Modification of citrulline residues with 2,3-butanedione facilitates their detection by liquid chromatography/mass spectrometry

Article abstract of DOI:10.1002/rcm.5015

Citrullination is a post-translational modification (PTM) that results from the deimination of the amino acid arginine into citrulline by Peptidyl Arginine Deiminase enzymes and occurs in a wide range of proteins in health and disease. This modification c

Full text of DOI:10.1002/rcm.5015

Research Article
Received: 13 January 2011
Revised: 15 March 2011
Accepted: 16 March 2011
Published online in Wiley Online Library
Rapid Commun. Mass Spectrom. 2011, 25, 1536–1542
(wileyonlinelibrary.com) DOI: 10.1002/rcm.5015
Modification of citrulline residues with 2,3‐butanedione
facilitates their detection by liquid chromatography/mass
spectrometry
Marlies De Ceuleneer, Vanessa De Wit, Katleen Van Steendam, Filip Van Nieuwerburgh,
*
Kelly Tilleman and Dieter Deforce
Laboratory for Pharmaceutical Biotechnology, Ghent University, Ghent, Belgium
Citrullination is a post‐translational modification (PTM) that results from the deimination of the amino acid arginine
into citrulline by Peptidyl Arginine Deiminase enzymes and occurs in a wide range of proteins in health and
disease. This modification causes a 1 Da mass shift, which can be used to identify citrullination sites in proteins by
the use of mass spectrometry. However, other PTMs, such as deamidation from asparagine to aspartic acid or from
glutamine to glutamic acid, can also cause a 1 Da mass shift, making correct interpretation of the data more difficult.
We developed a chemical tagging strategy which, combined with an open source search application, allowed us to
selectively pinpoint citrullinated peptides in a complex mixture after liquid chromatography/mass spectrometry (LC/
MS) analysis. After incubation of a peptide mixture with 2,3 butanedione, citrulline residues were covalently
modified which resulted in a 50 Da shift in singly charged mass. By comparison of the peptide mass fingerprint from
a modified and an unmodified version of the same sample, our in‐house search application was able to identify the
citrullinated peptides in the mixture. This strategy was optimized on synthetic peptides and validated on a digest of
in vitro citrullinated fibrinogen, where different proteolytic enzymes were used to augment the protein coverage.
This new method results in easy detection of citrullinated residues, without the need for complex mass spectrometry
equipment. Copyright © 2011 John Wiley & Sons, Ltd.
Citrulline is an amino acid that is not encoded in the DNA;
instead, when present in a protein, it originates from a post‐
translational modification (PTM) (deimination) of arginine.
This deimination, catalyzed by Peptidyl Arginine Deiminases
(PADs), results in a neutral amino acid with a mass increase
of 1 Da (Fig. 1).[1] It is clear that the loss of a positive charge
can have severe effects on the tertiary structure of a protein,
making citrullination an interesting modification to investi-
gate, especially when examining immunogenicity in auto-
immunity. After all, a difference in tertiary structure might
bring new epitopes to the surface of a molecule, triggering an
immune reaction. In addition, citrullination creates neo‐
epitopes that will activate the immune system.[2]
Citrulllination in the context of autoimmune disease is most
often linked to rheumatoid arthritis, where the presence of
citrullinated proteins and antibodies against these citrullinated
proteins (ACPA) in the affected joints has been well docu-
mented.[3–5] In fact, the determination of ACPA‐titres has
proven to be very important for diagnosis of rheumatoid
arthritis (RA), since they show a specificity of 90–98%.[6] The
identification of the exact antigen that elicits this specific
immune response, and most importantly the location of
citrullinated residues within this antigen, would be of great
interest in the research of the pathogenesis of RA.
A role for citrullination has also been demonstrated in
multiple sclerosis,[7] indicating that this small modification
and resulting conformation change might have an impact on
antigen‐TCR contact and the resulting antibody production
in a variety of diseases.[8,9]
Physiologically, the deimination of arginine residues on
cytoskeletal proteins causes loss of cell structure and plays a role
in apoptosis.[10] When also considering histone citrullination as
an important epigenetic regulatory mechanism,[11] it comes as
no surprise that the interest in characterization of citrullinated
proteins is on the rise. After all, the ability to localize the
citrullinated residue in a protein will provide invaluable
information about the changes in structure and function of this
protein, and will therefore improve current understanding of
the importance of this PTM and its role in several autoimmune
diseases.
Although citrullination as a PTM plays a role in many
physiological and pathological processes, relatively few
methods exist to easily identify deiminated arginine residues
in a protein. In most cases, citrullination is verified on
Western blot, either with an antibody against citrulline itself
or against an on‐blot chemically modified version of the
amino acid. However, using these detection methods, it is
only possible to determine the citrullination state of an entire
protein without being able to annotate the location of the
converted arginine. The latter can be achieved by mass
spectrometry (MS), but although a mass difference of 1 Da
* Correspondence to: D. Deforce, Laboratory for Phar-
maceutical Biotechnology, Ghent University, Ghent,
Belgium.
E-mail: dieter.deforce@ugent.be
Rapid Commun. Mass Spectrom. 2011, 25, 1536–1542
Copyright © 2011 John Wiley & Sons, Ltd.

Detection of citrulline residues by LC/MS
Figure 1. Modification of protein‐bound arginine and citrulline. In vivo or in vitro deimination by PAD enzymes
converts an arginine into a citrulline residue, which causes a mass shift of +1 Da. Afterwards, in vitro modification with
2,3‐butanedione at low pH results in the formation of an imidazolone derivative, with a specific mass shift of +50 Da.
Modification
can be distinguished by high resolution mass spectrometry,
there are other PTMs that also give rise to a 1 Da difference.
Deamidation of Asn to Asp and of Gln to Glu or even the
uptake by the peptide of a hydrogen atom can also account
for this small difference. Moreover, a 1 Da mass difference can
be especially difficult to detect on higher charged peptides.
Therefore, another, more specific method of detecting citrul-
line residues by means of MS is needed.
Modification of citrullinated peptides (1 nmol unless other-
wise specified) with 10 μL 50 mM 2,3‐butanedione (BD)
(Sigma‐Aldrich, Germany) in 30 μL trifluoroacetic acid (TFA)
(Sigma‐Aldrich, Germany) was carried out at 37°C for 16 h.
Other reaction conditions were tested and are specified
throughout the text.
To mimic more complex protein samples, peptides were
A recent report described the neutral loss of isocyanic acid
from citrulline and used this loss as a marker for citrullinated
peptides.[12] This technique could be useful; however, the
detection of a true neutral loss on a quadrupole time‐of‐flight
(QTOF) instrument is not possible[13] and limits the researcher
in instruments to a triple quadrupole mass spectrometer.
Similarly, Stensland et al. used a combination of Collision‐
Induced Dissociation (CID) and Electron‐Transfer Dissocia-
tion (ETD) for the fragmentation and identification of
modified citrullinated peptides;[14] this technique, though
very elegant, is not readily available to all research groups.
Therefore, we developed a modification‐based MS method for
citrulline detection which could universally be performed on
all electrospray ionization (ESI)‐based mass spectrometers.
This approach involved 2,3‐butanedione (BD), which is
used in chemical modification of both arginine[15] and citrul-
line[16] (Fig. 1). In the case of arginine, this reaction is usually
performed in basic environment in the presence of arylboronic
acid; when performed in acidic conditions and followed by
reaction with antipyrin, it is specific for citrulline‐containing
residues. The latter modification reaction was used in order to
design a simple yet specific liquid chromatography/mass
spectrometry (LC/MS) method for the detection of citrulline‐
containing peptides.
spiked into
Sunnyvale, CA, USA).
a synthetic cytochrome C digest (Dionex,
In vitro citrullination of human fibrinogen
In vitro citrullination was performed on human fibrinogen
purified from plasma (Calbiochem, Darmstadt, Germany) in
deimination buffer (0.1 M tris‐HCl, pH 7.4, 100 mM CaCl2,
5 mM dithiothreitol (DTT)), by adding 7 IU of PAD from
rabbit skeletal muscle (Sigma‐Aldrich, St. Louis, MO, USA)
per mg of protein. After 2 h at 37°C, the reaction was stopped
by adding EDTA to a final concentration of 50 mM.
In‐solution digest of proteins
Proteins were digested using both LysC endoprotease
(Sigma‐Aldrich) and GluC endoproteinase (Sigma‐Aldrich).
Briefly, a protein was dissolved in 50 mM ammonium
bicarbonate (Sigma‐Aldrich) and 6 M urea (Sigma‐Aldrich),
after which it was first reduced with 20 mM DTT (Invitrogen,
CA, USA) for 1 h at room temperature and subsequently
alkylated with 200 mM iodoacetamide (MP, Solon, OH, USA)
for 1 h at room temperature. Then, samples were diluted 1:10
with MilliQ ultrapure water and digested with GluC or LysC
endoproteinase for 4 h at 37°C (1 part enzyme to 50 parts
protein).
EXPERIMENTAL
Fractionation on HPLC
Synthetic peptides
Peptides originating from 10 to 20 µg of protein were separated
on a PepMap 100 (C18) column (i.d. 75 µm, length 15 or 25 cm;
Dionex, Sunnyvale, CA, USA) at 60°C, by use of a U3000 nano‐
LC system (Dionex) at a flow rate of 300 nL/min. Elution was
performed with 0.1% formic acid (FA) in water (buffer A) and
80% acetonitrile/0.1% FA (v/v) in water (buffer B), increasing
Two synthetic peptides derived from human vimentin, one
without citrulline (vim = NH2‐STSRSLYASSPG‐COOH (MW
1213.9)) and one citrullinated (vim cit = NH2‐STS Cit
SLYASSPG‐COOH (MW 1215.2)), were purchased from
Thermo Scientific (Rockford, IL, USA).
Rapid Commun. Mass Spectrom. 2011, 25, 1536–1542 Copyright © 2011 John Wiley & Sons, Ltd. wileyonlinelibrary.com/journal/rcm

M. De Ceuleneer et al.
from 4% to 65%B in 60 min, ending with 15 min of 100%B.
Fractions were collected every 5 min (n = 15), subsequently dried
in a SpeedVac and resuspended in 20μL MilliQ ultrapure water.
Half of each fraction was then modified as described above.
of 25 pmol/μL. The yield of the modification gradually
improved when the reaction was allowed to take place for
longer periods of time and, at 16 h, 96% of citrullinated
peptides were modified (Fig. 2(A)). In all further experiments,
incubation occurred overnight.
Mass spectrometric analysis
An attempt was made to optimize reaction conditions by
adjusting the concentration of reagents. To obtain the lowest
possible modification of arginine residues, a different volume
(60 μL instead of 30 μL) of TFA was tested, resulting in a
lower pH. Moreover, to minimize the background peaks in
the spectrum originating from the modification reagent BD, a
lower concentration (5 mM instead of 50 mM) was tested. To
ascertain whether concentration of peptides influenced the
quantity of reagents needed, different amounts of peptide
(16 pmol, 0.8 pmol and 0.16 pmol) were modified. After
analysing these different reaction conditions for traces of
unreacted unmodified citrullinated peptides, the best results
were obtained with 30 µl TFA and 10 µl 50 mM BD, as
described in literature (data not shown).[19] Even at low
concentration, full modification of the citrullinated peptide
could be observed after overnight incubation.
After modification of 1 nmol of the synthetic citrullinated
peptide (m/z 607.3 [2+]), new spectral peaks were observed at
m/z 631.3 [2+] and 632.3 [2+] in the modified sample when
compared to the unmodified sample (Fig. 2(B), vim cit, m/z
607.3 [2+], versus vim cit mod, m/z 632.3 [2+]). This shift of
50 Da (or in this case m/z 25 because of the doubly charged
nature of the peptide) matched with the formation of an
imidazolone derivative as proposed by Holm et al.[19] (Fig. 1).
The peptide showing the 50 Da mass shift was indeed the
citrullinated chemically modified peptide as MS/MS analysis
showed the 2,3‐butanedione being covalently attached to the
citrulline residue (Fig. 2(C)). Out of a total of 14 fragment
ions, y‐ions as well as b‐ions, nine fragment ions were
annotated in the MS/MS spectrum with an additional mass
of 50 Da (denoted by * in Fig. 2(B)). Due to the non‐tryptic
nature of the peptide, y‐ions were less frequently observed.
The amino acid sequence of the peptide could still be
established at a concentration of 25 pmol/μL.
Samples were introduced into the mass spectrometer (QqTOF,
Waters, Milford, MA, USA), through a nano‐ESI source by
infusion at a flow rate of 0.5 μL/min. LC/MS was used when
more complex mixtures were analyzed and was performed
using the U3000 system (Dionex, Sunnyvale, CA, USA), with a
C18 column (i.d. 75 µm, length 15 or 25 cm; Dionex,
Sunnyvale, CA, USA) at 25°C or 60°C. The same gradient
was used as described above at a flow rate of 300 nL/min.
For tandem mass spectrometry (MS/MS), ions were frag-
mented by CID, with a collision energy of 10eV for infused
samples and with a custom collision energy profile for LC/MS/
MS samples, ranging from 25 to 55 eV for doubly charged
peptides between m/z 400 and 1200, and ranging from 11 to
26 eV for triply charged peptides between m/z 435 and 1000.
All solvents were MS‐grade and purchased from Biosolve
(Westford, MA, USA).
Peak picking and analysis
For MS‐only samples, peak picking was performed using
Mascot Distiller version 2.3.2 (Matrix Science, London, UK)
for peaks with a maximum charge of 5. Correlation threshold
was set to 0.6 and signal‐to‐noise ratio had to be at least 2.
Afterwards, peak lists from unmodified and modified
samples were loaded into an in‐house R application MsMod
(available upon request). Basically, the program contains two
data‐filtering steps. The first step filters out peak data from the
unmodified sample matching the modified data within a user‐
defined experimental mass tolerance (we used 0.1 Da). The
second filter is based on the web‐based tool PeptideMass,[17]
a
protein digest simulator of the protein(s) under investigation
(in our case fibrinogen chain α, β and γ). The digestion
parameters allow up to 1 and 3 miscleaved sites for LysC and
GluC, respectively, and a user‐defined mass range, in our case of
350–8192Da. The theoretical isotopic profiles of the generated
peptides[18] are compared with the filtered measured data. The
filter parameters allow for a user‐defined mass deviation, a
fixed modification of carbamidomethylated cysteines, as well as
a variable N‐terminal carbamidomethylation, and a variable
citrullination.
The reaction was found to be specific for citrullinated
peptides: no shift could be detected when non‐citrullinated
peptides were subjected to the same reaction (data not shown).
Modification of citrullinated peptides in mixtures
To determine the behaviour of peptides in more complex
mixtures, both citrullinated and non‐citrullinated synthetic
peptides were spiked in a cytochrome C digest. The entire
mixture was then modified as described in the Experimental
section.
Manual confirmation of the data was performed using
MassLynx software (Waters, Milford, MA, USA). MS/MS
analysis was performed on a Mascot in‐house server.
Different dilutions of synthetic peptides in the cytochrome
C digest were tested (1:10, 1:20, 1:100); complete modification
of the citrullinated peptides could still be observed at a 1:100
dilution (160 fmol peptide/16 pmol cytochrome C digest)
(data not shown). In all modified samples, modification of
the citrullinated synthetic peptide vim cit could be detected.
RESULTS AND DISCUSSION
Modification of citrullinated peptides with BD is specific
and causes a mass shift of 50 Da
The reaction of citrullinated peptides with BD alone, as
opposed to the widely used combination of BD and antipyrine,
was evaluated on 1 nmol citrullinated synthetic peptide using
previously described reaction conditions.[19] Different time
points (30 min, 1 h, 2 h, 4 h, 8 h and 16 h) were tested and the
reaction was verified by infusion MS at a concentration
In a small portion of the runs (4/16) unmodified
citrullinated peptides could still be found, with a reaction
yield of >70% in all samples. This indicates an incomplete
reaction when the modification is carried out on more
complex samples, an observation also made by Stensland
et al.[14]
wileyonlinelibrary.com/journal/rcm Copyright © 2011 John Wiley & Sons, Ltd. Rapid Commun. Mass Spectrom. 2011, 25, 1536–1542

Detection of citrulline residues by LC/MS
Figure 2. Modification of a citrullinated synthetic peptide (H2N‐STSCitSLYASSPG‐COOH) with 2,3 butanedione in 60% TFA.
(A) Increasing incubation time results in higher amounts of modified peptides, with the highest (96%) after 16 h of incubation.
(B) Specific modification of citrulline residues causes a 25 Da mass shift for doubly charged peptides, in this case from m/z 607.3
to 632.3. (C) MS/MS spectrum of the modified peptide. Due to the non‐tryptic nature of the peptide, b‐ions are most abundant
in the spectrum. Seven out of 12 possible b‐ions and 2 out of 12 possible y‐ions have been found with a mass exceeding their
theoretical mass by 50 Da (denoted by *), proving the covalent bond between BD and the citrulline residue.
In order to clarify whether non‐citrullinated peptides
from cytochrome C were affected by the reaction, we inves-
tigated two peptides, one without an arginine residue
(KTGQAPGFSYTDANK, m/z 528.9 [3+]) and one with an
arginine residue (GEREDLIAYLKK, m/z 717.6 [2+]). These m/z
values were found in all samples, both before modification
and after, while no m/z value of 50 Da higher could be detected
in the modified sample (data not shown). This indicated that
no modification of random peptides took place, not even
when an arginine residue was present in the peptide.
in order to obtain a modified and unmodified version of the
same fraction. This fractionation was followed by consecutive
LC/MS analysis of first the modified and next the unmodified
sample, to minimize inter‐run variability.
To facilitate comparison of the modified with the unmod-
ified sample, an in‐house application was developed in R to
process the data.
Peak lists generated in Mascot Distiller for both the
unmodified and the modified sample were compared to each
other based on mass and charge, with a tolerance of 0.1 m/z,
and peaks that were the same in both files were discarded. The
resulting peak list was then compared to a theoretical digest of
the citrullinated form of all three fibrinogen chains. Based
on the list resulting from this semi‐automated comparison,
manual analysis was performed to check if the modified form
of the peptide was present in the modified sample. Manual
analysis was necessary because of the low ion counts of the
modified peptides, which prevented accurate peak picking of
some of these peptides.
Detection of citrullinated peptides originating from an
in vitro citrullinated protein
As mentioned above, we noted that the reaction did not run to
completion when applied to a more complex sample. We
found that it was necessary to first fractionate more complex
samples, such as whole protein digests, prior to modification.
When utilizing our method for the elucidation of citrullinated
sites in in vitro citrullinated human fibrinogen, 10 µg (approxi-
mately 100 pmol) of a LysC digest was first fractionated by
HPLC (n = 2). Afterwards, the 15 fractions were split into two
This process of detection is illustrated in Fig. 3 for the
peptide ESSHHPGIAEFPSRGK. In its citrullinated form, this
Rapid Commun. Mass Spectrom. 2011, 25, 1536–1542 Copyright © 2011 John Wiley & Sons, Ltd. wileyonlinelibrary.com/journal/rcm

M. De Ceuleneer et al.
peptide should have a calculated mass of m/z 608.4 [3+], which
could be observed in the unmodified sample (Fig. 3(A)).
This, taken together with a loss of signal in the modified sample,
resulted in detection by the search algorithm. Afterwards,
manual analysis was performed to investigate modification.
In the modified sample, a peptide with m/z 625.2 [3+] could
be found (Fig. 3(B)). The difference in m/z between both
peptides mentioned was 16.6. Since the peptides were triply
charged, this corresponded to a 50 Da mass shift indicative for
a modified citrullinated residue. As can be seen in Fig. 3(C), a
residual amount of unmodified peptide could be observed in
the modified sample. However, the ion count was 85% lower,
indicating that at least part of the total peptide amount was
modified and preventing the peaks being efficiently picked by
Mascot Distiller. This large loss in signal was the primary
reason for detection by our search application MSMod.
No trace of the m/z signal for 625.2 [3+] could be found in
the unmodified sample (Fig. 3(D)), indicating that this
peptide was indeed a modification product, rather than a
peptide resulting from the digest.
again, this time including MS/MS analysis. In total, six
peptides originating from a LysC digest of citrullinated human
fibrinogen could be positively identified as citrullinated with
this method (see Table 1).
To ensure better coverage of the entire protein, theexperiment
was repeated using GluC as endoproteinase (n = 2). Using this
protease, proteolysis resulted in shorter peptides with more
easily detectable arginines/citrullines. Nine more peptides
could be identified (Table 1) after comparison with the
theoretical digest.
After both LysC and GluC digestion, 15 peptides, encom-
passing 17 citrullinated residues, were identified using the
search algorithm due to loss of signal after modification
(Table 1). For nine of the 15 peptides identified, the modified
form could be found in the modified sample (italicized in
Table 1); for the other six peptides, signal‐to‐noise was too low
(S/N <2 in the MS spectrum) for reliable manual peak picking.
Of the 15 peptides in total discovered by modification and
subsequent automatic comparison, eight could also be
identified by MS/MS followed by Mascot analysis (shown in
bold in Table 1). The other six could not be retrieved by a
conventional LC/MS/MS approach, indicating that our
method was more sensitive in picking up low‐abundance
citrullinated peptides in a complex mixture.
This approach was validated where possible by peptide
sequence analysis. To this end, any peaks from the unmodified
fraction not found in the modified fraction were compiled into
an include list, after which the unmodified samples were run
Figure 3. Detection of a citrullinated peptide in a LysC digest of in vitro citrullinated fibrinogen (single scan spectra are shown
at the time of highest ion count). (A) MS spectrum of the unmodified sample at retention time (Rt) 21.5 min: Peptide
ESSHHPGIAEFPSRGK should have a calculated mass of 608.1 [3+]. However, in the unmodified sample, a shift of +0.33 Da can
be observed on the triply charged peptide, which corresponds to a 1 Da mass shift from arginine to citrulline. (B) MS spectrum
of the modified sample at Rt 30.2 min: In the modified sample, a peptide with m/z 625.2 [3+] can be found. The difference in m/z
between both peptides mentioned is 16.6 Da. Since the peptides are triply charged, this corresponds to a 50 Da mass shift, also
indicative for a modified citrullinated residue. The observed shift in Rt amounts to 11 min (from 22.3 min in the unmodified
sample to 33.9 min in the modified sample). (C) MS spectrum of the modified sample at Rt 22.3 min: Modification is not
quantitative, as can be seen here by the presence of a residual amount of unmodified peptide in the modified sample. (D) MS
spectrum of the unmodified sample at Rt 30.2 min: No trace of the m/z 625.2 [3+] can be found in the unmodified sample,
demonstrating that this peptide is indeed a product of modification, rather than a peptide resulting from the digest.
wileyonlinelibrary.com/journal/rcm Copyright © 2011 John Wiley & Sons, Ltd. Rapid Commun. Mass Spectrom. 2011, 25, 1536–1542

Detection of citrulline residues by LC/MS
Table 1. Peptides identified as being citrullinated in in vitro citrullinated fibrinogen. m/z (cit) was found in the unmodified sample
by the search algorithm. Bold R denotes a citrullinated arginine residue, underlined amino acids denote carbamidomethylation.
Masses are italicized when found by manual analysis and displayed in bold when confirmed by MS/MS
Peptide
Strand
Protease
m/z (cit)
m/z (mod)
SRGSESGIFTNTKE
TFPGFFSPMLGEFVSETESRGSESGIFTNTK
VSGNVSPGTRRE
GGGVRGPRVVE
GGGVRGPRVVE
KLVTSKGDKELRTGKE
SGSFRPDSPGSGNARPNNPDWGTFEE
ESSSSHHPGIAEFPSRGK
QFTSSTSYNRGDSTFESK
FPSRGKSSSYSKQFTSSTSYNRGDSTFE
DLLPSRDRQHLPLIK (*)
VTSGSTTTTRRSCSKTVTK
LRTGKEKVTSGSTTTTRRSCSKTVTK
TVIGPDGHKE
A
A
A
A
GluC
LysC
GluC
GluC
757.4 (2+)
1129.9 (3+)
659.3 (2+)
542.8 (2+)
571.3 (2+)
895.5 (2+)
927,575 (3+)
608.5 (3+)
1022 (2+)
1047.4 (3+)
601.4 (3+)
687.2 (3+)
782.2 (3+)
782.5 (2+)
1146.6 (3+)
709.3 (2+)
567.8 (2+)
596.3 (2+)
920.5 (2+)
960.9 (3+)
625.1 (3+)
1047 (2+)
1080.6
A
A
A
A
A
A
A
A
GluC
GluC
LysC
LysC
GluC
LysC
LysC
GluC
618.1 (3+)
720.5 (3+)
815.7 (3+)
KREEAPSLRPAPPPISGGGYRARPAK
YCRTPCTVSCNIPVVSGKECEE
ASILTHDSSIRYLQE
B
B
C
LysC
GluC
GluC
921.2 (3+)
882.5 (3+)
867.5 (2+)
971.2 (3+)
899.2 (3+)
892.5 (2+)
(*)This peptide was citrullinated on one residue, although we could not determine on which one due to the close
proximity of both arginine residues.
When looking at other methods for elucidating citrullination
sites, such as enzymatic labelling[20] or accurate mass determi-
nation,[21] attention can be drawn to two clear advantages
of our method. Five of the citrullinated arginines we identi-
fied in the α‐strand of fibrinogen (QFTSSTSYNRGDSTFESK,
GGGVRGPRVVE, ESSSSHHPGIAEFPSRGK and VSGNVSP
GTRRE) have already been reported as being citrullinated
in vitro by Kubota et al.[20] In that paper, the authors identi-
fied citrullination sites in fibrinogen α by incorporation of O18
by using trypsin, GluC and chymotrypsin. Using only two
digestive enzymes, LysC and GluC, we were able to elucidate
seven more citrullination sites in fibrinogen α (SRGSES
GIFTNTKE, SGSFRPDSPGSGNARPNNPDWGTFEE, KLVT
SKGDKELRTGKE, VTSGSTTTTRRSCSKTVTK and DLLPS
RDRQHLPLIK) which were not covered in the analysis of
Kubota et al. Additionally, this enzymatic labelling approach
can only be used after in vitro labelling, while our method
could be extrapolated to in vivo citrullinated samples for
future research.
The most widely used approach to identifying citrullination
sites to date is the detection of a 1 Da mass shift, most often in
combination with MS/MS analysis. However, one of the
identified peptides in this analysis illustrates the advantage of
a specific modification of citrullinated residues. Peptide
QFTSSTSYNRGDSTFESK has an m/z of 1022 [2+] and was
correctly identified after MS/MS analysis with a 1 Da modi-
fication. However, since asparagine (N) and arginine (R) are in
such close proximity, the Mascot search algorithm could not
successfully distinguish between a deamidated asparagine
and a citrullinated arginine (scored 144 and 139, respectively,
from the same spectrum). The extra modification of the
citrulline residue with BD, and the resulting loss of signal in
the modified sample, dismissed the possibility that the 1 Da
modification resided anywhere else but on the arginine
residue.
CONCLUDING REMARKS
Different methods already exist for the detection of protein‐
bound citrulline. In most cases, these are based on the 1 Da
mass difference when compared to the same peptide with
arginine instead of citrulline. This analysis is often combined
with a tryptic miscleavage to confirm the findings, since
trypsin cannot cleave after a citrulline residue.[22] However,
this method may incorrectly interpret the 1 Da mass shift,
which will then be annotated as a deamidation of glutamine
or asparagine. Other methods, based on the detection of a
neutral loss[12] or chemical modification followed by a
combination of CID and ETD peptide fragmentation,[14] have
already been successfully used in the detection of protein‐
bound citrulline. Similarly, analysis of citrullinated peptides
from in vitro citrullinated fibrinogen has been performed
based on the observation of a 1 Da mass increase, combined
with the use of both dissociation techniques.[23] However, this
detection can only be performed by a selected group of mass
spectrometers, respectively a triple quadrupole or an ion trap
mass analyzer, thereby hampering widespread usage.
We set out to develop a simple, specific LC/MS method for
the detection of peptide‐bound citrulline based on the modi-
fication of citrulline with 2,3‐butanedione in acidic environ-
ment that can be carried out on any mass spectrometer.
This modification resulted in a specific 50 Da mass shift,
and could be carried out on peptide amounts as low as
160 fmol. In mixtures, the reaction was not complete, but
still sufficient to allow detection of citrullinated peptides in
a complex sample, as demonstrated by the identification of
Rapid Commun. Mass Spectrom. 2011, 25, 1536–1542 Copyright © 2011 John Wiley & Sons, Ltd. wileyonlinelibrary.com/journal/rcm

M. De Ceuleneer et al.
[5] D. Caspi, M. Anouk, I. Golan, D. Paran, I. Kaufman,
I. Wigler, D. Levartovsky, I. Litinsky, O. Elkayam. Arthritis
Rheum. 2006, 55, 53.
[6] B. Vander Cruyssen, I. Peene, T. Cantaert, I. E. Hoffman,
L. De Rycke, E. M. Veys, F. De Keyser. Autoimmunol. Rev.
2005, 4, 468.
[7] L. Tranquill, L. Cao, N. Ling, H. Kalbacher, R. Martin,
J. Whitaker. Mult. Scler. 2000, 6, 220.
[8] J. A. Hill, S. Southwood, A. Sette, A. M. Jevnikar, D. A. Bell,
E. Cairns. J. Immunol. 2003, 171, 538.
[9] A. Carrillo‐Vico, M. Leech, S. Anderton. J. Immunol. 2010,
184, 2839.
[10] M. Inagaki, H. Takahara, Y. Nishi, K. Sugawara, C. Sato.
J. Biol. Chem. 1989, 264, 18119.
[11] J. Dieker, S. Muller. Clin. Rev. Allergy Immunol. 2010, 39, 78.
[12] G. Hao, D. Wang, J. Gu, Q. Shen, S. S. Gross, Y. Wang. J. Am.
Soc. Mass Spectrom. 2009, 20, 723.
[13] D. Schwudke, J. Oegema, L. Burton, E. Entchev, J. Hannich,
C. Ejsing, T. Kurzchalia, A. Shevchenko. Anal. Chem. 2006,
78, 585.
15 citrullinated peptides from in vitro citrullinated fibrino-
gen. The method also avoids the incorrect interpretation of a
1 Da mass shift caused by deamidation of glutamine or
asparagine, which can be especially problematic if these
residues are situated in close proximity to the citrulline
residue.
To aid data processing a search algorithm, MSMod, was
developed, which facilitated the comparison of modified and
unmodified samples and might also be used for the detection
and evaluation of other post‐translational modifications in
different samples.
In conclusion, we developed a fast and simple method to
identify and analyze citrullinated peptides in a mixture by
selectively tagging the citrullinated residue with BD, fol-
lowed by LC/MS analysis. In the future, this specific tagging
strategy will be used in the analysis of in vivo citrullinated
samples, in order to map the citrullinated epitopes in
different autoantigens.
[14] M. Stensland, A. Holm, A. Kiehne, B. Fleckenstein. Rapid
Commun. Mass Spectrom. 2009, 23, 2754.
[15] A. Leitner, S. Amon, A. Rizzi, W. Lindner. Rapid Commun.
Mass Spectrom. 2007, 21, 1321.
Acknowledgements
[16] T. Senshu, T. Sato, T. Inoue, K. Akiyama, H. Asaga. Anal.
Biochem. 1992, 203, 94.
[17] http://www.expasy.org/tools/peptide‐mass.html.
[18] M. Palmblad, J. Buijs, P. Hakansson. J. Am. Soc. Mass
Spectrom. 2001, 12, 1153.
Marlies De Ceuleneer, Katleen Van Steendam and Kelly
Tilleman were supported by the Fund for Scientific Research
(FWO) Flanders.
[19] A. Holm, F. Rise, N. Sessler, L. M. Sollid, K. Undheim,
B. Fleckenstein. Anal. Biochem. 2006, 352, 68.
REFERENCES
[20] K. Kubota, T. Yoneyama‐Takazawa, K. Ichikawa. Rapid
Commun. Mass Spectrom. 2005, 19, 683.
[21] M. Hermansson, K. Artemenko, E. Ossipova, H. Eriksson,
J. Lengqvist, D. Makrygiannakis, A. Catrina, A. Nicholas,
L. Klareskog, M. Savitski, R. Zubarev, P. Jakobsson.
Proteomics Clin. Appl. 2010, 4, 511.
[22] A. Ordonez, I. Martinez‐Martinez, F. J. Corrales, C. Miqueo,
A. Minano, V. Vicente, J. Corral. FEBS J. 2009, 276, 6763.
[23] J. J. van Beers, R. Raijmakers, L. E. Alexander, J. Stammen‐
Vogelzangs, A. M. Lokate, A. J. Heck, R. B. Schasfoort, G. J.
Pruijn. Arthritis Res. Ther. 2010, 12, R219.
[1] B. Gyorgy, E. Toth, E. Tarcsa, A. Falus, E. I. Buzas. Int.
J. Biochem. Cell Biol. 2006, 38, 1662.
[2] P. Mydel, Z. Wang, M. Brisslert, A. Hellvard, L. E. Dahlberg,
S. L. Hazen, M. Bokarewa. J. Immunol. 2010, 184, 6882.
[3] L. De Rycke, A. P. Nicholas, T. Cantaert, E. Kruithof,
J. D. Echols, B. Vandekerckhove, E. M. Veys, F. De
Keyser, D. Baeten. Arthritis Rheum. 2005, 52, 2323.
[4] A. Kinloch, K. Lundberg, R. Wait, N. Wegner, N. H. Lim,
A. J. Zendman, T. Saxne, V. Malmstr, P. J. Venables. Arthritis
Rheum. 2008, 58, 2287.
wileyonlinelibrary.com/journal/rcm Copyright © 2011 John Wiley & Sons, Ltd. Rapid Commun. Mass Spectrom. 2011, 25, 1536–1542