The FASEB Journal • Research Communication
Comparing the effects of mitochondrial targeted
and localized antioxidants with cellular antioxidants
in human skin cells exposed to UVA and
hydrogen peroxide
Anne O. Oyewole,* Marie-Claire Wilmot,* Mark Fowler,† and Mark A. Birch-Machin*,1
*Department of Dermatological Sciences, Institute of Cellular Medicine, Medical School, Newcastle
University, Newcastle Upon Tyne, UK; and †Unilever Research and Development Colworth,
Sharnbrook, UK
ABSTRACT Skin cancer and aging are linked to
increased cellular reactive oxygen species (ROS), par-
ticularly following exposure to ultraviolet A (UVA) in
sunlight. As mitochondria are the main source of
cellular ROS, this study compared the protective ef-
fects of mitochondria-targeted and -localized antioxi-
dants (MitoQ and tiron, respectively) with cellular
antioxidants against oxidative stress-induced [UVA
and hydrogen peroxide (H2O2)] mitochondrial DNA
(mtDNA) damage in human dermal fibroblasts. With
the use of a long quantitative PCR assay, tiron (EC50 10
mM) was found to confer complete (100%) protection
(P<0.001) against both UVA- and H2O2-induced
mtDNA damage, whereas MitoQ (EC50 750 nM) pro-
vided less protection (17 and 32%, respectively;
P<0.05). This particular protective effect of tiron was
greater than a range of cellular antioxidants investi-
gated. The nuclear factor erythroid 2-related factor 2
(Nrf2) signaling pathway provides cellular protection
against oxidative stress. An ELISA assay for the Nrf2
target gene heme oxygenase-1 (HO-1) and studies using
Nrf2 small interfering RNA both indicated that tiron’s
mode of action was Nrf2 independent. The comet assay
showed that tiron’s protective effect against H2O2-
induced nuclear DNA damage was greater than the
cellular antioxidants and MitoQ (P<0.001). This study
provides a platform to investigate molecules with sim-
ilar structure to tiron as potent and clinically relevant
antioxidants.—Oyewole, A. O., Wilmot, M.-C., Fowler,
M., Birch-Machin, M. A. Comparing the effects of
mitochondrial targeted and localized antioxidants
with cellular antioxidants in human skin cells ex-
posed to UVA and hydrogen peroxide. FASEB J. 28,
485–494 (2014). www.fasebj.org
Key Words: DNA damage � mitochondria � ultraviolet radia-
tion � reactive oxygen species � ROS � Nrf2
Exposure to solar ultraviolet radiation (UVR)
plays a key role in the development of skin cancers and
aging (1). There is evidence that ultraviolet A (UVA)
may have a causative role in skin cancer (2, 3), but its
mechanism of action is thought to be different from
that of higher energy UVB occurring via an indirect
mechanism involving reactive oxygen species (ROS),
including hydrogen peroxide (H2O2; ref. 4). This oxi-
dative stress can subsequently exert downstream effects,
including lipid peroxidation and generation of DNA
damage.
As electrons leak from the electron transport chain
(ETC), mitochondria are the major cellular generator
of superoxide (1, 5), which can be readily converted
into other ROS. Mitochondrial DNA (mtDNA) is close
to the site of superoxide production, making it highly
vulnerable to oxidative damage. The accumulation of
mtDNA mutations, through oxidative stress mecha-
nisms, has been proposed as an underlying cause of the
carcinogenic and aging process in many tissues, includ-
ing skin (1, 4, 5). We and others have shown that
mtDNA damage accumulates in skin due to UVR irra-
diation, photoaging, and skin cancer (6–10), thereby
pioneering the use of mtDNA damage as a biomarker
of UV exposure (7, 10, 11).
1 Correspondence: Dermatological Sciences, Institute of
Cellular Medicine, Medical School, Framlington Pl., New-
castle University, Newcastle upon Tyne, NE2 4HH, UK.
E-mail: m.a.birch-machin@ncl.ac.uk
doi: 10.1096/fj.13-237008
Abbreviations: EC50, half-maximal effective concentration;
ETC, electron transport chain; HDFn, human dermal fibroblast
neonatal; HO-1, heme oxygenase 1; Keap1, kelch-like ECH-
associated protein 1; mtDNA, mitochondrial DNA; MTS, 3-(4,5-
dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-
sulfophenyl)-2H-tetrazolium, inner salt; NAC, N-acetyl cysteine;
nDNA, nuclear DNA; NF104, neonatal fibroblast 104; Nrf2,
nuclear factor erythroid 2-related factor 2; qPCR, quantitative
real-time PCR; RNAi, RNA interference; ROS, reactive oxygen
species; SED, standard erythemal dose; SF, sulforaphane; tiron,
4,5-dihydroxy-1,3-benzenedisulfonic; siRNA, short interfering
RNA; UVA, ultraviolet A; UVB, ultraviolet B; UVR, ultraviolet
radiation
4850892-6638/14/0028-0485 © FASEB
Antioxidants have been shown to provide protection
to the skin against oxidative stress. For example, caro-
tenoids protect against UVR-induced erythema, possibly by
acting as a ROS scavenger (12), and with collaborators
we have shown that UVR-induced mtDNA damage in
human skin is protected by a diet rich in tomato paste
(high in the antioxidant lycopene; ref. 13). However,
the ability of antioxidants to abrogate oxidative stress in
skin and indeed other tissues has been the subject of
much discussion in the face of little comparative data
for the different antioxidants and their targets, partic-
ularly at the level of mitochondria (4). As mitochondria
are a major source of ROS, this has led to the develop-
ment of several antioxidant compounds with the
ubiquinone structure and designed to target the mito-
chondria. A widely used compound is MitoQ, a ubiqui-
none derivative conjugated to triphenylphosphonium
that enables this molecule to enter the mitochondria
(14–16). An unrelated compound, 4,5-dihydroxy-1,3-ben-
zenedisulfonic (tiron), which chelates iron and other
metals (17) and exhibits ROS scavenging properties (17–
20), is also able to permeabilize the mitochondrial mem-
brane and thereby be localized rather than targeted to the
mitochondrion from the cytoplasm (18, 20).
The comparative effects of a mitochondria-targeted
antioxidant, MitoQ; an antioxidant preferentially local-
ized to the mitochondria, tiron; and cellular antioxi-
dants have not been previously investigated, particu-
larly in relation to protection against oxidative stress-
induced DNA (mitochondrial and nuclear) damage
under the same experimental conditions. Therefore, in
a novel approach, we have compared the protective
effects of mitochondria-targeted and localized antioxi-
dants (MitoQ and tiron, respectively) with each other
and with cellular antioxidants [including N-acetyl cys-
teine (NAC), resveratrol, curcumin, and sulforaphane
(SF)] against oxidative stress (UVA and H2O2)-induced
mtDNA damage in human dermal fibroblasts utilizing a
long quantitative PCR (qPCR) assay to amplify the
majority of the genome. We have investigated the wider
effects of these compounds through comparison of
their protective effects against H2O2-induced nuclear
DNA (nDNA) damage using the comet assay. The
nuclear factor erythroid 2-related factor 2 (NrF2) sig-
naling pathway significantly contributes to cellular
protection against oxidative stress in the skin (21–
24), and it may provide an “intramitochondrial anti-
oxidant network” (25, 26). Therefore, experiments
were performed [including an ELISA assay for the Nrf2
target gene heme oxygenase 1 (HO-1) and studies using
Nrf2 short interfering RNA (siRNA)] to determine
whether the mode of action of tiron (the most potent
antioxidant observed in our study) was Nrf2 pathway
dependent. Prevention of excessive ROS in a physiolog-
ical manner based on medicinal chemical studies of the
tiron structure may provide a tractable therapeutic
platform beyond skin to other tissues where the process
of aging and disease is associated with increased levels
of oxidative stress.
MATERIALS AND METHODS
Cell viability and half-maximal effective concentration
(EC50) determination
The 3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-
(4-sulfophenyl)-2H-tetrazolium, inner salt (MTS) assay (Pro-
mega, Southampton, UK) measures the ability of viable cells to
metabolize the tetrozolium salt MTT by succinate dehydroge-
nase. The absorbance of nontreated (control) cells on each
96-well plate was set as reference of 100% cell viability, and the
viabilities of cultured human dermal fibroblast cells treated in
the presence of various compounds (after 24 h in culture
medium) were calculated relative to this value. All compounds
were tested in at least 3 independent experiments, and for each
experiment the results for all concentrations assayed are the
means of 8 replicates. The EC50 values (concentration inducing
50% of cell death) were calculated using dose-response curves
from �3 independent experiments. Statistical significance was
performed using a 1-way ANOVA with Dunnett’s correction,
where P � 0.05 was considered significantly different from
control (untreated).
UVA/H2O2 treatment
Human dermal fibroblast neonatal (HDFn) cells were grown
in DMEM containing 10% fetal bovine serum, 5 IU/ml
penicillin, and 5 g/ml streptomycin. The cells were treated in
advance with the optimum, sublethal dose (determined by
the MTS assay of the antioxidant for 24 h in phenol red-free
DMEM) and washed before treatment with 200 �M H2O2
(Sigma, UK) or irradiated with a sublethal dose [9.2 J/cm2,
equivalent to 0.5 standard erythemal dose (SED)] of UVR
using a TL-09 lamp (315– 400 nM, peak emission at 360
nM; Philips, Amsterdam, The Netherlands). Exposure to
this physiologically relevant dose of UVA equated to �45
min exposure time. Total cellular DNA was extracted using
the Qiagen DNeasy tissue extraction kit (Qiagen, Man-
chester, UK).
qPCR
MtDNA damage was measured as amplification efficiency for
a large (11,095 bp) amplicon normalized to a short (83 bp)
amplicon as described previously (27) and performed using
the real-time PCR assay as described previously (28). At least
3 independent PCR analyses on �3 biological replicates were
performed.
DNA primer sequences for the 11,059-bp amplicon were as
follows: D1B (282-255), 5=-ATGATGTCTGTGTGGAAAGTG-
GCTGTGC-3=; and OLA (5756–5781), 5=-GGGAGAAGC-
CCCGGCAGGTTTGAAGC-3=. DNA primer sequences for the
83-bp amplicon were as follows: IS1 (16,042-16,066), 5=-
GATTTGGGTACCACCCAAGTATTG-3=; and IS2 (16,101-
16,124), 5=-AATATTCATGGTGGCTGGCAGTA-3=.
Quantification of the Nrf2 target HO-1 using ELISA
Fibroblast cells (3�104/well) were seeded and treated with
the optimum, sublethal dose of each antioxidant for 24 h.
This experimental work was conducted at Unilever Colworth,
where the standard operating protocol for the HO-1 ELISA
assay used human neonatal fibroblast 104 (NF104) cells.
These cells can be considered as interchangeable with the
HDFn cells used throughout the rest of the study, as both cells
are human neonatal fibroblast cells isolated from foreskin. In
addition, previous proof-of-concept studies between New-
486 Vol. 28 January 2014 OYEWOLE ET AL.The FASEB Journal � www.fasebj.org
castle University and Unilever have shown that both cell types
behave in the same way in the oxidative stress-induced
mtDNA strand-break assays. Cell lysates from the NF104 cells
were analyzed for HO-1 content using the DuoSet IC ELISA
kit (R&D Systems, Abingdon, UK) and a microplate reader at
450 nM [results expressed as picograms per microgram
protein using the Pierce bicinchoninic acid (BCA) assay;
Thermo Fisher Scientific, Waltham, MA, USA).
Comet assay
The comet assay was performed as described previously by
Atherton et al. (29). Nucleoids were selected (100/slide), and
the tail length (measure of DNA migration from head of
comet) for each treatment was measured and normalized
against the control (untreated). According to the established
conditions in our laboratory and reports from collaborators,
a lower concentration (i.e., 10 �M) of H2O2 was required in
the comet assay compared with the concentration (i.e., 200
�M) used in the long qPCR assay to detect mtDNA damage.
As a result, it was observed that the maximal concentration of
antioxidant required to provide the greatest protective effect
against H2O2-induced nDNA damage was much lower than
that required in the long qPCR mtDNA damage assay (with
the exception of tiron).
siRNA and Nrf2 studies
At 24 h after transfection (Lipofectamine RNAiMAX; Invitro-
gen, Paisley, UK) with siRNA (20 nM) against Nrf2 (or a
negative control using On-TargetPlus ontargeting pool;
Thermo Scientific), cells were treated with medium contain-
ing the compound for 24 h before UVA irradiation. Nrf2
mRNA levels were determined 48 h post-transcription follow-
ing cDNA synthesis (Applied Biosystems, Warrington, UK)
from isolated RNA (RNeasy; Qiagen) and quantification of
the Nrf2 mRNA transcripts (TaqMan Gene Expression Assay;
Applied Biosystems). Primer/probe sets for human Nrf2 and
�-actin (endogenous control) were from Applied Biosystems.
Nrf2 results were normalized to �-actin using the ��Ct
method (30, 31).
Statistical analyses
ANOVAs with correction for multiple groups were performed
using commercially available software (GraphPad Prism 5;
GraphPad, San Diego, CA, USA).
RESULTS
Tiron confers greater protection than MitoQ against
both UVA- and H2O2-induced mtDNA damage in
dermal fibroblasts
The first aim of the study was to determine and
compare the relative effects of MitoQ and tiron in
terms of providing protection against both UVA- and
H2O2-induced mtDNA damage in cultured HDFn cells.
First, it was crucial to derive a sublethal experimental
concentration of tiron and MitoQ by performing an
extensive range of cell viability dose-response experi-
ments (MTS assay). The derived maximum concentra-
tion of MitoQ and tiron without inducing significant
(not �5%) death in the HDFn cells was 150 and 3 mM,
respectively (1-way ANOVA with Dunnett’s correction;
treatments were not significantly different when com-
pared with untreated control, P�0.001; n�3). These
concentrations were then selected for the subsequent
experiments. The EC50 for cell viability for MitoQ and
tiron in the HDFn cells was derived from the experi-
mental series as 750 and 10 mM, respectively. There-
fore, the derived sublethal concentrations of 150 nM
MitoQ and 3 mM tiron are not only well below the EC50
values but do not cause any significantly observable cell
death (i.e., �5%).
The protective effect of these sublethal concentra-
tions of MitoQ (150 nM) and tiron (3 mM) against
UVA-induced mtDNA damage was determined in
HDFn cells utilizing the previously validated real-time
long qPCR assay (28, 32). The long qPCR assay mea-
sures DNA lesions in the majority (�70%) of the
mitochondrial genome. The cells were exposed to a
physiological dose of UVA (0.5 SED) that we have
previously shown to be well tolerated by the cultured
skin cells (28). First, the results in Fig. 1 show that
exposure to physiologically relevant levels of UVA re-
sulted in a statistically significant increase in the levels
of mtDNA damage (P�0.001; Fig. 1). This is expressed
as an increase in Ct value due to the decrease in the
relevant proportions of undamaged:damaged circular
mtDNA molecules as the UVA-induced DNA damage
blocks the progression of the DNA polymerase. The
data are expressed as a fold difference in C(t), with
each treatment corrected for fold difference in relation
to the value obtained from the nonirradiated control
sample (which is normalized as 1). For example, the
level of UVA-induced damage in Fig. 1B is calculated as
a fold increase of 1.72, which approximates to a 128-
fold increase in mtDNA damage. The second and most
important observation in Fig. 1 is that tiron appears to
provide a clear and complete (100%) protection
against the UVA-induced mtDNA damage. This obser-
vation is in sharp contrast to MitoQ, which provided
only a 17% protective effect.
UVA has been postulated as a stimulator of oxidative
stress in skin (1), and the data from Fig. 1 suggest,
therefore, that tiron may be affecting this pathway. To
address this issue, the experimental protocol repre-
sented in Fig. 1 was repeated with the only exception of
replacing UVA with H2O2 (200 �M as used in our
previous study; ref. 28) as a direct oxidative inducer of
mtDNA damage in the HDFn cells (Fig. 2). There are 2
key observations from the data presented in Fig. 2. First,
in a similar fashion to the results observed in Fig. 1
(UVA induction), the data in Fig. 2 show that 200 �M
H2O2 induced a statistically significant induction in
mtDNA damage, as represented by an increase in Ct
value. Second, and more important, tiron appears to
provide a clear and complete (100%) protection
against the H2O2-induced mtDNA damage (P�0.001).
Once again, this observation is in sharp contrast to the
results in Fig. 2B, where MitoQ provided a 32% protec-
tive effect (P�0.05).
487EFFECT OF MITOCHONDRIAL AND CELLULAR ANTIOXIDANTS
Protective ability of a range of cellular antioxidants
against UVA and H2O2-induced mtDNA damage is
lower than that exhibited by tiron
The next question to be addressed is how this greater
protection of mtDNA damage exhibited by tiron com-
pares to a range of cellular antioxidants that do not
penetrate the mitochondrial membrane with the same
experimental protocol used in Figs. 1 and 2. In the
same manner as the experiment described in Fig. 1A,
sublethal experimental concentration for each cellular
A
Co
ntr
ol
Tir
on
Tir
on
+
UV
A
UV
A
0.0
0.5
1.0
1.5
2.0
***
Fo
ld
D
iff
er
en
ce
C
(t)
Co
ntr
ol
Mi
toQ
Mi
toQ
+
UV
A
UV
A
0.0
0.5
1.0
1.5
2.0
*
Fo
ld
D
iff
er
en
ce
C
(t)
B
Figure 1. Protective effect of MitoQ and tiron against UVA-induced mtDNA damage in HDFn cells. Effects of tiron (A) and
MitoQ (B) against UVA-induced mtDNA damage in HDFn skin cells using the lesion-specific real-time PCR assay. Statistical
significance was assessed by performing a 1-way ANOVA with Newman-Keuls multiple comparison test. Bars represent means
sem (n�3). A) Tiron (3 mM) appears to completely prevent UVA-induced mtDNA damage (i.e., 100% protection). ***P� 0.001
vs. UVA treatment alone. B) In contrast MitoQ (150 nM) was able to offer less (17%) protection against UVA-induced mtDNA
damage. *P � 0.05 vs. UVA treatment alone.
A
Co
ntr
ol
Tir
on
Tir
on
+
H 2
O 2
0.0
0.5
1.0
1.5
***
Fo
ld
D
iff
er
en
ce
C
(t)
0.0
0.5
1.0
1.5
*
Fo
ld
D
iff
er
en
ce
C
(t)
H 2
O 2
Co
ntr
ol
Mi
toQ
Mi
toQ
+
H 2
O 2
H 2
O 2
B
Figure 2. Protective effect of MitoQ and tiron against H2O2-induced mtDNA damage in HDFn cells. Effects of tiron (A) and
MitoQ (B) against H2O2-induced mtDNA damage in HDFn skin cells using the lesion-specific real-time PCR assay. Statistical
significance was assessed by a 1-way ANOVA with Newman-Keuls multiple comparison test. Bars represent means sem (n�3).
A) Tiron (3 mM) appears to completely (i.e., 100%) prevent H2O2-induced mtDNA damage. ***P � 0.001 vs. H2O2 treatment
alone. B) In contrast, MitoQ (150 nM) was able to offer less (32%) protection against H2O2-induced mtDNA damage. *P� 0.05
vs. H2O2 treatment alone.
488 Vol. 28 January 2014 OYEWOLE ET AL.The FASEB Journal � www.fasebj.org
antioxidant was determined by performing an exten-
sive range of cell viability dose-response experiments
(MTS assay). The derived maximum concentration
(without inducing �5% cell death) was 100 �M for
resveratrol, 8 mM for NAC, and 8 �M for curcumin
(1-way ANOVA with Dunnett’s correction; treatments
were not significantly different when compared with
untreated control, P�0.001; n�3). The protective ef-
fect of these concentrations of antioxidants against
UVA and H2O2-induced mtDNA damage was deter-
mined in HDFn cells using the long qPCR strand-break
assay (Table 1). As the aim of the experiment was to
compare the protective effect of tiron with that of
known cytosolic antioxidants against UVA- and H2O2-
induced mtDNA damage, the degree of protection
offered by each compound was expressed as a percent-
age of the protection offered by tiron, which appeared
to be complete (