Content uploaded by Jean Cadet
Author content
All content in this area was uploaded by Jean Cadet on Dec 23, 2013
Content may be subject to copyright.
Cyclobutane pyrimidine dimers are predominant
DNA lesions in whole human skin exposed
to UVA radiation
Ste
´
phane Mouret*, Caroline Baudouin
†
, Marie Charveron
†
, Alain Favier*, Jean Cadet*, and Thierry Douki*
‡
*Commissariat a` l’Energie Atomique (CEA)兾La Direction des Sciences de la Matie` re (DSM)兾De´ partement de Recherche Fondamentale sur la Matie`re
Condense´ e, Service de Chimie Inorganique et Biologique UMR-E 3 (CEA-UJF), CEA-Grenoble, Laboratoire ‘‘Le´ sions des Acides Nucle´ iques,’’
38054 Grenoble Cedex 9, France; and
†
Laboratoire de Biologie Cellulaire, Institut de Recherche Pierre Fabre, Hoˆ tel Dieu Saint Jean,
2 rue Viguerie, 31025 Toulouse Cedex 3, France
Edited by James E. Cleaver, University of California, San Francisco, CA, and approved July 5, 2006 (received for review May 22, 2006)
Solar UV radiation is the most important environmental factor
involved in the pathogenesis of skin cancers. The well known
genotoxic properties of UVB radiation (290 –320 nm) mostly in-
volve bipyrimidine DNA photoproducts. In contrast, the contribu-
tion of more-abundant UVA radiation (320–400 nm) that are not
directly absorbed by DNA remains poorly understood in skin. Using
a highly accurate and quantitative assay based on HPLC coupled
with tandem mass spectrometry, we determined the type and the
yield of formation of DNA damage in whole human skin exposed
to UVB or UVA. Cyclobutane pyrimidine dimers, a typical UVB-
induced DNA damage, were found to be produced in significant
yield also in whole human skin exposed to UVA through a mech-
anism different from that triggered by UVB. Moreover, the latter
class of photoproducts is produced in a larger amount than 8-oxo-
7,8-dihydro-2ⴕ-deoxyguanosine, the most common oxidatively
generated lesion, in human skin. Strikingly, the rate of removal of
UVA-generated cyclobutane pyrimidine dimers was lower than
those produced by UVB irradiation of skin. Finally, we compared
the formation yields of DNA damage in whole skin with those
determined in primary cultures of keratinocytes isolated from the
same donors. We thus showed that human skin efficiently protects
against UVB-induced DNA lesions, whereas very weak protection
is afforded against UVA. These observations emphasize the likely
role played by the UVA-induced DNA damage in skin carcinogen-
esis and should have consequences for photoprotection strategies.
carcinogenesis 兩 DNA damage 兩 mutagenesis 兩 oxidative stress 兩
DNA repair
O
ccurrence of skin cancers, which mostly arise from exposure
to solar UV radiation, has constantly increased in the recent
years due to changes in life habits. Harmful effects of UV radiation
are mostly associated with both direct and indirect photoinduced
damage to DNA (1) that can lead to the induction of mutations. The
chemical nature and the formation efficiency of the DNA lesions
greatly depend on the wavelength of the incident photons. UVB
radiation (290–320 nm), the most energetic mutagenic and carci-
nogenic component of solar radiation, is directly absorbed by DNA,
giving rise to dimeric photoproducts between adjacent pyrimidine
bases. Two types of these bulky modifications are produced, namely
cyclobutane pyrimidine dimers (CPDs) and pyrimidine (6-4)
pyrimidone photoproducts (1). Evidence for the involvement of
these lesions in photocarcinogenesis is provided by the high pro-
portion of p53 mutations (TC to TT or CC to TT transitions)
detected at bipyrimidine site s in skin tumors (2–4).
UVA radiation (320–400 nm), the other component of terres-
trial UV radiation, is mutagenic in cultured cells (5, 6) and induces
skin tumors in mice (7, 8). In addition, UVA has been shown to be
involved in immunosuppression (9) and is suspected to play a major
role in the induction of melanoma (10, 11), the most severe type of
skin cancers. Consequently, the carcinogenic properties of UVA
have become a matter of concern. The incorrect use of sunscreens
that until recently afforded mostly UVB protection and allow
longer exposure periods to sunlight (12), together with the use of
artificial tanning equipments (13), lead to an increase in the overall
amount of UVA received by the population of wealthie st countries.
A better asse ssment of the role of UVA has regained attention
with recent mutagenesis studies in skin tumors (14). In contrast with
UVB, the lesions re sponsible for the UVA-induced mutations are
not clearly identified. UVA radiation is extremely poorly absorbed
by DNA and its genotoxic effects have been explained mostly by the
induction of oxidative stre ss. The formation of 8-oxo-7,8-dihydro-
2⬘-deoxyguanosine (8-oxodGuo) in much larger amount than
strand breaks and oxidized pyrimidine bases (15, 16) strongly
suggests a major role played by singlet oxygen (17). However, the
UVA mutation spectrum in mammalian cells doe s not exhibit a
predominance of G:C to T:A transversions (18, 19) that is consid-
ered as the mutagenic hallmark of 8-oxodGuo. This observation,
together with the lack of increase in mutation rate in cells deficient
in repair of 8-oxodGuo (20), sugge sts that other lesions are involved
in UVA mutagenesis.
Interestingly, data have been gathered on the induction of
cyclobutane pyrimidine dimers in DNA upon exposure of bacteria
(21), mammalian cells (22–24), and skin (25, 26) to UVA radiation.
As a striking trend, recent studies showed that the yield of CPD is
higher than that of 8-oxodGuo in rodent cell lines (15, 27, 28) and
in human skin cells (29) exposed to UVA. In contrast to UVB,
UVA radiation preferentially induce s the production of CPDs at
TT sites without any detectable formation of pyrimidine (6-4)
pyrimidone photoproducts (28–30). These observations clearly
indicate that UVA radiation doe s not induce CPDs via a direct
excitation pathway but more likely through a triplet energy transfer
photosensitization mechanism.
Until now, the predominant formation of CPDs over oxidatively
generated lesions has only been shown in cultured cells. The recent
observation that 8-oxodGuo is produced more efficiently than
CPDs in UVA-irradiated yeast (31) shows that the cell type and the
cellular context strongly influence the UVA photochemistry of
DNA. Therefore, the present study was designed to confirm in
whole human skin the trends observed in UVA-irradiated cultured
cutaneous cells. For this purpose, we accurately determined the
yield of formation of a series of relevant DNA lesions within whole
human skin exposed to UVA radiation. Results were compared
with those obtained upon UVB irradiation. Individual DNA lesions
were quantified by using the accurate HPLC analytical method
Conflict of interest statement: No conflicts declared.
This paper was submitted directly (Track II) to the PNAS office.
Abbreviations: CPD, cyclobutane pyrimidine dimmer; 8-oxodGuo: 8-oxo-7,8-dihydro-2⬘-
deoxyguanosine; MS兾MS, tandem mass spectrometry.
See Commentary on page 13567.
‡
To whom correspondence should be addressed. E-mail: thierry.douki@cea.fr.
© 2006 by The National Academy of Sciences of the USA
www.pnas.org兾cgi兾doi兾10.1073兾pnas.0604213103 PNAS
兩
September 12, 2006
兩
vol. 103
兩
no. 37
兩
13765–13770
GENETICS SEE COMMENTARY
associated either with tandem mass spectrometry (MS兾MS) detec-
tion for bipyrimidine photoproducts or electrochemical detection
for 8-oxodGuo. Using the same approach, information also was
gained on the repair kinetics of CPDs in skin.
Results
Formation of Bipyrimidine Photoproducts Within Human Skin Exposed
to UVB Radiation.
A first series of experiments was carried out to
quantify the yield of formation of the bipyrimidine photoproducts
within the skin of six donors immediately after exposure to UVB
radiation. As shown in Fig. 1A, the formation of photoproducts was
linear with respect to the applied UVB dose (0–0.2 J兾cm
2
). This
observation clearly indicate s that no secondary photoreaction oc-
curred under the experimental conditions used. The bipyrimidine
photoproducts were generated in the DNA of human skin upon
UVB irradiation in the following decreasing frequency: T⬍⬎T ⬎
T⬍⬎C ⬎ TC (6-4) ⬎ C⬍⬎T ⬎ C⬍⬎C ⬎ TT (6-4). The overall
yield of formation of photoproducts was calculated from the linear
regression of the dose-course plot for the sum of all bipyrimidine
photolesions and for each donor (Table 1). Thus, the mean yield of
formation was 518.8 lesions per 10
6
normal bases per J兾cm
2
within
the whole skin exposed to UVB radiation. For the six donors, the
difference between the highest (donor B) and the lowest (donor E)
individual yield of photoproducts was ⬍2-fold. It may be added that,
as previously shown in cell culture (32–34), none of the Dewar
valence isomers was detected, even at the highe st applied UVB
dose. The distribution of photoproducts in human skin was found
to be similar to the one obtained within primary cultures of
Fig. 2. Distribution of bipyrimidine photoproducts within human skin and
primary keratinocytes upon exposure to either UVB (A) or UVA (B) radiation.
The proportion (in a percentage) of each photoproduct was determined for
each donor, and results were represented by the average ⫾ SD.
Fig. 1. Formation of bipyrimidine photoproducts within human skin ex-
posed to UVB (A) or UVA (B) radiation. The presented data corresponds to one
representative donor. The results are expressed in lesions per 10
6
bases and are
the average ⫾ SD.
Table 1. Sum of the yield of formation of all bipyrimidine photoproducts
in irradiated human skin
Donors
UVA UVB
Skin
Cultured
keratinocytes Skin
Cultured
keratinocytes
A 0.065 ⫾ 0.017 0.121 ⫾ 0.024 431.0 ⫾ 104.1 12,934 ⫾ 278
B 0.113 ⫾ 0.018 0.129 ⫾ 0.004 656.8 ⫾ 142.0 11,304 ⫾ 124
C 0.077 ⫾ 0.008 0.117 ⫾ 0.013 546.7 ⫾ 71.8 11,139 ⫾ 620
D 0.076 ⫾ 0.008 0.136 ⫾ 0.006 511.4 ⫾ 30.4 12,464 ⫾ 318
E 0.080 ⫾ 0.008 0.106 ⫾ 0.010 425.1 ⫾ 89.5 7,505 ⫾ 215
F 0.076 ⫾ 0.013 0.143 ⫾ 0.007 542.2 ⫾ 76.4 12,409 ⫾ 253
Mean 0.081 ⫾ 0.017 0.125 ⫾ 0.014 518.8 ⫾ 78.4 11,293 ⫾ 1,811
Ratio keratinocytes兾skin 1.5 21.8
For each donor, the yield (expressed in lesions per 10
6
normal bases per Joules per centimeter squared) was
calculated from the linear regression for the sum of the level of all photoproducts with respect to the applied UV
dose. The reported value is the slope ⫾ SE. Mean represents the average ⫾ SD.
13766
兩
www.pnas.org兾cgi兾doi兾10.1073兾pnas.0604213103 Mouret et al.
keratinocyte isolated from the same donors (Fig. 2A). Emphasis
was placed on keratinocytes because other types of skin cells
(melanocytes and fibroblasts) are present at much a lower fre-
quency. The quantification of bipyrimidine photoproducts after
extraction of DNA from a whole biopsy thus mostly reflects damage
to keratinocytes of the different layers of the epidermis.
Formation of CPD in Skin Exposed to UVA Radiation. Then, the yield
of formation of bipyrimidine photoproducts was determined in the
whole human skin upon exposure to UVA radiation. The deter-
mined distribution of photoproducts was completely different from
that obtained after UVB irradiation. Indeed, exposure to UVA led
to the predominant formation of CPD at TT sites with much lower
amounts of the corresponding TC or CT cyclobutane dimers.
Neither C⬍⬎C, (6-4) photoproducts nor any of the Dewar valence
isomers were detected in the DNA of irradiated skin. The forma-
tion of the three types of CPDs was found to be linear with respect
to the applied UVA dose (0–200 J兾cm
2
) (Fig. 1B). For each donor,
the yield of formation was calculated from the linear regression of
the plot for the sum of all UVA-induced CPD (Table 1). The
difference between the highest and the lowest individual yield was
also ⬍2-fold, and the mean formation yield in human skin was 0.081
lesions per 10
6
normal bases per J兾cm
2
. Moreover, the distribution
of UVA-induced CPDs in the DNA of whole skin was strongly
similar to the one obtained within primary culture of keratinocytes
(Fig. 2B), where CPDs were produced in the following decreasing
order of abundance T⬍⬎T ⬎⬎ T⬍⬎C ⬎ C⬍⬎T, as reported for
other cell types (28, 29).
Larger Yield of T<>T than 8-oxodGuo in UVA-Irradiated Skin. In
another series of experiments, the induction of 8-oxodGuo, which
is oxidatively generated damage, was quantified within the whole
skin exposed to UVA radiation by using the HPLC-electrochemical
assay. UVA radiation induced a slight increase in the level of
8-oxodGuo in the DNA, even for the highest applied UVA dose of
200 J兾cm
2
. Although limited, the increase was found to be repro-
ducible from one donor to the other (Fig. 3). The induction of
T⬍⬎T also was determined by HPLC-MS兾MS assay in the same
samples. It was shown that T⬍⬎T was formed in a larger amount
than 8-oxodGuo (Fig. 3). The yield of formation for 8-oxodGuo was
0.0071 lesions per 10
6
normal bases per J兾cm
2
in human skin
exposed to UVA radiation, and the difference between the highest
and the lowest individual yields was ⬍4-fold (Table 2). The mean
yield of T⬍⬎T formation was 0.066 lesions per 10
6
normal bases
per J兾cm
2
in the DNA of UVA-irradiated whole skin (Table 2). The
ratio between the level of the two lesions shows that T⬍⬎Tare
induced with a 9-fold higher frequency than 8-oxodGuo in the
whole human skin exposed to UVA radiation.
For three donors of the series, the formation of 8-oxodGuo
also was quantified after UVB irradiation. Even at the highest
dose applied, namely 0.2 J兾cm
2
, only a very modest increase in
the level of 8-oxodGuo was observed (Fig. 4). In contrast, T⬍⬎T
was obtained in large amounts. Based on the sensitiv ity of the
assays, it can be estimated that the yield of CPDs is at least two
orders of magnitude higher than that of 8-oxodGuo in whole
human skin exposed to UVB.
Persistence of T<>T in UV-Irradiated Skin. Because T⬍⬎T was
found to be the major lesion produced af ter both UVA and UVB
irradiations in the DNA of human skin, the persistence of this
photoproduct was determined after an acute skin ex posure to
either UVA (100 J兾cm
2
) or UVB (0.1 J兾cm
2
) radiation. Inter-
estingly, T⬍⬎T remained present in large amount within the
DNA of human sk in 48 h after the end of the irradiations (Fig.
5). It may be noted that for the respective applied doses, UVA
produced ⬇2-fold less T⬍⬎T than UVB radiation. However, the
proportion of persistent T⬍⬎T 48 h after irradiation was
sign ificantly higher (P ⬍ 0.05) for UVA (72%) than for UVB
radiation (55%). These low repair rates are in agreement with
observations made by other groups on UV-irradiated human
sk in (35–37). The integ rity of the repair capacities in the skin
biopsies is shown by the efficient removal of TT (6-4) (Fig. 5).
Attenuation of UV-Induced Damage by Skin. Finally, we determined
the extent of protection provided by the skin against the formation
Fig. 3. Formation of 8-oxodGuo and thymine-thymine cyclobutane dimers
within human skin exposed to UVA radiation (200 J兾cm
2
). The results are
expressed in lesions per 10
6
bases and are the average ⫾ SD.
Table 2. Yield of formation of 8-oxodGuo and thymine
cyclobutane dimer (T<>T) within human skin exposed
to UVA radiation
Donors 8-oxodGuo T⬍⬎T
Ratio T⬍⬎T兾
8-oxodGuo
F 0.0085 ⫾ 0.0064 0.057 ⫾ 0.007 6.7
G 0.0080 ⫾ 0.0030 0.077 ⫾ 0.009 9.6
H 0.0087 ⫾ 0.0039 0.061 ⫾ 0.004 7.0
I 0.0066 ⫾ 0.0024 0.050 ⫾ 0.005 7.6
J 0.0081 ⫾ 0.0040 0.040 ⫾ 0.003 5.0
K 0.0023 ⫾ 0.0036 0.112 ⫾ 0.004 48.8
Mean 0.0071 ⫾ 0.0025 0.066 ⫾ 0.025 9.4
The results (expressed in lesions per 10
6
normal bases per Joules per centi-
meter squared) represent the slope ⫾ SE of the linear regression for 8-oxod-
Guo or T⬍⬎T with respect to the applied UV dose for each donor. The mean
is average ⫾ SD.
Fig. 4. Formation of 8-oxodGuo and thymine-thymine cyclobutane dimers
within human skin exposed to UVB radiation (0.2 J兾cm
2
). The results are
expressed in lesions per 10
6
normal bases and are the average ⫾ SD.
Mouret et al. PNAS
兩
September 12, 2006
兩
vol. 103
兩
no. 37
兩
13767
GENETICS SEE COMMENTARY
of bipyrimidine photoproducts induced upon exposure to UVA or
UVB. For this purpose, primary keratinocytes were cultured from
the skin of each of the donors (donors A–F), exposed to UVB and
UVA radiations, and then the yield of formation of photoproducts
was quantified. Although the relative proportions of the different
photoproducts were similar in vivo and in culture, for both UVA
and UVB radiations (Fig. 2), a major difference was observed
concerning the absolute yields of formation (Table 1). Indeed,
UVB radiation produced 22-fold more bipyrimidine photoproducts
in keratinocyte monolayers than in the whole human skin. In
contrast, the yield of CPD formation measured after UVA irradi-
ation was only 1.5-fold lower in skin with respect to keratinocytes.
Interestingly, the protection was the same for all photoproducts.
The values of the attenuation factor for T⬍⬎T, the major damage
induced by both UV components, is 22 and 1.7 for UVB and UVA,
respectively.
Discussion
The formation of CPD in higher yield rather than oxidatively
generated DNA lesions upon UVA irradiation represents recent
new information that challenges the hypothesis that UVA-induced
DNA damage are mostly mediated by oxidative stress. These data
were obtained mostly in cultured cells (15, 27–29) and remained to
be confirmed in whole skin. For this purpose, we determined the
yield of a wide array of dimeric pyrimidine photoproducts by using
an HPLC-MS兾MS assay that allows the simultaneous and individ-
ual quantification of the 12 possible cyclobutane pyrimidine dimers,
(6-4) photoproducts, and Dewar valence isomers (32). Quantitative
data on the formation of 8-oxodGuo, the main UVA-oxidatively
induced DNA damage, also were obtained.
The distribution of the bipy rimidine photoproducts induced by
UVB radiation in the DNA of whole human skin was found to
be very similar to that obtained in previous studies involving
isolated DNA, human monocytes, or rodent cell lines (28, 32).
The major lesion in human skin was T⬍⬎T, which was produced
in a 10-fold higher yield than the TT pyrimidine (6-4) pyrimi-
done photoproduct. T⬍⬎C also was generated, in a 2-fold higher
yield than TC (6-4) photoproducts. The overall amount of the
t wo latter TC photoproducts was similar to that of the TT lesions,
in agreement with previous HPLC-MS兾MS measurements in
human monocytes or human dermal fibroblasts exposed to the
same UVB source (32, 34). Similar to results obtained on
isolated DNA and cultured cells, CT and CC sites also were much
less reactive than TT and TC sequences in the DNA of human
sk in. The present results contrast with those of previous studies
in which the quantification of the photoproducts in the DNA of
human skin exposed to UVB radiation was achieved by using the
HPLC
32
P-postlabeling assay. These works led to the conclusion
that T⬍⬎C was the major lesion generated (38–40). However,
it may be pointed out that the same approach failed to provide
reproducible data concerning the formation of CPDs in skin
ex posed to simulated sunlight (36, 37, 41–46).
Our results clearly demonstrate that CPDs also are produced in
significant yield in the whole human skin exposed to UVA radia-
tion, in agreement with data obtained by immunohistochemistry
(25, 47, 48). CPDs were predominantly produced at TT sites after
UVA irradiation of human skin. Interestingly, the ratio between the
yield of CPDs produced at TT compared with those at TC and CT
sites was much higher than upon exposure to UVB radiation. In
addition, no other photoproducts such as pyrimidine (6-4) pyrimi-
done photoproducts and Dewar valence isomers could be detected
in human skin exposed to UVA radiation. These observations,
which are in agreement with results obtained in CHO (28, 30) and
human skin cells (29), show that the mechanism of formation of
CPDs upon UVA and UVB irradiation is different. The distribu-
tion of dimeric photoproducts suggests that a photosensitization
reaction involving a triplet energy transfer mechanisms rather than
a direct excitation proce ss takes place upon exposure of skin to
UVA, similarly to what is observed in the effects of some phototoxic
drugs (49–51). Moreover, in contrast to a widely accepted hypoth-
esis, we also found that, like in cultured cells (15, 28, 29), T⬍⬎T was
produced in a larger amount than 8-oxodGuo in human skin
exposed to UVA. UVA-mediated photosensitization processes are
expected to be partly involved in photocarcinogenesis and, in
particular, in the induction of melanoma (10, 52). Emphasis has
been placed in the past on photo-oxidative processes that may occur
in the dermis, but, in the light of our present re sults, photosensitized
formation of cyclobutane dimers should not be neglected.
The results reported above show that the UV photochemistry of
DNA is roughly the same in cultured mammalian cells and in whole
skin, at least in terms of relative frequency of the different
photoproducts. However, the structure of the skin was found to
greatly affect the relative sensitivity of DNA to UVA and UVB
radiations. It is universally presumed that the upper epidermal
layers in human skin, most particularly the stratum corneum, and
the melanin would protect the basal layer against DNA damage
formation by blocking the penetration of a significant portion of the
UV spectrum. However, to our best knowledge, only one study has
evaluated the role played by the multilayered structure of the skin
against the induction of CPD by either UVB radiation or simulated
sunlight in the DNA of engineered human skin (53). To determine
the extent of protection afforded by skin against the radiation
emitted by our UVA and UVB sources, we compared the yield of
formation of DNA damage within the whole human skin with that
determined in monolayer cultures of keratinocytes obtained from
the same donor. This approach allowed us to establish that the
organization of the human skin efficiently protects against UVB-
induced formation of DNA bipyrimidine le sions by a factor of ⬇22.
In contrast, a very weak protection is afforded by the skin against
UVA; the corresponding factor is 1.5. Interestingly, immunohisto-
chemical studies have shown that the frequency of DNA photo-
products is constant through all of the epidermis thickness of skin
exposed to UVB radiation (25, 48). It therefore is most likely that
the efficient UVB protection is provided by the components of the
stratum corneum (urocanic acid and aromatic amino acids) and by
melanin.
CPD is the major UVA-induced DNA lesion, and skin does not
afford efficient photoprotection against its formation. These two
features are likely worsened in terms of deleterious biological
effects by the fact that CPDs are persistent DNA lesions in
UVA-irradiated skin. Indeed, we observed that TT CPD remained
Fig. 5. Persistence of thymine-thymine cyclobutane dimers within human
skin exposed to UVA (100 J兾cm
2
) or UVB (0.1 J兾cm
2
) radiation. Repair of TT (6-4)
photoproducts within UVB irradiated skin also is shown. The results are
expressed in percentage of residual lesions and are the average ⫾ SD of data
obtained with four different donors. The statistical significance in the level of
remaining T⬍⬎T between UVA and UVB radiation was calculated by using the
Student t test. Difference between UVA and UVB was found to be statistically
significant at 24 h (P ⬍ 0.002) and 48 h (P ⬍ 0.05).
13768
兩
www.pnas.org兾cgi兾doi兾10.1073兾pnas.0604213103 Mouret et al.
present in a large amount in the DNA of human skin 48 h after
UVA or UVB irradiations have ceased. Similar observations
previously were made after exposure of skin to simulated solar
radiation (35–37, 42, 43, 54). Interestingly, we found that the level
of unrepaired UVA-induced TT CPD was higher than after UVB
irradiation. This result is reminiscent of a similar fact observed in
primary culture of skin keratinocytes (33). The effect of UVA on
the repair of CPD remains to be clearly understood. Obviously, TT
cyclobutane dimers are the same whether they are produced by
UVA or UVB. The reduced repair capacities of UVA-irradiated
cells could be explained by a different in-cell cycle arrest after
irradiation (55, 56) or by a degradation of DNA repair protein by
the oxidative stre ss induced by UVA.
Conclusion
The data obtained in the present work further emphasize the likely
role played by cyclobutane pyrimidine dimers in UVA genotoxicity.
Indeed, this class of damage was found to be produced in larger
amounts than oxidatively generated damage. In addition, CPDs
were shown to persist over a rather long period after UVA
irradiation. It thus may be anticipated that part of the UVA-
induced mutations are due to the presence of CPDs. This proposal
is in agreement with a series of recent mutagenesis data that clearly
implicate CPDs, unlike 8-oxodGuo, as major promutagenic DNA
photoproducts induced by UVA radiation (14, 19, 20, 30, 57). The
TT cyclobutane dimer is known to be rather poorly mutagenic.
However, a number of mutations at TT sites were observed in
rodent cells exposed to UVA (30). It may be added that TC and CT
CPDs also are produced upon UVA irradiation of skin, although
in an ⬇10-fold lower yield than the TT dimer. Cytosine-containing
CPDs are known to be mutagenic and the TC to TT transition is
a major mutagenic event in skin tumors (2–4). It is likely that these
transitions are mostly the re sult of UVB-induced damage, but the
present results show that a UVA contribution cannot be ruled out.
The observation that UVA may induce the formation of CPDs also
may have some mechanistic implications into its immunosuppres-
sive properties. Indeed, DNA damage has been shown to be
involved in UVA-induced immunosuppression (58). The likely
biological significance of the UVA-induced CPDs make s necessary
additional work to definitively establish the underlying mechanism
of formation. Determination of an action spectra in skin similar to
that obtained in rodent cell line (15) should provide at least a partial
answer to this question. In addition, the contribution of UVA-
induced CPDs to the overall DNA damage induced by sunlight also
remains to be determined within a complete terrestrial solar
spectrum.
Materials and Methods
Skin Sample Preparation and Cell Culture. Human skins were ob-
tained immediately after breast plastic surgery from healthy donors
with their inform consent (Centre Hospitalier Universitaire de
Grenoble and Clinique des Alpes, Grenoble, France). All donors
were Caucasian (16–62 year old), and their skin phototype was
between II and III according to the Fitzpatrick classification (59).
The whole skin was rinsed four times in PBS, for 30 sec in 70%
ethanol, and five times again in PBS. Then, 6-mm punch biopsie s
were made, placed in a 35-mm Petri dish with PBS, and stored in
the dark until irradiation.
For half of the donors, keratinocytes were isolated from whole
skin. Briefly, to obtain primary keratinocytes, the skin, after
washing with ethanol 70%, was rinsed 10 times in PBS containing
100 units兾ml penicillin, 100
g兾ml streptomycin, and 0.25
g兾ml
amphotericin B, and finally was immersed overnight in 0.25%
trypsin at 4°C. The action of trypsin was stopped by the addition of
DMEM with 10% FCS. Epidermis was detached from dermis and
homogenized. After centrifugation, harvested primary keratino-
cytes were seeded in Keratinocyte serum-free medium supple-
mented with 1.5 ng/ml EGF兾25
g/ml bovine pituitary extract兾75
g/ml primocin and then cultured at 37°C in a humidified atmo-
sphere containing 5% CO
2
. For all experiments, cells were used at
passages 2 or 3.
UV Sources and Irradiation Procedure. The UVB source used was a
VL 215 G irradiator (Bioblock Scientific, Illkirch, France) fitted
with t wo 15-W tubes with a spectrum distribution mostly emit-
ting at 312 nm as described in ref. 33. The irradiance was
measured by a VLX 3W radiometer (Vilbert Lourmat, Marne la
Valle´e, France) equipped with a 312-nm probe. The sample
ef fectively received an average irradiance of 0.3 mW兾 cm
2
with
the lamp placed 80 cm above the irradiated targets.
The UVA source used was a Waldman UVA 700L irradiator
fitted w ith a high pressure lamp MSR 700 (700 W) (Waldman,
Villingen-Schwenn ingen, Ger many) with an emission spectrum
providing mostly photons of wavelength ⬎330 nm as described
in ref. 29. The irradiance ef fectively received by the sample was
40 mW兾cm
2
.
Petri dish-containing skin biopsies were exposed immediately
and lid removed to UVB radiation or on ice to UVA radiation. Just
after irradiation, skin biopsie s were dripped dry, frozen in liquid
nitrogen, and kept at ⫺80°C until the DNA extraction step.
Sham-irradiated skin biopsies were treated similarly and kept in the
dark during irradiation. To study the persistence of DNA photo-
products in human skin, three biopsies were pooled per point.
Typically after irradiation with UVB (0.1 J兾cm
2
) or UVA (100
J兾cm
2
) radiation, PBS was replaced by fresh medium (DMEM兾
F12), and skin biopsies then were incubated at 37°C for increasing
period.
For cultured cells irradiation, keratinocytes were seeded at 5 ⫻
10
5
cells per 100-mm Petri dish (UVB irradiation) or 2 ⫻ 10
5
cells
per 60-mm Petri dish (UVA irradiation) and grown to subconflu-
ence during 5 days. Practically, just before irradiation, culture
medium was removed, and the cells were rinsed twice with PBS.
Irradiations then were performed in PBS with the lid removed at
room temperature for UVB or on ice for UVA. Immediately after
treatment, keratinocytes were trypsinized and recovered by cen-
trifugation. The cell pellet then was frozen and kept at ⫺80°C until
extraction. In all cases, irradiations of whole human skin or primary
keratinocytes were performed in triplicate.
DNA Extraction and Digestion. Two different protocols were used to
extract DNA from whole skin and primary keratinocytes. For
human skin, after a grinding step DNA was extracted by using the
DNEasy Tissue Kit obtained from Qiagen (Courtaboeuf, France).
Briefly, the first step was the cold grinding of the biopsy by using
a manual pe stle cooled down with liquid nitrogen. The resulting
powder then was recovered with 180
l of the first lysis buffer ATL.
After adding proteinase K, samples were incubated overnight at
55°C for a complete lysis of the tissue. An RNase A treatment and
a second lysis step involving buffer AL were performed before
loading the lysate sample s onto the DNEasy mini spin column.
DNA then was eluted in 200
l of deionized water after efficient
washing. The sample was freeze-dried overnight, and the resulting
DNA residue was dissolved in 50
l of a 0.1 mM deferroxiamine
mesylate solution. To study the persistence of DNA photoproducts
and the formation of 8-oxodGuo in human skin, the protocol was
modified slightly to allow the simultaneous treatment of three and
four biopsies, respectively. For this purpose, the volume of each
buffer was doubled and elution was performed in two successive
steps by using 200
l of a 0.1 mM deferroxiamine mesylate solution
each time. Samples then were freeze-dried overnight and dissolved
into 50
l of a 0.1 mM deferroxiamine solution for the measure-
ment of DNA photoproducts or concentrated with a speed-vac for
8-oxodGuo analysis.
For keratinocytes, DNA extraction was performed by using a
chaotropic method as reported in ref. 60. In brief, the cell pellet was
efficiently homogenized by pipeting in the presence of Triton X-100
Mouret et al. PNAS
兩
September 12, 2006
兩
vol. 103
兩
no. 37
兩
13769
GENETICS SEE COMMENTARY
to remove the plasma membrane. Nuclei were isolated by centrif-
ugation and made soluble by the addition of SDS. After successive
treatment by RNase s and proteinase, DNA was precipitated by
sodium iodide and 2-propanol, and the resulting pellet was dis-
solved in 50
l of a 0.1 mM deferroxiamine mesylate solution.
In all cases, DNA then was digested by incubation with 0.025
units of phosphodiesterase II兾2.5 units of DNase II兾0.5 units of
nuclease P1 at pH 6 for2hat37°C. An additional dige stion step
involving phosphodiesterase I and 2 units of alkaline phosphatase
at pH 8 was performed. After this second digestion step, 0.1 M HCl
was added, the sample was centrifuged, and then transferred into
HPLC vials. The solution contained normal nucleosides and 8-
oxodGuo, whereas the bipyrimidine photoproducts consisted of
modified dinucleoside monophosphates. Cytosine-containing cy-
clobutane dimers were obtained as their uracil derivatives after fast
and quantitative spontaneous deamination.
Analysis of DNA Lesions.
Analysis of bipyrimidine photoproducts.
The
samples were freeze-dried overnight, and the resulting residue s
were made soluble in 40
l of a 20 mM triethylammonium acetate
solution just before analysis by HPLC-MS兾MS. The sample s were
injected onto a series 1100 microHPLC system (Agilent Technol-
ogies, Massy, France) coupled to a API 3000 triple quadrupolar
mass spectrometer (PerkinElmer兾SCIEX, Thornhill, ON, Can-
ada). The separation was achieved on an Uptisphere ODB octa-
decylsilyl silica gel column (150 ⫻ 2 mm; 3-
m particle size) from
Interchim (Monluc¸on, France) with a gradient of acetonitrile in a
2 mM triethylammonium acetate solution. Chromatographic con-
ditions and mass spectrometry features were as described in ref. 32.
Normal nucleosides were quantified by HPLC UV with the detec-
tor set at 270 nm. The transitions used for the detection of the
different bipyrimidine photoproducts were as follows: 545 3 447
for the TT cyclobutane dimer (T⬍⬎T), 545 3 432 for the TT (6-4)
photoproduct [TT (6-4)], 531 3 195 for the TC and the CT
cyclobutane dimers (T⬍⬎C and C⬍⬎T, re spectively), 530 3 195
for the TC (6-4) photoproduct [TC (6-4)], and 517 3 195 for the
CC cyclobutane dimer (C⬍⬎C).
Analysis of 8-oxodGuo.
The samples were analyzed by HPLC asso-
ciated with a coulometric electrochemical detection with slight
modifications of the method described in ref. 27. Briefly, the
separation was performed on an Uptisphere ODB octadecylsilyl
silica gel column (250 ⫻ 4.6 mm, 5
m particle size) with an isocratic
eluent (25 mM potassium phosphate, 8% methanol). The coulo-
metric detection was provided by a Coulochem II detector
equipped with a 5011 cell (ESA, Chelmsford, MA) with the
potential of the two electrodes set at 200 and 450 mV, respectively.
Normal nucleosides were quantified by a UV detector at the output
of the HPLC column that was set at 254 nm.
The amount of each individual lesion determined in both HPLC-
MS兾MS and HPLC-electrochemical analyses was inferred from an
external calibration obtained by injecting known amounts of the
authentic compound. Similarly, the amount of analyzed DNA was
determined from the area of the peak of dGuo after appropriate
calibration of the UV detector.
1. Cadet J, Sage E, Douki T (2005) Mutat Res 571:3–17.
2. Brash DE, Rudolph JA, Simon JA, Lin A, McKenna GJ, Baden HP, Halperin AJ,
Ponten J (1991) Proc Natl Acad Sci USA 88:10124–10128.
3. Ziegler A, Leffell DJ, Kunala S, Sharma HW, Gailani M, Simon JA, Halperin AJ,
Baden HP, Shapiro PE, Bale AE, Brash DE (1993) Proc Natl Acad Sci USA
90:4216–4220.
4. Dumaz N, Drougard C, Sarasin A, Daya-Grosjean L (1993) Proc Natl Acad Sci
USA 90:10529–10533.
5. Stary A, Robert C, Sarasin A (1997) Mutat Res 383:1–8.
6. Drobetsky EA, Turcotte J, Chateauneuf A (1995) Proc Natl Acad Sci USA
92:2350–2354.
7. Pastila R, Leszczynski D (2005) Photoder matol Photoimmunol Photomed
21:183–190.
8. de Laat A, van der Leun JC, de Gruijl FR (1997) Carcinogenesis 18, 1013–1020.
9. Halliday GM (2005) Mutat Res 571:107–120.
10. Wang SQ, Setlow R, Ber wick M, Polsky D, Marghoob AA, Kopf AW, Bart RS
(2001) J Am Acad Der matol 44:837–846.
11. Setlow RB, Grist E, Thompson K, Woodhead AD (1993) Proc Natl Acad Sci USA
90:6666–6670.
12. Gasparro FP (2000) Environ Health Perspect 108(Suppl 1):71–78.
13. Westerdahl J, Ingvar C, Masback A, Jonsson N, Olsson H (2000) Br J Cancer
82:1593–1599.
14. Agar NS, Halliday GM, Barnetson RS, Ananthaswamy HN, Wheeler M, Jones
AM (2004) Proc Natl Acad Sci USA 101:4954–4959.
15. Kielbassa C, Roza L, Epe B (1997) Carcinogenesis 18:811–816.
16. Pouget JP, Douk i T, Richard M-J, Cadet J (2000) Chem Res Toxicol 13:541–549.
17. Ravanat JL, Di Mascio P, Martinez GR, Medeiros MH, Cadet J (2000) J Biol
Chem 275:40601–40604.
18. Sage E, Lamolet B, Brulay E, Moustacchi E, Chaˆteauneuf A, Drobetsky EA
(1996) Proc Natl Acad Sci USA 93:176–180.
19. Kappes UP, Luo D, Potter M, Schulmeister K, Runger TM (2006) J Invest
Dermatol 126:667–675.
20. Kappes UP, Runger TM (2005) Radiat Res 164:440–445.
21. Tyrrell RM (1973) Photochem Photobiol 17:69–73.
22. Kvam E, Tyrrell RM (1997) Carcinogenesis 18:2379–2384.
23. Perdiz D, Grof P, Mezzina M, Nik aido O, Moustac chi E, Sage E (2000) JBiol
Chem 275:26732–26742.
24. Freeman SE, Ryan SL (1990) Mutat Res 235:181–186.
25. Young A R, Chadwick CA, Harrison GI, Nikaido O, Ramsden J, Potten CS (1998)
J Invest Dermatol 111:982–988.
26. Freeman SE, Hacham H, Gange RW, Maytum DJ, Sutherland JC, Sutherland
BM (1989) Proc Natl Acad Sci USA 86:5605–5609.
27. Douki T, Perdiz D, Grof P, Kuluncsics Z, Moustacchi E, Cadet J, Sage E (1999)
Photochem Photobiol 70:184–190.
28. Douki T, Reynaud-A ngelin A, Cadet J, Sage E (2003) Biochemistry 42:9221–9226.
29. Courdavault S, Baudouin C, Charveron M, Favier A, Cadet J, Douki T (2004)
Mutat Res 556:135–142.
30. Rochette PJ, Therrien JP, Drouin R, Perdiz D, Bastien N, Drobetsky EA, Sage
E (2003) Nucleic Acids Res 31:2786–2794.
31. Kozmin S, Slezak G, Reynaud-Angelin A, Elie C, de Rycke Y, Boiteux S, Sage
E (2005) Proc Natl Acad Sci USA 102:13538–13543.
32. Douki T, Cadet J (2001) Biochemistry 40:2495–2501.
33. Courdavault S, Baudouin C, Charveron M, Canguilhem B, Favier A, Cadet J,
Douki T (2005) DNA Repair 4:836–844.
34. Courdavault S, Baudouin C, Sauvaigo S, Mouret S, Candeias S, Charveron M,
Fav ier A, Cadet J, Douki T (2004) Photochem Photobiol 79:145–151.
35. Young AR, Chadw ick CA, Harrison GI, Hawk JL, Nikaido O, Potten CS (1996)
J Invest Dermatol 106:1307–1313.
36. Xu G, Snellman E, Bykov VJ, Jansen CT, Hemminki K (2000) J Invest Dermatol
114:628–631.
37. Hemminki K, Xu G, Kause L, Koulu LM, Zhao C, Jansen CT (2002) Carcino-
genesis 23:605–609.
38. Bykov VJ, Marcusson JA, Hemminki K (1998) Cancer Res 58:2961–2964.
39. Bykov VJ, Marcusson JA, Hemminki K (2000) J Invest Dermatol 114:40–43.
40. Bykov VJ, Hemminki K (1996) Carcinogenesis 17:1949–1955.
41. Zhao C, Snellman E, Jansen CT, Hemminki K (2002) J Invest Dermatol
118:180–184.
42. Zhao C, Snellman E, Jansen CT, Hemminki K (2002) Int J Cancer 98:331–334.
43. Xu G, Snellman E, Jansen CT, Hemmink i K (2000) J Invest Dermatol 115:95–99.
44. Xu G, Snellman E, Bykov VJ, Jansen CT, Hemminki K (2000) Mutat Res
459:195–202.
45. Bykov VJ, Sheehan JM, Hemminki K, Young AR (1999) J Invest Der matol
112:326–331.
46. Bykov VJ, Jansen CT, Hemminki K (1998) Cancer Epidemiol Biomarkers Prev
7:199–202.
47. Young AR, Potten CS, Nikaido O, Parsons PG, Boenders J, Ramsden JM,
Chadw ick CA (1998) J Invest Dermatol 111:936–940.
48. Chadwick CA, Potten CS, Nikaido O, Matsunaga T, Proby C, Young AR (1995)
J Photochem Photobiol B 28:163–170.
49. Lhiaubet V, Paillous N, Chouini-Lalanne N (2001) Photochem Photobiol
74:670–678.
50. Sauvaigo S, Douki T, Odin F, Caillat S, Ravanat J-L, Cadet J (2001) Photochem
Photobiol 13:230–237.
51. Traynor NJ, Gibbs NK (1999) Photochem Photobiol 70:957–959.
52. Wood SR, Berwick M, Ley RD, Walter RB, Setlow RB, Timmins GS (2006) Proc
Natl Acad Sci USA 103:4111–4115.
53. Therrien JP, Rouabhia M, Drobetsky EA, Drouin R (1999) Cancer Res
59:285–289.
54. Katiyar SK, Matsui MS, Mukhtar H (2000) Photochem Photobiol 72:788–793.
55. Banrud H, Stokke T, Moan J, Berg K (1995) Carcinogenesis 16:1087–1094.
56. de Laat A, van Tilburg M, van der Leun JC, van Vloten WA, de Gruijl FR (1996)
Photochem Photobiol 63:492–497.
57. Besaratinia A, Synold TW, Chen H-H, Chang C, Xi B, Riggs AD, Pfeifer GP
(2005) Proc Natl Acad Sci USA 102:10058–10063.
58. Kuchel JM, Barnetson RS, Halliday GM (2005) Photochem Photobiol Sci
4:577–582.
59. Fitzpatrick TB (1988) Arch Dermatol 124:869–871.
60. Ravanat JL, Douki T, Duez P, Gremaud E, Herbert K, Hofer T, Lasserre L,
Saint-Pierre C, Favier A, Cadet J (2002) Carcinogenesis 23:1911–1918.
13770
兩
www.pnas.org兾cgi兾doi兾10.1073兾pnas.0604213103 Mouret et al.
