.. A, o
L ”Jun «3..
«in. a

\fv

.. . , N v
.Asihz.‘
53..

.3
( ‘5 :5

5.3L: 4. Aﬂlv

.3 tiazb. .v .

Li a.

 

 

 

 

 

.
. 0.. .I .-
, a.“ C.
on; .
L if
.1. .a.
...
t

 

 

 

ill-pg! 11 ‘
3.1;: ,
EZ...5PJ

l ﬁn.

7..sl.rl!#¢£mh

$30493:

 

. ..
:3. in
3?: ¢ 3 .3)

. ~.r§3uuﬁuai.lun

94.0.? 3.32
uh he) éve‘r-

is}.

, k 4 .
. l .a i}? . .
var ‘S.-D.'¢levx. . , X, . '
{dun 3.2.1. Hanummn a. g-
115...; .

Sign ‘
. l 53‘...

. u 1.!)
3”. .uxl...‘

 

ﬁaééagagggmgﬁﬁﬁ _

.v. vﬁu. 0.1

 

LIBRARY
Michigan State
University

This is to certify that the
dissertation entitled

APPLICATIONS AND COMPUTATIONAL
CONSIDERATIONS OF THE CYANYLATION-BASED
DISULFIDE MAPPING METHODOLOGY FOR CYSTINYL
PROTEINS

presented by
Wei Wu

has been accepted towards fulﬁllment
of the requirements for the

Ph.D degree In Chemistry

9%

ﬂ Major Professor’s Signature

[0 WM

Date

MSU is an Afﬁrmative Action/Equal Opportunity Institution

-.— A-c-n-I-C-l-I-C-I-

- .— 4-h-

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:/CIRC/DateDue.965-p. 15

APPLICATIONS AND COMPUTATIONAL CONSIDERATIONS OF THE
CYANYLATION-BASED DISULFIDE MAPPING METHODOLOGY FOR
CYSTINYL PROTEINS
By

Wei Wu

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

Department of Chemistry

2003

ABSTRACT
APPLICATIONS AND COMPUTATIONAL CONSIDERATIONS OF THE
CYANYLATION-BASED DISULFIDE MAPPING METHODOLOGY FOR
CYSTINYL PROTEINS
By

Wei Wu

Disulﬁde bond conformation is very important to maintain the stability and
function of a cystinyl protein. Cyanylation (CN)-based disulﬁde mapping methodology is
very robust and versatile to characterize disulﬁde structures of cystinyl proteins. The
acidic condition for partial reduction and cyanylation efﬁciently thwarts the thio-disulﬁde
exchange reaction that generates disulﬁde artifacts. Cleavage of the amide bond at the N-
terminal side of a cysteine makes it possible to analyze disulﬁde structures with adjacent
cysteines.

Two large proteins (pea chloroplast fructose bisphosphatase, F BPase, 39 kDa and
cobalamin-independent methionine synthase, MetE from E. coli, 84 kDa) were subjected
to the CN-based disulﬁde mapping. For FBPase, two oxidized structures were identiﬁed
by mass Spectrometry. For MetE, the mapping experiment showed that the reduced form
of the enzyme contains seven free cysteines, and that the oxidized form of the enzyme
has a glutathione adduct at Cys645.

A quantiﬁcation method using homologous alkylating reagents was proposed and
evaluated. The problem of over-alkylation was minimized by adjusting the reagent-to-

peptide ratio and reaction time. Concentration ratio of cysteine-containing peptides

between 1:5 and 5:1 can be reliably represented by the peak intensity ratios of two
differently alkylated peptides.

A computational analysis was performed to evaluate the CN-based disulﬁde
mapping methodology with generic cystinyl proteins. Signature sets, minimal subsets of a
fragment set to uniquely identify a disulﬁde structure, were generated from fragment sets
and differential sets. The conciseness and multiplicity of Signature sets ensure the
efﬁciency of the method even when the majority of the expected CN-induced fragments
were not detected. Analysis of the fragment sets showed that partial reduction should be
directed to maximize the yield of both singly and doubly reduced isoforms. Experimental
data from mapping experiments of ribonuclease A (RNase A), human epidermal growth
factor (hEGF) and its two disulﬁde isomers were used to validate the concept of

‘signature sets’.

ACKNOWLEDGEMENTS

I want to express my sincere gratitude to my mentor, Dr. J. Throck Watson, for
his guidance and support throughout my graduate career at Michigan State University. I
would like to thank Dr. John Allison for his constructive suggestions for this dissertation,
Drs. Doug Gage and James Jackson for serving on my research committee.

I want to thank my parents for their unconditional love and lifetime education, my
brother and sister-in-law for setting up two exceptional models ahead for me.

I want to thank my ﬁance, Wei Huang, for her love and support.

Thanks to Watson group members, J ianfeng, Chad, Yingda, Jose-Luis, Shelly, Yi-
Te, Robin and Charles, to mass spec facility personnel, Susi, Rhonda and Bev, for your

friendship and collaboration.

iv

TABLE OF CONTENTS

LIST OF TABLES ................................................................................... viii
LIST OF FIGURES .................................................................................... x
CHAPTER 1
INTRODUCTION OF DISULFIDE MASS MAPPING ......................................... 1
1. Biological Importance of Disulﬁde Bonds ................................................ l
2. Disulﬁde Mapping Approaches ............................................................. 3
2.1. Enzymatic Digestion-based Disulﬁde Mass Mapping ............................ 3
2.2. Partial Reduction/Alkylation-based Methods ...................................... 5
2.3. Cyanylation (CN)-induced Cleavage-based Disulﬁde Mass Mapping ......... 6
3. Computer—assisted Analysis of Disulﬁde Bonds ....................................... 10
3.1. Homology-based Approaches ...................................................... 10
3.2. Mass Spectrometry (MS)-based Data Analysis ................................... ll
6. Summary ...................................................................................... 11
Reference ......................................................................................... 13
CHAPTER 2

CYANYLATION-BASED DISULFIDE MAPPING OF TWO LARGE PROTEINS:
CHLOROPLAST FRUCTOSE BISPHOSPHATASE AND COBALAMINE-

INDEPENDENT METHIONINE SYNTHASE ................................................. 15
l. Chloroplast Fructose Bisphosphatase (FBPase) .......................................... 15

1 .1 . Introduction ............................................................................. l 5

1.2. Materials and methods ................................................................ 17

1.3. Results and Discussion ................................................................ 23

2. Cobalamin-independent methionine synthase (MetE) ................................. 39

2.1 . Introduction ............................................................................ 39
2.2. Materials and Methods ................................................................ 41
2.3. Results and Discussion ............................................................... 44
3. Conclusion .................................................................................. 54
References ....................................................................................... 55
CHAPTER 3
QUANTIFICATION OF CYSTEINE-CONTAINING PEPTIDES WITH
HOMOLOGOUS ALKYLATIN G REAGENTS ................................................ 57
1. Introduction ................................................................................... 57
2. Materials and Methods ...................................................................... 61
3. Results and Discussion ...................................................................... 63
3.1. Optimization of the Alkylation Reaction Conditions ............................ 63
3.2. Quantiﬁcation Analysis of Cysteine-Containing Peptides ...................... 70
3.3. Estimation of the Minimal Error and Standard Deviation ....................... 76
3.4. Using Homologous Reagent to Quantify Disulﬁde Isomers .................... 8O
4. Conclusion ..................................................................................... 82
References ....................................................................................... 83
CHAPTER 4

‘SIGNATURE SETS (Si)’, A MINIMAL POSITIVE SIGNATURE FEATURE FOR
IDENTIFYING PROTEIN DISULFIDE STRUCTURES WITH CYANYLATION (CN)-

BASED MASS MAPPING METHODS ........................................................ 84
1. Introduction .................................................................................. 84

2. Development of the Concept of Signature Sets ........................................ 85
2.1. Disulﬁde Structure and Fragment Set ............................................. 85

vi

2.2. Fragments from CN-induced Cleavage of Singly and Doubly Reduced

Isoforms Represent the Entire Fragment Set ..................................... 89
2.3. Differential Sets (D(i, j)) for a Disulﬁde Structure (F i) ......................... 91
2.4. Generation of Signature Sets for a Particular Disulﬁde Structure ............. 94

3. Evaluation of the Signature Sets Concept with Disulﬁde Mass Mapping Data
............................................................................................... 95

3.1. Output of Signature Sets for Generic Proteins with n Disulﬁde Bonds ...... 95
3.2. Human Epidermal Growth Factor (hEGF) and its Disulﬁde Isomers. . . . .....97

3.3. Further Evaluation of the Signature Sets with Ribonuclease A (RNase A)
Mapping Data ........................................................................ 103

3.4. Characteristics of Signature Sets .................................................. 105

4. Frequency Analysis of the Fragments from CN-induced Cleavage of a Four-

Disulﬁde Protein .......................................................................... 106

5. Homogeneity in Disulﬁde Structures of the Protein under Study .................. 109
6. Conclusions ............................................................................... 111
Reference ....................................................................................... 112

APPENDIX

DISULFIDE INDEX, FRAGMENT SETS, SIGNATURE SETS FOR A GENERIC 3-

DISULFIDE BOND PROTEIN ................................................................ 113
1. Disulﬁde Structure Index ................................................................ 113
2. Fragment Sets ............................................................................. 114
3. Signature Sets .............................................................................. 116

vii

Table 2.

.2.

LIST OF TABLES

. Assignment of peaks in the MALDI mass Spectrum after CN-induced

cleavage of the pea chloroplast fructose bisphosphatase ........................ 24

Assignment of peaks in the MALDI mass spectrum after CN-induced
cleavage of the recombinant spinach chloroplast fructose bisphosphatase

........................................................................................................ 28

Table 2.

Table 2.

Table 3.

Table 3.

Table 4.

Table 4.

Table 4.

.3.

Summary of energy values of the oxidized CyslSS-174, Cysl 74-179 and
Cy5155-179 forms of the Spinach chloroplast fructose bisphosphatase, and of
the reduced form, following energy minimization. ........................... 34

. Molecular weight calculation based on the 22 peaks marked between the two

arrows in the spectra of the oxidized MetE and the reduced MetE. Average
mass and its standard deviation were calculated from the 21 independent
calculations from two peaks that correspond to the enzyme with n protons and
n+1 protons. (n from 76 to 95). ..................................................... 47

. Peak assignment of the fractions collected from HPLC separation of the

reduced MetE protein. .............................................................. 51

. Precision and accuracy of the assay of peptide CQDSETRTFY, over a

dynamic range between 1:10 and 10:1. (5 measurements). .................... 74

. Precision and accuracy of the experimental ratio of the intensity of the

monoisotopic peak vs. that of the peaks with 1~4 (13C) atoms in the 2-BA
modiﬁed peptide. C0 represents the monoisotopic peak while C1~4 represents
the peptide with 1~4 13C atoms. The expected ratio was calculated from the
elemental composition of the product. (10 measurements). .................... 78

. Number of disulﬁde structures and number of CN-induced cleavage fragments

as a function of number of disulﬁde bonds. ...................................... 86

. Indexing of all 15 possible isomeric disulﬁde structures for a 3-disulﬁde

protein containing 6 cysteines. ................................................... 96

. Expected fragments from CN-induced mass mapping of native hEGF and

those detected by MALDI-MS. The ﬁrst ﬁve fragments are constituents of
four signature sets (see Table 4. 4), while the other nine theoretically possible
fragments do not contribute in distinguishing the disulﬁde structure of the
native protein from other theoretically possible disulﬁde structures. (n.d. =not

detected). .............................................................................. 99

viii

Table 4. 4. Signature sets for the three isomeric proteins resulting from the refolding of
hEGF. The sets marked with ‘X’ had all constituent fragments represented by
peaks in previously recorded mass Spectra during disulﬁde mass mapping..102

Table 4. 5. Signature sets for RNase A; compositions for fragments are expressed in
computer—shorthand according to relative position of Cys residues as
explained in text. ..................................................................... 106

Table 4. 6. Number of occurrences of fragments from CN-induced cleavage of Singly and

doubly reduced isoforms of all 105 theoretically possible disulﬁde structures
for a protein containing four disulﬁde bonds. ................................... 108

ix

LIST OF FIGURES

Figure 1.1. Chromatogram of 2nd dimension RP-HPLC run of one of the fractions
collected from the lSt dimension cation exchange separation of the digest of
the protein complex of pregnancy-associated plasma protein-A and the
proform of eosinophil major basic protein. Fraction P31 contains a pair of
peptides linked through a disulﬁde bond. ............................................ 4

Figure 2. 1. Structure of Spinach chloroplast fructose bisphosphatase in protein data bank
(PDB) ﬁle lspi. Ring D is highlighted. The a-carbons of the Cys residues in
the insertion into the chloroplast enzyme are represented as spheres.
Clockwise from left to right they are Cysl79, Cysl74 and Cy8155. In the
crystal structure of the pea chloroplast enzyme in PDB ﬁle 1d9q Cysl 53 and
Cysl73 (correspond to 155 and 174 in PDB ﬁle lspi) are involved in a
disulﬁde bond. The ﬁgure was constructed using MOLSCRIPT.
............................................................................................ 16

Figure 2. 2. MALDI mass spectrum of CN-induced cleavage products of pea chloroplast
fructose bisphosphatase, divided into three traces. Peaks were automatically
marked at the centroid by the Data Explorer software. The mass spectrum is
divided into three mass regions because the peaks in mass region two (middle
panel) are signiﬁcantly more intense than those in regions one and three. The
experimentally determined m/z values together with calculated masses for
expected protonated CN-induced cleavage fragments appear in Table 2.1.
........................................................................................... 25

Figure 2. 3. MALDI mass Spectrum of CN-induced cleavage products of recombinant
spinach fructose bisphosphatase. The mass spectrum is divided into two mass
ranges due to the intensity differences. The assignment of the labeled peaks is
Shown in Table 2.2. .................................................................... 29

Figure 2. 4. Mass chromatograms corresponding to the analysis of the CN-induced
cleavage products of the recombinant spinach fructose bisphosphatase by LC-
ESI-MS. The top panel is a chromatogram reconstructed from the entire mass
range (m/z 200-2000). The second panel is a mass chromatogram of ion
current at m/z 1306 corresponding to the triply charged itz-155-190 fragment
containing a disulﬁde bond between Cysl 74 and 179. The third panel is a
mass chromatogram at m/z 1959 corresponding to the doubly protonated CN-
induced cleavage fragment itz-155-190. .......................................... 32

Figure 2. 5. An averaged spectrum (of 40 spectra) acquired during the analysis of the CN-
induced cleavage products of the recombinant spinach fructose bisphosphatase
by LC-ESI-MS. The two most intense peaks in the mass spectrum correspond

to the doubly and triply charged itz-155-190 fragment, respectively, with a
disulﬁde bond between Cysl 74 and Cysl79. .................................... 33

Figure 2. 6. Modeled structures of the loop containing the redox-sensitive cysteine
residues in the reduced form of spinach chloroplast fructose bisphosphatase,
in the two oxidized forms observed in these experiments (‘b’ and ‘c’), and in
the oxidized form containing the 155-179 disulﬁde bond (‘d’) that apparently
can occur when Cysl 74 is replaced by mutation. In the oxidized form of the
pea leaf enzyme in PDB ﬁle 1d9q [8], the loop has contracted into a short
helix. ..................................................................................... 35

Figure 2. 7. ESI-MS Spectra of the oxidized and reduced MetE. The clusters of peaks

marked between two arrows are used to calculate the molecular weight (result
shown in Table 2.4). ................................................................. 46

Figure 2. 8. Microbore HPLC separation of CN-induced cleavage fragments from
cyanylated reduced (upper panel) and oxidized (lower panel) MetE. The
fraction marked with an asterisk was identiﬁed as containing two variants of a
glutathionated peptide and was subjected to total reduction and further mass
spectral analysis. ...................................................................... 50

Figure 2. 9. Direct MALDI-MS analysis of one of the chromatographic fractions
(marked with * in the lower panel of Figure 2.7) consisting of mixture of CN-
induced cleavage fragments from the CN-induced cleavage of oxidized MetE
(upper panel). MALDI analysis of the same fraction after treatment with an
excess of TCEP (lower panel). ..................................................... 52

Figure 3. 1. Structures of ICAT reagent pair used for protein quantiﬁcation. .............. 58

Figure 3. 2. The two reagents 2-BA and 2-BP modify the cysteine residues in a
peptide/protein. The reagents have a CH2 as the net difference in their
composition, as reﬂected by a mass difference of 14 Da in their differentially
modiﬁed products. .................................................................... 60

Figure 3. 3. MALDI-TOF mass spectra of the 2-BA/2-BP modiﬁcation reaction (peptide
HCKFWW) mixture. Before optimization (upper panel), up to ﬁve
modiﬁcations by 2-BA were observed for the peptide. There was no obvious
over-alkylation from the 2-BP modiﬁcation under that condition. After
optimization (lower panel), the desired products were the predominant Species
in the spectrum. No over-alkylation was observed. The insert is a ‘zoom-in’ of
the m/z range around the peaks for the desired products. '3 C isotopic peaks
were well resolved. .................................................................... 66

Figure 3. 4. Time course of the two modiﬁcation reactions of peptide HCKFWW. For 2-
BP modiﬁcation (upper panel), the desired singly modiﬁed peptide yield
increases relatively more Slowly. After 30 minutes, the yield is over 95%. No
over-alkylation was observed. For 2-BA modiﬁcation, the desired singly

xi

modiﬁed product yield reached 95% within 5 minutes. The over-alkylation
becomes Signiﬁcant ................................................................... 68

Figure 3. 5. Mass spectrum of 2-BA and 2-BP modiﬁed peptides (PHCKRM and
DRVYIHPCHLLYYS) at two different states (concentration ratio 1:1). ..... 72

Figure 3. 6. MALDI mass Spectra from quantiﬁcation analysis of a series of standard
solutions containing various ratios (10:1, 5:1, 5:2, 5:3, 1:1, 3:5, 2:5, 1:5 and
1:10) of preformed alkylation products (2-BA and 2-BP) of peptide
CQDSETRTF Y. Each Spectrum was averaged over 100 laser shots. The
mono-isotopic peaks for the two modiﬁed products have masses of 1306.4 Da
(2-BA) and 1320.4 Da (2-BP). ..................................................... 73

Figure 3. 7. Standard curve of quantiﬁcation of peptide CQDSETRTF Y, over a dynamic
range between 1:5 and 5:1 (5 measurements). ................................... 75

Figure 3. 8. Standard curve for quantiﬁcation of peptide HCKFWW and peptide
DRVYIHPCHLLYYS, over a dynamic range between 1:5 and 5:1. The ratio
is the intensity of the peak for the 2-BA modiﬁed product over that for the 2-
BP modiﬁed product. ............................................................... 76

Figure 3. 9. Isotopic distribution variance of the product peaks after 2-BA and 2-BP
modiﬁcation of peptide DRVYIHPCHLLYYS in three experiments (5:3, 5:2
and 5:1). The insert is a computer-generated distribution representing the
isotopic variants for the elemental composition of the 2-BA modiﬁed product
(from http://prospector.ucsf.edu). Peak (a) shows very similar distribution
pattern. The three isotopic clusters labeled with (b3) Show the intensity of the
monoisotopic peak higher than that of the mono-1 C peak. The isotopic cluster
(c) shows a bigger deviation from the expected distribution, especially for the
bis-13C peak. All spectra were the average of the accumulation of 100 shots.
........................................................................................... 83

Figure 4. 1. HPLC chromatogram showing three fully oxidized isomers of hEGF after 48
hours of oxidative refolding. ‘Native’ represents the native hEGF, while ‘III-

A’ and ‘III-B’ represent the misfolded versions of the protein. ............... 102

Figure 4. 2. The number of occurrences for a CN-induced cleavage fragment as a
function of its constituent free cysteines for a four-disulﬁde protein. l 09

xii

CHAPTER 1

INTRODUCTION OF DISULFIDE MASS MAPPING

1. Biological Importance of Disulfide Bonds

Disulﬁde bonds stabilize the 3-dimensional conformation of proteins, thus helping
to maintain their biological function. Each one of the disulﬁde bonds may contribute as
much as 5~6 kcal/mol to the stability of a folded protein at optimal temperature (1,2). In
protein engineering, efforts have been made to introduce one or several novel disulﬁde
bonds to make the target protein more stable (3-5). Some successfully increase the
stability of the target protein after inserting novel, non-native disulﬁde bonds, while
others have the reverse effect. Nonetheless, the characterization of disulﬁde bond
structures, in both the native and the mutant, is a very important part of these studies.

Disulﬁde bonds are implicated in ‘mad cow disease’, a transmissible spongiforrn
encephalopathetic disease (6). The disease, unlike other diseases that are induced by
bacterial or viral infections, is triggered by a conformational change of a key protein (the
prion protein, PrP) from its normal cellular isoform (PrPC) to an abnormal scrappie
isoform (PrPSC). Recently, there is evidence Showing that intermolecular disulﬁde bond
formation may play a role in this conformational change. Disulﬁde formation in many
other proteins is found as a key response to oxidative stress (7-9). In these proteins, the
formation and dissociation of disulﬁde bonds is a ‘redox switch’ that regulates the protein
activity.

In eukaryotic organisms, the secretory and membrane proteins, often objects of

protein folding studies, are synthesized in the rough endoplasmic reticulum (ER),

followed by translocation into the cistemal Space of the ER where the signal peptide is
cleaved and other post-translationally modiﬁed, including disulﬁde bond formation and
glycosylation. These assembled proteins are then translocated to the Golgi apparatus for
further modiﬁcation, and then either translocated to the cell surface or packaged into
secretory vesicles for secretion (10).

It is a common practice in protein engineering to insert the DNA that encodes
therapeutically important proteins into the prokaryotic host’s DNA and use the host’s
transcription and translation machineries to express the proteins (known as the
recombinant technology or ‘cloning’). Bacteria, such as. E. coli, are usually used as the
host expression system due to their fast production rate and convenient handling. Due to
the lack of cytoplasmic organelles in prokaryotic organisms that facilitate protein folding,
the desired over-expression of proteins often leads to the formation of insoluble
aggregates, oﬁen referred to as inclusion bodies, in the cytoplasmic or periplasmic
spaces. TO solubilize these inclusion bodies and to restore the biological function of the
protein, a refolding process is required. The inclusion body is usually ﬁrst dissolved in
buffer with a high concentration of the denaturant. Then, a step-wise dialysis lowers the
concentrations of denaturant and eventually solubilizes the protein under physiological
pH and ionic strength. A critical oxidative refolding is followed to establish the disulﬁde
structure by exposing the protein to a redox buffer, typically containing oxidized
glutathione and reduced glutathione with a concentration ratio between (1:3 to 1:1), the
ratio present in the eukaryotic endoplasmic reticulum to hold ER slightly more oxidizing
than the rest of the cytoplasm (11). A slightly basic pH (around 8) in the redox buffer

promotes the formation of thiolate anion and assists the thio-disulﬁde exchange reaction.

Addition of molecular chaperones, e.g., heat-shock proteins and protein disulﬁde
isomerase (PDI) may assist the refolding process (12). Isomers with non-native disulﬁde
structures are often present along with the protein with the native disulﬁde structure after
the folding process. It is important to separate them and determine the disulﬁde
structures of the folding products so that only the one with the correct disulﬁde structure

is subjected to further puriﬁcation and therapeutic purposes.

2. Disulfide Mapping Approaches

2.1. Enzymatic Digestion-based Disulfide Mass Mapping

Mass Spectrometry (MS)—based disulﬁde mapping is based on controlled
degradation of the target cystinyl protein, followed by the characterization of the
degradation products by mass spectrometry. Disulﬁde connectivity, retained throughout
the proteolysis, is explored by analyzing the digestion products, particularly those
containing disulﬁde bonds. Chemical degradations, such as cyanogen bromide digest,
Speciﬁc to a methionine residue, are often employed. Because the sequence of the
protein is known, and because the cleavage sites of the controlled degradations are
usually speciﬁc, the masses of the fragments, including those containing disulﬁde bonds,
can be calculated prior to the mapping experiment.

A typical example of this enzymatically based mass mapping can be found in the
recent publication by Overgaard et al. (13). They studied the disulﬁde structure of a
protein complex of pregnancy-associated plasma protein-A and the proform of eosinophil

major basic protein (a 2:2 complex with a total M.W. of 500 kDa). A combination of

cyanogen bromide digestion and trypsin digestion was utilized to fragment the protein
complex. Due to the unusually large size of this protein complex, the peptide digest
mixture is separated with a 2-dimensional liquid chromatography system (cation
exchange + reverse phase) due to its complexity. In Figure 1.1, fraction P31 off the 2"d
dimensional reverse phase high performance liquid chromatography (RP-HPLC) was
identiﬁed as a disulﬁde-containing peptide by matrix-assisted laser desorption/ionization
time of ﬂight mass Spectrometry (MALDI-TOF-MS). By identifying the disulﬁde-
containing peptides in the remaining HPLC fractions, the disulﬁde structure of the

protein complex was resolved.

 

 

L o 5 P31
E ' QGQCSVPNELNSNLK
c |
o / WCTAGLK
5}! 0.3
EU
8 »O.1 1
g JLALLi I L—___E
'0 WV

 

0 5 1O 15 20 25 30
time (min)

Figure 1.1. Chromatogram of 2"d dimension RP-HPLC run of one of the fractions
collected from the 1St dimension cation exchange separation of the digest of the protein
complex of pregnancy-associated plasma protein-A and the proform of eosinophil major
basic protein. Fraction P31 contains a pair of peptides linked through a disulﬁde bond.

This protease-based disulﬁde mapping method has several potential drawbacks.
First, the disulﬁde/cysteine containing peptides are only a very small fraction of the entire

protein digest, due to the fact that cysteine has a natural occurrence of 2% among all the

20 natural amino acids. Separating and identifying these small fractions of peptides of
interest poses a great challenge to analysts. Secondly, most of the Speciﬁc proteases,
such as trypsin, Lys-C, Glu-C, etc., function at a Slightly alkaline pH, under which the
thiolate anion (RS'), a strong nucleophile, is promoted from deprotonation of a sulthydryl
group and may attack a neighboring native disulﬁde bond to form an artiﬁcial disulﬁde
isomer. Control of pH and time of digestion is often critical to retain the disulﬁde
structure of the protein in the fragments. Lastly, in order to get a complete disulﬁde map,
this method requires a sufﬁcient number of cleavage sites distributed between cysteines.
However, as two cysteines become closer to each other in the amino acid sequence, the
chance of ﬁnding an available enzymatic cleavage Site reduces accordingly. In the case
of adjacent cysteines, cleavage between them by a protease becomes impossible, as no
known proteases cleave at a cysteine residue. Protease-based disulﬁde mass mapping

methods generate inconclusive disulﬁde structures in such cases.

2.2. Partial Reduction/AIkylation-based Methods

This method was proposed by Gray to study disulﬁde structure of small, highly
bridged proteins (14). Those proteins are often resistant to proteolytic digestion due to
their highly knotted structure. The method involves step-wise partial reduction, with
different alkylating reagents capturing the reduced cysteine pairs at each step, and then
using a protein sequence analyzer to identify the differently alkylated cysteines. The
method works best for proteins that expose different disulﬁde bonds sequentially for
reduction and alkylation under a successively more severe reducing environment.

However, a protein’s behavior under reducing conditions is less controlled by the

concentration of the reducing agent than by the protein’s intrinsic property, i.e., its
sequence. Additional degradation and identiﬁcation steps of the differentially alkylated
protein are needed if the protein size exceeds the limit of a protein sequencer. Variation
of this method, in which mass Spectrometry (LC/MS/MS) was used to identify the

alkylation groups, was developed by Yen et al. (15).

2.3. Cyanylation (CN)-induced Cleavage-based Disulfide Mass Mapping

The method is based on selective cyanylation of the sulfhydryl group of a cysteine
residue, and subsequent nucleophilic attack by NH3 to promote cyanylation (CN)-induced
cleavage of the peptide backbone. This Speciﬁc chemical cleavage at the N-terminal side
of a cysteine residue involves two steps, as illustrated in Scheme 1.1. In the ﬁrst step, the
hypothetical peptide (a 20mer with a cysteine at residue 10) is cyanylated with the
cyanylating reagent, cyanodiaminopyridinium (CDAP) tetraﬂuoroborate. In the second
step, the peptide bond between residue 9 and cyanylated Cyle (in the middle of the
scheme) breaks upon alkaline incubation with ammonia, resulting in a truncated peptide

(1-9), and an iminothiazolidine (itz)-terminated peptide (itle-20).

 

 

 

 

 

1 1
C an lation CN-induced
y y Cleavage 9
10 -—-SH , 10 l_ SCN >
CDAP, NH3 itle
PH=3
20 20 20

Scheme 1.1. Cyanylation-induced cleavage of a hypothetical cysteine-containing peptide
generates two cleavage fragments.

Because cystines (disulﬁde bonds) are nonreactive with the cyanylating reagent
CDAP, a multi-cystinyl protein must be partially reduced (to generate speciﬁc pairs of
free sulﬂrydryl groups) using triS-carboxyethylphosphine (TCEP), as illustrated in
Scheme 1.2 for a hypothetical 2—disulﬁde protein. In this example, the four constituent
cysteines are located at residues 10, 20, 30 and 40. The four cysteines form two disulﬁde
bonds (Cyle with Cys30, and Cys20 with Cys40). When subjected to reduction, the
disulﬁde bond between Cy320 and Cys40 may be reduced to form a ‘singly reduced
isoform’, as illustrated on the right side of the equation in Scheme 1.2. The other singly
reduced isoform, in which the disulﬁde between Cyle and Cys30 is reduced, may be
formed at the same time, as well as the doubly/totally reduced isoform, in which both
disulﬁde bonds are reduced (structures not shown). The partially reduced isoforms are
immediately exposed to the cyanylyating reagent, CDAP, to covalently modify all the
free sulfhydryl groups for the subsequent cleavage reaction. The low pH during both
partial reduction and cyanylation suppresses the thiOl-disulﬁde exchange reaction that

could lead to disulﬁde artifacts. Reduction of one disulﬁde bond generates two nascent

sulfhydryl groups (SH) and causes a +2-Da shift; cyanylation of the cysteine pair results
in a +50-Da mass Shift ( [SH —> SCN] x2, [26-1]x2=50 Da ). Thus, the net mass shift in
going from an oxidized sulfur (as in a disulﬁde bond) is 26 Da (25+1) per sulfur involved
in the chemical modiﬁcation. The degree of partial reduction can be determined with a
mass measurement of the partially reduced and cyanylated isoform, e.g., a doubly
reduced/cyanylated isoform would have a +104-Da mass shift from that of the fully

oxidized protein.

 

 

 

 

 

 

 

 

Partial Reduction SH SH
TCEP l
l l ————. l L
1 10 20 30 4o 50 1 10 20 30 4o 50

Scheme 1.2. Generation of one of the two possible singly reduced isoforms of a
hypothetical 2-disulﬁde protein.

Once cyanylated and subjected to nucleophilic attack in 1M NH40H, the
polypeptide backbone of the singly reduced and cyanylated isoform of the hypothetical
protein Shown in Scheme 1.3 is cleaved at the N-terminal side of the two cyanylated
cysteine residues. However, as Shown in the middle of Scheme 1.3, a residual disulﬁde
bond, in this example, holds two of the three cleavage products together. Thus, the
cleavage reaction mixture represented by the middle of Scheme 1.3 is treated with an
excess of reducing reagent (TCEP) to reduce the residual disulﬁde bond, thereby
ensuring that all CN-induced cleavage products, 1-19, it220-39, and itz40-50, are free as
shown at the end of Scheme 1.3. With the mass mapping of these three fragments and
the knowledge that they are from a singly reduced isoform, we can deduce that CysZO

and Cys40 are involved in a disulﬁde bond in the original protein.

Similarly, when subjected to cyanylation, cleavage and total reduction, the other
possible partially reduced isoform (not illustrated), in which the disulﬁde bond between
Cyle and Cys30 is reduced, generates three fragments, 1-9, itle-29 and itz30-50. By
detecting them through mass mapping, the other disulﬁde bond is conﬁrmed; of course,
in such a case involving two disulﬁde bonds in a protein with only four cysteines, a
determination of the connectivity of cysteines in one disulﬁde bond allows the

connectivity of the other two cysteines to be established by default.

 

 

 

 

 

SH SH
L . 1
SCN SCN 1 10 19 ”‘20 30 39
l l
10 20 30 4o 50 ,
rtz40 50

 

\ /'T

 

 

NH4OH, otal Reduction
1 hr, rt 1 10 19 TCEP
it22(% I 39
30
+
itz40 50

 

Scheme 1.3. Cleavage of a singly reduced and cyanylated isoform, aﬁer total reduction
(to break the residual disulﬁde bond), generates three fragments.

In summary, the original strategy for CN-based mass mapping(16) involves three
key steps: partial reduction/cyanylation, HPLC separation of the partially reduced
isoforms/mass measurement to determine the reduction status, and alkaline cleavage/total
reduction and mass spectral analysis. The overall disulﬁde structure is established by

determining the individual disulﬁde bonds with the detection of sets of three fragments

from the cleavage reaction of Singly-reduced isoforms. This method has been used
successfully to solve the disulﬁde structures of ribonuclease A (13 kDa) and sillucin (3

kDa with three adjacent cysteines) (16,17).

3. Computer-assisted Analysis of Disulfide Bonds
3.1. Homology-based Approaches

A study of the cystinyl proteins in the protein databank has shown that the
symmetric and reducible (disulﬁde structure has at least two non-overlapping domains)
disulﬁde structures occur at a much higher rate than just by chance (18). Attempts have
been made to predict the cysteine status (free or disulﬁde-bonded) based on the amino
acid sequence based on the assumption that the local secondary structure affects the
propensity of a cysteine to be free or oxidized. Sequence alignment using 7-17 residues
around the cysteine residues to all the sequences in the cysteine-containing protein
database suggests predictors of that cysteine being free or being oxidized. After tuning
the parameters with a set of protein sequences with known structures, the prediction
efﬁciency was reported to be as high as 81% (19). However, this program only predicts
whether a cysteine is free or connected, not how it is connected when it is in the form of a
cystine.

Another homology-based algorithm using integer linear programming (ILP)
predicts both the beta Sheet and the disulﬁde bridges (connectivity) successfully with
model proteins such as bovine pancreatic trypsin inhibitor (BPTI) (20). However, the

chance of success of these homology-based methods highly depends on the presence of a

10

highly homologous protein, the structure of which has already been solved either by

chemical means or by crystallography data, in the existing protein database.

3.2. Mass Spectrometry (MS)-based Data Analysis

Computer-assisted data analysis has been utilized to analyze the CN-based
disulﬁde mapping experimental data. We observed that the two terminal cysteines that
deﬁne a particular cleavage fragment could not have been involved in a disulﬁde bond
with the internal cysteines in that fragment. The m/z value of the ion representing the
CN-induced cleavage fragment containing internal free cysteines is referred to as a
‘negative signature mass (N SM)’ (21). A computer algorithm, negative signature mass
algorithm (NSMA), was developed to determine protein disulﬁde structures based on
‘ruling out’ certain disulﬁde linkages and reconstructing the disulﬁde structure with the
surviving disulﬁde linkages. NSMA has been used successfully to solve the disulﬁde
structures of a four—disulﬁde protein, ribonuclease A, and a six-disulﬁde protein,

transforming growth factor [3 type II receptor extracellular domain.

4. Summary

In chapter 2 of the dissertation, results from the cyanylation-based disulﬁde mass
mapping experiments of reduced and oxidized forms of two large proteins (chloroplast
fructose bisphosphatase, F BPase, 39 kDa; cobalamin-independent methionine synthase
MetE from E. coli, 85 kDa ) are presented and discussed. These applications represented
the ﬁrst attempts to use this method to analyze disulﬁde structures of proteins over 15

kDa. Both enzymes have a dynamic disulﬁde structure: the reduced form and the

11

oxidized form. Change in disulﬁde structure, as a response to the changes (light in
FBPase and redox potential in MetE) in the environment, directly regulates the activity of
the enzymes.

Chapter 3 explores a novel mass spectrometry-based quantiﬁcation method of
cysteinyl peptides. The method involves alkylation of a cysteine-containing peptide at
two different states with two homologous reagents. The relative quantity ratio of the same
cysteine-containing peptide at two different states is represented by area ratio of the mass
spectral peaks corresponding to the two differently alkylated peptides. The over-
alkylation problem was observed, and minimized by adjusting the reaction conditions
aﬁer a survey of the kinetics of the two alkylation reactions. This quantiﬁcation method
could be used to quantify the two isomeric structures identiﬁed in the oxidized forms of
F BPase.

Chapter 4 presents the theoretical aspects of the cyanylation(CN)-based mass
mapping method. The concept of ‘signature sets’ is proposed, followed by a careful
study of the fragment set associated with a disulﬁde structure. The conciseness of the
Signature sets proves the robustness of this methodology: The disulﬁde structure of a
fully-oxidized cystinyl protein can be determined by the CN-based mapping approach
with the detection of a very small fraction of all the possible cleavage fragments. The
‘signature sets’ concept was validated with the data from the disulﬁde mapping
experiments of human epidermal growth factor (hEGF) and its two disulﬁde iomers

(generated from oxidative refolding) and ribonucleaes A (RNase A).

Reference:

12

10.

ll.

12.

13.

14.

15.

16.

Betz, S. F. (1993) Protein Sci 2, 1551-1558
Darby, N., and Creighton, T. E. (1997) M01 Biotechnol 7, 57-77

Futami, J., Tada, H., Seno, M., Ishikarni, S., and Yamada, H. (2000) Journal of
Biochemistry 128, 245-250

Johnson, C. M., Oliveberg, M., Clarke, J., and Fersht, A. R. (1997) J Mol Biol
268, 198-208

Hinck, A. P., Truckses, D. M., and Markley, J. L. (1996) Biochemistry 35, 10328-
10338

Welker, E., Wedemeyer, W. J ., and Scheraga, H. A. (2001) Proc Natl Acad Sci U
S A 98, 4334-4336

Linke, K., and Jakob, U. (2003) Antioxidants & Redox Signaling 5, 425-434
Eaton, P., Jones, M. E., McGregor, E., Dunn, M. J ., Leeds, N., Byers, H. L.,
Leung, K. Y., Ward, M. A., Pratt, J. R., and Shattock, M. J. (2003) J Am Soc
Nephrol 14, 8290-8296

Harding, H. P., Zhang, Y. H, Zeng, H. Q., Novoa, 1., Lu, P. D., Calfon, M., Sadri,
N., Yun, C., Popko, B., Paules, R., Stojdl, D. F., Bell, J. C., Hettmann, T., Leiden,
J. M., and Ron, D. (2003) Mol Cell 11, 619-633

Ruddon, R. W., Sherman, S. A., and Bedows, E. (1996) Protein Sci 5, 1443-1452
Gilbert, H. F. (1997) J Biol Chem 272, 29399-29402

Thomas, J. G., Ayling, A., and Baneyx, F. (1997) Appl Biochem Biotechnol 66,
197-238

Overgaard, M. T., Sorensen, E. S., Stachowiak, D., Boldt, H. B., Kristensen, L.,
Sottrup-Jensen, L., and Oxvig, C. (2003) J Biol Chem 278, 2106-2117

Gray, W. R. (1993) Protein Sci 2, 1732-1748
Yen, T. Y., Yan, H, and Macher, B. A. (2002) J Mass Spectrom 37, 15-30

Wu, J ., and Watson, J. T. (1997) Protein Sci 6, 391-398

13

l7.

18.

19.

20.

21.

Qi, J. F., Wu, J., Somkuti, G. A., and Watson, J. T. (2001) Biochemistry 40, 4531-
4538
Benham, C. J ., and Jafri, M. S. (1993) Protein Sci 2, 41-54

Fariselli, P., Riccobelli, P., and Casadio, R. (1999) Proteins-Structure Function
and Genetics 36, 340-346

Klepeis, J. L., and F loudas, C. A. (2003) Journal of Computational Chemistry 24,
191-208

Qi, J. F ., Wu, W., Borges, C. R., Hang, D. H., Rupp, M., Tomg, E., and Watson,

J. T. (2003) Journal of the American Society for Mass Spectrometry 14, 1032-
1038

14

CHAPTER 2
CYANYLATION-BASED DISULFIDE MAPPING OF TWO LARGE PROTEINS:
CHLOROPLAST FRUCTOSE BISPHOSPHATASE AND COBALAMINE-

INDEPENDENT METHIONINE SYNTHASE

1. Chloroplast Fructose Bisphosphatase (FBPase)
1.1. Introduction

The chloroplast fructose bisphosphatase (Enzyme Classiﬁcation, EC. 3.1.3.11)
participates in the Calvin cycle (photosynthetic CO2 ﬁxation). The enzyme is reductively
activated in the light in vivo (1). Oxidative inactivation of fructose bisphosphatase,
glyceraldehyde-3-P dehydrogenase, sedoheptulose bisphosphatase, and
phosphoribulokinase prevents the futile operation of the Calvin cycle in the dark; light
activation reverses the inactivation and allows photosynthetic CO2 ﬁxation to proceed (2-
4). In chloroplast fructose bisphosphatases, there is an insertion that contains redox-
sensitive cysteines (Figure 2. 1) in the shape of a loop that protrudes from the surface of
the reduced form of the enzyme. Although Site-directed mutagenesis experiments
indicate that each of the three Cys residues in the loop can form disulﬁde bonds with the
other Cys residues in the loop (5-7), there are only two, Cy5153 and Cysl73, close
enough to be joined by a disulﬁde bond in the crystal structures of the oxidized pea
chloroplast enzyme (8). We reasoned that mass mapping Should allow us to determine
whether there are additional disulﬁde bonds in the oxidized form of the enzyme. The
results indicate that there is an alternate disulﬁde bond between Cysl73 and Cysl78 in

some of the molecules of the pea chloroplast enzyme, as isolated from pea leaves, and

15

between the corresponding cysteine residues in the recombinant spinach enzyme, as

isolated.

 

Figure 2. 1. Structure of spinach chloroplast fructose bisphosphatase in protein data bank
(PDB) ﬁle lspi (9). Ring D is highlighted. The a-carbons of the Cys residues in the
insertion into the chloroplast enzyme are represented as spheres. Clockwise from left to
right they are Cysl79, Cysl74 and CyslSS. (For more detail see Figure 2. 6) In the
crystal structure of the pea chloroplast enzyme in PDB ﬁle 1d9q (8) Cy5153 and Cysl73
(correspond to 155 and 174 in PDB ﬁle lspi) are involved in a disulﬁde bond. The ﬁgure
was constructed using MOLSCRIPT (10).

1.2. Materials and methods

Plant material

Pea (Pisum sativum L. var Little Marvel) plants were grown from seed in the
University of Illinois at Chicago greenhouse as described previously (11). Seeds were

purchased from Old’s Seed Company, Madison WI.

Enzyme isolation

All steps in the puriﬁcation of the pea leaf enzyme were performed at or below
pH 6.5, in order to avoid thiol-disulﬁde exchange and scrambling of disulﬁde bonds. The
enzyme is stabilized by phosphate, hence potassium phosphate was used as the buffer
throughout. All operations were carried out on ice or at 0 to 4°C. Shoots (500 to 700 g)
from 12 to 14-day pea plants were harvested, washed in deionized water, and frozen
overnight at -30°C. The following morning the frozen tissue was blended in twice its
weight of 50 mM, pH 5.5 potassium phosphate and 0.4-times its weight of
polyvinylpolypyrollidone, ﬁltered through a wire-mesh strainer, 4 layers of cheesecloth
and one layer of lutrasil (Lutravil Co., Durham NC) and centrifuged (13,200g, 20 min).
The green supernatant solution was slowly made 40% saturated with respect to
ammonium sulfate by addition of 0.231 g pulverized (NH4)2SO4 per mL, stirred for 25
min and centrifuged (13,200g, 30 min). The straw colored supernatant solution was

made 60% saturated with respect to ammonium sulfate (by addition of 0.125 g

17

(NI-102804 per mL), stirred for 15 min and centrifuged (5,860g, 20 min). The pelleted
protein was suspended in 50 mM, pH 6.5 potassium phosphate (the phosphate buffer) and
dialyzed overnight against 1 L of the buffer with one change of buffer, and frozen.
Freezing did not appear to affect the activity of the enzyme at this point. When the
isolation was continued, the retentate in the dialysis bag was thawed, centrifuged
(27,000g, 15 min), and the supernatant solution was applied to a 2.54 x 9 cm column of
F ast-ﬂo DEAE-cellulose in the phosphate buffer. The column was washed with 500 mL
of the buffer and then with 5 L of the buffer containing 0.05 M KCl. The enzyme was
eluted with a linear 0.05 to 0.4 M KCl gradient (200 mLs + 200 mLs) in the phosphate
buffer. An equal volume of 1 M pH 6.5 potassium phosphate buffer that was 40%
saturated with respect to ammonium sulfate was added to the combined active fractions
and the solution was applied to a 1 mL phenyl Sepharose CL-4B column in 0.5 M
potassium phosphate (pH 6.5) that was 20% saturated with respect to ammonium sulfate.
The column was washed with 50 mLs of that buffer and then the enzyme was eluted in
0.5 M pH 6.5 potassium phosphate. The active fractions were concentrated on Millipore
Centriprep and Centricon 30 ﬁlters and applied to a 1 x 22 cm Superose 12 10/20 column
in 50 mM pH 6.5 potassium phosphate buffer. Active fractions were collected and
concentrated on a Centricon 30 ﬁlter. Purity was assessed by SDS gel electrophoresis.
This method was not particularly efﬁcient in terms of enzyme recovery.
Reproducibility was a problem. The yield was never more than 8% of the activity in the
dialyzed ammonium sulfate fraction that was applied to the DEAE cellulose column.
There was considerable loss of activity at the phenyl-Sepharose step. The highest

Speciﬁc activity Obtained was about 150 units per one mg of protein. Many preparations

18

were not homogeneous and had to be discarded. This puriﬁcation procedure did allow
puriﬁcation of the enzyme at a pH unfavorable for thiol-disulﬁde exchange.

Prior to cyanylation, any extraneous proteins present in the enzyme preparation
were separated from the fructose bisphosphatase by reverse phase HPLC on a Waters
system (Waters 2695 module with a Waters 2487 dual wavelength detector). 100 pg of
protein in 25mM pH 6.5 potassium phosphate buffer was injected onto a Vydac C18
(catalog number 218TP54, 4.6 mm id, 250 mm length) column. The gradient was 2% to
90% solvent B in 50 min, with solvent A being 0.1% TFA (triﬂuoroacetic acid) in water
and solvent B being 0.1% TFA, 90% acetonitrile in water. The ﬂow rate was lmL/min.
The detector was set at 214 nm. Fractions were collected manually. The masses of the
fractions were determined by MALDI (matrix-assisted laser desorption ionization) mass
spectrometry. Fractions that contained proteins with masses corresponding to the mass of
fructose bisphosphatase were combined and taken to dryness by lyophilization.

The recombinant spinach fructose bisphosphatase protein was kindly provided by
Renate Scheibe, Universitéit Osnabrﬁck. It was produced and puriﬁed as described by
Reichert et al. (5). All operations were performed at pH values of 6.5 or lower. The

protein appeared to be homogeneous on SDS gel electrophoresis.

Activity assay, protein estimation

Activity of the pea leaf enzyme was routinely assayed at pH 8.8. At this pH, changes in

the redox status of the protein do not affect activity. Cuvettes contained 100 mM Tris-

HCl, 10 mM MgCl2, 10 mM fructose-1,6-bisphosphate, 1 mM NADP, and 0.5 units

19

glucose-6-phosphate dehydrogenase and 1.5 units glucose-6-phosphate isomerase per
mL. Activity was followed by change in absorbance at 340 nm on a Varian Cary 210
spectrophotometer at room temperature (23°C). Protein was assayed according to

Bradford (12).

C yanylation (CN) and CN-induced cleavage

For cyanylation of the pea chloroplast fructose bisphosphatase, 30 DL of 1-cyano-4-
dimethylamino-pyridinium (CDAP) tetraﬂuoroborate solution (150 mM CDAP, 8 M
urea, 100 mM sodium citrate at pH 3) was added to the combined lyophilized fructose
bisphosphatase fractions. After 20 min at room temperature, the cyanylated protein was
separated from the cyanylating reagents by reverse phase HPLC, as described above.
The volume of the protein-containing fraction from the HPLC column was reduced to 10
pl under vacuum and 10 DL of 6 M guanidine hydrochloride was added. Cleavage was
accomplished by adding 20 pL of 1 M NH40H and incubating the mixture at room
temperature for an hour. Excess ammonia was removed under vacuum. After addition of
40 DL of 30% (v/v) acetonitrile/0.1% TFA to the dried cleavage products, the samples
were analyzed by MALDI-TOF-MS.

The recombinant spinach enzyme (insoluble in water) was washed with 500 pL
distilled water to remove the residual sodium acetate from the lyophilization step, and
was collected by centrifugation (14,000g, 8 min). 50 pL of 50 mM CDAP, 6M
guanidine-HCI, 100 mM sodium citrate buffer (pH 3) was added to the pellet to dissolve

and cyanylate the protein. After 15 min, the reaction was terminated by separation of the

20

cyanylated protein from the reaction mixture on a Bio-Rad 6 spin column that had been
pre-equilibrated with 8 M urea. For the alkaline cleavage reaction, 50 pL of 4 M NI-I40H
was added to the cyanylated protein (50 pL, in 8 M urea) to make the ﬁnal NH40H
concentration 2 M. After incubation for 1 hour at room temperature, the cleavage
reaction was terminated by removing the ammonia by vacuum aspiration. The cleavage

products were dissolved in 40 pL of 0.1% aqueous TFA.

Mass spectrometry

MALDI mass spectra were obtained on a Voyager DE-STR mass Spectrometer (Applied
Biosystems, Foster City, CA) equipped with a 337-nm nitrogen laser. The accelerating
voltage in the ion source was set at 25 kV. A solution of saturated Ot-cyano-4-
hydroxycinnamic acid (Aldrich Chemical Co.) in 1:2 of acetonitrile/water was diluted
with isopropanol (1:4 v/v). 30pL of the diluted solution was applied to the ﬂat MALDI
probe using Cadene’s protocol (13) to form an ‘ultra-thin layer’. The matrix solution
consisted of a saturated solution of or-cyano-4-hydroxycinnamic acid (Aldrich Chemical
Co.) in 1:1 of acetonitrile/water. 1 pL of the cleaved peptide solution was diluted with 9
pl of matrix solution and 0.5 pL of the peptide/matrix mixture was deposited onto the
probe. Co-crystallization was usually observed in 1 min. Residual liquid above the co-
crystallized peptide/matrix solid was removed by vacuum aspiration and 10 pL of water
was deposited on the crystals; after 15 seconds, the water was removed by vacuum
aspiration. This washing process was repeated two more times in order to remove the

majority of the salt contaminant, the guanidine hydrochloride (pea enzyme) or urea

21

(spinach enzyme), which, at a high concentration, suppresses the protein signals in
MALDI. External calibration was used with both the intact enzymes and their CN-
induced cleavage products.

For the ESI-LC/MS experiment with the recombinant spinach protein, the
cleavage peptide mixture was mixed 1:1 with 10% acetic acid (vzv). 6 pL of sample was
loaded onto a peptide capillary trap (Michrom 004/25108/32) through the auto sampler of
the CapLC system (Waters Co., Milford, MA). An auxiliary pump was used to wash the
trap with 1% formic acid in water at a ﬂow rate of 20 pL/min for 5 min. At the end of
the ﬁrst 5—min wash, an automated valve on the CapLC system switched allowing the
gradient mixture of solvent A and B (Solvent A: 1% formic acid in water; Solvent B: 1%
formic acid in acetonitrile) to ﬂow through the trap and onto the nano-spray column
(PicoFrit column from NewObjective, PFC7515-AQ-5, 15-pm tip i.d., 75-pm column
id, 5 cm of C18 packing before the frit at the tip). The gradient started at 5% of B at 5
min and went to 70% in 85 min, then it ramped up to 90% in 5 min and remained at 90%
for 20 min. The ﬂow rate from the PicoFrit column tip was 0.1 pL/min. The ﬂow from
the pump was 9 pL/min, allowing the gradient to be ramped up or down smoothly. A T-
Splitter between the CapLC pumps for solvent A/B mixture and the capillary trap
interfaced the two ﬂow rates. The ﬂow from the PicoFrit column tip was directly Sprayed
into a Finnigan LCQ-Deca mass spectrometer (Thermo Finnigan, San Jose, CA). The
source voltage was set at 2.7 kV and the heated capillary temperature was set at 200°C.
The quadruple ion trap mass spectrometer scanned from m/z 200 to m/z 2000 at a rate of
1.5 sec/scan. No data-dependent scans were used. The tune method was created by

infusion of l pmol/pL of angiotensin peptide at the same ﬂow rate (0.1 pL/min).

22

Modeling

In order to evaluate the suitability of Sites identiﬁed for the formation of disulﬁde
bonds, tertiary structure diagrams were made and displayed with Accelrys (Biosym
Technologies, San Diego, CA) on a Silicon Graphics work station. Subsequent energy
minimizations using the Discover Molecular Dynamics Package (Biosym Technologies)
were performed as previously described (14). Brieﬂy, minimization was initiated with all
heavy atoms subject to a tethering force of 2000 kcal/Az. The tethering force was
gradually reduced during steepest descents minimization, which was followed by
subsequent rounds of complex gradient minimization. The distance-dependent dielectric

constant directly proportional to atom separation was used throughout.
1.3. Results and Discussion
Mass Mapping

The ﬁnal reverse phase HPLC step in the puriﬁcation of the pea chloroplast
enzyme yielded a broad peak at the retention time corresponding to fructose
bisphosphatase protein. When this component was analyzed by MALDI-TOF-MS, ﬁve

peaks were detected: mass Spectral (MS) peaks at m/z 39059.0, 19529.8, 12999.8, 9735.4

and 7791.9, corresponding to the intact enzyme carrying 1 to 5 protons. After

23

cyanylation and HPLC separation, the major protein peak consisted of cyanylated

fructose bisphosphatase, as determined by MALDI-MS analysis.

 

 

 

Peak Observed mass Calculated mass

Identity Deviation

(Da) (Da)
1 doubly protonated itz-306-357 2889.33 2891.265 -0.067%
2 itz-153-189 with a disulﬁde
3953.98 3955.30 -0.033%

bond between 173-178
3 itz-49-91 4657.16 4658.19 -0.022%
4 2—48 5183.22 5182.98 0.005%
5 itz-306-357 5781.51 5781.53 0.000%
6 itz-92-1 52 6426.83 6426. 90 -0.001 %
7 itz-92-177 with a disulﬂde bond

9164.86 9162.85 0.022%

between 1 53-173
8 itz-92-189 with a disulﬁde bond

and a B-elimination among 10303.85 10304.20 -0.003%

153,173,178
9 112-306-357 dimer 11558.61 11562.02 -0.029%
10 itz-190-305 13290.20 13289.27 0.007%
Table 2. 1. Assignment of peaks in the MALDI mass Spectrum (Figure 2. 2) after CN-

induced cleavage of the pea chloroplast fructose bisphosphatase.

24

5°00 1 1 2889.33

1
4000 ,.

3000 l

Intensity

2000 3

2 3953.98

1000

    

0

2500 2700 2900 3100 3300 3500 3700 3900 4100 4300 4500

m I:

    

30000
25000
20000
1 5000
10000

5000

5 5781.51

lntenslty

5183.22

4500 4700 4900 5100 5300 5500

5700 5900

ml:

7 9164.86

 
 
       
     

8 10303.85
6 6428.83

Intensity
8
O
O

9 11558.61

10 13290.20

6000 7000 8000 9000 10000 1 1000 12000 13000 14000

mi:

Figure 2. 2. MALDI mass spectrum of CN-induced cleavage products of pea chloroplast
fructose bisphosphatase, divided into three traces. Peaks were automatically marked at
the centroid by the Data Explorer software. The mass spectrum is divided into three mass
regions because the peaks in mass region two (middle panel) are signiﬁcantly more
intense than those in regions one and three. The experimentally determined m/z values

together with calculated masses for expected protonated CN-induced cleavage fragments
appear in Table 2. 1.

25

Structure A

1—43 r1249—9r
SCN SCN SCN ISCN ISCN Cleavage, NH3
—4I9-si2 —153"L73—]78 -190-306- —-—-* “392—152 ”2190—305 moo—357
itzlS3—IZ3_L78—l89 ‘-

(signature fragment)

Structure 8

1—48 it249—91
lSCN ISCN ISCN ISCN ISCN Cleavage, NH3

—__> itzl78_l89 112190—305 172306—357

itz92_15E—173_l77 *

(signature fragment)

_49— 92 "153—I73 -178 —l90—306_

Scheme 2. 1. The oxidized form of the native pea chloroplast fructose bisphosphatase is
subjected to cyanylation and alkaline cleavage. All the free cysteines at the starting point
are cyanylated and speciﬁcally cleaved, resulting in six fragments. In the upper panel, the
observation of the signature fragment, marked with an asterisk, supports the existence of
Structure A, in which there is a disulﬁde bond between Cysl73 and Cysl78. Likewise,
the observation of the fragment itz-92-l77 (in the lower scheme for Structure B) indicates
a disulﬁde bond between Cysl 53 and Cysl 73.

Cyanylation of the sulﬂrydryl group of a single free cysteine residue in a peptide
followed by CN-induced cleavage in aqueous ammonia will produce two fragments (15).
If the sequence of the protein is known, mass mapping analysis (comparing the
experimentally-determined masses from mass Spectrometry against the calculated masses
of the expected cyanylation-induced cleavage fragments) will allow identiﬁcation of the
cysteine status. AS shown in Figure 2. 2 and Table 2. 1, mass mapping analysis of the

CN-induced cleavage products of cyanylated pea chloroplast fructose bisphosphatase

26

indicated the presence of CN-induced fragments consisting of residues 2-48, itz-49-91,
itz-92-152, itz-92-177, itz-92-189, itz-153-189, itz-190-305, itz-306-357, itz-306-357
(doubly charged), and an itz-306-357 dimer (where itz = iminothialzolidine-carboxyl
blocked amino terminus). CN-induced cleavage occurred at six cyanylated cysteine
residues as illustrated in Scheme 2. 1. Cleavage did not occur at Cysl73. One possibility
could be that in the original enzyme as isolated, Cysl73 is disulﬁde bonded and is never
free, or free Cysl73 might be present in the original enzyme, yet its CN-induced cleavage
products itz-153-172 and itz-173-177 were not detected, possibly due to the signal
suppression from other peptides in MALDI, or from interference from matrix ion
background. In itz-92-177, Cy5153 must be involved in a disulﬁde bond with Cysl73,
because these are the only two otherwise unmodiﬁed in this fragment, and they must be
oxidized to match the calculated mass value of the fragment. In itz-153-189, Cysl73
must be involved in a disulﬁde bond with Cysl78, as both of these residues are oxidized
to ﬁt the mass calculation (and neither is available for cyanylation). Itz-92-189
represents a peptide containing a disulﬁde bond between two Cys residues and a third
Cys residue that has undergone B-elimination. Details about B-elimination and other side
reactions have been described previously (16). These data are consistent with Cysl73
forming two alternate disulﬁde bonds in this enzyme: CyslS3-Cysl73 and Cysl73-
Cysl78. Fragments that would have indicated a CyslS3-Cysl78 disulﬁde bond, namely
itz-92-172 and itz-173-189 linked by the disulﬁde bond (m/z 10381.1) and its prompt
fragmentation(17) products, itz-92-172 (m/z 8636.1) and itz-173-l89 (m/z 1748.0), were

not observed.

27

Peak 4 of the MALDI spectrum of the pea chloroplast fructose bisphosphatase
(Table 2. 1) corresponds to a CN-induced cleavage product consisting of residues 2
through 48. The transit peptide must then be cleaved after, not before, the methionine
shown as residue 1 in PDB ﬁle ldcu (8) when the enzyme is transported into the
chloroplast. That amino acid sequence is based on DNA sequencing. This is the ﬁrst
identiﬁcation of the transit peptide cleavage site and the N-terminal amino acid for pea
chloroplast fructose bisphosphatase. To avoid confusion, the numbering from the PDB

ﬁles has been retained for the crystal structure of the pea enzyme.

 

 

Peak Identity Observed mass Calculated mass Deviation
(Da) (Da)
1 doubly-protonated itz—51-93 2285.35 2286.57 0.053%
2 doubly-protonated 112-307-358 2904.13 2904. 30 0.005%
3" ltz-155-190 with 174 and 179
3917.06 3918.21 -0.029%
oxidized
4 itz-51-93 4571.21 4572.13 0.020%
5 itz-3074558 5807.80 5807.60 0.003%
6" itz-94-178 with 155 and 174
9097.61 9097.76 -0.002%
oxidized

 

Table 2. 2. Assignment of peaks in the MALDI mass spectrum after CN-induced
cleavage of the recombinant spinach chloroplast fructose bisphosphatase

“Peak 6 indicates a disulﬁde bond between CyslSS and Cysl74, consistent with the
crystallography data. The alternative disulﬁde bond between Cysl74 and Cysl79 is
indicated by peak 3.

28

 

 

 

 

  

 

 

 

 

 

9001 2 2904.13
.5
g 600, 1 2285.35
3
E
300- 3 3917.06
0 1.. . .... .JL... illu. . ...
2000 2600 3200 3800 4400
mlz
.4 4571.21
22°00 5 5807.80
17600 -
E 13200 ~
2
3
.E
5" 9097.61
4400 5400 6400 7400 8400 9400 10400
mlz

Figure 2. 3. MALDI mass spectrum of CN-induced cleavage products of recombinant
spinach fructose bisphosphatase. The mass spectrum is divided into two mass ranges due
to the intensity differences. The assignment of the labeled peaks is shown in Table 2. 2.

29

The signal to background ratio for the itz-153-189 peak (peak 2 in Table 2. 1)
indicating the alternative disulﬁde bond between Cysl73 and Cysl78 in the MALDI
spectrum for the pea leaf fructose bisphosphatase cleavage products was relatively low.
This could be the result of suppression of the ionization of that particular CN-cleavage
fragment by the other peptides during analysis by MALDI. Another possibility is that
that fragment was less abundant in the mixture. To seek more convincing evidence, a
Similar mapping experiment was performed on recombinant spinach chloroplast fructose
bisphosphatase, a homologue of the pea fructose bisphosphatase, using ESI-LC/MS (as
described below) as well as MALDI-MS. In ESI-LC/MS, the cleavage fragments were
separated on a C18 column prior to MS analysis, thereby reducing the chance of signal
suppression between the individual components. ESI also has a better detection limit
(tens of femtomoles) than MALDI. The ion trap mass spectrometer serves as the detector
for the LC. It scans over the m/z range (m/z 200-2000) at a rate of ~0.7 scans/sec.
Because the column was relatively short (5 cm), a long elution program (150 min) with a
shallow gradient was used to achieve a better separation. A total of 5983 scans was
collected during the chromatographic separation.

The MALDI spectrum of the CN-induced cleavage products of the recombinant
Spinach chloroplast fructose bisphosphatase was shown in Figure 2. 3. The peak
assignments are detailed in Table 2. 2. The results are similar to those obtained from
analysis of the pea leaf enzyme. Among the CN-induced cleavage fragments are
protonated itz-155-190 (analogous to itz-153-189 described above for the pea leaf

fructose bisphosphatase cleavage products), with a disulﬁde bond between Cysl74-179

30

and a calculated mass of 3918.2 Da, and protonated itz-94-l78, with a disulﬁde bond
between Cysl 55-174 and a calculated mass of 9097.76 Da.

The ESI-LC/MS results are shown in Figure 2. 4. The upper frame is the mass
chromatogram plotting the relative intensity of the base peak in the entire m/z range (m/z
200-2000) over the time of spectrum acquisition. The middle frame and the lower frame
in Figure 2. 4 are mass chromatograms plotting the total ion current in a speciﬁed m/z
range against the time of spectrum acquisition. The middle frame plots m/z 1305.8-
1306.4 that corresponds to the triply charged itz-155-190 fragment with a disulﬁde bond
between Cysl 74 and 179. The bottom frame plots m/z 19586-1959.] that corresponds to
the doubly charged itz-155-190 fragment. The fact that the two chromatographic peaks
Share the same retention time (51.3 min) suggests that the two peaks represent the same
peptide with different degrees of protonation. Figure 2. 5 shows the mass Spectrum of the
CN-induced cleavage fragments eluted from the column in the chromatographic window
indicated in Figure 2. 4. The major peaks correspond to the double and triply charged itz-
155-190 fragment. These results indicate that there is a second disulﬁde bond in the
recombinant Spinach ﬁ'uctose bisphosphatase, and that bond joins Cysl74 and Cysl79,
consistent with the results obtained from the disulﬁde mapping of the native pea
chloroplast enzyme. Clearly, two alternate disulﬁde bonds are present in the oxidized
form of the Spinach enzyme, as well as in the pea chloroplast enzyme, as isolated. As
with the spinach enzyme, fragments indicative of a CyslSS-Cysl79 disulﬁde bond,
namely itz-94-173 and itz-174-190 linked by the disulﬁde bond (m/z 10344.1), and the
prompt fragmentation products itz-94-173 (m/z 8585.1) and itz-174-190 (m/z 1762.0),

were not found.

31

100% 4

 

 

 

 

 

 

 

 

g 80% - mass chromatogram 1

5 60% , mass range m/z200-2000

C

3 40% l

E 20% «

a m—ﬂhq
0% .

0 15 3O 45 60 75 90
peak at 51.3 m‘n that
corresponds to the doubly

100% . charged and trny charged
3 80% q mass chromatogram 2 112-155-190wnh174 and
'2 mass range m/z1305.8-1306.4 179 oxidzed
3 60% 1
.5
o 4
g 410/0
:3 20% 1
0% -
0 15 30
5‘ 100%-
g 80% - mass chromatogram 3
.3 60% ~ mass range m/z1958.6-1959.1
'- 40% ~
g 20% 4
3
2 0% A3 A ---LL‘_L..~_‘_.M
0 15 30 45 60 75 90

 

 

time (mln)

Figure 2. 4. Mass chromatograms corresponding to the analysis of the CN-induced
cleavage products of the recombinant spinach fructose bisphosphatase by LC-ESI-MS.
The top panel is a chromatogram reconstructed from the entire mass range (m/z 200-
2000). The second panel is a mass chromatogram of ion current at m/z 1306
corresponding to the triply charged itz-155-190 fragment containing a disulﬁde bond
between Cysl74 and 179. The third panel is a mass chromatogram at m/z 1959
corresponding to the doubly protonated CN-induced cleavage fragment itz-155-190.

32

1306.05 (+3)
100% l

80% l 1958.87 (+2)
60% 2

40% ~

 

relative intensity

20% -

 

 

 

0% ~ .
200 400 600 800 1000 1200 1400 1600 1800 2000

mlz

Figure 2. 5. An averaged Spectrum (of 40 spectra) acquired during the analysis of the CN-
induced cleavage products of the recombinant spinach fructose bisphosphatase by LC-
ESI-MS. The two most intense peaks in the mass Spectrum correspond to the doubly and
triply charged itz-155-190 fragment, respectively, with a disulﬁde bond between Cysl74
and Cysl 79.

Modeling

We used the coordinates of ring D of the spinach enzyme in PDB ﬁle lspi (9) to
estimate the energy values of the reduced enzyme and of the oxidized forms of the
enzyme with the two alternate disulﬁde bonds, and with the 155-179 disulﬁde bond that
was not detected in the wild type pea and spinach enzymes (Table 2. 2). All of these
disulﬁde bonds are energetically possible according to these calculations, but the CyslSS-
Cysl79 disulﬁde bond is energetically less favorable (by ~270 kcal/mol, enthalpy), and,
therefore, it would not be predicted to be present in the predominant oxidized species.
The positions of the three Cys residues in the modeled structures are shown in Figure 2.

6. In the crystal structure of the oxidized pea enzyme, the loop has contracted into a Short

33

helix. In ring A Of that structure, the C01 separation between Cysl73 and Cysl78 is 15.7

A and the Sy separation is 8.58 A.

 

 

 

 

Form of
Enthalpy
Enzyme
Final Bond Non-Bond Coulombic
(kcal/mol)
Fully Reduced 2107.0 1004.3 632.2 -1767.7
155-174 cystine 1443.3 980.4 923.2 -2549.3
174-179 cystine 1439.7 979.3 889.7 -2550.2
155-179 cystine 1713.8 991.7 922.3 -2387.5

 

Table 2. 3. Summary of energy values of the oxidized Cy3155-174, Cysl74-179 and
Cy5155-179 forms of the spinach chloroplast fructose bisphosphatase, and of the reduced

form, following energy minimizationa

‘”The minimization protocol is described earlier in ‘methods’ section.

represents a total of 9000 to 12000 steps.

34

Each series

155
174 155 ‘74
y, 179 179 '
c) d
155
174

 

Figure 2. 6. Modeled structures of the loop containing the redox-sensitive cysteine
residues in the reduced form of spinach chloroplast fructose bisphosphatase, in the two
oxidized forms observed in these experiments (‘b’ and ‘c’), and in the oxidized form
containing the 155-179 disulﬁde bond (‘d’) that apparently can occur when Cysl74 is
replaced by mutation. In the oxidized form of the pea leaf enzyme in PDB ﬁle 1d9q (8),
the loop has contracted into a short helix.

In these cyanylation-based disulﬁde mass mapping experiments, the protein is
exposed to the reagent, 1-cyano-4-dimethylamino-pyridinium (CDAP), which cyanylates
all the free cysteine residues in a protein, yet remains non-reactive to the disulﬁde bonds
(cystines). The modiﬁed protein, under alkaline conditions, is cleaved Speciﬁcally at the
N-terminal side of the cyanylated cysteine residues, resulting in an iminothialzolidine-
carboxyl blocked N-terrninus and an amide C-terrninus. The acidic condition of the
cyanylation reaction effectively thwarts the disulﬁde scrambling reactions. Blocking of
the free cysteines with a cyano group further prevents disulﬁde scrambling at the alkaline
cleavage step. Scheme 2. 1 provides a structural correlation between the sites of the
cysteines involved in the putative dual disulﬁde bonds in pea fructose bisphosphatase and
the observed CN-induced cleavage fragments; in this case, deductive reasoning indicates

that the native structure of pea fructose bisphosphatase contains a disulﬁde bond between

35

CyslS3 and Cysl73. Consideration of the fact that no free cysteines exist after
cyanylation, coupled with the capability to differentiate between the mass of a cystine
and that of two cysteines, infers the existence of a disulﬁde bond between the two
internal cysteines in the CN-induced cleavage products marked with an asterisk. During
the CN-based disulﬁde mass mapping of pea leaf fructose bisphosphatase, a peak
corresponding to itz-92-177 with a disulﬁde bond between CyslS3 and Cysl73 was
detected in the MALDI mass spectrum of the CN-induced cleavage reaction mixture. In
addition, another fragment, itz-153-189 with a disulﬁde bond between Cysl73 and
Cysl78, was observed. Based on this information, the dual disulﬁde structure in Scheme
2. 2 is proposed.

The results of analyzing the recombinant spinach enzyme by the CN-based
disulﬁde mapping methodology indicated that this protein has a disulﬁde bond structure
analogous to that of the pea enzyme. The observation of two CN-cleavage fragments,
namely itz-94-178, with a disulﬁde bond between Cy3155 and Cysl74, and itz-155-190,
with a disulﬁde bond between Cysl74 and Cysl79, provided direct evidence for the

proposed disulﬁde structure of the recombinant Spinach protein.

36

The CyslS3-Cysl73 disulﬁde bond is obvious in the protein crystals used for the
determination of the structure of the oxidized enzyme (PDB ﬁles 1d9q and ldcu (8)).
The second disulﬁde bond, between Cysl73 and Cysl78, is not evident in those crystals.
In ring D of the structure of the reduced form of the spinach enzyme (PDB ﬁle lspi (9))
these residues are well positioned to form a disulﬁde bond. This disulﬁde bond was also
suggested by Balmer et al. (7) on the basis of Site-directed mutagenesis experiments. We
do not ﬁnd evidence for a third disulﬁde bond, joining CyslS3-Cys178, suggested by
Jacquot et al. (6), Balmer et al. (7), and Reichert et al. (5) on the basis of their
mutagenesis experiments, nor do we ﬁnd evidence for the disulﬁde bond in the interior of
the protein between Cys49 and Cys190 suggested by Li et al. (18).

Notably, replacement of CyslS3 with a serine results in the formation of an
enzyme that is permanently active (6,7). This suggests that if the Cysl73-Cysl78
disulﬁde bond does form, it does not affect the activity of the enzyme. Balmer et al. (7)
suggest that the Cysl73-Cysl78 disulﬁde bond is non-physiological. It formed slowly in
the recombinant enzyme in which Cy5153 was replaced by a serine. In the C153S
mutant, this disulﬁde bond can only be formed by oxidation, but in the wildtype enzyme,
with three Cys residues in the 1705 loop, this bond could be formed by thiol-disulﬁde

exchange.

37

SH SH SH SH

 

 

—4|9 —9L —153 —l73 -—l78 —— l9|0 —i06 —

Scheme 2. 2. The proposed disulﬁde structure of the oxidized form of pea chloroplast
fructose bisphosphatase. The protein is a mixture of two Species, each with a single
disulﬁde bond. In one species, as illustrated by the solid-line bracket, Cysl 53 and Cysl 73
form a disulﬁde bond. In the second species, as illustrated by the broken-line bracket,
Cysl73 and Cysl78 form a disulﬁde bond.

It is reasonable to assume that the inactivating CyslS3-173 disulﬁde bond is
formed initially and that the enzyme containing this disulﬁde bond is the predominant
species in the chloroplast in the dark. The 1705 loop forms a very negative Shield that
will protect the Cy5153-Cysl73 disulﬁde bond from external attack by thiolates and
decrease the possibility of the formation of mixed disulﬁdes. Cysl78 is buried on the
other side of the Cysl 53-Cysl73 disulﬁde bond. When dissociated to a thiolate, Cysl78
should be able to attack the CyslS3-Cysl73 disulﬁde bond. The pH in the stroma rises
about 1 unit when chloroplasts are illuminated. As the pH rises, the Cysl78 sulﬂrydryl
should begin to dissociate, thereby gaining the ability to attack any disulﬁde bonds in the
vicinity. A pH-dependent shift from one disulﬁde form to the other could account for the
difference in reductive activation seen when the enzyme is assayed in the presence of
high levels of substrate and Mg++ at pH 7.8 (full activation) and at pH 8.8 (only about
20% activation) (19).

We cannot exclude the possibility that the Cysl73-178 disulﬁde bond was formed
during enzyme isolation, either by direct oxidation or by thiOl-disulﬁde exchange. It is

reasonable to assume that if the Cysl73-178 disulﬁde bond is present in some of the

38

enzyme molecules in those preparations, that it will also be present in some of the
enzyme molecules in the chloroplast in the green leaf after several hours of darkness.

There is no evidence for the presence of the Cysl73-Cysl78 disulﬁde bond in the
crystal structure of the recombinant pea chloroplast enzyme in Protein Data Bank ﬁles
1d9q and ldcu. The fully reduced form is absent. There are no disulﬁde bonds apparent
in the structure of the reduced form of the spinach enzyme in Protein Data Bank ﬁle lspi,
and there is no indication in the description of the puriﬁcation of the spinach enzyme (9)
that it was protected from oxidation. Apparently, fructose bisphosphatase crystallizes out
as the reduced form or as the oxidized form, and, in the case of the oxidized enzyme, the
two isoforms do not co-crystallize. Neither crystallography nor mass Spectrometry
provide quantitative data, and therefore the relative amounts of the two disulﬁde-
containing isomers in vivo, and in the enzyme as isolated, remain to be determined.

These experiments further illustrate the utility of cyanylation-induced
cleavage/mass mapping in determining the location of protein disulﬁde bonds. Clearly,
where alternate disulﬁde bonds are possible, a crystal structure showing only one of the

possible disulﬁde bonds is not deﬁnitive.

2. Cobalamin-independent methionine synthase (MetE)
2.1. Introduction

Cobalamin-independent methionine synthase (MetE, EC. 2.1.1.14) catalyzes the
transfer of a methyl group from 5-methyltetrahydropteroylpolyglutamate (CH3-

H4PteGlun, n 2 2) to the thiol group of L-homocysteine (Hcy) to form

tetrahydropteroylpolyglutamate (H4PteGlun, n 2 2) and L-methionine (1) in the terminal

39

step of methionine biosynthesis in Escherichia coli. The 753-residue enzyme contains
seven cysteines, located at residues 323, 353, 516, 560, 643, 645 and 726. It has been
established that the MetE enzyme contains one equiv of zinc, which is essential for the
enzymatic activity, and that the homocysteine thiol group directly ligates to the active site
zinc(20,21). MetE from E. coli has high sequence homology to methionine synthases
from plants, fungi, and oether eubacteria. Sequence alignment Shows that His641,
Cys643, and Cys726 are the only conserved amino acid residues throughout the
alignment. Binding assays and activity assays of the wild type and the four mutants
(Cys726Ser, Cys6438er, His641Gln, and His641Asn ) suggest that these three residues
bind with zinc in the native enzyme. It is hypothesized that the zinc in MetE also ligates
with a molecule of water, which is replaced by the sulfur in homocysteine during the

methyl transfer reaction (22).

CH3 — H ,PteGlun +L_homocysteine(RSH) ——> H 4PteGlun +L_methionine(RSCH3) (l)

The activity of many proteins is regulated by the redox state in the environment.
Zinc ﬁnger transcription factors such as members of the Spl family have been
demonstrated to be redox-regulated both in vitro and in vivo(23). The zinc ﬁnger of
replication protein A as well as the zinc ring ﬁnger protein SAG have been shown to be
highly sensitive toward oxidation and reduction processes (24,25). Metallothionein, an
important cystosolic zinc storage protein, has been shown to utilize disulﬁde bond
formation to release its stored zinc upon oxidative stress (26,27). This allows the efﬁcient
transfer of zinc from the high-afﬁnity zinc center of metallothionein to proteins with

much lower zinc binding afﬁnity. Reversible disulﬁde bond formation is a very

40

convenient way to translate environmental changes into changes in protein activity. Here,
we investigate the disulﬁde structure of oxidized MetE using mass spectrometry and

correlate the result to the observation of zinc loss during oxidation.

2.2. Materials and Methods

Enzyme isolation

The recombinant MetE was overexpressed by growing the appropriate E. coli
strains aerobically in Luria-Bertani medium supplemented with 100 pg/mL ampicillin,
0.5 mM zinc sulfate, and 0.4 mM isopropyl B-D-thiogalactopyranoside (IPTG) as
previously described (20). Most strains were grown at 37 ° C to late log-phase or early
stationary phase (A420 about 8.0). MetE was then puriﬁed to homogeneity by a single step
DEAE-Sepharose ion-exchange chromatography using a linear gradient of potassium
phosphate buffer (180-500 mM at pH 7.2 containing 500 pM DTT) as previously
described (20). Protein purity was analyzed by electrophoresis in 12% polyacrylamide
gels in the presence of sodium dodecyl sulfate (SDS-PAGE) and visualized by
Coomassie blue staining. Puriﬁed proteins were concentrated to about 100 mg/mL and

stored at -80 C.

Oxidation and Protein Estimate

Oxidized glutathione (GSSG) was added to the reduced MetE to make a ﬁnal
solution containing 70 pM of MetE, 10 mM of GSSG and 20 mM of triS-buffer (pH 7.2).
The Protein/GSSG mixture was incubated at 37 C for 90 min. The GSSG was removed at

the end of the incubation by running the mixture through a Bio-Rad 6 spin column that

41

had been equilibrated with 20mM tris buffer. Protein concentration was determined by

the absorption coefﬁcients at 280 nm (1.62 mg'1 cmz).

C yanylation (CN) and CN-induced cleavage and HPLC

For cyanylation of the reduced and oxidized MetE, cyanylation reagent, 20 pL of
1-cyano-4-dimethylamino-pyridinium (CDAP) tetraﬂuoroborate solution (100 mM
CDAP, 8 M urea, 100 mM sodium citrate at pH 3), was added to 50 pL of reduced or
oxidized enzyme (70 pM, with 20 mM of tris-buffer). After 20 min at room temperature,
the reaction was terminated by separation of the cyanylated protein from the reaction
mixture on a Bio-Rad 6 Spin column that had been pre-equilibrated with 8 M urea. For
the alkaline cleavage reaction, 70 pL of 2 M NH40H was added to the cyanylated protein
(70 pL, in 8 M urea) to make the ﬁnal NH40H concentration 1 M. After 1 hour at room
temperature, the cleavage reaction was terminated by removing the ammonia by vacuum
aspiration. The cleavage mixture was dissolved in 0.1% TFA solution for separation. A
Michrom BioResourceS Ultrafast Microprotein Analyzer (Auburn, CA) (microbore
HPLC system) was used to separate the cleavage mixture. A gradient of 15% to 75% B
(solvent A: 0.1% triﬂuoroacetic acid, 95% water, 5% acetonitrile; solvent B: 0.1%
triﬂuoroacetic acid, 5% water and 95% acetonitrile) over 50 min was ramped at a ﬂow
rate of 50 pL/min through a Michrom microbore column (150-mm length, l-mm i.d.,
packed with 8-u, 4000A polymeric stationary phase, part number 901/00711/00). UV
detector was set at 214 nM. The manually collected fractions were lyophilized to dryness.
Aqueous solvent (5 pL of 0.1% triﬂuoroacetic acid) was added to the dried sample for

further MALDI analysis. For the fractions identiﬁed as with glutathionated peptides, a

42

total reduction was also performed by adding 2 pL of 100 mM TCEP to 1 pL of the

reconstituted HPLC fraction.

Mass spectrometry

MALDI mass spectra were obtained on a Voyager DE-STR mass spectrometer
(Applied Biosystems, Foster City, CA) equipped with a 337-nm nitrogen laser. The
accelerating voltage in the ion source was set at 25 kV. A solution of saturated Ot-cyano-
4-hydroxycinnamic acid (Aldrich Chemical Co.) in 1:2 of acetonitrile/water was diluted
with isopropanol (1:4 v/v). The diluted solution (30pL ) was applied to the ﬂat MALDI
probe using Cadene’s protocol (13) to form an ‘ultra-thin layer’. The matrix solution
consisted of a saturated solution of a-cyano-4-hydroxycinnamic acid (Aldrich Chemical
Co.) in 1:1 of acetonitrile/water. The reconstituted cleavage fragment solution (1 pL)
from HPLC was mixed with 1 pL of matrix solution, and 0.5 pL of the peptide/matrix
mixture was deposited onto the probe. Co-crystallization was usually observed after 1
min. External calibration was used during analysis of the intact enzymes and their CN-
induced cleavage products by MALDl-MS.

A nanospray-ESI-MS infusion assay of the intact oxidized and reduced MetE was
developed using a Micromass Q-TOF II (Micromass, Wythenshawe, UK.) mass
spectrometer equipped with an orthogonal electrospray source (Z-Spray). Tire intact
protein solution (2.5 pmol/pL MetE in 5% formic acid, 50% methanol) was directly
infused into the mass spectrometer at a ﬂow rate of 0.1pL/min by a syringe pump
through an empty nano-spray column (PicoFrit column from NewObjective, PF360-75-

15-CE-5, 15-pm tip, 75-pm column id). The source voltage was set at 2.0 kV and source

43

temperature was set at 160°C. The linear quadruple (Q1) was set to pass ions from m/z
500 to 2000 through into the pusher region of the time-of-ﬂight (TOF) mass analyzer.
MaSSLynx software was used to control the settings and View the spectrum. The ﬁnal

spectrum was an average of over 40 scans for both reduced and oxidized MetE.

2.3. Results and Discussion
Mass Measurement of the Intact Protein

We used ESI infusion to measure the intact enzyme, in both the reduced form and
the oxidized form. Multiply protonated enzyme with number of charges up to 120 was
observed. An envelop of peaks, corresponds to the same recombinant enzyme associated
with different numbers of protons. The number of charges on the peaks in the series, from
left to right, corresponds to a series of continuously decreasing integers. The m/z values

of the two neighboring peaks can be determined from the following equations:

 

m/zl = M+n (l)
n

m/z, = M+n+1 (2)
n+1

The two neighboring peaks are designated as m/z. and m/Z2 (m/zl being on the
right) in the Spectrum. M is the molecular weight of the protein and n is the number of
charges on the peak on the right (m/zl). The mass of the proton is 1 Da for illustrative
purposes; in calculations in Table 2. 4, a mass of 1.0079 Da was used for hydrogen. The
number of charges and the molecular weight can be solved from the previous two

equations:

44

n- m/zZ—l (3)

m/zl—m/z2

 

M =nx(m/zl —l) (4)

The number of charges (n) calculated from (3) iS rounded to the closest integer,
with a few exceptions when the knowledge of the charge series being a continuous
integers forces n to the second closest integer. The molecular weight is calculated from
equation (4) using the integer numbers of n. The expected mass of the reduced form of
the enzyme is 84542.4 Da, while the observed mass for reduced MetE is 84548.5 Da,
corresponding to residue 2 through residue 753 in the protein sequence from protein
database (SwissProt, P25665). The transit pro-enzyme must be cleaved after, not before,
the methionine in E. coli before the enzyme is expressed into the cytoplasm. To avoid
confusion, we have retained the numbering consistent with that in the protein databank.
From ESI-MS result, we also noticed that the mass difference between the reduced MetE
and oxidized MetE (309.07 Da) corresponds well with the mass Shift caused by a single
glutathionation event (305.32 Da = reduced glutathione GSH, MW: 307.32 Da — 2H from
disulﬁde bond formation).

S-Glutathionation is a post-translational modiﬁcation event when a cysteine
residue is covalently linked to a reduced glutathione molecule to form a mixed disulﬁde
bond, usually as a cellular response to oxidative stress(28-30) . The enzymatic reduction
of the mixed disulﬁde bond between GSH and protein (reverse reaction of
glutathionation) is mediated by thioltransferases. The reversible modiﬁcation of cysteines
with GSH disulﬁde linkages is believed to serve as a protection mechanism under
oxidative stress, i.e., it masks critical protein sulfhydryl groups from more damaging,

irreversible oxidation.

45

     
 
  

ljl
11:

 

E Oxidized M31133

 

tn 1 : i .
CI . _-‘. —-—--_+—-
8 ~ .
g ‘1 1 1 Reduced MetE
g . : i . ’
Q) l t ' 5 1
‘3‘ ' l ‘ 111 l' l ‘
' ' 1 1
l 1 ‘
‘. I V l J I i .
1 .11111111111111“lllll . , 1 i ; , 1“ : 1‘ 1‘
700 900 1 100 1300 1500 1700 m/z

Figure 2. 7. ESI-MS spectra of the oxidized and reduced MetE. The clusters of peaks
marked between two arrows are used to calculate the molecular weight (result shown in

Table 2. 4).

46

 

ESl of reduced MetE

ESl of oxidized MetE

 

 

mlz n n(integer) M (Da) m/z n Minn-31g) M (DaL

1113.44 76.52 76 84544.84 1117.53 76.16 76 84855.68
1099.09 76.60 77 84552.32 1103.06 77.05 77 84858.01
1084.94 78.06 78 84546.70 1088.94 78.01 78 84858.70
1071.23 79.17 79 84547.55 1075.17 78.98 79 84858.81
1057.88 79.99 80 84549.77 1061.74 79.79 80 84858.57
1044.83 81.13 81 84549.59 1048.61 81.23 81 84855.77
1032.12 81.75 82 84551.19 1035.87 81.99 82 84858.69
1019.66 82.77 83 84548.12 1023.40 82.80 83 84858.54
1007.50 84.02 84 84545.34 1011.20 84.00 84 84856.14
995.66 85.01 85 84545.60 999.32 85.08 85 84856.19
984.10 86.15 86 84545.66 987.72 85.92 86 84857.07
972.82 87.05 87 84547.39 976.37 86.91 87 84856.15
961.78 88.61 88 84547.94 965.27 88.46 88 84855.15
951.06 88.21 89 84554.46 954.49 88.69 89 84860.08
940.41 90.08 90 84546.01 943.86 90.13 90 84856.78
930.09 91.58 91 84546.84 933.52 91.20 91 84858.15
920.06 91.78 92 84552.70 923.40 92.07 92 84860.17
910.15 92.80 93 84550.49 913.49 92.38 93 84860.84
900.46 93.98 94 84548.59 903.72 94.28 94 84854.75
890.99 95.09 95 84548.39 894.24 95.02 95 84857.43
881.73 95.87 96 84549.23 884.94 96.05 96 84857.58
872.64 875.83

average 84548.51 average 84857.58

standard deviation 2.62 standard deviation 1.70

 

Table 2. 4. Molecular weight calculation based on the 22 peaks marked between the two
arrows in the spectra of the oxidized MetE and the reduced MetE. Average mass and its
standard deviation were calculated from the 21 independent calculations from two peaks
that correspond to the enzyme with n protons and n+1 protons. (n from 76 to 95)

47

CN-based Disulfide Mass Mapping

The mixture of cleavage fragments was separated with microbore HPLC. In
analysis of MetEmb 16 fractions were collected, concentrated and subjected to MALDI-
MS analysis. Eight fragments are expected as CN-induced cleavages occur at the seven
cyanylated cysteines along the polypeptide chain (Shown in Scheme 2. 3). Seven
expected fragments were detected by mass Spectrometry. The one that was not observed
has only two amino acid residues. The ionization of that peptide was very likely being
suppressed by the matrix ions, or a peak for a matrix ion interfered with observation of a
peak for the cleavage fragment. However, information about the status of the two
terminal cysteines that deﬁne this fragment, Cys643 and Cys645, can be readily retrieved
from the observation of two other cleavage fragments, namely, itz-560-642 and itz-645-
726. Observation of itz-560-642 indicates that Cyst560 and Cys643 were cyanylated and,
thus, had been free in the original molecule of the reduced enzyme. Similarly,
observation of itz-645-725 indicates Cys645 and Cys726 are free in the reduced MetE.
The deduction of the cysteine status in the reduced MetE by CN-induced mass mapping
is illustrated in Scheme 2. 3. We conclude from the mapping experiment that in the
reduced MetE, all seven cysteines are free (not engaged in any kind of disulﬁde

linkages).

48

Reduced MetE
2 ‘— 322 11:323— 352 11:353— 515

112516— 559 172560— 642
lSCN ISCN ISCN SCN SCN ISCN ISCN Cleavage.NH3

""323 — 353 "516—560—643 '—645 —726 — —_—_’ it:643"'"' 644 itz645— 725 "5726— 753

Not Detected

SCN
itz645 —726'—753

Incomplete Cleavage at C ys726

Scheme 2. 3. In the disulﬁde mass mapping of reduced MetE, seven out of eight expected
fragments were detected. Detection of a CN-induced cleavage fragment, itz-645-753 with
incomplete cleavage helps to determine the glutathionation sites between Cys645 and
Cys726 in the oxidized form (see Scheme 2. 4).

In the analysis of oxidized MetE, Sixteen chromatographic fractions where
collected and analyzed by MALDI. Two key fragments: itz-643-753 with one
glutathionation and one incomplete cleavage (observed mass: 12855.95, expected mass:
12855.48), and itz-643-753 with one glutathionation and one beta elimination (a Side
reaction during alkaline treatment in which HSCN is eliminated from the cyanylated
cysteine to form a dehydroalanine) (observed mass: 12798.08, expected mass: 12796.48),
were identiﬁed from mass mapping of the HPLC fraction marked with an asterisk in the
lower panel in Figure 2. 8. Only a portion of each mass Spectrum (m/z 11800-13800) is
shown in Figure 2. 9 because a peak at m/z 9402.00 corresponding to the co-eluted itz-
560-642 is about ten times more intense than the key peaks described above.

Beta-elimination reaction is a side reaction that competes with the cleavage
reaction during alkaline incubation of the cyanylated peptide. The yield of beta-
elimination varies depending on the primary sequence of the cyanylated peptide.

Incomplete cleavage, a seemingly sequence-dependent event, was observed and reported

49

in the analysis of a small, highly knitted protein before. These Side reactions, though
causing the yield of the desired cleavage products to decrease, do not pose a great threat
to the overall mapping strategy, because products from these two reactions have a
different mass from the properly-cleaved fragments, and, thus, are discemable by mass
spectrometry. In the oxidized MetE disulﬁde mass mapping, detection of these
incomplete-cleavage products essentially provided useful information for disulﬁde

structure elucidation.

 

 

 

 

 

1.E+06 -
8.E+05 -
E
g 5.9051
.5
g 4.E+05 1
8
E 2.E+05 4
O.E+00 L T . . T . .
10 20 30 40 50 60 70
min
8 E+05 l
2; 6.E+05
.E 4.E+05 —
2
'3
E 2.E+05 ‘ a
0.E+00 — .1 I Y I r ﬁ
10 20 30 40 50 60 70
min

Figure 2. 8. Microbore HPLC separation of CN-induced cleavage fragments from
cyanylated reduced (upper panel) and oxidized (lower panel) MetE. The fraction marked
with an asterisk was identiﬁed as containing two variants of a glutathionated peptide and
was subjected to total reduction and further mass Spectral analysis.

50

In addition to direct mass identiﬁcation, a complete reduction experiment was
performed on the same HPLC fraction to validate the glutathionation, as shown in
Scheme 2. 4. The loss of a glutathione molecule upon reduction of the mixed disulﬁde
bond would cause a mass shift of —305 Da for each of the two variants of the itz-643-753
fragment described in the previous paragraph. In the analysis after reduction, as expected,
the incomplete cleavage itz-643-753 fragment showed —303.27-Da mass shift, while the

beta-elimination itz-643-753 fragment showed —305.90-Da mass shift.

 

Observed mass

 

Identity Expected mass (Da) (D a) Relative Error

2-322 36264.00 36243.91 -0.055%
itz-323-352 3430.93 3429.19 -0.051%
itz-353-515 18411.56 18411.70 0.001%
itz-516-559 4730.66 4729.95 -0.015%
itz-560-642 9400.59 9399.80 -0.008%
itz-643-644 310.34 n.d. n.d.
itz-645-725 9207.30 9204.06 -0.035%
itz-726-753 3095.57 3095.96 0.013%

itz-645-753 with

incomplete 12283.85 12283.86 0.000%

cleavage at 726

 

Table 2. 5. Peak assigmnent of the fractions collected from HPLC separation of the

reduced MetE protein.

51

3500 —
3000 ~
2500 -
2000 ~
1500 ~
1000 —
500 -

0 """" T r 5'“ ‘ "
1 1800 12300 12800 13300 1 3800

mlz

12855. 95

12798.08

relative intensity

 

 

4500 4
4000 4 12552.68
3500 ~
3000 «
2500 —
2000 - l
1500 -
1000 1
500 4
o - . .
11800 12300 12800 13300 13800

mlz

12492.18

 

relative intensity

 

 

Figure 2. 9. Direct MALDI-MS analysis of one of the chromatographic fractions
(marked with "' in the lower panel of Figure 2. 8) consisting of mixture of CN-induced
cleavage fragments from the CN-induced cleavage of oxidized MetE (upper panel).

MALDI analysis of the same fraction after treatment with an excess of TCEP (lower
panel).

52

Oxidized MetE SSG SC N

”2643 —645 —726 _ 753

ISCN iSCN lSCN SCN SCN ISSG ISCN Cleavage, NH3

_ 323 _ 353 _5 16— 560-643 —645 "726 _ __’ SSG Beta Elimination
113643 —645—726 —'753

Total Reduction with
TC EP

SH SCN
it2643 —645_726 —753

SH Beta Elimination

"2643 —645_ 726 —753

Scheme 2. 4. Detection of two key variants of a CN-induced cleavage fragment itz-643-
753, limits the glutathionation site to Cys645 and Cys726. Total reduction experiment
conﬁrmed the glutathionation. The disulﬁde structure of the oxidized MetE was proposed
considering the observation from the controlled experiment of the reduced MetE, in
which incomplete cleavage at Cys726 was observed (Scheme 2. 3).

From the CN-induced cleavage mapping of the oxidized MetE, candidacy for the
glutathionation Site is reduced from seven cysteines to the two internal cysteines (Cys645
and Cys725) in fragment itz-643-753. In the mapping experiment on the reduced form of
the enzyme, we observed fragment itz-645-753 with an incomplete cleavage at Cys726,
but had not observed evidence for fragment itz-643-726 with an incomplete cleavage at
Cys645, suggesting that Cys726 is more likely to suffer incomplete cleavage under
alkaline condition than Cys645. This controlled experiment strongly suggests that
fragment itz-645-753 (before complete reduction) has the glutathionation at Cys645 and
the incomplete cleavage at Cys726. However, we do not have direct, conclusive

experimental evidence to support this.

53

Mass spectral analysis of the intact oxidized MetE and reduced MetE suggests
that there is one glutathionation event on one of the seven cysteine residues in the
oxidized form of the protein. The CN-induced mass mapping experiment limits the single
glutathione site to Cys645 or Cys726, but most likely Cys645. There is no experimental
evidence to support the formation of a disulﬁde bond between Cys643 and Cys726, the
two zinc binding sites, under oxidative stress. Zinc loss and corresponding loss of protein
activity during oxidation may be caused by the glutathionation on Cys645 that disrupts

the zinc binding pocket in its vicinity.

3. Conclusion

CN-induced disulﬁde mass mapping of both reduced MetE and oxidized MetE
Showed that CN-induced disulﬁde mass mapping methodology is not only applicable to
small, tightly-folded proteins, but also to proteins as big as 84.5 kDa. Use of a high
concentration of denaturant is critical to solubilize large, and usually more hydrophobic,
proteins. Size exclusion-based spin column proves to be a fast and efﬁcient way to
separate the cyanylated proteins (MetE and F BPase) from the cyanylating reagent.

In the analysis of the intact proteins, ESI-MS showed great mass accuracy and
could provide useful information as regards to the number of post-translational events
(glutathionation) over the entire protein sequence. However, such analysis can not
pinpoint the locus of such event. Controlled degradation of the protein is required to map
the post-translational events. CN-induced disulﬁde mass mapping proved to be valuable
in analysis of cysteine-related modiﬁcations, such as disulﬁde bond formation and

glutathionation, in that all the fragments from speciﬁc cleavages at cysteines contain

54

information regarding to the cysteine status. Fragments from beta elimination and

incomplete cleavage reactions can be used in some cases to elucidate disulﬁde structures.

Reference:

10.

ll.

12.

13.

Chueca, A., Sahrawy, M., Pagano, E. A., and Gorge, J. L. (2002) Photosynth Res
74, 235-249

Kobayashi, D., Tamoi, M., Iwaki, T., Shigeoka, S., and Wadano, A. (2003) Plant
Cell Physiol 44, 269-276

Anderson, L. E., Huppe, H. C., Li, A. D., and Stevens, F. J. (1996) PlantJ 10,
553-560

Qi, J. F., Isupov, M. N., Littlechild, J. A., and Anderson, L. E. (2001) J Biol Chem
276, 35247-35252

Reichert, A., Dennes, A., Vetter, S., and Scheibe, R. (2003) Biochimica Et
Biophysica Acta-Proteins and Proteomics 1645, 212-217

Jacquot, J. P., LopezJaramillo, J ., MiginiacMaslow, M., Lemaire, S., Cherﬁls, J .,
Chueca, A., and LopezGorge, J. (1997) FEBS Lett 401, 143-147

Balmer, Y., Stritt-Etter, A. L., Hirasawa, M., Jacquot, J. P., Keryer, lE., Knaff, D.
B., and Schurmann, P. (2001) Biochemistry 40, 15444-15450

Chiadmi, M., Navaza, A., Miginiac-Maslow, M., Jacquot, J. P., and Cherﬁls, J.
(1999) Embo Journal 18, 6809-6815

Villeret, V., Huang, S. H., Zhang, Y. P., Xue, Y. F ., and Lipscomb, W. N. (1995)
Biochemistry 34, 4299-4306

Kraulis, P. J. (1991) Journal of Applied Crystallography 24, 946-950

Anderson, L. E., Goldhabergordon, I. M., Li, D., Tang, X. Y., Xiang, M. H., and
Prakash, N. (1995) Planta 196, 245-255

Bradford, M. M. (1976) Anal Biochem 72, 248-254

Cadene, M., and Chait, B. T. (2000) Anal Chem 72, 5655-5658

55

14.

15.

16.

17.

18.

19.

20.

21.

22.

23.

24.

25.

26.

27.

28.

29.

30.

Stevens, F. J ., Li, A. D., Lateef, S. S., and Anderson, L. E. (1997) Photosynth Res
54, 185-197

Wu, J ., and Watson, J. T. (1998) Anal Biochem 258, 268-276

Qi, J. F., Wu, J ., Somkuti, G. A., and Watson, J. T. (2001) Biochemistry 40, 4531-
4538

Patterson, S. D., and Katta, V. (1994) Anal Chem 66, 3727-3732

Li, D., Stevens, F. J ., Schiffer, M., and Anderson, L. E. (1994) Biophys J 67, 29-
35

Ashton, A. R. (1998) Arch Biochem Biophys 357, 207-224

Gonzalez, J. C., Peariso, K., PennerHahn, J. E., and Matthews, R. G. (1996)
Biochemistry 35, 12228-12234

Peariso, K., Goulding, C. W., Huang, S., Matthews, R. G., and Penner-Hahn, J. E.
(1998) JAm Chem Soc 120, 8410-8416

Zhou, Z. H. S., Peariso, K., Penner-Hahn, J. E., and Matthews, R. G. (1999)
Biochemistry 38, 15915-15926

Wu, X. S., Bishopric, N. H., Discher, D. J ., Murphy, B. J ., and Webster, K. A.
(1996) Mol Cell Biol 16, 1035-1046

Park, J. S., Wang, M., Park, S. J., and Lee, S. H. (1999) JBiol Chem 274, 29075-
29080

Swaroop, M., Bian, J. H., Aviram, M., Duan, H. J., Bisgaier, C. L., Loo, J. A., and
Sun, Y. (1999) Free Radical Biology and Medicine 27, 193-202

Maret, W., and Vallee, B. L. (1998) Proc Natl Acad Sci U S A 95, 3478-3482
Jakob, U., Eser, M., and Bardwell, J. C. A. (2000) J Biol Chem 275, 38302-38310
Klatt, P., and Lamas, S. (2000) Eur J Biochem 267, 4928-4944

Thomas, J. A., Chai, Y. C., and Jung, C. H. (1994) in Oxygen Radicals in
Biological Systems, Pt C Vol. 233, pp. 385-395

Thomas, J. A., Poland, 8., and Honzatko, R. (1995) Arch Biochem Biophys 319,
1-9

56

CHAPTER 3
QUANTIFICATION OF CYSTEINE-CONTAINING PEPTIDES WITH

HOMOLOGOUS ALKYLATING REAGENTS

1. Introduction

Sensitive methods to quantify gene expression at the mRNA level have been
developed and widely used to identify clusters of genes for which the transcription levels
are linked at a Speciﬁc state (1,2). However, these methods do not reﬂect directly the
expression level of proteins, the ultimate biological effector molecules. Quantitative
proteome analysis, especially one that monitors changes in protein expression upon
perturbation, may provide insight into the mechanisms of disease and may lead to the
discovery of novel therapeutically important proteins that new drugs can target.

Conventional protein quantiﬁcation methodology is based on 2-D polyacrylamide
gel electrophoresis (PAGE), followed by gel excision, in-gel digestion and LC/MS/MS to
identify the proteins in a particular gel spot (3-5). The PAGE method is problematic
when two proteins cannot be resolved fully from the labor intensive, hard-to-automate 2-
D gel separation; furthermore, this method is biased toward highly abundant proteins, but
against membrane proteins, very large or very small proteins, very acidic or very basic
proteins, and most importantly, proteins with low expression levels, such as transcription
factors and protein kinases.

New quantitative methods based on protein modiﬁcation with amino acid-Speciﬁc
reagents, proteolytic digestion and mass Spectrometry have been developed to quantify

complex protein mixtures (6,7). Isotope-coded afﬁnity tags (ICAT) reagents were

57

developed by Gygi et al. for such purposes (7). The structure of a representative ICAT
reagent is shown in Figure 3. 1. Labeled and unlabeled forms of the reagent can be used
to quantitatively compare the same proteins in two different samples. The reagent
contains three moieties: a thiol-speciﬁc reactive group attaches the reagent to proteins via
alkylation, a linker group that can be labeled with eight deuterons to distinguish the

reagent pair, and a biotin group to enrich cysteine-containing peptides by afﬁnity capture.

Heavy Reagent: d8-1CAT (X=deuterium)
)1 Light Reagentde-1CAT(X=hydrogen)

NH NH

Wit/\i/l\°/VO\/\°/i\jAﬁ)l\/I

S

Thiol-speciﬁc

Biotin Linker (heavy or light) Reactive Group

Figure 3. 1. Structures of ICAT reagent pair used for protein quantiﬁcation.

The ICAT quantiﬁcation method includes three steps: (1) Use of the light (or
heavy) reagent to label an aliquot of totally reduced protein mixture at state one; use of
the other reagent to label an equal aliquot of totally reduced protein mixture at state two.
(2) Mix the two aliquots and subject them to proteolytic digestion. (3) Use afﬁnity
chromatography based on the biotin-avidin interaction to capture the peptides that are
modiﬁed with ICAT reagents and use mass spectrometry to analyze the captured
cysteine-containing peptides. The protein concentration ratio at the two states is reﬂected
by the relative intensity ratio of the mass Spectral peaks corresponding to the differently

labeled (do vs. d3) peptides of the same protein.

58

A similar approach, named MCAT (mass-coded abundance tag) (6), uses a lysine-
speciﬁc reagent to modify the amino group of the lysine residues in the protein of the ﬁrst
state. In this approach, the proteins in the second state are not modiﬁed. Because
modiﬁcation at lysine resides will disable tryptic cleavage at such residues, the proteins
of the two different states, unlike in the ICAT approach, are digested separately, prior to
being modiﬁed. After the modiﬁcation, the peptides are mixed for MS analysis. The two
differently-modiﬁed peptides used for quantiﬁcation have a mass difference of 42 Da,
corresponding to the mass Shift caused by the modiﬁcation group.

In both ICAT and MCAT approaches, quantiﬁcation of particular proteins is
accomplished through quantiﬁcation their differently modiﬁed proteolytic peptides. In
this chapter, we developed a method that uses two brominated, structurally homologous
reagents to quantify cysteinyl peptides. The two reagents used are 2-bromoacetamide (2-
BA) and 2-bromopropionamide (2-BP), with their structures shown in Figure 3. 2. A
peptide modiﬁed by 2-BA has a 57-Da mass difference from the original peptide with a
free sulfhydryl group; A peptide modiﬁed by 2-BP has a 71-Da mass difference to the
original peptide. The net difference of 14 Da between the two modiﬁed products arises

from the overall elemental composition difference of a CH2 between the two reagents.

59

 

O Rl—CIZH— R2
A 1__ c.
R,——— C1~I—-——R2 Br_ CHT—C NH2 l 2 ii
i 2-BA S —CH2—C— NHZ
CH,
| , + 0 +57 Da
SH ll R,——-CH—— R2
Cysteinyl Peptide Br —(l:H _ C — NH2 —> (lin 0
CH3 S TH C NH2
2-BP
CH3
+71 Da

Figure 3. 2. The two reagents 2-BA and 2-BP modify the cysteine residues in a
peptide/protein. The reagents have a CH2 as the net difference in their composition, as
reﬂected by a mass difference of 14 Da in their differentially modiﬁed products.

To quantify a cysteinyl peptide in two states, a different brominated homologous
reagent is used to label the peptide in each state, as illustrated in Scheme 3. 1. 2-BA
causes a +57 Da mass shift to the original peptide; 2-BP causes a mass Shift of +71Da.
After the modiﬁcation, the two products are mixed together for MALDI-MS analysis.
The relative concentration ratio of the two peptides is represented by the intensity ratio of
mass spectral peaks representing the two products.

This analytical method, Sharing the same working principle as those for ICAT and
MCAT, provides a Simpliﬁed system to explore the potential problems of such
quantiﬁcation processes and to evaluate the various assumptions under which ICAT and
MCAT work. This method, successful in quantiﬁcation at the cysteinyl peptide level,
could be utilized to quantify protein complexes by adding a biotin-like afﬁnity tag into

the structure of the reagents. A modiﬁed method is proposed that could be used to

60

quantify proteins with different disulﬁde structures, such as the two oxidized FBPase

isomers discussed in chapter two.

State 1: State 2:
Cys-peptide:
+ 2-BA \ / + 2-BP
———+71Da --------------------

—-—-+57Da ''''' 1 ............. I

 

 

 

Scheme 3. 1. A cysteine-containing peptide from a protein at two different states is
modiﬁed with 2-BA and 2-BP, respectively. The two modiﬁed peptides are then
combined for analysis by MS. The relative intensity ratio of the two mass spectral peaks
that correspond to the two products represents the quantity ratio of the peptide/protein in
two different states.

2. Materials and Methods

Peptide Modiﬁcation

Peptides (American Peptides CO., Sunnyvale, CA) were dissolved in aqueous
buffer (100 mM tris, pH8.5, 10 mM dithiothreitol). 2-BA (100 mM dissolved in
methanol) was added to the peptide solution with a molar ratio of 300: 1. 2-BP (700 mM

dissolved in methanol) was added to the peptide solution with a molar ratio of 2000:].

61

Alkylation with 2-BA is allowed to proceed at room temperature for 5 min, while 2-BP is
allowed to proceed at room temperature for 30 min. The reaction was terminated with the
addition of a quencher solution (saturated or-cyano-4-hydroxycinnamic acid in 50%
aqueous acetonitrile, 1% acetic acid) to the reaction mixture (1: 1 v/v) after the
designated period of time for reaction at room temperature. The pH of the ﬁnal solution is
~3.5, which efﬁciently prevents the sulﬁrydryl groups from deprotonation. MS analysis
of the peptide mixture showed that no additional alkylation occurs under such pH

condition.

Mass Spectrometry

The MALDI target was coated with a thin layer of Ot-cyano-4-hydroxycinnamic
acid matrix, as described by Cadene et al. (8). 0.5 pL of peptide/quencher mixture was
directly deposited onto the target. The co-crystallization formed almost instantaneously.
The residue liquid evaporated at ambient conditions in about 10 minutes. The target was
inserted into a MALDI mass Spectrometer (MALDI DE-STR, Applied Biosystems,
Foster City, CA) with a reﬂector. All MS spectra were acquired in reﬂectron mode and
averaged on 100 laser Shots (9). Data Explorer software (Applied Biosystems, Foster
City, CA) was used to view and analyze data. Information such as peak height and peak

area was directly derived from the software.

Alkylation Reaction Yield Estimation

The yield of the desired singly modiﬁed peptide (YS) can be represented by the

peak height (monoisotopic) of the unmodiﬁed peptide (Hu), the peak height

62

(monoisotopic) of the singly-modiﬁed peptide (HS) and sum of the peak height

(monoisotopic) of the multiply modiﬁed peptide (Hm) in the following equation:

Hs

Ys=
Hu+Hs+Hm

x100%

 

Similarly, the yield of the multiply modiﬁed peptide (Ym) can be represented by this

equation:

Hm

Ym=
Hu+Hs+Hm

x100%

 

3. Results and Discussion

3.1. Optimization of the Alkylation Reaction Conditions

The desired reaction Should be speciﬁc to cysteine and it should go to completion.
Other favorable characters involve a mild pH condition and relatively short incubation
time as to minimize peptide decomposition. The reactivity of brominated alkylating
reagents, such as 2-BP and 2-BA, is much lower than that of their iodinated counterparts
due to the fact that bromide (Br') is a poorer leaving group than iodide (I'). Brominated
alkylating reagents were reported to work at a higher reagent-to-thiol ratio (400), a longer
incubation time (2 hours), and a higher pH (9.0) (10,11). We evaluated the effect of pH
on the reaction yield. Within three differently buffered conditions (pH 8.5, 9, and 9.5)
studied, we found no Signiﬁcant difference in their effects on the yield. This result can be

explained as the thiols are sufﬁciently deprotonated under those conditions. In later

63

experiments, pH 8.5 was used. Methanol was used as a solvent to dissolve reagents 2-BP
and 2-BA, rather than DMF(N,N-dimethylformamide) as reported by Gardner et al. (10)
for two reasons: First, 2-BA and 2-BP have a good solubility in methanol. Second and
more importantly, methanol is more volatile than DMF. The low volatility of DMF was
found to interfere with the co-crystallization process between matrices and peptides,
resulting in very poor mass spectrometric signals for the peptides. Methanol does not
seem to affect the co-crystallization or interfere with the ionization during MALDI, and,

thus, the reaction mixture can be directly analyzed by MS without further puriﬁcation.

Alkylation with a 2000-molar excess of 2-BA and 2-BP was used to modify the
peptide HCKFWW (mono-isotopic MH+: 906.4 Da) separately, under above-mentioned
conditions for 2 hours. Then, the reaction mixtures were combined and directly analyzed
by MALDI-TOF-MS. The 2-BP modiﬁcation proceeded as expected: the desired mono-
alkylation (with +71-Da mass Shift) was the predominant product, as shown in the
Spectrum in the top panel of Figure 3. 3. However, for 2-BA modiﬁcation under the same
condidtion, products with one to ﬁve alkylation events (with +57xn-Da mass Shift, 11:
1~5) were observed. Besides the cysteine residue, the side chains of histidine and
tryptophan, as well as the N-terminal amino group, were modiﬁed by 2-BA at the same

time.

Over-alkylation with iodoacetamide was reported to occur more preferentially at
N-terminal NH2 of the peptide and at the side chain of the histidine residue than at the
Side chain of lysine at relatively low pH (pH 7-7.5), sometimes, even than alkylation at

the thiol of cysteine. At pH 7-8.5, alkylation at thiol occurs much faster, yet competition

64

from lysine side chain increases at the same time (12). Over-alkylation with

iodoacetamide at methionine residues was also reported by Lapko et al. (1 3).

Experiments with peptide HCKFWW showed that over-alkylation with 2-BA can
occur on tryptophan residues as well under certain conditions. It is reasonable to believe
that iodoacetamide, a stronger alkylating reagent than 2-BA, could modify the side chain

of tryptophan under similar conditions.

Stoichiometric modiﬁcation of cysteine residues is very important for both the
ICAT method and the brominated homologous reagent method discussed in this chapter,
because only peaks representing cysteine-modiﬁcation products are used to represent the
peptide concentration. If over-alkylation happened, we would have underestimated the
quantity of the peptide modiﬁed by 2-BA by using the intensity ratio of the mono-

alkylated peaks.

65

Alkylation of Peptide HCKFWW (MW:905.4)

 

 

 

 

 

 

 

 

 

 

 

    

 

 

 

100% - 977.4
80% ~
E 60% 3 1020.4
g 9634 1077.4
1: 40% - 996-4 .78. - 57°: 3
"" i2 7103 < > 5703 1134.4
20% 5‘
0%
880 930 980 1030 1080 1 130 1 180 1230 1280
mlz
964.1
100%
100% _ 964.1 5. 80%
8 °°°/° 978.1
80% — ,3 40%
" 1,._ , 211..
is: 60% 7 962 967 972 977 982
I:
§ 40% 3 907.1
"' 7108 978.1
20% - E
l 5705
0%' J."“'+‘IAA"T“T‘A‘L‘TF-T‘TT 1‘ T ”I
880 930 980 1030 1080 1 130 1 180 1230 1280
mlz

Figure 3. 3. MALDI-TOF mass spectra of the 2-BA/2-BP modiﬁcation reaction (peptide
HCKF WW) mixture. Before optimization (upper panel), up to ﬁve modiﬁcations by 2-
BA were observed for the peptide. There was no Obvious over-alkylation from the 2-BP
modiﬁcation under that condition. After optimization (lower panel), the desired products
were the predominant Species in the spectrum. No over-alkylation was observed. The
insert is a ‘zoom-in’ of the m/z range around the peaks for the desired products. l3C
isotopic peaks were well resolved.

66

A time-course study was performed for both 2-BA and 2-BP modiﬁcation of the
same peptide HCKFWW to explore the optimal condition for alkylation so that only
cysteine residues in a peptide are stoichiometrically modiﬁed. Both alkylation reactions
proceeded with a reagent-to-peptide ratio of 2000. Aliquots of the reaction mixture were
removed and quenched by mixing with acidic matrix solutions at different time intervals.
Analysis of the reaction mixture by MALDI-MS was used to assess the reaction yield at a
particular time point with the assumption that the intact peptide and its differently-
alkylated forms have the same ionization efﬁciency. Under this condition, 2-BA
modiﬁcation and 2-BP modiﬁcation took a vastly different pace, as shown in Figure 3. 4.
For 2-BP, the desired mono-alkylation reaction proceeded slowly to near-completion
after 30 minutes of incubation. A noticeable level of under-alkylation was perceptible
within 5 min, yet no signiﬁcant over-alkylation was observed even after an hour of
incubation. For 2-BA modiﬁcation of HCKFWW, the desired mono-alkylation was
achieved almost instantaneously, however, the over-alkylation began to be more

signiﬁcant after 10 min.

67

Time Course of 2-BP Modiﬁcation of HCKFWW
100%

 

 

 

 

 

 

 

 

 

 

 

i I E I ' I '
80%
60° /0 o intact peptide
3 I singly modiﬁed peptide
>. 40% A doubly modiﬁed peptide
20%
’ i
0% ‘ i i i g 1 f
0 10 20 30 40 50 60
time
Time Course of 2-BA Modification of HCKFWW
1 00%
it i - .
I I I
80% « f
60° /0 . o intact peptide
3 I singly modiﬁed peptide
>. 40% g A doubly modiﬁed peptide
20°/ ~
0 I I t I
I
0% 333—4 e e e I 9
0 10 20 30 40 50 60
time

Figure 3. 4. Time course of the two modiﬁcation reactions of peptide HCKFWW. For 2-
BP modiﬁcation (upper panel), the desired singly modiﬁed peptide yield increases
relatively more slowly. After 30 minutes, the yield is over 95%. No over-alkylation was
observed. For 2-BA modiﬁcation, the desired singly modiﬁed product yield reached 95%
within 5 minutes. The over-alkylation becomes signiﬁcant after 10 minutes.

68

From the time course analysis, we recognized that different reaction conditions
were needed for 2-BA modiﬁcation and 2-BP modiﬁcation to achieve stoichiometric
conversion to mono-alkylated products. After more experiments, we determined that for
2-BA modiﬁcation, a 300-fold molar excess of reagent and a 5-minute of incubation time
is appropriate to achieve over 95% conversion to the desired product; for 2-BP
modiﬁcation, a 2000-fold molar excess of reagent and a 30-minutes incubation time is
sufficient to achieve >95% yield. This modiﬁed condition was used to analyze the same
peptide HCKFWW at two states with a concentration ratio of 5:2. We used Z-BA to
modify the ﬁrst state and 2-BP to modify the second state. Then, we combined the
peptides in the two states for analysis by MALDI-TOF-MS. In the spectrum shown in the
lower panel of Figure 3. 3, the problem of over-alkylation for 2-BA was minimized, and

the peak intensity ratio closely represented the peptide concentration ratio.

The overall mass difference of a ‘CHZ’ in the two modiﬁed peptide products is
quite small, compared with the average size of a tryptic peptide (600-1500 Da). However,
the net structural difference of a CH2 has a relatively big effect on the reactivity
difference between a 2-carbon reagent (2-BA) and a 3-carbon reagent (2-BP). The
reactivity difference, more apparent in structural homologous reagents, also exists
between isotope-labeled isomeric reagents, but at a much lower level, such as for the
ICAT reagent pair. For example, the differently modiﬁed peptides by ICAT reagents

showed noticeable difference in retention times during HPLC analysis, suggesting the

69

two ICAT reagents have different hydrophobicity. Besides reactivity difference, another
important factor that affects the stoichiometric modiﬁcation of a speciﬁc amino acid
residue is the reagent-to-substrate ratio used during modiﬁcation reactions. A low ratio
may cause under-modiﬁcation, while a high ratio may cause over-modiﬁcation. As
concentrations of peptides/proteins are unknown in these applications, it is impossible to
maintain the same reagent-to-substrate ratio for both states (unless the peptide/protein
level at two different states is similar). Perturbation (disease)-caused changes in protein
expression level may accompany other changes, such as in pH, redox potential, etc. Care
should be taken during modiﬁcation of proteins/peptides in the perturbed state and in the

normal state so that these changes would not affect the stoichiometry of the reaction.

Results of the studies described above show that 2-BA is much more reactive than
2-BP, yet the desired result of complete mono-alkylation for both reactions can be

achieved by adjusting the reagent-to-thiol ratio as well as incubation time in each case.

3.2. Quantiﬁcation Analysis of Cysteine-Containing Peptides

Two peptides, PHCKRM (expected MH+: m/z 770.4 Da) and DRVYIHPCHLL-
YYS (expected MH+: m/z l777.9 Da) were combined at a l: 3 molar ratio. An aliquot of
the peptide mixture was modiﬁed with 2-BA, and an equal aliquot was modiﬁed with 2-
BP, with aforementioned protocols. The two differently modiﬁed aliquots were combined
and analyzed by MALDI-MS. For peptide PHCKRM, the peak intensity ratio of the two
differently-modiﬁed products (2-BA modiﬁed product with a calculated monoisotopic

mass for MH+ = 827.37 Da and 2-BP modiﬁed product with a calculated monoisotopic

7O

mass for MH+ = m/z 841.37 Da) was 1:1, as expected (shown in Figure 3. 5). MS peaks
representing two modiﬁcation products of peptide DRVYIHPCHLLYYS (2-BA modiﬁed
product with a calculated monoisotopic mass for MH+ = m/z 1834.9 Da and 2-BP
modiﬁed product with a calculated monoisotopic mass for MH+ = m/z 1848.9 Da) had an
expected intensity ratio of 1:1 as well (shown in Figure 3. 5). Though the intensities of
the two peaks representing two peptides differ by a factor of 10, the intensity ratio (1:1)
between the differently-alkylated products correctly represented the concentration ratio in
the two aliquots (1:1) in both peptides. Note that the two peaks near rn/z 1900 in the mass
spectrum of DRVYIHPCHLLYYS alkylation products (marked with an asterisk in Figure
3. 5) do not represent over-alkylation products. These two peaks have a mass shift of
+119-Da and +133-Da mass shift from that of the protonated intact peptide, respectively.
As the intensity ratio of these two peaks is 1:1 as well, these two peaks are most likely
the 2-BA and 2-BP alkylation products of a modiﬁed peptide that has a +62-Da mass
shift from the intact peptide DRVYIHPCHLLYYS. The unknown +62-Da mass shift
may arise from single or multiple modiﬁcation events, possibly due to prolonged

exposure in air at room temperature.

71

 

 

 

 

 

 

 

 

 

 

 

 

14000 7 peptidezDRVYlHPCHLLYYSﬂ N
12000 n 1403
Z >
10000 ‘ peptidezPHCKRM 1L
. _L

_ 8000 “ 14Da
n: <

6000 ~

(.112. r“ .1“ 4..“ 1L. 4:.
4000 -
2000 — \ / ﬂ *
O JMJ1¥1 ﬁn“. A...“ A“. _- i“ A L A?) A i _ 111-1
500 700 900 1100 1300 1500 1700 1900
mlz

Figure 3. 5. Mass spectrum of 2-BA and 2-BP modiﬁed peptides (PHCKRM and
DRVYIHPCHLLYYS) at two different states (concentration ratio 1:1).

Peptide CQDSETRTF Y (calculated monoisotopic mass of MH)r = 1249.5 Da) was
subjected to modiﬁcation with 2-BA and 2-BP with the above-mentioned protocol to
assess the dynamic range and standard deviation of this method. After quenching of the
reaction, the two modiﬁed products were mixed to create nine different concentration
ratios at 10:1, 5:1, 5:2, 5:3, 1:1, 3:5, 2:5, 1:5 and 1:10. The nine mixtures were analyzed
by MALDI-TOF-MS. In Figure 3. 6, the peak intensity ratios corresponded closely to the
peptide concentration ratios in each of the standard mixtures of preformed alkylated

peptides.

72

10:1 5:] 5:2

I .
1.11. H 11

 

 

 

5:3 1:1 3:5
.11. ,, -11-.--1l.
2:5 1:5 1:10

“m ll . 1 r # ‘LJ‘ T , r ‘7‘ 44“

Figure 3. 6. MALDI mass spectra from quantiﬁcation analysis of a series of standard
solutions containing various ratios (10:1, 5:1, 5:2, 5:3, 1:1, 3:5, 2:5, 1:5 and 1:10) of
preformed alkylation products (2-BA and 2-BP) of peptide CQDSETRTFY. Each
spectrum was averaged over 100 laser shots. The mono-isotopic peaks for the two
modiﬁed products have masses of 1306.4 Da (2-BA) and 1320.4 Da (2-BP).

In the ‘thin layer’ MALDI sample preparation method, the peptide and matrix co-
crystallize almost instantaneously (<10 seconds) after being spotted on to the probe. As
the co-crystallization is homogeneous, laser shots at any position within the sample well
generate a good and reproducible signal. For the experiments described here, we
randomly chose 5 spots to analyze. The spectra acquired at different spots within one
sample well have different base peak intensity and signal to noise ratio for the base peak.
There was a 5-fold difference between the highest value to the lowest value for both peak
intensity and signal to noise ratio in different spectra acquired across the sample well.
However, we noticed that the deviation of the individual peak intensity ratio of the

differentially modiﬁed peptides from the mean does not correlate to either peak intensity

or signal-to-noise ratio in that spectrum. With repetitive measurements, a standard

73

deviation can be measured (results in shown in Table 3. 1). The intensity ratio of the two
differently-modiﬁed products is plotted against the concentration ratio of peptides in the
standard mixtures, as shown in Figure 3. 7. A standard curve was generated from linear
regression. We only graphed the dynamic range from 1:5 to 5:1 in Figure 3. 7, as 1:10

and 10:1 points have relatively big deviations and standard deviations.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

concentration ratio intensity ratio standard relative . .
(B A/BP) experimental devnation standard devration
(BA/BP) (experimental) deviation

0.10 0.14 0.016 11.87% 38.41%
0.20 0.23 0.015 6.57% 17.46%
0.40 0.43 0.036 8.50% 6.85%
0.60 0.62 0.029 4.69% 3.33%
1.00 1.07 0.080 7.50% 6.65%
1.67 1.76 0.032 1.82% 5.43%
2.50 2.53 0.251 9.91% 1.30%
5.00 4.86 0.413 8.49% -2.79%
10.00 8.12 0.584 7.20% -18.84%

 

Table 3. 1. Precision and accuracy of the assay of peptide CQDSETRTFY, over a
dynamic range between 1:10 and. 10:1. (5 measurements)

74

6.0 4

y = 0.9659x + 0.0742
R2 = 0.9992

5.0 1
4.0 —

3.0 a

intensity ratio

2.0 d

1.0 a

 

 

0.0 ,

0.0 1.0 2.0 3.0 4.0 5.0 6.0
concentration ratio

 

Figure 3. 7. Standard curve of quantiﬁcation of peptide CQDSETRTFY, over a dynamic
range between 1:5 and 5:1 (5 measurements).

The same analysis was performed on two other peptides: (HCKFWW and
DRVYIHPCHLLYYS). Two standard curves plotting the peak intensity ratio vs. the
concentration ratio are shown in Figure 3. 8. All three curves showed good linearity,
revealing that peak intensity ratio from analysis by MALDI-MS can be used to represent
the concentration ratio of the modiﬁed peptides in standard solutions. However, as two
reactions may have different yields for particular peptides, the concentration ratio of the
modiﬁed peptides does not necessarily reﬂect the concentration ratio of the intact
peptides at two states accurately. The degree of deviation of the slope of the standard
curve from one reﬂects the degree of yield difference from the two modiﬁcation

reactions.

75

y=1.1358x+0.144

a] #:09974 81 y=1.07x+0.1282

R2 = 0.9962

mtensuty ratio
&.

intensity ratio
A

 

 

 

o I I I I 0 I I I I

0 2 4 _ . 6 8 0 2 4 6
concentration ratio concentration ratio

 

Figure 3. 8. Standard curve for quantiﬁcation of peptide HCKFWW and peptide
DRVYIHPCHLLYYS, over a dynamic range between 1:5 and 5:1. The ratio is the
intensity of the peak for the 2-BA modiﬁed product over that for the 2-BP modiﬁed
product.

3.3. Estimation of the Minimal Error and Standard Deviation

We used the 13 C isotope peak intensity distribution for a 2-BA modiﬁed peptide
(DRVYIHPCHLLYYS) to estimate the minimal error and standard deviation of this
quantiﬁcation method. In this analysis, we focus on the ratio of peak height for the
monoisotopic form of the peptide and that of the same peptide with n 13 C carbons (n =
1~4). The concentration (abundance) ratio of the mono-isotopic species and those with n
13 C carbons should be a constant, and can be estimated from the natural abundance of ‘3 C
and the number carbons in the peptide. Because there is no modiﬁcation reaction
involved in this estimation and because ionization efﬁciency difference is minimal

between these species, the error and standard deviation from this measurement should

76

represent a best case that we can accomplish when performing homologous alkylation-
based quantiﬁcation.

Three spectra from the quantiﬁcation analysis, with known ratio of 5:3, 5:2 and
5:1 (ratio of 2-BA product to 2-BP product) of peptides at two different states, were
shown in Figure 3. 9 to demonstrate the deviation of the expected isotopic ratio and the
measured isotopic ratios. We shall focus on the isotopic peak intensity ratio among peaks
representing a peptide, but not focus on the intensity ratio between the two peaks that
represent the two differently-modiﬁed peptides. All spectra were acquired by averaging
the accumulated ion current from 100 laser shots. The insert spectrum is a computer-
generated spectrum with the isotopic distribution calculated from the elemental
composition (C35H122N220228) of the 2-BA modiﬁed peptide DRVYIHPCHLLYYS. The
isotopic distribution of 2-BP modiﬁed peptide is not shown here. However, it should be
very similar to that of the 2-BA modiﬁed peptide, as the contribution from a ‘CHZ’ to the
isotopic distribution of all other atoms is minimal. In Figure 3. 9, the isotopic cluster (a)
has a similar distribution to the one in the insert spectrum. Three peak clusters labeled (b)
vary from the expected distribution in that the intensity of the monoisotopic peak exceeds
that of the peptide with one l3C. Peak cluster (c) shows a big deviation for the 2-‘3C

peptide.

Accuracy and precision was assessed by comparing the experimental isotopic
distribution of the 2-BA modiﬁed peptide from ten measurements with the calculated
isotopic distribution (results shown in Table 3. 2). The concentration ratio of mono-
isotopic and multiple-13 C should be a constant across the spot on the MALDI plate. The

error and standard deviation are large because the relatively small number of ions

77

produced in each laser shot lead to poor ion-counting statistics and the isotopic ratio in
this small population may not represent the bulk isotopic ratio very well. Whatever the
reason might be, this error and standard deviation from isotopic ratio analysis represents a

best case that can be achieved for homologous quantiﬁcation analysis.

 

 

 

 

 

ex erimental standard relative

expected ratio p . . . standard deviation
ratio devnatlon . .

devuahon
C1/C0 1.067 0.982 0.061 6.25% -8.02%
CZ/CO 0.653 0.608 0.035 5.75% -6.86%
C3/C0 0.291 0.275 0.025 9.18% -5.68%
04/00 0.104 0.118 0.010 8.21% 13.50%

 

 

 

 

 

 

 

 

Table 3. 2. Precision and accuracy of the experimental ratio of the intensity of the
monoisotopic peak vs. that of the peaks with 1~4 (13C) atoms in the 2-BA modiﬁed
peptide. CO represents the monoisotopic peak while C 1~4 represents the peptide with 1~4
3 C atoms. The expected ratio was calculated from the elemental composition of the
product. (10 measurements)

78

100.00% 1
80.00% 1
60.00% 1
40.00% -
20.00% 1

intensity

 

 

 

5:3

3

.2 A L
I

 

0.00%
1830

100.00% 1
80.00% 1
60.00% 1
40.00% 1
20.00% 1

intensity

 

1835

A‘

 

ﬂ

1840 1845 1850 1855 1860 1865

mlz

j

5:2

 

 

0.00%
1830

100.00% 1
80.00% 1
60.00% 1
40.00% «
20.00% «

intensity

1835

.“

 

b
L -1 I- U111“ ﬂ
1840 1845

1850 1855 1 860 1 865

mlz

5:1

C

 

 

 

0. 00%
1 830

Figure 3. 9. Isotopic distribution variance of the product peaks after 2-BA and 2-BP
modiﬁcation of peptide DRVYIHPCHLLYYS in three experiments (5:3, 5:2 and 5:1).
The insert is a computer-generated distribution representing the isotopic variants for the
elemental composition of the 2-BA modiﬁed product (from http://prospector.ucsf.edu).
Peak (a) shows very similar distribution pattern. The three isotopic clusters labeled with
(b) show the intensity of the monoisotopic peak higher than that of the mono-13C peak.
The isotopic cluster (c) shows a bigger deviation from the expected distribution,
especially for the bis-13C peak. All spectra were the average of the accumulation of 100

shots.

1 835

., -1111..-
1840 1845 1850 1855 1860 1865
mlz

79

3.4. Using Homologous Reagent to Quantify Disulfide Isomers

Homologous alkylating reagent pairs may be used to quantify simple mixtures of
disulﬁde isomers. The analytical strategy is illustrated in Scheme 3. 2, using an example
of the two disulﬁde isomers identiﬁed from CN-induced mass mapping of oxidized
FBPase. Structure 1 has a sole disulﬁde bond between CyslSS and Cysl74; structure 2
has a sole disulﬁde bond between Cys] 74 and Cysl79. Assume the concentration ratio of
structure 1 and structure 2 in oxidized FBPase is 2:1. On the left, the oxidized enzyme is
directly modiﬁed with tagA (2-BA), as Cysl 55 is only free and available for alkylation in
structure 2, one of the three molecules are labeled with tagA. On the right, the same
aliquot is totally reduced to generate a second state. All cysteines are free and available
for alkylation after total reduction. During modiﬁcation with tagB (2-BP), Cy3155 in all
three molecules is labeled. Then, the two aliquots are combined and proteolytically
digested. Analysis of the digestion mixture would focus on the differentially-modiﬁed
peptide containing Cys155. In this example, the tagA-modiﬁed peptide and tagB-
modiﬁed peptides would have an intensity ratio of 1:3. In general, if the intensity ratio of
the tagA-modiﬁed CyslSS peptide and tagB-modiﬁed CyslSS is X, the concentration

ratio of structure 1 and structure 2 can be expressed as (l-X)/X.

80

Structure 1 Structure 2
SCN SCN SCN SCN SCN SCN SCN SCN iCN iCN

L__J |_J

1/. /| l/l /. 3/| :B/|
155 155 reduction 155

l/l /' l/‘ /| _, 3/' :B/|
155 155 155

2 A 2 3 B
L 155 l/ l/ 155 l/ l/ 155 l/
\ combine

155
155

‘ /l/ Vl/ "L
155 '3
f‘l "

._.E_A
155 —’

 

 

 

 

 

 

3—578/1/

Scheme 3. 2. Analytical strategy to quantify relative concentration ratio of two disulﬁde
isomers in oxidized FBPase using homologous alkylating regent pair.

81

 

4. Conclusion

For the relative quantiﬁcation of cysteine-containing peptides at two different
states to be successful, the modiﬁcation reaction needs to proceed to completion (14).
Cysteine-speciﬁc alkylating reagents, such as 2-BP, 2-BA, iodoacetamide and ICAT,
may incompletely modify cysteine residues as well as overly modify other non-cysteine
residues, such as the N-terminal amino group, the amino group of a lysine residue, or the
side chains of histidine, methionine and tryptophan. After adjusting reaction conditions,
such as reaction time and reagent-to-substrate ratio, stoichiometric alkylation at cysteines
is achieved for the 2-BA and 2-BP reagents.

The lack of knowledge and control over the reagent-to-substrate (peptide or
protein) ratio, which may lead to over-modiﬁcation or under-modiﬁcation, is recognized
as a major threat to the accuracy and the dynamic range of this homologous alkylation-
based quantiﬁcation methodology, as well as to other similar quantiﬁcation technologies
such as ICAT-based methods and MCAT-based methods.

The quantiﬁcation at the peptide level with the homologous reagent pair (2-BA
and 2-BP) was successful in that the peptide concentration ratio at two states can be
represented by the peak intensity ratio of the two differently alkylated peptides. An
assessment of the minimal error and standard deviation was made by studying the
isotopic distribution of a peptide. An analytical strategy based on this homologous
alkylating reagent pair method is proposed to analyze the relative concentration ratio of

disulﬁde isomers.

82

References:

l.

10.

11.

12.

13.

14.

Roth, F. P., Hughes, J. D., Estep, P. W., and Church, G. M. (1998) Nat Biotechnol
16, 939-945

DeRisi, J. L., Iyer, V. R., and Brown, P. O. (1997) Science 278, 680-686

Link, A. J ., Hays, L. G., Carmack, E. B., and Yates, J. R. (1997) Electrophoresis
18,1314-1334

Garrels, J. I., McLaughlin, C. 8., Warner, J. R., Futcher, B., Latter, G. I.,
Kobayashi, R., Schwender, B., Volpe, T., Anderson, D. S., MesquitaFuentes, R.,
and Payne, W. E. (1997) Electrophoresis 18, 1347-1360

Boucherie, H., Sagliocco, F., Joubert, R., Maillet, 1., Labarre, J ., and Perrot, M.
(1996) Electrophoresis 17, 1683-1699

Cagney, G., and Emili, A. (2002) Nat Biotechnol 20, 163-170

Gygi, S. P., Rist, B., Gerber, S. A., Turecek, F., Gelb, M. H., and Aebersold, R.
(1999) Nat Biotechnol 17, 994-999

Cadene, M., and Chait, B. T. (2000) Anal Chem 72, 5655-5658

Gobom, J ., and Nordhoff, E. (2002) Mass Spectrometry and Hyphenated
Techniques in Neuropeptide Research, 415-429

Gardner, J. A., and Matthews, K. S. (1987) Anal Biochem 167, 140-144

Uchida, K., Tanaka, K., and Hiratsuka, T. (1972) Biochim Biophys Acta 256, 132-
141

Boja, E. S., and Fales, H. M. (2001) Anal Chem 73, 3576-3582
Lapko, V. N., Smith, D. L., and Smith, J. B. (2000) J Mass Spectrom 35, 572-575

Smolka, M. B., Zhou, H., Purkayastha, S., and Aebersold, R. (2001) Anal
Biochem 297, 25-31

83

 

CHAPTER 4
‘SIGNATURE SETS (Si)’, A MINIMAL POSITIVE SIGNATURE FEATURE FOR
IDENTIFYING PROTEIN DISULFIDE STRUCTURES WITH CYANYLATION

(CN)-BASED MASS MAPPING METHODS

1. Introduction

The conventional methodology for characterizing disulﬁde structures relies on a
combination of Edman sequencing, proteolytic digestion, and mass mapping (1,2). The
drawbacks of these conventional methods include the inability to analyze proteins with
adjacent cysteines and a risk of forming artifacts due to disulﬁde scrambling during the
alkaline proteolysis. The cyanylation (CN)-based mass mapping methodology avoids the
problem of disulﬁde scrambling by covalently modifying the free sulﬂiydryls at low pH
(3); furthermore, this methodology is amenable to the analytical challenge presented by
proteins containing closely spaced cysteines as demonstrated in the analysis of cystinyl
protein folding intermediates (4), and of a tightly knotted protein containing three
adjacent cysteines among those forming four disulﬁde bonds (5).

One potential problem of using the CN-based method is that some CN-induced
cleavage fragments may be undetected, partly because cleavage yield varies among
different local sequences (6), and partly because matrix ions may interfere with detection
of a peptide ion during the MALDI process. We introduce the ‘signature set’ (Si) concept
to alleviate this concern. Speciﬁcally, we show that the correct disulﬁde structure can be

derived as long as a small, critical fraction of CN-induced cleavage fragments is

84

obtained. Furthermore, we analyze the structure of signature sets in order to derive the
optimal conditions for producing them. For example, we observe the surprising fact that
signature sets typically do contain fragments from doubly reduced isoforms, so it is
advisable to adjust conditions so that both singly and doubly reduced isoforms will be

generated.

2. Development of the Concept of Signature Sets

2.1. Disulfide Structure and Fragment Set

For a cystinyl protein with 2n cysteines, when all the cysteines are involved in
disulﬁde bonds, there are (2n-1)*(2n-3)*...*3*1 possible, isomeric disulﬁde
structures(7). The number of CN-induced cleavage fragments has an upper bound of
C22,,+2 = (n+1)*(2n+1), a combination of picking two points (two ends of a cleavage
fragment) out of 2n+2 (number of cysteines + 2 termini of the protein) possible points.
Note that the number of disulﬁde structures in Table 4. 1 increases at a faster-than-
exponential rate in the number of disulﬁde bonds (n! grows faster than a".), while the

number of cleavage fragments increases in a polynomial fashion in the number of

disulﬁde bonds.

85

number of 1 2 3

disulﬁde bonds 4 5 5 7 8 9 10

number of 3

disulﬁde structures 15 105 945 10,395 135,135 2,027,025 34,459,425 654,729,075

number of
possible CN- 6
induced cleavage
fragments

15 28 45 66 91 120 153 190 231

Table 4. 1. Number of disulﬁde structures and number of CN-induced cleavage
fragments as a function of number of disulﬁde bonds.

In Scheme 4. 1, ‘N’ represents the N-terminus of the protein, ‘C’ represents the C-
terminus, and ‘1’ represents a token cysteine residue. This new symbolism is proposed
because it provides an efﬁcient shorthand code to represent various CN-induced cleavage
fragments for the computational considerations described in this chapter. Upon CN-
induced cleavage, the structure represented at the left in Scheme 4. 1 produces two
fragments: [N,1] and [LC]: [N,l] represents the N-terminal fragment, which consists of
all residues from the N-terminus of the protein up to, but not including the ﬁrst (in this
case, the only) cysteine; [1,C] represents the second fragment consisting of the modiﬁed
cysteine, in the form of an iminothiazolidine derivative, and all remaining residues

through the C-terminus of the protein.

86

 

 

 

 

 

N N N
[NJ]
C an lation CN-induced
y y Cleavage 1
1 "SH > 1 — SCN g
CDAP, NH3 1
PH=3 [LC]
C C C

 

Scheme 4. 1. Symbolism for cleavage products resulting from cleavage of a cyanylated
cysteine-containing peptide.

Whereas the symbolism introduced in Scheme 4. 1 is attractive from the
standpoint of computational coding, it does not readily represent the well-documented
chemistry involved in the cyanylation-based mass mapping methodology (16). For
example, the numeral ‘1’ appears twice in Scheme 4. 1 to code two different fragments.
When a numeral (corresponding to the relative position of a given cysteine from the N-
terminus) is the ﬁrst entry in a set of two coordinates that code a given fragment, such as
‘1’ in the case of [1,C], that numeral represents the modiﬁed cysteine residue that deﬁnes
the N-terminus of that CN-induced cleavage fragment. When a numeral is the second
deﬁning coordinate for a given fragment, such as ‘1’ in the case of [N,1], that numeral
refers to the amino acid residue on the N-terminal side of the designated cysteine (#1 in
this case); thus, the residue preceding the designated cysteine becomes the C-terminus of
the given CN-induced cleavage fragment. When applied to a protein sequence, this
generic representation can be decoded to the amino acid residue numbers (see hEGF
example in this chapter). In general, when X and Y are two cysteines in the protein
sequence (X being the one that is closer to the N-terminus of the protein), fragment [X,

Y] represents the amino acid sequence of the CN-induced cleavage fragment starting with

87

the derivatized CysX residue and extending to the residue on the N-terrninal side of Cst
(but not including Cst). This generic representation of CN-induced fragments is used

throughout this chapter.

There are three possible disulﬁde structures for a generic 2-disulﬁde bond protein,
all shown in Scheme 4. 2. Disulﬁde structure A (with two disulﬁde bonds: Cysl-Cys3
and CysZ-Cys4), upon partial reduction and cyanylation, generates three isoforms: two
singly reduced isoforms and one doubly/totally reduced isoform. One of the two singly
reduced isoforms in which the disulﬁde bond between Cys2 and Cys4 is reduced and
cyanylated, upon CN-induced cleavage, will generate three fragments: [N,2], [2,4], [4,C];
the other possible singly reduced isoform, in which the disulﬁde bond between Cysl and
Cys3 is reduced and cyanylated, will generate three other fragments: [N,1], [1,3] and
[3,C]. Similarly, the CN-induced cleavage of the doubly/totally reduced isoform

generates ﬁve fragments: [N,1], [1,2], [2,3], [3,4] and [4,C].

A fragment set associated with a disulﬁde structure is deﬁned as a set containing
CN-induced cleavage fragments, with replicated fragments only appearing once, from all
possible partially-/totally reduced isoforms.. The fragment set associated with disulﬁde
structure A in Scheme 4. 2 contains 9 fragments: [N,1], [1,2], [2,3], [3,4], [4,C], [N,2],
[2,4], [1,3] and [3,C]. Note that fragment [4,C] may be derived from a singly reduced
isoform, as well as from a doubly reduced isoform. Such redundant fragments are only
reported once in a fragment set. Mass spectrometry, a tool to determine the identity of a
fragment by means of measuring its mass, is incapable of differentiating the pathways

from which a fragment is generated. The fragment sets for structures associated with

88

disulﬁde structure B and disulﬁde structure C can be generated in a similar fashion and

are listed in Scheme 4. 2.

Fragment Set:

A [NJ] [12] [2,3] [3,4] [4,C] [N,Z]
' [2,4] [1,3] [3,C]

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

N l 2 3 4 C
Fragment Set:
B. [NJ] [1.2] [2,3] [3,4] [4,C] [N,3]
[2,C1
N l 2 3 4 C
Fragment Set:
0 [N,11[1,2][2,3l [3,4] [4,C] [1,41
[N2] [3,Cl
N l 2 3 4 C

Scheme 4. 2. Three possible disulﬁde structures for a two-disulﬁde protein and the
corresponding fragment sets.

2.2. Fragments from CN-induced Cleavage of Singly and Doubly Reduced Isoforms
Represent the Entire Fragment Set

During partial reduction of a protein containing more than four disulﬁde bonds,
triply reduced isoforms may be generated. As reduction and cyanylation of one disulﬁde
bond generates two cleavage sites along the poly-peptide backbone, a triply reduced

isoform has 6 cleavage sites and may generate 7 CN-induced cleavage fragments

89

4 —5
CN-induced
SCN SCN SCN SCN SCN SCN Cleavage I -—2
_, 5 -—6
N 1 2 3 4 5 6 C 2 _..3
6 —C

Scheme 4. 3. CN-induced cleavage fragments from a triply reduced/cyanylated isoform
of a multi-cystinyl protein (n23). All the seven fragments can be generated from the CN-
induced cleavage of a singly or a doubly reduced isoform from the same protein.

The structure on the left in Scheme 4. 3 represents a triply reduced and
cyanylated isoform of a multi-cystinyl protein (11 2 4). Note that the residual n-3
disulﬁde bonds that are not reduced during partial reduction are not shown in this
scheme. The six dots along the polypeptide backbone represent the six cysteines that are
reduced and cyanylated; these six cysteines had been involved in three disulﬁde bonds in
the original structure. Upon cleavage and total reduction, this isoform generates 7
fragments: [N,1], [1,2], [2,3], [3,4], [4,5], [5,6] and [6,C]. These seven fragments can
all be generated from the CN-induced cleavage of a singly or a doubly reduced and
isoform from the same structure: Fragment [N ,1] can be generated from the cleavage
reaction of a reduced isoform in which the disulﬁde bond involving Cysl is reduced. By
the same reasoning, fragment [6,C] can also be generated from another singly reduced
isoform in which the disulﬁde bond involving Cys6 is reduced. Fragment [1,2] can be
generated from either a singly reduced isoform or a doubly reduced isoform depending
on how Cysl and Cys2 are connected: if Cysl and CysZ are connected to each other to
form a disulﬁde bond, fragment [1,2] can be generated from a singly reduced isoform

when disulﬁde bond between Cysl and Cys2 is reduced; if Cysl and Cy52 are connected

9O

to two of the rest of the four cysteines (Cys3~Cys6) to form two disulﬁde bonds,
fragment [1 ,2] can be generated from a doubly reduced isoform when those two disulﬁde
bonds are reduced. By the same reasoning, fragments [2,3], [3,4], [4,5], [5,6] can all be
generated from the CN-induced cleavage reactions of either a singly or a doubly reduced
isoform. Thus, we proved that all fragments from the cleavage of a triply reduced isoform
can all be generated from singly/doubly reduced isoforms. The conclusion is applicable
to more highly-reduced isoforms by the same reasoning. This theoretical insight provides
a guidance for the partial reduction experimental procedure: it is not advisable to overly
reduce the cystinyl protein during partial reduction to generate mostly triply or more
highly reduced isoforms, as the CN-induced cleavage fragments from these isoforms
would provide no more information regarding to the disulﬁde structure than those from

singly/doubly reduced isoforms.

2.3. Differential Sets (D(i, j)) for a Disulﬁde Structure (Fi)

In a comparison between any two fragment sets (F g and Fj), there are some
fragments that are common to both structures. Importantly, there are fragments that are
unique to only one structure. We deﬁne a subset of the original fragment set (Fi) that
contains all the fragments that are present in F3 but absent in Fj, the differential set D“, j) .
Similarly, another differential set Do, 3) contains all the fragments that are present in
fragment set Fj but absent in fragment set Fi. In a particular pair-wise comparison
between two fragment sets (Fi and F] )generated from two isomeric 2-disulﬁde proteins,
as illustrated in Scheme 4. 4, D“, j) contains four distinguishing fragments, while DU. 0

contains two fragments.

91

 

 

 

 

 

 

A.
N 1 2 3 4 C N 1 2 3 4 C
Fragment Set i: Fragment Set j:
[NJ] [12] [2,3] [3,4] [4,C] [NJ] [2,4] [NJ] [1,2] [2,3] [3,4] [4,C] [N,3] [2,C1

[1,3] [3,C]

 
    

[NJ] [1.2]
[2,3] [3,4]
[4,C]

    
  
 

Differential Set DUJ) Differential Set 004)

Common Fragments

Scheme 4. 4. Two differential sets (D(iJ) and D0,], ) generated in the comparison between
the fragment sets of two isomeric disulﬁde structures for a 2-disulﬁde protein.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

d2
d1
Disulﬁde .. dr ..
Structure i l l
N x Y C
d, i- i. _______
Disulﬁde d! i ........ 1 l
Stwcturej -- 9.1117199. ' 1
'l ‘ ”'l T I
N X Y C

Scheme 4. 5. ‘df, in disulﬁde structure i represents the ﬁrst disulﬁde bond from the N-
tenninus that connects differently between the two distinct disulﬁde structures.

We observed that regardless of how many disulﬁde bonds are there in a cystinyl

protein, the differential set D“, j) in any pair-wise comparison is always populated. Two

92

isomeric, fully disulﬁde bonded structures, structure i and structure j as shown in Scheme
4. 5, have to have at least two different disulﬁde bonds. In disulﬁde structure i, d;
represents the ﬁrst disulﬁde bond numbered from the N-terminus that is absent in
disulﬁde structure j. X and Y are the two cysteine residues that join to form df. All the
cysteines to the N-terminal side of X have the same connectivity in both structure i and
structure j. Because none of the aforementioned cysteines is connected to either X or Y in
structure i, neither are those cysteines in disulﬁde structure j connected to either X or Y.
In disulﬁde structure j, either X or Y forms a disulﬁde bond with another cysteine in the
region either between X and Y (region 1) or to the C-terrninal side of Y (region 2 in
Scheme 4. 5). Consider the cysteine that binds with Y to make disulﬁde structure j; if it is
a cysteine in region 1, fragment [X, Y], a cleavage product of a singly reduced isoform of
disulﬁde structure 1, becomes a fragment in the differential set D“, j). Disulﬁde structure j,
upon partial reduction, cyanylation, cleavage and total reduction, can never produce
fragment [X, Y]; any fragment terminated with Y for disulﬁde structure j will always be
shorter than fragment [X, Y]. The other possibility is for Y in structure j to bind with a
cysteine that is to the C-terminal side of Y (region 2). In that case, fragment [Y, C] from
structure i will be a member of in the differential set D“, j) in a pair-wise comparison with
disulﬁde structure j: for disulﬁde structure j to generate a CN-induced cleavage fragment
terminated with Y, the fragment will always be shorter than [Y, C]. In either case, the
differential set D“, j) contains at least one fragment. For a protein with n disulﬁde bonds,
there are [(2n-l)*(2n-3)*...*3*1] different structures when all cysteines are involved in
disulﬁde bonds. For a particular disulﬁde structure i, the differential set D“, j) (j =1 ~

[(2n-1)*(2n-3)*...*3* 1], j i i ) is always populated.

93

2.4. Generation of Signature Sets for a Particular Disulfide Structure

The detection of any one of the fragments in a differential set is sufﬁcient to
distinguish between two structures. However, to pinpoint one disulﬁde structure from
among all theoretical possibilities, a signature set (8,) can be constructed from
comparisons of the fragment set Fi of one disulﬁde structure with the fragment sets

characterizing all the other theoretically possible disulﬁde structures.

Scheme 4. 6 shows how the signature sets for a particular disulﬁde structure (i)
are generated. For a particular fragment set (i), there are [(2n-l)*(2n-3)*...*3*1]-1
differential sets, which are all subsets of fragment set (Fi), generated in the comparisons
between F; and all the other possible fragment sets from possible disulﬁde structures. We
can take one fragment from each differential set to form a new set. The new set, after
duplicated are removed, is deﬁned as a signature set (Si). There are [(2n-1)*(2n-
3)*...*3*1]-1 fragments in this set, with many replicates. It is noticed that this signature
set (83), is also a subset of fragment set (F3), in fact, it is a minimal subset of F] that leads
to an unequivocal identiﬁcation of disulﬁde structure i. As there are often multiple
fragments in a differential set (D(i, j)), there are usually multiple signature sets (8.)

associated with one disulﬁde structure.

94

 

 

Diﬂcrenth' 5“ #1 fragment set \

 

 

 

 

 

 

 

#2 fragment set

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

# i # i-l fragment set [(2n-l)*(2n-3)*...*3*l]-l
com arisons
fragment # i+l fragment set p
SCI ; :
ﬁ # x-l fragment set
S # x fragment set
Signature Set (Si) x = [(2n-l)*(2n-3)*..."‘3*l]-l

 

 

 

Scheme 4. 6. A signature set (8:) is formed by combining one fragment from each of the
[(2n- 1 )*(2n-3)* . . . *3 *1]-1 differential sets.

3. Evaluation of the Signature Sets Concept with Disulfide Mass Mapping Data
3.1. Output of Signature Sets for Generic Proteins with n Disulﬁde Bonds

Computer programs are written to generate disulﬁde structures, fragment sets,
differential sets and signature sets, sequentially, for generic cystinyl proteins with n
disulﬁde bonds. An example of the output for all isomeric disulﬁde structures is shown in
shorthand symbols in Table 4. 2 for a generic 3-disulﬁde protein. The 15 possible
structures for a three-disulﬁde protein are indexed from 0 to 14; likewise, for a four-
disulﬁde protein, the 105 possible disulﬁde structures are indexed from 0 to 104. Because
a cystinyl protein containing three disulﬁde bonds has only 15 isomeric structures, they
are listed in Table 4. 2 for illustrative purposes. Structure 0 in Table 4. 2 has three

disulﬁde bonds: (1, 2) (3, 4) (5, 6). Note that the symbol (1, 2) is different from [1, 2]: the

95

former symbol refers to a disulﬁde bond formed by linking Cysl and Cys2; the latter
symbol refers to a CN-induced cleavage fragment consisting of residues from Cysl up to

and including the residue that is on the N-terminal side of CysZ, as deﬁned earlier.

 

 

 

l

| .

| Strugtttlfnebindex Disulﬁde Structure

1 o . (1,2) (3,4) (5, 6)

’ 1 3 (1,2) (3,5) (4, 6)
2 I (1.2) (3,6) (4,5)

I 3 ' (1,3) (2,4) (5, 6)

1 4 l (1.3) (2.5) (4, 6)

l 5 (1,3) (2,6) (4, 5)
6 (1,4) (2,3) (5,6)
7 (1,4) (2, 5) (3,6)
8 (1,4) (2,6) (3, 5)
9 (1,5) (2,3) (4,6)
10 (1,5) (2,4) (3,6)
11 (1,5) (2,6) (3,4)
12 (1,6) (2, 3) (4,5)
13 (1,6) (2,4) (3, 5)
14 (1,6) (2,5) (3,4)

 

 

Table 4. 2. Indexing of all 15 possible isomeric disulﬁde structures for a 3-disulﬁde
protein containing 6 cysteines.

A second program is responsible to generate fragment set, differential sets and
signature sets for each disulﬁde structure. For a particular disulﬁde structure, the program
composes all possible singly reduced and doubly reduced isoforms. For each isoform, the
program performs the cleavage and total reduction operation in silico to generate three
CN-induced cleavage fragments from a singly reduced isoform and ﬁve fragments from a
doubly reduced isoform; it then arranges these fragments into a fragment set. Before

including a CN-induced cleavage fragment in a fragment set, the program ﬁrst

96

determines whether the fragment has already appeared in the fragment set. Thus, all
fragment sets are ‘non-redundant’; i.e., they contain no replicates. From previous
discussion, we showed that triply or more highly reduced isoforms do not contribute new
information to a fragment set; thus, they are not considered by the program. A differential
set (D(i,], ) is generated by comparing fragment set i and fragment set j and contains all
the fragments that are present in fragment set i but not in fragment set j. Signature set (8;)
is then generated by taking one fragment from each of the Di,- ( j from 1 to [(2n-1)*(2n-
3)*...*3*l] and j at i ) and combine them together. Repetitive fragments in Si are only
recorded once. As there can be multiple fragments in Di], there are multiple signature sets
(8,) for each structure (i) and each one of these signature sets (Si) is equally sufﬁcient to
differentiate disulﬁde structure (1) from all the other possible disulﬁde structures. The
output of all theoretically possible disulﬁde structures for a generic 3- and 4-disulﬁde
protein (from program 1) and the signature sets associated with these disulﬁde structures
(from program 2) can be obtained from following URL:

http://www.bch.msu.edu/facilities/massspec/disulﬁde/index.html

3.2. Human Epidermal Growth Factor (hEGF) and its Disulfide Isomers

6 14 20 31 33 42
NSDSE(IIPLSHDGYCLHDGVCMYIEALDKYACNCVVGYIGERIfQYRDLKWWELR

 

 

 

 

Scheme 4. 7. Primary structure and native disulﬁde linkages of human epidermal growth
factor (hEGF)

97

We re-examined data obtained from analysis of three isomeric forms of a cystinyl
protein containing three disulﬁde bonds using our partial reduction, CN-induced
cleavage, and mass mapping methodology. The proteins are human epidermal grth
factor (hEGF, structure shown in Scheme 4. 7) and two of its non-native three-disulﬁde
isomers, as produced during an oxidative refolding experiment with reduced hEGF and
separated by HPLC as illustrated in Figure 4. l (4). In the following, we coordinate the
computed signature sets (Si) with experimentally detected fragments to efﬁciently
identify the disulﬁde structures of the three isomeric species. In the 53-residue sequence
of hEGF, the six cysteines reside at positions 6, 14, 20, 31, 33 and 42. These cysteines in
the computer-generated disulﬁde structures, fragment sets, differential sets and signature
sets are enumerated ‘1-6’, according to their relative positions in the sequence. The
computer-shorthand representation of the 15 possible disulﬁde structures are listed and
indexed in Table 4. 2. The computer-shorthand representation of the CN-induced
cleavage fragments in Table 4. 3 needs to be carefully transformed into a chemical
representation for agreement with the mass spectral results of analysis(4). For example,
[N,3] corresponds to CN-induced cleavage at the N-terminal side of the 3rd cysteine
(residue 20 in hEGF), and thus, corresponds to a peptide containing all residues from the
N-terminus of hEGF up to residue 19. Fragment [2,5] corresponds to CN-induced
cleavages at the N-terminal side of Cys2 (residue 14) and of CysS (residue 33), and thus,
consists of residues from the iminothiazolidine-blocked residue 14 up to residue 32 in

hEGF .

98

Ctiiaggigigrgigt' Cleavage fragments, Observed Expected Fragirrr:ents
residue numbers in numbered byrelatlve masses masses Signature
hEGF Cys posutlons (Da) (Da) Sets
1-32 [N,5] 3692.9 3694.1 X
itz-6-19 [1,3] 1540.1 1541.7 X
itz-14-30 [2,4] 1969.2 1970.3 X
itz-20-53 [3,C] 4218.3 4218.0 X
[itz-3153,, [4,9] __72917__.1_ __ 2916.4 _ X
1-5 [N,1] 682.7 n.d.
1-13 [N,2] 1555.6 1554.6
itz-33-41 [5,6] 1020.1 1020.2
itz-42-53 [6,C] 1721.0 1722.0
itz-6-13 [1.2] 915.9 915.9
itz-14-19 [2,3] 667.7 n.d.
itz-20-30 [3.4] 1345.6 1345.0
itz-20-32 [3.5] 1562.8 n.d.
itz-31-32 [4,5] 260.2 n.d.

Table 4. 3. Expected fragments from CN-induced mass mapping of native hEGF and
those detected by MALDI-MS. The ﬁrst ﬁve fragments are constituents of four signature
sets (see Table 4. 4), while the other nine theoretically possible fragments do not
contribute in distinguishing the disulﬁde structure of the native protein from other
theoretically possible disulﬁde structures. (n.d. =not detected)

99

 

 

 

‘ Partial Reduction,

Cyanylation,
‘ Cleavage,
‘ Total reduction

N o 0 ® 0 itz-6+ IIIIIIIIII

[N,5], m/z 3694.] [1,3], m/z 1541.6 12 other fragments
¥ J

V
Signature Set 1

 

Scheme 4. 8. The two fragments, [N,5] and [1,3], in the ﬁrst signature set for native
hEGF are generated in the CN-induced mass mapping experiment, along with 12 other
theoretically possible fragments.

Native hEGF has a disulﬁde structure identical to structure 3 in Table 4. 2. From
computational analysis, there are four signature sets for structure 3, as listed in Table 4. 4,
two consisting of two fragments, and the other two consisting of three fragments.
Analysis of the experimental data show that peaks representing fragments in all four sets
were detected in the MS mapping. The generation of the two fragments in the ﬁrst
signature set from CN-induced mass mapping experiment as well as the shorthand
representation of these two fragments are shown in Scheme 4. 8. In the mass spectral
analysis of the mixture of CN-induced cleavage fragments, a peak at m/z 3692.9 was
observed and assigned to fragment [N,5], consisting residues from the N-terminus of
hEGF to the alanine at residue 31 ( calculated mass = 3694.1 Da); another peak at m/z
1540.1 was observed and correspondingly mapped to fragment [1,3], consisting of

residues from the iminothiazolidine derivative of Cys6 to Va119 ( calculated mass =

100

1541.6 Da). Detection of these two fragments satisﬁes one of the signature sets, and,
thus, constitutes identiﬁcation of hEGF as disulﬁde structure 3. Other peaks in the mass
spectrum also represent all the fragments in three other signature sets, thereby
corroborating the identiﬁcation of hEGF as structure 3. Whereas, it is possible for some
other disulﬁde structures to generate one of the two fragments listed in the ﬁrst two
signature sets in Table 4. 4, it is impossible for any one of the other 14 possible disulﬁde
structure to generate both fragments simultaneously. From observation of the mass
spectra, ten of the fourteen expected fragments in the fragment set for structure 3 were
identiﬁed by mass mapping, as shown in Table 4. 3. Three of the four undetected
fragments have a very low calculated mass. Possible reason is that their mass spectral
peaks may have been obscured by those for matrix ions during analysis by MALDI.
Ionization of the fourth fragment [3, 5], a larger cleavage fragment from a doubly
reduced isoform, might have been suppressed by other peptide components, causing the
fragment to be undetected. However, failure to detect these four peaks does not affect the
outcome of the disulﬁde structure determination. Signature Sets analysis shows the
intrinsic robustness of the cyanylation mapping approach: even when a majority of the
cleavage fragments are not detected, one can still unequivocally determine the disulﬁde

SII'UCIUI'C .

Similarly, when analyzing the experimental data on folding intermediate IIIB, we
detected all members of three of the ﬁve signature sets, each containing only two of the
twelve possible fragments (in Table 4. 4). In the analysis of the data on the folding

intermediate IIIA, we detected both fragments for the only signature set.

101

Naﬂve

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Figure 4. 1. HPLC chromatogram showing three fully oxidized isomers of hEGF after 48
hours of oxidative refolding. ‘Native’ represents the native hEGF, while ‘III-A’ and ‘III-

minutes

B’ represent the misfolded versions of the protein.

 

 

Signature sets for

Signature sets for

Signature sets for

 

 

 

 

structure 3 Detected structure 0 Detected structure 14 Detected
(native) (Ill-B) (lIl-A)
[N,5] [1,3] X [N,3] [N,5] X [1,6] [2,5] X
[N,5] [2,4] X [N,5] [2,5]
[1,3] [3,C] [4,C] X [N,5] [2,C] X
,1 [2,4] [3,C] [4,C] X [2,5] [2,C]
_ _‘_*_____ [2,0] [4,C] x

Table 4. 4. Signature sets for the three isomeric proteins resulting from the refolding of
hEGF. The sets marked with ‘X’ had all constituent fragments represented by peaks in

 

 

previously recorded mass spectra during disulﬁde mass mapping.

102

 

3.3. Further Evaluation of the Signature Sets with Ribonuclease A (RNase A)
Mapping Data

RNase A is consisted of 124 amino acid residues, 8 of which are cysteines that
link to one another to form of 4 disulﬁde bonds (disulﬁde structure shown in Scheme 4.
9). After partial reduction, cyanylation, cleavage and total reduction, 27 possible CN-
induced cleavage fragments can be generated in silico to form its theoretical fragment set.
The differential sets, generated by comparing this fragment set with the fragment sets of
the other 104 theoretically possible disulﬁde structures, contain from 2 to 15 fragments
(results not shown). With the aforementioned algorithm, the computer outputs 25
signature sets of native RNase A in Table 4. 5 (shown in Table 4. 5). Detection of the
combinations of fragments in any one of the 25 signature sets is sufficient to distinguish
the native disulﬁde structure of RNase A from the other 104 possible isomeric disulﬁde
structures. Mass spectra obtained during previous mass mapping of RNase A showed

three signature sets (marked with ‘X’ in Table 4. 5 ) were detected.

 

 

 

 

 

 

 

 

 

Scheme 4. 9. Representation of the native disulﬁde structure of RNase A: Eight cysteines
at residues 26, 40, 58, 65, 72, 84, 95 and 110 are numerated ‘1-8’ according to their
relative position to the N-terminus of the protein.

103

 

Signature Sets for RNase A Identiﬁed 1n

 

Experiments
[1,6] [2,6] [3,8] X
[1,6] [2,7] [3,6]
[1,6] [2,7] [3,7]
[1,6] [2,7] [3,8] X
[1,6] [3,7] [3,8]
[2,6] [2,7] [3,8] X

[2,7] [3,6] [3,8]
[0,3] [2,6] [2,7] [3,6]
[0,3] [2,6] [2,7] [3,7]
[0,3] [2,7] [3,6] [3,7]
[1,4] [2,6] [2,7] [3,6]
[1,4] [2,6] [2,7] [3,7]
[1,4] [2,7] [3,6] [3,7]
[1,6] [2,6] [2,7] [3,6]
[1,6] [2,6] [2,7] [3,7]
[1,6] [2,7] [3,6] [3,7]
[2,6] [2,7] [3,6] [3,8]
[2,6] [2,7] [3,6] [5,8]
[2,6] [2,7] [3,6] [6,9]
[2,6] [2,7] [3,7] [3,8]
[2,6] [2,7] [3,7] [5,8]
[2,6] [2,7] [3,7] [6,9]
[2,7] [3,6] [3,7] [3,8]
[2,7] [3,6] [3,7] [5,8]
[2,7u3,6] [3,7] [6,9]

NNNNNNI—‘u—th—ﬁh—ib—‘t—ﬁr—ﬁu—‘r—II—I
M-ﬁbJNh—OOooqog/‘gwwﬂo\OOONONMADJNH

Table 4. 5. Signature sets for RNase A; compositions for fragments are expressed in
computer-shorthand according to relative position of Cys residues as explained in text.

104

3.4. Characteristics of Signature Sets

An analysis of the signature sets from all 105 disulﬁde structures for a cystinyl
protein with eight cysteines shows that most of the CN-induced cleavage fragments in a
set are those that contain some internal cysteines. For example, most fragments in a
signature set contain at least two free internal cysteines. Another observation is that most
of the fragments are internal fragments, i.e., they do not contain either the N-terminus or
the C-terminus of the protein. These two observations are in agreement with the ‘rule-
out’ concept of the ‘negative signature mass algorithm’(8). Larger fragments with more
internal free cysteines rule out more invalid linkages; fragments not terminated by the N-
or C-terminus of the protein are capable of ruling out invalid linkages from both ends of
the cleavage fragment, as opposed to the N-or C-terrninal fragment, which can only rule
out invalid linkages from the opposing end of the original molecule. Both of these
observations are manifested in the signature sets for native RNase A, as shown in Table

4.5.

Another observation of the signature sets is that most often, a signature set
consists of a combination of fragments from singly and- doubly reduced isoforms of the
protein. For example, out of the 25 signature sets for RNase A, only one set consists of
fragments derived solely from a singly reduced isoform; in all of the other 24 sets, the
fragments are derived from both singly and doubly reduced isoforms. This observation
suggests that during the partial reduction, it behooves the analyst to adjust conditions so

that both singly and doubly reduced isoforms are generated.

105

4. Frequency Analysis of the Fragments from CN-induced Cleavage of a Four-
Disulfide Protein

A particular CN-induced cleavage fragment is characterized by its two ‘ends’,
either a modiﬁed (iminothiazolidine derivative) cysteine residue (or a residue to the N-
terrninal side of a cysteine) or the N- or C- terminus of the protein. For a protein with
four disulﬁde bonds, a total of 10 points (8 cysteines plus the N- and C- terminus) can
serve as these ‘ends’ for a CN-induced cleavage fragment. The number of all the
theoretically possible CN-induced cleavage fragments for such a protein is 45, derived
from a combination of selecting two points (‘ends’) out of the 10 available points. For a
randomly selected disulﬁde structure, we examined all four possible singly reduced
isoforms and the CN-induced cleavage fragments that could be derived from them;
cumulatively, 12 of the 45 possible fragments were derived from all the singly reduced
isoforms. We also considered all six possible doubly reduced isoforms and their CN-
cleavage products; cumulatively, 30 fragments were derived from the doubly reduced
isoforms of the particular disulﬁde structure. Then, we projected this list of 42 fragments,
some of them replicates, onto the list of the 45 theoretically possible fragments and
recorded the number of occurrences for each fragment. In analogous fashion, we can
assess the occurrences of each of the 45 theoretically possible fragments in cleavage
fragment sets for all the other 104 theoretically possible disulﬁde structures. Totally,
105*42 = 4410 occurrences are distributed among the 45 fragments.

If a fragment has a high number of occurrences, many different disulﬁde
structures are capable of generating that particular fragment, i.e., such a fragment would
not be especially useful in distinguishing between disulﬁde structures. Fragments with a

low occurrence rate, on the other hand, has more distinguishing power and are more

106

likely to validate a given disulﬁde structure. The distribution of occurrences among the
45 theoretically possible fragments is shown in Table 4. 6. Fragments [N,8], [LC] and
[N,C] never appear once; this is an expected result because such fragments derive from
either no or only one cleavage site, a condition that is incompatible with our partial
reduction methodology in which a minimum of two cysteines (cleavage sites) is
produced.

From Table 4. 6, we note that ‘longer’ fragments with more internal free cysteines
have a lower number of occurrences, and thus, are more informative in distinguishing
between disulﬁde structures. This observation is graphically illustrated in Figure 4. 2.
The fragments with fewer internal free cysteines occur more frequently than ones with
more free cysteines, a phenomenon more obvious for the fragments derived from doubly
reduced isoforms than from singly reduced isoforms.

Another observation of the results of frequency analysis shown in Table 4. 6 is
that a given internal fragment has a lower number occurrences than a terminal fragment
with the same number of free cysteines. For example, consider terminal fragments [N,4]
and [5,C]; each has three internal free cysteines and each has occurred 96 times. On the
other hand, fragments [1,5], [2,6], [3,7] and [4,8], each with three internal free cysteines,
have occurred 69 times, and are thus are more informative in distinguishing between
disulﬁde structures. These observations agree with those made from the signature sets

analysis.

107

 

 

Number of Number of Number of

Fragment Fragment Fragment

 

Occurrences Occurrences Occurrences
[N,1] 420 [1,8] 15 [4,5] 150
[N2] 270 [1 ,C] 0 [4,6] 105
[N,3] 165 [2,3] 150 [4,7] 69
[N,4] 96 [2,4] 105 [4,8] 42
[N,5] 54 [2,5] 69 [4,C] 54
[N6] 30 [2,6] 42 [5,6] 150
[NJ] 15 [2,7] 24 [5,7] 105
[N,8] 0 [2,8] 15 [5,8] 69
[N,C] 0 [2,0] 15 [5,C] 96
[1 ,2] 150 [3,4] 150 [6,7] 150
[1 ,3] 105 [3,5] 105 [6,8] 105
[1.4] 69 [3,6] 69 [6,C] 165
[1,5] 42 [3,7] 42 [7,8] 150
[1 ,6] 24 [3,8] 24 [7,C] 270
___l1_.7l__ 15_ [3.01 30 [8.01 420

 

 

Table 4. 6. Number of occurrences of fragments from CN-induced cleavage of singly and
doubly reduced isoforms of all 105 theoretically possible disulﬁde structures for a protein
containing four disulﬁde bonds.

108

 

 

 

 

 

 

 

 

2000
II)
§ 1600 I .
e OSingly reduced
3 1200 . ‘ IDoubly reduced
I ASingly/doubly reduced
‘3' 800 -
g A
g 400 1 I A
Z . 9 I ‘
0 . T I t I I

0 2 4 6 8

Number of internal cysteines in CN-induced
cleavage fragments

Figure 4. 2. The number of occurrences for a CN-induced cleavage fragment as a
function of its constituent free cysteines for a four-disulﬁde protein.

5. Homogeneity in Disulﬁde Structures of the Protein under Study

Similar analysis of the fragment sets generated from combinations of two
different disulﬁde structures of a protein with 3 disulﬁde bonds showed that the
distinguishing feature between the sets is lost in some cases. From 15 possible disulﬁde

structures for a protein with three disulﬁde bonds, 105 different heterogeneous
combinations of a pair of disulﬁde structures can be derived (Cf5 =105). We generate the

fragment sets for all the 105 structure pairs. Like a fragment set for an individual
disulﬁde structure, the fragment set for a disulﬁde structure pair does not contain
duplicates. Some fragment sets associated with the structure pairs, in at least one pair-
wise comparison with other fragment sets, do not have any fragments in a differential set.

In Scheme 4. 10, we generated the fragment sets from two pairs: pair A consisting

of disulﬁde structures (1, 2) (3, 4) (5, 6) and (l, 4) (2, 3) (5, 6); Pair B consisting of

109

disulﬁde structures (1, 2) (3, 4) (5, 6) and (1, 5) (2, 3) (4, 6). The fragment set of pair
A consists of 16 members: [0,1], [1,2], [2,7], [0,3], [3,4], [4,7], [0,5], [5,6], [6,7], [2,3],
[2,5], [4,5], [1,4], [0,2], [3,7] and [3,5]; the fragment set of pair B consists of 20
members: [0,1], [1,2], [2,7], [0,3], [3,4], [4,7], [0,5], [5,6], [6,7], [2,3], [2,5], [4,5], [1,5],
[5,7], [0,2], [3,7], [0,4], [4,6], [3,5] and [1,4]. We noticed that fragment set A is a subset
of fragment set B. In the comparison between these two sets, set A does not have any
fragments in its differential set (BM, 3)), while pair B has four fragments: [1,5], [5,7],
[0,4] and [4,6] in its differential set (D(A, 3)). For Pair A, because of the lack of at least
one CN-induced cleavage in this differential set, the signature set can not be constructed.
Pair B, however, has a populated differential set in every one of the comparisons with all

the other 104 sets. Signature sets for structure pair B can be constructed.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Scheme 4. 10. Pair A does not have any fragments in its differential set when its fragment
set is compared with that of Pair B. Pair B has at least one fragment in its differential set
when compared with the fragment sets from 104 other possible pairs.

110

 

The major limitation of the CN-induced disulﬁde mass mapping methodology is
the requirement for starting with one single, pure disulﬁde structure, free from other
molecules with isomeric disulﬁde structures. Of course, this is a requirement for all
methodologies used to solve disulﬁde structures because some conﬂicting evidence
(observed or not) will always be produced by isomeric disulﬁde structures using any
distinguishing technique. Two disulﬁde structures in certain combinations for a generic
3-disulﬁde protein, such as the two forming pair B in Scheme 4. 10, can be identiﬁed

simultaneously by CN-based mapping method without any separation of the two isomers.

6. Conclusions

Cyanylation-induced cleavage of the polypeptide backbone of cystinyl proteins
generates sequence-speciﬁc fragments that can be used to mass-map the linkages of
speciﬁc cysteines involved in a particular disulﬁde structure. A proof is given herein for
the existence of at least one unique CN-induced cleavage fragment in Dm), in a pair-wise
comparison between two fragment sets. Further, we have shown that certain
combinations of CN-induced fragments form a series of signature sets (8,), that uniquely
characterize a given disulﬁde structure. By inspection, we noted that signature sets
consist of fragments that contain internal cysteines within their sequences. We also note
that triply and more highly reduced isoforms of the cystinyl protein do not contribute
unique fragments to a fragment set; thus, it behooves the analyst to seek conditions that
promote formation of the singly and doubly reduced isoforms of the protein. Selected
generic signature sets for a three-disulﬁde cystinyl protein were transposed to the

sequence of hEGF to illustrate the capacity to distinguish the native protein and two mis-

111

folded isomers based on combinations of as few as three CN-induced cleavage fragments
as detected by mass spectrometry. The concept of signature sets provides the basis for
designing analytical methodology to unambiguously identify speciﬁc disulﬁde structures

for isolated cystinyl proteins.

Reference:

1. Li, J. H., Yen, T. Y., Allende, M. L., Joshi, R. K., Cai, J., Pierce, W. M.,
Jaskiewicz, E., Darling, D. S., Macher, B. A., and Young, W. W. (2000) J Biol
Chem 275, 41476-41486

2. Zhou, Z., and Smith, D. L. (1990) J Protein Chem 9, 523-532

3. Wu, J ., and Watson, J. T. (1997) Protein Sci 6, 391-398

4. Wu, J ., Yang, Y., and Watson, J. T. (1998) Protein Sci 7, 1017-1028

5. Qi, J., Wu, J ., Somkuti, G. A., and Watson, J. T. (2001) Biochemistry 40, 4531-
4538

6. Wu, J ., and Watson, 1. T. (1998) Anal Biochem 258, 268-276
7. Benham, C. J., and Jafri, M. S. (1993) Protein Sci 2, 41-54

8. Qi, J., Hang, D., Rupp, M., Tomg, E., Borges, C. R., Wu, W., and Watson, J. T.
(2003) Journal of the American Society for Mass Spectrometry 14, 1032-1038

112

 

APPENDIX
DISULFIDE INDEX, FRAGMENT SETS, SIGNATURE SETS FOR A GENERIC

3-DISULFIDE BOND PROTEIN

l. Disulfide Structure Index

Disulﬁde
Linkages ofthe 6 Cysteines, Numbered by their
Structure
Relative Position within Sequence

Number

0 (1,2) (3,4) (5,6)

1 (1,2) (3,5) (4,6)

2 (1,2) (3,6) (4, 5)

3 (1,3) (2,4) (5, 6)

4 (1,3) (2, 5) (4, 6)

5 (1,3) (2, 6) (4, 5)

6 (1,4) (2, 3) (5, 6)

7 (1,4) (2, 5) (3,6)

8 (1,4) (2,6) (3, 5)

9 (1,5) (2,3) (4,6)

10 (1,5) (2,4) (3,6)

11 (1,5) (2,6) (3,4)

12 (1,6) (2,3) (4, 5)

13 (1,6) (2,4) (3, 5)

14 (1,6) (2, 5) (3,4)

113

2. Fragment Sets

Fragment Set for Disulﬁde Structure 0:

[0,1] [1,2] [2,7] [0,3] [3,4] [4,7] [0,5] [5,6] [6,7] [2,3] [2,5] [4,5]

Fragment Set for Disulﬁde Structure 1:

[0,1] [1,2] [2,7] [0,3] [3,5] [5,7] [0,4] [4,6] [6,7] [2,3] [2,4] [3,4] [4,5] [5,6]

Fragment Set for Disulﬁde Structure 2:

[0,1] [1,2] [2,7] [0,3] [3,6] [6,7] [0,4] [4,5] [5,7] [2,3] [2,4] [3,4] [5,6]

Fragment Set for Disulﬁde Structure 3:

[0,1] [1,3] [3,7] [0,2] [2,4] [4,7] [0,5] [5,6] [6,7] [1,2] [2,3] [3,4] [3,5] [4,5]

Fragment Set for Disulﬁde Structure 4:

[0,1] [1,3] [3,7] [0,2] [2,5] [5,7] [0,4] [4,6] [6,7] [1,2] [2,3] [3,5] [3,4] [2,4] [4,5] [5,6]

Fragment Set for Disulﬁde Structure 5:

[0,1] [1,3] [3,7] [0,2] [2,6] [6,7] [0,4] [4,5] [5,7] [1,2] [2,3] [3,6] [3,4] [2,4] [5,6]

Fragment Set for Disulﬁde Structure 6:

[0,1] [1,4] [4,7] [0,2] [2,3] [3,7] [0,5] [5,6] [6,7] [1,2] [3,4] [4,5] [3,5]

114

Fragment Set for Disulﬁde Structure 7:
[0,1] [1,4] [4,7] [0,2] [2,5] [5,7] [0,3] [3,6] [6,7] [1,2] [2,4] [4,5] [1,3] [3,4] [4,6] [2,3]

[3,5] [5,6]

Fragment Set for Disulﬁde Structure 8:
[0,1] [1,4] [4,7] [0,2] [2,6] [6,7] [0,3] [3,5] [5,7] [1,2] [2,4] [4,6] [1,3] [3,4] [4,5] [2,3]

[5,61

Fragment Set for Disulﬁde Structure 9:

[0,1] [1,5] [5,7] [0,2] [2,3] [3,7] [0,4] [4,6] [6,7] [1,2] [3,5] [1,4] [4,5] [5,6] [3,4]

Fragment Set for Disulﬁde Structure 10:

[0,1] [1,5] [5,7] [0,2] [2,4] [4,7] [0,3] [3,6] [6,7] [1,2] [4,5] [1,3] [3,5] [5,6] [2,3] [3,4]

[4,61

Fragment Set for Disulﬁde Structure 11:

[0,1] [1,5] [5,7] [0,2] [2,6] [6,7] [0,3] [3,4] [4,7] [1,2] [2,5] [5,6] [1,3] [4,5] [2,3] [4,6]

Fragment Set for Disulﬁde Structure 12:

[0,1] [1,6] [6,7] [0,2] [2,3] [3,7] [0,4] [4,5] [5,7] [1,2] [3,6] [1,4] [5,6] [3,4]

Fragment Set for Disulﬁde Structure 13:

[0,1] [1,6] [6,7] [0,2] [2,4] [4,7] [0,3] [3,5] [5,7] [1,2] [4,6] [1,3] [5,6] [2,3] [3,4] [4,5]

115

Fragment Set for Disulﬁde Structure 14:

[0,1] [1,6] [6,7] [0,2] [2,5] [5,7] [0,3] [3,4] [4,7] [1,2] [5,6] [1,3] [4,6] [2,3] [4,5]

3. Signature Sets

Signature Sets for Structure 0:
[N3] [N,5]

[N5] [25]

[N5] [2,C]

[2,5] [2,C]

[2,C] [4,C]

Signature Sets for Structure 1:
[2,C] [3,5]

[2,C] [4,6]

[N3] [N41 [35]

[N3] [N4] [4,6]

[N,3] [2,C] [35}

[N3] [2,C] [4,6]

Signature Sets for Structure 2:

[2,C] [3,6]

116

[N3] [N4] [3,6]
[N,3] [2,C] [3,6]

Signature Sets for Structure 3:
[NS] [1.3]

[NS] [2.4]

[NS] [1,3] [3,C]

[N,5] [2,4] [3,C]

[1,3] [3,C] [4,C]

[2,4] [3,C] [4,C]

Signature Sets for Structure 4:
[N4] [2,5]

[2,5] [3,C]
[N,4] [1,3] [2,5]
[N,4] [1,3] [3,5]
[N,4] [1,3] 14,6]
INA] [2,4] [2,5]
[1,3] [2,5] [3,C]
[1,3] [3,C] [4,6]
[2,4] [2,5] [3,C]
[2,4] [3,C] [4,6]

[er1 [N,4] 12,4] [3,5]

117

[N,2] [N,4] [2.4] [4,6]
[N,4] [1.3] [2,4] [35]
[N4] [1.31124] [4.6]
[N,4] [1,3] [3.5] [3,C]
[N,4] [2.4] [2,5] [35}
[N4] [2.4] [2.5] [4,6]
[N,4] [2,4] [3,5] [3,C]
[N,4] [2.411343] [4,61
[1,3] [2,5] [3,5] [3,C]
[1,3] [3,5] [3,C] [4,6] -
[13113.5] [3,C] [5,C]
[2,4] [2,5] [3,5] [3,C]
[2,4] [3,5] [3,C] [4,6]

[2,4] [3,5] [3,C] [5,C]

Signature Sets for Structure 5:
[N,4] [2.6]

[2,6] [3,6]

[2,6] [3,C]

[N,4] [1,3] [3,6]

[1,3] [2,6] [3,6]

[1,3] [3,6] [3,C]

[2,4] [2,6] [3,6]

118

[2,4] [3,6] [3,C]

[N2] [N4] [2,4] [3,6]
[N,4] [1,3] [2,4] [3,6]
[N,4] [2,4] [2,6] [3,6]

[N,4] [2,4] [3,6] [3,C]

Signature Sets for Structure 6:
[N,5] [1.4]

[N,5] [1,4] [3,C]
[1,4] [3,C] [4,C]

Signature Sets for Structure 7:
[1,4] [2,5]

[2,5] [3,6]

[N,3] [1,4] [3,6]
[N,3] [2,4] [25}
[N3] [2,5] [3,51
[1,3] [1,4] [3,6]
[1,4] [2,4] [2,5]
[1,4] [2,4] [3,6]
[1,4] [2,5] [3,5]
[1,4] [2,5] [3,6]

[1,4] [3,5] [3,6]

119

[1,4] [3,6] [4,6]
[1,4] [3,6] [4,C]
[2,4] [2,5] [3,6]
[2,4] [2,5] [4,C]
[2,5] [3,5] [3,6]

[2,5] [3,5] [4,C]

Signature Sets for Structure 8:
[1,4] [2,6]

[2,6] [3,5]

[N,3] [2,41 [2,61

[1,4] [2,4] [2,6]

[2,4] [2,6] [3,5]

[2,4] [2,6] [4,6]

[2,4] [2,6] [4,C]

Signature Sets for Structure 9:
[N4] [151

[1,4] [1,5]

[1,5] [3,C]

[N,4] [1,4] [3,5]

[N,4] [1,41 [4,6]

[1,4] [1,5] [3,C]

120

[1,4] [3,C] [4,6]
[N,4] [1.41135] [3,C]
[1,4] [1,5] [3,5] [3,C]
[1,4] [3,5] [3,C] [4,6]

[1,4] [3,5] [3,C] [5,C]

Signature Sets for Structure 10:
[1,5] [2,4]

[1,5] [3,6]

[N3] [15} [3,51

[1,3] [1,5] [3,5]

[1,5] [2,4] [3,51

[1,5] [3,5] [3,6]

[1,5] [3,5] [4,C]

Signature Sets for Structure 11:
[1,51 [2,5]
[1.5] [2.6]

[2,5] [2,6]

Signature Sets for Structure 12:

[N24] [1,61

[1,4] [1,6]

121

[1,6] [3,6]
[1,6] [3,C]
[N,4] [1.4] [3,6]
[1,4] [1,6] [3,6]

[1,4] [3,6] [3,C]

Signature Sets for Structure 13:

[1,6] [2,4]

[1 ,6] [3,5]

Signature Sets for Structure 14:

[1,6] [2,5]

122

 

   

S

1[11111111111111111111l

1
1
477