If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
* This work was supported by grants from la Ligue Nationale Française contre le Cancer, Institut Curie (Genotoxicology Program).The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. § Recipient of a fellowship from Ministere de l'Education Nationale de la Recherche and from Académie Nationale de Médecine. Present address: CNRS UMR 2027, Institut Curie, Centre Universitaire, F-91405 Orsay, France. ‖ Supported by a joint CNRS/Hungarian Academy of Sciences program.
In order to better understand the relative contribution of the different UV components of sunlight to solar mutagenesis, the distribution of the bipyrimidine photolesions, cyclobutane pyrimidine dimers (CPD), (6-4) photoproducts ((6–4)PP), and their Dewar valence photoisomers (DewarPP) was examined in Chinese hamster ovary cells irradiated with UVC, UVB, or UVA radiation or simulated sunlight. The absolute amount of each type of photoproduct was measured by using a calibrated and sensitive immuno-dot-blot assay. As already established for UVC and UVB, we report the production of CPD by UVA radiation, at a yield in accordance with the DNA absorption spectrum. At biologically relevant doses, DewarPP were more efficiently produced by simulated solar light than by UVB (ratios of DewarPP to (6-4)PP of 1:3 and 1:8, respectively), but were detected neither after UVA nor after UVC radiation. The comparative rates of formation for CPD, (6-4)PP and DewarPP are 1:0.25 for UVC, 1:0.12:0.014 for UVB, and 1:0.18:0.06 for simulated sunlight. The repair rates of these photoproducts were also studied in nucleotide excision repair-proficient cells irradiated with UVB, UVA radiation, or simulated sunlight. Interestingly, DewarPP were eliminated slowly, inefficiently, and at the same rate as CPD. In contrast, removal of (6-4) photoproducts was rapid and completed 24 h after exposure. Altogether, our results indicate that, in addition to CPD and (6-4)PP, DewarPP may play a role in solar cytotoxicity and mutagenesis.
cyclobutane pyrimidine dimers
pyrimidine (6-4) pyrimidone photoproducts
simulated solar light
nucleotide excision repair
global genomic repair
Chinese hamster ovary
phosphate-buffered saline plus Tween 20
Overwhelming evidence associates the steadily increasing incidence of skin cancer with an increased exposure to the UV components of sunlight (
). In the mutated p53 tumor suppressor gene of skin cancer, the majority of mutations harbor the “UV mutational signature,” i.e. C → T transitions and CC → TT tandem double mutations, which occur at sites where major DNA photoproducts are formed (
). The photolesions that are readily produced at these sites by UVC (254 nm) and UVB (280–320 nm), such as cyclobutane pyrimidine dimers (CPD)1 and pyrimidine (6-4) pyrimidone photoproducts ((6–4)PP), are thought to represent the predominant forms of premutagenic damage (
Until recently, the investigation of DNA photolesions has been predominantly conducted using UVC at 254 nm or UVB. However, wavelengths lower than 290 nm do not reach the Earth's surface due to absorption by the stratosphere. Consequently, the human population is mostly exposed to longer wavelength UVB (λ > 295 nm; Ref.
) and UVA (320–400 nm) radiation, which constitute about 5% and 95%, respectively, of the solar spectrum. The relatively weak UVB component is believed to be responsible for most of the biological effects of sunlight, which are mediated by direct absorption of UVB by DNA. Since UVA is poorly absorbed by DNA, its genotoxic effect has been attributed to indirect photosensitizing reactions (
). At the DNA level, photosensitization induces oxidative damage either by charge transfer from excited endogenous chromophores, or by reactions with reactive oxygen species that are generated at these wavelengths. Indeed, photo-induced oxidative DNA modifications, such as single strand breaks, DNA-protein cross-links, alkali-labile sites, 8-oxo-7,8-dihydro-2′-deoxyguanosine (8-oxodGuo) have been observed (see Ref.
In mammalian cells, UV-induced bipyrimidine photoproducts are removed via nucleotide excision repair (NER), either by transcription-coupled repair (TCR) or global genomic repair (GGR), while oxidative photolesions and single strand breaks are most likely eliminated by base excision repair (
). The repair kinetics and efficiencies of UV-induced damage vary considerably from one type of photolesion to another, and between species, such as rodent and human. For example, (6-4)PP are rapidly and efficiently removed in both organisms, probably via GGR, whereas CPD are repaired rather slowly and incompletely (
The mutational specificities of UVC, UVB, and UVA radiation and simulated solar light (SSL) were previously determined at theaprt locus in Chinese hamster ovary (CHO) cells that were proficient or deficient in DNA repair (
). Briefly, it was found that: 1) in the NER-proficient cell line, GC → AT transitions and CC →TT tandem double mutations were the major types of events following SSL exposure, whereas the former type of mutation largely predominated following UVB irradiation; 2) while GC → AT transitions still contributed up to 27% of the changes after UVA exposure, they were no longer preponderant; 3) in contrast, in the NER-deficient cell line, GC → AT transitions, all of which occurred at bipyrimidine sites, represented a large proportion of the mutational events induced by UVA. Altogether, these results suggested the formation of bipyrimidine photoproducts by UVA radiation and a role for such damage in UVA-induced mutagenesis. Indeed, we later observed a significant production of CPD in plasmid DNA irradiated with UVA (
In order to better understand the relative contribution of the different UV components to solar mutagenesis, we previously examined their respective DNA damage profiles. In particular, we investigated the relative spectral effectiveness in the induction of CPD and 8-oxodGuo. We showed that CPD are formed at least as frequently as 8-oxodGuo by UVA radiation (
). Here, using specific antibodies, we determine the yield of three bipyrimidine photoproducts, CPD, (6-4)PP, and DewarPP, in the genomic DNA of CHO cells exposed to physiologically relevant doses of UVC, UVB, UVA, and SSL. An estimation of the levels of these three types of photolesions was previously obtained by immunostaining cells irradiated with UVC, broad or narrow band UVB, or natural sunlight with monoclonal antibodies similar to those used here (
) assigned the increase in arbitrary gray scale values to the induction of antibody binding sites and estimated the relative production of (6-4)PP and DewarPP by assuming an equal luminescence signal for all three antibodies. However, their data took into account neither the very different affinities of the three antibodies for their respective substrates nor the possibility of multimeric forms of the immunoglobulins. In addition, comparison of the photolesion induction by various UV lamps was made possible only by expressing the fluence rates as dimer-equivalent fluences. In contrast to estimating relative values, determination of the absolute amount of different damage due to UV radiation allows a comparison of the biological effectivenesses of two or more types of photobiological action, e.g. induction of CPD and other types of photolesion. This is not the case for the determination of relative lesion induction, even if it can provide a precise action spectrum for a given photobiological effect. In this report, we describe the results of a method that allows a determination of the absolute amounts of these photoproducts. This method relies on the calibration of the immunochemical signals for each antibody using specific DNA repair enzymes. We present the first absolute determination of all three bipyrimidine photoproducts using one detection method. Our data show that the distribution of the bipyrimidine photolesions varies greatly depending on the wavelength region considered, and we demonstrate that UVA radiation produces significant amounts of CPD at doses comparable to those of human exposure. In addition, the relatively high production of DewarPP by SSL observed in this study, suggests that DewarPP may be a biologically relevant photolesion. However, the cytotoxic and mutagenic properties of this damage are not well established, mainly due to the difficulties in detecting DewarPP. In this respect, we analyzed the capacity of CHO cells to remove DewarPP, as well as CPD and (6-4)PP, using the same immunological approach. Furthermore, in order to investigate the possible effects of other types of lesions, i.e. DNA strand breaks and oxidative DNA damage, on the repair of the bipyrimidine photoproducts, CHO cells were irradiated with UVB, UVA radiation, or SSL. We demonstrate that DewarPP are removed with similar kinetics and efficiency as CPD, although they are valence isomers of (6-4)PP. In contrast, the removal of (6-4)PP is almost complete within 6 h. The similarity between the responses of DewarPP and CPD suggests that DewarPP may contribute significantly to the mutagenesis that occurs after exposure to sunlight.
UVC irradiation was performed using a germicidal lamp emitting primarily at 254 nm. Broad band UVB irradiation (290–320 nm) was carried out with a set of six 15-W fluorescent tubes (Vilber Lourmat, Torcy, France) having a spectral irradiance very similar to that of FS20 lamps. The incident light was filtered through a Schott WG 305 cut-off filter (thickness, 2 mm), which efficiently blocks contaminating wavelengths below 290 nm. Polychromatic UVA radiation was obtained from an Osram Ultramed 400-W gas discharge lamp, emitting through a 2-cm water layer held in Pyrex glass, and through a Schott WG345 filter (thickness, 3 mm) and an anticaloric KG1 filter. The delivered radiation comprised 60.8% UVA radiation (320–400 nm) including 0.05% UVA2 wavelengths (320–340 nm), 39.2% visible light, and less than 5 × 10−4% UVB (280–320 nm). The simulated solar light (SSL) was produced by a 2500-W xenon compact arc lamp (Conrad-Hanovia Inc., Newark, NJ) and passed through a Schott WG320 filter (thickness, 3 mm). The incident light was composed of 0.8% UVB, 6% UVA, 44.5% visible light, 48.7% infrared, and less than 10−5 % UVC. The proportions of UVB, UVA, visible light, and infrared in natural terrestrial sunlight are approximately 0.3%, 5.1%, 62.7%, and 31.9%, respectively. Emission spectra after filtration were previously presented (
). Percentage spectral irradiance values, which were normalized for the UV region (290–400 nm) only, are given under “Results.”
UVC fluence rate (0.16 J·m−2·s−1) was measured with a Latarjet dosimeter, and the irradiation time did not exceed 4 min. UVB (12.5 J·m−2·s−1) and UVA (135 J·m−2·s−1) fluence rates were measured with a radiometer VLX 3W equipped with interferential filters (Vilber Lourmat, Torcy, France); the exposure time did not exceed 15 and 120 min, respectively. According to a YSI Kettering 65A thermopile (Yellow Spring Instruments, OH), the fluence rate for SSL was 1250 J·m−2·s−1. Irradiation lasted no longer than 45 min. The UV wavelengths contributed to approximately 6.8% of the total energy measured with the thermopile.
Irradiation of CHO Cells and Genomic DNA Isolation
CHO cell lines AT3–2 and UVL9 (kindly provided by G. Adair, University of Texas, Smithville, USA), which are, respectively, proficient and deficient in nucleotide excision repair (mutated in ERCC1gene), were used. Cells were routinely grown in α-minimal essential medium containing 10% fetal calf serum (Life Technologies, Inc.) and 16 μg/ml gentallin. 2–4 h prior to irradiation, 107cells were seeded on 60-mm dishes (Costar). Cells were then washed and irradiated with UVB, UVA, and SSL in phosphate-buffered saline (PBS) on ice to prevent repair during exposure. Irradiation with UVC was performed at room temperature, since short exposures were required.
For photolesion quantification, NER− cells were used to avoid repair during long irradiation periods. After irradiation, cells, maintained on ice, were immediately scrapped from the dishes into ice-cold PBS buffer and pellets were stored at −20 °C until use. For repair experiments, confluent NER+ cells received 1 kJ·m−2 UVB, 1000 kJ·m−2 UVA, or 4500 kJ·m−2 SSL (equivalent to 306 kJ·m−2 UV energy, a conversion that was used throughout the experiments described here). The duration of irradiation corresponded to 160 s, 120 min, and 45 min, respectively. To study the repair of the DewarPP after UVB exposure, a dose of 5 kJ·m−2 was given to ensure sufficient induction of this photoproduct. In order to get enough DNA for the immuno-dot-blot (IDB) assay, one dish per repair time was used. Immediately after irradiation, cells were scrapped into cold PBS buffer and one half was pelleted and stored at −20 °C for the determination of the photolesion at time t = 0, whereas the other half was allowed to undergo repair for a set time in fresh medium at 37 °C. At times t = 2, 4, 6, and 24 h after irradiation, cells were harvested, washed, pelleted, and stored at −20 °C. At the repair time of 24 h, cellular growth was not observed.
For DNA extraction, cells were lysed for 1 h at 37 °C in lysis buffer (20 mm Tris-HCl, pH 8, 20 mm NaCl, 20 mm EDTA, and 0.5% SDS). Samples were then incubated 4 h at 37 °C with 100 μg/ml RNase A, 5 units/ml RNase T1, and overnight at 37 °C with 10 μl of proteinase K (25 mg/ml, Roche Molecular Biochemicals). DNA was purified by two extractions with phenol and chloroform/isoamyl alcohol (25:24:1 v/v/v) and further precipitated by the cold ethanol procedure. The amount of DNA was determined spectrophotometrically (Shimadzu UV-160A spectrophotometer) on the basis of its absorbance at 260 nm.
Detection of the Bipyrimidine Photoproducts by IDB Assay
Cyclobutane pyrimidine dimers, (6-4)PP, and DewarPP were detected by using TDM-2, 64M-2, and DEM-1 monoclonal antibodies, respectively (
), a triplicate of 500 ng of heat-denatured DNA per dot was loaded (Hybri-dot Manifold, Life Technologies, Inc.) on nitrocellulose membrane (0.2 μm, BA83; Schleicher & Schuell). After blotting, the dots were rinsed twice with 100 μl of PBS. Membranes were saturated overnight at 4 °C in PBS containing 5% nonfat dry milk (NFM) and 0.1% Tween 20 (Sigma) and then incubated for 1 h at 37 °C with TDM-2, 64M-2, or DEM-1 antibody (dilution 1/1000, 1/250, and 1/1000, respectively, in 0.5% NFM, 0.1% Tween 20, PBS). After extensive washing with 0.5% NFM, 0.1% Tween 20, PBS (NFM-TPBS), membranes were incubated 1 h at room temperature with a 1/2000 dilution of a second anti-mouse horseradish peroxidase-conjugated antibody (Calbag, San Francisco, CA) in NFM-TPBS buffer. Blots were then washed extensively with NFM-TPBS buffer, and peroxidase activity was revealed with the enhanced chemiluminescence blotting detection system (RPN2106, ECL™, Amersham Pharmacia Biotech). Membranes were immediately exposed to x-ray films (Kodak XAR) for different times depending on the antibody (for a given antibody, the same exposure time was always used). Relative luminescence intensity was determined using a Biocom image analyzer and Macroautorag software (Biocom, Les Ulis, France).
In repair experiments, a streptavidin/biotin system was used to increase the luminescence signal for detection of the (6-4)PP after UVB. In this case, after incubation with 64M-2 antibodies in TPBS, membranes were washed extensively with TPBS and incubated 1 h at room temperature with biotinylated second antibody (1/5000 in TPBS, Calbag). Next, membranes were washed three times for 15 min with TPBS, incubated 30 min with streptavidin-peroxidase solution (1/5000 in TPBS), and further processed as described above. For each repair time within a given experiment, the percentage of remaining lesions was deduced by comparing the decrease in the luminescence intensity with the luminescence at time t = 0 h using the same set of irradiated cells. The fraction of damage remaining at various times was expressed as a mean value of eight determinations as follows: two series of irradiation experiments and two IDB assays per experiment were performed, and each sample was dotted in duplicate.
Quantitative Determination of CPD in Plasmid DNA by Plasmid Relaxation and IDB Assays
The pZ189 plasmid DNA (5500 base pairs) was UVB-irradiated on ice in 10 mm sodium phosphate buffer, pH 7.5, at doses ranging between 0.5 and 5 kJ·m−2. The number of CPD per plasmid was determined by measuring the conversion of irradiated supercoiled plasmid (form I) DNA to the open circular (form II) following digestion with the pyrimidine dimer-specific enzyme T4 endonuclease V (DenV protein, obtained from Applied Genetics and from Dr. J. Brouwer (Leiden University, The Netherlands)). Briefly, 150 ng of pZ189 DNA was incubated for 30 min at 37 °C in 10 μl of reaction buffer (50 mm KH2PO4, 100 mm NaCl, 1 mm EDTA, 1 mm dithiothreitol, 0.1 mg/ml bovine serum albumin) with or without 0.2 μg of DenV protein. Samples were then electrophoresed through a 0.8% agarose gel in the presence of ethidium bromide. Photographic negatives of the gels were scanned using a Biocom image analyzer. The number of enzyme-sensitive sites per plasmid as a function of the dose was calculated from the Poisson distribution and corrected for differential binding of ethidium bromide to supercoiled versus relaxed DNA.
CPD were also detected by TDM-2 antibody in irradiated plasmids after linearization and denaturation (triplicates of 500 ng) using the IDB assay as described above. The linearization procedure ensured complete denaturation of the plasmid molecules. The luminescence intensity was determined as a function of the dose. Finally, a calibration curve, which represented the luminescence intensity as a function of CPD/kbp, was obtained and used to calculate the number of CPD in the genomic DNA of irradiated cells.
Quantitative Determination of (6-4)PP in Plasmid DNA by Plasmid Relaxation and IDB Assays
400 ng of pCAT plasmid DNA (4610 base pairs; Promega, Madison, WI) was subjected to 10–40 J·m−2 UVC radiation. CPD were removed under UVA illumination by incubation of (250 ng) irradiated DNA for 1 h in 50 mm Tris-HCl, pH 7.4, 50 mm NaCl, 1 mm EDTA, 10 mm dithiothreitol, in the presence of 30 ng of photolyase from Escherichia coli (a generous gift from Dr. A. Sancar, University of North Carolina, Chapel Hill, NC). The UVA source used was a HPW 125 Philips lamp, which emits mainly at 365 nm; a dose of 50 kJ·m−2 was necessary for complete photoreversion of CPD. An aliquot of 100 ng of the treated plasmid was then digested with DenV protein to check for the completion of the photoreversion. After photoreversion of CPD, remaining photolesions (essentially (6-4)PP) were detected by UvrABC endonuclease. Briefly, DNA (125 ng) was incubated 5 min at 37 °C with 0.75 pmol of UvrA protein and 1.8 pmol of UvrB protein in reaction buffer (50 mm Tris-HCl, pH 7.5, 10 mmMgCl2, 85 mm KCl, 1 mmdithiothreitol, and 2 mm ATP). Then, 0.75 pmol of UvrC protein was added for 30 min at 37 °C and the reaction was stopped by adding 0.05% SDS and heating for 5 min at 65 °C. The samples were then electrophoresed, stained, photographed, and analyzed as described above. The number of (6-4)PP per kbp could therefore be established as a function of the UVC doses, which were chosen to produce less than one (6-4)PP per plasmid molecule. At each step a control reaction was performed on irradiated or unirradiated plasmid DNA in the absence of the protein. In the case of UvrABC digestion, a mock reaction was also carried out in the presence of only two proteins.
In parallel, (6-4)PP were also detected in UVC-irradiated plasmid by monoclonal 64M-2 antibody by using the IDB assay, as described above. Similarly, the luminescence intensity obtained with 64M-2 antibody was determined as a function of the UVC dose, and a calibration curve (luminescence intensity as a function of (6-4)PP/kbp) was established.
Quantitative Determination of DewarPP by Photoreversion of (6-4)PP
Genomic DNA from cells exposed to UVB radiation at doses ranging from 2 to 10 kJ·m−2 was used as substrate. Triplicates of 500 ng of DNA/dose were then subjected to IDB using 64M-2 and DEM-1 antibodies. The exact number of initial (6-4)PP per kbp was calculated from the calibration curve obtained using the 64M-2 antibody. In order to convert (6-4)PP into their DewarPP isomers, the remaining DNA was re-irradiated with a single dose (450 kJ·m−2) of UVA radiation (Osram Ultramed 400-W gas discharge lamp without cut-off filters; thus, at least 0.3% of the emitted photons were in the range of 295–316 nm; see above). This DNA was again subjected to IDB with the two antibodies. The number of (6-4)PP that disappeared following isomerization was deduced from the calibration curve for the 64M-2 antibody luminescence signal. Since one DewarPP is assumed to result from the photoisomerization of one (6-4)PP at wavelengths around 320 nm (
), the number of (6-4)PP that disappeared during the UVA re-irradiation step corresponded to the number of DewarPP produced. For each initial UVB dose, the increase in the luminescence intensity obtained with the DEM-1 antibody after the photoisomerization process was correlated with the number of DewarPP per kbp, thus establishing a calibration curve for the DEM-1 luminescence signal.
For cell survival determination, the CHO repair-proficient and deficient cells were irradiated by UVB radiation (0–5 kJ·m−2) or SSL (0–306 kJ·m−2) in PBS in 60-mm Petri dishes as described above, washed with PBS and trypsinized. Aliquots of 102 to 106 cells were seeded in triplicate into 60- or 100-mm dishes. After 10 days of incubation in growth medium, colonies were stained with methylene blue in methanol and counted. Survival was calculated as the ratio of the cloning efficiencies of irradiated over unirradiated cells × 100. The surviving fraction was plotted versus the amount of each of the three photoproducts. To calculate the correlation coefficient, all the experimental points obtained with each source of irradiation were considered as a single population, and the corresponding linear regression was determined. The correlation between damage induction and cell survival was considered as statistically significant whenp < 0.05 (from correlation table, for a risk of first order equal to 0.05 and a degree of freedom = n − 2; Ref.
We first determined the yield of formation of the three bipyrimidine photoproducts, CPD, (6-4)PP, and DewarPP in cellular DNA irradiated with the different UV components of solar radiation. NER-deficient CHO cells were irradiated with UVC, UVB, UVA radiation, and SSL. Bipyrimidine photoproducts were revealed by using the IDB technique with TDM-2, 64M-2, and DEM-1 antibodies, which recognize CPD, (6-4)PP, and DewarPP, respectively (
), as described under “Experimental Procedures.” To establish the relationship between the antibody binding site and the absolute amount of photoproduct formed, a calibration procedure was developed.
Quantification of the Bipyrimidine Photoproducts
Antibody luminescence intensity was calibrated with respect to the absolute amount of CPD and (6-4)PP on the irradiated plasmid DNA according to the following protocol. Various amounts of CPD and (6-4)PP were produced by irradiating supercoiled plasmid DNA with various doses of UV irradiation (UVC or UVB). The number of CPD or (6-4)PP per plasmid molecule was determined as a function of the dose, using purified, specific DNA repair enzymes and a plasmid relaxation assay. In parallel, photoproducts were detected by using specific TDM-2 and 64M-2 antibodies in the IDB assay, and the relative luminescence intensity was determined as a function of the dose. These two determinations allowed us to establish a linear relationship between the luminescence signal and the number of CPD or (6-4)PP per kbp of DNA. The calibration curves were then used to calculate the amount of CPD or (6-4)PP in the genomic DNA of irradiated cells.
To calibrate the luminescence signal of the TDM-2 antibody, plasmid DNA was UVB-irradiated and the number of CPD was determined following digestion with T4 DenV protein in the plasmid relaxation assay (data not shown). The luminescence signal induced by IDB increased as a function of the UVB dose. A calibration curve of the luminescence intensity as a function of CPD per kbp is given in Fig.1A.
Since a specific enzyme for (6-4)PP was not available, a two-step protocol was employed to quantify this photolesion. We used the UvrABC endonucleolytic complex from E. coli, which recognizes a number of different lesions, including the UV-induced photoproducts, CPD, (6-4)PP, and DewarPP (
). Prior to incision by the UvrABC complex, CPD were photoreversed by E. coli photolyase plus light (365 nm). To avoid the presence of DewarPP in the DNA substrate, UVC radiation was used as the damaging agent and the enzymatic reactions were performed immediately after UVC exposure of the plasmid. As a matter of fact, the photoreactivating light by itself did not generate a detectable amount of (6-4)PP or DewarPP. Other UV photolesions, i.e. monomeric cytosine hydrates, which can be recognized by UvrABC proteins, are at least 100 times less frequent than CPD and, thus, were not expected to interfere with the quantification of the dimeric photoproducts at the doses used (see Ref.
for review). The complete photoreversion of CPD was verified using the T4 DenV protein (data not shown). In addition, (6-4)PP were detected by IDB on the same irradiated plasmid DNA samples, allowing us to establish the calibration dose-response curve for the 64M-2 antibody luminescence signal (Fig. 1B).
The procedure for quantifying DewarPP is described in Fig.2. It is based on the photoisomerization of (6-4)PP into DewarPP by radiation around 320 nm (
), we used for photoconversion an unfiltered broad-band UVA source whose emitted radiation was composed of at least 15% of the 320–340 nm wavelengths. Genomic DNA from CHO cells was irradiated at various doses of UVB, and an aliquot of the irradiated DNA was subjected to a first IDB analysis with 64M-2 and DEM-1 antibodies. The remaining DNA was subsequently re-irradiated with a UVA source for photoconversion (see “Experimental Procedures”) and subjected to a second IDB analysis with the same antibodies. According to the calibration curve for the 64M-2 antibody shown in Fig. 1B, the decrease in luminescence obtained with the 64M-2 antibody after photoisomerization is a measure of the number of photoisomerized (6-4)PP. Assuming a quantitative conversion of (6-4)PP into DewarPP by the photoisomerization process (
), the increase in luminescence obtained with the DEM-1 antibody can be expressed as the number of DewarPP produced or (6-4)PP lost by photoisomerization. Since several UVB doses were employed, the calibration curve for luminescence signal obtained with DEM-1 antibodies could be established (Fig.1C). It must be noted that the UVA dose for photoisomerization was not sufficient to elicit the formation of either (6-4)PP or DewarPP (see below).
The CPD induction obtained with the four types of radiation is displayed in Fig. 3. The saturation observed for UVB and to a lesser extent for UVC is likely to be due to photoreversion of this lesion. Results in TableI indicate that the rate of CPD formation by UVB radiation was 100 times lower than for UVC, and about 1000 times higher than for UVA. When only the UV portion (6.8% of the total emitted energy) is taken into account in the determination of the yield of CPD, SSL was 100 times less efficient than UVB in producing CPD. Such yields of CPD in the DNA of irradiated cells were very similar to those obtained on isolated plasmid DNA irradiated under the same conditions (
The yields of photoproducts were calculated taking into account only the UV region (6.8%) of the whole emitted radiation.
2.1 × 10−6
3.7 × 10−7
1.3 × 10−7
The yields were calculated from the slope of the regression lines given in Figs. Figure 3, Figure 4, Figure 5. For UVC- and UVB-induced CPD and (6-4)PP, only the linear parts of the curves were used. ND, not detectable at the physiological doses used.
1-a The yields of photoproducts were calculated taking into account only the UV region (6.8%) of the whole emitted radiation.
Since the formation of CPD by UVA radiation has been a matter of debate for over 20 years, we calculated the relative efficiency of different regions of the UV spectrum for the production of this damage using the action spectrum reported by Kielbassa et al. (
). TableII shows that the majority of CPD (86.7%) formed with our UVA irradiation device were produced at wavelengths above 345 nm and that the very weak UVB emission did not contribute significantly to their formation (5 × 10−4% of total emission). Furthermore, since a WG345 Schott filter was used to eliminate UVC and UVB wavelengths, the contribution of UVA2 wavelengths (320–345 nm) to the formation of this photolesion was low (1.8%). In contrast, among CPD produced by SSL, 89% were produced by the UVB component and 11% in the UVA range that comprised 8% by UVA2 wavelengths (320–345 nm). As expected, virtually all the CPD (99.6%) induced by the UVB source were formed by the absorption of photons in the 290–320-nm range of the spectrum.
Table IIPhysical characteristics (% irradiance) and calculated biological effectiveness (% CPD yield) of the UV sources used
UVB (290–320 nm)
CPD yield (%)
UVA (320–400 nm)
CPD yield (%)
UVA2 (320–345 nm)
CPD yield (%)
UVA1 (345–400 nm)
CPD yield (%)
The details of the calculations have been extensively described by Douki et al. (
). Values of spectral irradiances for the various intervals are based on measurements of the spectral irradiances in 1-nm steps for the given UV source. CPD spectral effectiveness was calculated on the basis of the action spectrum for CPD formation from Ref.
These data clearly demonstrate that the production of CPD not only depends on the energy distribution and the dose received by the sample, but also on the interaction between DNA and each type of photon. We next calculated an enhancement factor for the relative efficiency of the 290–320 nm wavelengths in producing CPD. TableIII gives the ratio of irradiance in the UVB region to irradiance in the UVA range for the different lamps and for comparison, the ratio of CPD production in the UVB and UVA regions. Interestingly, the very low contribution of UVB energy in the filtered emission spectrum of our UVA lamp (the ratio UVB/UVA equals 7.7.10−5) leads to a relatively high efficiency of CPD formation (a ratio UVB/UVA of 0.13). In other words, the rare UVB photons emitted by our UVA source are far more efficient than the abundant UVA photons in producing CPD. Indeed, an enhancement factor of 1695 was calculated. This enhancement factor is only 406 and 169 for the UVB source and the solar simulator, respectively. In addition, it appears that even though the filtered emission of our UVB lamp is composed of about 60% of UVA wavelengths, the biological effect studied here is purely a UVB-induced effect (Tables II and III). Altogether, these calculations substantiate the importance of the spectral distribution of a given source in the manifestation of a biological effect.
Table IIIRelative spectral effectiveness for CPD formation
Ratio of CPD yields due to UVBversus UVA region. The CPD yields were calculated as the products of the measured spectral distribution of the given source and the CPD's spectral sensitivity for the interval considered.
The enhancement factor represents the ratios of CPD yields normalized by the irradiance ratios. It highlights the relative importance of the UVB component of a lamp's emission spectrum for CPD induction.
3-a UVB, UVA, and SSL correspond to the UV sources used in the experiments, and as specified under “Experimental Procedures.”
3-b Ratio of the sum of irradiances between 290 and 320 nm (UVB) to those between 320 and 400 nm (UVA) for the different UV sources.
3-c Ratio of CPD yields due to UVBversus UVA region. The CPD yields were calculated as the products of the measured spectral distribution of the given source and the CPD's spectral sensitivity for the interval considered.
3-d The enhancement factor represents the ratios of CPD yields normalized by the irradiance ratios. It highlights the relative importance of the UVB component of a lamp's emission spectrum for CPD induction.
Under our conditions, (6-4)PP were readily produced by UVC, UVB, and SSL (Fig.4) but not by UVA radiation. Fig.5 shows the linear induction of DewarPP as a function of UVB and SSL doses. DewarPP was not detected after UVC or UVA exposure. Using the same UVA doses as those used for eliciting CPD, both (6-4)PP and DewarPP were detected in trace amounts when a more sensitive IDB method was employed (data not shown). For reliable quantification, very high, non-biologically relevant UVA doses would have been necessary. Table I indicates that UVB is about 2 × 10−2 times less effective than UVC and about 60 times more efficient than SSL in producing (6-4)PP, but only 20 times more efficient than SSL for DewarPP formation.
Relative Distribution of Bipyrimidine Photoproducts at Different UV Wavelengths
Table I demonstrates that the relative induction of the three bipyrimidine photolesions varies depending on the type of radiation. For example, the ratio of CPD to (6-4)PP is about 4–5 for UVC and about 8 and 6 for UVB and SSL respectively. This is in favor of a partial photoisomerization of (6-4)PP into DewarPP by long UV wavelengths. Indeed, the ratio of (6-4)PP to DewarPP is about 8 for UVB and 3 for SSL. These data demonstrate that DewarPP are relatively more frequent in cellular DNA upon SSL exposure than upon UVB irradiation. Consequently, this last photolesion should be taken into account when considering the genotoxicity of sunlight.
Repair Rates of CPD, (6-4)PP, and DewarPP
In order to compare the rates of removal of CPD, (6-4)PP, and DewarPP from the genomic DNA of NER-proficient CHO cells exposed to UVB, UVA radiation, or SSL, the photoproducts were measured at various times after exposure by the IDB assay. In general, a unique dose of radiation was chosen to ensure the detection of all three photolesions and their progressive removal in the same cellular population. Cells received 1 kJ·m−2 UVB radiation, 1000 kJ·m−2 UVA radiation, and 306 kJ·m−2 SSL (corresponding to 3.5–4 h of sun exposure at midday in summer in term of CPD induction (Ref.
)). These treatments led to survival rates of 15%, 30%, and 10%, respectively. In order to provide a sufficient initial amount of DewarPP and to follow the fate of these lesions, a dose of 5 kJ·m−2 UVB radiation was also used.
Fig. 6A shows that CPD induced by UVA were slowly and inefficiently removed in CHO cells. At similar survival levels (10–30%), their repair rates were not significantly different from the rates of CPD produced by UVB or SSL (Fig.6A). However, UVB radiation leads to a large excess of CPD (
) and are rapidly repaired) does not interfere with the repair of CPD. The CPD induction is 0.22 and 0.25 per kbp for UVB and UVA irradiation, respectively (Table IV), which corresponds to the level of CPD obtained with about 15 J·m−2UVC. Between 51% and 76% of CPD were still present 24 h after cell exposure to the different UV sources. This high residual level of CPD is probably not due to saturation of NER pathways since (6-4)PP were removed quickly and efficiently under the same conditions (see below).
Table IVNumber of bipyrimidine photoproducts (per kbp) immediately after irradiation and at t = 24 h repair time
Quantitatively, the (6-4)PP lesion represents the second major type of UV-induced photolesion (Tables I and IV). Fig. 6B shows that 90% of (6-4)PP were removed 6 h after exposure and that removal was almost complete after 24 h. The repair kinetics of cells exposed to UVB and SSL are undistinguishable and fit a first-order exponential decay function. The half-life of (6-4)PP is between 2 and 2.5 h. These data confirm that (6-4)PP are rapidly and efficiently eliminated compared with the poor repair of CPD (
The capacity of CHO cells to remove DewarPP was also examined after exposure to UVB and SSL. The two repair kinetics shown in Fig.6C are not statistically different (p > 0.05), and resemble those of CPD. The repair rate of DewarPP follows a two-step process. The repair rates of CPD and DewarPP (Fig. 6,A and C) do not exhibit significant differences.
Relationship between Induction of Bipyrimidine Photoproducts and Cytotoxicity
Since the role of each of the three bipyrimidine photoproducts in UV-induced lethality may vary depending on the cell type (
), we examined the response of the repair-proficient CHO AT3–2 cells. Survival, assayed by determining clonogenic efficiencies, was plotted as a function of the amount of CPD, (6-4)PP, and DewarPP produced by UVB radiation and SSL, which produce significant amounts of all three lesions (Fig. 7, A,B, and C). Insets in Fig. 7 show that cell killing was well correlated with the formation of CPD (r = −0.71, p < 0.05) and also of (6-4)PP (r = −0.68, p < 0.05). Since we observed that 90% of (6-4)PP were removed within 6 h after UV exposure, the contribution of this lesion to the loss of the clonogenic efficiency of the irradiated cells is questionable. This discrepancy may be due to the parallel induction of CPD and (6-4)PP (
). However, high correlation coefficients for CPD and (6-4)PP were observed while investigating the role of these lesions in the lethality of the repair-deficient UVL9 cell line upon UVB and SSL irradiation. Ther values were −0.85 and −0.77 (p < 0.05) for CPD and (6-4)PP, respectively. In the absence of repair, the lethality at a given dose of radiation depends on the intrinsic cytotoxicity of the lesion and its amount. Consequently, both lesions seem to be cytotoxic.
There was no statistically significant correlation between the induction of DewarPP and the lethality of irradiated NER+cells (r = −0.25, p > 0.05), whereas a statistically significant but loose correlation was observed in irradiated NER− cells (r = − 0.64,p = 0.05). This weak (or null for NER+) correlation may not be surprising considering the low production of DewarPP in comparison to CPD (Table I), at least in the case of UVB exposure (ratio 1:70). DewarPP formation depends on the formation of (6-4)PP, but the action spectra of these lesions are very different (see in the discussion). The LD10 values (amount of damage leading to 10% survival, or lethal dose) calculated for SSL-irradiated NER− cells are 0.12 CPD, 0.015 (6-4)PP, and 0.006 DewarPP per kbp. This apparently suggests that the former photolesion is the least cytotoxic. However, the ratios between the LD10 values for both the NER+ and NER− cell lines follow exactly the same ratios for the induction of these lesions upon SSL exposure. It seems that the lethality is governed by the extent of each lesion and not only by the intrinsic cytotoxicity of these lesions. These data clearly demonstrate that quantitative analysis of damage induction is important for our understanding of how each photoproduct contributes to biological repercussions of solar UV radiation.
We quantitatively analyzed the induction of the three bipyrimidine photoproducts, i.e. CPD, (6-4)PP, and DewarPP, in CHO cells exposed to the different UV components of solar radiation. The present study provides information about the absolute amounts of three photolesions produced by various regions of the UV spectrum and about how efficiently these types of DNA damage are produced by certain wavelengths. Furthermore, the repair rate of each of these photolesions was also examined at the genomic level.
The yield of each of the bipyrimidine photoproducts was assessed using previously described monoclonal antibodies TDM-2, 64M-2, and DEM-1 (
) and expectedly TC DewarPP predominate among these types of photoproducts and are ideal substrates for the 64M-2 and DEM-1 antibodies, respectively, but these antibodies can also recognize TT derivatives (
), we found that they were sensitive enough to detect the photoproducts induced at physiologically relevant doses of radiation, i.e. those that gave low toxicity in cultured cells or those of the normal human environment. However, in contrast to other studies (
), the detection method reported here allowed us to examine for the induction of all three bipyrimidine photoproducts within the same dose range and for each of the UV sources. Importantly, the three photolesions could be quantified in the same irradiated DNA sample.
CPD and UVA Radiation: Role in Solar Mutagenesis
The yields of CPD formed by UVC and UVB radiation shown in Table I are in excellent agreement with those reported in the literature (
). We further demonstrated that CPD are produced mostly by UVA1 wavelengths in the interval between 345–400 nm, which corresponds to the spectral irradiance of our UVA source (Table II). The yield of CPD produced by UVA follows exactly the DNA absorption at these wavelengths (
). Thus, compared with UVC, 105 higher doses of UVA are necessary to produce the same amount of CPD. Despite this difference, these UVA doses are still biologically relevant since they correspond to a few hours of sunlight exposure at zenith in summer (
Interestingly, we show in Table II that about 10% of CPD induced by SSL irradiation was formed by the UVA component. This is an additional argument indicating that UVA radiation may partly contribute to the DNA damage induced by sunlight, and consequently, UVA may participate to solar light mutagenesis more than previously expected. Indeed, a majority of the mutations induced by sunlight and found in skin tumors (
) are GC to AT transitions and tandem double CC to TT mutations, all of which can be attributed to bipyrimidine photoproducts (mainly to CPD in repair-proficient non-XP individuals). Furthermore, the mutational specificity of UVA comprises a large amount of GC to AT transition located at bipyrimidine sites (
(no value available either for UVB or DewarPP). We observed that (6-4)PP were 200 times less efficiently formed by UVB than by UVC, which is in accordance with the published action spectrum for the induction of CPD and (6-4)PP (
The comparative rates of the formation of CPD, (6-4)PP and DewarPP were 1:0.25 for UVC, 1:0.12:0.014 for UVB, and 1:0.18:0.06 for SSL. This illustrates that the distribution of the different classes of photolesion varies depending on the UV source, as already observed (
), and clearly demonstrates that the biological effects of sunlight cannot be deduced solely on the basis of studies with UVC. Furthermore, DewarPP was formed with high yield, making it the third major photolesion induced by SSL (about 10 times more frequent than 8-oxodGuo (Ref.
), although the repair mechanism of DewarPP has not been elucidated in mammalian cells. We first observed that the rate of removal of all of these photolesions did not depend on the source of irradiation. This indicates that the presence of relatively large amounts of single strand breaks and 8-oxodGuo or other oxidized bases in UVA- or SSL- irradiated cells (
) does not impair the processing of bipyrimidine photoproducts by the cellular machinery. On the other hand, SSB and oxidative damage are processed by repair pathways other than NER, i.e. base excision repair (
). The rate of removal of base damage is similar to that observed for (6-4)PP.
The repair rate of CPD has been extensively studied in both rodent and human cells. We observed that, at most, 20% of CPD was eliminated after a repair time of 6 h and 25–50% after 24 h. This is within the range of published repair times for global genomic repair in cultured rodent cells or rodent skin under similar conditions of irradiation (
). We found that the repair rate of DewarPP was very similar to that of CPD, but much lower than that of (6-4)PP. After a repair time of 6 h, CHO cells did not remove more than 15–30% of DewarPP, and the repair efficiency was no more than 20–50% after 24 h. Using radioimmunoassay, Rosenstein and Mitchell (
) observed a slightly faster repair rate of DewarPP in human skin fibroblasts than in the rodent cells described here. This is in agreement with reported differences in the repair kinetics between the two species (
A conformational basis for the differences in the repair rates of the three bipyrimidine photoproducts may be tentatively considered. Most likely, the recognition of the photolesion by NER proteins, which bind to sites of DNA damage, is the rate-limiting step in the repair process. Most of the investigations report greater DNA unwinding by (6-4)PP and DewarPP than by cis-syn CPD, and a smaller distortion of DNA by DewarPP than by (6-4)PP (
), and at the same rate and extent as CPD in CHO cells (Fig. 6, A and C). The fast removal of (6-4)PP by GGR may be due to the tight binding of XPC/hHR23B to the highly distorted region carrying (6-4)PP, whereas UV-DDB activity, which is absent from certain xeroderma pigmentosum E groups and hamster cells (
), may be required for targeting CPD and DewarPP for GGR. It is tempting to speculate that DewarPP, as well as CPD, are efficiently repaired by TCR but not by GGR, and that (6-4)PP are mainly repaired by GGR. Another possibility could be that, due to the distortion of the DNA helix at the DewarPP site, the DewarPP is still able to bind XPC/hHR23B and other associated proteins, but with lower affinity compared with the affinity of these proteins for (6-4)PP. Thus, according to this hypothesis, the DewarPP lesion would be repaired by GGR, but with slow kinetics.
Biological Relevance of DewarPP
The relative contribution of CPD and (6-4)PP to UVC- and solar UV- induced mutagenesis has been a matter of debate for almost 20 years (
). In short, (6-4)PP and DewarPP are more mutagenic than the corresponding CPD. In eucaryotic cells, it is likely that these lesions produce mutations via error-prone bypass by DNA polymerase ζ lacking proofreading activity (see Ref.
As our results show that DewarPP are no longer quantitatively negligible compared with CPD lesions, especially at TC sites, it is legitimate to question about biological contribution of this type of lesion to solar UV-mutagenesis. Considering the fast and efficient repair of (6-4)PP and the slow and inefficient removal of CPD and DewarPP, most of the mutations that we observed in the repair-proficient CHO cells after SSL (
), it is tempting to suggest that at least part of the CC to TT double events occurring after irradiation of CHO cells with SSL are due to DewarPP. Interestingly, such mutational events have also been recovered in the mutated p53 gene of skin tumors from normal and XP individuals exposed to sunlight (
). The determination of the yield of formation of these photolesions by the different UV components of solar light should help to estimate the risks associated with exposure to environmental sunlight or artificial sources such as sunbeds.
We are grateful to Yvette Rolland for excellent technical assistance and to Dr. P. Hughes and Y. Cohen for critical reading of the manuscript.
Monographs on the Evaluation of Carcinogenic Risks to Human: Solar and UV Radiation. IARC 55,