RESEARCH ARTICLE
Analysis of protein glycation using phenylboronate
acrylamide gel electrophoresis
Marta P. Pereira Morais1, Julia D. Mackay1, Savroop K. Bhamra1,
J. Grant Buchanan2, Tony D. James2, John S. Fossey3 and Jean M. H. van den Elsen1
1 Department of Biology and Biochemistry, University of Bath, Bath, UK
2 Department of Chemistry, University of Bath, Bath, UK
3 School of Chemistry, University of Birmingham, Edgbaston, Birmingham, UK
Received: April 28, 2009
Revised: September 24, 2009
Accepted: September 29, 2009
The incorporation of the specialized carbohydrate affinity ligand methacrylamido phenyl-
boronic acid in polyacrylamide gels for SDS-PAGE analysis has been successful for the
separation of carbohydrates and has here been adapted for the analysis of post-translationally
modified proteins. While conventional SDS-PAGE analysis cannot distinguish between
glycated and unglycated proteins, methacrylamido phenylboronate acrylamide gel electro-
phoresis (mP-AGE) in low loading shows dramatic retention of d-gluconolactone modified
proteins, while the mobility of the unmodified proteins remains unchanged. With gels
containing 1% methacrylamido phenylboronate, mP-AGE analysis of gluconoylated recom-
binant protein Sbi results in the retention of the modified protein at a position expected for a
protein that has quadrupled its expected molecular size. Subsequently, mP-AGE was tested
on HSA, a protein that is known to undergo glycation under physiological conditions. mP-
AGE could distinguish between various carbohydrate-protein adducts, using in vitro glycated
HSA, and discriminate early from late glycation states of the protein. Enzymatically glyco-
sylated proteins show no altered retention in the phenylboronate-incorporated gels, rendering
this method highly selective for glycated proteins. We reveal that a trident interaction between
phenylboronate and the Amadori cis 1,2 diol and amine group provides the molecular basis of
this specificity. These results epitomize mP-AGE as an important new proteomics tool for the
detection, separation, visualization and identification of protein glycation. This method will
aid the design of inhibitors of unwanted carbohydrate modifications in recombinant protein
production, ageing, diabetes, cardiovascular diseases and Alzheimer’s disease.
Keywords:
Carbohydrate structure / Electrophoresis / Glycoproteins / Glycoproteomics / Protein
adducts / Serum proteins
1 Introduction
Covalent glycosylation occurs in a significant set of proteins
in eukaryotic cells. Both N-glycosylation and O-glycosylation
of proteins contain remarkable complexity in the oligo-
saccharide chains that are attached in their transit through
the endoplasmic reticulum and Golgi. Multiple glycoforms
of the same protein not only add to the diversity and
complexity of the glycoproteome but these carbohydrate
modifications can control a protein’s localization, turnover
and active state and also manipulate their three-dimensional
structure and interactions with other proteins [1]. Undesired
carbohydrate modifications, however, may also occur in the
form of glycation, where sugar molecules, such as glucose,
lactose or fructose, covalently bind to a protein or lipid
molecule without the controlling action of an enzyme.
Abbreviations: 6PGL, 6-phosphatogluconolactone; AGE,
advanced glycation endproduct; AGH, aminoguanidine hydro-
chloride; mP-AGE, methacrylamido phenylboronate acrylamide
gel electrophoresis; MPBA, methacrylamido phenylboronic acid These authors contributed equally to this work.
Correspondence: Dr. Jean M. H. van den Elsen, Department of
Biology and Biochemistry, University of Bath, Bath, BA2 7AY, UK
E-mail: J.M.H.V.Elsen@bath.ac.uk
& 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.proteomics-journal.com
48 Proteomics 2010, 10, 48–58DOI 10.1002/pmic.200900269

During glycation the carbonyl group on the open-chain
carbohydrate reacts with amino groups of lysine, arginine or
N-terminal amino acid residues of the protein to form an
unstable Schiff base intermediate, followed by an Amadori
rearrangement to form a stable keto-amine. The reaction
reaches equilibrium after about 28 days and the extent of
glycation will therefore be affected by the half-life of the
protein [2]. Subsequently, the Amadori product can undergo
a number of chemical reactions to form advanced glycation
endproducts (AGEs), consisting of a broad range of hetero-
geneous fluorescent and yellow-brown products, including
nitrogen-containing and oxygen-containing heterocycles,
resulting from subsequent oxidation, dehydration, cycliza-
tion and condensation reactions with other reactive amino
groups [3, 4]. AGEs are important long-term biomarkers for
ageing and age-related chronic disease states such as
diabetes, cardiovascular diseases, Alzheimer’s disease,
autoimmune disease and cancer [5–10].
Endogenous as well as exogenous AGEs have been
implicated in disease states [11, 12]. These dietary or pre-
formed AGEs can result from glycation by numerous sugars,
including monomeric ketoses and aldoses or by dimeric and
polymeric sugars. Many of these have been little studied
compared with glucose. The oxidized glucose derivative
d-gluconolactone, for instance, has recently been shown to
adversely affect the quality of recombinant proteins. Sponta-
neous a-N-6-Phosphogluconoylation has been described in
recombinantly expressed proteins fused to a histidine affinity
tag [13–15]. 6-phosphatogluconolactone (6PGL), an inter-
mediate of the pentose phosphate pathway, which is
produced by glucose-6-phosphate dehydrogenase, is a potent
electrophile that reacts with the N-terminal amino group of
histidine-tagged proteins forming amine-linked products [15].
This modification has been shown to have effects on protein
activity [16] and interferes with crystallization of proteins [17].
It may also impair the structure of the expressed protein and
increase its immunogenicity, which would greatly obstruct
the use of recombinantly produced histidine-tagged proteins
in research, diagnostics and therapy.
Although glycation in proteins has not been well char-
acterized in proteomics [18], the detection identification and
quantitation of Amadori and AGE adducts is in great
demand and of great clinical importance in studies of age-
related chronic disease states and for the use of recombinant
proteins in therapy. Although early glycation products can
be identified and purified using aminophenylboronate affi-
nity chromatography, this method can be hampered by the
co-elution of early AGEs and Amadori products with some
enzymatically glycosylated proteins, including human
immunoglobulins, complement proteins C3 and C4 [19],
human platelet glycocalicin and membrane glycoproteins
from human lymphocytes (see [20] and references therein).
Here we present a simple technique, based on a widely used
method for protein analysis, PAGE, for the analysis of early
as well as late non-enzymatic carbohydrate modifications in
proteins. The presented methacrylamido phenylboronate
acrylamide gel electrophoresis (mP-AGE) method, initially
designed for the separation of carbohydrates [21] has here
been adapted to specifically detect, analyze and separate
early (Amadori) as well as long-term (AGE) glycation
products from unaffected proteins.
2 Materials and methods
2.1 Glyconoylation of Sbi-III-IV
Sbi-III-IV was freshly expressed and purified as described
previously [22] and 10mg/mL of protein was incubated with
100mM freshly prepared D-(1)-gluconic acid d-lactone (Sigma
Aldrich) in PBS, in a water bath at 371C for 15min to 16h.
2.2 Glycation of HSA
Minimally glycated HSA [23] was obtained by incubating
HSA (20mg/mL; Sigma Aldrich, St Louis, MO, USA) with
50mM glucose, fructose, lactose, mannose, maltose or
glucose-6-phosphate in PBS, in a water bath at 371C for up
to 55 days under aseptic conditions (in the presence or
absence of 50mM aminoguanidine hydrochloride (AGH)).
Formation of fluorescent AGE adducts during the in vitro
glycation was followed by means of fluorescence spectro-
scopy. Fluorescence spectra from incubated samples were
obtained between 400 and 550 nm, using an excitation
wavelength of 370 nm (see [24]).
2.3 Preparation of methacrylamido phenylboronic
acid (MPBA) polyacrylamide gels
MPBA (Fig. 1A) was synthesized as described recently
[21, 25].
Polyacrylamide electrophoresis gels used for protein
separation are commonly a hydrogel assembled by co-poly-
merizing acrylamide with the cross-linker methylene bis-
acrylamide. Acrylamide gels incorporating phenylboronate
were easily prepared by adding a small percentage (0–1%) of
MPBA to the electrophoresis gel preparation solution prior
to polymerization. Polyacrylamide resolving gels were
polymerized in the absence or presence of MPBA in
concentrations ranging from 0–1% by mixing MPBA
powder with an 8% acrylamide solution (from 40% stock
solution of acrylamide:bis-acrylamide, 29:1; Fisher Scien-
tific, Fair Lawn, NJ, USA) in 40mM Tris buffer at pH 8.8
and cast in a gel casting cassette (height: 100mmwidth:
100mm thickness: 0.75mm; Invitrogen, Carlsbad, CA,
USA). After polymerization of the resolving gel, using 10%
ammonium persulfate (Sigma Aldrich) and TEMED (Sigma
Aldrich) the stacking gel, containing no boronic acid, was
prepared with 10% acrylamide (from 40% stock solution of
acrylamide:bis-acrylamide, 29:1; Fisher Scientific) in 10mM
Proteomics 2010, 10, 48–58 49
& 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.proteomics-journal.com

improving the separation of saccharides, it was reasoned
that inclusion of boronic acid saccharide receptors that
reversibly interact with saccharides could also be used to
analyze carbohydrates attached to proteins and may advan-
tageously affect their retention characteristics. Again we
have chosen a phenyl boronic acid because of its capacity to
function as saccharide receptors in aqueous solution, as
previously attested by the many sensory systems reported
[26–30], thereby forming cyclic boronic esters with a variety
of carbohydrates under equilibrium conditions, via rever-
sible covalent interactions in aqueous media (see also [21]).
3.1 Analysis of non-enzymatically modified proteins
using mP-AGE
3.1.1 Gluconoyl adducts in recombinant proteins
For the analysis of d-gluconolactone modifications in mP-
AGE, we used a protein construct based on Staphylococcus
aureus immune-subversion protein Sbi. This protein has
been shown to inhibit the innate immune system [22] and is
currently being developed as a therapeutic for complement-
mediated acute inflammatory diseases. The Sbi-III-IV
construct of this protein has a 25-residue N-terminal tag
with sequence MSYHHHHHHDYDIPTTENLYFQGAM
and MS analysis of similar constructs containing this tag
have shown that this sequence is specifically prone to 6-
phosphogluconoylation. So far these undesired N-terminal
adducts could only be detected by MS analysis of the protein
and reveal itself as a 258Da mass addition, representing
6PGL, and/or a 178Da d-gluconolactone adduct, a result of
de-phosphorylation of 6PGL [15].
Figure 1A shows mass spectra of freshly purified
recombinant Sbi-III-IV (16527Da, left) and Sbi-III-IV with
exogenously added d-gluconolactone (100mM) confirming
the presence of gluconoylated protein (16 705 kDa, right) in
addition to the expected unmodified protein (see Fig. 1B for
the structure of the N-terminal d-gluconolactone adduct). No
6PGL adduct could be detected.
Figure 1D compares a normal SDS-PAGE profile with a
mP-AGE gel both showing the migration profile of the
mixture, analyzed in Fig. 1A, of unmodified and glyconoy-
lated Sbi-III-IV (lane 1) after expression in Escherichia coli
and purification using nickel-ion chelating chromatography
as described previously [22]. Lanes 2 and 3 show migration
profiles of Sbi-III-IV with exogenously d-gluconolactone
(100mM) and incubated for 15min and 16 h, respectively.
Under normal SDS-PAGE conditions the gluconoylated Sbi-
III-IV fraction in the freshly expressed and purified protein
cannot be distinguished from the unglycated protein, even
when large amounts of protein (7.5 mg) are loaded onto the
gel as shown in Fig. 1D. In the lanes containing Sbi-III-IV
incubated with d-gluconolactone, a faint shadow band
appears just above the 17 kDa Sbi-III-IV band that may hint
the presence of the modified protein.
In contrast, the migration profile in the MPBA (see Fig. 1B
for MPBA structure) incorporated gel shows a dramatic
separation of modified and unmodified proteins, with
the boronate affinity greatly affecting the mobility of the
gluconoylated Sbi-III-IV, retaining it at a position expected
for a protein that has quadrupled its expected molecular
size. While MicrOTOF analysis failed to detect the gluco-
noylated Sbi-III-IV peak in the fresh sample (Fig. 1A, left
panel) the modified protein could be detected in mP-AGE
(Fig. 1D, lane 4) using less than half the amount of total
protein (7.5mg of Sbi-III-IV was loaded in mP-AGE analysis,
20mg in MicrOTOF analysis). However, by using ultra-
sensitive MS methods such as MS/MS, with sensitivities in
the pico/femto molar range, such modifications would have
been readily detected.
While the majority of the molecular weight markers used
in the gels presented in Fig. 1D are not affected in their
mobility by the presence of the MPBA monomer in the gel
(see Supporting Information Fig. 1 for detailed analysis), the
retention of gluconoylated Sbi-III-IV is strongly correlated
with the concentration of MPBA incorporated in the gel
(0; 0.05; 0.1; 0.16; 0.5 and 1%), as is shown in Fig. 2A and B,
with the highest degree of retention observed in the 1%
MPBA gel. Even at concentrations as low as 0.16% of
incorporated MPBA, gluconoylated proteins are retained in
the gel at a virtual molecular size twice their actual size (Fig.
2B) while the electrophoresis profile of unmodified Sbi-III-
IV remains unchanged.
The gluconoylated protein band was identified as Sbi in a
Western blot analysis (not shown) and Supporting Infor-
mation Fig. 2A shows that the gluconoylation site is indeed
located in the N-terminal histidine tag. In Supporting
Information Fig. 2B it is evidenced that the detection and
separation of d-gluconolactone modified and unmodified
protein can also be achieved in other recombinantly
expressed proteins (Sulfolobus sulfataricus E2 enzyme, SSE)
containing this specific modification (Dr. Karl Payne,
personal communication).
With the need for control of post-translational gluconoyla-
tion in recombinant proteins becoming more significant in
the production of proteins of pharmaceutical and medical
applications [31], mP-AGE analysis could provide an impor-
tant tool for easy detection and separation of d-gluconolactone
modified proteins and for the development of methods and
substances that can prevent this modification.
Although the reason why d-gluconoylation specifically
targets N-terminal histidine metal-affinity tags in recombi-
nant proteins is not yet known, recent studies using
b-amyloid peptide oligomers suggest that perhaps histidine
tag-bound metal ions could be involved in the acceleration of
this process. b-Amyloid deposits, the hallmarks of Alzhei-
mer’s disease, contain both sugar-derived AGEs and copper
ions, and it has been shown in vitro that the formation of
covalently cross-linked high-molecular-mass a-amyloid
peptide oligomers, using synthetic b-amyloid peptide and
glucose or fructose, is accelerated by micromolar amounts of
Proteomics 2010, 10, 48–58 51
& 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.proteomics-journal.com

copper (and iron) ions [32]. mP-AGE analysis of these
proteins may therefore also provide an ideal method for the
analysis of b-amyloid formation and design new inhibitors
and drugs that can remove these undesired adducts.
3.1.2 Glycation of HSA
Human blood proteins like hemoglobin [33] and serum
albumin [34] undergo slow glycation, mainly by the formation
of a Schiff base between glucose molecules in the blood and
amino groups of lysine, arginine or N-terminal amino acid
residues of these proteins. After the formation of this labile
Schiff-base the glucose adduct undergoes an Amadori rear-
rangement to form a fructosamine. We chose the most
abundant serum protein HSA to investigate if these early
glycation products could be identified using mP-AGE. Figure
3 compares the SDS-PAGE migration profiles of sugar incu-
bated HSA (21 days at 371C) in a normal acrylamide gel
(Fig. 3A) and an MPBA incorporated gel (Fig. 3B). While the
normal gel shows an expected profile for HSA with a single
protein band at approximately 67 kDa in all samples, apart
from multimeric HSA bands and a faint higher molecular
weight band identified as transferrin, a common contaminant
in HSA purifications [35]. In the mP-AGE gel profile an
additional HSA band can be observed in all samples from day
1. As can be seen in the PBS incubated control sample, (Fig.
3A, lane 1) this extra band runs at a ‘‘molecular weight’’ of
80kDa ( ) on a 0.2% mP-AGE gel. Interestingly, while the
PBS, fructose and glucose 6-phosphate mP-AGE profiles do
not change over time, only the glucose- and lactose-incubated
HSA samples (lanes 2 and 4) show a shift of this additional
band to a higher position in the gel (90kDa; ), accom-
panied by an increase in intensity and broadening. This
change of the HSA profile in the glucose and lactose samples
in the mP-AGE gels can already be observed from day 6 after
incubation, starting with a slight retention of the lower gel
band (67 kDa; ) and increased intensity of the 80kDa
band (see Supporting Information Fig. 3, day 14; ).
3.1.2.1 HSA-Amadori adduct
The increase in intensity of the protein band running at
80 kDa ( ), after incubation with glucose and lactose, hints
to the identity of this additional band in the mP-AGE profile
of normal HSA: naturally glycated HSA. This (fructosa-
mine) Amadori product (Fig. 5A, right) results from the
reaction between glucose in the blood and amino groups
from HSA and separation of glycated and non-glycated HSA
using phenylboronate has been observed before with boro-
nate chromatography [20]. MicrOTOF MS analysis of the
HSA used in our experiments confirms the presence of
glycated protein. Fatty acid-free HSA had a mean molecular
weight of 66 420Da. In addition to native HSA we also
identified a peak corresponding to (naturally) glycated HSA
with a molecular weight of 67 954.5Da. This peak lies
between molecular masses found for minimally glycated
HSA (66 845Da) and highly glycated HSA (73 226Da)
described by Thornalley and co-workers [23]. In our mP-
AGE system the retention of this fructosamine HSA adduct
is significantly less than is observed with the linear gluco-
nolactone adduct in Sbi. The observed difference in gel
mobility therefore implies that the fructosamine-HSA band
is more likely to contain the cyclic hemiacetal form of the
fructosamine Amadori product, rather than the linear
(unstable) Schiff base or the acyclic Amadori product (Fig.
5A, middle).
3.1.2.2 HSA-AGE adducts
In the first days after incubation with glucose and lactose
(from day 3 onwards) the glycated HSA band (80 kDa in a
0.2% mP-AGE; ) intensifies and occupies a position
corresponding to a molecular weight of 90 kDa ( ). This
early stage process appears not to be affected by the
presence of AGH, a glycation inhibitor that is known to
impinge on the early glycosylation Amadori product
[36] (Supporting Information Fig. 3, day 14, far top right
panel)
In time, the retention of the 90 kDa band ( ) increases
occupying a position around 100kDa ( ), intensifies and
relativevirtual
BA
MWMW
70
80
4
4.5
40
50
60
2
2.5
3
3.5
10
20
30
0.5
1
1.5
MPBA monomer concentration (%) MPBA monomer concentration (%)
0 0
0 0.2 0.4 0.6 0.8 1 1.20 0.2 0.4 0.6 0.8 1 1.2
Figure 2. Retention of gluco-
noylated Sbi-III-IV in SDS
PAGE as a function of MPBA
content. The position of the
glyconoylated Sbi-III-IV band is
expressed as virtual molecular
weight (as compared with the
molecular weight marker gel
profile; A) and as re-
lative molecular weight (as
compared with the actual
molecular weight of gluconoy-
lated Sbi-III-IV; B).
52 M. P. Pereira Morais et al. Proteomics 2010, 10, 48–58
& 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.proteomics-journal.com

broadens while the lower band (67 kDa; ) completely
disappears around HSA’s turnover time, day 35. This is best
seen in Supporting Information Fig. 3 (day 42), where in a
0.5% MPBA gel this new HSA glycation form ( ) is sepa-
rated from unglycated HSA ( ), the natural HSA Amadori
product ( ), and in vitro glycated fructosamine-HSA ( ).
This late stage glycated HSA form ( ) is not seen in the AGH
containing samples. Since AGH inhibits the formation of
AGEs but not the formation of Amadori products we conclude
this glycated HSA form ( ) contains certain AGE adducts,
with higher affinity for phenylboronate.
3.2 Analysis of other post-translational
modifications using mP-AGE
After the successful identification and separation of some
early and late glycation protein products, we decided to also
analyze the effect of gel-incorporated boronate in mP-AGE
on the mobility of other post-translational modifications,
including phosphorylation, glycosylation and combinations
thereof. In Fig. 4 are shown the SDS-PAGE relative mobi-
lities of b-casein (phosphorylated/not glycosylated), human
hemoglobin (not phosphorylated/not glycosylated), chicken
ovalbumin (phosphorylated/glycosylated), human comple-
ment fragment C4c and human b2-glycoprotein I (not
phosphorylated/glycosylated). A minor shift in relative
mobility can be observed only with phosphorylated proteins
b-casein and ovalbumin in the 0.5% MPBA gel, when
compared with their positions in the control gel and those of
non-glycosylated hemoglobin in both gels. In contrast with
boronate affinity chromatography no significant retention of
enzymatically glycosylated proteins (C4c or b2-glycoprotein
I) was observed in mP-AGE. Boronate affinity chromato-
graphy is used to purify both glycated as well as normal
glycosylated proteins based on the presence of cis diols in
both proteins [37]. Our results indicate that the presence of
boronate in SDS-PAGE is specific for the detection of
glycation, without significant interference caused by post-
translational modifications such as phosphorylation or
glycosylation.
3.3 Molecular basis of mP-AGE selectivity for
glycated proteins
Our results indicate that mP-AGE is an ideal technique to
identify and separate early and late glycation products
especially in proteins containing gluconolactone and
glucose adducts. However, the lack of retention in the
phenylboronate containing gels of the fructose and glu-
cose-6-phosphate incubated HSA samples appears to
contradict previous findings reporting not only a similar
(glucose and fructose) [38] or even enhanced rate of
glycation (fructose and glucose-6-phosphate) [39, 40]
under physiological conditions but also a greater increase
in the fluorescence intensity of AGE formation (fructose)
[41, 42]. We indeed confirm a significantly greater (3-fold)
increase in fluorescence in the fructose incubated samples
compared to the glucose incubated HSA (not shown).
Below we present a model that provides an explanation
for our contradictory findings in the behavior of HSA in
mP-AGE analysis after exposure to different carbohydrates
(see Fig. 5):
116.0
66.2
45.0
1 2 3M 4 5 1 2 3M 4 5
0% MPBA 0.2% MPBA
6 7 8M 9
116.0
66.2
45.0
6 7 8M 9
+ AGH + AGH
A B
Figure 3. mP-AGE analysis of glycated HSA. SDS-PAGE analysis comparing gel profiles in the absence (A) and presence of MPBA (0.2%, B)
of HSA incubated for 21 days with PBS buffer (lane 1), glucose (lane 2), fructose (lane 3), lactose (lane 4) and glucose-6-phosphate (lane 5),
glucose with glycation inhibitor aminoguanidine hydrochloride (AGH, lane 6), fructose with AGH (lane 7), lactose with AGH (lane 8) and
glucose-6-phosphate with AGH (lane 9). The proposed positions of non-glycated HSA ( ; 67 kDa), naturally glycated Amadori-HSA ( ;
80 kDa), in vitro glycated fructosamine Amadori-HSA ( ; 90 kDa) and first signs of AGE-HSA ( ; 100 kDa) are indicated. The degree of
retention of the specified HSA modifications (lane 1: naturally glycated Amadori-HSA; lane 2: in vitro glycated fructosamine Amadori-
HSA) is indicated using black arrows. Also indicated is a minor shift in the position of the Amadori-HSA band in glucose-6-phosphate
incubated HSA compared to the positions of this modification in the neighboring lanes. We speculate that this shift may be caused by the
presence of a phosphate group in this product.
Proteomics 2010, 10, 48–58 53
& 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.proteomics-journal.com

can only result in a five-membered furanose-phosphate
protein adduct, as is shown in Fig. 5C. In its pyranose form,
the phosphate in glucose 6-phosphate is in the equatorial
position of the chair conformation, distant from the
anomeric carbon. In the furanose form, however, it has been
described that in a lower energy conformation the phos-
phate is found perched over the sugar ring, promoting a
electrostatic interaction with the protein’s amine [44].
Similar to the b-furanose form of fructosamine, the fructo-
samine 6-phosphate adduct contains an anomeric cis 2,3
diol, but the phosphate positioned over the sugar ring will
hinder the interaction between phenylboronate and this
Amadori product (Fig. 5C, right). A possible interaction
between the phosphate moiety and the protonated Amadori
amino group will further encroach on this interaction and
hamper a band shift of glucose 6-phosphate incubated HSA
in the phenylboronate gel.
The so-called Heyns rearrangement in a protein incu-
bated with fructose is believed to proceed through two routes
[24]: (i) involving carbon 1, leading to the six-membered
hemiacetal form shown in Fig. 5D (upper right), and (ii)
involving carbon 3 resulting in a five-membered form
(Fig. 5D, lower right). The a-furanose form of the latter
contains an anomeric cis 3,4 diol and although its b-furanose
form also has a cis 4,5 diol, the absence of a band shift in
mP-AGE suggests that the stabilization by the additional
boronate-amino group interaction (seen in fructosamine-
HSA) may be hindered by the proximity of the C1 hydroxyl
to the amino group. Alternatively, the formation of the six-
membered xylosamine form of the fructose-HSA adduct
may be preferred and its 4,6 diol can only form a much
weaker six-membered cyclic ester with the gel-incorporated
MPBA (indicated by squares).
3.3.1 The role of the anomeric cis diol
To further investigate the role of the sugar adduct cis diol
interactions with the phenylboronate, proposed in our
model, we decided to incubate HSA with two additional
sugars. First, we chose mannose to confirm the results
observed with glucose. Glycation by mannose should result
in an identical fructosamine-HSA adduct as seen with
glucose (Fig. 5E) and therefore create a similar band shift
effect as seen in the glucose incubated HSA. Secondly,
glycation by maltose in our model should result in a
maltulosamine HSA adduct (Fig. 5F) that, in contrast with
the lactulosamine product produced during incubation with
lactose, only has an anomeric cis diol (1,2). This single diol
adduct will help to elucidate whether the band shift seen
with glucose and lactose requires two cis diol interactions (in
the fructosamine adduct: cis diols 2,3 and 4,5; in the lactu-
losamine adduct: cis diols 2,3 and 30,40) with MPBA.
An mP-AGE analysis of glucose, fructose, mannose and
maltose incubated HSA is shown in Fig. 6. The substantial
band shift resulting from the incubation with mannose (lane 3)
is comparable with the retention observed with glucose-HSA,
confirming our prediction that the same fructosamine-HSA
adduct has formed. The band shift seen in the maltulosamine-
HSA adduct, resulting from glycation by maltose, clearly
indicates that the presence of an anomeric cis diol is sufficient
for this effect. Based on these results, we can conclude that the
Amadori product in its b-furanose form, stabilized by the
amino group is paramount for the retention of a glycated
protein in mP-AGE. Such a stable interaction is hindered in
fructose and glucose 6-phosphate protein adducts. Interest-
ingly, maltulosamine-HSA is retained slightly higher than the
two fructosamine-HSA bands, indicating that mP-AGE can
also distinguish between the two different glycation products.
3.3.2 N- and O-glycans
The absence of this essential anomeric cis diol in N- and
O-glycan structures attached to glycosylated proteins
explains why their mobility is not affected in mP-AGE
(see Supporting Information Fig. 4 for glycan structures).
Boronate affinity chromatography involves boronates
containing both five- and six-membered rings derived from
Amadori, early AGE products and N- and O-glycans.
Conversely, in mP-AGE the current applied to the proteins
in the gel appears to only accept the strongest interaction,
selecting the trident interaction between phenyl boronate
and the anomeric cis diol, stabilized by the Amadori amino
group (Fig. 5A). Similar to our results comparing lactose
and maltose incubated HSA, the presence of cis 3,4 diols
in galactose or fucose end groups in complex sugars are
unlikely to affect their mobility in the methacrylamido
phenylboronate gels.
1 2 3M 4 5
116.0
66.2
45.0
35.0
Figure 6. mP-AGE analysis comparing gel profiles in the
presence of MPBA (0.5%) of HSA incubated for 21 days with
glucose (lane 1), fructose (lane 2), mannose (lane 3), maltose
(lane 4) and PBS (lane 5). Helper lines are added to aid the
comparison of the relative retention of the sugar incubated HSA
bands with the PBS-incubated control band.
56 M. P. Pereira Morais et al. Proteomics 2010, 10, 48–58
& 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.proteomics-journal.com

3.3.3 Glucose versus fructose AGE products
The presence of an a-hydroxy carbonyl function in Amadori
products, compared with an aldehydo group in Heyns
products, enables Amadori products to undergo a variety of
reactions, more readily than Heyns products [45]. The
Amadori product can undergo a number of subsequent
rearrangements, which lead to the formation of linear
a-dicarbonyl compounds [46] via enolization along the entire
carbohydrate backbone [47]. While in fructose or glucose
6-phosphate adducts this keto-enol tautomerization along the
carbohydrate backbone is prevented, it can occur in the
glyconoyl and fructosaminyl adducts leading to a linear adduct
with flexible hydroxyls near the aminogroup available for a
trident complex with phenylboronate.
4 Concluding remarks
Our results show that mP-AGE analysis has great potential
as a new proteomics technique for the detection of
post-translational modifications in proteins. Because the
retention of glycosylated proteins is not affected and that of
phosphorylated proteins only slightly by the presence of
phenylboronate in mP-AGE, the method proves to be a
highly selective and sensitive technique for the detection,
identification and separation of early and late glycation
products in proteins. A maximum band shift in the mP-
AGE gels is achieved with the linear gluconoyl adducts
followed by glucose and lactose induced AGE and Amadori
products. In addition, mP-AGE proofs are an ideal and
simple method to distinguish between various carbohy-
drate–protein adducts. Although a significant part of our
results are obtained with in vitro glycated proteins, this
paper also shows evidence that this method is applicable to
in vivo glycated material, by identifying the presence of
naturally occurring Amadori-HSA and detecting gluconoy-
lated protein modifications, resulting from the production of
recombinant proteins in E. coli.
The mP-AGE technique is an easy-to-use and cost-effec-
tive method that can complement available MS strategies
used in the identification and analysis of early and late
glycation products. It will thereby advance the study of
spontaneous glucose-mediated glycation processes in
ageing, diabetes, cardiovascular and Alzheimer’s disease by
detecting known and new glycoxi-adducts, analyze potential
inhibitors of the accumulation of AGEs and design new
drugs that can remove these undesired adducts.
This work was supported in part by Royal Society Research
2007/R2 (to J. S. F.) and the research on Sbi was funded by
Biotechnology and Biological Sciences Research Council
(BBSRC) Follow-on Fund Grant BB/F528014/1 (to J. M. H.
vd E.). We are also grateful to the University of Bath Research
and Innovation Services (RIS) for additional funding. M. P. P.
M. is funded by a BBSRC studentship and J. S. F. thanks the
Leverhulme Trust (F/00351/P). We thank Dr. Anneke Lubben
for micrOTOF MS analysis. Professor David E. Isenman,
Professor Flip G. de Groot and Dr. Karl Payne are thanked for
providing C4c, b2-glycoprotein I and SSE, respectively.
The authors have declared no conflict of interest.
5 References
[1] Mann, M., Jensen, O. N., Proteomic analysis of post-trans-
lational modifications. Nat. Biotechnol. 2003, 21, 255–261.
[2] Brownlee, M., Cerami, A., Vlassara, H., Advanced products of
nonenzymatic glycosylation and the pathogenesis of diabetic
vascular disease. Diabetes Metab. Rev. 1988, 4, 437–451.
[3] Fu, M. X., Wells-Knecht, K. J., Blackledge, J. A., Lyons, T. J.
et al., Glycation, glycoxidation, and cross-linking of
collagen by glucose kinetics, mechanisms, and inhibition of
late stages of the Maillard reaction. Diabetes 1994, 43,
676–683.
[4] Wells-Knecht, K. J., Zyzak, D. V., Litchfield, J. E., Thorpe, S.
R. et al., Mechanism of autoxidative glycosylation: identifi-
cation of glyoxal and arabinose as intermediates in the
autoxidative modification of proteins by glucose.
Biochemistry 1995, 34, 3702–3709.
[5] Kurien, B. T., Scofield, R. H., Autoimmunity and oxidatively
modified autoantigens. Autoimmun. Rev. 2008, 7, 567–573.
[6] Wells-Knecht, M. C., Thorpe, S. R., Baynes, J. W., Pathways
of formation of glycoxidation products during glycation of
collagen. Biochemistry 1995, 34, 15134–15141.
[7] Rahbar, S., Novel inhibitors of glycation and AGE forma-
tion. Cell. Biochem. Biophys. 2007, 48, 147–157.
[8] Unoki, H., Yamagishi, S., Advanced glycation end products
and insulin resistance. Curr. Pharm. Des. 2008, 14, 987–989.
[9] Vitek, M. P., Bhattacharya, K., Glendening, J. M., Stopa, E.
et al., Advanced glycation end products contribute to
amyloidosis in Alzheimer disease. Proc. Natl. Acad. Sci.
USA 1994, 91, 4766–4770.
[10] van Heijst, J. W., Niessen, H. W., Hoekman, K., Schalkwijk,
C. G., Advanced glycation end products in human cancer
tissues: detection of Nepsilon-(carboxymethyl)lysine and
argpyrimidine. Ann. NY Acad. Sci. 2005, 1043, 725–733.
[11] Vlassara, H., Advanced glycation in health and disease: role
of the modern environment. Ann. NY Acad. Sci. 2005, 1043,
452–460.
[12] Vlassara, H., Uribarri, J., Glycoxidation and diabetic
complications: modern lessons and a warning? Rev.
Endocrinol. Metab. Disord. 2004, 5, 181–188.
[13] Geoghegan, K. F., Dixon, H. B., Rosner, P. J., Hoth, L. R.
et al., Spontaneous alpha-N-6-phosphogluconoylation of a
‘‘His tag’’ in Escherichia coli: the cause of extra mass of 258 or
178 Da in fusion proteins. Anal. Biochem. 1999, 267, 169–184.
[14] Yan, Z., Caldwell, G. W., McDonell, P. A., Jones, W. J. et al.,
Mass spectrometric determination of a novel modification of
the N-terminus of histidine-tagged proteins expressed in
bacteria. Biochem. Biophys. Res. Commun. 1999, 259, 271–282.
Proteomics 2010, 10, 48–58 57
& 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.proteomics-journal.com

[15] Yan, Z., Caldwell, G. W., McDonell, P. A., Identification of a
gluconic acid derivative attached to the N-terminus of
histidine-tagged proteins expressed in bacteria. Biochem.
Biophys. Res. Commun. 1999, 262, 793–800.
[16] Beranek, M., Drsata, J., Palicka, V., Inhibitory effect of
glycation on catalytic activity of alanine aminotransferase.
Mol. Cell. Biochem. 2001, 218, 35–39.
[17] Kim, K. M., Yi, E. C., Baker, D., Zhang, K. Y., Post-transla-
tional modification of the N-terminal His tag interferes with
the crystallization of the wild-type and mutant SH3 domains
from chicken src tyrosine kinase. Acta Crystallogr. D Biol.
Crystallogr. 2001, 57, 759–762.
[18] Priego Capote, F., Sanchez, J. C., Strategies for proteomic
analysis of non-enzymatically glycated proteins. Mass
Spectrom. Rev. 2009, 28, 135–146.
[19] Zhang, Q., Tang, N., Brock, J. W., Mottaz, H. M. et al.,
Enrichment and analysis of nonenzymatically glycated
peptides: boronate affinity chromatography coupled
with electron-transfer dissociation mass spectrometry.
J. Proteome Res. 2007, 6, 2323–2330.
[20] Liu, X. C., Boronic acids as ligands for affinity chromato-
graphy. Se Pu 2006, 24, 73–80.
[21] Jackson, T. R., Springall, J. S., Rogalle, D., Masumoto, N. et al.,
Boronate affinity saccharide electrophoresis: A novel carbo-
hydrate analysis tool. Electrophoresis 2008, 29, 4185–4191.
[22] Burman, J. D., Leung, E., Atkins, K. L., O’Seaghdha, M. N. et al.,
Interaction of human complement with Sbi, a Staphylococcal
immunoglobulin-binding protein: indications of a novel
mechanism of complement evasion by Staphylococcus
aureus. J. Biol. Chem. 2008, 283, 17579–17593.
[23] Thornalley, P. J., Argirova, M., Ahmed, N., Mann, V. M.
et al., Mass spectrometric monitoring of albumin in uremia.
Kidney Int. 2000, 58, 2228–2234.
[24] Suarez, G., Rajaram, R., Oronsky, A. L., Gawinowicz, M. A.,
Nonenzymatic glycation of bovine serum albumin by fruc-
tose (fructation). Comparison with the Maillard reaction
initiated by glucose. J. Biol Chem. 1989, 264, 3674–3679.
[25] D’Hooge, F., Rogalle, D., Thatcher, M. J., Perera, S. P. et al.,
Polymerisation resistant synthesis of methacrylamido
phenylboronic acids. Polymer 2008, 49, 3362–3365.
[26] James, T. D., in: Hall, D. G. (Ed.), Boronic Acids in Organic
Synthesis and Chemical Biology, Wiley-VCH, Weinheim
2005, pp. 441–480.
[27] James, T. D., Phillips, M. D., Shinkai, S., Boronic Acids in
Saccharide Recognition, RSC Publishing, Cambridge, UK 2006.
[28] James, T. D., Shinkai, S., Artificial receptors as chemosensors
for carbohydrates. Top. Curr. Chem. 2002, 218, 159–200.
[29] Klenk, D. C., Hermanson, G. T., Krohn, R. I., Fujimoto, E. K.
et al., Determination of glycosylated hemoglobin by affinity
chromatography: comparison with colorimetric and ion-
exchange methods, and effects of common interferences.
Clin. Chem. 1982, 28, 2088–2094.
[30] Scrafton, D. K., Taylor, J. E., Mahon, M. F., Fossey, J. S.,
James, T. D., ‘‘Click-fluors’’: modular fluorescent saccharide
sensors based on a 1,2,3-triazole ring. J. Org. Chem. 2008,
73, 2871–2874.
[31] Aon, J. C., Caimi, R. J., Taylor, A. H., Lu, Q. et al.,
Suppressing posttranslational gluconoylation of hetero-
logous proteins by metabolic engineering of Escherichia
coli. Appl. Environ. Microbiol. 2008, 74, 950–958.
[32] Loske, C., Gerdemann, A., Schepl, W., Wycislo, M. et al.,
Transition metal-mediated glycoxidation accelerates cross-
linking of beta-amyloid peptide. Eur. J. Biochem. 2000, 267,
4171–4178.
[33] Rajbar, S., An abnormal hemoglobin in red cells of diabet-
ics. Clin. Chim. Acta 1968, 22, 296–298.
[34] Day, J. F., Thorpe, S. R., Baynes, J. W., Nonenzymatically
glucosylated albumin. In vitro preparation and isolation from
normal human serum. J. Biol. Chem. 1979, 254, 595–597.
[35] Dillner-Centerlind, M. L., Hammarstrom, S., Perlmann, P.,
Transferrin can replace serum for in vitro growth of mitogen-
stimulated T lymphocytes. Eur. J. Immunol. 1979, 9, 942–948.
[36] Posthuma, N., ter Wee, P. M., Niessen, H., Donker, A. J.
et al., Amadori albumin and advanced glycation end-
product formation in peritoneal dialysis using icodextrin.
Perit. Dial. Int. 2001, 21, 43–51.
[37] Liu, X. C., Scouten, W. H., Boronate affinity chromato-
graphy. Methods Mol. Biol. 2000, 147, 119–128.
[38] McPherson, J. D., Shilton, B. H., Walton, D. J., Role of
fructose in glycation and cross-linking of proteins.
Biochemistry 1988, 27, 1901–1907.
[39] Haney, D. N., Bunn, H. F., Glycosylation of hemoglobin
in vitro: affinity labeling of hemoglobin by glucose-6-
phosphate. Proc. Natl. Acad. Sci. USA 1976, 73, 3534–3538.
[40] Bunn, H. F., Higgins, P. J., Reaction of monosaccharides
with proteins: possible evolutionary significance. Science
1981, 213, 222–224.
[41] Sakai, M., Oimomi, M., Kasuga, M., Experimental studies on
the role of fructose in the development of diabetic compli-
cations. Kobe J. Med. Sci. 2002, 48, 125–136.
[42] Schalkwijk, C. G., Stehouwer, C. D., van Hinsbergh, V. W.,
Fructose-mediated non-enzymatic glycation: sweet
coupling or bad modification. Diabetes Metab. Res. Rev.
2004, 20, 369–382.
[43] Rohovec, J., Maschmeyer, T., Aime, S., Peters, J. A., The
structure of the sugar residue in glycated human serum
albumin and its molecular recognition by phenylboronate.
Chemistry 2003, 9, 2193–2199.
[44] Sandwick, R., Johanson, M., Breuer, E., Maillard reactions
of ribose 5-phosphate and amino acids. Ann. NY Acad. Sci.
2005, 1043, 85–96.
[45] Yeboah, F. K., Alli, I., Yaylayan, V. A., Reactivities of D-
glucose and D-fructose during glycation of bovine serum
albumin. J. Agric. Food Chem. 1999, 47, 3164–3172.
[46] Ulrich, P., Zhang, X., Pharmacological reversal of advanced
glycation end-product-mediated protein crosslinking.
Diabetologia 1997, 40, S157–159.
[47] Reihl, O., Rothenbacher, T. M., Lederer, M. O., Schwack, W.,
Carbohydrate carbonyl mobility--the key process in the
formation of alpha-dicarbonyl intermediates. Carbohydr.
Res. 2004, 339, 1609–1618.
58 M. P. Pereira Morais et al. Proteomics 2010, 10, 48–58
& 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.proteomics-journal.com