A method to produce fully characterized ubiquitin covalently modified by 4-hydroxy-nonenal, glyoxal, methylglyoxal, and malondialdehyde

Article abstract of DOI:10.3109/10715762.2015.1118472

Reactive carbonyl species (RCS) and the corresponding protein adducts (advanced glycoxidation or lipoxidation end products, i.e. AGEs and ALEs) are now widely studied from different points of view, since they can be considered as biomarkers, pathogenic factors, toxic mediators and drug targets. One of the main limits of the research in this field is the lack of standardized and fully characterized AGEs and ALEs to be used for biological, toxicological, and analytical studies. In this work, we set up a procedure to prepare and fully characterize a set of AGEs and ALEs by incubating ubiquitin - a model protein selected as target for carbonylation - with four different RCS: 4-hydroxy-trans-2-nonenal (HNE), methylglyoxal (MGO), glyoxal (GO), and malondialdehyde (MDA). After 24 h of incubation, the extent of protein carbonylation was estimated using a recently developed quantitative strategy based on high-resolution mass spectrometry. The resulting AGEs and ALEs were fully characterized by both intact protein and bottom-up analyses in terms of: stoichiometry of the total amount of modified protein, elucidation of the structure of the RCS-deriving adducts, and localization of the RCS-modified amino acids. Each RCS exhibited different reactivity toward ubiquitin, as detected by quantifying the extent of protein modification. The order of reactivity was MGO > GO > HNE > MDA. A variety of reaction products was identified and mapped on lysine, arginine, and histidine residues of the protein. In summary, a highly standardized and reproducible method to prepare fully characterized AGEs/ALEs is here presented.

Full text of DOI:10.3109/10715762.2015.1118472

Free Radical Research
ISSN: 1071-5762 (Print) 1029-2470 (Online) Journal homepage: http://www.tandfonline.com/loi/ifra20
A method to produce fully characterized ubiquitin
covalently modified by 4-hydroxy-nonenal, glyoxal,
methylglyoxal, and malondialdehyde
Mara Colzani, Angela Criscuolo, Gaia Casali, Marina Carini & Giancarlo Aldini
To cite this article: Mara Colzani, Angela Criscuolo, Gaia Casali, Marina Carini & Giancarlo
Aldini (2016) A method to produce fully characterized ubiquitin covalently modified by 4-
hydroxy-nonenal, glyoxal, methylglyoxal, and malondialdehyde, Free Radical Research, 50:3,
328-336, DOI: 10.3109/10715762.2015.1118472
To link to this article: http://dx.doi.org/10.3109/10715762.2015.1118472
Accepted author version posted online: 10
Nov 2015.
Published online: 14 Jan 2016.
Submit your article to this journal
Article views: 33
View related articles
View Crossmark data
Full Terms & Conditions of access and use can be found at
http://www.tandfonline.com/action/journalInformation?journalCode=ifra20
Download by: [RMIT University]
Date: 01 February 2016, At: 06:31

FREE RADICAL RESEARCH, 2016
VOL. 50, NO. 3, 328–336
http://dx.doi.org/10.3109/10715762.2015.1118472
RESEARCH ARTICLE
A method to produce fully characterized ubiquitin covalently modified by
4-hydroxy-nonenal, glyoxal, methylglyoxal, and malondialdehyde
Mara Colzani, Angela Criscuolo, Gaia Casali, Marina Carini and Giancarlo Aldini
`
Department of Pharmaceutical Sciences, Universita degli Studi di Milano, Milan, Italy
ABSTRACT
ARTICLE HISTORY
Received 17 July 2015
Revised 12 October 2015
Accepted 5 November 2015
Published online
Reactive carbonyl species (RCS) and the corresponding protein adducts (advanced glycoxidation or
lipoxidation end products, i.e. AGEs and ALEs) are now widely studied from different points of view,
since they can be considered as biomarkers, pathogenic factors, toxic mediators and drug targets.
One of the main limits of the research in this field is the lack of standardized and fully characterized
AGEs and ALEs to be used for biological, toxicological, and analytical studies. In this work, we set up
a procedure to prepare and fully characterize a set of AGEs and ALEs by incubating ubiquitin – a
model protein selected as target for carbonylation – with four different RCS: 4-hydroxy-trans-2-
nonenal (HNE), methylglyoxal (MGO), glyoxal (GO), and malondialdehyde (MDA). After 24 h of
incubation, the extent of protein carbonylation was estimated using a recently developed
quantitative strategy based on high-resolution mass spectrometry. The resulting AGEs and ALEs
were fully characterized by both intact protein and bottom-up analyses in terms of: stoichiometry
of the total amount of modified protein, elucidation of the structure of the RCS-deriving adducts,
and localization of the RCS-modified amino acids. Each RCS exhibited different reactivity toward
ubiquitin, as detected by quantifying the extent of protein modification. The order of reactivity was
MGO4GO4HNE4MDA. A variety of reaction products was identified and mapped on lysine,
arginine, and histidine residues of the protein. In summary, a highly standardized and reproducible
method to prepare fully characterized AGEs/ALEs is here presented.
8 December 2015
KEYWORDS
4-hydroxy-trans-2-nonenal;
glyoxal; high-resolution
mass spectrometry;
malondialdehyde;
methylglyoxal; reactive
carbonyl species
Introduction
Reactive carbonyl species (RCS) generated by lipid and
sugar oxidation are bioactive molecules involved in
several toxicological mechanisms. The main damaging
effects of RCS are due to their ability to covalently modify
nucleophilic macromolecules such as proteins, whose
covalent modifications induce cellular and tissue dam-
ages by different mechanisms, including protein function
derangement, protein oligomerization, immunogenicity,
and activation of specific receptors (e.g. RAGE) [1,2]. RCS
and the corresponding protein adducts (advanced
glycoxidation or lipoxidation end products, i.e. AGEs
and ALEs) [3] are now widely studied at different levels
and in particular as biomarkers, pathogenic factors, toxic
mediators, and drug targets [1,4–6].
exogenously, resulting in the formation of a great variety
of reaction products (AGEs/ALEs) that differ not only on
the basis of the modifying moiety, but also on their
ability to target different proteins on different amino
acids. For instance, a,b-unsaturated aldehydes can react
with histidine or cysteine residues through C3 and form
the corresponding Michael adducts or can react with the
primary amino group of lysine leading to the corres-
ponding Schiff bases [7]. Di-aldehydes such as glyoxal
(GO) can react with the side chain of lysine to yield
Ne-(carboxymethyl)lysine (CML) or with the side
chain of arginine forming two possible dihydroxyimida-
zolidine intermediates that can open to give
N7-(carboxymethyl)arginine or dehydrate to yield three
different imidazolone derivatives [3].
One of the main limits of research in this field is the
lack of standardized and fully characterized AGEs and
ALEs to be used for biological, toxicological, and
analytical studies. This is due to the wide chemical
diversity of RCS that can be formed endogenously or
Many studies in this field utilize AGEs/ALEs-albumin
of different sources (human, rat, or bovine), prepared
using
different
reactants,
including
reducing
sugars (glucose, ribose, fructose) or aldehydes
CONTACT Mara Colzani
Italy
Supplemental data for this article can be accessed http://dx.doi.org/10.3109/10715762.2015.1118472.
mara.colzani@unimi.it
Department of Pharmaceutical Sciences, Universita degli Studi di Milano, Via Mangiagalli 25, Milan 20133,
ß 2015 Taylor & Francis

FREE RADICAL RESEARCH
329
(glycolaldehyde or glyoxylic acid, GO, methylglyoxal
(MGO), 4-hydroxy-trans-2-nonenal [HNE], acrolein), and
different methods in which several variables (e.g. the
buffer, the incubation temperature, and the incubation
time) can vary considerably [8,9]. Moreover a detailed
chemical characterization of most of the so far reported
albumin reaction products is lacking, as is information
on the reproducibility and standardization of the
procedure. Additionally, model proteins other than
albumin have been used in some studies [10,11].
to obtain HNE. The concentration of HNE was estimated
by UV absorbance at 224 nm (molar extinction coeffi-
cient ¼ 13,750 Mꢁ1 ꢃ cmꢁ1).
In vitro generation of protein adducts by RCS
Ubiquitin was dissolved in phosphate buffer (10 mM
sodium dihydrophosphate NaH2PO4ꢀH2O buffered with
200 mM disodium phosphate Na2HPO4ꢀ2H2O to reach
pH 7.4) to a final concentration equal to 10 mM. Ubiquitin
was incubated at 37 ꢂC with HNE, MGO, GO, malondial-
dehyde (MDA) or hexane-2,5-dione at increasing con-
centrations (50 mM, 100 mM, 200 mM, 500 mM, 1 mM,
2 mM, and 5 mM) corresponding to protein:RCS molar
ratios equal to 1:5, 1:10, 1:20, 1:50, 1:100, 1:200, and
1:500. About 10 mM ubiquitin was incubated without
RCS, as control sample. The reactions were stopped after
24 h by diluting the excess of RCS by ultrafiltration, using
Amicon YM3 filter units (Millipore, Milan, Italy). To do so,
70 ml of solution was diluted with 430 ml of water, loaded
on the filter units, and centrifuged at 14,000 ꢃ g for
15 min. The eluates containing the excess reagents were
discarded, while the ubiquitin retained by the filter
(about 50 ml of solution) was rinsed with 450 ml of water.
The eluate was discarded and the washing procedure
was repeated twice. After the last centrifugation, the
eluates were discarded and the filters were flipped to
recover the ubiquitin solution (about 70 ml). If necessary,
the sample was brought to the initial volume (70 ml) with
water. Ubiquitin was either enzymatically digested or
directly injected into the mass spectrometer for intact
protein analysis (see below).
Hence, a highly standardized and reproducible prep-
aration of AGEs/ALEs is required together with their full
characterization in terms of stoichiometry, structure
elucidation of the modifying moieties, and identification
of the nucleophilic sites undergoing the modifications.
Moreover the method should also be able to modulate
the extent of the protein modification – a parameter that
should be easily determined in quantitative or semi-
quantitative terms.
We believe that the availability of a well-characterized
series of AGEs/ALEs is essential for different types of
studies, for instance to compare their biological effects
and in more detail to establish the contribution of the
RCS and the degree of the modifications on the
biological/toxicological properties of the protein adduct.
In this paper, we set up a procedure to prepare and
fully characterize a set of AGEs and ALEs derivatives by
incubating a model protein (ubiquitin) with HNE, GO,
MDA, and MGO, which were selected since they repre-
sent the most studied lipid and sugar oxidation products
and they belong to different chemical classes of RCS [3].
Materials and methods
Intact protein analysis by microflow automated
loop injection ESI-MS
Reagents
Ultrapure water was prepared with a Milli-Q H2O
purification system (Millipore, Bedford, MA). Sodium
dihydrophosphate (NaH2PO4ꢀH2O), disodium phosphate
(Na2HPO4ꢀ2H2O), ammonium bicarbonate (NH4HCO3),
formic acid (HCOOH), sodium borohydride (NaBH4), LC-
MS grade H2O/0.1% formic acid, LC-MS grade aceto-
nitrile/0.1% formic acid, lyophilized ubiquitin from
bovine erythrocytes (U6253), MGO (67028), GO (50649),
malondialdehyde tetrabutylammonium salt (63287), and
2,5-hexanedione (165131) were purchased from Sigma-
Aldrich (Milan, Italy).
The analysis of ubiquitin upon incubation with RCS in
presence or absence of carbonyl quenchers was per-
formed accordingly to a strategy already described for
HNE-modified ubiquitin [13]. Briefly, 40 ml of ubiquitin
solution recovered from filter units were mixed with
40 ml of denaturing solution (H2O/CH3CN/HCOOH; 40/60/
0.2; v/v/v). About 5 ml aliquots of the samples were
injected into a LTQ-Orbitrap XL mass spectrometer using
an ESI source (Thermo Scientific, Milan, Italy). Samples
were automatically injected by an Ultimate 3000
RSLCnano system that pumped an isocratic mobile
phase (H2O/CH3CN/HCOOH; 70/30/0.1; v/v/v) at a con-
stant flow rate of 10 ml/min. Sample injection and spectra
acquisition were fully automated and controlled by the
software Xcalibur (version 2.0.7, Thermo Scientific) and
Chromeleon Xpress (Dionex, version 6.80). Source par-
ameters were set as follows: spray voltage 1.8 kV,
4-Hydroxy-2-trans-nonenal
dimethylacetal
was
synthesized according to the literature [12] and stored
at ꢁ20 ꢂC. For each experiment, fresh HNE was prepared
starting from stored HNE-dimethylacetal, which was
evaporated under nitrogen stream and hydrolyzed with
10 mM HCl, pH 3 for 1 h at room temperature in the dark,

330
M. COLZANI ET AL.
capillary temperature 220 ꢂC, capillary voltage 35 V, tube
lens offset 120 V. A list of 20 background ions [14] was
used as lock mass values for real-time mass calibration.
Mass spectra were acquired by the Orbitrap analyzer in
positive ion mode using the following settings: profile
mode, scan range m/z 110-2000, AGC target 5 ꢃ 105,
maximum inject time 500 ms, resolving power 100,000
(FWHM at m/z 400). A dedicated Xcalibur processing
method was set to quantify the area under the
11 + multicharged peaks, localized in the m/z
range 779.00–783.50 for unmodified ubiquitin, 793.00–
Germany) and solubilized in 20 ml of 0.1% formic acid for
the subsequent LC-MS/MS analysis. In-gel digestion was
preferred to in-solution digestion because, under our
experimental conditions, it produced a higher number of
observable peptides (data not shown).
NanoLC–ESI-MS/MS analysis of peptides
Digested peptides (5 ml) diluted in 0.1% formic acid were
loaded on a C18 HALO PicoFrit column (75 mm ꢃ 10 cm,
˚
2.7 mm particle size, 100A pore size, New Objective,
Woburn, MA) for reverse phase chromatography.
Samples were injected by an Ultimate 3000 RSLCnano
system (Dionex, CA) controlled by the software
Chromeleon Xpress 6.80 (Dionex, CA) and electrosprayed
using a nanoESI source equipped with a dynamic
nanospray probe (both Thermo Scientific, Milan, Italy).
Peptide separation was performed by a reverse phase
linear gradient from 99% buffer A (water with 0.1%
formic acid), 1% buffer B (acetonitrile with 0.1% formic
acid) to 60% buffer A, 40% buffer B over 60 min, at flow
rate of 300 nl/min. The instrument was an LTQ-Orbitrap
XL (Thermo Scientific, Waltham, MA), controlled by the
software Xcalibur 2.0.7 and operated in data-dependent
mode to acquire both full MS and MS/MS spectra. Full
MS spectra were acquired in profile mode by the
Orbitrap analyzer in a scan range equal to m/z 300-
1500, using capillary temperature 220 ꢂC, AGC scan
5 ꢃ 105 and resolution 60,000 FWHM at m/z 400. Tandem
mass spectra were acquired by the LTQ analyzer for the 2
most intense ions exceeding 1 ꢃ 104 counts. MS/MS
spectra acquisition was set as follows: centroid mode,
resolution 15,000, precursor ions isolation width ¼ m/z 2,
AGC target 1 ꢃ 104 and normalized collision energy
30 eV. Dynamic exclusion was set as follows: 3 repeat
counts, 30 s repeat duration, 45 s of exclusion duration.
Monoisotopic precursor selection was enabled; singly
and unassigned charged ions were excluded from
fragmentation.
797.50
for
HNE-modified
ubiquitin,
784.25–
785.75 + 789.50–791.00 + 794.75–796.25 for GO-modified
ubiquitin, 784.00–787.00 for MGO-modified ubiquitin,
and 784.00–785.50 + 788.10–789.60 for MDA-modified
ubiquitin.
Quantification of adducts formation by RCS by
intact protein analysis
Peak areas were automatically detected and quantified
post-acquisition using Xcalibur Quan Browser. The
percentage of modified ubiquitin obtained upon incu-
bation with every concentration of the different RCS was
computed as:
% modified ubiquitin
AUC modified ubiquitin
¼
AUC unmodified ubiquitin þ AUC modified ubiquitin
ꢃ 100
Protein digestion
Ubiquitin solutions recovered from the filter units (25 ml,
corresponding to about 2 mg of protein) were incubated
with 500 mM sodium borohydride at 37 ꢂC, to stabilize
adducts. After 1 h, 25 ml of LDS sample buffer (2 ꢃ solu-
tion, Bio-Rad, Segrate, Italy) was added to each sample.
The samples were heated for 5 min at 99 ꢂC and loaded
on a Mini-PROTEAN gel (TGX, any Kd, Bio-Rad, Segrate,
Italy) for SDS-PAGE. After 20 min run, the migration was
stopped and the gel was stained using Coomassie blue
(Bio-Safe G-250 Stain, Bio-Rad) accordingly to the
manufacturer protocol. Gel bands corresponding to
ubiquitin (around 8 kDa) were cut, washed with 200 ml
ammonium bicarbonate (50 mM) and digested with 1 mg
of trypsin or Glu-C (both Roche Diagnostics, Monza, Italy)
in 50 mM ammonium bicarbonate buffer. After overnight
incubation at 37 ꢂC, peptides were extracted by 10 min
incubation with extraction solution (acetonitrile/TFA/
water; 30/3/67; v/v/v) and by additional 10 min incuba-
tion with 100% acetonitrile. For each sample, collected
peptides were dried using a vacuum system (Christ,
Identification and localization of protein adducts
The software Proteome Discoverer (version 1.3.0.339,
Thermo Scientific, Waltham, MA) was used to extract
peaks from spectra and to match them to the primary
sequence of bovine ubiquitin (Uniprot entry
spjP0CG53j1-76, i.e. corresponding to amino acids 1-76,
forming the ubiquitin monomer). Accordingly to the
enzyme used for the digestion, trypsin or Glu-C was
selected as proteases, allowing a maximum of three
missed cleavages. Peptide and fragment ion tolerances
were set to 10 ppm and 0.5 Da, respectively. Allowed
variable modifications were methionine oxidation in

FREE RADICAL RESEARCH
331
combination with RCS-specific modification, as follow:
HNE-deriving Michael adducts (+156.11502
variety of RCS-deriving modifications on other nucleo-
philic residues (i.e. lysine, arginine, and histidine), which
would otherwise be masked by the high reactivity of
cysteine [15]. Moreover ubiquitin also represents an
interesting target of RCS-deriving modification, due to
the involvement of lysine residues (e.g. Lys 48) in mono-
and poly-ubiquitination, which regulate important cel-
lular events such as protein degradation, trafficking,
inflammation, and translation.
and +158.13067 for its reduced form) and Schiff bases
(+140.12011, reduced form) on Lys, Arg, and His; GO-
deriving carboxymethyl-Arg and Lys (+58.00548) and
hydro-imidazolone on Arg (+39.99492); MGO-deriving
methyl-hydro-imidazolone (+54.01056), tetra-hydro-pyr-
imidine (+144.04226), and argpyrimidine (+80.02622) on
Arg residues; MGO-deriving carboxyethyl-Arg and Lys
(+72.02113); or MDA-deriving N-(2-propenal)lysine
(+54.01056). As quality filter, only peptide matches
with ‘‘high’’ confidence (FDR ꢄ 0.01) were considered
as genuine identifications. The MS/MS spectra of
modified peptides were requested to match the
expected ions (b and/or y) neighboring the modified
amino acid both at the N- and C-termini, to obtain a
confident mapping of the modification site.
Four RCS belonging to different chemical classes were
selected as modifying agents: the a,b-unsaturated-alde-
hyde HNE, the keto-aldehyde MGO, and the two di-
aldehydes GO and MDA. These RCS were selected since
they are the most abundant and studied endogenous
and exogenous RCS acting as precursor of AGEs
and ALEs.
Ubiquitin carbonylation by RCS: quantification of
the modified forms of the protein
Relative quantification of adduct’s abundance
The signal deriving from RCS-modified peptides was
relatively quantified, as compared to their non-modified
counterpart, by calculating the AUC (area under the
curve) of the extracted ion chromatograms (XIC) for the
adducted and native forms of the same peptide
sequence, using m/z values corresponding to the
either [M + 2H]2+ or [M + 3H]3+, accordingly to the
charge state reported by the Proteome Discoverer-
based identification. The extent of peptide modification
was expressed as percentage using the following
formula:
The first step of the work was to monitor the RCS-
mediated ubiquitin carbonylation and to identify the
main protein adducts by intact protein analysis. For this
purpose, ubiquitin samples were incubated in the
absence and presence of RCS, filtered, diluted in
denaturing conditions, and analyzed by MS as intact
protein. The mass spectrum of control, non-modified
ubiquitin (5 mM concentration) is characterized by a
typical multi-charged ion pattern, with charge states
spanning from +7 to +13, with the +11 peak at m/z
779.61064 representing the most abundant signal, as
shown in Supplementary Figure 1A. After deconvolution
of the spectrum, the experimental average mass of
unmodified ubiquitin is 8564.64 Da (Supplementary
Figure 1B), in agreement with the expected average
mass (8564.74 Da) with an experimental error of 11.6
ppm on the deconvoluted mass value.
% modified peptide
AUC modified peptide
¼
AUC unmodified peptide þ AUC modified peptide
ꢃ 100
When multiple peaks were present in the XIC, the AUC
was calculated for the peak at the retention time
reported by Proteome Discoverer for peptide
identification.
After incubating ubiquitin for 24 h with RCS, the
corresponding MS spectra presented additional multi-
charged ion patterns attributed to the RCS-modified
ubiquitin forms. RCS induced a dose-dependent increase
of ubiquitin modification: the intensity and the number
of the additional peaks increased at increasing concen-
trations of RCS (Figure 1 and Supplementary Figures 2–
4), indicating the presence of different protein variants
generated by protein carbonylation, especially at the
highest concentrations of RCS.
Results and discussion
An in vitro method to generate fully characterized AGEs
and ALEs was set up, which was also able to modulate
the extent of protein carbonylation. Ubiquitin was
selected as model protein for several reasons: (i) it is
commercially available in a pure form and at a low price;
(ii) its molecular weight allows it to be analyzed in intact
form by using the most common MS analyzers, thus
permitting the determination of the relative content of
the modified isoforms; (iii) the absence of cysteine
residues in its sequence allows the observation of a
The percentage of RCS-modified ubiquitin was
determined by analysing the +11 charged ions (Supple-
mentary Figure 1C). Figure 2 and Supplementary Table I
report the percentage of modified ubiquitin obtained
upon 24 h incubations with the different RCS. MGO was
the most reactive RCS since it was able to modify about

332
M. COLZANI ET AL.
Figure 1. Mass spectra of intact ubiquitin acquired after 24 h incubation with increasing concentration of HNE. The molar excess of
HNE in respect to ubiquitin concentration is indicated on the right side of the figure. Excess HNE was removed after the incubation by
ultrafiltration.
necessary to obtain an appreciable amount of carbony-
lated protein. Hence, the order of reactivity was
MGO4GO4HNE4MDA.
Identification of protein adducts by intact protein
analysis
In pursuit of clarity, we focused on the +11 peak as
representative of the different multi-charged ion pat-
terns, since it represents the most intense peak in our
experimental conditions, as already reported in Ref. [13].
Figure 3(A) shows the +11 peak(s) present in the mass
Figure 2. Percentage of modified ubiquitin obtained upon the in
spectra of the unmodified ubiquitin and of the samples
vitro incubation of the proteins (10 mM) with increasing
incubated with an amount of RCS able to modify about
concentration of MGO, GO, HNE, or MDA. The percentage of
50% of the protein. In these mass spectra, the +11 peak
corresponding to unmodified ubiquitin is detectable at
nominal m/z 779, flanked by additional peaks at higher
m/z values, corresponding to ubiquitin variants that
underwent covalent modification by HNE, MGO, GO, and
MDA. All modifications detected on ubiquitin modified
by HNE, GO, MGO, or MDA are summarized in Table I.
Each RCS induced well-defined mass shifts (DM) in
respect to native ubiquitin, as highlighted in the
modified ubiquitin is plotted against the concentration of RCS
(mM). Error bars represent the standard errors estimated by three
independent replicate samples obtained by different
incubations.
50% of ubiquitin when present at a 10-fold molar excess
in respect to ubiquitin. A higher concentration (about a
50-fold excess) of HNE and GO was necessary to reach
the same amount of modified ubiquitin. MDA was the
least reactive RCS, since a 500-fold molar excess was

FREE RADICAL RESEARCH
333
Figure 3. Mass spectra obtained upon in vitro incubation of ubiquitin with: phosphate buffer (control sample), glyoxal (GO),
methylglyoxal (MGO), 2-hydroxy-trans-nonenal (HNE), malondialdehyde (MDA), and the unreactive diketone 2-5-exanedione. Spectra
are reported for selected RCS concentrations that generated about 50% of modified ubiquitin. (A) Focus on the z11 peaks present in
multi-charged mass spectra. (B) Deconvoluted mass spectra, showing the uncharged mass of ubiquitin in its unmodified and RCS-
modified forms.
Table I. Protein adducts corresponding to the mass shifts observed on ubiquitin by MS analysis
of the intact protein.
RCS
Theoretical DM monoisotopic
Adduct type
Residues
Ref.
HNE
GO
MGO
156.11502
58.00548
54.01056
72.02113
144.04226
54.01056
Michael adducts
Carboxymethyl
Methyl-hydro-imidazolone
Carboxyethyl
Tetrahydropyrimidine
N-propenal
KRH
KRC
R
KRC
R
[7]
[16,17]
[18]
[19]
[20]
[21]
MDA
K
Theoretical mass shifts and known target amino acids (described by the literature) are reported. Reactive
carbonyl species legend: HNE, 4-hydroxy-trans-nonenal; GO, glyoxal; MGO, methylglyoxal; MDA,
malondialdehyde.
deconvoluted spectra shown in Figure 3(B) and as
GO-induced modifications – The incubation of ubiqui-
tin with 50-fold molar excess of GO induced
the appearance of two new peaks characterized by DM
of 58.00 and 116.00 Da, indicating the presence of
one and two carboxymethyl modifications on the
protein [16].
described in the following paragraphs.
HNE-induced modifications – After incubating ubiqui-
tin with a 50-fold molar excess of HNE, the deconvoluted
mass spectrum presented a new peak at 8720.74 Da,
characterized by a DM of 156.11 Da in respect to the
native form of the protein. Consistently, the +11 peak in
the original mass spectrum is shifted by 14.2 m/z.
The new peak was attributed to the Michael adduct of
HNE [7].
MGO-induced modifications – 10-fold molar excess of
MGO generated three additional main peaks, showing
DM equal to 54.00, 72.01, and 144.03 Da. These values
correspond to the known adducts methyl-

334
M. COLZANI ET AL.
Table II. Amino acids targeted by HNE, MGO, GO, and MDA on
ubiquitin sequence, as detected by the analysis of peptides
obtained from the digestion of the ubiquitin samples shown in
Figure 3.
peptides bearing a variety of modifications on lysine,
arginine, or histidine residues, as reported in Table II.
In general, HNE and MDA preferentially targeted
lysine residues, while MGO and GO mainly targeted
arginine residues. In more detail, HNE-deriving Michael
adducts were localized on lysine residues K6, K11,
K33, K48, and K63, and on the histidine residue H68.
The HNE-deriving Schiff base was found on K6; this
modification was not detected in the intact protein,
likely because it is a reversible modification that can be
observed only after its stabilization.
RCS
Adduct type
Targeted amino acids
HNE
Michael adducts
K6, K11, K33, K48, K63,
H68
K6
Schiff base
MGO
Methyl-hydro-imidazolone
Carboxyethyl
Argpyrimidine
R42, R54, R72, R74
R54, R72
R72
GO
Carboxymethyl
Hydro-imidazolone
N-propenal
K6, R42, R54
R72
K6, K11, K27, K29, K33,
K48, K63
MDA
The GO-deriving carboxymethyl adducts, also identi-
fied on the intact protein, were localized on the arginine
residues R42 and R54 and on lysine K6. Additionally,
the GO-deriving imidazolone derivative was identified
on R72.
hydroimidazolone [18], carboxyethyl [19] and two
carboxyethyl and/or tetrahydropyrimidine [20] (isobaric
modifications), respectively.
The MGO-deriving carboxyethyl (already detected on
the intact protein) was localized on R54 and R72; the
methyl-hydroimidazolone adduct was observed on the
same amino acids and additionally on R42 and R74.
Additionally, the MGO-deriving argpyrimidine was
detected in R72. Based on these results we can consider
that the peak at m/z 792 visible in Figure 3 probably
corresponds to a double carboxyethyl adduct and not to
the tetrahydropyrimidil-arginine, since the latter was not
identified by the bottom-up analysis. In any case, it is not
possible to exclude the presence of tetrahydropyrimidil-
arginine, due to its isobaricity with the double carbox-
yethyl adduct. The MDA-deriving adduct N-propenal was
detected in all seven lysine residues present in the
ubiquitin sequence.
MDA-induced modifications – MDA (500-fold molar
excess) generated two peaks with a fixed mass incre-
ment equal to 54.01 Da. This value corresponds to two
N-(2-propenal)lysine adducts [21].
To test whether the protein adducts were artefacts
generated by the ionization process or resulted from
non-covalent interactions, ubiquitin was incubated with
hexane-2,5-dione, a diketone which does not act as RCS.
Unlike the samples incubated with HNE, GO, MGO, or
MDA, no additional peaks were observed by incubating
ubiquitin with hexane-2,5-dione even at the highest
concentrations (Figure 3 and Supplementary Figure 5).
Localization of RCS-deriving adducts on ubiquitin
sequence
Analysis of the preferentially modified
amino acids
In order to confirm the presence of the modifications
observed in the intact protein and to map the amino
acids targeted by the different RCS, aliquots of the
ubiquitin samples shown in Figure 3 were analyzed
The intensities of the signals corresponding to the
unmodified and RCS-modified peptides were quantified,
in order to identify the amino acids preferentially
modified by each RCS. To do so, the area under the
curve (AUC) of the ion chromatograms extracted for the
modified and unmodified forms of the peptide (XIC)
were determined and the results reported as a percent-
age of the modified peptide.
using
a bottom-up approach. The samples were
stabilized with sodium borohydride and split into two
halves that were digested using trypsin and Glu-C,
respectively. The peptides resulting from the two types
of digestion were analyzed by nanoLC-MS in parallel, in
order to increase the sequence coverage. The identifi-
cation of the RCS-modified peptides was obtained by
matching the acquired spectra to the ubiquitin
sequence, allowing the modifications observed on the
intact proteins (Table I). Moreover, additional adducts
described in the literature (listed in the Materials and
Methods section) were allowed as possible peptide
modification. Quality filters were applied to obtain
genuine peptide identification and to unambiguously
map the modified residues on the basis of fragmentation
spectra. By applying these criteria, we identified different
Supplementary Table II lists, for each RCS-modified
peptide, the modification site and its relative abundance
in respect to the corresponding unmodified form.
Peptides originating from Glu-C digestion were prefer-
entially considered in respect to those generated by
trypsin, since RCS-modified lysine and arginine appeared
to affect the efficiency of trypsin digestion on those
residues. As a consequence, in the sample digested with
trypsin, peptides modified on their C-terminus were not
detected in both their unmodified and modified forms,

FREE RADICAL RESEARCH
335
Figure 4. Localization of the amino acids preferentially modified by the four RCS. The percentage of modification for each amino acid
is color-coded, accordingly to the legend. Adducts legend: MA, Michael addition; SB, Schiff base; MGH, MGO-deriving methyl-
hydroimidazolone; CE, carboxyethyl; Arg-pyr, argpyrimidine; CM, carboxymethyl; GH, GO-deriving hydroimidazolone; NPK, N-propenal-
lysine.
preventing relative quantification. Peptides were quan-
tified as tryptic peptides only when the modifications
involved an internal amino acid and when they were not
quantified as Glu-C-deriving peptides.
efficient analyses of the structure–activity relationship.
For instance, the role of each specific RCS on the
biological activity of the resulting AGEs/ALEs is not well
defined yet, as well as the impact of the extent of
protein modification on cellular and tissue toxicity. The
characterization of the different types of AGEs/ALEs is
critical not only to better understand the biological
effects of RCS, but also to identify novel drug targets.
The method reported here can be easily adapted to
other proteins to create other well-characterized AGEs
and ALEs. As example of their application, standardized
AGEs and ALEs can be tested as ligand of the receptor
for AGEs (RAGE). This would allow comparison of the
affinity of the RAGE receptor for the same protein
bearing different RCS-deriving modifications, thus allow-
ing the detection of the moiety of AGEs/ALEs respon-
sible of the ligand:receptor interactions – an information
that is currently missing. The complete structure char-
acterization provides a solid background to study the
structure–activity relationship and to evaluate the bio-
logical activity of the modified moieties arising from the
condensation of the different RCS with a target protein.
In conclusion, we report an in vitro method for the
generation of fully characterized AGEs and ALEs, where
the carbonylation extent can be modulated and
determined. The method is highly reproducible and
represents an important tool to reach a better under-
standing of the structure–activity relationship of AGEs
and ALEs.
The preferential sites of modification are summarized
in Figure 4. H68 represents the amino acid preferentially
targeted by HNE (Michael addition). GO preferentially
generated the carboxymethyl adduct on R42. MGO was
able to react with different amino acids and to form
different adducts, but overall it preferentially generated
the methyl-hydroimidazolone adducts on arginine R42,
R72, and R74; R72 was also modified at relatively high
amounts as carboxyethyl-arginine. MDA preferentially
modified the residues K6 and K29 (N-propenal-lysine).
The adducted moieties detected in these experiments
are in agreement with previous observations; all of them
have already been reported as major AGEs/ALEs in vitro
or in vivo [7,10,13,16,17,21–25]. No modification of lysine
residues was detected for MGO, unlike previous reports
[19]. Moreover, cross-linked forms of ubiquitin (e.g.
MOLD, GOLD, Lys-Lys enamine) were not detected in
the acquired spectra. However, it is possible that protein
aggregates were present in some samples, but not
detected by our instrumental setup due to their high
mass values. Notably, protein precipitation was not
observed in any sample and ubiquitin multimers
were not observed by SDS-PAGE analysis of aliquots
of different ubiquitin samples incubated with RCS (data
not shown), indicating that protein aggregation was
either absent or rare in our experimental conditions.
Acknowledgements
G.A. acknowledges the colleagues of the COST action whose
fruitful discussion led to the conception of this paper. M.C.
received a post-doctoral fellowship (Dote Ricerca: FSE, Regione
Lombardia) enabling her to carry the present investigation.
Conclusions
Several pieces of evidence indicate that both endogen-
ous and exogenous AGEs and ALEs are involved in the
pathogenic mechanisms of different diseases and that
they are important modulators of toxic responses.
We believe that a great limit to research in this field is
the lack of well-characterized standards allowing
Declaration of interest
The authors declare no conflicts of interest.

336
M. COLZANI ET AL.
[14] Keller BO, Sui J, Young AB, Whittal RM. Interferences and
contaminants encountered in modern mass spectrom-
etry. Anal Chim Acta 2008;627:71–81.
[15] Schaur R. Basic aspects of the biochemical reactiv-
ity of 4-hydroxynonenal. Mol Aspects Med 2003;24:149–
159.
[16] Reddy S, Bichler J, Wells-Knecht KJ, Thorpe SR, Baynes JW.
N epsilon-(carboxymethyl)lysine is a dominant advanced
glycation end product (AGE) antigen in tissue proteins.
Biochemistry 1995;34:10872–10878.
[17] Iijima K, Murata M, Takahara H, Irie S, Fujimoto D.
Identification of N(omega)-carboxymethylarginine as a
novel acid-labile advanced glycation end product in
collagen. Biochem J 2000;347:23–27.
References
[1] Dalle-Donne I, Giustarini D, Colombo R, Rossi R, Milzani A.
Protein carbonylation in human diseases. Trends Mol
Med 2003;9:169–176.
[2] Barlovic DP, Soro-Paavonen A, Jandeleit-Dahm KAM.
RAGE biology, atherosclerosis and diabetes. Clin Sci
(Lond) 2011;121:43–55.
[3] Vistoli G, Maddis De D, Cipak A, Zarkovic N, Carini M,
Aldini G. Advanced glycoxidation and lipoxidation end
products (AGEs and ALEs): an overview of their mechan-
isms of formation. Free Radic Res 2013;47:3–27.
[4] Januszewski AS, Alderson NL, Metz TO, Thorpe SR, Baynes
JW. Role of lipids in chemical modification of proteins and
development of complications in diabetes. Biochem Soc
Trans 2003;31:1413–1416.
[5] Aldini G, Dalle-Donne I, Colombo R, Maffei Facino R,
Milzani A, Carini M. Lipoxidation-derived reactive car-
bonyl species as potential drug targets in preventing
protein carbonylation and related cellular dysfunction.
ChemMedChem 2006;1:1045–1058.
[18] Lo TWC, Westwood ME, Mclellan AC, Selwood T,
Thornalleys PJ. Binding and modification of proteins by
methylglyoxal under physiological conditions. A kinetic
and mechanistic study with N alpha-acetylarginine, N
alpha-acetylcysteine, and
N alpha-acetyllysine, and
bovine serum albumin. J Biol Chem 1994;269:32299–
32305.
[19] Ahmed MU, Brinkmann Frye E, Degenhardt TP, Thorpe SR,
Baynes JW. N-epsilon-(carboxyethyl)lysine, a product of
the chemical modification of proteins by methylglyoxal,
increases with age in human lens proteins. Biochem J
1997;324:565–570.
[6] Aldini G, Vistoli G, Stefek M, Chondrogianni N, Grune T,
Sereikaite J, et al. Molecular strategies to prevent, inhibit
and degrade advanced glycoxidation and advanced
lipoxidation end products. Free Radic Res 2013;47:93–
137.
[20] Oya T, Hattori N, Mizuno Y, Miyata S, Maeda S, Osawa T,
Uchida K. Methylglyoxal modification of protein: chem-
ical and immunochemical characterization of methyl-
glyoxal-arginine adducts. J Biol Chem 1999;274:18492–
18502.
[21] Uchida K, Sakai K, Itakura K, Osawa T, Toyokuni S.
Protein modification by lipid peroxidation products:
formation of malondialdehyde-derived N(epsilon)-(2-
propenal)lysine in proteins. Arch Biochem Biophys
1997;346:45–52.
[22] Thornalley PJ, Battah S, Ahmed N, Karachalias N, Agalou
S, Babaei-Jadidi R, Dawnay A. Quantitative screening of
advanced glycation endproducts in cellular and extracel-
lular proteins by tandem mass spectrometry. Biochem J
2003;375:581–592.
[23] Henle T, Walter AW, Haeßner R, Klostermeyer H.
Detection and identification of a protein-bound imida-
zolone resulting from the reaction of arginine residues
[7] Sayre LM, Lin D, Yuan Q, Zhu X, Tang X. Protein adducts
generated from products of lipid oxidation: focus on HNE
and one. Drug Metab Rev 2006;38:651–675.
[8] Valencia JV, Mone M, Koehne C, Rediske J, Hughes TE.
Binding of receptor for advanced glycation end products
(RAGE) ligands is not sufficient to induce inflammatory
signals: lack of activity of endotoxin-free albumin-derived
advanced glycation end products. Diabetologia
2004;47:844–852.
[9] Ott C, Jacobs K, Haucke E, Navarrete Santos A, Grune T,
Simm A. Role of advanced glycation end products in
cellular signaling. Redox Biol 2014;2:411–429.
[10] Cotham WE, Metz TO, Ferguson PL, Brock JWC, Hinton
DJS, Thorpe SR, et al. Proteomic analysis of arginine
adducts on glyoxal-modified ribonuclease. Mol Cell
Proteomics 2004;3:1145–1153.
[11] Uchida K, Stadtman ER. Modification of histidine residues
in proteins by reaction with 4-hydroxynonenal. Proc Natl
Acad Sci USA 1992;89:4544–4548.
[12] Rees MS, van Kuijk FJGM, Siakotos AN, Mundy BP.
Improved synthesis of various isotope labeled 4-hydro-
xyalkenals and peroxidation intermediates. Synth
Commun 1995;25:3225–3236.
and methylglyoxal.
1994;199:55–58.
Z
Lebensm Unters Forsch
[24] Klo¨pfer A, Spanneberg R, Glomb MA. Formation of
arginine modifications in a model system of Na-tert-
butoxycarbonyl (Boc)-arginine with methylglyoxal. J Agric
Food Chem 2011;59:394–401.
[13] Colzani M, Criscuolo A, De Maddis D, Garzon D, Yeum K-J,
Vistoli G, et al. A novel high resolution MS approach for
the screening of 4-hydroxy-trans-2-nonenal sequestering
agents. J Pharm Biomed Anal 2014;91:108–118.
[25] Shipanova IN, Glomb MA, Nagaraj RH. Protein modifica-
tion by methylglyoxal: chemical nature and synthetic
mechanism of a major fluorescent adduct. Arch Biochem
Biophys 1997;344:29–36.