The Similarity of Action Spectra for Thymine Dimers inHuman Epidermis and Erythema Suggests that DNA is theChromophore for Erythema

Antony R. Young, Caroline A. Chadwick,* Graham I. Harrison, Osamu Nikaido,† Jonathan Ramsden,* andChristopher S. Potten*Department of Photobiology, St John’s Institute of Dermatology, Guy’s, King’s and St Thomas’ School of Medicine, King’s College London, University ofLondon, London, U.K.; *CRC Department of Epithelial Biology, Paterson Institute for Cancer Research, Christie Hospital, NHS Trust, Manchester, U.K.;†Kanazawa University Division of Radiation Biology, Kanazawa, Japan

The location of DNA photodamage within the epidermisis crucial as basal layer cells are the most likely to havecarcinogenic potential. We have determined the actionspectra for DNA photodamage in different human epi-dermal layers in situ. Previously unexposed buttock skinwas irradiated with 0.5, 1, 2, and 3 minimal erythemadoses of monochromatic UVR at 280, 290, 300, 310, 320,340, and 360 nm. Punch biopsies were taken immediatelyafter exposure and paraffin sections were prepared forimmunoperoxidase staining with a monoclonal antibodyagainst thymine dimers that were quantitated by imageanalysis. Dimers were measured at two basal layer regions,the mid and the upper living epidermis. The slopes ofdose–response curves were used to generate four actionspectra, all of which had maxima at 300 nm. Dimeraction spectra between 300 and 360 nm were independent

Solar ultraviolet radiation (UVR) causes skin cancer (Greenand Williams, 1993; de Gruijl and van der Leun, 1994),especially in sun-sensitive people (skin types I/II) who tanpoorly. DNA is a major epidermal chromophore (Young,1997) and there is increasing evidence that DNA photodam-

age, such as cyclobutane pyrimidine dimers (CPD) and consequentmutation (e.g., p53), have a direct role in the initiation of skin tumors(Ziegler et al, 1994). Mouse studies provide evidence that UVR-induced immunosuppression, which specifically inhibits the normalimmunosurveillance of UVR-induced tumor cells, may be mediatedvia epidermal CPD (Kripke et al, 1992). Overall, DNA is suspected ofbeing a chromophore for many of the photobiologic effects of UVR.These include cytokine induction (Yarosh and Kripke, 1996), whichis likely to be important in UVR-induced immunosuppression andacute inflammation (erythema/sunburn) (Ley, 1985), and tanning(Gilchrest et al, 1996).

The location of DNA photolesions within the epidermis is likelyto be very important in tumor initiation. Mutation to suprabasal

Manuscript received August 18, 1997; revised July 30, 1998; accepted forpublication August 27, 1998.

Reprint requests to: Dr. Antony R. Young, Department of Photobiology, StJohn’s Institute of Dermatology, University of London, St Thomas’ Hospital,London SE1 7EH, U.K.

Abbreviations: CPD, cyclobutane pyrimidine dimer; MED, just perceptibleminimal erythema dose; MOD, mean optical density; TT, thymine dimer.

0022-202X/98/$10.50 · Copyright © 1998 by The Society for Investigative Dermatology, Inc.

982

of epidermal layer, indicating comparable epidermaltransmission at these wavelengths. Furthermore, weobserved 300 nm-induced dimers in dermal nuclei; how-ever, there was a marked effect of epidermal layer between280 and 300 nm, showing relatively poor transmission of280 and 290 nm to the basal layer. These data indicatethat solar UVB (µ295–320 nm) is more damaging tobasal cells than predicted from transmission data obtainedfrom human epidermis ex vivo. The epidermal dimeraction spectra were compared with erythema actionspectra determined from the same volunteers and ultravi-olet radiation sources. Overall, these spectral comparisonssuggest that DNA is a major chromophore for erythemain the 280–340 nm region. Key words: cyclobutane pyrimidinedimer/dermal DNA photodamage. J Invest Dermatol 111:982–988, 1998

keratinocytes committed to terminal differentiation is likely to beinconsequential, whereas DNA damage and mutation to stem cells inthe basal layer may have serious long-term consequences. Most studiesof UVR-induced DNA damage in human skin have been based ontechniques that require the digestion of the epidermis (Freeman et al,1989; Bykov and Hemminki, 1996) prior to DNA extraction, therebypreventing the localization of damage. We have recently developedmonoclonal antibody-based image analysis techniques to quantitateDNA photodamage in human epidermis in situ (Potten et al, 1993;Young et al, 1996). A major advantage of this approach is that specificDNA lesions and cell types can be localized within the epidermis(Chadwick et al, 1995; Young et al, 1998).

We have exploited this advantage to determine action spectra forthymine dimers (TT), a specific type of CPD, in different layers ofhuman epidermis. In addition, we have determined the action spectrafor erythema, with qualitative and quantitative techniques, in the samegroup of volunteers. An action spectrum, which demonstrates therelative photobiologic efficiency at different wavelengths per se, can beused to identify chromophores and/or as a biologic weighting functionto calculate the relative importance of different wavelengths underactual environmental conditions. The latter could, for example, evaluatethe effects of ozone layer depletion on basal layer DNA photodamage.Whereas several human erythema action spectra have been published(Parrish et al, 1982; McKinlay and Diffey, 1987; Anders et al, 1995),our specific aim was to evaluate the significance of epidermal DNA asa chromophore for erythema by comparing TT and erythema action

VOL. 111, NO. 6 DECEMBER 1998 THYMINE DIMER ACTION SPECTRA IN HUMAN EPIDERMIS 983

Table I. Source attributes, demographic analyses, and median MED at differentwavelengthsa

Sex Skin type

λ (nm) Bandwidth (nm) M F I II I/II Median age (y) Median MED (J per cm2)

280 5 1 5 1 5 20.0 0.05290 5 1 5 6 28.5 0.035300 5 1 5 6 29.0 0.025310 5 2 2 1 3 24.0 0.25320 10 5 1 3 2 1 24.0 2.4340 20 5 1 6 25.5 24.0360 20 3 3 5 1 25.0 32.0All λ 18 22 5 33 2 25.5

aNote that at 320, 340, and 360 nm WG320 1 mm filters were used to reduce stray UVB radiation. In two cases skin typeswere defined as type I/II as we could not fit them into either category. The age range of volunteers was restricted to exclude anypossible effects of aging. The mean age was 25.8 6 5.7 (SD). Note that bandwidth refers to full-width half-maximum, which isthe bandwidth at which 50% maximal irradiance occurs. Median rather than mean MED is given because MED has been reportedto be log-normally distributed in a normal population (Mackenzie, 1983).

spectra in the same volunteer pool with the same irradiation sourcesand protocols. Apart from the mechanistic information that such acomparison might provide, we were interested in the value of erythemaas a noninvasive clinical surrogate for DNA photodamage.

MATERIALS AND METHODS

Volunteers Forty skin type I/II volunteers (healthy young adults) wererecruited for studies at 280, 290, 300, 310, 320, 340, and 360 nm (n 5 6 foreach wavelength except at 310 nm with n 5 4). Demographic analyses areshown in Table I. The study was approved by the Ethics Committee of St.Thomas’ Hospital (London, U.K.), and written informed consent was obtainedfrom all participants.

UVR sources and radiometry Monochromatic spectra were obtained froma Photo-irradiation System (Applied Photophysics, London, U.K.) with a 2kW xenon arc. Delivery of UVR was with a liquid light guide (Oriel,Leatherhead, U.K.) with a 5 mm diameter exit. Irradiance was routinelymeasured with a wide-band thermopile radiometer designed to accommodatethe light guide (Medical Physics, Dryburn Hospital, Durham, U.K.). Irradiancesincreased with wavelength, ranging from about 1.0–1.5 mW per cm2 atø 300 nm to about 50 mW per cm2 at 360 nm. Source attributes are shownin Table I, and their normalized emission spectra, determined with a BenthamDM 150 double monochromator spectroradiometer (Bentham Instruments,Reading, U.K.) calibrated with a deuterium source measured by the NationalPhysics Laboratory (Teddington, U.K.), are shown in Fig 1.

Skin irradiation and erythema assessment A perspex applicator with sixapertures, designed to accommodate the exit end of the liquid light guide, wasattached with adhesive tape to a flat area of previously unexposed buttock skin.After each exposure the light guide was moved to the next aperture. Eachperson was exposed to an assigned wavelength and also to 300 nm, which wasused as a reference for comparing the responses of the different wavelengthgroups. At each waveband, a geometric series of six exposures was given witha dose increment of √2. Dose ranges were determined in pilot studies.

Twenty-four hours after exposure, erythema intensity at each site was assessedvisually by at least two observers according to the following scale: 0, no reactionor observers not sure; 0.5, minimal perceptible erythema (no clear borders anddefined as the MED); 1.0, definite erythema with clear borders; 2, erythemawith edema; and 3, edema or blisters. In addition, there was triplicate quantitativeassessment of erythema intensity (arbitrary units) at each site, as well as anadjacent nonirradiated control site to provide a background erythema level,made with a PC-linked Dia-Stron erythema (reflectance) meter (Dia-Stron,Andover, U.K.). Individual MED was used as the fundamental exposure unitfor the subsequent dose–response studies for TT. For these, irradiation procedureswere as before, on the contralateral buttock, and each person was exposed to0.5, 1.0, 2.0, and 3.0 MED. The maximum time taken to deliver 1 MEDranged from about 15–30 s with wavelengths ø 300 nm to about 15 min at360 nm. Punch biopsies (4 mm) were taken from each site, under localanesthesia, immediately after exposure (in practice within 5 min) as well as acontrol biopsy from an adjacent nonexposed site.

Assessment of thymine dimers The details have been published elsewhere(Potten et al, 1993; Chadwick et al, 1995; Young et al, 1996). Briefly, within5 min, the biopsies were sliced into 1 mm strips and fixed in methanol at 4°C

Figure 1. Emission spectra of the seven UVR sources normalizedat unity. Larger half-bandwidths are used in the UVA region to reduceexposure times.

for 18 h, from which 3 µm sections were cut. These were immunostained witha monoclonal antibody for TT (TDM-1) (Mizuno et al, 1991) followed by aDAB-peroxidase immune reaction that gives brown nuclear coloration. Nucleiwere counter-stained with thionine. The sections were analyzed using theDiscovery automated image analysis system (Becton Dickinson, Leiden, TheNetherlands) in which the nuclei are identified and defined with the thionineblue stain using a 620 nm filter and the antibody reaction product is quantitatedusing a 460 nm filter, within the defined nuclear mask, and expressed as a meanoptical density (MOD) per nucleus. Standard sections were always included ineach staining run to verify staining, i.e., MOD, reproducibility. The intensityof the light was always set at a constant level and the MOD of a referencesample was confirmed to insure that intergroup comparisons could be made.Four epidermal layers were defined by an interactive process with the imageanalysis software: (i) upper, the outermost two nucleated layers; (ii) mid, thenext two cell layers down; (iii) basal 5–7, basal cells at a depth of 5–7 cells; and(iv) basal 81, basal cells at a depth of eight or more layers. Two assessmentswere made in the basal layer to accommodate its undulating nature. At least150 nuclei were assessed per cell layer per skin section. In each volunteer, thespecific background MOD (i.e., zero MED) was subtracted from the MOD ofthe UVR exposed sites/layers.

Data analysis In all cases, the individual biologic dose units (MED) wereconverted to physical units (J per cm2). Using data generated from the reflectance

984 YOUNG ET AL THE JOURNAL OF INVESTIGATIVE DERMATOLOGY

Figure 2. Quantitative and qualitative action spectra for erythema givesimilar results. Our data show that the use of a quantitative endpoint (∆E 550) has no advantage over the classic qualitative endpoint (MED). They aregenerally similar to those of the CIE reference spectrum (relative data) (McKinlayand Diffey, 1987; Anders et al, 1995), which confirms that our volunteer poolgives the expected spectral reponse. Note that quantitative analysis was notpossible at 280 nm because of irregular dose–response curves and the greatestinterperson MED variation was seen at this wavelength. Error bars: 6 SD of mean.

meter, the quantitative measurement of erythema was determined by thedifference (∆E) between the reading from a UVR-exposed site and thebackground value from a nonexposed site. UVR dose–response curves wereconstructed as a logit function of log10 UVR dose according to Diffey and Farr(1991). The dose to achieve ∆E 5 50 (D∆E, equivalent to µ1 MED assessedby eye) was calculated from the logit regression parameters. Erythema actionspectra were determined with qualitative and quantitative data using the classiclog101/MED endpoint and log101/D∆E, respectively.

Linear regression analysis was used to determine UVR dose–response curvesfor TT at each wavelength and each epidermal layer for each volunteer. It wasconsidered best to fit all data to the same model and linear regression waschosen as the simplest. Individual data were pooled to generate mean log10slope for TT induction 6 SD for each wavelength at each epidermal location.Action spectra for TT induction were generated by plotting mean log10 slopeagainst wavelength.

RESULTS

All volunteer groups show similar 300 nm MED The median300 nm MED for the whole volunteer group (n 5 39, one not done)was 0.032 J per cm2. The values at 300 nm for the groups also exposedto 280, 290, 310, 320, 340, and 360 nm were 0.030, 0.035, 0.032,0.028, 0.035, and 0.040 J per cm2, respectively, with a ratio of 1.6between the groups with the highest (360 nm) and the lowest (300 nm,see Table I) MED. This is about 14% more than a single exposure-dose increment (i.e., 31.4), indicating that all groups showed acomparable erythema response at 300 nm.

Erythema action spectra assessed by qualitative and quantitativeendpoints are similar Figure 2 shows that erythema action spectradetermined by clinical assessment of MED and by analysis of reflectancedata are the same. The quality of reflectance data at 280 nm was notas good as with other wavelengths, i.e., increased dose did not alwaysresult in increased erythema. We suspect that this is due to localvariations in stratum corneum thickness.

Epidermal action spectra analyses, supported by histology, showthat wavelengths ù 300 nm are readily transmitted through theepidermis Typical individual linear regression analysis of TT dose–response for each wavelength are shown in Fig 3. In general, theregression fits were good with R2 ù 0.85 in 32% cases, ù 0.75 in22% cases, and ù 0.65 in 10% cases. The data points represent the

individualised doses of 0.5, 1, 2, and 3 MED. They show that thereis a clear epidermal layer effect on the dose–response curves at 280and 290 nm with less steep slopes in the deeper layers. It should benoted that the quality of the data at 280 and 290 nm was less goodthan at other wavelengths, probably because of local variations instratum corneum thickness. Figure 4(a) shows histologic evidence forthe lack of epidermal penetration at 290 nm; however, at wavelengthsù 300 nm the dose–response curves for all epidermal layers can besuperimposed, even at the lower doses. This clearly demonstrates thatsaturation of the response has not occurred and that TT are inducedby suberythemal exposures at all wavelengths. Figure 4(b, c) showshistologic evidence for the lack of epidermal screening at 300 nm aswell as the presence of TT in dermal cells.

The action spectra for TT in the four different epidermal layers areshown in Fig 5. This overall analysis, based on the slopes of theregression curves, shows that between 300 and 360 nm, the actionspectra are independent of epidermal depth; however, epidermallocation modifies the spectra between 280 and 300. The shorter UVBwavelengths are much less effective at damaging the deep basal cellsindicating decreased epidermal transmission.

Comparisons between erythema and TT action spectra Acomparison between the erythema and TT action spectra is shown inFig 6. Between 280 and 340 nm, the erythema action spectrum showsexcellent concordance with the TT action spectra of the upper andmid epidermis, but deviation from the basal layers. All spectra showexcellent concordance between 300 and 340 nm. At 360 nm there isevidence of deviation between the erythema and TT action spectra.

Comparisons between in vivo and in vitro TT action spectra canbe used to evaluate the optical properties of epidermis Figure 7shows the relationship between our TT epidermal action spectra(quantum corrected) and the quantum corrected TT action spectrumin naked DNA using the same antibody (Matsunaga et al, 1991). Thelatter, as expected, shows a good match with the DNA absorptionspectrum (Sutherland and Griffin, 1981). There is a good in vivo/in vitro match of TT action spectra between 300 and 320 nm but thelack of match at wavelengths less than 300 nm demonstrates the markedscreening effect of epidermal chromophores. At 320–360 nm higherdoses are required for TT induction in vivo than would be expectedfrom the in vitro TT data.

DISCUSSION

We have determined action spectra for TT induction in different layersof human epidermis in vivo. Our data show that 300 nm is the mosteffective wavelength in the solar UVR range, irrespective of epidermallayer. Maximal efficacy in human epidermis at 300 nm was alsoreported by Freeman et al (1989) in a study using the endonucleasesensitive site technique to determine CPD levels in whole humanepidermis. A mouse study assessing epidermal CPD showed a peak at293 nm (Ley et al, 1983).

TT action spectra were dependent on epidermal depth in the 280–300 nm range, with 280 nm causing almost 40-fold less damage (slopecomparisons) to the deep basal layer than the upper layer. DNAdamage, however, was independent of epidermal layer at wavelengthsbetween 300 and 360 nm, clearly demonstrating a lack of screening,and therefore damage gradient, within the epidermis in this spectralregion. We have previously reported a lack of a gradient within humanepidermis for TT induced by 300 nm but a steep gradient was seen at260 nm (Chadwick et al, 1995). These observations are confirmed inthis study, in which a gradient was seen at 280 nm and 290 nm butnot with longer wavelengths. A similar lack of gradient was alsoobserved for 300 nm-induced epidermal p53 expression, although agradient was seen with 254 nm (Campbell et al, 1993). Bykov andHemminki (1996) compared CPD levels in naked DNA and humanepidermis induced by 0.02 J UVB and UVC per cm2 (similar to ourMED). Although UVC-induced CPD levels in epidermis were lowerthan naked DNA, a greater number of UVB-induced CPD/nucleotidewas observed in the epidermis. Overall, our data and those of othersshow that UVR transmission properties determined in vitro from human

VOL. 111, NO. 6 DECEMBER 1998 THYMINE DIMER ACTION SPECTRA IN HUMAN EPIDERMIS 985

Figure 3. Slopes of 280 and 290 nm dose–response curves vary with epidermal layerbut are layer independent with ù 300 nm(typical individual data). Note that 280 and290 nm dose–response curves are typically more‘‘erratic’’ than longer wavelengths. This was alsothe case with erythema at 280 nm and may berelated to local variations in stratum corneumthickness. The slopes of the 280 and 290 nmcurves are layer dependent. At wavelengths ù300 nm the slopes are layer independent. TTlevel is synonymous with MOD as describedin Materials and Methods.

epidermal sheets (Bruls et al, 1984) underestimate UVB-induced basallayer damage and may not be relevant to human skin in vivo.Furthermore, our data suggest that the use of in vitro transmission datato correct mouse action spectra for nonmelanoma skin cancer forhuman risk assessment (de Gruijl and van der Leun, 1994), mayunderestimate the potential carcinogenic effects of ozone layerdepletion.

As expected from our data, visual inspection of the skin sectionsshows evidence of UVB (300 nm)-induced DNA photolesions indermal cells, although this has not been quantitated nor have the cellsbeen identified; however, we believe that it is likely that solar UVBmay have a direct effect of the skin’s vasculature, especially in thepapillary region.

DNA shows an absorption maximum at about 260 nm with anabsorption tail in the UVB region (Sutherland and Griffin, 1981). Thein vitro action spectrum for TT, with the same antibody as our in vivostudy, is generally similar in shape to the DNA absorption spectrumwith a maximum at 260 nm (Matsunaga et al, 1991). At wavelengthsless than 300 nm, the differences between the epidermal TT actionspectra and the DNA absorption spectrum or the TT in vitro actionspectrum, are likely to be accounted for by the optical screeningproperties of the epidermis. Differences in the optical properties ofmouse and human epidermis, especially the stratum corneum, mayaccount for the mouse CPD action spectrum maximum at 293 nm(Ley et al, 1983). Our data show that the living part of the epidermishas very little effect on UVR transmission in the 300–360 nm range,but attenuates wavelengths less than 300 nm. The 300 nm peak in theTT action spectrum of the upper living epidermis, although less

pronounced than the lower layers, provides evidence of considerablescreening of the shorter wavelengths (i.e., , 300 nm) by chromophoresin the nonviable epidermis. Urocanic acid, which is found in uniquelyhigh concentrations (6–12 nmole per cm2) in the stratum corneum, isthe most likely candidate because of its peak absorption at about268 nm and, like DNA, its absorption tail in the UVB region. Ourdata show that, in the 320–360 nm region, higher doses are requiredfor TT induction in vivo than would be expected from in vitro data.We cannot explain this but it is possible that our data provide indirectevidence for UVA-induced enzymatic photoreactivation (namely TTmonomerization), which has been reported to occur within 30 min(Sutherland et al, 1992) although the role of this process in humanskin remains controversial.

We have demonstrated the presence of the 6–4 photoproduct inhuman skin in vivo after irradiation with solar simulated radiation andwith 260 nm (Chadwick et al, 1995; Young et al, 1996). The actionspectrum for the 6–4 photoproduct in human skin is not known but,based on in vitro data (Rosenstein and Mitchell, 1987; Matsunaga et al,1991), it is reasonable to assume that it is essentially similar to that forTT in the solar UVB region.

Erythema action spectra determined by visual assessment or byanalysis of dose–response data determined by reflectance were verysimilar, confirming the value of the MED concept. These action spectrawere similar, in qualitative and quantitative terms, to those publishedby others (Parrish et al, 1982; McKinlay and Diffey, 1987) and, forexample, to a recent study, based on 252 normal volunteers of skintypes I, II, and III, which reported a median 300 nm MED of 0.027 Jper cm2 with a 95% range of 0.015–0.051 J per m2 (Diffey, 1994).

986 YOUNG ET AL THE JOURNAL OF INVESTIGATIVE DERMATOLOGY

Figure 4. UVB (300 nm) induces DNA damage to basal and dermal cells. (a) Exposure to 2 MED 290 nm results in TT (brown stain) in upper epidermalnuclei but virtually no damage in the basal layer (blue counterstain only), indicating that this wavelength is readily absorbed by epidermal chromophores. (b) Incontrast, exposure to 2 MED 300 nm induces basal layer TT with no evident epidermal damage gradient. A TT gradient, however, is seen in hair folliclekeratinocytes and TT are apparent in dermal nuclei. (c) Additional evidence for 2 MED 300 nm-induced TT in dermal nuclei. Note that some dermal nuclei showblue counterstain only. Scale bars: 50 µm.

This is close to our median 300 nm MED of 0.032 J per cm2. Arecent human erythema action spectrum shows a steeper decline inthe 300–320 nm region (Anders et al, 1995) than earlier studies. Thereason for this is almost certainly the use of true monochromaticradiation from a laser source, which prevents the ‘‘contaminating’’effects of wavelengths lower than peak emission that are present inspectra from conventional monochromators.

There was a striking concordance between the erythema and the midand upper layer TT action spectra between 280 and 340 nm. Thisagreement was independent of cell layer between 300 and 340 nm. Ourendpoint for erythema was based on a threshold response, whether

assessed by eye or machine. It has long been established that erythemadose–response curves become less steep with wavelengths decreasingfrom UVB towards UVC (reviewed by Diffey and Farr, 1991) andconfirmed by our 280 and 290 nm erythema dose–response data (notshown). An important consequence of this is that the shape of the actionspectrum for erythema, at wavelengths less than 300 nm, depends on thedegree of erythema that is used as the endpoint. Relative to 300–310 nm,a much higher dose of shorter wavelengths is necessary to induce anintense erythema. Thus, the calculated action spectrum for an intenseerythema (Farr and Diffey, 1985; McKinlay and Diffey, 1987) is quitesimilar in trend to the action spectra for TT in the basal layers.

VOL. 111, NO. 6 DECEMBER 1998 THYMINE DIMER ACTION SPECTRA IN HUMAN EPIDERMIS 987

Figure 5. Action spectra (not quantum corrected) for TT in differentepidermal layers show lack of epidermal layer dependence from 300 to360 nm. These data indicate no significant screening by epidermalchromophores at 300–360 nm, but there is evidence of a marked epidermaltransmission gradient with shorter wavelengths, especially at 280 nm. Error bars:6 SD of mean.

Figure 6. Epidermal DNA is a likely chromophore for human erythema.The nonquantum corrected action spectra for 24 h erythema (1/MED) andTT are essentially the same at (a) 300–340 nm in all epidermal layers and (b)280–340 nm in the upper and mid-epidermal layers. The data show a trendfor lower MED between 340 and 360 nm in comparison with TT. This isconsistent with an additional non-DNA chromophore for erythema in thisspectra region (Anders et al, 1995). Note that data are normalized at 300 nm.This treatment is appropriate for our data as we show that epidermal screeningdoes not occur at this wavelength (see Sutherland, 1995 for discussion).

Direct evidence that DNA is a chromophore for erythema has beenobtained from animal studies (Ley et al, 1985). The overall similarityof the TT and erythema action spectra in our studies, especially in the300–340 nm range, provide good circumstantial evidence that epidermalDNA is a chromophore for human erythema; however, it is highly

Figure 7. In vivo and in vitro TT action spectra comparisons showmarked attenuation of wavelengths less than 300 nm by stratum coreum.The quantum-corrected action spectrum for TT in naked DNA (Matsunagaet al, 1991) is similar to the DNA absorption spectrum (Sutherland and Griffin,1981) and to the in vivo epidermal TT action spectra (after quantum correction)at all epidermal depths at wavelengths ù 300–320 nm. All the action spectradiffer at wavelengths , 300 nm, demonstrating the screening effect ofchromophores in the stratum corneum and the viable epidermis. In vivo, higherdoses of µ320–360 nm are needed to induce TT than would be expected fromin vitro data, suggesting the possibility of photoreactivation. Note that data arenormalized at 300 nm. This treatment is appropriate for our data as we showthat epidermal screening does not occur at this wavelength (see Sutherland,1995 for discussion).

improbable that there are different chromophores for different degreesof erythema induced by the same wavelength as might be suggestedby spectral comparisons at 280–290 nm. A more likely explanation isthat the precise location of the chromophore within the epidermis isthe determining factor, but our data cannot exclude a role for DNAdamage to dermal cells, e.g., endothelial cells, especially at wavelengthsof 300 nm and higher. We speculate that a UVC and a shorter UVBwavelength MED is due to DNA damage in the mid to superficialepidermis. This results in the release of mediators, e.g., cytokines,which diffuse into the epidermis and activate the erythema response;however, a more intense erythema by these wavelengths may be dueto DNA damage to deeper epidermal layers. An erythema actionspectrum by Anders et al (1995) provides evidence for a distinct UVAabsorbing chromophore with a peak at about 360 nm. Our data showevidence of a discrepancy between the erythema and TT action spectrabetween 340 and 360 nm, also providing some evidence for achromophore other than DNA in the UVAI (340–400 nm) part of thespectrum. We speculate that at about 360 nm, erythema may resultfrom absorption by at least two chromophores, one of which is DNAand the other unknown. We believe that we can exclude stratumcorneum urocanic acid, which undergoes trans to cis isomerizationafter UVR exposure, as a possible chromophore for erythema inthe UVB and UVA range. The action spectrum for urocanic acidphotoisomerization in human skin is flat from about 280–310 nm,with activity falling off by less than an order of magnitude at 340 nm(Gibbs et al, 1997). Figure 2 shows that 340 nm is three orders ofmagnitude less effective than 300 nm at erythema induction.

988 YOUNG ET AL THE JOURNAL OF INVESTIGATIVE DERMATOLOGY

As we have recently shown that the TT dose–response curves(expressed in individual MED units) for basal keratinocytes andmelanocytes in skin types I/II are virtually identical at 300, 320, 340,and 360 nm (Young et al, 1998), it is reasonable to conclude that theTT action spectra for both cell types are the same. Thus, our data alsoindirectly support the hypothesis that DNA, a chromophore formelanogenesis (Gilchrest et al, 1996) as the human action spectra forthe 24 h MED, and melanogenesis are the same in the 300–340 nmregion (Parrish et al, 1982).

The MED is the most widely used endpoint in clinical andexperimental skin photobiology. It is also the endpoint in the determina-tion of the sun protection factors of sunscreens. The role of sunscreensin the prevention of nonerythema endpoints, such as skin cancer, is ofconsiderable topical interest (McGregor and Young, 1996). Our datasuggest that human erythema is a good spectral surrogate for the UVBand UVAII (320–340 nm) component of sunlight-induced epidermalDNA damage to basal keratinocytes and melanocytes, which is animportant factor in skin cancer. This conclusion, however, is notsupported by a mouse study in which a UVA/UVB sunscreen wasmore effective than a UVB sunscreen, with comparable sun protectionfactors, at protecting from solar simulated radiation-induced CPD (Leyand Fourtanier, 1997). In any case, it must be stressed that suberythemaldoses of solar simulated radiation induce epidermal TT (Young et al,1996), and data from this study, using monochromatic sources, confirmthis observation. Thus the prevention of erythema alone by sunscreenscannot guarantee the prevention of epidermal DNA photodamage.

We are grateful to the UK Government’s Department of Health, the UK NationalRadiological Protection Board (NRPB), and Cancer Research Campaign UK for theirgenerous support of this study. We thank Mrs. Jacqui Nagel for her excellent technicaland administrative assistance with these studies. We are grateful to Dr. John Sutherlandand Dr. Tsukasa Matsunaga for generously providing the raw data for the DNAabsorption and TT in vitro action spectra, respectively (Fig 7). We also thank Drs.Grant Bellany, Sabina Pfeifer, Charlotte Proby, Jane McGregor, and SriramuluTharakarum for taking the biopsies and John Sheehan for preparing the figures andreferences. Finally, we express our gratitude to all the volunteers for their participation inthis study.

REFERENCES

Anders A, Altheide H-J, Knalmann M, Knalmann M, Tronnier H: Action spectrum forerythema in humans investigated with dye lasers. Photochem Photobiol 61:200–205, 1995

Bruls WA, Slaper H, van der Leun JC, Berens L: Transmission of human epidermis andstratum corneum as a function of thickness in the ultraviolet and visible wavelengths.Photochem Photobiol 40:485–494, 1984

Bykov VJ, Hemminki K: Assay of different photoproducts after UVA, B and C irradiationof DNA and human skin explants. Carcinogenesis 17:1949–1955, 1996

Campbell C, Quinn AG, Angus B, Farr PM, Rees JL: Wavelength specific patterns of p53induction in human skin following exposure to UV radiation. Cancer Res 53:2697–2699, 1993

Chadwick CA, Potten CS, Nikaido O, Matsunaga T, Proby C, Young AR: The detectionof cyclobutane thymine dimers, (6–4) photolesions and the Dewar photoisomers insections of UV-irradiated human skin using specific antibodies, and the demonstrationof depth penetration effects. J Photochem Photobiol B: Biol 28:163–170, 1995

Diffey BL: Observed and predicted minimal erythema doses: a comparative study. PhotochemPhotobiol 60:380–382, 1994

Diffey BL, Farr PM: Quantitative aspects of ultraviolet erythema. Clin Physics PhysiolMeasurement 12:311–325, 1991

Farr PM, Diffey BL: The erythemal response of human skin to ultraviolet radiation. Br JDermatol 113:65–76, 1985

Freeman SE, Hacham H, Gange RW, Maytum DJ, Sutherland JC, Sutherland BM:Wavelength dependence of pyrimidine dimer formation in DNA of human skinirradiated in situ with ultraviolet light. Proc Natl Acad Sci USA 86:5605–5609, 1989

Gibbs NK, McLoone PR, Simics E: An action spectrum for urocanic acid isomerisationin human skin in vivo. Photochem Photobiol 65S:103S–104S, 1997

Gilchrest BA, Park H-Y, Eller MS, Yaar M: Mechanisms of ultraviolet light-inducedpigmentation. Photochem Photobiol 63:1–10, 1996

Green A, Williams G: Ultraviolet radiation and skin cancer: Epidemiological data fromAustralia. In: Young AR, Bjorn LO, Moan J, Nultsch W (eds). Environmental UVPhotobiology. New York: Plenum Press, 1993, pp. 233–254

de Gruijl FR, van der Leun JC: Estimate of the wavelength dependency of ultravioletcarcinogenesis in humans and its relevance to the risk assessment of a stratosphericozone depletion. Health Physics 67:1–8, 1994

Kripke ML, Cox PA, Alas LG, Yarosh DB: Pyrimidine dimers in DNA initiate systemicimmunosuppression in UV-irradiated mice. Proc Natl Acad Sci USA 89:7516–7520, 1992

Ley RD: Photoreactivation of UV-induced pyrimidine dimers and erythema in themarsupial Monodelphis domestica. Proc Natl Acad Sci USA 82:2409–2411, 1985

Ley RD, Fourtanier A: Sunscreen protection against ultraviolet radiation-inducedpyrimidine dimers in mouse epidermal DNA. Photochem Photobiol 65:1007–1011, 1997

Ley RD, Peak MJ, Lyon LL: Induction of pyrimidine dimers in epidermal DNA of hairlessmice by UVB: an action spectrum. J Invest Dermatol 80:188–191, 1983

Mackenzie LA: The analysis of ultraviolet radiation doses required to produce erythemaresponses in normal skin. Br J Dermatol 108:1–9, 1983

Matsunaga T, Hieda K, Nikaido O: Wavelength dependent formation of thymine dimersand (6–4) photoproducts in DNA by monochromatic ultraviolet light ranging from150 to 365 nm. Photochem Photobiol 54:403–410, 1991

McGregor JM, Young AR: Sunscreens, suntans, and skin cancer. Br Med J 312:1621–1622, 1996

McKinlay AF, Diffey BL: A reference action spectrum for ultraviolet induced erythema inhuman skin. CIE J 6:17–22, 1987

Mizuno T, Matsunaga T, Ihara M, Nikaido O: Establishment of a monoclonal antibodyrecognizing cyclobutane-type thymine dimers in DNA: a comparative study with64M–1 antibody specific for (6–4) photoproducts. Mutation Res 254:175–184, 1991

Parrish JA, Jaenicke KF, Anderson RR: Erythema and melanogenesis action spectra ofnormal human skin. Photochem Photobiol 36:187–191, 1982

Potten CS, Chadwick CA, Cohen AJ, Nikaido O, Matsunaga T, Schipper NW, Young AR:DNA damage in UV-irradiated human skin in vivo: automated direct measurement byimage analysis (thymine dimers) compared with indirect measurement (unscheduledDNA synthesis) and protection by 5-methoxypsoralen. Int J Radiation Biol 63:313–324, 1993

Rosenstein BS, Mitchell DL: Action spectra for the induction of pyrimidine (6–4)pyrimidone photoproducts and cyclobutane pyrimidine dimers in normal humanskin fibroblasts. Photochem Photobiol 45:775–778, 1987

Surtherland BM: Action spectroscopy in complex organisms: potentials and pitfalls inpredicting the impact of increased environmental UVB. J Photochem Photobiol B: Biol31:29–34, 1995

Sutherland JC, Griffin KP: Absorption spectrum of DNA for wavelengths greater than300 nm. Radiation Res 86:399–409, 1981

Sutherland BM, Hacham H, Gange RW, Sutherland JC: Pyrimidine dimer formation byUVA radiation: Implications for photoreactivation. In: Urbach F (ed.) BiologicalResponses to Ultraviolet A Radiation. Kansas: Valdenmar Publishing, 1992, pp. 47–58

Yarosh DB, Kripke ML: DNA repair and cytokines in antimutagenesis andanticarcinogenesis. Mutation Res 350:255–260, 1996

Young AR: Chromophores in human skin. Physics Med Biol 42:789–802, 1997Young AR, Chadwick CA, Harrison GI, Hawk JLM, Nikaido O, Potten S: The in situ

repair kinetics of epidermal thymine dimers and 6–4 photoproducts in human skintypes I and II. J Invest Dermatol 106:307–1313, 1996

Young AR, Potten CS, Nikaido O, Parsons PG, Boenders J, Ramsden JM, ChadwickCA: Human melanocytes and keratinocytes exposed to UVB or UVA in vivo showcomparable levels of thymine dimers. J Invest Dermatol 111:936–940, 1998

Ziegler A, Jonason AS, Leffell DJ, et al: Sunburn and p53 in the onset of skin cancer.Nature 372:773–776, 1994