A mitochondria-targeted mass spectrometry probe to detect glyoxals: Implications for diabetes

Article abstract of DOI:10.1016/j.freeradbiomed.2013.11.025

The glycation of protein and nucleic acids that occurs as a consequence of hyperglycemia disrupts cell function and contributes to many pathologies, including those associated with diabetes and aging. Intracellular glycation occurs after the generation of the reactive 1,2-dicarbonyls methylglyoxal and glyoxal, and disruption of mitochondrial function is associated with hyperglycemia. However, the contribution of these reactive dicarbonyls to mitochondrial damage in pathology is unclear owing to uncertainties about their levels within mitochondria in cells and in vivo. To address this we have developed a mitochondria-targeted reagent (MitoG) designed to assess the levels of mitochondrial dicarbonyls within cells. MitoG comprises a lipophilic triphenylphosphonium cationic function, which directs the molecules to mitochondria within cells, and an o-phenylenediamine moiety that reacts with dicarbonyls to give distinctive and stable products. The extent of accumulation of these diagnostic heterocyclic products can be readily and sensitively quantified by liquid chromatography-tandem mass spectrometry, enabling changes to be determined. Using the MitoG-based analysis we assessed the formation of methylglyoxal and glyoxal in response to hyperglycemia in cells in culture and in the Akita mouse model of diabetes in vivo. These findings indicated that the levels of methylglyoxal and glyoxal within mitochondria increase during hyperglycemia both in cells and in vivo, suggesting that they can contribute to the pathological mitochondrial dysfunction that occurs in diabetes and aging.

Full text of DOI:10.1016/j.freeradbiomed.2013.11.025

Free Radical Biology and Medicine 67 (2014) 437–450
Contents lists available at ScienceDirect
Free Radical Biology and Medicine
journal homepage: www.elsevier.com/locate/freeradbiomed
Original Contribution
A mitochondria-targeted mass spectrometry probe to detect
$
glyoxals: implications for diabetes
a
a
b
b
Pamela Boon Li Pun , Angela Logan , Victor Darley-Usmar , Balu Chacko ,
b
b
a
a
Michelle S. Johnson , Guang W. Huang , Sebastian Rogatti , Tracy A. Prime ,
c
c
a
d
d
Carmen Methner , Thomas Krieg , Ian M. Fearnley , Lesley Larsen , David S. Larsen ,
Katja E. Menger , Yvonne Collins , Andrew M. James , G.D. Kishore Kumar ,
Richard C. Hartley , Robin A.J. Smith , Michael P. Murphy
MRC Mitochondrial Biology Unit, Wellcome Trust/MRC Building, Cambridge CB2 0XY, UK
Department of Pathology, Centre for Free Radical Biology, University of Alabama at Birmingham, Birmingham, AL 35294, USA
Department of Medicine, University of Cambridge, Addenbrooke’s Hospital, Cambridge CB2 2QQ, UK
Department of Chemistry, University of Otago, Dunedin, New Zealand
a
a
a
e
e
d
a,
n
a
b
c
d
e
Centre for the Chemical Research of Ageing, WestCHEM School of Chemistry, University of Glasgow, Glasgow G12 8QQ, UK
a r t i c l e i n f o
a b s t r a c t
Article history:
Received 3 August 2013
Received in revised form
The glycation of protein and nucleic acids that occurs as a consequence of hyperglycemia disrupts cell
function and contributes to many pathologies, including those associated with diabetes and aging.
Intracellular glycation occurs after the generation of the reactive 1,2-dicarbonyls methylglyoxal and
glyoxal, and disruption of mitochondrial function is associated with hyperglycemia. However, the
contribution of these reactive dicarbonyls to mitochondrial damage in pathology is unclear owing to
uncertainties about their levels within mitochondria in cells and in vivo. To address this we have
developed a mitochondria-targeted reagent (MitoG) designed to assess the levels of mitochondrial
dicarbonyls within cells. MitoG comprises a lipophilic triphenylphosphonium cationic function, which
directs the molecules to mitochondria within cells, and an o-phenylenediamine moiety that reacts with
dicarbonyls to give distinctive and stable products. The extent of accumulation of these diagnostic
heterocyclic products can be readily and sensitively quantified by liquid chromatography–tandem mass
spectrometry, enabling changes to be determined. Using the MitoG-based analysis we assessed the
formation of methylglyoxal and glyoxal in response to hyperglycemia in cells in culture and in the Akita
mouse model of diabetes in vivo. These findings indicated that the levels of methylglyoxal and glyoxal
within mitochondria increase during hyperglycemia both in cells and in vivo, suggesting that they can
contribute to the pathological mitochondrial dysfunction that occurs in diabetes and aging.
22 November 2013
Accepted 25 November 2013
Available online 4 December 2013
Keywords:
Mitochondria
Exomarker
Methylglyoxal
Glyoxal
Hyperglycemia
MitoG
Free radicals
&
2013 The Authors. Published by Elsevier Inc. All rights reserved.
Glycation, the nonenzymatic formation of sugar–protein and
sugar–nucleotide adducts, plays a major role in disrupting cell function
and causing tissue damage in a range of pathologies such as diabetes,
aging, and neurodegeneration [1–3]. Glycation increases in response to
the elevation in glucose that occurs in unregulated diabetes and is a
major cause of diabetic complications [4,5]. Within the cell excessive
glucose can lead to molecular damage through the formation of
phosphate intermediates of glycolysis [1,6] or from the metabolism
of acetone generated during ketosis [7]. These reactive 1,2-dicarbonyls
often exist in modified chemical forms in situ including reversible
hemiacetals, hemithioacetals, and hemiaminals with small biomole-
cules and with reactive moieties on proteins and nucleic acids [8,9].
In addition they can react directly with free amine functions on
proteins and nucleic acids, thereby generating substantial permanent
modifications such as arginine-derived hydroimidazolones and lysine
cross-links on proteins [10] and guanine-derived imidazopurinones on
DNA [11]. Such modifications are thought to result in biochemical
dysfunction by altering protein structure and activity and by inducing
genomic mutations [2]. These markers of glycation damage are
elevated in many clinical samples from diabetic patients and also in
animal models of diabetes and aging [2,4,9,12,13], consistent with
a contribution from these reactions to cell damage and pathology.
1,2-dicarbonyl compounds such as methylglyoxal from the triose
☆
This is an open-access article distributed under the terms of the Creative
Commons Attribution-NonCommercial-ShareAlike License, which permits non-
commercial use, distribution, and reproduction in any medium, provided the
original author and source are credited.
n
Corresponding author.
E-mail address: mpm@mrc-mbu.cam.ac.uk (M.P. Murphy).
0891-5849/$ - see front matter & 2013 The Authors. Published by Elsevier Inc. All rights reserved.
http://dx.doi.org/10.1016/j.freeradbiomed.2013.11.025

438
P.B.L. Pun et al. / Free Radical Biology and Medicine 67 (2014) 437–450
An important role for methylglyoxal and glyoxal in pathology is
further supported by the existence of the glyoxalase enzyme system,
which specifically degrades these two dicarbonyls [14]. Loss of the
glyoxalase degradation pathway renders organisms more susceptible
to glycation and subsequent damage, whereas its overexpression
increases life span in Caenorhabditis elegans [15]. Thus dicarbonyl-
associated glycation of proteins and nucleic acids is a significant
contributing factor in a range of pathologies, particularly those
associated with diabetes or aging.
In hyperglycemia, there is considerable evidence for mitochon-
drial damage and elevated oxidative stress that contribute to
pathology, and this has been in part ascribed to mitochondrial
glycation due to methylglyoxal and glyoxal [16–21]. Furthermore
these reactive dicarbonyls disrupt mitochondrial function in vitro
uncertain, and it has proven difficult to specifically evaluate the
importance of these processes. This is in part due to the uncer-
tainties related to the distribution of methylglyoxal and glyoxal
between the cytosol and the mitochondria. To assess the impor-
tance of mitochondrial damage caused by methylglyoxal and
glyoxal we have developed a mitochondria-selective molecule,
MitoG, to assess relative changes in the levels of these damaging
species within mitochondria in cells and in vivo.
To target mitochondria we used the lipophilic triphenylpho-
sphonium (TPP) cation functionality, which has been shown to
direct a wide variety of antioxidants, probes, and bioactive
molecules to mitochondria in cells, animal models, and patients
after intravenous, oral, or intraperitoneal administration [25–27].
Uptake occurs directly through the phospholipid bilayer and does not
require a protein carrier. The extent of accumulation in mitochondria is
determined by the membrane potential and can be adequately
described by the Nernst equation, which indicates an ꢀ10-fold
increase in accumulation per 60 mV membrane potential under
[22–24]. Therefore, understanding the contribution from glycation
damage by reactive dicarbonyls to mitochondrial dysfunction is
important for analyzing and understanding the pathology asso-
ciated with hyperglycemia. However, the mechanistic details are
Fig. 1. Rationale and mechanism for the detection of intramitochondrial dicarbonyls. (A) MitoG, a mitochondria-targeted glyoxal and methylglyoxal trap, consists of a
mitochondria-targeting TPP moiety and a phenylenediamine group that reacts with 1,2-dicarbonyls. The TPP moiety of MitoG leads to its uptake into tissues where it
accumulates within mitochondria, driven by both the plasma and the mitochondrial membrane potentials. (B) Within mitochondria MitoG can then react with glyoxal or
methylglyoxal to form the quinoxaline products, QE and MQE (present as two isomers, MQE1 and MQE2). These products can then be quantified by LC–MS/MS relative to
deuterated internal standards to provide a measure of the amount of free glyoxal and methylglyoxal present within mitochondria in cells and in vivo.

P.B.L. Pun et al. / Free Radical Biology and Medicine 67 (2014) 437–450
439
typical biological conditions [25–27]. Consequently, TPP compounds
accumulate ꢀ1000-fold within mitochondria in vivo assuming
plasma and mitochondrial membrane potentials of 30 and 160 mV,
respectively [25–27] (Fig. 1A).
MitoG–dicarbonyl quinoxaline reaction products (Fig. 1B). Here we
describe the development of MitoG, a mitochondria-targeted probe
for methylglyoxal and glyoxal. We show that it can be used for the
evaluation of 1,2-dicarbonyl production within mitochondria in cells
and in vivo. These findings are consistent with mitochondrial
glycation contributing to the underlying pathology of hyperglycemia
in diabetes and related disorders.
Alkoxy-substituted phenylenediamines have been used for
detecting 1,2-dicarbonyls because of the enhanced reactivity due
to the electron-donating alkoxy substituent [28,29]. These are
commonly used derivatizing agents for the detection of dicarbo-
nyls, reacting to form stable quinoxaline products [12,13,28].
Therefore, to make a mitochondria-targeted molecule that reacts
selectively with methylglyoxal and glyoxal, we conjugated the
TPP moiety through an oxygen atom to an o-phenylenediamine
moiety. The mode of action of this molecule, called MitoG, is
shown in Fig. 1B. The ability to quantify the accumulation of the
quinoxaline products from the in situ reaction of MitoG with
methylglyoxal and glyoxal provides an opportunity to assess
changes in the levels of these compounds in mitochondria in cells
and in vivo. This can be done by an extension of an approach we
recently developed to assess levels of mitochondrial hydrogen
peroxide in vivo by the use of a mitochondria-targeted peroxide
reactive compound, MitoB [30,31]. In this methodology, a diagnostic
exomarker product, MitoP, was formed from MitoB, and its levels
were determined ex vivo by liquid chromatography–tandem mass
spectrometry (LC–MS/MS) of tissue homogenates relative to deut-
erated internal standards [30,31]. The sensitivity of such an approach
is greatly enhanced by the inherent positive charge of the TPP
moiety that decreases the threshold for detection by MS, enabling
the analysis of fmol compound/g wet wt tissue [30,31]. Thus relative
changes in methylglyoxal and glyoxal levels within mitochondria
can be assessed based upon the extent of accumulation of the
Materials and methods
Chemical syntheses
A schematic of the syntheses of MitoG, quinoxaline ether (QE),
and the methylquinoxaline ethers 1 and 2 (MQE1/MQE2) is shown
in Fig. 2. In summary, 6-(4-aminophenoxy)hexanol (1) was
synthesized using the reported method [32]. Nitration of 1 was
achieved as described [33] and involved conversion of 1 to the
acetamide followed by nitration with concentrated nitric acid to
give 2. Deprotection then gave the nitroaniline 3 in 48% overall
yield from 1. The basic o-phenylenediamine skeleton was obtained
by catalytic hydrogenation of the nitroaniline 3 over palladium on
carbon. The air- and light-sensitive diamine 4 was immediately
protected with tert-butyloxycarbonyl (Boc) groups by treatment
with di-tert-butyldicarbonate in tetrahydrofuran [34]. The primary
alcohol in 5 was mesylated to give 6 and then converted to the
phosphonium functional group by reaction with triphenylpho-
sphine and sodium iodide in acetonitrile. The product 7 was
obtained by precipitation from ether and column chromatography
to give a white solid in 80% yield. To obtain a robust analytical
Fig. 2. Syntheses of MitoG, MQE, and QE. A detailed description of the synthetic procedures and product characterization and yields for all the reactions is given in the
supplementary materials.

4
40
P.B.L. Pun et al. / Free Radical Biology and Medicine 67 (2014) 437–450
sample, anion exchange to the tetraphenylborate was carried out
by treatment of 7 with sodium tetraphenylborate in dichloro-
methane. Deprotection of the amino groups to give MitoG was
accomplished by treatment of 7 in 1,4-dioxane with 9.8 M hydro-
chloric acid. MitoG was then reacted with either glyoxal to give the
quinoxaline QE or with methyl glyoxal to give the two methylqui-
noxaline products, MQE1 and MQE2, which were formed in a ratio
was determined using the biuret assay relative to bovine serum
albumin and was typically 40–60 mg/ml. A Clark-type oxygen elec-
trode (Rank Brothers, Bottisham, Cambridge, UK) connected to a
Powerlab 2/20 data acquisition system (ADInstruments, Bella Vista,
NSW, Australia) was used to measure respiration rates and was
calibrated with air-saturated water (210 nmol O
(2 mg protein/ml) were suspended in 120 mM KCl, 10 mM Hepes,
1 mM EGTA, 1 mM MgCl , and 5 mM KH PO , pH 7.2 (KOH) in a
2
/ml at 37 1C). RLM
1
of 10:1 (by H NMR), although they gave a single HPLC peak. Data
2
2
4
are quoted for the major isomer. Further details of the syntheses,
including the associated synthesis of 4-hexyloxyphenylene-1,2-
diamine (HP), are given in the supplementary material.
The glyoxalase I inhibitor, bromobenzyl glutathione cyclopentyl
diester, was synthesized by alkylation of glutathione [35], followed
by standard Boc protection, coupling with cyclopentanol using
thermostated 1-ml electrode chamber with stirring. MitoG was then
added and after 5 min glutamate and malate (5 mM each) were
added, followed 3 min later by 400 mM ADP. Oxygen consumption
rates were determined from the slopes using Chart version 5.5.6. for
Mac (ADInstruments). An electrode selective for the TPP moiety of
MitoG was constructed and used as previously described [38].
2
-(1H-benzotriazole-1-yl)-1,1,3,3-tetramethyluronium tetrafluoro-
borate [36], trifluoroacetic acid (TFA) removal of the Boc group,
and treatment with sodium bicarbonate to give the free base.
Other reagents were obtained from Sigma–Aldrich, unless other-
Cell culture
All cells were incubated at 37 1C in a humidified atmosphere of
95% air and 5% CO and culture media used were supplemented
2
with 10% (v/v) fetal calf serum (FCS), 100 U/ml penicillin, and
100 mg/ml streptomycin. C2C12 cells (mouse myoblast cell line;
European Collection of Animal Cell Cultures) were cultured in
wise stated. 1H, C, and P spectra were acquired on Varian
INOVA-400 or Varian INOVA-500 spectrometers. Chemical shifts
are reported downfield from TMS, relative to the solvent residual
signal. For the chemical synthesis components of this work,
high-resolution mass spectra were recorded on a Bruker microTOF
electrospray mass spectrometer and HPLC analysis was carried out
on an Agilent HP1100: column Phenomenex Prodigy 250 ꢁ 3 mm,
13
31
low-glucose (1000 mg/L
D-glucose) Dulbecco’s modified Eagle’s
medium (DMEM; Invitrogen). Cells were maintained at subcon-
fluency (o80%) to prevent differentiation. BAECs (bovine aortic
endothelial cells; Cell Applications, San Diego, CA, USA) were
gradient elution 10% acetonitrile/water (0.1% TFA) to 100% acet-
ꢂ1
onitrile over 12.5 min at 0.5 ml min
54 nm.
with detection at 210 and
maintained in
ing 5 mM -glucose. All culture flasks and assay plates used for
experiments were precoated with fibronectin (25 g/ml; Sigma) in
Hanks’ balanced salt solution (Cell Applications) at 1–4 g fibronec-
α
-minimum essential medium (Invitrogen) contain-
2
D
μ
Assessment of compound properties
μ
2
tin/cm . After 1 h at room temperature excess coating solution was
removed, and flasks and plates were then seeded with cells for
experiments. BAECs were used for experiments at passages 4–6.
To assess cell viability, C2C12 cells or BAECs were seeded at a
density of 10,000 or 40,000 cells/well, respectively, in 96-well
plates. After an overnight incubation, the medium was replaced
with fresh medium containing test compounds and incubated
for 24 h. To determine cell survival, cells were washed twice
with medium, then fresh medium (100 ml per well) was added
and mixed with 20 ml 3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxy-
methoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium (MTS)/phena-
zine methosulfate (Promega, Madison, WI, USA). After 2 h
absorbance was read at 490 nm in a plate reader (SpectraMax
Plus 384; Molecular Devices). All treatments were conducted in
triplicate wells.
TPP-conjugated compounds were made up as 10 mM stock
solutions in dimethyl sulfoxide (DMSO), flushed with argon,
and stored as aliquots at ꢂ20 1C. UV/visible spectra and kinetic
analyses were done using a Shimadzu UV-2501PC spectrophot-
ometer in a 1-ml cuvette containing KCl buffer (120 mM KCl,
1
0 mM Hepes, and 1 mM EGTA, pH 7.2 (KOH)). The molar extinc-
tion coefficients were calculated from the absorbance at maximal
absorption ( max) for the compounds for solutions of known
λ
concentration. Reaction rates between MitoG and 1,2-dicarbonyls
were determined from the initial linear slope over the first 30 s at
the
λmax for the formation of the relevant quinoxaline. Fluores-
cence spectra were obtained in 2.5 ml KCl buffer using a Shimadzu
RF-301PC fluorimeter with slit widths of 3 nm. Kinetic assays used
excitation and emission wavelengths of 344 and 433 nm, respec-
tively; emission spectra used an excitation wavelength of 344 nm
and excitation spectra used an emission wavelength of 433 nm.
RP-HPLC was performed using a Gilson 321 pump with a C18
column (Jupiter 300 Å, Phenomenex) with a Widepore C18 guard
column (Phenomenex). Samples (1 ml) were injected through a
The Seahorse XF24 extracellular flux analyzer was used to
assess cellular oxygen consumption rate (OCR) [39–41]. Cells
(40,000 BAECs/well) were cultured and subjected to experimental
treatment in Seahorse XF24 V7 assay plates that were coated with
fibronectin as described above. OCR was determined as follows:
cells were washed twice in assay medium (4.15 g/L DMEM base,
1.85 g/L NaCl, 1 ꢁ Glutamax (Invitrogen), 1 mM sodium pyruvate,
0
.22- m polyvinylidene difluoride (PVDF) filter (Millex, Millipore).
μ
HPLC buffer A (0.1% TFA in water) and HPLC buffer B (90%
acetonitrile and 0.1% TFA) were used and a gradient was run at
5 mM D-glucose, 15 mg/L phenol red, 20 mM Hepes, and 0.4% (v/v)
1
2
ml/min at room temperature as follows: 0–2 min, 5% B;
–17 min, 5–100% B; 17–19 min, 100% B; 19–22 min, 100–5% B.
FCS, pH 7.4) and incubated in 630 ml of assay medium for 1 h at
37 1C in air. After 20 min equilibration, basal OCR was determined,
and after 3 h incubation with MitoG, OCR was determined after
sequential injections of oligomycin (1 mg/ml), carbonylcyanide
p-trifluoromethoxyphenylhydrazone (FCCP; 2 mM), and rotenone/
antimycin A (4 mg/ml and 5 mM, respectively). OCRs were normal-
ized to cell number as measured using the sulforhodamine B (SRB)
assay as follows: cells were fixed (200 ml of 5% trichloroacetic acid
at 4 1C for 1 h), washed with water (3 ꢁ ), stained for 20 min with
Peaks were detected by absorbance at 220 nm (UV/Vis 151; Gilson)
and by fluorescence (excitation and emission wavelengths of 344
and 433 nm; RF-10AXL; Shimadzu). Partition coefficients between
phosphate-buffered saline (PBS) and octan-1-ol were determined
as previously described [37].
Mitochondrial preparation and incubations
5
0 ml of 0.4% (w/v) SRB in 1% (v/v) acetic acid, and washed (3 ꢁ 1%
Rat liver mitochondria (RLM) were prepared at 4 1C in STE buffer
acetic acid) and the incorporated SRB dye was solubilized in 100 ml
of 10 mM Tris base for 5 min. Then 50 ml of each sample was
transferred in duplicate to a 96-well plate and mixed with 50 ml of
(250 mM sucrose, 5 mM Tris, and 1 mM EGTA, pH 7.4 (HCl)) by
homogenization and differential centrifugation. Protein concentration

P.B.L. Pun et al. / Free Radical Biology and Medicine 67 (2014) 437–450
441
1
0 mM unbuffered Tris base and absorbance was read at 565 nm in
PVDF; Millex; Millipore), and dried under vacuum (Savant SpeedVac).
The dried samples were resuspended in 150 l 20% acetonitrile/0.1%
formic acid by vortexing for 5 min, followed by centrifugation at
16,000g for 10 min. Samples were then transferred to silanized
autosampler vials (Chromacol 1.5HRRV(S)), flushed and sealed under
argon, and stored at ꢂ80 1C until LC–MS/MS analysis.
a plate reader (SpectraMax Plus 384; Molecular Devices). A
standard curve was constructed by seeding a known number of
cells in quadruplicate wells and fixing and staining them in
parallel with sample wells. To calculate the proportion of oxygen
consumption attributable to proton leak and reserve capacity,
OCRs after injection of oligomycin, FCCP, and rotenone/antimycin
A were expressed as a percentage of the total basal OCR [40,42].
To assess total methylglyoxal formation in cells by RP-HPLC, the
cell layers were washed in PBS (1 ml), scraped into 1.5 ml PBS, and
pelleted by centrifugation (16,000g for 2 min). The cell pellet was
resuspended in 500 ml KCl buffer, 500 ml 0.1% formic acid was
added, the protein was pelleted by centrifugation as above, and the
supernatant was collected. Methylglyoxal was then derivatized to
μ
LC–MS/MS analysis
The MS fragmentation patterns of TPP compounds were deter-
mined by direct infusion of compounds (1 mM in 20% acetonitrile
(v/v)) at 2 ml/min into a triple-quadrupole mass spectrometer
(Waters Quattro Ultima). Electrospray ionization in positive ion
mode was used with the following settings: source spray voltage,
3 kV; cone voltage, 100 V; ion source temperature, 80 1C; collision
energy, 50 V. Nitrogen and argon were used as the curtain and
collision gas, respectively.
2
-methylquinoxaline by addition of 100 mM o-phenylenediamine
followed by incubation at 37 1C for 4 h. Samples were dried under
vacuum and assessed for 2-methylquinoxaline as previously
described [28]. Briefly, dried samples were resuspended in 1 ml
LC–MS/MS analyses were carried out using a triple-quadrupole
mass spectrometer (Waters Xevo TQ-S) with an I-class Aquity LC
system (Waters) attached. Samples were kept at 4 1C before injec-
HPLC buffer (68% (v/v) 10 mM KH
trile), filtered (0.22- m PVDF; Millex; Millipore), and separated
by isocratic RP-HPLC in the HPLC buffer above at a flow rate of
ml/min at room temperature using a Gilson 321 pump with a
2 4
PO , pH 2.5, and 32% acetoni-
μ
tion by the autosampler of 10 l flow-through needle
and RP-HPLC at 30 1C using a Luna 5 Phenyl-Hexyl column
m; Phenomenex) with a Phenyl-Hexyl guard
μ
l into a 15-
μ
2
μ
C18 column (Jupiter 300 Å; Phenomenex) and a Widepore C18
guard column (Phenomenex). Peaks were detected fluorimetrically
(1ꢁ 50 mm,
5
μ
column (2 ꢁ 4 mm; Phenomenex). Buffers used were MS buffer A
(
excitation and emission wavelengths of 352 and 385 nm, respec-
(0.1% (v/v) formic acid in water) and MS buffer B (95% acetonitrile/
tively; RF-10AXL; Shimadzu).
0.1% formic acid (both v/v)). A gradient was run at 50 l/min as
μ
For LC–MS/MS analysis of MitoG reaction products cells were
follows: 0–2 min, 5% B; 2–3 min, 5–25% B; 3–5 min, 25–75% B; 5–
7 min, 75–100% B; 7–10 min, 100% B; 10–12 min, 100–5% B; 12–
20 min, 5% B. Eluant was diverted to waste from the mass spectro-
meter at 0–5 and 16–20 min acquisition time using an in-line divert
valve. For MS analysis, electrospray ionization in positive ion mode
was used: source spray voltage, 2.5 kV; cone voltage, 25 V; ion
source temperature, 100 1C; collision energy, 38 V. Nitrogen and
argon were used as the curtain and the collision gas, respectively.
Multiple reaction monitoring in positive ion mode was used for
compound detection. Transitions used for quantification were as
follows: MQE, 5064262; d15-MQE, 5214277; QE, 24262; and
d15-QE, 5074277. For each experiment standard curves were
prepared using known amounts of MQE and QE, which were spiked
with IS and extracted in parallel with the samples. Standards and
samples were quantified using MassLynx 4.1 software to determine
the peak area for MQE, QE, and ISs, and the standard curves were
used to determine the amounts of MQE and QE present in samples.
incubated in T25 flasks (Nunc) with 2 M MitoG at 37 1C. Cell
monolayers were washed (1 ml PBS), scraped into 1.5 ml PBS, and
pelleted by centrifugation (16,000g for 2 min). The pellet was
μ
resuspended in 250 l 100% acetonitrile/0.1% spiked with deuter-
μ
ated internal standards (ISs; 100 pmol each d15-MQE and d15-QE),
vortexed, and centrifuged (2 ꢁ 16,000g for 15 min). Samples were
dried under vacuum (Savant SpeedVac), resuspended in 100 l
μ
2
0% acetonitrile/0.1% formic acid, vortexed, centrifuged (16,000g
for 10 min), transferred to silanized autosampler vials (Chromacol
.5HRRV(S)), flushed and sealed under argon, and then stored at
1
ꢂ80 1C until LC–MS/MS analysis.
Where specified, cells were subjected to acute glycemia treat-
ment and were incubated in low-glucose (5 mM
glucose (30 mM -glucose), or osmotic control medium (5 mM
-glucoseþ25 mM -glucose) for 4 h.
D-glucose), high-
D
D
L
Mouse experiments
Statistics
þ/ꢂAkitaJ
The Akita mouse model of type I diabetes (Ins2
) [43–45]
was assessed at the University of Alabama. All procedures were
performed in accordance with Guide for the Care and Use of
Laboratory Animals and were approved by the Institutional Animal
Data analysis was performed with the R software environment
for statistical computing and graphics (R Foundation for Statistical
Computing, Wien, Austria). All data were analyzed using t tests or
one-way analysis of variance followed by post hoc Dunnett’s test
as appropriate and represented as the mean7standard error
of mean. P values equal to or less than 0.05 were taken to be
statistically significant.
Care and Use Committee at the University of Alabama at Birming-
þ/ꢂAkitaJ
ham. Male Ins2
and wild-type (C57BL/6) mice (4–8 weeks of
age) from The Jackson Laboratory (Bar Harbor, ME, USA) were
maintained on laboratory chow and water ad libitum until 14 weeks
of age, when they were used for experiments. For this MitoG
(100 nmol in 100 ml saline) was administered by tail vein injection.
After 4–6 h the mice were killed and urine samples taken and snap-
frozen for subsequent analysis of MQE/QE content. Blood glucose
levels were measured using an Accu-Chek Advantage blood glucose
meter (Roche Diagnostics). Urine creatinine levels were determined
at the Core Biochemical Assay Laboratory (Addenbrooke’s Hospital,
Results and discussion
Synthesis and characterization of MitoG and its reaction products
The syntheses of MitoG, its predicted quinoxaline products
upon reaction with methylglyoxal and glyoxal, and their deuter-
ated versions are summarized in Fig. 2. The UV/Vis absorption
spectra of MitoG and its component TPP and phenylenediamine
moieties are shown in Fig. 3A. As expected, the spectrum of MitoG
was a summation of those of the phenylenediamine HP and a
simple alkyl-TPP salt. As o-phenylenediamine undergoes oxidative
Cambridge, UK). To extract MQE and QE, 20
μ
l urine was mixed with
l 60% acetonitrile/0.1% formic acid and incubated on ice for
0 min. The extracts were then spiked with deuterated ISs (100 pmol
500
μ
3
each d15-MQE and d15-QE), vortexed for 30 s, incubated on ice for
0 min with vortexing every 10 min, and then centrifuged (16,000g
for 10 min). The supernatants were collected, filtered (0.22-
3
μm

4
42
P.B.L. Pun et al. / Free Radical Biology and Medicine 67 (2014) 437–450
degradation [2], and because there was a literature precedent for
the additional reactivity of alkoxy-substituted phenylenediamines,
we assessed the stability of MitoG by RP-HPLC and found that
stock solutions were stable under storage at ꢂ20 1C. To assess its
stability under biologically relevant conditions, we measured the
decomposition of dilute solutions of MitoG at 37 1C and observed
0.5
0.8
HP
0
0
0
0
0
0
0
.7
.6
.5
.4
.3
.2
.1
0
0
0
0
0
.4
.3
.2
.1
0
MQE
QE
MitoG
ButylTPP
350
MitoG
330
2
50
300
400
450
500
550
250
270
290
310
350
370
390
Wavelength (nm)
Wavelength (nm)
900
600
300
0
1.8
100
90
1
00
MQE
Excitation
Emission
MQE
QE
1.5
1.2
MQE
80
8
0
0
0
0
7
6
0
0
6
4
2
MitoG
0.9
50
40
QE
QE
0.6
3
2
1
0
0
0
0
0
0
.3
MitoG
320 340
MitoG
420 440 460
0
3
00
360
380 400
Wavelength (nm)
480
5
6
7
8
9
10
11
12
Time (min)
0
0
0
0
0
0
0
0
.8
.7
.6
.5
.4
.3
.2
.1
0
100
Excitation
Emission
MitoG + methylglyoxal
8
6
4
2
0
0
0
0
MitoG + methylglyoxal
MitoG + glyoxal
MitoG + glyoxal
380 400 420
Wavelength (nm)
0
300
250
270
290
310
330
350
370
390
320
340
360
440
460
480
Wavelength (nm)
9
6
3
00
00
00
0
1.8
.5
1.2
.9
900
600
300
0
1.8
1.5
1.2
0.9
0.6
0.3
0
MitoG + methylglyoxal
MitoG + methylglyoxal + MQE
1
0
0.6
0
0
.3
5
6
7
8
9
10
11
12
5
6
7
8
9
10
11
12
Time (min)
Time (min)
Fig. 3. In vitro characterization of MitoG, MQE, and QE. (A) UV/Vis scanning spectra of 100 μM MitoG show the characteristics of its component TPP and HP moieties. (B) UV/Vis
scanning spectra of 100 μM MitoG, MQE, and QE. (C) Fluorescence excitation spectra (emission 433 nm) and emission spectra (excitation 344 nm) of 10 μM MitoG, MQE, and QE. MitoG
was not fluorescent. MQE and QE had peak excitation and emission wavelengths of 344 and 433 nm, respectively. (D) RP-HPLC profile of 10 nmol each of MitoG, MQE, and QE.
Absorbance (red) at 220 nm, fluorescence (blue) was observed at excitation and emission wavelengths of 344 and 433 nm, respectively. (E) UV/Vis scanning spectra of 100 μM MitoG
and 1 mM methylglyoxal or glyoxal in KCl buffer after incubation at 37 1C for 2 h. (F) Fluorescence excitation and emission spectra of 10 μM MitoG and 20 μM methylglyoxal, or of
20 μM MitoG and 40 μM glyoxal, in KCl buffer after incubation at 37 1C for 2 h. (G) MitoG (5 mM) and 10 mM methylglyoxal were incubated in 10 μl KCl buffer at 37 1C for 2 h, then 1 μl
of the mixture was assessed by RP-HPLC. (H) The identity of the product peak was confirmed by spiking the reaction mixture with 10 nmol of MQE standard. Experiments with glyoxal
gave similar results (data not shown).(For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article).

P.B.L. Pun et al. / Free Radical Biology and Medicine 67 (2014) 437–450
443
minimal loss by 4 h with significant loss by 24 h; therefore MitoG
Accumulation of MitoG within energized mitochondria
is sufficiently stable for biological experiments lasting up to ꢀ4 h.
The expected products of the reaction of MitoG with methyl-
glyoxal (MQE) and glyoxal (QE) were synthesized and have UV/Vis
absorption spectra distinct from that of MitoG (Fig. 3B). Both QE
and MQE were fluorescent, whereas MitoG was not (Fig. 3C). The
three compounds were resolved by RP-HPLC (Fig. 3D) although the
MQE regioisomers (MQE1, MQE2; Fig. 1B) were not distinguished.
Both MQE and QE were stable after 1 week at 37 1C as assessed by
RP-HPLC (data not shown). The properties of MitoG, MQE, and QE
are summarized in Table 1. We conclude that MitoG is sufficiently
stable for biological experiments involving trapping of glyoxal and
methylglyoxal within mitochondria in cells and in vivo, and the
reaction products MQE and QE were both robustly stable for
chemical isolation and subsequent LC–MS/MS analyses.
To be a mitochondria-selective probe for 1,2-dicarbonyls in
cells and in vivo, the TPP moiety of MitoG should promote its
uptake within mitochondria. We measured the uptake of MitoG by
isolated mitochondria using an electrode selective for the TPP
moiety of MitoG (Fig. 5). After addition of MitoG to calibrate the
electrode response, subsequent addition of mitochondria led to a
small decrease in MitoG concentration due to the expected
adsorption of MitoG to unenergized mitochondria [38]. Addition
of the respiratory substrate succinate generated a membrane
potential and led to the substantial uptake of MitoG as shown by
a decrease in the external concentration. Dissipation of the
membrane potential with the uncoupler FCCP led to the release
of MitoG from the mitochondria. The membrane potential-
dependent uptake of MitoG into mitochondria was ꢀ2 nmol/mg
protein, which, assuming a mitochondrial matrix volume of
0
.5
mitochondria of ꢀ4 mM, whereas the external MitoG concentra-
tion was ꢀ2 M. This ꢀ2000-fold accumulation of MitoG
indicates that MitoG, like other TPP-conjugated compounds, is
μ
l/mg protein [31], gives a concentration of MitoG within
Reactivity of MitoG with glyoxal and methylglyoxal
μ
MitoG reacted with methylglyoxal and glyoxal to form products
with absorption, fluorescence, and RP-HPLC properties (Figs. 3E–H)
identical to those of independently synthesized and characterized
MQE or QE (Figs. 3B–D). Incubation of MitoG with the biologically
important and reactive aldehydes 4-hydroxynonenal and acrolein
followed by RP-HPLC analysis indicated that although MitoG
did react, a number of products were formed, some of which were
unstable (data not shown). Thus the reaction of MitoG with these
aldehydes is not diagnostically useful, in contrast to that with
0
0
0
0
0
.5
.4
.3
.2
.1
Methylglyoxal
1,2-dicarbonyls, which generates stable, dominant products.
Dicarbonyl
Progress of the reaction between MitoG and methylglyoxal could
be observed by UV/Vis spectrophotometry, but that of MitoG with
glyoxal was too slow (Fig. 4A). However, both reactions could be
assessed using a more sensitive fluorimetric method (Fig. 4B).
The greater reactivity of methylglyoxal over glyoxal with o-phenyle-
nediamines is consistent with the enhanced toxicity of methylglyoxal
MitoG
0
Glyoxal
[46–48]. Aqueous glyoxal exists predominantly as the dihydrate, which
is an unreactive tetraol, whereas methylglyoxal is predominantly the
monohydrate with only the aldehyde converted into a 1,1-diol. This
accounts for the greater reactivity of methylglyoxal, which requires
only a single dehydration to generate the highly reactive dicarbonyl
0
50
100
150 200 250
Time (s)
300
[
8]. The rate constants for the MitoG–methylglyoxal and the MitoG–
ꢂ1 ꢂ1
250
glyoxal reactions were 1971 and 0.670.3 M
s
(means7SE,
n¼3) at 37 1C, respectively, compared with the reported value of
ꢂ1 ꢂ1
1.7 70.3 M
s
for the reaction of methylglyoxal with o-phenyle-
200
Methylglyoxal
nediamine at 25 1C [49]. A competitive rate experiment for methyl-
1
glyoxal between MitoG and o-phenylenediamine, monitored by
H
1
1
50
00
NMR (0.12 mM, DMSO, 25 1C), showed that MitoG was ꢀ1.8 times
more reactive than o-phenylenediamine, whereas a similar competi-
tion for glyoxal between 4-methoxyphenylene-1,2-diamine and
o-phenylenediamine showed that the 4-methoxy substitution
enhanced reactivity ꢀ3.8-fold. Together these data confirm that
MitoG reacts with 1,2-dicarbonyls to form stable diagnostic products
and that the electron-donating ether linkage in MitoG enhances the
reactivity compared to unsubstituted o-phenylenediamine.
Dicarbonyl
MitoG
5
0
Glyoxal
0
0
50
100
150 200 250
Time (s)
300
Table 1
M extinction coefficients and partition coefficients of MitoG, MQE, and QE.
Compound Molar extinction coefficient at wavelength
Partition
coefficient
Fig. 4. In vitro reaction of MitoG with methylglyoxal and glyoxal. (A) The reactions
between MitoG (100 μM) and 1,2-dicarbonyls (25 μM) at 37 1C were followed from
the formation of the quinoxaline products at 345 nm. (B) Fluorimetric detection of
the reaction between MitoG and methylglyoxal (40 μM each), and between MitoG
and glyoxal (200 μM each), was done by monitoring the quinoxaline products at
excitation and emission wavelengths of 344 and 433 nm, respectively. The changes
in fluorescence caused by MQE and QE formation were quantified from calibration
curves constructed by plotting fluorescence (excitation and emission wavelengths
of 344 and 433 nm, respectively) against known amounts of either quinoxaline and
were linear up to at least 20 mM.
ꢂ1
ꢂ1
λ (M
cm
)
λ 267 nm
λ 274 nm
λ 299 nm
λ 345 nm
MitoG
MQE
QE
40697206 37217193 3679 7188
64637139 55097127
52757293 45577245
2.470.1
74667158 12.370.5
69017375 13.670.9
Data are means7SE of three determinations.

4
44
P.B.L. Pun et al. / Free Radical Biology and Medicine 67 (2014) 437–450
RLM
Succinate
MitoG as probe for mitochondrial methylglyoxal and glyoxal in cells
FCCP
For MitoG to be an effective probe it should react with
methylglyoxal and glyoxal within a biological system to give the
diagnostic products, MQE and QE, which can then be extracted and
analyzed by LC–MS/MS. To assess whether this was possible in
cells, BAECs were preincubated with MitoG for 1 h, then methyl-
glyoxal or glyoxal was added, and after a further 3 h the cell layers
were extracted and analyzed by LC–MS/MS to assess the amounts
of MQE (Fig. 9A) and QE (Fig. 9B). The levels of both MitoG-derived
products, MQE and QE, initially increased with the concentration
of exogenous 1,2-dicarbonyls added before showing saturation at
supraphysiological 1,2-dicarbonyl concentrations, at which MitoG
became limiting (Figs. 9A and B). Treatment with the 1,2-dicarbo-
nyl scavenger aminoguanidine (AG), or decreasing MitoG mito-
chondrial uptake using the uncoupler FCCP, reduced the amounts
of MQE and QE detected (Figs. 9A and B). As some MitoG is present
in the culture medium, there will also be a contribution from
MQE/QE formation in the supernatant that is subsequently accu-
mulated by the cells. These findings are consistent with reaction
of MitoG with methylglyoxal and glyoxal within cells to form
MQE/QE and this reaction being decreased by AG or by lowering
the extent of MitoG uptake into mitochondria within cells by
dissipating the membrane potential. We conclude that MitoG
reacts with methylglyoxal or glyoxal in a biological context to
form MQE and QE and that these products can be extracted from
cells and quantified by LC–MS/MS.
!
"#$%
MitoG
additions
1
min
Fig. 5. Uptake of MitoG by isolated mitochondria. A TPP-selective electrode was
placed in a stirred chamber at 37 1C containing 1 ml KCl buffer supplemented with
μg/ml rotenone. After calibration with five additions of 1 mM MitoG (arrowheads),
RLM (1 mg protein/ml) were added, followed by 10 mM succinate and 500 nM FCCP
where indicated. This trace is representative of three independent experiments.
4
selectively accumulated by mitochondria in a membrane potential-
dependent manner.
We next utilized MitoG to determine relative mitochondrial
levels of methylglyoxal and glyoxal under hyperglycemia, a
condition to which damaging glycation by 1,2 dicarbonyls is
thought to contribute. First we confirmed that hyperglycemia
did increase the production of cellular methylglyoxal in our
system. To do this we incubated BAECs under conditions of high
(30 mM) and low (5 mM) glucose for 4 h and then measured the
formation of methylglyoxal by derivatization with o-phenylene-
diamine to generate 2-methylquinoxaline, which was assessed by
RP-HPLC. This analysis showed that hyperglycemia in BAECs did
indeed increase the formation of methylglyoxal ꢀ2-fold com-
pared to controls (data not shown, n¼3). To see whether MitoG
could be used to assess a change in mitochondrial methylglyoxal/
glyoxal under hyperglycemic conditions we next compared the
formation of MQE and QE in cells after incubation with low
(5 mM) or high (25 mM) glucose for 4 h (Figs. 9C and D). There is
a gradual increase in the amount of MQE detected over time and
this increases substantially on going from low to high glucose,
consistent with the increase in methylglyoxal within mitochon-
dria under conditions of hyperglycemia. This formation of MQE/
QE was blocked by the uncoupler FCCP (Figs. 9C and D) and the
formation of MQE was also decreased by the methylglyoxal trap
AG (Fig. 9E). The increase in MQE upon hyperglycemia did not
For MitoG to be useful, it should not disrupt mitochondrial
function or cause cell toxicity at the concentrations used. To test
this we incubated MitoG with isolated mitochondria and assessed
its effect on respiration under different conditions. At 5 M MitoG
μ
there was a slight increase in proton leak as indicated by an
increase in coupled respiration (Fig. 6A). This is expected as high
levels of TPP cations within biological membranes eventually
cause increased proton leak. In contrast, the oxidative phosphor-
ylation complexes themselves were insensitive to MitoG up to 5
μ
M as there was no effect on phosphorylating respiration (Fig. 6B).
MitoG only decreased the viability of C2C12 cells and BAECs at
concentrations above 10 and 50 M, respectively (Figs. 6C and D).
μ
To assess the effect of MitoG on mitochondrial function within
cells, we measured the OCR using a Seahorse XF24 flux analyzer
(
Figs. 6E–H). Concentrations of MitoG above 10
μ
M slightly
decreased the OCR owing to ATP synthesis, and lower concentra-
tions in the range 2–5 mM showed a trend toward an increased
OCR due to proton leak and a decreased respiratory reserve
capacity. Therefore, we routinely used a MitoG concentration of
2 μM for most cell experiments.
occur when the high concentration of
with 25 mM nonphysiological -glucose (along with 5 mM
-glucose to maintain cell viability). This suggests that the
increase in MQE caused by a high concentration of -glucose
D-glucose was replaced
Quantification of reaction of MitoG with 1,2-dicarbonyls
by LC–MS/MS
L
D
D
To use MitoG to probe the local concentration of methylglyoxal
and glyoxal, it was necessary to measure the amounts of the
reaction products (MQE, QE) by LC–MS/MS relative to deuterated
internal standards [31]. The fragmentations of MQE and QE and their
deuterated versions during tandem MS were determined (Fig. 7) and
were as expected for TPP compounds [31]. This fragmentation pattern
requires metabolism of the glucose to generate methylglyoxal
and is not due to nonspecific effects of a high carbohydrate
concentration in the culture medium (Fig. 9E). The reaction
between MitoG and methylglyoxal/glyoxal may also occur out-
side mitochondria, with the subsequent uptake of MQE/QE
within mitochondria. However, as the reaction between MitoG
and the dicarbonyls is second order, the rate of MQE/QE forma-
tion within mitochondria is expected to be ꢀ500- to 1000-fold
greater than in other compartments, even if the methylglyoxal
and glyoxal concentrations were the same. Therefore these data
are consistent with the formation of MQE/QE occurring primarily
within the mitochondria.
—
to the triphenylphosphorus cation (Fig. 8A)—was used to establish
an LC–MS/MS assay for the products of the reaction of MitoG with
,2-dicarbonyls, and typical standard curves are shown in Fig. 8B. We
1
conclude that MitoG reaction products can be very sensitively
detected, facilitating the use of MitoG to assess mitochondrial methyl-
glyoxal and glyoxal in cells and in vivo.

P.B.L. Pun et al. / Free Radical Biology and Medicine 67 (2014) 437–450
445
1
1
1
40
20
00
140
120
100
80
8
0
0
0
0
6
4
2
60
40
20
0
0
1
3
5
1
3
5
1
1
1
40
20
00
140
120
100
80
8
0
0
0
0
*
6
4
2
60
*
40
*
*
20
*
*
0
0
0
.1
0.5
1
5
10
50
100
500
0.1
0.5
1
2
5
10
50
100
200
8
0
0
0
0
0
0
0
0
20
15
7
6
5
4
3
2
1
10
5
0
0
Untreated
controls
1
2
5
10
Untreated
controls
1
2
5
10
3
2
2
1
1
00
50
00
50
00
30
25
20
15
10
5
5
0
0
0
Untreated
controls
1
2
5
10
Untreated
controls
1
2
5
10
Fig. 6. Effects of MitoG on mitochondrial and cell function. (A, B) RLM respiring on glutamate/malate at 37 1C were incubated with various concentrations of MitoG for 7 min
in an oxygen electrode (A) to measure coupled respiration before ADP was added (B) to measure phosphorylating respiration. Data are the percentage of the respiration rates
of untreated controls (dashed lines). (C, D) (C) C2C12 cells or (D) BAECs were incubated with MitoG for 24 h and cell survival was then determined using the MTS assay.
Data are expressed as a percentage of the untreated controls (dashed line). (E–H) BAECs were incubated with MitoG for 2 h at 37 1C and the cellular oxygen consumption rate
(
OCR) after the sequential additions of oligomycin, FCCP, and antimycin A/rotenone was then measured using a Seahorse XF24 analyzer. (E) OCR due to ATP synthesis, (F) OCR
n nn
due to proton leak, (G) reserve capacity, and (H) nonmitochondrial oxygen consumption. Results are means7SE of three independent experiments. Po0.05 or Po0.01
relative to untreated controls.

4
46
P.B.L. Pun et al. / Free Radical Biology and Medicine 67 (2014) 437–450
Fragment ions
Precursor ion
100
100
0
0
400
450
500
550
600
650
100
150
200
250
300
350
100
100
0
0
400
450
500
550
600
650
100
150
200
250
300
350
100
100
0
0
400
450
500
550
600
650
100
150
200
250
300
350
100
100
0
0
400
450
500
550
600
650
100
150
200
250
300
350
Fig. 7. Fragmentation of MQE and QE by tandem mass spectrometry. Compounds (1 μM in 20% acetonitrile) were infused, at 2 μl/min, into a triple-quadrupole mass
spectrometer. The indicated parent ions of MQE and QE and their corresponding d15 variants were fragmented to generate the indicated daughter ions.
When BAECs were incubated under conditions of high glucose in
the presence of the glyoxalase I inhibitor bromobenzyl glutathione
cyclopentyl diester [35], to suppress degradation of glyoxal and
methyl glyoxal, the amount of MQE and QE increased (Fig. 9F).
Together these data indicate that MitoG is an effective probe for
assessing changes in mitochondrial 1,2-dicarbonyl production within
cells. Furthermore, these findings suggest that the mitochondrial
methylglyoxal concentration increases approximately threefold
under hyperglycemia. Therefore mitochondrial glycation by elevated
reactive dicarbonyls is a strong candidate to contribute to the

P.B.L. Pun et al. / Free Radical Biology and Medicine 67 (2014) 437–450
447
Fig. 8. LC–MS/MS analysis of MQE and QE. (A) QE and both isoforms of MQE have parent ions with distinctive m/z ratios that fragment to form characteristic daughter ions.
B) Standard curves based on the analyses of MQE and QE by LC–MS/MS analysis relative to the corresponding deuterated ISs.
(
disruption of mitochondrial function that occurs during pathological
hyperglycemia.
goal, as it is known that after intravenous injection TPP com-
pounds rapidly distribute from the blood to mitochondria within
tissues and are then slowly excreted over several hours into the
urine and bile [27,52,53]. Therefore it may be possible to admin-
ister MitoG to mice and then analyze urine to see if there is
elevated production of the MQE/QE products under certain con-
ditions, indicating an elevation in reactive dicarbonyls within
MitoG as a probe for mitochondrial levels of methylglyoxal and
glyoxal in vivo
Previously we have shown that the mitochondria-targeted
hydrogen peroxide mass spectrometry probe MitoB can be used
to assess the production of hydrogen peroxide within mitochon-
dria in living fruit flies [30,31] and mice [50]. In this, MitoB is
acting as a probe to generate the exomarker MitoP [51], which is an
exogenous probe compound that when administered to an experi-
mental animal is converted to a diagnostic marker that can be
assessed ex vivo and used to infer the production of reactive
species within the living organism [51]. Therefore we next set out
to see if MitoG could also be used to generate the exomarkers MQE
and QE to assess the formation of methylglyoxal/glyoxal within
mitochondria in vivo. The TPP component of MitoG facilitates this
mitochondria in vivo.
þ/ꢂAkitaJ
To do this we used the Akita mouse model (Ins2
) in
which a mutation in proinsulin leads to chronic hyperglycemia
and consequent pathological complications similar to those found
in type I diabetes [43–45]. To assess whether there were changes
in mitochondrial methylglyoxal/glyoxal in the Akita mice com-
pared to wild type, we administered MitoG (100 nmol) as a tail
vein injection and after 4–6 h urine samples were taken and the
MQE and QE contents measured relative to creatinine. These data
are shown as a function of blood glucose levels for individual mice
(Figs. 10A and B). As expected, blood glucose was far higher in the

4
48
P.B.L. Pun et al. / Free Radical Biology and Medicine 67 (2014) 437–450
3
3
2
2
50
00
50
00
70
60
50
40
30
1
1
50
00
+ FCCP
+
AG
20
10
0
5
0
0
+
FCCP
300
+
AG
0
50
100
150
200
250
0
50
100
150
200
250
300
[Methylglyoxal] (µM)
[Glyoxal] (µM)
6
5
4
3
2
1
0
0
0
0
0
0
60
50
40
30
20
10
0
Low glucose
High glucose
Low glucose
High glucose
0
15 min
2 h
4 h
4 h + FCCP
15 min
2 h
4 h
4 h + FCCP
6
0
0
0
0
0
0
0
80
5
4
3
2
1
6
4
2
0
0
0
0
0
10
25
Low
glucose
High
glucose
L-glucose
High glucose +
Aminoguanidine
[Inhibitor] (µM)
Fig. 9. Quantification of MQE and QE in cells. (A, B) BAECs were incubated with 2 μM MitoG for 1 h and sometimes supplemented with 10 mM AG or 2 μM FCCP.
A) Methylglyoxal or (B) glyoxal were then added, and after a further 3 h incubation, the levels of MQE and QE in the cell layers were determined by LC–MS/MS relative to
deuterated ISs. (C, D) BAECs were incubated in media containing low (5 mM) or high (30 mM) -glucose with 2 μM MitoG for the times indicated. Levels of (C) MQE and
D) QE in the cell layers were determined by LC–MS/MS relative to deuterated ISs. Results are means7SE of four independent experiments. (E) BAECs were incubated for 4 h
in the presence of 2 μM MitoG in media containing 5 mM -glucose (low), 30 mM -glucose (high), 5 mM -glucose/25 mM -glucose ( -glucose), or 30 mM -glucose/10 mM
aminoguanidinium. Levels of MQE in the cell layers were determined by LC–MS/MS relative to deuterated ISs. (F) BAECs were incubated in media containing high (30 mM)
-glucose with 2 μM MitoG for 4 h with the indicated concentrations of the glyoxalase I inhibitor bromobenzyl glutathione cyclopentyl diester. Levels of MQE in the cell
(
D
(
D
D
D
L
L
D
D
n
nn
layers were then determined by LC–MS/MS relative to deuterated ISs. Results are means7SE of three (A, B) or four (C–E) independent experiments. Po0.05 or Po0.01,
relative to untreated (A, B) or indicated (C–E) controls; #Po0.05 or ##Po0.01 relative to cells treated with 250 μM 1,2-dicarbonyl.
Akita mice than in wild-type mice, and this increase in blood
glucose correlated with a significant increase in both the MQE and
the QE MitoG adducts, normalized to creatinine (Figs. 10A and B).
These data suggest that MitoG can be used as a probe for the
formation of methylglyoxal and glyoxal under conditions of
pathologically relevant hyperglycemia in vivo.
developed a mass spectrometry approach to measure changes in
the levels of these reactive 1,2-dicarbonyls within mitochondria in
cells and in vivo.
We found that the combination of an o-phenylenediamine with a
TPP moiety led to a molecule that accumulated within mitochondria,
where it reacted with methylglyoxal and glyoxal to form stable
products. These products could then be extracted from cells and
biological fluids and analyzed by LC–MS/MS relative to stable isotope
internal standards. The selectivity of this assessment for methyl-
glyoxal and glyoxal was based on the identification of the products
by tandem mass spectrometry, and this was further demonstrated by
the changes in their levels on pharmacological inhibition of the
glyoxalase system and by addition of the dicarbonyl trap aminogua-
nidine. The ability of the TPP moiety to locate compounds to
mitochondria in cells and in vivo is well established, suggesting that
the formation of MQE/QE occurs predominantly because of the
reaction with methylglyoxal/glyoxal within the mitochondrial matrix.
This is supported by the inhibition of the formation of MQE/QE by
Discussion
Disruption of mitochondrial function has been proposed to play
a role in pathologies associated with hyperglycemia. One plausible
pathway by which this may occur is through damaging glycation
reactions to mitochondrial components by the reactive 1,2-dicar-
bonyls methylglyoxal and glyoxal. However, the importance of this
pathway was difficult to assess in cells and in vivo because of the
uncertainties about the levels of methylglyoxal and glyoxal that
occur within mitochondria during hyperglycemia. Here we have

P.B.L. Pun et al. / Free Radical Biology and Medicine 67 (2014) 437–450
449
hyperglycemia both in cells and in vivo, indicating that there is a
large increase in the amount of methylglyoxal/glyoxal in mitochondria
under these conditions. This is consistent with mitochondrial glycation
due to the accumulation of methylglyoxal/glyoxal in the matrix
contributing to the mitochondrial disruption seen in hyperglycemia.
To conclude, we have developed a new mitochondria-targeted mass
spectrometric approach to assess levels of reactive dicarbonyls within
mitochondria in cells and in vivo. This will be of use in assessing the
contribution of these damaging species to mitochondrial dysfunction
in diabetes and aging.
Wild type
Akita
6
5
4
3
2
1
0
0
5
10
15
20
25
30
35
40
45
Mean values
Blood glucose (mmol/l)
Acknowledgments
This work was supported by the Medical Research Council (UK),
by the Foundation for Research, Science and Technology (NZ), by
Grant BB/I012826/1 from the BBSRC (to R.C.H. and M.P.M.), and by
the NIH (Grant DK075865 to V.D.U.). We thank Dr. Helena
Cochemé for assistance with the figures.
1
1
2
0
8
6
4
2
0
Appendix A. Supplementary material
Supplementary data associated with this article can be found in
the online version at http://dx.doi.org/10.1016/j.freeradbiomed.
2013.11.025.
0
5
10
15
20
25
30
35
40
45
Mean values
Blood glucose (mmol/l)
Fig. 10. Quantification of mitochondrial dicarbonyls in vivo. The levels of (A) MQE
and (B) QE were quantified in urine of wild-type and Akita mice relative to
creatinine and compared with blood glucose levels of these mice. MitoG (100 nmol)
was administered by tail vein injection in both wild-type and Akita mice. After
References
[
[
1] Brownlee, M. Negative consequences of glycation. Metab. Clin. Exp 49:9–13;
000.
2] Thornalley, P. J. Protein and nucleotide damage by glyoxal and methylglyoxal
4
–6 h, blood glucose, urinary creatinine, and urinary MQE/QE levels were mea-
2
sured. The data to the right of the plots are the means7SE of the two conditions.
nn
Po0.01.
in physiological systems—role in ageing and disease. Drug Metab. Drug Interact
23:125–150; 2008.
[
[
3] Pun, P. B.; Murphy, M. P. Pathological significance of mitochondrial glycation.
Int. J. Cell Biol. 2012:843505; 2012.
4] Brownlee, M. Biochemistry and molecular cell biology of diabetic complica-
tions. Nature 414:813–820; 2001.
addition of the uncoupler FCCP, which decreases the uptake of MitoG
into mitochondria. Furthermore, as the reaction between MitoG and
methylgloxal/glyoxal is second order, the ꢀ1000-fold concentration
of MitoG within mitochondria will mean that the rate of formation
of MQE/QE will be greater inside mitochondria by a similar factor.
Thus formation of MQE/QE within a cell or tissue is largely due to
changes in the amount of methylglyoxal/glyoxal within mitochon-
dria. However, other reactive species within biological systems such
as nitric oxide or reactive aldehydes such as 4-hydroxynonenal may
also react with and deplete MitoG, although these reactions will not
produce the diagnostic products formed from the reaction with
glyoxal/methylglyoxal. Furthermore, although this approach will
report on mitochondrial dicarbonyl exposure, it does not indicate
the cellular source of the methylglyoxal/glyoxal, and they may be
largely produced in the cytosol with subsequent diffusion into
mitochondria.
[5] Rabbani, N.; Thornalley, P. J. Glyoxalase in diabetes, obesity and related
disorders. Semin. Cell Dev. Biol. 22:309–317; 2011.
[
6] Phillips, S. A.; Thornalley, P. J. The formation of methylglyoxal from triose
phosphates—investigation using a specific assay for methylglyoxal. Eur. J.
Biochem. 212:101–105; 1993.
[7] Casazza, J. P.; Felver, M. E.; Veech, R. L. The metabolism of acetone in rat. J. Biol.
Chem. 259:231–236; 1984.
[
8] Lo, T. W. C.; Westwood, M. E.; Mclellan, A. C.; Selwood, T.; Thornalley, P. J.
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. 269:32299–32305; 1994.
[
9] Chaplen, F. W.; Fahl, W. E.; Cameron, D. C. Evidence of high levels of
methylglyoxal in cultured Chinese hamster ovary cells. Proc. Natl. Acad. Sci.
USA 95:5533–5538; 1998.
[
10] Thornalley, P. J.; Battah, S.; Ahmed, N.; Karachalias, N.; Agalou, S.; Babaei-Jadidi, R.;
Dawnay, A. Quantitative screening of advanced glycation endproducts in cellular
and extracellular proteins by tandem mass spectrometry. Biochem. J. 375:581–592;
2003.
The ability to use MitoG within living mice to generate MQE/QE
as exomarkers to assess the changes in dicarbonyls that occurred
within a model of type I diabetes was an important step. This
suggests that this approach will be useful in assessing the role of
reactive dicarbonyls in the mitochondrial damage associated with
diabetes and in developing specific therapies. One limitation of
MitoG is that it is not possible to normalize for tissue uptake
because of the decomposition of MitoG over time in aerobic
tissues. Thus changes in the uptake of MitoG into the tissue, for
example by pathological alterations in mitochondrial membrane
potential due to hyperglycemia, might lead to changes in the
accumulation of the exomarker distinct from changes in organelle
glyoxal concentration, and work is under way to circumvent this
limitation.
[
11] Thornalley, P. J.; Waris, S.; Fleming, T.; Santarius, T.; Larkin, S. J.; Winklhofer-
Roob, B. M.; Stratton, M. R.; Rabbani, N. Imidazopurinones are markers of
physiological genomic damage linked to DNA instability and glyoxalase
1
2
-associated tumour multidrug resistance. Nucleic Acids Res. 38:5432–5442;
010.
[12] Kingkeohoi, S.; Chaplen, F. W. R. Analysis of methylglyoxal metabolism in CHO
cells grown in culture. Cytotechnology 48:1–13; 2005.
[13] Dhar, A.; Desai, K.; Liu, J. H.; Wu, L. Y. Methylglyoxal, protein binding and
biological samples: are we getting the true measure? J. Chromatogr.
77:1093–1100; 2009.
14] Thornalley, P. J. The glyoxalase system: new developments towards functional
characterization of metabolic pathway fundamental to biological life.
Biochem. J 269:1–11; 1990.
15] Morcos, M.; Du, X.; Pfisterer, F.; Hutter, H.; Sayed, A. A.; Thornalley, P.; Ahmed, N.;
B
8
[
a
[
Baynes, J.; Thorpe, S.; Kukudov, G.; Schlotterer, A.; Bozorgmehr, F.; El Baki, R. A.;
Stern, D.; Moehrlen, F.; Ibrahim, Y.; Oikonomou, D.; Hamann, A.; Becker, C.;
Zeier, M.; Schwenger, V.; Miftari, N.; Humpert, P.; Hammes, H. P.; Buechler, M.;
Bierhaus, A.; Brownlee, M.; Nawroth, P. P. Glyoxalase-1 prevents mitochondrial
protein modification and enhances lifespan in Caenorhabditis elegans. Aging Cell
7:260–269; 2008.
The findings in Figs. 9 and 10 show that the formation of MQE
and QE from MitoG increased dramatically under conditions of

450
P.B.L. Pun et al. / Free Radical Biology and Medicine 67 (2014) 437–450
[
16] Green, K.; Brand, M. D.; Murphy, M. P. Prevention of mitochondrial oxidative
[36] Twibanire, J. D.; Grindley, T. B. Efficient and controllably selective preparation
of esters using uronium-based coupling agents. Org. Lett. 13:2988–2991; 2011.
[37] Kelso, G. F.; Porteous, C. M.; Coulter, C. V.; Hughes, G.; Porteous, W. K.;
Ledgerwood, E. C.; Smith, R. A. J.; Murphy, M. P. Selective targeting of a redox-
active ubiquinone to mitochondria within cells—antioxidant and antiapoptotic
properties. J. Biol. Chem. 276:4588–4596; 2001.
[38] Asin-Cayuela, J.; Manas, A. R.; James, A. M.; Smith, R. A. J.; Murphy, M. P. Fine-
tuning the hydrophobicity of a mitochondria-targeted antioxidant. FEBS Lett.
571:9–16; 2004.
damage as a therapeutic strategy in diabetes. Diabetes 53(Suppl. 1):S110–118;
2004.
[
[
[
17] Yoon, Y.; Galloway, C. A.; Jhun, B. S.; Yu, T. Mitochondrial dynamics in diabetes.
Antioxid. Redox Signaling 14:439–457; 2011.
18] Newsholme, P.; Gaudel, C.; Krause, M. Mitochondria and diabetes: an intri-
guing pathogenetic role. Adv. Exp. Med. Biol. 942:235–247; 2012.
19] Baynes, J. W.; Thorpe, S. R. Role of oxidative stress in diabetic complications—a
new perspective on an old paradigm. Diabetes 48:1–9; 1999.
[20] Rosca, M. G.; Mustata, T. G.; Kinter, M. T.; Ozdemir, A. M.; Kern, T. S.; Szweda, L. I.;
[39] Choi, S. W.; Gerencser, A. A.; Nicholls, D. G. Bioenergetic analysis of isolated
Brownlee, M.; Monnier, V. M.; Weiss, M. F. Glycation of mitochondrial proteins
from diabetic rat kidney is associated with excess superoxide formation. Am. J.
Physiol. 289:F420–F430; 2005.
cerebrocortical nerve terminals on a microgram scale: spare respiratory
capacity and stochastic mitochondrial failure. J. Neurochem. 109:1179–1191;
2009.
[
40] Dranka, B. P.; Benavides, G. A.; Diers, A. R.; Giordano, S.; Zelickson, B. R.; Reily, C.;
Zou, L.; Chatham, J. C.; Hill, B. G.; Zhang, J.; Landar, A.; Darley-Usmar, V. M.
Assessing bioenergetic function in response to oxidative stress by metabolic
profiling. Free Radic. Biol. Med. 51:1621–1635; 2011.
[
[
[
[
21] Ceriello, A.; Ihnat, M. A.; Thorpe, J. E. The "metabolic memory": is more than
just tight glucose control necessary to prevent diabetic complications? J. Clin.
Endocrinol. Metab. 94:410–415; 2009.
22] Ray, S.; Dutta, S.; Halder, J.; Ray, M. Inhibition of electron flow-through
complex I of the mitochondrial respiratory-chain of Ehrlich ascites-carcinoma
cells by methylglyoxal. Biochem. J 303:69–72; 1994.
23] Biswas, S.; Ray, M.; Misra, S.; Dutta, D. P.; Ray, S. Selective inhibition of
mitochondrial respiration and glycolysis in human leukaemic leucocytes by
methylglyoxal. Biochem. J. 323(Pt 2):343–348; 1997.
24] Rosca, M. G.; Monnier, V. M.; Szweda, L. I.; Weiss, M. F. Alterations in renal
mitochondrial respiration in response to the reactive oxoaldehyde methyl-
glyoxal. Am. J. Physiol. 283:F52–59; 2002.
[
41] Hill, B. G.; Benavides, G. A.; Lancaster Jr J. R.; Ballinger, S.; Dell'Italia, L.;
Jianhua, Z.; Darley-Usmar, V. M. Integration of cellular bioenergetics with
mitochondrial quality control and autophagy. Biol. Chem. 393:1485–1512;
2012.
[
[
42] Brand, M. D.; Nicholls, D. G. Assessing mitochondrial dysfunction in cells.
Biochem. J. 435:297–312; 2011.
43] Yoshioka, M.; Kayo, T.; Ikeda, T.; Koizumi, A. A novel locus, Mody4, distal to
D7Mit189 on chromosome 7 determines early-onset NIDDM in nonobese
C57BL/6 (Akita) mutant mice. Diabetes 46:887–894; 1997.
[
[
[
25] Murphy, M. P. Development of lipophilic cations as therapies for disorders due
to mitochondrial dysfunction. Expert Opin. Biol. Ther 1:753–764; 2001.
26] Smith, R. A. J.; Hartley, R. C.; Cocheme, H. M.; Murphy, M. P. Mitochondrial
pharmacology. Trends Pharmacol. Sci. 33:341–352; 2012.
[
[
44] Izumi, T.; Yokota-Hashimoto, H.; Zhao, S.; Wang, J.; Halban, P. A.; Takeuchi, T.
Dominant negative pathogenesis by mutant proinsulin in the Akita diabetic
mouse. Diabetes 52:409–416; 2003.
45] Chacko, B. K.; Reily, C.; Srivastava, A.; Johnson, M. S.; Ye, Y.; Ulasova, E.; Agarwal,
27] Smith, R. A. J.; Porteous, C. M.; Gane, A. M.; Murphy, M. P. Delivery of bioactive
A.; Zinn, K. R.; Murphy, M. P.; Kalyanaraman, B.; Darley-Usmar, V. Prevention of
molecules to mitochondria in vivo. Proc. Natl. Acad. Sci. USA 100:5407–5412;
diabetic nephropathy in Ins2
(þ/)(AkitaJ)
mice by the mitochondria-targeted
2003.
therapy MitoQ. Biochem. J. 432:9–19; 2010.
[
[
28] Chaplen, F. W.; Fahl, W. E.; Cameron, D. C. Method for determination of free
intracellular and extracellular methylglyoxal in animal cells grown in culture.
Anal. Biochem. 238:171–178; 1996.
29] Yamaguchi, M.; Hara, S.; Nakamura, M. Determination of methylglyoxal in
mouse blood by liquid chromatography with fluorescence detection. Anal.
Chim. Acta 221:163–166; 1989.
[
[
[
[
46] Murata-Kamiya, N.; Kamiya, H. Methylglyoxal, an endogenous aldehyde,
crosslinks DNA polymerase and the substrate DNA. Nucleic Acids Res.
29:3433–3438; 2001.
47] Murata-Kamiya, N.; Kamiya, H.; Kaji, H.; Kasai, H. Mutations induced by
glyoxal and methylglyoxal in mammalian cells. Nucleic Acids Symp. Ser.
4
4:3–4; 2000.
[30] Cochemé, H. M.; Logan, A.; Prime, T. A.; Abakumova, I.; Quin, C.; McQuaker, S. J.;
48] Pampati, P. K.; Suravajjala, S.; Dain, J. A. Monitoring nonenzymatic glycation of
human immunoglobulin G by methylglyoxal and glyoxal: a spectroscopic
study. Anal. Biochem. 408:59–63; 2011.
Patel, J. V.; Fearnley, I. M.; James, A. M.; Porteous, C. M.; Smith, R. A. J.; Hartley, R. C.;
Partridge, L.; Murphy, M. P. Using the mitochondria-targeted ratiometric mass
2 2
spectrometry probe MitoB to measure H O in living Drosophila. Nat. Protoc.
49] Fedoronko, M.; Konigstein, J.; Linek, K. Determination of dl-glyceraldehyde,
7:946–958; 2012.
dihydroxyacetone and methylglyoxal in
a mixture. J. Electroanal. Chem.
[
31] Cochemé, H. M.; Quin, C.; McQuaker, S. J.; Cabreiro, F.; Logan, A.; Prime, T. A.;
Abakumova, I.; Patel, J. V.; Fearnley, I. M.; James, A. M.; Porteous, C. M.;
Smith, R. A. J.; Saeed, S.; Carre, J. E.; Singer, M.; Gems, D.; Hartley, R. C.;
Interfacial Electrochem. 14:357–367; 1967.
[50] Chouchani, E. T.; Methner, C.; Nadtochiy, S. M.; Logan, A.; Pell, V. R.; Ding, S.;
James, A. M.; Cochemé, H. M.; Reinhold, J.; Lilley, K. S.; Partridge, L.; Fearnley, I. M.;
Robinson, A. J.; Hartley, R. C.; Smith, R. A. J.; Krieg, T.; Brookes, P. S.; Murphy, M. P.
Cardioprotection by S-nitrosation of a cysteine switch on mitochondrial complex I.
Nat. Med. 19:753–759; 2013.
2 2
Partridge, L.; Murphy, M. P. Measurement of H O within living Drosophila
during aging using a ratiometric mass spectrometry probe targeted to the
mitochondrial matrix. Cell Metab. 13:340–350; 2011.
[
32] Carrigan, C. N.; Bartlett, R. D.; Esslinger, C. S.; Cybulski, K. A.; Tongcharoensir-
ikul, P.; Bridges, R. J.; Thompson, C. M. Synthesis and in vitro pharmacology of
substituted quinoline-2,4-dicarboxylic acids as inhibitors of vesicular gluta-
mate transport. J. Med. Chem. 45:2260–2276; 2002.
[
51] Logan, A.; Cochemé, H. M.; Boon Li Pun, P.; Apostolova, N.; Smith, R. A. J.;
Larsen, L.; Larsen, D. S.; James, A. M.; Fearnley, I. M.; Rogatti, S.; Prime, T. A.;
Finichiu, P.; Dare, A.; Chouchani, E. T.; Pell, V. R.; Methner, C.; Quin, C.;
McQuaker, S. J.; Krieg, T.; Hartley, R. C.; Murphy, M. P. Biochim. Biophys. Acta
1840:923–930; 2014.
[
[
33] Fanta, P. E.; Tarbell, D. S. 2-Nitro-4-methoxyaniline. Org. Synth. 25:78–80;
1
945.
34] Barton, J. K.; Shao, F.; Elias, B.; Lu, W. Synthesis and characterization of iridium
III) cyclometalated complexes with oligonucleotides: insights into redox
[52] Porteous, C. M.; Logan, A.; Evans, C.; Ledgerwood, E. C.; Menon, D. K.;
Aigbirhio, F.; Smith, R. A.; Murphy, M. P. Rapid uptake of lipophilic triphenyl-
phosphonium cations by mitochondria in vivo following intravenous injec-
tion: implications for mitochondria-specific therapies and probes. Biochim.
Biophys. Acta 1800:1009–1017; 2010.
[53] Li, Y.; Zhang, H.; Fawcett, J. P.; Tucker, I. G. Effect of cyclosporin A on the
pharmacokinetics of mitoquinone (MitoQ10), a mitochondria-targeted anti-
oxidant, in rat. Asian J. Pharm. Sci. 5:106–113; 2010.
(
reactions with DNA. Inorg. Chem. 46:10187–10199; 2007.
[
35] Thornalley, P. J.; Edwards, L. G.; Kang, Y.; Wyatt, C.; Davies, N.; Ladan, M. J.;
Double, J. Antitumour activity of S-p-bromobenzylglutathione cyclopentyl
diester in vitro and in vivo: inhibition of glyoxalase I and induction of
apoptosis. Biochem. Pharmacol. 51:1365–1372; 1996.