Exploring post-translational arginine modification using chemically synthesized methylglyoxal hydroimidazolones

Article abstract of DOI:10.1021/ja301994d

The methylglyoxal-derived hydroimidazolones (MG-Hs) comprise the most prevalent class of non-enzymatic, post-translational modifications of protein arginine residues found in nature. These adducts form spontaneously in the human body, and are also present at high levels in the human diet. Despite numerous lines of evidence suggesting that MG-H-arginine adducts play critical roles in both healthy and disease physiology in humans, detailed studies of these molecules have been hindered by a lack of general synthetic strategies for their preparation in chemically homogeneous form, and on scales sufficient to enable detailed biochemical and cellular investigations. To address this limitation, we have developed efficient, multigram-scale syntheses of all MG-H-amino acid building blocks, suitably protected for solid-phase peptide synthesis, in 2-3 steps starting from inexpensive, readily available starting materials. Thus, MG-H derivatives were readily incorporated into oligopeptides site-specifically using standard solid-phase peptide synthesis. Access to synthetic MG-H-peptide adducts has enabled detailed investigations, which have revealed a series of novel and unexpected findings. First, one of the three MG-H isomers, MG-H3, was found to possess potent, pH-dependent antioxidant properties in biochemical and cellular assays intended to replicate redox processes that occur in vivo. Computational and mechanistic studies suggest that MG-H3-containing constructs are capable of participating in mechanistically distinct H-atom-transfer and single-electron-transfer oxidation processes. Notably, the product of MG-H3 oxidation was unexpectedly observed to disassemble into the fully unmodified arginine residue and pyruvate in aqueous solution. We believe these observations provide insight into the role(s) of MG-H-protein adducts in human physiology, and expect the synthetic reagents reported herein to enable investigations into non-enzymatic protein regulation at an unprecedented level of detail.

Full text of DOI:10.1021/ja301994d

Article
pubs.acs.org/JACS
Exploring Post-translational Arginine Modification Using Chemically
Synthesized Methylglyoxal Hydroimidazolones
Tina Wang,† Rendy Kartika,‡ and David A. Spiegel*,†,§
†Department of Chemistry, Yale University, 225 Prospect Street, New Haven, Connecticut 06511, United States
‡Department of Chemistry, Louisiana State University, 707 Choppin Hall, Baton Rouge, Louisiana 70803, United States
§Department of Pharmacology, Yale University School of Medicine, 333 Cedar Street, New Haven, Connecticut 06510, United States
S
* Supporting Information
ABSTRACT: The methylglyoxal-derived hydroimidazolones
(MG-Hs) comprise the most prevalent class of non-enzymatic,
post-translational modifications of protein arginine residues
found in nature. These adducts form spontaneously in the
human body, and are also present at high levels in the human
diet. Despite numerous lines of evidence suggesting that MG-
H−arginine adducts play critical roles in both healthy and
disease physiology in humans, detailed studies of these
molecules have been hindered by a lack of general synthetic
strategies for their preparation in chemically homogeneous
form, and on scales sufficient to enable detailed biochemical and cellular investigations. To address this limitation, we have
developed efficient, multigram-scale syntheses of all MG-H−amino acid building blocks, suitably protected for solid-phase
peptide synthesis, in 2−3 steps starting from inexpensive, readily available starting materials. Thus, MG-H derivatives were
readily incorporated into oligopeptides site-specifically using standard solid-phase peptide synthesis. Access to synthetic MG-H-
peptide adducts has enabled detailed investigations, which have revealed a series of novel and unexpected findings. First, one of
the three MG-H isomers, MG-H3, was found to possess potent, pH-dependent antioxidant properties in biochemical and cellular
assays intended to replicate redox processes that occur in vivo. Computational and mechanistic studies suggest that MG-H3-
containing constructs are capable of participating in mechanistically distinct H-atom-transfer and single-electron-transfer
oxidation processes. Notably, the product of MG-H3 oxidation was unexpectedly observed to disassemble into the fully
unmodified arginine residue and pyruvate in aqueous solution. We believe these observations provide insight into the role(s) of
MG-H−protein adducts in human physiology, and expect the synthetic reagents reported herein to enable investigations into
non-enzymatic protein regulation at an unprecedented level of detail.
cytokine release,9−14 decreases in the adhesive properties of
INTRODUCTION
Advanced glycation end-products (AGEs) comprise a structur-
ally diverse class of post-translational protein modifications that
■
vascular basement membrane,15 alterations in chaperone
function,16 and numerous other processes.17,18 MGO concen-
trations have also been observed to increase in various disease
form through non-enzymatic chemical processes in all living
states, including diabetes,19−22 cancer,23 cardiovascular dis-
ease,24 and renal failure.25 Taken together, these observations
suggest that MG-H modifications may play a role in the
pathophysiology of these and perhaps other illnesses.
organisms, primarily on lysine and arginine side chains.1,2 Out
of more than eight arginine-derived AGEs reported to date, the
methylglyoxal-derived hydroimidazolones (MG-Hs, Figure 1A)
are believed to be the most prevalent in humans.3 Estimates
suggest that at least one MG-H adduct is present on 3−13% of
the proteins found in the human body,4,5 and that MG-H-
modified arginine comprises 1.3% by weight of total enzymatic
protein hydrosylate derived from alkali-treated baked goods
such as pretzels.6
MG-H-protein adducts form both intra- and extracellularly as
a mixture of three regioisomers (Figure 1A), via rapid reactions
between arginine and methylglyoxal (MGO), a dicarbonyl
metabolite produced by all living cells.7 Notably, elevated levels
of MGO and MG-H adducts have been shown to induce a
range of cellular and subcellular effects including perturbations
in cell signaling, inhibition of protein synthesis and cell
growth,8 induction of oxidative stress and pro-inflammatory
Despite their putative biological properties and their ubiquity
in humans, MG-Hs, and AGEs in general, have only been the
subject of a limited number of synthetic studies. Available
synthetic procedures for MG-H adducts are limited to
providing products on analytical scales,6 in very low purities,26
or as a subset of the natural isomers.27 The vast majority of
published data regarding the biological functions of MG-Hs
have been obtained using highly heterogeneous mixtures
prepared via prolonged incubation of model proteins (such as
albumins) with MGO.28 Although these mixtures have
Received: February 28, 2012
Published: May 16, 2012
© 2012 American Chemical Society
8958
dx.doi.org/10.1021/ja301994d | J. Am. Chem. Soc. 2012, 134, 8958−8967

Journal of the American Chemical Society
Article
Figure 1. Stuctures and retrosynthesis of the methylglyoxal-derived hydroimidazolone (MG-H) class of AGEs. (A) The MG-Hs consist of three
isomeric structures as shown, each formed through condensation of methylglyoxal with arginine. Note that MG-H3 is believed to exist as two low-
energy tautomers, as shown. For simplicity, only the tautomer drawn on the left will be depicted throughout the remainder of the document. (B)
Proposed “divergent” retrosynthesis of all three isomers derived from a single, readily available starting material.
Scheme 1
provided useful information regarding certain properties and
functions of MGO−protein adducts, their heterogeneous
nature complicates efforts to identify and characterize the
specific structure(s) responsible for bioactivity.29 Moreover,
these samples are prone to contamination with varying levels of
bioactive impurities (e.g. bacterial lipopolysaccharide).30
Here we report preparative-scale synthetic routes that for the
first time afford access to the entire class of methylglyoxal-
derived hydroimidazolones, as both amino acids and peptide
conjugates. These routes to protected MG-H building blocks
are concise (2−3 steps starting from readily accessible
materials) and high yielding (44−52% overall yield), and
MG-H monomers can readily be incorporated site-specifically
into synthetic oligopeptides using automated Fmoc solid-phase
peptide synthesis (SPPS). Critically, access to chemically
characterized MG-H-containing peptide constructs has enabled
us to perform biochemical and cellular evaluations, which have
revealed a series of notable findings. First, we have discovered
that MG-H3-containing oligopeptides are redox-active, both in
vitro and in a cellular system. Notably, this redox activity is pH-
dependent and has a molar potency comparable to that of
ascorbic acid. Computational and mechanistic studies suggest
that the MG-H3 core heterocycle possesses a versatile reactivity
profile and is capable of participating in both one- and two-
electron oxidation processes. Also, we unexpectedly found that
the product of MG-H3 oxidation disassembles spontaneously in
aqueous solution to provide the fully unmodified arginine side
chain and pyruvate. Taken together, these data suggest that
MG-H3 adducts may serve as pH-sensitive reducing agents
under physiological conditions, leading to spontaneous
8959
dx.doi.org/10.1021/ja301994d | J. Am. Chem. Soc. 2012, 134, 8958−8967

Journal of the American Chemical Society
Article
Scheme 2
deglycation of the hydroimidazolone core to regenerate the
native arginine side chain. These results underscore the
significant potential of chemically homogeneous MG-H adducts
to facilitate chemical and biological investigations into these
poorly understood species at an unprecedented level of detail.
on the bis(4-methoxyphenyl)methyl (Dod) group (Scheme
1).36,37 Application of this derivative to Fmoc-ornithine-
OBn·HCl (8), followed by HgCl2-promoted coupling with
Cbz-thiourea (9), smoothly provided the corresponding
guanidine (10) in good yield (50% over two steps).35 Notably,
this two-step sequence could be executed on preparative scales
(11.0 g) without difficulty. Subjection of 10 to catalytic
hydrogenolysis cleaved the Cbz and benzyl ester groups in a
single operation, and the resulting intermediate (11) under-
went spontaneous intramolecular cyclization to provide the
MG-H1 building block (12) as the exclusive product in 88%
yield. Overall, this sequence provides 12 in 43% overall yield in
only three synthetic steps, starting with >12 g of ornithine
derivative 8. Removal of the remaining protecting groups in
compound 12 afforded the free MG-H1−arginine adduct as a
TFA salt (12free·TFA), whose spectral properties (1H NMR,
13C NMR, and HRMS) agreed well with literature values
reported for a sample isolated from bakery products.6
Assignment of the hydroimidazolone ring in 12 and 12free as
the indicated tautomer was based on comparisons with
previously reported data in homologous systems and detailed
computational studies (Table S2).38,39
Having identified a successful route to the MG-H1 monomer
(12), we next turned our attention to MG-H2 (18, Scheme 2).
Toward this end, we developed a simple sequence consisting of
reductive amination (8 → 14), guanidylation (14 → 16), and
cyclization (16 → 17), which afforded iminohydantoin 17 in
55% overall yield. Chemoselective cleavage of the benzyl ester
in 17 provided side-chain- and backbone-protected MG-H2
monomer 18 in only three synthetic steps starting from 8 and
in 48% overall yield. Global protecting group removal
proceeded smoothly to yield MG-H2-arginine adduct
18free·TFA, which matched previously reported spectroscopic
data.33,40 Connectivity of 18free was also independently
confirmed through HMBC experiments, which indicated strong
couplings between the ring methine proton and the adjacent
side-chain C-atom (Figure S5). Tautomeric forms of 18 and
18free·TFA were assigned based on comparison to previously
reported model systems,38,39 and results of DFT calculations
(Table S3).
RESULTS
■
Synthetic Studies. The arginine-derived hydroimidazoles
MG-H1, MG-H2, and MG-H3 (Figure 1A) present substantial
synthetic challenges. These compounds have been reported to
undergo decomposition, spontaneous ring-opening, H/D
exchange, isomerization, and rearrangement reactions.31 Also,
because MG-Hs likely exert their biological actions as peptide
or protein conjugates in vivo, we required a synthesis that
provided AGE-amino acid monomers suitably protected for
Fmoc SPPS (Scheme 1). We therefore designed the
retrosynthetic analysis shown in Figure 1B. We envisioned
accessing protected isomeric MG-H monomers 1−3 via the
regioselective cyclization of intermediates 4−6. Regiocontrol in
these ring-forming processes was anticipated to derive from the
strategic placement of functional groups R1, R2, and R3. We also
expected that the appropriate protecting groups could be used
to stabilize the acid- and base-labile hydroimidazolone ring
during SPPS. Compounds 4−6 could then be derived from the
commercial reagent Nα-Fmoc-Nδ-Boc-L-ornithine (7) via
simple alkylation and/or guanidylation processes.32 Due to
published evidence indicating that the ring methine proton in
all three MG-H isomers rapidly undergoes epimerization or
exchange in water,6,33 we decided to target these compounds as
diastereomeric mixtures.
Initial synthetic efforts focused on accessing a suitably
protected MG-H1−arginine adduct. Based on our preliminary
studies, and prior observations suggesting that the MG-H1
isomer is kinetically disfavored,34 we hypothesized that a
blocking group on the ornithine side chain (e.g., R1 in 4) would
be necessary to direct cyclization through the desired pathway.
Furthermore, we discovered that guanidylation required the
side-chain N-atom to be electron rich,35 thus ruling out the
majority of common N-protecting groups. After initial experi-
ments, which revealed that sterically unhindered substituents
on the ornithine side-chain amine (e.g., trimethoxybenzyl) led
to spontaneous internal removal of the Fmoc group, we settled
Finally, MG-H3 precursor 22 was obtained simply by
rearranging chemical transformations that had proved success-
8960
dx.doi.org/10.1021/ja301994d | J. Am. Chem. Soc. 2012, 134, 8958−8967

Journal of the American Chemical Society
Article
Scheme 3
ful in the routes above. This sequence (Scheme 3) began with
mercury-mediated guanidylation of 8 with 2,2,4,6,7-pentam-
ethyldihydrobenzofuran-5-sulfonyl (Pbf)-derived thiourea 19,46
which afforded an intermediate N,N′-bis-alkyl-guanidine (20).
Although, in theory, pathways for forming either MG-H1 or
MG-H3 ring systems were both accessible via 20, in situ
cyclization provided MG-H3 precursor 21 as a single
regioisomer, likely due to the steric influence of the bulky
Pbf group. With 21 in hand, benzyl ester removal provided
compound 22 in 64% yield. Overall, this sequence allowed us
to access monomer 22 in only two steps and 52% overall yield
starting from 8. Protecting group removal provided 22free·TFA,
whose spectral properties matched previously reported
values.33,40 Notably, as for the MG-H1 series (Scheme 1),
synthetic sequences for accessing both MG-H2 and MG-H3
precursors could be carried out effectively on large scales (6.7−
10.7 g). Connectivity assignments in 22free·TFA were inferred
from HMBC couplings between the side-chain protons and the
ring carbonyl carbon (Figure S6). However, the tautomeric
preference of 22free could not be unequivocally established (see
Supporting Information).38
With protected monomeric hydroimidazolone derivatives 12,
18, and 22 in hand, we set out to construct several AGE-
containing oligopeptide sequences. Toward this end, we
synthesized various oligopeptides, site-specifically substituting
the arginine residues with the corresponding hydroimidazo-
lones (Table 1). HSA fragments (residues 111−120 and 407−
415) were chosen because of their established propensities
toward glycation in vitro,41,42 and were prepared in good to
excellent yields, with purities universally greater than 95%.38
Spectroscopic data for peptides 23−25 were found to match
with data previously reported for these compounds.33,40
Overall, this strategy represents the first general route for
site-specifically incorporating MG-H residues into oligopep-
tides, is effective on large scales, and also enables straightfor-
ward access to multiply AGE-derivatized constructs (Table 1,
entries 8−11).33,40
Table 1. Synthesis of Hydroimidazolone-Containing
Peptides
isolated
HSA
yield
a
entry compd
fragments
peptide sequences
LG-(MG-H1)-AG
(%)
1
2
3
4
5
6
7
8
9
23
24
25
26
27
28
29
30
31
n/a
32
33
47
46
59
37
55
33
39
n/a
LG-(MG-H2)-AG
n/a
LG-(MG-H3)-AG
407−415
407−415
407−415
407−415
111−120
111−120
LLV-R-YTKKV
LLV-(MG-H1)-YTKKV
LLV-(MG-H2)-YTKKV
LLV-(MG-H3)-YTKKV
NLP-R-LV-R-PEV
NLP-(MG-H1)-LV-(MG-H1)-
PEV
10
11
32
33
111−120
111−120
NLP-(MG-H2)-LV-(MG-H2)-
36
49
PEV
NLP-(MG-H3)-LV-(MG-H3)-
PEV
a
Peptides were isolated in >95% purity following HPLC purification.
as monitored by 1H NMR spectroscopy. The half-lives of H/D
exchange in these peptides were calculated to equal 40.5, 28.5,
and 15.4 h (Figures S1−S3), respectively.38 Other workers have
observed an increase in the rate of epimerization at pH 5.0
versus 7.4, suggesting that this process may occur rapidly under
physiological conditions.34 Taken together, these data suggest
that MG-Hs likely exist as thermodynamic mixtures of
diastereomers in the body.
Biochemical Investigations. Strong correlations between
oxidative stress and AGE formation have been observed both in
vitro and in vivo.43−47 In particular, MGO−protein adducts have
been suggested to induce the formation of reactive oxygen
species (ROS) in tissue culture,9−14 and treatment of bovine
serum albumin (BSA) with MGO has been shown to cause free
radical generation by EPR spectroscopy.48 Interestingly,
glycated proteins believed to contain MG-H adducts have
also been shown to possess antioxidant properties in vitro.49,50
Taken together, these observations led us to hypothesize that
hydroimidazolone adducts might be capable of participating in
redox processes.
Although the ring stereocenters in MG-H1 and MG-H3 have
been reported previously to undergo epimerization,6 the rate of
epimerization has not yet been quantified. Thus, we
investigated rates of H/D exchange using constructs 23−25.
Dissolution of these peptides as the corresponding formate salts
in neat D2O revealed a gradual disappearance of the
hydroimidazolone ring α-proton for all three MG-H isomers
To test this hypothesis, we first evaluated the redox activity
of MG-H−peptide conjugates 26−29 in vitro, in the MTT and
DPPH assay systems (Figure 2). MTT (3-(4,5-dimethylthiazol-
2-yl)-2,5-diphenyltetrazolium bromide) is a tetrazolium salt
8961
dx.doi.org/10.1021/ja301994d | J. Am. Chem. Soc. 2012, 134, 8958−8967

Journal of the American Chemical Society
Article
Figure 2. MG-H3-containing peptides possess antioxidant activity comparable with that of ascorbic acid (AA). (A) Peptides 26−29, 30−33, and AA
were evaluated for their ability to reduce yellow 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) to purple formazan at room
temperature. Data represent the mean of triplicate experiments standard deviation. Data in the indicated columns (***, P < 0.0001) were found to
differ from all other data sets in the figure using a one-way analysis of variance (ANOVA) with Tukey’s multiple comparison posthoc test. (B)
Radical quenching activity of peptides 26−29 was determined by treatment with the stable radical 1,1-diphenyl-2-picryl-hydrazyl (DPPH). Values
represent the mean of triplicate experiments standard deviation. (C) Effect of pH on the observed rate of DPPH quenching by both MG-H3−
peptide and MG-H3−amino acid conjugates. Note that determination of DPPH quenching rate of free MG-Hs at pH 7 was impeded by high assay
background. Values represent the mean of triplicate experiments standard deviation. (D) Antioxidant activity of MG-H−peptide conjugates in
cellular assays. RAW 264.7 macrophages were pre-incubated with dihydrorhodamine 123, a cell-permeable fluorogenic probe converted to a highly
fluorescent product on exposure to various cellular oxidants, then treated with oxidant (H2O2) and additive (peptide conjugates or AA). Two
treatment regimes were explored; in the first, oxidant concentration was varied at a fixed level of additive (left-hand panel), while in the second,
additive was varied at a fixed amount of oxidant (right-hand panel). Reported “Antioxidant Activity” values reflect the extent to which each additive is
capable of suppressing H2O2-induced fluorescence. These values ( standard error) were calculated from linear least-squares fits of fluorescence
intensity data plotted versus additive concentration as detailed in the Supporting Information. P-values (two-tail) represent the probabilities that
pair-wise differences in slope (plus versus minus additive) could have arisen from randomly chosen data points, and are reported as follows: ***, P <
0.0001; *, P < 0.05. All reported trends were observed on at least two separate occasions.
that can be reduced to a purple formazan by mitochondrial
enzymes,51 superoxide,52 or direct electron transfer mecha-
nisms.53 This indicator has been used to measure cell viability,54
superoxide production,52 and the antioxidant activity of natural
products.55 DPPH (1,1-diphenyl-2-picryl-hydrazyl), on the
other hand, is a stable radical that can be quenched by either
hydrogen atom abstraction or proton-coupled electron transfer,
and is commonly used to estimate the ability of substrates to
serve as H-atom donors in aqueous media.56 Peptide conjugates
containing the MG-H3 modification (29 and 33) exhibited
antioxidant activities in both MTT (Figure 2A) and DPPH
(Figure 2B) assays at comparable molar potencies to ascorbic
acid (vitamin C), an essential nutrient that serves as both an
antioxidant57 and enzyme cofactor.58 Surprisingly, despite their
high structural similarities to MG-H3, neither MG-H1 nor MG-
H2 exhibited any redox activities under these conditions.
We also evaluated the effect of pH on the rate of DPPH
quenching in the presence of various MG-H constructs (Figure
2C). Once again, MG-H3 adducts demonstrated significant
activity in this assay, both as peptide conjugate 29 and as the
free amino acid monomer (22free), while MG-H1 and MG-H2
constructs were inactive. Furthermore, the activity of MG-H3
conjugates was observed to increase at pH values approaching
7, which is consistent with the results of pKa titrations and
computational studies (see Table 2, below). Interestingly, pH
changes were found to exert a far greater effect on DPPH
reduction rate for MG-H3−peptide conjugate 29 than the
corresponding free amino acid (22free). Taken together with
observations in MTT assays (Figure 2A), this context-
dependent pH sensitivity suggests that the chemical properties
of MG-Hs may be influenced by neighboring amino acid
residues, and that such effects may be important determinants
of AGE effects in vivo.
To determine the cellular relevance of this redox behavior,
we explored the effects of MG-H−peptide conjugates using a
well-established cellular assay protocol.59 Thus, RAW 264.7
macrophages were treated with dihydrorhodamine 123, a cell-
permeable fluorogenic probe converted to a highly fluorescent
product on exposure to cellular oxidants, in the presence of
varying concentrations of oxidant (H2O2) and MG-H−peptide
conjugates.60 As shown in Figure 2D, MG-H3 conjugate 29 was
found to possess significant cellular antioxidant activity, which
was both concentration-dependent (right panel) and robust
over a range of physiologically relevant H2O2 concentrations
(left panel, see Supporting Information for more detail).
Interestingly, these effects were slightly smaller in magnitude
than those of ascorbic acid, perhaps due to the relatively poor
membrane permeability of this specific peptide sequence
8962
dx.doi.org/10.1021/ja301994d | J. Am. Chem. Soc. 2012, 134, 8958−8967

Journal of the American Chemical Society
Article
Figure 3. LC/MS and LC/MS-MS analyses of hydroimidazolone−peptide conjugates from MTT antioxidant assays. (A) LC traces of MTT assay
reaction mixtures. Two new peaks are observed in assays involving peptide 29, while experiments employing peptides 26−28 indicate only starting
materials. (B) Mass spectra corresponding to material contained in the LC peaks labeled 1 and 2 in panel A. These experiments indicate a loss of two
mass units in newly formed material versus the parent peptide 29. (C) LC/MS-MS fragmentation analysis of material in peak 2. These data suggest
that the decrease in reaction product mass originates at the MG-H3-modified residue. (D) Schematic reaction mechanism illustrating the overall
transformation from MG-H3-modified peptide 29 to the corresponding Arg-modified sequence (26) and pyruvate (35) by way of imidazolone 34.
(Figure S17). Because MGO and corresponding MG-H
adducts are present physiologically both inside and outside of
the cytosol, these data may underestimate the effects of MG-H3
adducts in vivo. Interestingly, MG-H2 conjugate 28 was found
to possess detectable antioxidant effects, which were only
revealed under conditions of variable H2O2 concentration. This
observation may suggest that MG-H2 possesses cellular redox
activity that emerges under conditions of high oxidative stress.
As before, MG-H1 conjugate 27 was found to be inactive in this
assay. Taken together, these studies demonstrate the capacity of
MG-H3−peptide conjugates to serve as antioxidants in a
cellular milieu in the context of physiologically relevant reactive
oxygen species.
indicated the formation of a new species with a mass 2 units
less than that of parent peptide 29 (Figure 3A,B), suggesting
oxidation to the corresponding imidazolone (MG-I3, 34).61
Tandem mass spectrometry (MS/MS) fragmentation analysis
(Figure 3C) further confirmed this decrease in mass to
originate at the MG-H3 residue. Detailed NMR experiments,
performed on the oxidized intermediate 34, further support our
proposed structural assignment (Figure S8). Interestingly, an
analogous MG-H1-derived imidazolone was postulated in prior
work to form under aerobic conditions in model reactions
between MGO and BSA or model peptides;61,62 however,
subsequent studies revealed this structural assignment to be
incorrect, and the adduct initially believed to be the
imidazolone was actually the argpyrimidine modification.62
Based on the electron-deficient nature of the imidazolone
oxidation product, we speculated that perhaps this material
To characterize the products of redox reactions involving
MG-H species, we next performed LC/MS analyses of MTT
assay mixtures using peptides 26−29. These experiments
8963
dx.doi.org/10.1021/ja301994d | J. Am. Chem. Soc. 2012, 134, 8958−8967

Journal of the American Chemical Society
Article
could react with arginine and/or lysine residues to form inter-
or intramolecular cross-links. Formation of such cross-links is
common in protein glycation processes, and has been cited as a
critical contributing factor in the impact of AGE formation on
protein structure and function.63 Unexpectedly, dissolution of
oxidized peptide 34 in physiological buffer led to spontaneous
hydrolysis to provide the parent arginine adduct 26 and
pyruvate (35, Figure 3D, Figures S9−S11). This hydrolysis
process was found to proceed at pH 7.4 with first-order kinetics
(kobs = 1.84 × 10−5 s−1; t1/2 = 10.5 h) at a rate compatible with
the lifetime of many intracellular proteins (Figure S11).
Hydrolysis of imidazolone adducts to provide unmodified
arginine residues is therefore kinetically competent to occur
under physiological conditions.64 Notably, these results provide
a novel mechanistic framework through which MG-H3-
modified proteins can directly participate in cellular redox
processes and recycle back to unmodified species. Furthermore,
the observation of the pyruvate byproduct in these studies
supports the intermediacy of MG-I3 (34), and also provides
evidence that MG-H adducts can non-enzymatically cleave to
give rise to a metabolically useful intermediate.
Table 2. pKa Values and Calculated Energies for MG-H
Derivatives
a b
,
pKa
f
g
h
BDE
IE
ΔGox
d
e
compd exptl
calcd
(kcal/mol)
(kcal/mol)
(kcal/mol)
36
37
38
39
4.58
3.58
3.81
7.44
n/r
76.2
76.5
68.3
76.4
153.2
150.6
141.0
186.9
26.0
22.3
23.8
n/r
4.68
7.70
n/r
a
R = Orn for compounds employed in pKa measurements, and R =
Me for structures used in calculations. Details of energetic calculations
for relevant tautomeric and protonated forms can be found in the
b
Supporting Information (SI). All calculations were performed
We next conducted mechanistic studies to gain insight into
the origin of MG-H3’s antioxidant activity. We first prepared
derivatives of 29 in which the methine C−H bond in the MG-
H3 core heterocycle was replaced with deuterium. Indeed, the
effect of isotopic substitution on reduction rate differed
substantially between the MTT and DPPH assays; a kinetic
isotope effect (KIE) of 1.38 was observed using the former
assay system, and 2.27 using the latter (Figures S15 and S16).
The relatively large KIE observed in DPPH assays supports a
mechanism involving C−H/D bond homolysis as the rate-
limiting step,65 while the relatively small KIE for the MTT assay
is consistent with speculations that electron-transfer pathways
are rate-limiting under these conditions.53 Taken together,
these data suggest that MG-H3 is capable of functioning as an
antioxidant by way of two distinct mechanistic pathways.
To shed light on the observed reactivity differences between
MG-H isomers, we performed a series of pKa titrations and
Density Functional Theory (DFT) calculations (B3LYP/6-
311++G(2df, 2p), Table 2). These experiments revealed a
number of intriguing features, which may provide explanations
for many of MG-H3’s unique properties. For example,
indications that the MG-H3 core heterocycle (38) possesses
C−H bond-dissociation energy (BDE) and ionization energy
(IE) values that are both significantly lower than those of
isomeric compounds 36 and 37 may explain the multimodal
activity of MG-H3 derivatives in both MTT and DPPH assays.
Furthermore, the observation that 38 possesses a pKa value
near neutrality, taken together with calculations suggesting that
protonation of the core heterocycle (39) increases both BDE
and IE, may explain the deleterious effect of decreasing pH on
antioxidant activity (Figure 2C). Interestingly, the observed pKa
trends for various MG-H isomers are in agreement with
experimental observations of structurally homologous isomers
of creatinine.39,66 Because the observed reactivity differences
between MG-H isomers do not correlate with differences
between free energy of oxidation (ΔGox) values (Table 2), we
speculate that they reflect kinetic rather than thermodynamic
factors.
utilizing the Gaussian 09 program suite. Geometry optimizations
were performed using the 6-31+G(d) basis set, and single-point
energies determined at the 6-311++G(2df,2p) level. All calculations
implemented a continuum model to account for the effects of water
solvent. l.p. = lone pair. Additional details regarding pKa titrations and
c
calculations can be found in the SI. Calculations pertaining to other
d
MG-H3 tautomers can be found in the SI. pKa values were obtained
e
experimentally via NMR titration. pKa values were calculated as
f
g
h
detailed in the SI. Bond dissociation energy. Ionization energy. Data
correspond to the free energy difference between the indicated MG-H
and the corresponding imidazolone oxidation product, as described in
the SI.
hydroimidazolone-modified arginine residues into peptides via
automated Fmoc SPPS. These constructs have allowed us to
undertake amongst the first detailed biochemical investigations
into chemically homogeneous MG-Hs. Previously, such studies
have been hampered due to limited availability of preparative-
scale synthetic strategies for peptide-MG-H constructs. Our
investigations have revealed a series of unexpected findings.
First, using well-characterized assays to measure redox behavior,
we have determined that adducts containing MG-H3 can serve
as potent antioxidants in vitro and in cellular systems, while
isomeric MG-H1 and MG-H2 conjugates are significantly less
active under identical conditions. Based on kinetic isotope
effects and computational results, we believe that MG-H3
conjugates can participate in redox reactions through two
distinct mechanisms. In DPPH assays, we postulate this
mechanism to involve rate-limiting H-atom transfer,65 while
we speculate the MTT reduction to proceed by way of rate-
limiting single-electron transfer, as previously described.53
Furthermore, results from both experimental and computa-
tional studies support that protonation of MG-H3’s core
heterocycle takes place within the physiological range and
diminishes its reactivity, suggesting a possible role for pH
fluctuations in regulating MG-H oxidation in vivo. Finally, we
have observed that MG-H3 oxidation leads to spontaneous
decomposition to pyruvate and fully unmodified arginine in
aqueous solution. This result provides strong support for a
previously unreported “deglycation” mechanism for MG-H-
modified arginine residues, which may be relevant in vivo.
DISCUSSION
■
Here we disclose robust synthetic routes that enable
straightforward incorporation of all known variations of
8964
dx.doi.org/10.1021/ja301994d | J. Am. Chem. Soc. 2012, 134, 8958−8967

Journal of the American Chemical Society
Article
Figure 4. Proposed model for the regulation of arginine glycation. Accessible Arg side chains react with MGO to form a mixture of isomeric MG-H
adducts, of which MG-H3 is believed to be the kinetic product. MG-H3 is sensitive to local perturbations in pH or redox balance. Path A: At neutral
pH, MG-H3 undergoes rapid oxidative conversion to MG-I3, which then spontaneously hydrolyzes to regenerate Arg and pyruvate. Path B: At low
local pH, or in proximity to acidic residues, MG-H3-H+ is formed, and this species is expected to be relatively insensitive to changes in redox balance.
Path C: At high pH or in proximity to basic residues, on the other hand, ring-opening of MG-H3 is expected to afford carboxyethylarginine (CEA)
and MG-H1, both of which are predicted to be relatively insensitive to redox changes.
In light of our observations, as well as previous reports, we
hypothesize that non-enzymatic post-translational modification
of protein arginine residues may occur in vivo through the
pathway shown in Figure 4. First, based on previous studies,
unmodified arginine residues are hypothesized to form MG-H
isomers at a rate related to MGO concentration, pH, and other
protein- and context-specific variables (e.g., subcellular local-
ization, sequence-dependent rate accelerations, etc.). Our data
suggest that, once formed, MG-H3 adducts are greatly
influenced by their local environment. At physiological pH,
MG-H3 adducts are expected to exist as a nearly equimolar
mixture of unprotonated and protonated (MG-H3-H+) forms
on the basis of pKa determinations (Table 2). Thus, upon
encountering appropriate oxidizing agents (e.g., reactive oxygen
species or free radicals), unprotonated MG-H3 can undergo a
net two-electron oxidation to yield MG-I3 (Figure 4, path A),
which spontaneously hydrolyzes to afford arginine and
pyruvate. Because oxidation is most likely a faster process
than hydrolysis at neutral pHfor example, DPPH scavenging
by peptide 29 has an observed half-life of approximately 6 min
(Figure S11), while hydrolysis has a half-life of 10.5 h (Figures
S5−S7)the possibility exists that significant concentrations of
MG-I3 can accumulate physiologically, perhaps leading to
protein−protein cross-linking or oxidative stress.67
In situations where protonation is favored (Figure 4, path B),
MG-H3 adducts are protected from oxidation (Figure 2C,
Table 2) and ring-opening.31 They are therefore believed to be
quite stablefor example, the half-life of 22free is approximately
24 days in DPPH assays run at pH 4.0 (Figure 2C) and 14.8
days in hydrolysis assays conducted at pH 5.4 and 37 °C.31 One
might also speculate that the presence of a positive charge in
MG-H3-H+ serves to mitigate destabilizing effects of side-chain
modification on protein structure,68 as opposed to other MG-H
modifications, which would be uncharged at neutral pH.
Further studies will be necessary to investigate these
possibilities in detail. Finally, in more basic environments
(Figure 4, path C), MG-H3 has a half-life of approximately 20
min in pH 9.4 buffer at 37 °C,31 and the primary products of
hydrolysis are carboxyethylarginine (CEA) and MG-H1.34 MG-
H1 is also expected to be the most thermodynamically stable
MG-H isomer; our calculations suggest it to be approximately
2.39 kcal/mol lower in energy than MG-H3 (Supporting
Information, Table S5), a trend which is supported by previous
experimental studies.34 MG-H1 is therefore hypothesized to
represent a “long-lived” modification that is susceptible neither
to hydrolytic nor oxidative removal. Indeed, MG-H1 is often
identified as the major MG-H formed in vivo,33 perhaps because
of its stability relative to MG-H3, and the scarcity of MG-H2.33
Also worth noting are the apparent functional similarities
between MG-H3 and ascorbic acid in redox assays (Figure 2).
Intriguingly, like MG-H3, ascorbic acid has been documented
to serve as a reducing agent through both single-electron- and
H-atom-transfer mechanisms, leaving open the possibility that
these species perform similar functions in vivo.69 Similarly,
ascorbate’s paradoxical oxidative activity, observed in vitro in
the presence of transition metals,70 may be mechanistically
linked to pro-oxidant effects reported for MGO-modified
proteins.71 Based on reported literature values,17,72 we estimate
that concentrations of MG-H−arginine adducts can range from
100 nM to 3 μM in plasma or cytosol (higher levels are
observed among diabetic patients), while ascorbate is found in
the plasma at approximately 6−60 μM and in cytosol at
approximately 35−500 μM.70,73 These rough estimates suggest
that MG-H3 may function as a bulk antioxidant physiologically.
Also, the high cellular permeability of MGO suggests the
existence of MG-H−protein adducts both intra- and extracell-
ularly, although these modified proteins themselves may be
impermeable.
In conclusion, we hypothesize that both formation and
removal of MG-H adducts occur in a pH- and/or redox-
regulated manner, and that these processes could be involved in
normal physiology and perturbed in various disease processes.
8965
dx.doi.org/10.1021/ja301994d | J. Am. Chem. Soc. 2012, 134, 8958−8967

Journal of the American Chemical Society
Article
(17) Thornalley, P. J. Drug Metabol. Drug Interact. 2008, 23, 125−
Validation of this theory will require follow-up studies of
significant scope, which will be enabled by chemically
homogeneous AGE derivatives, and have the potential to
implicate MG-Hs and other AGE family members in a much
broader range of life processes than previously suspected.
150.
(18) Rabbani, N.; Thornalley, P. J. Ann. N.Y. Acad. Sci. 2008, 1126,
124−127.
(19) Fosmark, D. S.; Bragadottir, R.; Stene-Johansen, I.; Berg, J. P.;
Berg, T. J.; Lund, T.; Sandvik, L.; Hanssen, K. F. Acta Ophthalmol.
Scand. 2007, 85, 618−622.
ASSOCIATED CONTENT
* Supporting Information
■
(20) Fosmark, D. S.; Torjesen, P. A.; Kilhovd, B. K.; Berg, T. J.;
Sandvik, L.; Hanssen, K. F.; Agardh, C. D.; Agardh, E. Metabolism
2006, 55, 232−236.
S
Detailed experimental procedures and compound character-
izations. This material is available free of charge via the Internet
at http://pubs.acs.org.
(21) Kilhovd, B. K.; Giardino, I.; Torjesen, P. A.; Birkeland, K. I.;
Berg, T. J.; Thornalley, P. J.; Brownlee, M.; Hanssen, K. F. Metabolism
2003, 52, 163−167.
(22) Han, Y.; Randell, E.; Vasdev, S.; Gill, V.; Curran, M.; Newhook,
L. A.; Grant, M.; Hagerty, D.; Schneider, C. Clin. Biochem. 2009, 42,
562−569.
AUTHOR INFORMATION
Corresponding Author
david.spiegel.@yale.edu
■
(23) Thornalley, P. J. Gen. Pharmacol. 1996, 27, 565−573.
(24) Kilhovd, B. K.; Juutilainen, A.; Lehto, S.; Ronnemaa, T.;
Torjesen, P. A.; Hanssen, K. F.; Laakso, M. Atherosclerosis 2009, 205,
590−594.
Notes
The authors declare no competing financial interest.
(25) Agalou, S.; Ahmed, N.; Babaei-Jadidi, R.; Dawnay, A.;
Thornalley, P. J. J. Am. Soc. Nephrol. 2005, 16, 1471−1485.
(26) Gruber, P.; Hofmann, T. J. Pept. Res. 2005, 66, 111−124.
(27) Hellwig, M.; Geissler, S.; Matthes, R.; Peto, A.; Silow, C.;
Brandsch, M.; Henle, T. Chembiochem 2011, 12, 1270−1279.
(28) Lieuw-a-Fa, M. L. M.; Schalkwijk, C. G.; Engelse, M.; van
Hinsbergh, V. W. M. Thromb. Haemostasis 2006, 95, 320−328.
(29) Waanders, F.; van den Berg, E.; Schalkwijk, C.; van Goor, H.;
Navis, G. Nephrology, Dialysis, Transplantation: Off. Publ. Eur. Dialysis
Transplant Assoc.−Eur. Renal Assoc. 2007, 22, 3093−3094.
(30) Valencia, J. V.; Mone, M.; Koehne, C.; Rediske, J.; Hughes, T. E.
Diabetologia 2004, 47, 844−852.
ACKNOWLEDGMENTS
■
We thank Prof. K. Wiberg for invaluable discussions and
assistance regarding mechanistic studies and computational
chemistry. We also thank Prof. Seth Herzon and Christopher
Parker for helpful discussions. This work was supported by the
Ellison Medical Foundation through the New Scholar Award in
Aging Research and by the National Institutes of Health
through the NIH Director’s New Innovator Award Program
(DP22OD002913). Finally, we would like to acknowledge the
SENS foundation for financial support, and Drs. William Bains
and Aubrey de Grey for helpful discussions.
(31) Ahmed, N.; Argirov, O. K.; Minhas, H. S.; Cordeiro, C. A. A.;
Thornalley, P. J. Biochem. J. 2002, 364, 1−14.
REFERENCES
(32) Linton, B. R.; Carr, A. J.; Orner, B. P.; Hamilton, A. D. J. Org.
Chem. 2000, 65, 1566−1568.
■
(1) Henle, T. Protein Pept. Lett. 2010, 35, S32−S37.
(2) Maillard, L. C. C. R. Acad. Sci. 1912, 154, 66−68.
(3) Ahmed, N.; Thornalley, P. J.; Dawczynski, J.; Franke, S.; Strobel,
J.; Stein, G.; Haik, G. M. Invest. Ophthalmol. Vis. Sci. 2003, 44, 5287−
5292.
(4) Thornalley, P.; Battah, S.; Ahmed, N.; Karachalias, N.; Agalou, S.;
Babaei-Jadidi, R.; Dawnay, A. Biochem. J. 2003, 375, 581−592.
(5) Ahmed, N.; Dobler, D.; Dean, M.; Thornalley, P. J. Biol. Chem.
2005, 280, 5724−5732.
(6) Henle, T.; Walter, A. W.; Haessner, R.; Klostermeyer, H. Z.
Lebensmittel-Untersuch. Forsch. 1994, 199, 55−58.
(7) Gomes, R. A.; Vicente Miranda, H.; Silva, M. S.; Graca, G.;
Coelho, A. V.; Ferreira, A. E.; Cordeiro, C.; Freire, A. P. FEBS J. 2006,
273, 5273−5287.
(8) Egyud, L. G.; Szent-Gyorgyi, A. Proc. Natl. Acad. Sci. U.S.A. 1966,
56, 203−207.
(9) Godbout, J. P.; Pesavento, J.; Hartman, M. E.; Manson, S. R.;
Freund, G. G. J. Biol. Chem. 2002, 277, 2554−2561.
(10) Westwood, M. E.; Thornalley, P. J. Immunol. Lett. 1996, 50, 17−
21.
(11) Abordo, E. A.; Westwood, M. E.; Thornalley, P. J. Immunol. Lett.
1996, 53, 7−13.
(12) Fan, X.; Subramaniam, R.; Weiss, M.; Monnier, V. Arch.
Biochem. Biophys. 2003, 409, 274−286.
(13) Glorieux, G.; Helling, R.; Henle, T.; Brunet, P.; Deppisch, R.;
Lameire, N.; Vanholder, R. Kidney Int. 2004, 66, 1873−1880.
(14) Van Herreweghe, F.; Mao, J. Q.; Chaplen, F. W. R.; Grooten, J.;
Gevaert, K.; Vandekerckhove, J.; Vancompernolle, K. Proc. Natl. Acad.
Sci. U.S.A. 2002, 99, 949−954.
(15) Thornalley, P. J.; Rabbani, N. Seminars Dialysis 2009, 22, 400−
404.
(16) Gangadhariah, M. H.; Wang, B.; Linetsky, M.; Henning, C.;
Spanneberg, R.; Glomb, M. A.; Nagaraj, R. H. BBA-Mol. Basis Dis.
2010, 1802, 432−441.
(33) Ahmed, N.; Thornalley, P. J. Biochem. J. 2002, 364, 15−24.
(34) Klopfer, A.; Spanneberg, R.; Glomb, M. A. J. Agric. Food Chem.
̈
2011, 59, 394−401.
(35) Levallet, C.; Lerpiniere, J.; Ko, S. Y. Tetrahedron 1997, 53,
5291−5304.
(36) Jonsson, D. Tetrahedron Lett. 2002, 43, 4793−4796.
(37) Jonsson, D.; Unden, A. Tetrahedron Lett. 2002, 43, 3125−3128.
(38) Details of these experiments can be found in the Supporting
Information.
(39) Kenyon, G. L.; Rowley, G. L. J. Am. Chem. Soc. 1971, 93, 5552.
(40) Gruber, P.; Hofmann, T. J. Pept. Res. 2005, 66, 111−124.
(41) Ahmed, N.; Thornalley, P. J. Maillard Reaction: Chem. Interface
of Nutrition, Aging, Disease 2005, 1043, 260−266.
(42) Ahmed, N.; Dobler, D.; Dean, M.; Thornalley, P. J. J. Biol. Chem.
2005, 280, 5724−5732.
(43) Wolff, S. P.; Jiang, Z. Y.; Hunt, J. V. Free Radic. Biol. Med. 1991,
10, 339−352.
(44) Schleicher, E.; Friess, U. Kidney Int. Suppl. 2007, S17−S26.
(45) Fan, X.; Subramaniam, R.; Weiss, M. F.; Monnier, V. M. Arch.
Biochem. Biophys. 2003, 409, 274−286.
(46) Schmitt, A.; Bigl, K.; Meiners, I.; Schmitt, J. Biochim. Biophys.
Acta 2006, 1763, 927−936.
(47) Deuther-Conrad, W.; Loske, C.; Schinzel, R.; Dringen, R.;
Riederer, P.; Munch, G. Neurosci. Lett. 2001, 312, 29−32.
(48) Lee, C.; Yim, M. B.; Chock, P. B.; Yim, H. S.; Kang, S. O. J. Biol.
Chem. 1998, 273, 25272−25278.
(49) Dittrich, R.; El-Massry, F.; Kunz, K.; Rinaldi, F.; Peich, C. C.;
Beckmann, M. W.; Pischetsrieder, M. J. Agric. Food Chem. 2003, 51,
3900−3904.
(50) Murakami, M.; Shigeeda, A.; Danjo, K.; Yamaguchi, T.;
Takamura, H.; Matoba, T. J. Food Sci. 2002, 67, 93−96.
(51) Slater, T. F.; Sawyer, B.; Straeuli, U. Biochim. Biophys. Acta 1963,
77, 383−393.
8966
dx.doi.org/10.1021/ja301994d | J. Am. Chem. Soc. 2012, 134, 8958−8967

Journal of the American Chemical Society
Article
(52) Burdon, R. H.; Gill, V.; Rice-Evans, C. Free Radic. Res. Commun.
1993, 18, 369−380.
(53) Marques, E. P.; Zhang, J. J.; Tse, Y. H.; Metcalfe, R. A.; Pietro,
W. J.; Lever, A. B. P. J. Electroanal. Chem. 1995, 395, 133−142.
(54) Mosmann, T. J. Immunol. Methods 1983, 65, 55−63.
(55) Liu, Y.; Nair, M. G. J. Nat. Prod. 2010, 73, 1193−1195.
(56) Brandwilliams, W.; Cuvelier, M. E.; Berset, C. Food Sci. Technol.:
Lebensmittel-Wissenschaft Technol. 1995, 28, 25−30.
(57) Padayatty, S. J.; Katz, A.; Wang, Y. H.; Eck, P.; Kwon, O.; Lee, J.
H.; Chen, S. L.; Corpe, C.; Dutta, A.; Dutta, S. K.; Levine, M. J. Am.
Coll. Nutr. 2003, 22, 18−35.
(58) Arrigoni, O.; De Tullio, M. C. Biochim. Biophys. Acta-Gen.
Subjects 2002, 1569, 1−9.
(59) Kim, C. Y.; Lee, C.; Park, G. H.; Jang, J. H. Arch. Pharm. Res.
2009, 32, 869−881.
(60) Emmendorffer, A.; Hecht, M.; Lohmann-Matthes, M. L.;
Roesler, J. J. Immunol. Methods 1990, 131, 269−275.
(61) Lo, T. W.; Westwood, M. E.; McLellan, A. C.; Selwood, T.;
Thornalley, P. J. J. Biol. Chem. 1994, 269, 32299−32305.
(62) Uchida, K.; Khor, O. T.; Oya, T.; Osawa, T.; Yasuda, Y.; Miyata,
T. FEBS Lett. 1997, 410, 313−318.
(63) Goh, S. Y.; Cooper, M. E. J. Clin. Endocrinol. Metab. 2008, 93,
1143−1152.
(64) Dice, J. F. FASEB J. 1987, 1, 349−357.
(65) Baciocchi, E.; Calcagni, A.; Lanzalunga, O. J. Org. Chem. 2008,
73, 4110−4115.
(66) Matsumoto, K.; Rapoport, H. J. Org. Chem. 1968, 33, 552−558.
(67) Westwood, M. E.; Thornalley, P. J. J. Protein Chem. 1995, 14,
359−372.
(68) Biswas, S.; Chida, A. S.; Rahman, I. Biochem. Pharmacol. 2006,
71, 551−564.
(69) Halliwell, B. Free Radic. Res. 1999, 31, 261−272.
(70) Duarte, T. L.; Lunec, J. Free Radic. Res. 2005, 39, 671−686.
(71) Halliwell, B. Free Radic. Res. 1999, 31, 261−272.
(72) Yatscoff, R. W.; Tevaarwerk, G. J.; MacDonald, J. C. Clin. Chem.
1984, 30, 446−449.
(73) Korcok, J.; Yan, R.; Siushansian, R.; Dixon, S. J.; Wilson, J. X.
Brain Res. 2000, 881, 144−151.
8967
dx.doi.org/10.1021/ja301994d | J. Am. Chem. Soc. 2012, 134, 8958−8967