Photoprotection in Human Skin—A Multifaceted SOS Response


  • This paper is part of a special issue dedicated to Professor Hasan Mukhtar on the occasion of his 60th birthday.

*Corresponding author email: (Barbara A. Gilchrest)


Human skin has developed elaborate defense mechanisms for combating a wide variety of potentially damaging environmental factors; principal among these is UV light. Despite these defenses, short-term damage may include painful sunburn and long-term UV damage results in both accelerated skin aging and skin cancers such as basal cell carcinoma, squamous cell carcinoma and even malignant melanoma. While UV radiation damages many cellular constituents, its most lasting effects involve DNA alteration. The following sections briefly review UV-inducible protective responses in bacteria and in skin, thymidine dinucleotides (pTT) as a powerful probe of DNA damage responses, and potential means of harnessing these inducible responses therapeutically to reduce the now enormous burden of cutaneous photodamage in our society.

UV Radiation, DNA Damage and Skin Cancer

Acute UV damage may cause painful sunburn and cumulative long-term UV damage is implicated in photoaging and skin cancer (1–7), due in large part to DNA alterations (8,9). Increased exposure to sunlight and other UV sources has thus resulted in an alarming rise in rates of skin cancer (10–12). While all forms of skin cancer are on the rise, malignant melanoma is of greatest concern because melanoma is responsible for most skin cancer deaths (11,13,14). Unrepaired or incorrectly repaired photolesions may lead to mutations. Gain of function mutations in oncogenes as well as loss-of-function mutations in tumor suppressor genes such as p53 or p16 may result in inappropriate cellular proliferation (10,14–17).

Loss of p53 function plays a particularly critical role in photocarcinogenesis for several reasons. Normal p53 function is required for optimal DNA repair (18) and for apoptotic elimination of severely UV-damaged keratinocytes (19). p53 also participates in transient cell cycle arrest after UV irradiation (20), understood to reduce mutation risk by allowing more time for DNA repair before resumption of DNA synthesis (20). Chronically photodamaged epidermis is characterized by clones of p53-mutant cells whose survival advantage in the face of continued UV irradiation provides rich soil for accumulation of further DNA damage, greatly increasing the likelihood of malignant progression (19,21). However, p53 also plays a central but less well recognized role in the integrated set of “SOS-like” inducible responses discussed below that function to protect genomic integrity in mammalian tissues exposed to repeated environmental insults.

The balance between the rate of DNA damage and the rate and fidelity of repair determines mutation frequency. Mutations have been shown to increase exponentially with age (22–28) and the incidence of skin cancer, like most cancers in the general population, also increases exponentially with age (29,30). These increases are due, at least in part, to a decrease in DNA repair capacity as the organism ages (31–35).

The Bacterial SOS Response

Bacteria evolved in a UV-rich environment and demonstrate an ability to increase their repair of UV-induced DNA damage following an acute exposure. In the inducible repair system termed the save our ship (SOS) response by Radman (36) in 1974, DNA damage and stalled replication forks derepress genes involved in DNA repair, mutagenesis and survival (37). Weigle first observed that sublethal UV irradiation of host Escherichia coli before infection with UV-irradiated bacteriophage enhanced DNA repair and survival of these phage (38). Because the phage rely completely on host repair proteins to correct the pre-existing damage, this strongly implied that sublethally irradiated bacteria have better DNA repair than nonirradiated bacteria. However, this enhanced repair was associated with a higher mutation frequency, indicating that the inducible repair system has a lower fidelity (38,39).

The bacterial SOS system has been extensively studied at the molecular level. Single-stranded DNA, produced largely at stalled replication forks, binds the RecA protein and the DNA–protein complex then induces genes that are normally repressed (reviewed in Eller and Gilchrest [40] and references therein). The derepressed genes encode DNA repair proteins that are then produced in increased amounts until the basal state is re-established. Two of the induced proteins, umuC and umuD, mediate rapid but relatively inaccurate DNA repair, promoting mutations during the SOS response period (41). The response thus promotes short-term bacterial survival in the face of ongoing DNA damage while increasing the probability that the bacterial progeny may “adapt” in some way to the injurious environment.

The Eukaryotic SOS Response

There is substantial and growing evidence that eukaryotic cells have an inducible DNA repair capacity, functionally analogous to the well-documented SOS response in prokaryotic cells. In all eukaryotic cells, responses to DNA damage include not only DNA repair, but cell cycle arrest and induction of specific genes. One of the major mediators of these responses in mammalian cells is the p53 tumor suppressor and transcription factor, the “guardian of the genome” (42). Although the regulation and activation of p53 by DNA damage is beyond the scope of this article, there are many excellent reviews on this subject (43–49). However, the concept that mammalian cells, like bacterial cells, have a coordinated inducible DNA damage-based mechanism for temporarily enhancing DNA repair capacity has only recently gained acceptance (50,51).

Evidence for a coordinately regulated eukaryotic SOS response to DNA damage and replication stress has been accumulating for decades. In 1976, Lytle et al. (52) showed that UV-irradiated human fibroblasts supported the growth (an indirect measure of repair) of UV-irradiated herpes virus better than nonirradiated fibroblasts. Similarly, UV-irradiated monkey kidney cells supported the growth of irradiated SV40 virus better than did nonirradiated cells (53). Interestingly, increased repair of irradiated viral DNA in these mammalian cells was not associated with an increase in mutation frequency, in contrast to the prokaryotic SOS response (53), consistent with selective pressure during the evolution of higher organisms for high fidelity repair. That is, single-celled prokaryotes might benefit from better adaptation to environmental challenges through mutation, but enhanced mutagenesis in multicellular organisms would be expected instead to lead to cell transformation, cancer and eventual death of the organism. Jeeves and Rainbow (54,55) found that enhancement of mammalian DNA repair could also be stimulated by ionizing radiation (IR) and that this enhanced repair was effective against both UV- and IR-induced DNA damage (54,55). They also found that cells from patients with the genetic disease ataxia telangiectasia (AT) failed to exhibit an IR-inducible DNA repair capacity (56), consistent with the subsequently identified role of the ataxia telangiectasia-mutated (ATM) protein kinase in mediating cellular responses to DNA strand breaks generated by IR (57–59). A central role for p53 in this response was also demonstrated, as UV irradiation-enhanced DNA repair could not be induced in p53 null tumor cells (60,61) or in otherwise normal cells expressing dominant negative p53 (60).

A Model for Studying Inducible Mammalian DNA Damage Responses

Studies of the UV-inducible DNA damage responses in mammalian cells are necessarily confounded by the multiple well-documented targets for UV irradiation in addition to DNA such as cell membranes and cytoplasmic proteins. However, insights have been afforded by what, in retrospect, was a serendipitous observation.

In 1994 we observed that treatment of pigment cells or intact guinea pig skin with thymidine dinucleotides (abbreviated pTT to indicate a 5′ phosphate group), the major target for UV-induced photoproduct formation (62), could induce “tanning” clinically and histologically indistinguishable from that following UV irradiation (63). Subsequent studies demonstrated that DNA damage alone, for example treatment of cells with DNA restriction enzymes, also induced tanning (64), implicating UV-induced DNA damage (as opposed to UV-induced membrane or cytoplasmic damage) as a major stimulus of this response.

We therefore interpreted pTT as a selective mimic of DNA damage caused by UV irradiation that stimulates pigmentation at least in part by increasing mRNA and protein levels of tyrosinase, a critical and rate-limiting enzyme in melanin biosynthesis (63,65). The ability of pTT to stimulate pigmentation was later confirmed and expanded by Pedeux et al. (66), who also demonstrated that pTT treatment resulted in a temporary S-phase arrest and that pTT effects were not caused by thymidine monophosphate (66), excluding the possibility that this breakdown product of pTT was responsible for its activities. The upregulation of tyrosinase following UV irradiation was later shown by our group (65) and others (67) to be regulated by p53, and, in the case of pTT treatment, to require p53 (68). These data are highly consistent with the understood importance of tanning as a DNA damage response and its central role in the inducible SOS-like response with presumptive evolutionarily conserved function of protecting the genome against further DNA damage.

Cui et al. recently extended these observations by showing that the pro-opiomelanocortin (POMC) gene and its derived peptide α-melanocyte stimulating hormone (α-MSH), a major stimulator of pigmentation, contains a p53 consensus sequence (69). α-MSH content in UV-irradiated keratinocytes and murine skin was greatly increased as a result of p53-mediated transcription of the POMC gene after UV irradiation and α-MSH stimulation of melanocytes resident in the epidermis led to tanning through binding to the melanocortin-1 receptor (MC1R) (69). These events did not occur in p53 knock-out mice (69). This work delineates at least one molecular pathway by which p53 stimulates tanning and underscores the already recognized critical importance of paracrine stimulation of melanogenesis in melanocytes by neighboring keratinocytes (70). Interestingly, there is also evidence that, in addition to stimulating photoprotective tanning, α-MSH enhances DNA repair capacity. Böhm et al. first demonstrated enhanced survival and DNA repair of melanocytes, keratinocytes and fibroblasts supplemented with α-MSH (71). Furthermore, Hauser et al. showed that melanocytes with loss-of-function mutations in the MC1R were more prone to induced apoptosis and exhibited reduced DNA repair after UV irradiation (72). The precise pathways linking α-MSH/MC1R signaling to DNA repair have not been elucidated.

The impact of the increased epidermal production of α-MSH is likely enhanced by increased α-MSH binding to melanocytes, presumably due to increased MC1R expression, observed after pTT treatment (64) and hence also interpretable as part of the DNA damage-inducible response. Of note, cultured melanocytes can tan in isolation, aside from keratinocytes (73), in a p53-dependent manner (65), but whether POMC/α-MSH mediates the response is unknown. Furthermore, a role for p53 in regulating the several other well-established UV-induced keratinocyte-derived factors, other than α-MSH, has not been thoroughly explored.

Because p53 is known to participate directly in DNA repair and to transcriptionally regulate other repair proteins (reviewed in Helton and Chen [74]), and because pTT induces p53 (50), it was also of interest to examine the repair of DNA damage in pTT-treated cells. We found that pretreatment of normal fibroblasts with pTT enhanced removal of both thymine dimers and (6-4) photoproducts and enhanced cell survival after UV irradiation (50). The activation and induction of p53 was demonstrated by the accumulation of p53 in the nuclei of pTT-treated fibroblasts, the upregulation of the p53-responsive SDI1 (p21) gene in these cells, and the enhanced binding of p53 to its consensus sequence DNA in extracts from pTT-treated cells (50). Goukassian et al. demonstrated that these effects of pTT on p53 and p53-regulated gene expression were measurable within 24–48 h and were quantitatively and qualitatively similar to those following exposure to UV radiation (75). pTT-treated human cells were also shown to have an enhanced ability to repair DNA damaged by the chemical carcinogen benzo(a)pyrene, in a manner also dependent on p53 (76). The pTT-induced increase in repair capacity for both UV- and chemical-induced DNA damage was approximately 100% of basal levels (75,76). Such a doubling of DNA repair would be expected to have a profound protective effect against carcinogenesis because a decrease of only 5–8% characterizes patients with early-onset basal cell carcinoma, a malignancy highly associated with sun exposure, compared to age-matched cancer-free controls (28). Furthermore, an age-associated decrease in DNA repair capacity of only 15% during adulthood has been proposed to account at least in part for the age-associated increase in malignancy (28).

Together, these data strongly support the existence of a UV-inducible p53-mediated photoprotective SOS-like response of substantial magnitude in human cells and suggest that this response can also be activated following treatment with specific DNA oligonucleotides. However, unlike the highly mutagenic prokaryotic SOS response, the enhanced repair rate induced by pTT in cultured human cells is associated with a reduced mutation frequency following exposure to a DNA-damaging agent such as UV irradiation and no increase in the low constitutive mutation rate in nonirradiated pTT-treated cells (77,78). Critically, intermittent topical pTT treatment of intact repeatedly UV-irradiated murine skin increased the rate of photoproduct removal and decreased mutation frequency, as detected by recovery and sequencing of a transgenic lacZ/pUR288 reporter plasmid (79).

Together, these studies demonstrate that cells in human skin respond to DNA damage by increasing their capacity to repair subsequent damage as well as by increasing epidermal melanin, a large polymer arranged in supranuclear caps within keratinocytes and capable of absorbing UV photons and reactive oxygen species before they can interact with DNA to create photoproducts (80,81). Thus, responses induced by both UV and UV-mimetic oligonucleotides reduce DNA damage incurred following a subsequent UV irradiation, minimizing net damage from future insults. The level of damage produced by the subsequent exposure is reduced and this lesser damage is removed more rapidly and more completely. Functionally, this protective response resembles the prokaryotic SOS response, yet it differs in several fundamental ways. First, prokaryotes are not known to contain p53 or any homolog of this protein. Second, there is no damage-induced pigmentation response in prokaryotes. Third, only the prokaryotic SOS response is highly mutagenic, providing a mechanism for bacteria to respond rapidly to environmental challenges and to adapt a strategy not expected to yield a survival advantage in higher organisms, but rather to create an unacceptable cancer risk. Fourth, as discussed below, the mammalian SOS response as studied using the pTT probe in murine and human skin alters not only the damaged cell but also its tissue environment in ways that promote overall homeostasis and genomic integrity of the organism.

Molecular Mechanisms

The dinucleotide pTT was originally chosen to probe UV-inducible responses because it is the principal substrate for cyclobutane pyrimidine dimer formation by UV light. A control adenine dinucleotide, the complementary sequence, was inactive (63). However, it later became apparent that certain other, but not all, DNA oligonucleotides also induce these DNA damage-like responses (68,82). Comparing active to inactive oligonucleotides, we noticed that the activity paralleled the degree of homology to the mammalian telomere sequence, repeats of TTAGGG. Active oligonucleotides generally show a greater than 50% homology with the telomere sequence, whereas inactive oligonucleotides fail to meet this criterion (82). In general, longer sequences are more effective than the shorter sequences and the presence of cytosine residues greatly reduces the activity (68,82,83). The originally studied pTT may be considered 100% homologous to one-third of the TTAGGG repeat sequence. Because of the sequence similarity between the active oligos and the telomere sequence, we termed them “T-oligos.” Compared to pTT, a 11-base oligo 100% homologous to the TTAGGG sequence has a higher molar efficacy and induces a more dramatic S-phase arrest in treated cells (83) and a 16-base 100% homolog is even more potent in a variety of bioassays (82). Same-size oligos complementary to the telomere sequence or unrelated to this sequence are inactive (83).

Telomeres are tandem repeats of TTAGGG and its complement at the ends of chromosomes in all mammalian cells (reviewed in Blackburn [84]). Telomeres do not encode genes but rather are understood to “cap” (85) chromosomes and prevent chromosomes from being recognized as broken ends of DNA, thus avoiding inappropriate fusion between chromosomes (84). Because of the “end replication problem” (86,87), telomeres shorten with each round of cell division, ultimately resulting in critically short telomeres that cause cells to permanently withdraw from the cell cycle, a state termed senescence (88). Telomeres thus serve as the biologic clock, limiting the cell’s replicative capacity, understood to be a cancer prevention mechanism (89).

Telomeres are known to form a loop structure, held in place by insertion of a single-stranded overhang on the 3′ strand composed of TTAGGG repeats into the proximal duplex telomere and held in place by binding proteins (90). Interestingly, disruption of the normal telomere loop structure by ectopic expression of a dominant-negative form of the telomere repeat binding factor TRF2, required for maintaining the telomere loop structure, exposes the telomere 3′ TTAGGG overhang and induces ATM- and p53-mediated apoptosis or senescence, depending on the cell type (91–93). We speculate that T-oligos mimic the exposed telomere TTAGGG, such as in the overhang, and induce similar DNA damage-like responses (83). We further speculate that acute DNA damage and critical telomere shortening may be physiologic settings in which single-stranded TTAGGG repeats, otherwise never “visible” in the cell, become exposed, serving as a signal for cells to initiate DNA damage signaling (94,95) (Fig. 1). As discussed in more detail below, exposure of single-stranded TTAGGG repeats must logically also occur at times of telomere replication, when the complementary base pairs separate to allow DNA polymerase to generate new strands.

Figure 1.

 Proposed mechanisms for induction of DNA damage responses. Experimental disruption of the normal telomere loop structure through transfection and expression of a dominant negative telomere binding protein (TRF2DN) results in exposure of the 3′ telomere overhang, approximately 100–400 bases composed of TTAGGG repeats, and activation of ATM and its effector protein p53, followed by senescence or apoptosis of treated cells (91). Removal of a second TTAGGG binding protein protection of telomeres-1 (POT-1) has recently been shown to activate the ATM-related kinase ATR with similar consequences (140). We hypothesize that critical shortening of the telomere during serial cell divisions renders the loop structure stochastically unstable, after which it tends to spontaneously open, exposing the overhang and initiating the same signaling. Similarly, we postulate that acute DNA damage, such as formation of thymine dimers or 8-oxo-guanine by UV irradiation, distorts the loop structure, pulling the TTAGGG strand away from its complementary strand at lesional sites (indicated by triangles on the 3′ strand). Extensive damage might also result in opening of the loop itself, exposing the 3′ overhang, with TTAGGG sequences then initiating signaling through ATM/ATR and p53. Replication of the telomere during late S phase of the cell cycle would also be expected to separate the DNA strands, exposing TTAGGG repeats, with the same consequences. In each of these scenarios the exposed (non base-paired) TTAGGG repeats are indicated with asterisks. Finally, T-oligos, known to concentrate in the nucleus (83), also provide the putative signal. The consequence of initiating signaling through this pathway would be determined by the intensity and duration of the signaling. Exposure of numerous TTAGGG repeat sequences without prompt restoration of the telomere loop structure, as might be expected with TRF2DN treatment or at the time of cell entry into senescence with critically short telomeres, is predicted to produce the extreme responses of apoptosis and senescence; while brief exposures of limited TTAGGG repeats, such as experienced during sublethal DNA damage following UV exposure or during telomere replication, might produce less dramatic genome-protective responses compatible with later resumption of cell division, such as transient cell cycle arrest, enhanced DNA repair, enhanced melanogenesis or other protective responses described in the text. The consequence of T-oligo treatment might similarly depend on T-oligo dose and duration of cell treatment, as well as the underlying physiologic condition of the cell that is treated.

Regardless of the initiating event, DNA damage-like responses elicited by T-oligo treatment were found to be quite extensive. In normal human fibroblasts, T-oligo treatment resulted in an S-phase arrest, p53 induction and phosphorylation and phosphorylation of the p95/Nbs1 protein (96), known to mediate the ATM-regulated S-phase arrest in response to IR (97). The S-phase arrest in T-oligo-treated cells was found to be greatly diminished in cells lacking p95/Nbs1 (96) and in normal cells in the presence of serum stimulation to gradually proceed to a G1 arrest (95), characteristic of senescent or acutely DNA damaged cells (98,99). However, treatment with T-oligos did not decrease telomere length or in loss of the telomere overhang (96), in contrast to experimental telomere loop disruption (93) or entry of cells into senescence following multiple cell divisions (100). These observations suggest that T-oligos do not act by disrupting the telomere loop or otherwise affecting genomic DNA. Nevertheless, prolonged treatment of fibroblasts with T-oligo induces a p53- and pRb-mediated senescent phenotype characterized by a large, spread morphology and positive staining for senescence-associated galactosidase activity, upregulation and activation (phosphorylation) of p53, induction of p16 and inability to phosphorylate pRb and re-enter the cell cycle after serum stimulation (95,101), identical to the changes observed in cells entering senescence after multiple rounds of cell division (102). More recently, treatment of fibroblasts with T-oligos was shown to lead to phosphorylation of the variant histone H2AX protein (103). Phosphorylated H2AX (γH2AX), a common marker of sites of DNA damage and phosphorylation, is thought to facilitate recruitment of DNA repair and cell cycle regulatory proteins to such sites (reviewed in 104,105) and has also been reported in senescent cells (106). In T-oligo-treated cells γH2AX forms discrete foci that co-localize with the telomere-specific protein TRF1 or a telomere-specific PNA probe (103), strongly suggesting a telomeric site of action for these oligos. Aside from the predominant telomeric location of these foci in T-oligo-treated cells (103), they are indistinguishable in appearance from those induced by UV irradiation (Fig. 2).

Figure 2.

 DNA damage foci in UV-irradiated versus T-oligo-treated cells. Human fibroblasts were exposed to a moderately damaging dose of UVC (16 mJ cm−2) and harvested 4 h later or supplemented once with T-oligo (GTTAGGGTTAGGGTTA) 20 μm and harvested 48 h later. Cells were fixed in paraformaldehyde and reacted with an antibody specific for γH2AX, followed by an FITC-tagged secondary antibody, then counterstained with DAPI (blue) to show total nuclear DNA. Numerous γH2AX foci are present after both treatments.

Although we originally hypothesized that T-oligos mimic the exposed telomere 3′ overhang following experimental telomere loop disruption, at times of acute DNA damage, or when telomeres reach a critically short length (proliferative senescence) (94), they may also mimic or enhance other events. For example, in addition to the responses following gross experimental telomere disruption (91), it was recently shown that telomeres of exponentially dividing cells elicit modest but detectable DNA damage responses in S-phase, as they are replicated (107,108). It was postulated that telomeres may thus recruit proteins to facilitate their own replication and to reform the proper loop structure, proteins that also function in recombinant repair of DNA (107,108). Consistent with this concept, we have also found that T-oligos, presumptive mimics of exposed TTAGGG repeats, are most effective when provided to replicating cells and are largely inactive in serum-starved quiescent cells (M. S. Eller, M. Yaar and B. A. Gilchrest, unpublished observation), suggesting a connection between T-oligo-induced DNA damage-like responses and the S-phase of the cell cycle.

Recently, we found that the protein mutated in the progeroid cancer-prone Werner syndrome (WRN) plays a major role in the T-oligo-induced DNA damage-like responses (103). Reduction in WRN levels by RNAi in fibroblasts and U2OS osteosarcoma cells dramatically reduces p53 and H2AX phosphorylation after T-oligo treatment, and cells from patients with Werner syndrome in which WRN function is lacking are minimally responsive to T-oligos (103).

An essential feature of the most active T-oligos is a G-rich sequence and an absence of cytosine residues (82) and far more effective than pTT (Fig. 3). In fact, some oligonucleotides, such as repeats of GGTT, are as effective if not more effective than equal length 100% telomere homologs (82). Such oligonucleotides would be expected to form G-quadruplex structures, non-Watson–Crick base-paired stacks of guanine quartets, known to form in G-rich DNAs such as telomeres (109) and a preferred substrate for WRN (110). These considerations and recent experiments (111) suggest that T-oligos may form intra- or intermolecular G-quadruplex structures, interacting with WRN at telomeres, where this protein is known to localize at times of DNA replication (112–114) (Fig. 4). WRN is also known to cooperate with DNA polymerase delta to synthesize DNA from G-rich templates (115) and both WRN (116,117) and DNA polymerase delta (118) function in telomere replication. We postulate that telomeric G-rich DNA may form G-quadruplex structures during telomere replication and conceivably at other times when TTAGGG single-stranded repeats are generated, creating localized mini-blocks to replication. WRN interaction with these structures would then send a signal of replicative stress, temporarily inhibiting replication and generating localized DNA damage responses through ATM and possibly also the DNA-PK and ATR kinases (119,120). Additional G-quadruplex structures present in the nucleus at this time in the form of T-oligos could exaggerate this signal, generating more robust responses, including extensive H2AX phosphorylation and strong activation of DNA repair and cell cycle regulatory pathways, as observed after T-oligo treatment (103). A major role for WRN in initiating DNA damage signaling is thus consistent with much recent data about this protein’s function, as well as with our T-oligo data.

Figure 3.

 Comparative response of human skin to pTT versus (GGTT)4. Normal abdominal skin of a 45-year-old, fair-skinned woman obtained at the time of elective surgery was processed to make multiple 5 × 5 mm fragments as described (122) and each fragment was placed in a multiwell chamber with a porous base that provided medium to the dermal surface of the skin. The multiwell plate was maintained at 37°C in a 100% humidity, 5% CO2 environment and at 24 h intervals fragments were treated topically with pTT 400 μm (GGTT)4 50 μm or vehicle alone. After 96 h (four applications) the skin fragments were harvested and processed for Fontana-Masson silver staining to show the degree of melanization.

Figure 4.

 Hypothetical model for the mechanism of action of T-oligos. Single-stranded G-rich telomeric DNA generated under conditions described in Fig. 1 can form G-quadruplex structures and interfere with DNA repair or replication. Interaction of these structures with WRN generates a replicative stress signal to the cell via ATM/ATR kinases and their effector proteins. This results in localized DNA damage responses. Additional G-quadruplex DNA present at telomeres in the form of T-oligos could amplify this signal of replicative stress, leading to more robust DNA damage-like responses including extensive histone phosphorylation, cell cycle arrest and activation of the p53 pathway.

The above hypothesis is consistent with our 11-16 base T-oligo data, but does not appear to explain the activity of pTT, which is incapable of forming G-quadruplex structures. Conceivably, TT sequences within telomeric TTAGGG repeats are “presented” to a critical domain of WRN or a WRN interacting protein during resolution of G-quadruplex structures and this event is mimicked by pTT. Alternately, the mechanism by which this dinucleotide stimulates DNA damage-like responses may be unrelated to that of the longer, G-rich oligos. More studies will be required to resolve this apparent paradox.

Tanning—The Classic View

For centuries dark skin color has been recognized to afford good protection against sun damage. Both constitutive (baseline) and facultative (sun-induced) pigmentation afford such protection; and individuals with dark baseline skin color generally tan most proficiently (121). In the 1900s melanin was characterized as a large polymer and found to absorb UV photons as well as reactive oxygen species and chemical toxins. Light and electron microscopic studies localized melanin-containing melanosomes within the skin to the melanin “parasol” or supranuclear cap where the pigment could absorb incident photons that would otherwise reach the nucleus and damage DNA (80,81). These observations are consistent with melanin and UV-induced melanogenesis (tanning) as evolutionary adaptations to damaging UV irradiation, but failed to explain the discrepancy between the measurable sun protection factor of a heavily melanized epidermis and the far greater protection against chronic photodamage observed in dark-skinned “good tanners”versus fair-skinned “poor tanners” (122).

Tanning Includes Enhanced DNA Repair and Antioxidant Capacity

One explanation for the fact that “good tanners” are protected against skin cancer and photoaging to a greater degree than predicted by the ability of their epidermal melanin to absorb photons is the UV-induced additional DNA repair capacity discussed above and assumed to parallel UV-induced melanogenesis, as both responses are driven by p53 activation (50,65,75). Consistent with this speculation, using pTT as a probe for the DNA damage component of UV irradiation, we were able to induce photoprotective responses in human skin explants (123,124). In this system, oligonucleotide pretreatment raised both constitutive and UV-induced p53 levels and the degree of phosphorylation of p53 on serine-15 (123), indicative of p53 activation (125). pTT pretreatment also accelerated removal of cyclobutane pyrimidine dimers following UV irradiation and increased the proportion of photoproducts removed during the “safe” period of epidermal arrest (124). Furthermore, as expected, treatment of these human skin explants with pTT induced pigmentation and expression of melanogenic proteins similar to UV irradiation; and both pTT and UV irradiation decreased by approximately half the amount of DNA damage introduced by a subsequent UV exposure, compared to damage immediately postirradiation in never previously UV-exposed explants (124).

Much UV-induced damage results from the direct interaction between photons and DNA bases, producing dimers and other photoproducts (11), but DNA is also damaged indirectly via reactive oxygen species created by photon interactions with other cellular constituents (30). It was therefore of interest to determine whether the inducible DNA damage response to UV irradiation includes protection against oxidative stress. We first studied keratinocytes and found that UV irradiation resulted over several days in upregulation of both the constitutively expressed and inducible superoxide dismutases (SOD-1 and SOD-2), responsible for converting highly reactive oxygen species to less damaging ones (126). Again using pTT as a probe for UV-induced DNA damage selectively, we demonstrated upregulation of SOD-1 and SOD-2, as well as enhanced cell survival of pTT-treated cells following challenge with hydrogen peroxide (127,128). This response is also mediated by p53 activation (127,128) and therefore likely parallels DNA damage-induced melanogenesis in proficiency within individuals. More recently, in a mouse model of chronic UV irradiation (three times weekly for 7 months), pTT treatment was found to result in strikingly lower levels of 8-oxo-guanine (129), the major form of oxidative DNA damage (130), 24 h after the final UV exposure. These data strongly suggest that the mammalian SOS response has evolved to minimize DNA damage from oxidative stress, as well as from direct UV-induced base damage.

Tanning Includes Anti-Inflammatory Responses

A further damaging consequence of UV irradiation in mammalian skin, not observed in bacteria, is inflammation. After an acute UV exposure, “rubor, dolor, calor and loss of function” evolve over 12–24 h or longer, depending on UV dose and the individual’s skin type, constituting the familiar sunburn reaction (131). In chronically UV-irradiated skin, sunburn is less readily detected, but “heliodermatitis” or “dermatoheliosis” is well described (132), especially in those who tan poorly. In addition, UV-induced inflammation is implicated in progression of photocarcinogenesis (133–136). Many cell types and inflammatory mediators have been shown to participate in this reaction, prominently including prostaglandins (137). In particular, the inducible form of cyclo-oxygenase (COX-2) that mediates formation of prostaglandins and eicosanoids (137) has been implicated in both acute and chronic responses to UV. Of particular interest in the present context, increased expression of COX-2 promotes photocarcinogenesis (as well as tumor progression in other settings, such as colon carcinoma) and inhibition of COX-2 in mice decreases UV-induced squamous cell carcinomas (SCCs) (135,138).

To determine whether the inducible SOS-like response to UV irradiation involves activation of anti-inflammatory pathways, human keratinocytes and fibroblasts and intact mouse skin were treated with the UV-mimetic dinucleotide pTT and then UV irradiated (139). In controls, COX-2 levels were induced several fold within 24 h. However, in pTT-pretreated cells and mice, within 24 h COX-2 protein levels were strikingly lower. Further studies revealed that this downregulation occurred at the transcriptional level, the result of competition between the proinflammatory NFκB and anti-inflammatory p53 for the p300 co-activator required by both transcription factors (139).

In subsequent mouse experiments involving chronic UV irradiation (three exposures weekly for 7 months), strong COX-2 protein expression was observed in the control murine epidermis 24 h after the final exposure, as expected based on the literature (135,138), but in mice treated 5 days week−1 throughout the irradiation period with topical pTT, epidermal expression of COX-2 was reduced by ∼90% (129). This reduction was associated with a highly significant approximately 50% reduction in photocarcinogenesis (129), although cause-and-effect were not established, and pTT-treated mice also demonstrated enhanced DNA repair (129).

These data suggest that the well-known inverse tendency to sunburn prominently and then suntan poorly or, conversely, to sunburn minimally and then suntan darkly, reflects more than simply basal levels of melanin pigment in the UV-irradiated epidermis. Indeed, the observation that skin phototype (historical tendency to sunburn versus suntan) may be quite independent of basal melanin content was the major motivation for development of Fitzpatrick’s very useful classification system that does not rely on untanned skin color to predict vulnerability to either acute or chronic UV damage (121). If one considers both sunburn and suntan as reflections of proficiency of the multifaceted inducible DNA damage response, skin phototype I/II individuals can be seen to have a poor SOS-like response while skin phototype IV-VI individuals can be seen to have a very good to excellent SOS response, manifested during the periods of repeated UV exposure as minimal inflammation, rapid DNA repair and marked enhancement of melanogenesis and anti-oxidant defenses (Fig. 5). Long term, the different degrees of protection to genomic integrity afforded by poor versus excellent SOS responses may largely determine the risk of photocarcinogenesis in these individuals. Although still quite speculative, such a formulation is consistent with laboratory-based, clinical and epidemiologic findings and deserves thorough study.

Figure 5.

 Tanning as a multifaceted inducible response to UV-induced DNA damage. In addition to increased melanogenesis, the damaged epidermis is characterized by enhanced DNA repair and antioxidant defenses, as well as diminished inflammation. This state is maximally induced with 24–48 h and then persists for at least several days. The intensity of this tanning response depends on the severity of the initial damage and the individual’s genetic endowment, but is highly protective against future UV damage within the save our ship period.

Clinical Implications and Applications

The evolutionarily conserved SOS response is best viewed as an adaptive response to environmental insults, a means of committing more of a cell’s precious energy resources to genomic protection at times of greater-than-baseline threats to the integrity of its DNA. It is a superb biologic solution to cell survival in an unpredictable environment in which intermittently present threats are assumed far more likely to occur during a period of hours to days after an initial episode of damage than during extended periods of no damage. In the current world environment, however, human beings can often anticipate periods of increased risk. In the case of UV exposure, examples might include a planned sunny vacation during a long winter or a summer weekend outing after days of indoor activities.

The work summarized above demonstrates that the physiologic SOS-like inducible DNA damage response can also be induced by pTT or other small DNA fragments, without incurring damage to genomic DNA. This provides a theoretic opportunity to reduce cumulative DNA damage in human skin by inducing the response prior to an anticipated first episode of DNA damage, or even by inducing the response to a higher-than-otherwise level during periods of repetitive frequent damage. This approach has already been validated in mouse models of photocarcinogenesis, in which topical intermittent pTT applications during months of frequent UV exposures substantially and significantly reduce the tumor burden for both SCC (79) and basal cell carcinoma (129). In combination with conventional sunscreen use and other standard safe-sun practices, induction of the innate SOS-like response by pTT thus offers the prospect of a much needed new approach to photoprotection.

Conflict of Interest Statement— This review describes discoveries that are patented in part and licensed to SemaCo, Inc., a for-profit company in which Dr. Eller and Dr. Gilchrest hold equity.


Acknowledgements— The authors thank Ingrid Persson and Jacy Bernal for excellent secretarial assistance. Drs. Hee-Young Park, Mina Yaar and Simin Arad prepared and photographed the skin explants shown in Fig. 3. This work was supported by NIH grant R01 CA 10515 (BAG).