Sequencing 3926, 56-60 (2000)

ll. Beavis, C. R.; Bridson, N. J. J. Phys. D: Appl. Phys. 26, 442-447 (1993)

12. Homeffer, V.; Dreisewerd, K.; Liidemann, C-H.; Hillenkamp, F.; Lage, M.;
Strupat, K. Int. J. Mass Spectrom. 185/186/187, 859-870 (1999)

13. Gliickmann, M.; Pfenninger, A.; Kruger, R.; Thierolf, M.; Karas, M.; Homeffer,
V.; Hillenkamp, E.; Strupat, K. Int. J. Mass Spectrom. 210/211, 121-132 (2001)

14. Tsukihara, T.; Aoyama, H.; Yamashita, E.; Tomizaki, T.; Tamaguchi, H.;
Shinzawa-Itoh, K.; Nakashima, R.; Yaono, R.; Yoshikawa, S. Science 272,
1136-1144(1996)

15. Tanaka, M.; Haniu, M.; Yasunobu, T. K.; Yu, A. C.; Yu, L.; Wei, H. Y.; King E.
T. J. Biol. Chem. 254, 3879-3885 (1979)

76

l6.

17.

18.

19.

20.

21.

Biewald, R.; Buse, G. Hoppe-Seyler's Z. Physio]. Chem. 363, 1141-1153 (1982)

Smith, E. O.; Bement, D. M.; Grossman, L. 1.; Lomax, M. 1. Biochim. Biophys.
Acta 1089, 266-268 (1991)

Erdweg, M.; Buse, G. Biological Chemistry Hoppe-Seyler 366, 257-263 (1985)
Seelan, S. R.; Grossman, I. L. Biochemistry 31, 4696-4704 (1992)

Suarez, D. M.; Revzin, A.; Narlock, R.; Kempner, S. E.; Thompson A. D.;
Ferguson-Miller, S. J. biol. Chem. 259, 13791-13799 (1984)

Marx, K. M.; Mayer-Posner, F .; Soulimane, T.; Buse, G. Anal. Biochem. 256,
192-199 (1998)

77

Chapter Five: The Study of Cytochrome C Oxidase Complexes Using
MALDI-TOF-MS

Experimental Section

Cytochrome c oxidase sample C (See chapter 4 experimental section) was used in all the
experiments. The compounds 2,5-dihydroxybenzoic acid (DHB), 6-aza-2-thiothymine
(ATT), sinapinic acid (SA), and alpha-cyano-4-hydroxycinnamic acid (CHCA) were
purchased from Sigma-Aldrich (St. Louis, MO). The standard solutions of these MALDI
matrices were prepared using water. Ammonium acetate, purchased from Sigma-Aldrich
(St. Louis, MO), was added to adjust the solution pH. Lauryl maltoside was used as a
detergent to solubilize cytochrome c oxidase. Bovine serum albumin (BSA), purchased

from Sigma-Aldrich (St. Louis, MO), was used for calibration.

Linear MALDI mass spectra were recorded on a PerSeptive Biosystems (Framingham,
MA) Voyager delayed extraction (DE) time-of-flight (TOF) mass spectrometer with a
nitrogen laser (337 nm, 3 ns pulse length). Typically, 50 laser shots were averaged for
each spectrum. The accelerating voltage was 25 kV, the delayed time was 800 ns, the
grid voltage was 90% of the accelerating voltage, and the magnitude of the guide wire

voltage was 0.3% of the accelerating voltage.

Results and discussion
The first problem encountered in the analysis of cytochrome c oxidase (CcO) complexes
using MALDI-TOF-MS is the calibration of the high mass range. The molecular weight

of C00 is above 200,000 Da. The calibration species usually used for the high mass

78

range in MALDI MS are: bovine serum albumin (M.W. 66,431), cytochrome c (M.W.
12,361) and Immunoglobulin G (lgG) (M.W. 144, 391). None of these are appropriate

to calibrate spectra of cytochrome c oxidase complexes.

It was observed that, in the positive MALDI mass spectrum of insulin, the insulin
monomer forms the base peak, but a multitude of oligomers is seen as well [1]. When the
analyte concentration is high enough, non-specific aggregation can occur. This can be a
solution to achieving the high molecular weight calibration. Figure 5.1 shows the BSA
dimer and trimer detected in MALDI-TOF-MS, which can be used to calibrate the

spectrum for m/z values as high as that for the cytochrome c oxidase monomer peak.

132,863
[2M+H]+

 

 

 

 

 

 

 

 

 

E‘
M
H
a
.5
9
E
.e
:12
99,648
|3M+2H11+
199,307
|3M+H|+
80001.0 ' 1234883 I 1669766 - 2104644 ' 2539522 Y 2974400

m/z
Figure 5.1: Positive-ion linear MALDI-TOF spectrum of BSA. SA was used as matrix.
Concentration of BSA: 50 11M.

79

In the conventional MALDI analysis of CcO, no ions above 100,000 Da were detected.
Cytochrome c oxidase does dissociate and only individual subunits were observed. The
MALDI experimental conditions could have possibly disturbed the native structure of the
protein. Organic solvents, matrix, and low pH could weaken the non-covalent
interactions. In order to eliminate these detrimental effects, variations in the

experimental conditions were tested.

As mentioned in chapter one, the matrix 6-aza-thiothymine (ATT) is close to neutral in
pH and has found success for the detection of duplex-DNA, protein and peptide
complexes. The first advantage of the use of ATT is that it does not create an acidic
environment. The denaturation of CcO usually occurs at a pH lower than 5 [4]. Another

advantage of ATT is that it is readily soluble in aqueous solution.

Different combinations of water and ACN were tested. When pure water was used as the
solvent, several complexes were detected. The results become more reproducible by the
addition of ammonium acetate. It is not well known why the addition of ammonium
acetate helps the detection of cytochrome c oxidase complexes in MALDI MS. It might
be due to the pH (pH is about 6.5), better cocrystallization, or facilitation in the
desorption/ionization process. Other ammonium salts were also evaluated, such as

ammonium citrate and ammonium bicarbonate. They don’t provide much improvement.

Since cytochrome c oxidase is a membrane protein, it is found that some lipids are still

attached to the protein after it is isolated. Nine phospholipids, five phosphatidyl

80

ethanolamines (PE), three phosphatidyl glycerols (PG) and one phosphatidyl choline
(PC) have been clearly found in cytochrome c oxidase from bovine heart. In addition to
the lipids mentioned above, cardiolipin has long been known as one of the phospholipids
that cannot be removed without the loss of the enzyme activity. Five cardiolipins were

found to be bound to the beef heart enzyme [6].

In an early study of Ferguson-Miller et al. [5], an apparent molecular weight, 300,000-
350,000 resulting from gel filtration, was obtained for CcO from bovine heart in the
presence of lauryl maltoside. In order to identify this species as a monomer or a dimer,
the sedimentation equilibrium analysis in solvents of different densities was performed.
It gave the molecular weight of both the protein moiety and the protein-detergent
complex. The difference between them provided an estimate of the amount of associated
detergent. The mass of lauryl maltoside bound to the beef heart oxidase is about 106,000

amu.

Table 5.1 shows the main species and their approximate molecular weights in CcO from

bovine heart.

81

Table 5.1: The Main Species in Cytochrome c Oxidase Enzyme from Bovine Heart and
Their Approximate Molecular Weights

 

 

 

 

 

 

 

Species Approximate molecular weight
Cardiolipin 1500

5PE+3PG+1PC 6,750

Cytochrome c oxidase monomer 205,000

Heme 852

Lauryl maltoside micelle around CcO 106,000 + 20,000
CcO+5CL+5PE+3PG+1PC+2Heme 326,954

 

 

 

 

Using ATT as matrix, ammonium acetate and laury maltoside as matrix additives, some
peaks representing cytochrome c oxidase complexes are successfully detected in

MALDI-TOF-MS.

82

207,470

Relative Intensity

.4._*__ eh.» 4. ‘—o—
r”
I
1

l

r

l \‘ \J'~./'= - /‘ 1‘

l

l

l
L_.- ________ - “3.3 :33-.. --._ _-- __ __-:-”_- iii-:3
128057.11 1792158 230374.6 2815334 3326922 383851.11

 

m/z

Figure 5.2: Positive-ion linear MALDI-TOF spectrum of CcO complexes. ATT (in 60
mM N1140Ac) was used as matrix. Concentration of CcO: 3 11M.

The peak shown in Figure 5.2 could possibly represent the monomer of cytochrome c
oxidase with thirteen subunits in it. The expected molecular weight is 204,751 while the
m/z value from the MALDI spectrum is 207, 470, which is 1% higher than the expected
value. It is possible that some phospholipids are bound to the monomer and make the

m/z value higher than the actual molecular weight of CcO monomer.

83

414,764

1
1
l
l

‘7- A - - -

E‘

m

=

8 l

.5 i l

o .

.2. :

:5 l 1

0 i

a: 1 E
l 1 1
11'. 1
t ' 1‘ 1 WV.“ . K r I IV ' ‘ ‘

214774.0 309957.2 405140.4 500323.6 95506.8 90690.0

m/z

Figure 5.3: Positive-ion linear MALDI-TOF spectrum of CcO complexes. ATT (in 60
mM NH40Ac) was used as matrix. Concentration of CcO: 3 uM.

The peak shown in Figure 5.3 could represent the dimer of cytochrome c oxidase
((subunits I-XIII)2). The expected molecular weight is 409,502, while the m/z detected in
MALDI-MS is 414,764. The m/z value for the peak in MALDI mass spectrum is 5,264
higher than the expected value. It could have seven phospholipids or three cardiolipins

attached to the protein.

84

112,744

 

.‘E‘ i l
"=1 + l
E. I l
o . ,
.2 1 1 1
‘3 J , ,
Ta 1 1 i
9‘ » i r
1 l
1 5 1
. a
i u
. i
L? -2. —- --- 1 ~ , *g my _ ”"‘:.'.___:"”“'“V —_\;f_.\i‘xu,./;10
104205.0 128587.4 152969.8 77352.2 201734.6 226ll7.0

m/z

Figure 5.4: Positive-ion linear MALDI-TOF spectrum of CcO complexes. ATT (in 60
mM NH4OAc) was used as matrix. Concentration of CcO: 3 uM.

The peak shown in Figure 5.4 could represent the complex of subunit I binding to
subunits 11 and III. Subunits I, II and III are the three core subunits in cytochrome c
oxidase. They are located mainly in the transmembrane space. Subunit I is directly
interacting with subunits II and III. The hydrophobic interactions between them are
stronger than the interactions among the other smaller subunits. The partial dissociation
of the enzyme leads to the smaller subunits falling off while the three largest ones still
remain together. The expected molecular weight for subunits I + II +111 is 112,744 while

the m/z detected in MALDI-MS is 113,500. The difference is 756 Da.

85

321,535

- 'L .Vg.‘

 

5. 147,963 ,1 1
2 1 1 *1 .
3 11 ,1 ‘ 1
,5. 1 1 1 i
1 1
E 1 1 1 1
N 1 1 1 1
a 1 ~ * 1
a: i, 1 1
1 ' 1 1 1 1
. 1.‘ 1 1 1: 1 1
1 . 1 1 1 1
1 1 1
1032810 l66827.8 230374.6 293921.4 3574682 4210150

m/z

Figure 5.5: Positive-ion linear MALDI-TOF spectrum of C00 complexes. ATT (in 60
mM NI-hOAc) was used as matrix. Concentration of CcO: 3 uM.

The peak shown in Figure 5.5 at m/z 147,963 could represent the subunit III-depleted
enzyme. To study the biological function of subunit III, subunit III has been removed by
the use of detergent, incubation, ion exchange or affinity chromatography [1]. These
methods yield preparations containing less than 15% of the normal complement of
subunit 111, but also cause the removal of several of the smaller peptides (subunits VI, VII

and VIII) [1]. This will give a molecular weight of 145,000 Da (205,000-29,918-10,670-

10,067-9436).

86

The other peak in Figure 5.5 at m/z 321,535 could represent the detergent micelle binding
to the monomer. It was found that the lauryl maltoside micelle around beef heart
cytochrome c oxidase has a molecular weight of approximately 106,000 as determined by
the sedimentation equilibrium method [2]. That will give a molecular weight of 327,000
while the mass detected in MALDI MS is 321,535. The difference is about 5,465. This

could mean that there are some phospholipids bound to the protein.

 

115,069
1 1 1
f 1* .
I 1
1 1
3' 1 ., 5
g 1 *1 357,249
.5. 1
2 1
z: 1 . 215,234 3 1
'5 1 1* . ,, 1
o ~ 1
a: 1 1 J i 1 1
|
. ' 1 i
1| 323,430 1 1
79999.0 2040004 328001.8 4520032 576004.15 700006.0
m/z

Figure 5.6: Positive-ion linear MALDI-TOF spectrum of CcO complexes. ATT (in 60
mM N1140Ac) was used as matrix. Concentration of CcO: 3 uM.

87

The peak shown in Figure 5.6 has m/z value of 215,234, which is 10,000 Da higher than
the molecular weight of CcO monomer. This peak could represent five cardiolipins and
three phospholipids attached to the monomer. It could also represent fourteen

phospholipids and two cardiolipin lipids bound to the monomer.

The other peak in Figure 5.6 at m/z 357,249 could present the subunit III-depleted dimer.
The calculated molecular weight is 350,000 (410,000-29,932-10,670-10,067-9436). If it
has five cardiolipins attached, this will give a molecular weight of about

357,500(350,000+7,500).

 

618,632
1 1
1 1 1 1
. 1 1 1
>5 1 1 1
r: 1 5 . 1
2 1 ' 1 1
g 1 1 1 1
1— 1 1 1
9 1 1 1 1
.2 . ‘
3 1 1 1 1
O 1 1
M 1 1 1 1
1 1
1 311,215 1 1
. _..__ ___--M__T___ ._ _._,;. --_, -__-—-:‘_____,'; ~; ;1--_.___ _. . _ ..,2___2. ”,0
272357.0 3595222 4 87.4 5338526 6210173 708183.0
m/z

Figure 5.7: Positive-ion linear MALDI-TOF spectrum of CcO complexes. ATT (in 60
mM NH40Ac) was used as matrix. Concentration of CcO: 3 uM.

88

The peak shown in Figure 5.7 at m/z 311,215 could represent the lauryl maltoside micelle
binding to the cytochrome c oxidase monomer. The calculated molecular weight is

311,000 (205,000+106,000). That matches the mass detected in MALDI MS very well.

The other peak in Figure 5.7 at m/z 618,632 could represent the cytochrome c oxidase

trimer. The expected molecular weight is 615,000.

Actually all these peaks representing CcO complexes detected using MADLI MS were
from a heterogeneous target. MALDI has a notorious reputation for its variability.
Although the analyte and matrix mix very well in the solution, upon the solvent
evaporation, they form heterogenous sample crystals. The radius of the sample well is
about 1 mm, which is much smaller than the size of the laser beam. When the laser
irradiates different sample areas, different MALDI spectra will be obtained, with varying
signal intensity and resolution. In our study of the cytochrome c oxidase complexes,
many different forms of the enzyme have been detected: the intact monomer and dimer,
some phospholipids binding to the protein, the detergent micelle binding to the protein,
and the protein losing some smaller subunits. These complexes were detected in different
MALDI sample spots and made it hard to get consistent results. Why are different
complexes formed in different areas of the sample plate? Consider that the MALDI
crystallization of the matrix and analyte mixtures occurs in a very short time, usually one
to three minutes. The fast solvent evaporation has great influence on the crystal
formation. In some areas of the surface, the solvent evaporates very fast and the

concentrations of analyte molecules and matrix are very high. They get saturated on the

89

surface even they are not saturated in the bulk solution yet. The crystals will be formed
on the surface first. The crystals formed at the beginning of the crystallization process

are very difference from those that are formed at the end of the crystallizatio.

Another problem is the low resolution at the high m/z values. Figure 5.8 shows the peak

representing the cytochrome c oxidase monomer detected in MALDI MS.

205,300

Relative Intensity

 

 

 

 

 

170,000 1 1 230,000
Heme attached Phospholipids m/z
CcO " attached
monomer v
Cardiolipin
attached

Figure 5.8: Low resolution at high m/z values

The half width of this peak is about 8,000 Da. The resolution is about 26. The poor

resolution makes it very hard to tell the actual form of the enzyme detected. The

molecular weight of each phospholipid is about 750 Da. We can not distinguish the

90

monomer from monomer-phospholipids complexes based on this peak. Similarly, there
is no way to determine whether the heme is still bonded to the protein or not. We lack

information to differentiate the different forms of the enzyme due to this wide peak.

The use of the matrix ATT in an aqueous ammonium acetate solution successfully
stabilized the cytochrome c oxidase complexes for the MALDI analysis. Our
experimental results suggest that organic solvents and acids are the primary reasons for
the dissociation of the cytochrome c oxidase complexes. Meanwhile we found evidence
from the lower mass range to prove that some CcO complexes were stabilized in
MALDI. Under this condition, the individual subunit could still be detected, but the

intensity has been dramatically decreased.

Different sample preparation methods have been evaluated. When the cytochrome c
oxidase solution and matrix solution were mixed in a vial first and then applied to the
sample plate, it could increase the sensitivity and reproducibility. This method yielded
more homogeneous crystals than if the cytochrome c oxidase and matrix solutions were

sequentially deposited onto the sample well.

It was reported by our lab that several peaks representing complexes of cytochrome c
oxidase subunits from R. Sphaeroides were detected when sucrose was added to the
cytochrome c oxidase solution [3]. The presence of sucrose can stabilize proteins in

solution. Unfortunately, sucrose doesn’t help stabilize bovine heart enzyme very well.

91

Conclusion

Several cytochrome c oxidase complexes from bovine heart have been detected using
appropriate matrix, solvent, and pH. Our experiments suggest that the organic solvents
and low pH, rather than the desorption/ionization process itself, are the primary reasons
that limit the detection of the cytochrome c oxidase complexes using MALDI MS.
Careful attention to the effect of solvent compositions and pH will improve the capability

of detecting weakly bound non-covalent complexes using MALDI MS.

92

References
1. Gregory, C. L. Dissertation. Michigan State University 1988

2. Suarez, D.M.; Revzin, A.; Narlock, R.; Kempner, S.E.; Thompson, A.D.; Ferguson-
Miller, S. J. Biol. Chem. 259, 13791-13799 (1984)

3. Distler, M. A.; Qin, L.; Hilmi, Y.; Hiser, C.; Ferguson-Miller, 8.; Allison, J.
Unpublished.

4. Gregory, L. Ferguson-Miller, S. Advances in membrane biochemistry and
bioenergetics 1988a, 301 -309

5. Suarez, D.M.; Revzin, A.; Narlock, R.; Kempner, S. E.; Thompson, A.D.; Ferguson-
Miller, S.; J. biol. Chem. 259, 13791-13799 (1984)

6. Tsukihara, T.; Aoyama, H.; Yamashita, E.; Tomizaki, T.; Tamaguchi, H.; Shinzawa-
Itoh, K.; Nakashima, R.; Yaono, R.; Yoshikawa, S. Science 272, 1136-1144 (1996)

93

Chapter Six: Conclusions and Future Work

While the analysis of membrane proteins by mass spectrometry has proven to be difficult,
the multi-component membrane protein cytochrome c oxidase from bovine heart was
successfully detected using MALDI MS. To prevent the hydrophobic subunits from
aggregating, the detergent lauryl maltoside was used. By selecting the right detergent
and appropriate concentration, all of the 13 subunits have been observed and their
molecular weights were determined by MALDI MS and some of them were detected by
ESI FTMS. Another challenge is that the enzyme dissociates in the conventional
MALDI experiment. In this work, cytochrome c oxidase was stabilized by the use of a
neutral matrix and aqueous solvent. As a consequence, some intact complexes were

detected by MALDI MS.

This work is focused on the protein part of this enzyme. However, when analyzing
membrane proteins, in addition to the protein part, it is also essential to characterize the
lipids present. Lipids are important to stabilize the protein and maintain the activity of
the protein. Lipids play a significant role in the structure of the enzyme and should be
considered as a part of it. Many phospholipids are found to be bound to the protein after
the isolation and purification of bovine heart cytochrome c oxidase. Some proof can be
found in the analysis of the cytochrome c oxidase complexes. Peaks with m/z values
between the molecular weigh of the C00 monomer and dimer could be assigned to the
species that have phospholipids attached to them. However, this alone doesn’t provide

information on what types of lipids are present and what their structures are. More direct

94

information can be obtained from the low mass range of the MALDI mass spectrum. The
lipids have molecular weights from 700 Da to 1,500 Da. However there are some
problems we should realize. Lipids are amphiphilic molecules and not soluble in aqueous
solutions, which are usually used in the MALDI experiment. The addition of detergent
could be one of the solutions. But if detergent is used to help the dissolution, the analysis

using MALDI is possibly complicated by the presence of matrix and the detergent.

MALDI MS not only can provide molecular information, it can also determine the type
of lipids present and further elucidate their structure by the performance of post-source

decay (PSD) experiments.

The purification process usually strips some lipids out of the enzyme. As a consequence,
different amount of lipids will be present in different stages of purification. Mass
spectrometry will be a quick way to determine the types of lipids present. The relative
intensities of the peaks can help determine the amount of lipids present. Monitoring the
amount and type of lipids present in the enzyme at various levels of purification will be a

new application of mass spectrometry.

These are all the things to be considered, and additional work needs to be done to

continue the study of bovine cytochrome c oxidase.

95

Appendix

96

 

Appendix: Cytochrome c Oxidase purification
After Yoshikawa et al. with minor modifications

Prepare :a) 1 beef heart (~lkg)

b) l-2L 0.2M NaPi, pH 7.4
c) ~9L of cold water

[:1 grind the whole heart

 

450g

D add

 

1!
2.0 L ice+cold water
+ 0.27 L 0.2M NaPi
homogenize 10’ at ““.Low t=_°C

13
D centr.

supernatan

20' @ 2,800 rpm

sediment

[:1 add:

1.35 L ice+cold water
+ 0.15 L 0.2M NaPi

D homogenize 10' at “Low",t=_°C

[:1 com. 15' @ 2,800 rpm
supernatant

sediment

><

 

<— Prepare container
with ice bag

 

4508 L

D add

2.0 L ice+cold water
+ 0.27 L 0.2M NaPi

1:] homogenize 10‘ at “Low", t=_°C
1:] centr. 20’ @ 2,800 rpm.

supematan

sediment

[3 add:

1.35 L ioe+cold water
+ 0.15 L 0.2M NaPi

D homogenize 10' at “Low" , t= °C

_—

4. Put washing botth

El centr. 15' @ 2.800 rpm
with dist. H20 on ice

supernatant

 

V
A

 

A
‘

sediment

 

1/2 of supematants ( ~3 L):
add 30% acetic acid up to pH 5.15

D centr. 15‘ @ 2,800 rpm
tam
X supema

sediment

wash out with cold dist. H20
(use washing bottles)

”2 of supernatant: (~3L):
add 30% acetic acid up to pH 5.15

1:] centr. 15' @ 2,800 rpm

tant
X supema

sediment

wash out with cold dist. H20

 

volumes are per one preparation (1 beef heart)

97

1

combine the samples in the 3L beaker

 

separate pellets from suspension

homogenizc pellets for l‘ (@ 3000 rpm) in a small blender

 

 

mix with suspension <

rinse the blender & add ice cold dist. H20 up to 2.5 L

DECIDED

centr. l5' @ 2,500 rpm

X supernatant

sediment

transfer into blender with rubber spatula
[:1 add: 100 mL 0.2 M NaPi
1:] homogenize 3‘ (@ 5,000 rpm) in a small blender

[:1 centr. 10‘ @ 2,500 rpm

 

supernatant
sediment
'3 measure volume (V: mL)
[:1 wash homogenizer & centr. tubes: 75 mL 0.2M NaPi (v: mL)

1:] wash cylinder: 52 mL or less 0.2 M NaPi to keep total volume under 454 ml

[:1 add ice cold dist. H20 up to 454 mL

 

1‘ (sample from the second heart. do not mix)
stand overnight

[:1 Prepare Na cholate solution for the second day:

for 20% solution for 30 % solution
75 mL 0.2 M NaPi, pH 7.4 or 75 mL 0.2 M NaPi, pH 7.4..or 50 mL 0.2 M NaPi
+2.94 g NaOH +3 1.6g Na cholate + 3 1 .6g Na cholate
+30g cholic acid
add water up to 150 ml water up to 100 ml

 

volumes are per one preparation (1 beef heart)

98

We;
(if two hearts are used, keep them separately)

adjust pH from to 7.4

add Na cholate to final concentration of 3.2 % (86.4 mL of 20% Na cholate into 454 ml of sample)

! maintain pH 7.3 - 7.4 using lN NaOH !

stir 30' on ice < 5 Prepare 200 mI/hefl of 2.0% and fl
1:] mUheart of 0.5% cholate Na

Prepare dialysis buffer (4Uheart)
Prepare dialysis membrane (321??

131:1 DEC]

El
[:1
D add (NH4)2803: 0-33% or 19.6 g/lOO mL (5.4 dL )4 19.6 = 105.84 g)
C1
[:1

centr. 20’ @ 11,000 rpm

 

supernatant cm/heart)
sediment
[:1 D measure volume (V: ml) (V: ml)
[:1 1:] add (NH4)2SO3: 33-50% or 10.7 g/lOO mL (' g) (__g)
[j centr. 25° @ 11,000 rpm

X supernatant
. sediment

[:1 [:1 add 0.5% cholate. final volume 150 mL
recuspend pellets using glass rod

1:] dialysis 90' against 4 Uhcart of 0.04 M NaPi, pH 7.4

[31:1

E] centr. 30' @ 45,000 rpm

X supernatant
sediment

1:1 1:] add 2.0% cholate, final volume 200 mL
break pellets with glass rod & homogenize in glass homogenizer

1:] 1:] add (NH4)2SO3: 0-25% or 14.4 g/lOO mL (28.8g)
monitor pH 7.3»7.4

1:] [:1 incubate 20' on ice without stirring

 

volumes are per one preparation (1 beef heart)

99

DUE] 1:11:11313 [11:11:]
><

DEC]
1:11:113

1:1

supernatant

centr. l0' @ l7,000 rpm

sediment

 

 

..1

E] measure volume (V: ml)

D add (NH4)2SO3: 2.5-45% or 12.5 g/100 ml.

1:1

supernatant

centr. 5' @ l7.000 rpm

sediment

add 0.5% cholate, final volume 200 mL

add (NH4)2SO3: 0-25% or 14.4 gllOO mL (28.8g)
monitor pH 7.3-7.4

incubate on ice without 20' stirring

[11:1 DE]

centr. 5‘ @ 17,000 rpm

supernatant

sediment

 

1.

[:1 measure volume (V: ml)
E] add (NH4)2SO3: 25-40% or 9.3 g/IOO mL

1:] centr. 5' @ 17,000 rpm
tant
X supema

sediment

add 0.5% cholate. final volume 130 mL
add (NH4)2SO3: 025% or 14.4 g/lOO mL (18.72 g)

incubate 20‘ on ice without stirring

(V: ml)

(__g)( _g)

(___D Make Brij 35 solutions:

LLHLULCLIfl—loi Brij 35

200 mI/heart 0.34%
6.8 mL 10% Brij 35
93.2 mL H20

100 mL 0.2 M NaPi

[:1 200 cart 0.21%
4.2 leO% Brij 35

95.8 mL H20

100 mL 0.2 M NaPi
(V: ml)
(__g)( __g)

 

volumes are per one preparation (I beef heart)

100

DEC]

[31:]

1:11:

[:1 centr. 10' @ 17,000 rpm Prepare dialysis buffer
Prepare dialysis membrane

supernatant

sediment

 

X
1:] measure volume (V: ml) (V=___ml)

[:1 add (NH4)2803: 25-35% or 6.1 g/100 mL (__g)( ___g)

1:] centr. 5’ @ 17,000 rpm
Xsupernatant

sediment

1:1 add 0.34% Brij 35. final volume 200 mL

1:] add (NH4)2803: 0-25% 14.4 yroo mL (28.8 g)
monitor pH 7.3-7 .4

if any aggregation '——-* D [:1 centr: 10‘ @ 17,000 rpm
supernatant

sediment

if not X
l

1 DD measure volume (V=___m|)
1

1:] add (NH4)2SO325-35% (12.6 g) [:1 [:1 add (NH.),S0,: 25-35% or
D J 6.1g/100mL( _g)

centr. 10 @ 19,000 rpm
X supernatant

 

 

 

 

sediment

1:1 1

add 0.21% Brij 35
final volume 160 mL

 

volume up to 100 ml

stand overnight

0RD
1:]

final volume 200 ml

 

1:]
1:1
1 1
1:]

add (NH4)2SO3: 0-25% or 14.4 gllOO mL (___g)

monitor pH 7.3-7.4

volumes are per one preparation (1 beef heart)

lOl

1111111111111111111111111111111111111111
3 1293 02470 0126