Revisiting the photochemistry of solar UVA in human skin

The UV portion of sunlight has traditionally been divided into three wavelength spectra: UVC (290 nm and below), UVB (290–320 nm), and UVA (320 nm-visible). The boundaries between these three ranges are somewhat arbitrary, but the divisions have been convenient for categorizing the response of human skin to solar exposure. Wavelengths <290 nm (UVC) are blocked by stratospheric ozone but include the peak absorption of nucleic acids. Because near-monochromatic UVC (254 nm) is emitted by low-pressure mercury germicidal lamps, these wavelengths have been the historical mainstay of DNA damage and repair studies until the more recent use of longer-wavelength “more relevant” UV sources. Solar UVB contains the shortest wavelengths that penetrate the ozone layer and are absorbed by nucleic acids and, thus, constitute the greatest threat to humans, inducing erythema (sunburn), DNA damage, and ultimately skin cancer. DNA absorbs energy from both UVC and UVB, and a small proportion is converted into covalent dimer photoproducts between adjacent pyrimidine bases. It is widely believed that these lesions are the sites of the sunlight-induced “signature” mutations associated with the initiation of carcinomas (1). Like the shorter-wavelength spectra, UVA is carcinogenic (2) and is even suggested to be of greater relative importance than UVB for melanoma formation (3). However, because the amount of UVA energy absorbed by DNA is several orders of magnitude lower than UVB (4), its mechanism of action is thought to be different from UVB, proceeding by secondary free radical pathways rather than by direct absorption. Thus, until recently, the role of oxidative damage in DNA has been considered paramount in UVA carcinogenesis and UVA-associated pathologies. Recent work from the laboratory of Thierry Douki in Grenoble, culminating in the paper by Mouret et al. in this issue of PNAS (5) and in the 2006 New Investigator Award from the American Society for Photobiology, promises to foster a major change in perspective and a re-evaluation of the importance of UVA in human health and its mechanism of action on human tissues.

Cyclobutane Dimers Predominate UVA

The observation that cyclobutane pyrimidine dimers (CPDs) are induced by UVA is not new (6). However, the observation that the CPD is the predominant lesion in the DNA in human skin explants and keratinocytes after exposure to UVA is revolutionary, and its implications are important in both the human health and environmental sciences. The Douki laboratory (5) has applied the versatile and powerful technique of HPLC coupled with tandem mass spectrometry to quantify all 12 of the different possible bipyrimidine photoproducts induced by UV radiation. As mentioned, these include the CPDs as well as (6-4) photoproducts [(6-4)PDs] and the secondary photoisomers of the (6-4) photoproduct, the Dewar pyrimidinones. In addition, they used HPLC and electrochemical detection to quantify the induction of 8-oxodeoxyguanosine (8-oxodGuo) in the same samples. Surprisingly, they found that in both human skin and cultured keratinocytes, both UVB and UVA induced significant amounts of T<>T, T<>C cyclobutane dimers. However, only T<>T and T<>C CPDs were produced in any significant amount by UVA, and there were no (6-4)PDs. The authors suggest (5) that CPD formation by UVA proceeds by some form of photosensitization rather than through direct absorption of UVA energy. They also found that there were no 8-oxodGuo products induced by either wavelength, indicating that the nucleus was well protected against reactions involving UV-induced singlet oxygen.

UVA- and UVB- induced skin cancers may be initiated by the same type of DNA damage.

UVA and Skin Cancer

The implications of these new findings are important and serve to further clarify long-standing questions as well as open up new lines of inquiry. The importance of UVA in skin cancer is undeniable. UVA dominates the solar UV spectrum, and this spectral preponderance has been used to argue for its biological importance even in the face of minimal molecular impact. Indeed, using a hybrid fish model, it has been suggested that UVA may be the primary cause of sunlight-induced melanoma and that without UVA sunscreens, we are afforded little protection against this deadly disease (3). However, with these new data from the Douki laboratory, it appears that reactive oxygen species may not play as significant a role in the biological effects of UVA as previously believed, at least so far as DNA damage is concerned. In fact, these data suggest that many, if not most, UVA- and UVB-induced skin cancers, including carcinoma and melanoma, may be initiated by the same type of DNA damage. Published mutation spectra support a role of CPDs in UVA mutagenesis; mouse studies show that C to T transition mutations predominate in both UVA-exposed skin (7) and squamous cell carcinomas from mice exposed to UVB or simulated sunlight containing both UVA and UVB (8). A dependence of melanoma on CPDs would be consistent with that observed in squamous cell carcinoma. Melanomas in Xeroderma pigmentosum patients, who are unable to repair CPDs, display a high incidence of C to T transition mutations (9). Similar to mutation spectra in UVB-irradiated mammalian cells and UVB-induced SCC (10), approximately half of the CDKN2A mutations in various human melanoma cell lines are C to T transitions, with approximately half of the remaining mutations being CC to TT tandem transitions (11). UVB-signature mutations at the CDKN2A locus have also been detected in benign and metastatic melanomas in the South American opossum (12). Finally, although BRAF mutations are not typically found at the expected dipyrimidine sites, BRAF mutations are usually only found in melanomas occurring in nonexposed skin (13). Although these new data suggest a major role of CPDs in UVA photobiology and carcinogenesis, they do not exclude reactive oxygen species as important intermediates in DNA photochemistry and mutation induction in response to UVA exposure (14, 15).

UVA, the Environment, and Sun Safety

Another interesting consequence of the new photochemical data from the Douki laboratory involves the relative abundance of CPDs and (6-4)PDs after exposure to solar UV radiation. Differences in mechanisms of formation and structure of the CPD and (6-4)PD are significant and may potentially determine their different molecular and biological effects in human skin (16). Both CPDs and (6-4)PDs can inhibit DNA synthesis and gene transcription, but because of structural differences, the (6-4)PD is considerably more efficient at blocking the progression of DNA polymerase than the CPD and may be responsible for most of the lethal effects of UV. In contrast, because of the relatively greater bypass efficiency, the CPD may be responsible for most of the mutagenic effects. The Mouret et al. paper (5) in the current issue shows that UVA induces CPDs but not (6-4)PDs. Hence, in an environment in which the UVB:UVA ratio is variable or in a state of transition, the relative lethality and mutagenicity of sunlight may also vary. For example, longer UVA wavelengths penetrate deeper in skin than shorter UVB wavelengths. Hence, at greater depth in the skin, the ratio of CPDs to (6-4)PDs will increase, and the mutagenic effects of sunlight caused by CPDs should show a corresponding increase at the expense of the lethal effects caused by (6-4)PDs. The opposite would be expected under conditions of stratospheric deozonation. In this instance, the increased UVB (but not UVA) associated with thinning of the ozone layer could serve to increase the relative proportion of (6-4)PDs compared with CPDs and thus increase the killing effects of sunlight with respect to its mutagenic effects. It is an ironic speculation that anthropogenic destruction of the ozone layer would actually serve to increase cell killing and reduce mutagenesis in sunlight-exposed skin, a combination of events that would reduce the risk for skin cancer. It should be kept in mind, however, that there is an epidemiological link between childhood sunburn (i.e., cell killing and inflammation) and melanoma (17). Furthermore, the increase in UVB over Antarctica and its subsequent lethal effects have had a significant impact on productivity in the Southern Ocean (18), effects that will ripple through the oceanic food web and ultimately impact the human population.

Because of the significant induction of CPDs by UVA, the need for additional protection against the entire spectrum of solar UV irradiation is now even more important. The overwhelming predominance of UVA in sunlight compensates in part for the lower efficiency of production of CPDs by UVA. Hence, the use of sunscreens with a high sunlight-protection factor (SPF) in the UVA spectral range is more important than ever, and the recent approval of Mexoryl is certainly a positive step forward. Indeed, the Douki data support the idea that sunscreens lacking UVA protection may actually increase the risk for skin cancer. Because SPFs are often based on the ability of a sunscreen to reduce erythema, which is mainly a UVB phenomenon, they may not be relevant in reducing the risks for UVA. The current work (5) suggests an alternative approach that would incorporate these new results. The wavelength dependence of CPDs could be quantified and used to design and use sunscreens with SPFs based on the frequency of CPD formation. Using this strategy, the calculation of SPF would be built on a firm scientific foundation and its value directly linked to the reduction of DNA damage and, by implication, reduction of cancer. Along similar lines, the data from the Douki group (5) make a strong case against the use of artificial tanning with UVA. The claim that the use of long wavelengths for tanning is “safe” is no longer valid if those same wavelengths induce DNA damage that is associated with the initiation of skin cancer. There are now, indeed, no safe UV wavelengths.