Singlet Oxygen Mediates the UVA-induced Generation of the Photoaging-associated Mitochondrial Common Deletion*

Mutations of mitochondrial (mt) DNA accumulate during normal aging. The most frequent mutation is a 4,977-base pair deletion also called the common deletion, which is increased in photoaged skin. Oxidative stress may play a major role in the generation of large scale mtDNA deletions, but direct proof for this has been elusive. We therefore assessed whether the common deletion can be generated in vitro through UV irradiation and whether reactive oxygen species are involved in this process. Normal human fibroblasts were repetitively exposed to sublethal doses of UVA radiation and assayed for the common deletion employing a semiquantitative polymerase chain reaction technique. There was a time/dose-dependent generation of the common deletion, attributable to the generation of singlet oxygen, since the common deletion was diminished when irradiating in the presence of singlet oxygen quenchers, but increased when enhancing singlet oxygen half-life by deuterium oxide. The induction of the common deletion by UVA irradiation was mimicked by treatment of unirradiated cells with singlet oxygen produced by the thermodecomposition of an endoperoxide. These studies provide evidence for the involvement of reactive oxygen species in the generation of aging-associated mtDNA lesions in human cells and indicate a previously unrecognized role of singlet oxygen in photoaging of human skin.

Mutations of mitochondrial (mt) DNA accumulate during normal aging. The most frequent mutation is a 4,977base pair deletion also called the common deletion, which is increased in photoaged skin. Oxidative stress may play a major role in the generation of large scale mtDNA deletions, but direct proof for this has been elusive. We therefore assessed whether the common deletion can be generated in vitro through UV irradiation and whether reactive oxygen species are involved in this process. Normal human fibroblasts were repetitively exposed to sublethal doses of UVA radiation and assayed for the common deletion employing a semiquantitative polymerase chain reaction technique. There was a time/dose-dependent generation of the common deletion, attributable to the generation of singlet oxygen, since the common deletion was diminished when irradiating in the presence of singlet oxygen quenchers, but increased when enhancing singlet oxygen half-life by deuterium oxide. The induction of the common deletion by UVA irradiation was mimicked by treatment of unirradiated cells with singlet oxygen produced by the thermodecomposition of an endoperoxide. These studies provide evidence for the involvement of reactive oxygen species in the generation of aging-associated mtDNA lesions in human cells and indicate a previously unrecognized role of singlet oxygen in photoaging of human skin.
Oxidative phosphorylation in mitochondria is carried out by five protein complexes encoded by both the nuclear DNA and the mitochondrion's own genome, the mitochondrial (mt) 1 DNA. Mutations of mtDNA have been shown previously to play a role in a variety of degenerative diseases mainly affecting muscle and nerve tissues (1-3) as well as diseases such as familial diabetes mellitus (4). Their relevance is not restricted to degenerative diseases, however; e.g. mtDNA mutations are also critically involved in the normal aging process (5)(6)(7)(8).
The most frequent and best characterized mutation in mtDNA is a deletion of 4,977 base pairs in length, also called the common deletion. This common deletion is considered to be a marker for mutations in the mitochondrial genome, and substantial efforts have been made to elucidate the mechanism by which it is generated. A modified slip-replication mechanism has been proposed (9 -12) involving the misannealing of direct repeats (Scheme 1). Hotspots for the common deletion exhibit structural abnormalities facilitating the misannealing of direct repeats from the light to the heavy strand of the mtDNA (13). This then leads to loop formation of both the heavy and the light strand of mtDNA. The initiation of loop exclusion is thought to be mediated by reactive oxygen species (13,14).
Reactive oxygen species can damage mtDNA (15)(16)(17), and damage by hydrogen peroxide is more extensive in mtDNA than in nuclear DNA (18). Furthermore it has been shown recently that increased oxidative stress is correlated to an altered mitochondrial function in vivo (19). In addition, oxidative stress induced by solar radiation may also be responsible for the increased frequency of mtDNA mutations in photoaged human skin (14, 20 -22). Evidence for a direct link between reactive oxygen species and the generation of large deletions of the mitochondrial genome, however, has not yet been provided.
In the present study we demonstrate that the common deletion can be generated under defined conditions in human dermal fibroblasts through repetitive UVA irradiation. By employing this in vitro model we provide evidence that UVA radiationinduced mtDNA deletions are caused by the generation of singlet oxygen.
Generation of the Common Deletion by UVA Irradiation-For UVA irradiation, medium was replaced by PBS, lids removed, and cells were exposed to radiation from a UVASUN 5000 Biomed irradiation device (Mutzhas, Munich, Germany). The emission was filtered with UVACRYL (Mutzhas) and UG1 (Schott Glaswerke, Munich, Germany) and consisted of wavelengths greater than 340 nm. The UVA output was determined with a UVAMETER (Mutzhas) and found to be approximately 70 milliwatts/cm 2 UVA at a tube-to-target distance of 30 cm.
In order to generate the common deletion, cells were irradiated three times daily with 8 J/cm 2 UVA for 4 consecutive days and checked for viability by trypan blue exclusion. Cells were then aliquoted 1:1 with one aliquot stored at Ϫ80°C until extraction of mtDNA. The other aliquot was plated to a 10-cm culture dish for ongoing culture and irradiation as indicated.
Chemical Treatments and Singlet Oxygen Generation-All chemicals were purchased from Sigma except for sodium azide (Merck) and applied as described previously (23). Vitamin E (as ␣-tocopheryl succinate) was dissolved in ethanol and added to cells 24 h before irradiation at a concentration of 25 M. For irradiation in the presence of heavy water, deuterium oxide (99.9 atom % 2 H) was used at a final concentration of 95% in PBS (24). Singlet oxygen was generated by thermal decomposition of the endoperoxide of the disodium salt of 3,3Ј-(1,4-naphthyli-* This work was supported by a grant from the Deutsche Forschungsgemeinschaft, SFB 503, project B2 and B1, and the Biomedizinische Forschungszentrum of the Heinrich-Heine-University, Düsseldorf, Germany. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. ¶ To whom correspondence should be addressed. Tel.: 49-211-811-7627; Fax: 49-211-811-8830; E-mail: krutmann@rz.uni-duesseldorf.de. 1 The abbreviations used are: mtDNA, mitochondrial DNA; PCR, polymerase chain reaction; PBS, phosphate-buffered saline; NDP, 3,3Ј-(1,4-naphthylidene)dipropionate. dene)dipropionate (NDPO 2 ). Cells were incubated with 0.3 mM NDPO 2 in PBS for 1 h in the dark at 37°C, which yielded singlet molecular oxygen and 3,3Ј-(1,4-naphthylidene)dipropionate (NDP) (25). This singlet oxygen system was shown to be well suited for use in cell cultures, because it is water-soluble and nontoxic for these cells up to 40 mM for 1-h incubation (24). Infrared emission of singlet oxygen was measured with a liquid nitrogen-cooled germanium photodiode detector (model EO-817L, North Coast Scientific, Santa Rosa, CA) as described previously (25). The rate of singlet oxygen generation was monitored by the formation of NDP. Fifteen minutes after addition of NDPO 2 , the rate of singlet oxygen generation was 1 M/min. DNA Extraction-Total cellular DNA was extracted from normal human fibroblasts employing the QIAamp Tissue Kit (Qiagen, Hilden, Germany).
PCR Analysis-For estimation of mtDNA content, PCR with primers C1 and C2 was carried out, amplifying a product of 247 base pairs in length. Amplification of fragments representing the common deletion was carried out as described previously (20,26). In brief, primer oligonucleotides (A1/A2) were designed to anneal outside the common deletion. During DNA amplification the polymerase extension time was designed to be too short for the amplification of wild-type PCR-products only allowing the efficient amplification of the shorter and deleted mtDNA-fragments. To increase sensitivity and specificity a secondary, nested PCR was performed (B1/B2) from the primary PCR product.
Linear amplification conditions for each primer pair used were determined as described in detail previously (27). Amplification was found to be linear up to 31 cycles for primers C1/C2 and A1/A2 and up to 30 cycles for B1/B2. Therefore primary PCR (primers A1/A2 and C1/C2) was carried out in 100-l reaction volume with 0.1-0. Amplification products were visualized in a 1% agarose gel stained with ethidium bromide (0.25 g/ml).
Quantification by Ion-Exchange Chromatography-Products were quantified by ion-exchange chromatography connected to an on-line ultraviolet spectrophotometer (Gynkotek, Germering, Germany), which allowed exact quantification of amplification products at 260 nm (27,29), and quantification of deleted mtDNA was carried out as described previously (20). The amplification product generated by primer pair C1 and C2 represents a segment of the mtDNA (30), carrying no known mutations of the mtDNA. Thus, it served as a reference fragment representing the overall amount of mtDNA molecules present in the cultured cells. Values determined for PCR products representing the common deletion were normalized to values of the reference fragments of the same cell lines.
Restriction Enzyme Analysis-To confirm their identity, PCR products were subjected to diagnostic digestion with the restriction enzyme XbaI (New England Biolabs GmbH, Schwalbach, Germany).

Generation of the Common Deletion in Normal Human Fibroblasts by
Repetitive UVA Irradiation-In order to determine the maximal UVA dose that allows sublethal repetitive irradiation, normal human fibroblasts were exposed to UVA doses of 0, 4, 8, and 16 J/cm 2 three times daily (Fig. 1). Repetitive irradiation with 16 J/cm 2 induced cell death, whereas irradiation with doses of 0, 4, and 8 J/cm 2 had no effect on cell viability. In the subsequent experiments cells were therefore exposed to repetitive doses of 8 J/cm 2 .
Amplification of total DNA extracts from normal human fibroblasts with primer oligonucleotides C1/C2 by PCR yielded the expected fragment of 247 base pairs in all experiments. The amplification of this fragment confirmed the effective extraction of mtDNA and served as a reference fragment for subsequent quantification (Fig. 2a). Nested PCR of fibroblasts exposed to repetitive UVA irradiations followed by ion-exchange chromatography showed no signal for the common deletion after 12 irradiations. Exposure of cells to 24 irradiations yielded the first detectable signal (Fig. 2, b and c), and maximal induction was observed after an irradiation of cells for 36 times. Sham-irradiated control cells showed no increase of the common deletion at any of the irradiation intervals. Digestion of PCR products with the restriction enzyme XbaI confirmed the identity of the amplified fragments (data not shown).
Role of Singlet Oxygen in the Generation of the Common Deletion-The role of singlet oxygen as a mediator of UVA radiation effects has been shown for several systems (31,32). Therefore, in the present study, reagents capable of quenching (sodium azide, vitamin E) or enhancing (deuterium oxide) singlet oxygen effects were examined for their capacity to affect UVA radiation-induced generation of the common deletion. As shown in Fig. 3, repetitive irradiation of fibroblasts in the presence of sodium azide significantly suppressed the induction of the common deletion in a dose-dependent manner. Conversely, suppression of UVA-mediated mtDNA mutagenesis was also observed when cells were treated with vitamin E during UVA radiation exposure. Moreover, irradiation of normal fibroblasts in the presence of deuterium oxide (D 2 O) resulted in a slight but consistent increase of signal intensity for the common deletion when compared with cells exposed to UVA radiation alone (Fig. 4).
The capacity of sodium azide and vitamin E to suppress and of deuterium oxide to enhance the UVA radiation-induced generation of the common deletion indicated a prominent role for singlet oxygen in this system. Therefore it was assessed whether these UVA radiation-induced effects could be mimicked by stimulating unirradiated cells with singlet oxygen generated via thermal decomposition of NDPO 2 (23,25). As shown in Fig. 4, incubation of cells with NDPO 2 led to the induction of the common deletion in normal human fibroblasts parallel to the observed UVA induced mutagenesis, whereas incubation with NDP did not generate a signal for the common deletion.

Generation of the Common Deletion in Vitro with
Physiological Doses of UVA-Large scale deletions of mitochondrial DNA such as the common deletion are thought to play a central role in aging of human tissues. Analysis of the mechanism by which the common deletion is generated in human cells during the normal aging process would require an assay system allowing the generation of mitochondrial mutations under standardized conditions and in a reproducible manner by exposing relevant target cells to physiologically occurring oxidative stress. In the present study we have therefore assessed whether it is possible to generate mitochondrial DNA mutations in cultured human dermal fibroblasts by exposing them in vitro to UVA radiation.
The assay system is based on the following. Chronically sun-exposed human skin has a higher content of mtDNA mutations than sun-protected skin, and a role for mtDNA mutations in photoaging of human skin has been proposed (14,20,22). Comparative analysis of the epidermal versus dermal compartment of photoaged skin revealed that the common deletion FIG. 2. UVA radiation generates the common deletion in human fibroblasts. Cells were irradiated with 8 J/cm 2 of UVA 12, 24, and 36 times. Total cellular DNA was extracted and subjected to PCR, amplifying either the reference fragment representing the total mitochondrial genome or the common deletion. Positive control represents a sample of a patient with known disease caused by the common deletion. Sham-irradiated cells were treated identically with regard to medium change and passaging. Data are given as mean Ϯ S.D. of relative content of the common deletion of three experiments and plotted in arbitrary units. a, representative agarose gel of PCR amplifications of the reference fragment. b, representative agarose gel of PCR products representing the common deletion. c, quantification by ion-exchange chromatography. Values of deleted mtDNA molecules were normalized for total amount of mtDNA molecules (reference fragment).

FIG. 3.
Suppression of UVA-mediated mtDNA mutagenesis by sodium azide or vitamin E. Cells were exposed to 8 J/cm 2 of UVA for 12, 24, and 36 times in the absence or presence of NaN 3 (20 and 50 mM) and vitamin E (25 M). In order to assure diffusion into cells and hydrolysis by esterases, cells were pretreated with vitamin E 24 h prior to UVA irradiation. For subsequent irradiations vitamin E was present continuously. The positive control represents a sample of a patient with known disease caused by the common deletion. Sham-irradiated cells were treated identically with regard to medium change and passaging. Data are given as mean Ϯ S.D. of relative content of the common deletion of three experiments and are plotted in arbitrary units. a, representative agarose gel of PCR products representing the common deletion. b, quantification by ion-exchange chromatography.
is almost exclusively present in dermal cells (14). Thus, in the present study, dermal fibroblasts were used and exposed to UVA radiation, which can penetrate into the dermis at significant dose levels. The UVA radiation doses employed for in vitro irradiation of fibroblasts are of physiological relevance, since they are similar to those administered to human skin during the course of a 15-30-min sun exposure on a summer day at noon at the northern latitude of 30 -35° (33). Human cells are capable of repairing damage to mtDNA induced by reactive oxygen species (34,35), and therefore an irradiation regimen has been developed that shifts the steady-state in cells between irradiation-induced damage and ongoing repair toward the damage side. By employing this protocol, it has indeed been possible to reproducibly generate the common deletion in human fibroblasts under standard conditions (Fig. 2).
A possible explanation for this could be that cells harboring the common deletion show a selective growth advantage, thus leading to increasing amounts of this deletion. It has been described, however, that there are selection mechanisms in cells containing large scale mtDNA deletions that lead to a growth disadvantage rather than an a growth advantage (26,36). Therefore it is very unlikely that the observed increase of the common deletion is simply due to differences in growth. Furthermore, the induction of the common deletion was a function of the number of UVA radiation exposures given to cells, indicating that it resulted from cumulative photodamage. We therefore propose that UVA radiation-induced generation of the common deletion in human dermal fibroblasts, as described here, represents a novel in vitro model to study the mechanism (i) by which the common deletion is generated in human cells in general and (ii) by which UVA radiation contributes to aging in particular.
Single Strand Breaks by Singlet Oxygen-It has been proposed that oxidative stress is responsible for the generation of mitochondrial DNA mutations in human cells (5), but a direct link between reactive oxygen species and large scale deletions of mtDNA has thus far been elusive. Here we have demonstrated that UVA radiation causes the generation of the common deletion in human fibroblasts through an oxidative mechanism that depends on the generation of singlet oxygen. Evidence for the involvement of singlet oxygen in biological processes is based on the use of quenchers to diminish singlet oxygen-mediated effects, on strategies that allow extension of the half-life of singlet oxygen to enhance singlet oxygen-mediated effects, and on the use of singlet oxygen-generating systems. By employing these three different strategies we here provide evidence that generation of the common deletion in human dermal fibroblasts through repetitive UVA irradiation critically depends on the generation of singlet oxygen. Formation of mitochondrial DNA mutations is thought to occur by misannealing of direct repeats situated in the D-loop (Scheme 1) during replication of mtDNA, thereby leading to excluded loops (9). In order to complete the deletion process, single strand breaks need to be generated to permit exonucleolytic degradation of the looped-out DNA. Based on the present study we propose a model in which singlet oxygen generation is responsible for base damage and subsequent strand break formation. In DNA, guanosines represent the prime targets for modification by UVA, and singlet oxygen (37,38) and G stretches are extremely susceptible to modification (39 -41). In mtDNA, replication is started at the replication origin of the heavy strand (OH) generating the D-loop (Scheme 1), which after misannealing leads to the excluded loop (9). The heavy strand contains several guanosine stretches of 3 to 5 bases of which many are in direct proximity to the direct repeats. Therefore it is tempting to speculate that the required strand breaks are indeed generated in this excluded loop via singlet oxygen.
Relevance for Photoaging-Our studies identify a previously unrecognized biological function of singlet oxygen. In addition, they demonstrate that oxidative stress is indeed responsible for the generation of large scale deletions of mitochondrial DNA in human cells that have been exposed to UVA radiation, which induces tissue aging under normally occurring conditions. In previous studies, singlet oxygen was found to mediate other UVA radiation-induced biological effects as well. In particular, UVA radiation-induced expression of metalloproteinases I, II, and III expression in human dermal fibroblasts was mediated through the generation of singlet oxygen (24). The increased expression of matrix metalloproteinases in human skin fibroblasts is thought to be partially responsible for the decreased content of collagen fibers in photoaged human skin. Furthermore recent findings indicate that singlet oxygen is linked with the in vivo UVA action spectrum, which is responsible for photoaging of mouse skin (42). Taken together with the present observation that UVA radiation-induced singlet oxygen is capable of generating mitochondrial DNA mutations in UVAirradiated dermal fibroblasts, it is possible that the generation of singlet oxygen in human skin is of central importance for photoaging. Singlet oxygen quenching may thus represent an effective strategy to protect human skin from photoaging. FIG. 4. Singlet oxygen mediation of UVA effect. Coincubation of irradiated cells with D 2 O: for UVA irradiation of fibroblasts, in the presence of heavy water, medium was replaced with deuterium oxide (99.9 atom % 2 H) at a final concentration of 95% in PBS. After irradiation, D 2 O was removed again, and cells were kept in normal medium. Generation of singlet oxygen by NDPO2: unirradiated cells were treated with singlet oxygen generated by thermal decomposition of the endoperoxide of the disodium salt of NDPO 2 , 0.3 mM in PBS for 1 h in the dark at 37°C, and yielded singlet molecular oxygen and NDP. The rate of singlet oxygen generation was monitored by the formation of NDP. Fifteen min after addition of NDPO 2 , the steady-state rate of singlet oxygen generation was 1 M/min. Control cells were stimulated with NDP, which had been generated by thermal decomposition from the same batch of NDPO 2 used in these same experiments. Data are given as mean Ϯ S.D. of relative content of the common deletion of three experiments, and error bars are given where applicable. a, representative agarose gel of PCR products representing the common deletion. b, quantification by ion-exchange chromatography.