Thomas M.
Rünger
*,
Benyamin
Farahvash
,
Zsofia
Hatvani
and
Adam
Rees
Boston University School of Medicine, Department of Dermatology, 609 Albany Street, Boston, MA 02118, USA. E-mail: truenger@bu.edu; Fax: 001 617 638 5515; Tel: 001 617 638 5551
First published on 18th October 2011
Mechanisms of UVA-mutagenesis remain a matter of debate. Earlier described higher rates of mutation formation per pyrimidine dimer with UVA than with UVB and other evidence suggested that a non-pyrimidine dimer-type of DNA damage contributes more to UVA- than to UVB-mutagenesis. However, more recently published data on the spectra of UVA-induced mutations in primary human skin cells and in mice suggest that pyrimidine dimers are the most common type of DNA damage-inducing mutations not only with UVB, but also with UVA. As this rebuts a prominent role of non-dimer type of DNA damage in UVA-mutagenesis, we hypothesized that the higher mutation rate at UVA-induced pyrimidine dimers, as compared to UVB-induced ones, is caused by differences in the way UVA- and UVB-exposed cells process DNA damage. Therefore, we here compared cell cycle regulation, DNA repair, and apoptosis in primary human fibroblasts following UVB- and UVA-irradiation, using the same physiologic and roughly equimutagenic doses (100–300 J m−2 UVB, 100–300 kJ m−2 UVA) we have used previously for mutagenesis experiments with the same type of cells. ELISAs for the detection of pyrimidine dimers confirmed that much fewer dimers were formed with these doses of UVA, as compared to UVB. We found that cell cycle arrests (intra-S, G1/S, G2/M), mediated at least in part by activation of p53 and p95, are much more prominent and long-lasting with UVB than with UVA. In contrast, no prominent differences were found between UVA and UVB for other anti-mutagenic cellular responses (DNA repair, apoptosis). Our data suggest that less effective anti-mutagenic cellular responses, in particular different and shorter-lived cell cycle arrests, render pyrimidine dimers induced by UVA more mutagenic than pyrimidine dimers induced by UVB.
It is well established that solar ultraviolet light produces a variety of different types of DNA damage that in turn may result in mutation formation and ultimately photocarcinogenesis.2DNA damage profiles are dependent on wavelength. UVB generates DNA photoproducts (cis–syn cyclobutane pyrimidine dimers (CPDs), pyrimidine (6–4) pyrimidone photoproducts (6-4PPs), and Dewar valence isomers (DewPPs)) through direct absorption by DNA. While UVA is less energetic than UVB, it is also far more abundant than UVB and constitutes the majority of solar-available UV. While UVA is much less absorbed by DNA than UVB, it still generates pyrimidine dimers.2–11
Similar to UVB-induced mutations, mutations induced by UVA have also been shown to be predominantely C → T transitions, both in vitro and in vivo.11–14 As C → T transitions are mostly formed at sites of pyrimidine dimers, this indicates that pyrimidine dimers are the most common pre-mutagenic DNA lesions not only in UVB-, but also in UVA-mutagenesis.
In the past, findings of a higher yield of mutations per pyrimidine dimer with UVA, as compared to UVB,15 and of a second peak of skin cancer formation in the UVA2 range (340–400 nm) without a peak of pyrimidine dimer formation 16,17 were interpreted to support a contribution of a non-pyrimidine dimer-type DNA damage to UVA mutagenesis, including e.g. oxidative DNA base damage or DNA double strand breaks. However, the above mentioned predominance of C → T transitions with UVA is not consistent with that explanation, and we have proposed an alternative explanation:11,12
The vast majority of UV-induced DNA lesions are not translated into a mutation, because cells are equipped with a variety of anti-mutagenic mechanisms that prevent mutation formation at sites of DNA damage or limit survival of severely damaged cells. These antimutagenic cellular DNA damage responses include DNA repair (some of which is inducible18,19), cell cycle arrest, and apoptosis. We hypothesize that a UVA-induced pyrimidine dimer may be more mutagenic than a UVB-induced pyrimidine dimer, because UVA induced antimutagenic cellular responses are not as effective as those induced by UVB. And indeed, p53, a central regulator of antimutagenic cellular DNA damage responses, was found to be much less activated by UVA, as compared to UVB, when comparing equimutagenic doses.12
In order to further substantiate this hypothesis, we are describing here in much further detail the cellular DNA damage responses to equimutagenic doses of UVA and UVB, both by function (cell cycle arrest, DNA repair, and apoptosis) and by related DNA damage signaling. Toward that goal we have used the same model system we used for our studies with the hprt-mutagenesis assay12 (primary cultured human fibroblasts) and the same range of UVA and UVB doses. This avoids the problem that investigations from different laboratories can often not be directly compared because of the use of different model systems (mouse vs. human; transformed cellsvs. primary cells; cultured cellsvs. in vivo studies), because of different emission spectra of UVB and UVA lamps, and because of poorly standardized dosimetry. The use of primary, non-transformed cells assures that DNA damage responses are not altered, as is often the case in transformed cell lines, e.g. through mutation of p53. In addition, human cells are better suited for mutagenesis studies than e.g. rodent cells, as the latter are characterized by a reduced efficiency of DNA repair, which is an important DNA damage response following UV-exposure. In addition, the UVB- and UVA-doses used here can easily be reached by exposure to natural sunlight and are therefore clinically relevant. Despite all these advantages of our model system, it is nevertheless not ideal, because in humans, exposure to ultraviolet light is not known to induce fibroblast-derived malignancies and because cultured cells grow in an artificial environment that only poorly represents the in vivo situation.
A solar simulator (LH 153, Kratos Analytical, Ramsey, NJ) was used for UVB irradiations. The tissue culture lid removes contaminating UVC and only slightly reduces irradiance of the shorter wavelengths of the UVB range. The resulting emission spectrum was published by Werninghaus et al.21 The UVA emitted by this source at the maximum UVB doses used was less than 2.5 kJ m−2, consistent with previous reports that solar simulators often emit insufficient longwave UVA to adequately simulate natural sunlight.22,23
Radiometric measurements were performed for each experiment. For dosimetry, an IL-1700 Research Radiometer (International Light, MA) was used, equipped with a UVB-sensor (SED 240) or a UVA-sensor (SEF 015). A 137Cs gamma irradiator was used to expose cells to 10 Gy of gamma radiation.
Table 1 lists the doses of UVA and UVB used here and compares them to the range of UVA- and UVB-doses we had used previously in our mutagenesis experiments with the same type of cells and identical UV sources and dosimetry.12 It demonstrates that the UVA and UVB conditions used here induce mutations at rates that are in the same order of magnitude (roughly equimutagenic). These doses of UVA and UVB also have similar effects on cell growth/cell yields, as previously reported.12
Doses used to study UV-induced damage responses | Mutation frequencies (from Kappes et al.12) | |
---|---|---|
UV dose | UV dose | mutation frequency (number of mutants/106 colony forming cells) |
UVA | ||
100 kJ m−2 | 150 kJ m−2 | 5.9 |
200 kJ m−2 | 200 kJ m−2 | 70.4 |
300 kJ m−2 | 300 kJ m−2 | 136.4 |
UVB | ||
100 J m−2 | 70 J m−2 | 8.7 |
200 J m−2 | 200 J m−2 | 20.4 |
300 J m−2 | 400 J m−2 | 62.3 |
600 J m−2 | 97.7 |
0.25 μg of treated or untreated plasmid was transfected into exponentially growing cells in 35 mm tissue culture dishes at 50% confluence, using Lipofectamine Plus (Gibco/Invitrogen) and the manufacturer's protocol. Triplicate samples were used for each cell line and each dose of UV. After 2 days, luciferase activity was determined using the luciferase reagent kit and luminometer (Promega). In order to compensate for variations of transfection rates, plasmid replication or condition of the cells used, triplicate parallel reference samples of unirradiated plasmid were used in every experiment. The relative luciferase activity with the UV-irradiated samples, which reflects repair of DNA photoproducts on the plasmids by the host cells (host cell reactivation), was calculated in percent of the parallel control samples transfected with unirradiated plasmid. While the read-out of this assay depends on the transcription of the repaired reporter gene, defects in global-genome repair, as e.g. in cells from patients with xeroderma pigmentosum complementation group C, are also detectable with this assay.28,31 This indicates that this assay is not limited to the measurement of transcription-coupled repair, and does not distinguish between global genome and transcription-coupled repair. The unpaired Student t-test was used to test for differences.
Here, we used this assay to investigate the effects of UVA and UVB on DNA repair capacity. For that, cells were pre-irradiated with different doses of UVA (50, 100, or 200 kJ m−2) or UVB (50, 100, or 200 J m−2) 24 h before transfection of the damaged plasmids for the repair assay. This permits investigation of UVA- and UVB- effects on DNA repair capacity separate from UV-induced toxicity, which is much higher for UVA than for UVB when comparing doses that generate the same amount of pyrimidine dimers.
Fig. 1 UVB induces more cyclobutane pyrimidine dimers (CPDs) and 6–4 photoproducts (6,4-PPs) than UVA, when using roughly equimutagenic doses. Shown are means of quadruple samples ± standard deviation. |
Fig. 2 Bivariate BrdU/PI flow-cytometric analysis of cell cycle progression in non-synchronized cells reveals profound differences in cell cycle profiles following irradiation with roughly equimutagenic doses of UVB and UVA. Panel A depicts the attribution of cells to the different phase of the cells cycle. The cell cycle profiles at various time points after irradiation with sham, UVB (300 J m−2), or UVA (300 kJ m−2) are shown in panel B. The numerical analysis of the percentage of cells in the active S, in the arrested S, and in the G2/M phase are shown in panels C, D, and E respectively (averages of duplicate samples ± standard deviation). |
UVB-irradiated cells also demonstrate a long-lasting G2/M arrest (2–48 h), whereas a mild G2/M arrest with UVA is observed only at the 2 and 8 h time points (Fig.2, panels B and E).
A G1/S arrest was not observed with non-synchronized fibroblasts (Fig. 2, analysis of cells in G1/G0 not shown). In order to study the failure of UVB- or UVA-irradiated cells to enter S-phase (G1/S-arrest), we synchronized cells by serum starvation and irradiated them with UVB or with UVA just prior to release from growth arrest by re-feeding with complete medium (10% calf serum). By 24 h after irradiations and release from serum starvation, sham- and low-dose (100 J m−2) UVB-irradiated fibroblasts readily entered S-phase, whereas the intermediate dose of UVB (200 J m−2) significantly reduced the fraction of cells entering S-phase by that time (partial S-phase arrest; Fig. 3 A). The highest dose of UVB (300 J m−2) completely blocked S-phase entry (complete S-phase arrest; Fig. 3 A). In contrast, none of the UVA doses affected the fraction of cells with S-phase entry, indicating a complete lack of a G1/S-arrest (Fig. 3 A).
Fig. 3 Dose-dependent block of S-phase entry only after UVB-, but not after UVA-irradiation. Cells were synchronized by serum starvation and released from serum starvation immediately after irradiation. Cell cycle profiles show efficient S-phase entry 24 h after sham, low-dose UVB (100 J m−2), and all doses of UVA (100, 200, and 300 kJ m−2; panel A; shown are representative examples of duplicates, the numbers are averages of the percentage of cells in active S-phase ± standard deviation). The intermediate dose of UVB (200 J m−2) delayed and reduced S-phase entry, and the highest dose of UVB (300 J m−2) blocked S-phase entry completely. Panel B describes the hour-by-hour S-phase entry of sham-irradiated cells and that S-phase entry after release from serum starvation is first detectable after 13 h. This was not delayed after irradiation with the highest UVA-dose, as the S-phase entry was also observed at the same time and in an almost identical fraction of cells. |
In order to determine whether UVA delays S-phase entry temporarily, which may not be detectable anymore 24 h after release from serum starvation, we observed S-phase entry of sham-irradiated fibroblasts hour-by-hour and determined the number of cells in active S-phase after UVA-irradiation exactly at the time point at which sham-irradiated cells started entering S-phase (13 h after release from serum starvation; Fig. 3 B). At that time point, the number of cells in active S-phase was almost identical after sham- and after UVA-irradiation. This indicates that UVA does not delay S-phase entry at all, not even by one hour (Fig. 3 B).
The cell cycle data (Fig. 2 and 3) demonstrate that the responses of fibroblasts to equimutagenic doses of UVA and UVB are not only quantitatively different (more pronounced cell cycle arrests with UVB than with UVA), but also qualitatively different (e.g. an early and very transient intra-S-arrest only with UVA, whereas the intra-S-arrest with UVB is observed later and persists much longer; prominent G1/S-arrest with higher doses of UVB, but no G1/S-arrest with UVA).
The rapid slowing of DNA replication after UVB with the sustained intra-S arrest may be mediated by activation of p95, because phosphorylation of p95 can be detected as early as 15 min after UVB irradiation (Fig. 4 A). The rapid, but short-lived slowing of DNA replication with brief intra-S-arrest after UVA cannot be due to activation of p95, as phosphorylation of p95 was not detected after UVA (Fig. 4 A). It may be due to (temporary) replication stops at pyrimidine dimers. However, if this were the case, it should be more prominent with UVB, because UVB generates more pyrimidine dimers (as shown in Fig. 1). Brief replication stops at UVA-induced DNA lesions other than pyrimidine dimers, including oxidative base modifications or single-strand breaks, may therefore account for the brief and transient slowing of DNA replication early after UVA irradiation.
Fig. 4 Dose-dependent activation of p53 and p95 (NBS1) early after UVB, but not after UVA (panel A). Consistent with much shorter-lived and less prominent activation of p53 at the early time points with UVA, as shown in panel A, and as reported previously for later time points,12 the induction of XPC (113 kD), which is known to be p53-dependent,33,34 is less prominent and shorter-lived after UVA irradiation, as compared to UVB-irradiation (panel B). |
Activation of ATM, an upstream regulator of p95, was detected with both, UVA and UVB, both at an early and at a later time point, 30 min and 4 h after irradiation (Fig. 5). However, this was only seen in a subset of cells and therefore appears to be cell cycle dependent. This is in contrast to ionizing radiation, with activates ATM in all cells (Fig. 5). The difference in p95 activation between UVA and UVB (no p95 activation with UVA, prominent activation with UVB) cannot be attributed to differences in phosphorylation of ATM at serine 1981, because UVA and UVB both appear to activate ATM equally. Other upstream kinases, e.g. ATR, or phosphorylation sites of ATM other than serine 1981 may be differentially activated by UVA and UVB and account for the difference in p95 activation.
Fig. 5 No difference in the activation of ATM (phosphorylation at serine 1981) by UVB (300 J m−2) or UVA (300 kJ m−2). Shown are representative examples of immunostainings for serine 1981–ATM nuclear foci (green; counterstaining with DAPI, blue, to mark nuclei; panel A) and the numerical analysis of the percent of cells with nuclear foci of activated ATM early (30 min) and later (4 h) after irradiation. One hundred cells in three different areas of the slide were counted; shown are averages ± standard deviation. Activation of ATM was only seen in a subset of cells, both after UVB and UVA, whereas ionizing radiation activated ATM in all cells. |
Activation of the G1/S-checkpoint is commonly thought to be mediated by p21via p53. Consistent with the activation of the G1/S checkpoint with UVB and the lack of it with UVA, we already reported a much weaker and less long lasting activation of p53 with UVA, as compared to UVB (2–48 h after irradiation12). Here, we add to this data by describing that p53 activation is already detectable one hour after irradiation with UVB, and confirm the lack of p53 activation with UVA (Fig. 4 A).
Fig. 6 No difference in how UVA and UVB affect DNA repair capacity of fibroblasts. Cells were irradiated with intermediate doses of UVA (200 kJ m−2) or UVB (200 J m−2; panel A), with low doses of UVB (50 or 100 J m−2; panel B), or with low doses of UVA (50 or 100 kJ m−2; panel C), and DNA repair capacity was determined 24 h later using a host cell reactivation assay with plasmids containing increasing amounts of pyrimidine dimers generated by increasing doses of UVB to the plasmid prior to transfection. The repair of the DNA damage on the plasmids by the host cells' DNA repair machinery (host cell reactivation) was measured 24 h after transfection (= 48 h after pretreatment of host cells with UVA or UVB) by measuring expression of the plasmid-encoded luciferase. Shown are averages of triplicates samples ± standard deviation. Pre-irradiation with the higher doses of UVA and UVB (panel A) reduced DNA repair capacity of the host cells, whereas pre-irradiation with the lower doses of UVB (panel B) or UVA (panel C) did not affect DNA repair capacity. |
Fig. 7 Both UVA and UVB induce apoptosis only in a small fraction of cells irradiated with the maximum doses used in this study. The fraction of cells with sub-G1 DNA content (apoptotic cells) was determined at several time points after UVA (300 kJ m−2) or UVB (300 J m−2) irradiation. Shown are averages of duplicate samples ± standard deviation. |
Using an experimental design very similar to ours (similar doses of UVA and UVB; primary human fibroblasts), de Laat et al.26 also described suppression of DNA replication following both UVA and UVB. Very similar to our results, this effect was also more long-lasting with UVB than with UVA. However, they did not specifically analyze the fraction of cells arrested in S-phase, where we found an early, but short-lived intra-S arrest after UVA, and a late, but persistent and more profound intra-S arrest after UVB. Very similar to our results, they also did not see dramatic differences in the percentage of cells in G0/G1 when using non-synchronized cells. They did not examine S-phase entry in synchronized cells, with which we here describe a clear G1/S arrest after UVB (but not after UVA). A more prominent and longer-lasting suppression of DNA replication by UVB, as compared to UVA, was also described in vivo in epidermal cells of irradiated hairless mice.38 Using transformed human or mouse embryonal fibroblasts, Girard et al.39 also described slowing of DNA replication/BrdU incorporation following UVA-irradiation, again very similar to our results. We hypothesize that in our experiments this slowing is not due to replication stops at pyrimidine dimers, because it is more prominent after UVA, as compared to UVB, while our UVA doses produce less pyrimidine dimers than our UVB doses. Instead, we hypothesize that the slowing of DNA replication may be due to brief replication stops at sites of non-pyrimidine dimer types of DNA damage, including e.g. oxidative base modifications and single-strand breaks. In line with this interpretation, Girard et al.39 describe a partial rescue of UVA-inhibited DNA replication by N-acetyl-L-cystein (NAC), a scavenger of reactive oxygen species, confirming that oxidative DNA damage is indeed at least in part responsible for the immediate slowing of DNA replication after UVA exposure. Such slowing of replication is different from a complete block of replication, and may be due to recruitment of translesional DNA polymerases for bypassing oxidative DNA damage, including e.g.8-oxo-guanine.40 Given the lack of p53- and p95-activation early after UVA-irradiation, we conclude from our data that the early slowing of DNA replication after UVA is not due to activation of cell cycle checkpoints. Again very much in line with this interpretation, Girard et al.39 describe that in their transformed cells, the slowing of DNA replication after UVA is independent of the ATM/ATR/Chk1/Chk2 pathways, which are upstream regulators of p53 and p95. Using the same model system (primary human fibroblasts) and the same UVA sources as in this report, we recently published evidence that UVA does not generate DNA double strand breaks.41 This excludes the possibility that DNA double strand breaks cause the observed slowing of DNA replication after UVA exposure. Finally, given that UVA has been suggested to generate more pyrimidine dimers than oxidative DNA damage,3,42 mechanisms other than impaired replication at UVA-induced DNA lesions still need to be considered for the observed slowing of replication with UVA.
Our data support the hypothesis that UVA-induced dimers are more mutagenic than UVB-induced dimers because of less protective, anti-mutagenic cellular responses. This, however, only applies to exposure to pure UVA. For example, following exposure to natural solar radiation, which contains both UVA and UVB, the UVB-induced DNA damage responses will also protect against mutation formation at the additional UVA-induced dimers. Nevertheless, life in modern times does involve situations of exposure to pure UVA, including e.g. exposure to sunlight filtered through window glass, sun-exposure with sunscreens that effectively filter UVB, but do not (or less effectively) filter UVA, or UVA-phototherapy. In those circumstances, the mutagenic and with that the carcinogenic potential of UVA-induced pyrimidine dimers, in the absence of UVB-induced protective responses, may be of particular clinical significance.
Footnote |
† Contribution to the themed issue on the biology of UVA. |
This journal is © The Royal Society of Chemistry and Owner Societies 2012 |