The Efficacy of a Broad-spectrum Sunscreen to Protect Engineered Human Skin from Tissue and DNA Damage Induced by Solar Ultraviolet Exposure¹

Sunburn, pigmentation, hyperplasia, immunosuppression, and vitamin D synthesis represent acute responses of the skin to solar UVR,³ whereas photoaging and photocarcinogenesis constitute chronic effects. Clinical, experimental, and epidemiological evidence associates UV light exposure with the development of skin cancer (1). The incidence of nonmelanoma and melanoma skin cancers has been increasing in most parts of the world. Each year, >61,000 new cases of nonmelanoma skin cancer and 3,200 new cases of malignant melanoma are diagnosed in Canada (2). It has been estimated that a child born in Canada today has a 1 in 115 lifetime risk of contracting malignant melanoma and a 1 in 7 lifetime risk of having nonmelanoma skin cancer (2). Worldwide, the incidence of skin melanoma doubles every 10–15 years (3). Even more troubling is the 40–100% increase in the death rate from malignant melanoma between 1971 and 1996 (2).

Skin exposure to solar UVR induces significant damage to DNA, giving rise to several types of premutagenic DNA photoproducts (4, 5). These can be divided into three main classes: CPD, (6-4)photoproducts, and photooxidative damage (including strand breaks and various base modifications). The CPDs and the less frequent (6-4)photoproducts (15–30% of CPD levels) are produced by direct absorption of UVR by DNA (6, 7). A CPD results from the formation of a ring between C5 and C6 of two adjacent pyrimidines on the same DNA strand, whereas 6-4 photoproducts are characterized by a covalent bond between carbons 6 and 4 of two adjacent pyrimidines(e.g., TC or CC). Although it is evident that CPDs exert significant premutagenic potential because they are repaired at a much slower rate than (6-4) photoproducts, the later can also potentially cause skin cancer (8, 9). Indeed, either of these lesions can deliver misinformation during replicative bypass,leading to the fixation of mutations (6, 7, 10).

The general morbidity and mortality associated with skin cancer,therefore, represent a major health and economic problem. Several measures, such as sun avoidance, clothes, sunglasses, and sunscreens are available for attenuating the sun’s harmful effects. As evidenced by several studies, sunscreens can prevent sunburn (11, 12), immune suppression (13, 14, 15), actinic keratosis (16, 17), and UV-induced DNA damage (18, 19). The levels of UVB-induced CPD both in human (19, 20) and mouse models (21) were reduced when a sunscreen was applied prior to irradiation. Despite all of these reports, the effectiveness of sunscreens in preventing or reducing skin cancer still remains an open question. The relative effectiveness of sunscreens is evaluated based on their ability to prevent erythema. Although this is a convenient end point for such an assessment, it is nonetheless a crude indicator of UVR-induced damage. Hence, structural and molecular end points that play a role in carcinogenesis have to be studied to evaluate the ability of a sunscreen to prevent skin cancer.

The aim of our study was to evaluate the ability of a broad-spectrum SPF 30 sunscreen to protect engineered human skin against tissue alterations and DNA damage (photoproduct formation) after SSL exposure. For this purpose, we first performed immunohistochemical analyses to assess tissue structure, the formation and distribution of CPDs, and the deposition of basement membrane protein (laminin) after irradiation. We carried out molecular analyses to evaluate the frequency of CPDs, (6-4) photoproducts, and photooxidative damage formation after exposure of EHS to SSL.

Skin donors were healthy women, 15–20 years of age. Keratinocytes and fibroblasts were isolated from UV-unexposed normal human skin biopsies after breast reductive surgeries as described previously (22, 23). Both cell types were seeded in 75-cm²culture flasks (Falcon; Becton Dickinson, Lincoln Park, NJ). When the cultures reached 70–80% confluence for keratinocytes and 100% for fibroblasts, the cells were detached and used to design EHS. Tissues were produced by mixing calf skin type I and type III collagen (2 mg/ml) with normal human fibroblasts (1.5 ×10⁶ cells/ml) to produce the dermis. Tissues were cultured in 5% FCS-supplemented medium for 4 days and then seeded with keratinocytes (9 ×10⁴/cm²) to obtain the EHS. They were grown under submerged conditions for 7 days and then they were raised to an air-liquid interface for 5 more days to allow the differentiation of the epidermis into the different strata. Each series was conducted using keratinocytes and fibroblasts isolated from the same skin biopsy.

A broad spectrum SPF 30 chemical sunscreen and its vehicle, kindly provided by Bristol-Myers Squibb Co., Canada, were applied at a dose of 2 μl/cm² (24) on the stratum corneum of the EHS 30 min before irradiation. The sunscreen contained UVA filters (3.0% w/w of butyl methoxydibenzoylmethane and 3.0% w/w of oxybenzone) and UVB filters (5.0% w/w of octyl salicylate and 7.5%w/w of octyl methoxycinnamate). Three experimental conditions(untreated, vehicle-treated, or sunscreen-treated) were tested. Prior to irradiation, the culture medium was replaced by irradiation medium(DMEM supplemented with 12.5 μg/ml of bovine pituitary extract),without phenol red and hydrocortisone, to avoid the UV-induced formation of medium-derived toxic substances. Petri dishes containing the EHS were put on ice and uncovered to allow direct exposure of the EHS to UV rays. The SSL source used was a Kratos solar simulator equipped with a 2500-W Xenon compact arc lamp (Conrad-Hanovia, Inc.,Newark, NJ) delivering 1042 J/m²/s. The incident light was filtered through a sheet of cellulose acetate (Kodacel TA-407 clear 0.015 inch; Eastman-Kodak Co.), which efficiently blocks contaminating wavelengths <290 nm. As such, the term SSL will hereafter refer to the filtered wavelength incidence upon tissues. The SSL doses used were 0, 1000, 2000, 4000, and 6000 kJ/m² delivered at a fluence of 1042 J/m²/s. Approximately 1.5% of these doses fell into the UVB wavelength range (290–320 nm). The administrated doses were monitored using a YSI Kettering 65A radiometer (Yellow Springe Instruments, Dayton, OH). We chose low, intermediate, and high doses to facilitate the evaluation of the efficiency of the tested sunscreen in a broad dose spectrum.

Immediately after irradiation, biopsies were taken from each EHS. They were either fixed with Bouin’s solution and then embedded in paraffin or directly embedded in optimal cutting temperature, frozen in liquid nitrogen, and stored at −80°C until use. Thin cryostat sections (4μm) of the paraffin-embedded biopsies were stained with Masson Trichrome to evaluate the structure of the tissue as described elsewhere (22). Thin cryostat sections (4 μm) of the frozen biopsies were incubated for 45 min at room temperature with specific rat monoclonal antihuman laminin antibody (Chemicon, Temecula,CA) or mouse monoclonal CPD antibody (Biomedical Technologies,Stoughton, CA). The laminin antibody recognizes a conformational epitope localized on the laminin B1-B2 heterodimer and in the P1 fragment of laminin. The CPD antibody reacts specifically with UV-induced thymidine dimers in double or single-stranded DNA. Antibody dilutions were 1:100 for antilaminin and 1:50 for anti-CPD. Sections were then incubated in FITC-conjugated to goat antirat immunoglobulin(Chemicon) diluted 1:100 or goat antimouse immunoglobulin (Chemicon)diluted 1:100 for 30 min at room temperature. The sections were extensively washed with PBS between incubations. They were then mounted with a coverslip in 50% glycerol mounting medium and observed using epifluorescence microscopy and photographed. Results are representative pictures of the different EHS for each experimental condition(i.e., untreated, vehicle-treated, or sunscreen-treated). The experiments were repeated twice with four EHS samples per condition and per SSL dose. No major difference was observed between the histological and immunohistochemical results from unprotected and vehicle-treated EHS. We will not present pictures of vehicle-treated EHS, and we will hereafter refer to them as data not shown.

Immediately after irradiation, epidermal cells were isolated as described previously (22, 23). After homogenization, the cells were centrifuged, and the cellular pellet was resuspended in 2 ml of 0.15 m NaCl, 0.005 m EDTA (pH 7.8), and 2 ml of 0.02 m Tris-HCl (pH 8.0), 0.02 m NaCl, 0.02 m EDTA (pH 7.8), and 1% SDS. DNA was then purified as described previously (23), resuspended in distilled water, and used to evaluate the global frequency of photoproducts.

Global frequency refers to the average photoproduct density throughout the whole genome (e.g., 1 CPD/kb). To evaluate the global frequency of the photoproducts using agarose gel electrophoresis,photoproducts have to be specifically converted into DNA single-strand breaks. For each class of photoproducts, there is at least one method to specifically convert them into single-strand breaks. CPD can be enzymatically cleaved with the T₄ endonuclease V enzyme, and (6-4) photoproducts are converted using hot piperidine treatment (25). The photooxidative damage were specifically digested with Nth (also designed endonuclease III) and Fpg enzymes (also designed formamidopyrimidine glycosylase), both from Escherichia coli, as described previously (26). Digested DNA was resuspended at a final concentration of 1 μg/μl The global frequency for each class of photoproducts was determined with neutral agarose gel electrophoresis of glyoxal/DMSO-denatured genomic DNA as described previously in detail (26). Briefly, 5 μg of previously digested or treated DNA were dissolved in distilled water, and the following mix [2 μl of 100 mm sodium phosphate (pH 7.0), 3.5 μl of 6 m glyoxal (Sigma), and 10 μl of DMSO] was added. The DNA samples were then incubated at 50^oC for 1 h. Prior to loading, 3.8 μl of loading buffer [10 mm sodium phosphate (pH 7.0),50% glycerol, and 0.25% xylene cyanol FF] were added. The gels were run in 10 mm sodium phosphate (pH 7.0), running buffer at 3–4 V/cm with constant buffer circulation. The gels were stained for 2 h in a solution of 1× SYBR Gold nucleic acid gel stain (S-11494; Molecular Probes, Eugene, OR) in TAE (pH 8.0) and then photographed. No destaining or washing was required.

The overall adduct frequency was then estimated from these neutral-glyoxal gels after the enzymatic or chemical conversion of DNA phototproducts to single-strand breaks. The migration of the DNA fragments through the agarose gel allows their separation according to their molecular weight, the smaller the fragment the greater will be the distance of its migration. Willis et al.(27) have shown that when a randomly cleaved DNA molecule is gel fractionated, the mobility of each fragment is proportional to the log of the molecular weight throughout the middle of the mobility range. Using the same logic, we calculated the approximate mass of each DNA smear by estimating the molecular weight at the highest intensity of the DNA staining dye. The numbers obtained were divided by 2(because each fragment contains one photoproduct at each end) and then expressed as number of lesions per Mb.

Antisera were raised against DNA that was either irradiated with 100 kJ/m² UVC (254 nm) light for (6-4)photoproducts or dissolved in 10% acetone and irradiated with UVB light under conditions that have been shown to produce CPDs exclusively. For the RIA heat-denatured sample, DNA was incubated with 5–10 pg of poly(dA):poly(dT) (labeled to >5 ×10⁸ cpm/μg by nick translation with[³²P]dTTP) in a total volume of 1 ml of 10 mm Tris (pH 7.8), 150 mm NaCl, 1 mmEDTA, and 0.15% gelatin (Sigma). Antiserum was added at a dilution that yielded 30–60% binding to labeled ligand, and after incubation overnight at 4°C the immune complex was precipitated with goat antirabbit immunoglobulin (Calbiochem) and carrier serum from nonimmunized rabbits (University of Texas M. D. Anderson Cancer Center, Science Park/Veterinary Division, Bastrop, TX). After centrifugation, the pellet was dissolved in tissue solubilizer(NCS; Amersham), mixed with ScintiSafe (Fisher) containing 0.1%glacial acetic acid, and the ³²P was quantified by liquid scintillation spectrometry. Under these conditions, antibody binding to an unlabeled competitor inhibits antibody binding to the radiolabeled ligand. These details, as well as those concerning the specificities of the RIAs, have been described previously (28). Comparisons between photoproduct frequencies of unprotected and sunscreen protected tissues were made by using the Student’s t test. Results were considered significant if P < 0.05 and are presented as mean ± SD.

As shown in Fig. 1, the different strata(germinativum, granulosum, spinosum, and corneum) of EHS irradiated with 1000 kJ/m² were less distinguishable from each other compared with the unirradiated EHS (Fig. 1,a). As the SSL dose was increased, there was an increase in epidermal disorganization, as determined by the thickening of the stratum corneum and reduction in the number of epidermal cell layers. Morphologically differentiated keratinocytes (large cells with faint nuclei, large cytoplasms, and the presence of vacuoles) were also induced in these irradiated tissues. Furthermore, beginning at 1000 kJ/m² of SSL, vacuoles started to appear in the basal layer and persisted in a dose-dependent manner. At high doses of SSL (6000 kJ/m²), the basal cell layer was completely destroyed in the unprotected EHS (Fig. 1,d), and the dermis separated from the epidermis, which consisted mainly of dead cells forming the thickened stratum corneum. Comparable changes were observed in vehicle-treated EHS (data not shown). After exposure to 1000, 2000, and 4000 kJ/m² of SSL,sunscreen-treated EHS showed no tissue or cellular damage with the different epidermal layers of the sunscreen-protected EHS remaining intact (Fig. 1, f and g). Even at 6000 kJ/m², the sunscreen showed considerable attenuation of the epidermal damage caused by SSL (Fig. 1,h). For example, the basal layer as well as the dermis remained practically intact in sunscreen-treated irradiated (6000 kJ/m² SSL) EHS compared with the untreated and vehicle-protected EHS. The ability of the sunscreen to protect basement membrane proteins was also analyzed using immunofluorescence. In contrast to the unirradiated EHS, where there was a considerable deposition of laminin along the dermoepidermal junction (Fig. 2,a), SSL irradiation of unprotected EHS caused a dose-dependent decrease in laminin deposition with complete degradation of laminin after SSL doses as low as 1000 kJ/m². Application of the SPF 30 sunscreen clearly prevented this laminin degradation at low doses (Fig. 2, f and g, respectively). Laminin deposition was completely abolished at high doses in sunscreen-treated EHS (Fig. 2,h). In the presence of sunscreen, SSL irradiation resulted in a diffuse deposition of laminin along the dermoepidermal junction(arrows, Fig. 2,g) compared with the nonirradiated EHS (Fig. 2 a).

Using immunofluorescence micrography, we also evaluated the effect of the sunscreen on CPD formation and distribution after SSL exposure. As shown in Fig. 3, SSL irradiation of unprotected EHS revealed that the majority, if not all, of the epidermal cells contained CPD-stained nuclei. The CPD-positive nuclei were distributed throughout the full thickness of the epidermis, with a greater proportion of CPD-stained nuclei in the basal layer (Fig. 3, b–d). Furthermore, the higher the SSL dose, the greater the number of CPD-immunostained keratinocytes in unprotected EHS. Application of the SPF 30 sunscreen inhibited all CPD formation up to 4000 and at 6000 kJ/m², application of the sunscreen markedly reduced the CPD formation (Fig. 3 h).

In this paper, global frequency refers to the average frequency of photoproducts in genomic DNA (e.g., 255 CPDs per Mb of DNA sequence). The global frequency of CPDs and (6-4) photoproducts was quantified using two approaches: neutral glyoxal gel electrophoresis and RIA. Alkaline agarose gels are most commonly used for single-strand DNA electrophoresis. However, most (6-4) photoproducts and photooxidative damage are alkali labile, e.g., abasic sites are cleaved by β elimination in alkaline agarose conditions, and 8-oxoguanine bases are labile above pH 11 (29). Glyoxal-agarose gel electrophoresis of single-stranded DNA at neutral pH can be used to overcome this problem by size fractionating DNA, as in an alkali-agarose gel, while retaining alkali-labile sites intact (30).

The effects of SSL on the global frequency of CPDs in the epidermis of EHS are shown in Fig. 4. The analysis of DNA fragment mobility distribution showed (Fig. 4,A) that much smaller DNA fragments were found in the unprotected EHS compared with sunscreen protected EHS. Indeed, at an SSL dose of 1000 kJ/m², >1000 CPDs were formed per Mb. The global frequency of CPDs induced in the DNA of epidermal cells after SSL exposure was dose dependent, with the frequency ranging from 1000 CPDs per Mb at 1000 kJ/m² to >10,000 CPDs per Mb at 6000 kJ/m² (Fig. 4,A). The efficacy of the sunscreen in protecting EHS is evidenced by the striking reduction of the global CPD frequency. At 1000 kJ/m² of SSL, the global CPD frequency decreased from >1000 CPDs/Mb in unprotected EHS to <100 CPDs/Mb in protected EHS (Fig. 4,A). The quantification of CPD frequency in SSL-irradiated EHS was also performed by RIA. As shown in Fig. 4 B, 4000 J/m² of SSL induced ∼29,000 CPDs/Mb in unprotected EHS, whereas the same dose induced ∼12,000 CPDs/Mb in protected EHS. At 6000 kJ/m² of SSL, ∼70,000 CPDs were induced in unprotected EHS, whereas ∼30,000 CPDs/Mb were induced in protected EHS. Thus, the global CPD frequency was significantly reduced at all doses of SSL in sunscreen-protected EHS(P < 0.05 and P < 0.01).

The global frequency of (6-4) photoproducts was also evaluated using neutral glyoxal gel electrophoresis (data not shown) and RIA(Fig. 5). Like the CPD, the frequency of(6-4) photoproducts was also dose dependent. Indeed, in unprotected EHS, the frequency of (6-4) photoproducts increased 8-fold between 1000 and 6000 kJ/m², varying from 200 to 1600 (6-4)photoproducts/Mb. As shown in Fig. 5, relative to unprotected EHS, the global frequency of (6-4) photoproducts in the SPF 30-protected EHS was significantly reduced (P < 0.01). For example, 1000 and 4000 kJ/m² SSL induced 400 and 1040 (6-4)photoproducts/Mb in unprotected EHS and 200 and 485 (6-4) photoproducts per Mb in sunscreen-protected tissues.

SSL irradiation also produced a significant amount of photooxidative damage. The global frequency of photooxidative damage was quantified using neutral glyoxal gel electrophoresis of DNA digested with Fpg and endo III. As shown in Fig. 6, the density of low molecular weight DNA fragments was higher in unprotected EHS compared with protected tissues. Using this semiquantitative method, we determined that after irradiation with 6000 kJ/m²SSL, the frequency of photooxidative damage was reduced from 250 to 1000 lesions per Mb in unprotected EHS to 50–350 lesions per Mb in sunscreen-protected EHS (Table 1).

Although mounting evidence indicates that solar UVR is harmful to the skin, the number of sun worshippers is still increasing. Hence,there is an urgent need worldwide for protection against the deleterious effects of sunlight. Public health authorities recommend photoprotective measures, such as wearing protective clothing, reducing sun exposure, and using topical sunscreens. Over the past decade,intense research has been carried out to develop more efficient sunscreening agents, especially for wavelengths within the UVA (31). Nowadays, most sunscreens can efficiently absorb or reflect photons throughout the UVA and UVB spectra. Previous studies using animal models have shown that a SPF 15 sunscreen is effective in reducing SSL-induced connective tissue damage (32). Specifically, this study showed that the sunscreen was able to prevent elastosis, but elastic fibers were mildly hyperplastic. Collagen appeared undamaged and, although dermal cellularity was increased,massive inflammation did not occur. On the other hand, the glycoproteins were slightly, but not remarkably, increased in sunscreen-protected and irradiated animals compared with unirradiated animals. These studies did not address the efficacy of sunscreen at protecting the epidermal structure against UVR. Using a critical alternative model, we were able to shed some light on the efficiency of an SPF 30 sunscreen at protecting the different epidermal layers constituting the engineered human epidermis.

Previous studies have shown that exposure of the skin to sunlight leads to biochemical and ultrastructural changes in dermal collagen. Indeed,UVR induces the formation of stable collagen cross-links that render the protein insoluble (33, 34, 35). Furthermore, repeated exposure to UVR also affects BM components (34, 35). However, to our knowledge, no investigation has rigorously examined the consequences of a single exposure to UVR on preexisting BM proteins. In this regard, we find that SSL was able to immediately abolish preexisting laminin (a BM protein) and that an SPF 30 sunscreen was able to maintain the basement membrane protein deposition in irradiated EHS. By preventing the degradation of preexisting BM proteins such as laminin, the sunscreen preserves the interaction between the epidermis and the dermis and thus helps to maintain normal skin function, even under UV stress. The question that still needs to be answered is: Does the sunscreen really preserve the preexisting BM proteins or does it preserve the secretion of these proteins after UV irradiation? This issue is currently being addressed.

Exposure of cellular DNA to UV component of sunlight produces several types of premutagenic photoproducts (4-9) which, if unprevented and/or unrepaired, can lead to mutation and promote skin cancer development (36, 37). Because sunlight is responsible for DNA photolesions (38, 39), it is reasonable to assume that the ability of sunscreens to reduce UVR-induced DNA damage would be closely related to their ability to prevent skin cancer. Some research groups,using cell culture systems or experimental animals, have correlated sunscreen efficiency with prevention of DNA damage (20). They demonstrated that sunscreens with an SPF value of 10–15 could significantly reduce CPD formation. We addressed this issue in human skin using EHS and showed that in unprotected irradiated EHS, the global photoproduct frequencies [e.g., CPDs, (6-4)photoproducts, and photooxidative damage] increased in a dose-dependent manner. In sunscreen-protected EHS, the frequencies of these three types of photoproducts were significantly reduced. Such results suggest that the quantitative (RIA) and semiquantitative(glyoxal gel) evaluation of sunscreen efficiency against photoproduct formation provides a highly relevant biological end point with respect to DNA damage prevention. The in situ localization of CPDs together with our observation of significant reductions in photoproduct frequencies indicate that the sunscreen provided an efficient block against UVR penetration through skin and prevented the direct absorption of UV photons by DNA. The assessment of sunscreen efficiency based on the mitigation of DNA damage is a recent development. Indeed,Walter and coworkers (40, 41) showed that several organic sunscreens reduced DNA damage in the skin of hairless mice, and Freeman et al. (18) showed that a SPF 15 sunscreen was able to significantly reduce UV-induced CPDs. Our data complement these earlier reports and confirm the observation that sunscreens reduce the formation of all three types of photoproducts [e.g., CPDs,(6-4) photoproducts, and photooxidative damage] in the DNA following skin exposure to solar UV light.

SPF labeling is based on in vivo determinations of erythema. Because we used a SPF 30 sunscreen in a nonerythematous in vitro assay, our evaluation of this sunscreen as far as erythema is concerned is not relevant. To assess the protection factor of a sunscreen in vitro, other end points are required. By definition, the SPF 30 sunscreen we used should reduce the erythemally effective UVR from simulated sunlight radiation to 1/30th of its unprotected effectiveness. Using our quantitative damage results, we have calculated a DNA-PF as the frequency of CPDs induced in unprotected EHS divided by the frequency induced in sunscreen-protected EHS. Using this calculation, the DNA-PF was between 1 and 2. Similar values are obtained using (6-4) photoproduct frequencies. These calculations suggest that SPF 30 is equivalent to 1–2 DNA-PF. The observation that erythema is dependent on CPD induction (42, 43) supports the relationship between SPF and DNA-PF. To determine whether a 1–2 DNA-PF is sufficient to protect form solar UVR skin cancer, more studies should be performed with different experimental models, including human skin in vivo, to confirm the correlation between the SPF and the DNA-PF.

In conclusion, the SPF 30 sunscreen provided the EHS with significant protection against SSL-induced tissue damage and the induction of the major photoproducts in DNA, including CPDs, (6-4) photoproducts,and photooxidative damage. Our study is a powerful reinforcement for the importance of regular sunscreen use, which may in turn constitute a highly effective first line of defense against cutaneous photodamage and skin cancer development. Our current data support those from other laboratories (13, 14, 15, 16, 17, 18) that sunscreens are necessary for protecting the skin against UVR-induced damage but that the degree of protection is limited. Thus, the use of a high SPF sunscreen does not necessarily allow indefinite sunbathing without significant damage to the skin. On the other hand, the present study leads to an important question: Do sunscreens provide complete protection, or do they merely delay or attenuate the damaging effects of UVR? Answers to these questions will require time-course studies, which we have already begun.

We are extremely grateful to Geneviève Ross for excellent technical assistance and to R. S. Lloyd and S. Boiteux for kindly supplying T4 endonuclease V, and Nth and Fpg, respectively. We also thank Claude Marin of the audiovisual service of Saint-Sacrement Hospital, Québec, and Diane Lepage and Richard Couture of the audiovisual service of CHUQ-Hôpital Saint-François d’Assise, Québec.