Proteins of nucleotide and base excision repair pathways interact in mitochondria to protect from loss of subcutaneous fat, a hallmark of aging

We started by asking whether the CSA and CSB proteins, known to be present predominantly in the nucleus, are also localized in mt. Immunocytostaining with antibodies against CSA or CSB and against mitochondrial marker proteins, as well as nuclear staining with subsequent confocal laser scanning microscopy, was performed in normal and CSA- and CSB-deficient fibroblasts (see Tables S1, S3, and S4 for detailed cell lines and antibodies). In normal fibroblasts, CSA (Fig. 1 A) and CSB (Fig. 1 B), proteins were virtually undetectable in mt; however, these proteins were clearly visible in the nucleus with some minor localization at the cytoplasm (Fig. 1, A and B). Confocal images of cells without nuclear staining confirmed the presence of CSA and CSB proteins in the nucleus and cytoplasm (Fig. S1 and not depicted). Confocal microscopy of CSA and CSB cells with the CSA and CSB antibodies, respectively, validated the specificity of the antibodies (Fig. 1, A and B, and Fig. S1). Note that in the absence of a functional CSB protein (CSB cells; Fig. 1 A, bottom), an increased cytoplasmic staining of CSA was observed, whereas the nuclear staining was similar in both CSB and normal cells.

Genotoxic stress is known to increase CSA localization to the sites of nuclear DNA damage (Kamiuchi et al., 2002). To determine whether oxidative stress is capable of inducing the CSA and CSB protein levels in mt, we exposed human CSA- and CSB-deficient and normal fibroblasts to 25 µM H₂O₂ for 12 h. This treatment did not affect cellular viability. In contrast to the unstressed situation, CSA (Fig. 1 C) and CSB (Fig. 1 D) proteins were clearly detectable by confocal microscopy in mt after oxidative stress. The staining was highly specific, as the CSA or CSB proteins were undetectable in mt or nuclei of cells defective in CSA or CSB, respectively. CSA and CSB cells complemented with the respective hemagglutinin (HA)-tagged-WT CSA and HA-tagged-WT CSB proteins showed reconstitution of CSA (Fig. 1 E) and CSB (not depicted) in mt to WT levels. Confocal microscopy of cells without nuclear staining showed the presence of CSA (Fig. S2) protein in the nucleus and mt. Similar results were obtained when cells were exposed to a UVA-induced oxidative stress as reported previously (Berneburg et al., 1999, 2004, 2005; Lin and Beal, 2006) instead of H₂O₂ (unpublished data).

To confirm results at the molecular level and to quantify the amount of each protein in the different cell compartments, whole-cell and mitochondrial extracts were isolated from oxidatively stressed cells and subsequently analyzed by PCR and Western blotting. PCR revealed the absence of nuclear GAPDH-DNA in mitochondrial extracts (Fig. 1 F), whereas PCR using previously published mitochondrial primers (Berneburg et al., 1999, 2004, 2005; Lin and Beal, 2006) showed a strong signal for the presence of mtDNA (IS; Fig. 1 F). This was confirmed at the protein level by the presence of mitochondrial ATP-synthase in all extracts (Fig. 1 F) and the absence of nuclear phosphorylated NF-κB in mitochondrial extracts, but its presence in whole-cell extracts (Fig. 1 F). Immunoblot analysis of mitochondrial extracts for S6 ribosomal protein also revealed lack of this protein in mitochondrial extracts verifying the absence of cytoplasmic contamination (Fig. 1 F). Nevertheless, immunoblot analysis of stressed cells revealed the presence of CSA and CSB proteins in both whole-cell and mitochondrial extracts of normal cells, whereas CS-deficient cells remained negative (Fig. 1 F).

As the presence of the nuclear CSA and CSB proteins in mt has not been reported before, we used immunoelectronmicroscopy as a third method to verify our findings. Gold-labeled antibodies against CSA or CSB clearly demonstrated the localization of both proteins in oxidatively stressed mt (circles; Fig. 1 G).

To further demonstrate the location of CSA and CSB within the organelle, protease protection assays were performed (Fig. 1 H). CSA (Fig. 1 H, top) and CSB (Fig. 1 H, bottom) proteins in oxidatively stressed mt are protected from proteinase K digestion. However, disintegration of mitochondrial membranes by Triton X-100 renders CSA and CSB proteins sensitive to proteinase K digestion. These results show that CSA is located within mt and exclude unspecific association at the surface of the organelle.

The presence of mtOGG-1 and mt single-stranded DNA binding protein (mtSSBP)-1 in mt and their role in repair of oxidative mtDNA damage have previously been shown (Nishioka et al., 1999; de Souza-Pinto et al., 2001). Therefore, we investigated, if CSA and CSB directly interact with mtDNA, mtOGG-1 and mtSSBP-1 (Fig. 2). To assess binding to mtDNA and the proteins indicated, gel electrophoresis mobility shift assays (GEMSAs; Fig. 2 A), as well as coimmunoprecipitation (CoIP; Fig. 2, B–E) experiments were performed. In GEMSA experiments, labeled mtDNA shifted in the presence of extracts generated from oxidatively stressed mt with a supershift in the presence of CSA antibody (arrow; Fig. 2 A). This supershift is consistent with the presence of CSA in a mtDNA–protein complex. To confirm these results, CSA-containing protein complexes were immunoprecipitated and the presence of mtDNA was shown by DNA-specific biotin labeling. MtDNA could only be detected in CSA-containing protein complexes of oxidatively stressed mt from normal human cells in the absence of DNase digestion (Fig. 2 B). There was no detectable CoIP of CSA–mtDNA complexes in CSA-deficient extracts, in unstressed mitochondrial extracts, or in negative buffer controls, and treatment of the CSA precipitates with DNase digestion abolished the signal. CSB–mtDNA complexes were equally detectable by CoIP (Fig. 2 C). To assess direct interaction with other known mitochondrial proteins, CoIP was performed for mtOGG-1, mtSSBP-1, cytochrome oxidase (COX) IV, and ATP-Synthase (ATP-S) subunit β with CSA (Fig. 2 D) and CSB (Fig. 2 E). DNase digestion-resistant interaction of NER-associated CSA and CSB in oxidatively stressed mt could clearly be detected for mtOGG-1 and mtSSBP-1, but not for COX IV or ATP-Sβ. Such an interaction could not be shown in CSA- or CSB-deficient cells (Fig. 2, D and E, respectively). Interaction did not occur in nonstressed mt. Thus, simple binding of proteins on mtDNA at distant sites can be excluded (unpublished data).

The 4,977 bp “common” deletion of mtDNA can be generated through oxidative stress and is considered to be a marker for the presence of other mtDNA mutations (Balajee et al., 1999; Berneburg et al., 1997, 2004). The potential role of CSA and CSB proteins in protection from damage of mtDNA was assessed by repetitive exposure of normal human fibroblasts, as well as CSA- and CSB-deficient fibroblasts to sublethal UVA-irradiation and subsequent screening for mtDNA mutations using quantitative real-time PCR, as previously published (Berneburg et al., 1997, 1999, 2004, 2005; Koch et al., 2001).

We and others have shown that repetitive, nonlethal UVA irradiation induces the common deletion in normal human fibroblasts and keratinocytes after 2 wk of irradiation (Berneburg et al., 1999, 2005; Koch et al., 2001). When compared with normal human and XP-D fibroblasts, exposure of either CSA- or CSB-deficient fibroblasts to this treatment regimen prematurely induced the common deletion. More importantly, the relative frequency of mtDNA deletions was up to 10-fold higher than in normal human fibroblasts (Fig. 3). Addition of exogenous ROS scavengers can protect against UVA-induced mtDNA deletions (Berneburg et al., 1999). If the CS proteins are involved in protection from oxidative mtDNA damage, ROS scavengers should also compensate for the deficiency of CS fibroblasts. In fact, coincubation of repetitively irradiated CSA-deficient cells with vitamin E protected from induction of mtDNA mutagenesis, indicating that CSA is involved in protection from mtDNA damage resulting from oxidative stress (Fig. 3). To provide formal proof of this hypothesis, we transfected CSB-deficient fibroblasts with the WT CSB gene and exposed the cells to oxidative damage using repetitive UVA irradiation. Indeed, complementation of CSB deficiency reduced the frequency of mtDNA mutations to the level of normal human cells.

Photosensitivity of CS patients is currently explained by the known deficiency of CS cells in TC-NER. However, other age-associated symptoms such as reduced subcutaneous fat tissue cannot readily be deduced from this. To investigate whether mtDNA mutations are associated with this particular progeroid symptom in vivo, microdissected subcutaneous fat tissue from mice lacking the functional Csb or Xpa gene were assessed for the present levels of the most frequent mouse-mtDNA deletion in comparison to WT littermate controls, depending on their respective ages. Tissues from mice aged 13, 52, and 130 wk were investigated. Overall, there was an age-dependent accumulation of mtDNA mutations in all mice investigated. In addition to this, the Csb^m/m mice show a strong increase of the mtDNA deletion in subcutaneous fat tissue of aged mice that exceeded levels of 60% in all Csb^m/m animals (Fig. 4 A). This effect was highly significant (P < 0.01, Student's t test) when comparing Csb^m/m with WT and Xpa^−/− mice at 130 wk of age. Morphologically, this correlated with reduced subcutaneous fat tissue in histological sections of Csb^m/m mice at the age of 130 wk (Fig. 4 B), whereas WT and Xpa^−/− animals did not show this phenotype.

Increased mitochondrial mutations in subcutaneous fat tissue could also be detected in Csa^−/− mice (Fig. 4 C). There was a strong increase of mtDNA mutations in 104-wk-old Csa^−/− mice, whereas no increased levels were detected in younger animals, WT, as well as Xpd^TTD/TTD mice.

All PCR results could be confirmed by sequence analysis (not depicted), as well as by restriction enzyme digested with EcoRV (Fig. 4 D) and PacI (not depicted). Similar results were obtained with another mt deletion D1 (Fig. S3) as described by Tanhauser and Laipis (1995).

To address the question whether reduction of subcutaneous fat occurs via reduction of fat cell size or via reduction of fat cell number, we performed Ziehl-Neelsen staining for lipofuscin of subcutaneous fat tissue from 130-wk-old Csb^m/m mice containing high levels of mtDNA deletions (Fig. 5 A). There was an increased level of macrophages containing granular lipofuscin, whereas levels in normal fat cells were not increased (Fig. 5 B), indicating phagocytosis of cell debris. To investigate whether this may be caused by apoptosis of fat cells containing excessive levels of mtDNA mutations, we FACS-sorted UVA-irradiated fibroblasts (Fig. 5 C) and observed that CSA- and CSB-deficient fibroblasts containing high levels of mtDNA mutations undergo apoptosis at higher levels compared with WT cells containing lower levels of mtDNA mutations. This increased apoptotic cell death especially occurred in cells containing high levels of mtDNA deletions. Nonirradiated controls showed WT levels of apoptosis.

Analysis of unstressed normal human cells did not reveal a clear detectable signal for CSA and CSB proteins in mt. However, exposure of cells to oxidative stress induced by UVA or H₂O₂ showed a marked enrichment of both proteins in mt as shown by three different approaches, and in at least two different cell lines for each complementation group.

Whereas early work demonstrated no removal of cyclobutyl pyrimidine dimers and 6–4 photoproducts by NER in mt (Clayton et al., 1974; Pascucci et al., 1997), removal of oxidative damage in mt has been linked to presence of BER components (Bogenhagen et al., 2001; de Souza-Pinto et al., 2001). Interestingly, a role for CSA and CSB in repair of oxidative damage of nuclear DNA has been reported (Le Page et al., 2005; D'Errico et al., 2007; Stevnsner et al., 2008), and the presence of CSA and CSB in mt has been postulated previously, yet direct evidence has remained elusive, possibly because of low baseline levels. Low amounts at baseline and absence of mitochondrial targeting sequences (MTS) may account for lacking detection of CS proteins in mt under unstressed conditions by standard methods, including yeast proteomics (Reinders et al., 2006), although alternative import mechanisms remain possible (Stojanovski et al., 2007). Although the exact mechanism of mitochondrial import of CSA and CSB is not clear, one can speculate that upon stress conditions, a conformational change in these proteins and/or specific modification results in the formation of an internal mitochondrial targeting signal. Analysis of the primary sequence of both CSA and CSB with a program aimed at prediction of mitochondrial targeting signals (MitoProt II) suggested that such a canonical signal is not present in these proteins. The absence of an N-terminal cleavable targeting sequence is supported by our observation that the mitochondrial portion of the protein does not migrate faster on SDS-PAGE than the bulk of the protein which is obtained in the whole cell extract (Fig. 1 F). Furthermore, sequence analysis did not reveal splice variants which result in mitochondrial targeting signals. The majority of the CSB molecules were sensitive to protease treatment (Fig. 1 H). This observation suggests that only a small portion of the molecules are in internal compartments of the organelle, whereas the vast majority of the mitochondrial-associated protein is on the surface or in the process of being imported into the organelle. Thus, this outer-surface portion could represent a CSB pool of protein, ready to be imported into mt. Of note, CSA and CSB synthesized in a cell-free system could not be imported in vitro into mt isolated from fibroblasts (unpublished data). Thus, as observed in many cases in the past, it appears that the in vitro system cannot resemble the situation in intact cells.

Mitochondrial enrichment of CSA and CSB after exposure to different sources of oxidative stress indicates a functional role of CS proteins in damage processing. This enrichment cannot be explained by simple increased permeability of mt caused by treatment with H₂O₂ or UVA, as viability of cells was not affected by the oxidative stress applied and the mitochondrial marker ATP-synthetase was not detected in the cytoplasm by confocal microscopy. In addition to this, localization assays and immunoelectronmicroscopy experiments directly demonstrated localization of CSA and CSB proteins inside the mt rather than simple attachment to the outer mitochondrial membrane. This strongly supports our proposal that the CSA and CSB proteins are physically present inside mt, a localization in close proximity to mtDNA.

Mt do contain mtOGG-1, and this enzyme has been shown to be involved in repair of oxidative mtDNA damage (de Souza-Pinto et al., 2001). Furthermore, interaction of CSB with nuclear components of BER has been reported (Muftuoglu et al., 2009). In contrast, a direct interaction of mitochondrial NER-associated CSA and CSB and BER-associated mtOGG-1 and mtSSBP-1 in mt has not been shown previously. In the present study, GEMSA showed direct interaction of CSA with mtDNA of oxidatively stressed mt and CoIP experiments revealed a direct interaction of CSA and CSB with mtDNA, mtOGG-1, and mtSSBP-1. Recent work indicates separable roles for CSA in the cellular response to UV and oxidative DNA damage (Nardo et al., 2009) and mitochondrial dysfunction mediated by CSB deficiency in the pathology of CS (Osenbroch et al., 2009). Our experiments support this notion because neither CSA nor CSB interact with COX IV and ATP-Sβ upon oxidative stress. Furthermore, we could previously show that sublethal levels of oxidative stress do not significantly reduce cellular ATP-levels in human cells (Berneburg et al., 2005). Therefore, our data also argue for specific roles of CSA and CSB in response to oxidative stress by direct interaction of NER-CSA and NER-CSB with BER-mtOGG-1.

Mutations of mitochondrial DNA are involved in neurodegenerative processes (Kraytsberg et al., 2006; Lin and Beal, 2006), as well as normal aging (Schriner et al., 2005; Trifunovic et al., 2004) and premature aging of the skin (Pang et al., 1994; Berneburg et al., 1997, 1999, 2004, 2005; Koch et al., 2001; Durham et al., 2003). Accelerated induction of the “common” mtDNA deletion in deficient CSA and CSB cell lines provides evidence for a role of CS proteins in counteracting mitochondrial mutagenesis. The exact role of CSA and CSB in the mitochondrion is not clear, but delay of mtDNA mutagenesis by vitamin E indicates a role in protection from oxidatively induced mtDNA damage. Although it has been shown that in Ogg1^−/−/Csb^−/− double-knockout mice the basal levels of 7,8-dihydro-8-oxoguanine (8-oxoG) and other oxidative modifications in mtDNA are low (Trapp et al., 2007), others demonstrated deficient mitochondrial repair of 8-oxoG in Csb^−/− mouse cells and CSB-deficient human fibroblasts (Stevnsner et al., 2002). This may be caused by higher levels of 8-oxoG lesions on transfected plasmids. In our system, under oxidative stress we found premature induction of mtDNA deletions reported to be triggered by oxidative stress to be directly dependent on CS proteins because it was not detectable in NER-deficient XP-D mutated cells, but it was in CSB-deficient cells, and it was normalized by stable expression of HA-tagged CSB in CSB-deficient fibroblasts.

The fact that XP-D–mutated cells show WT levels of mtDNA deletions upon irradiation (Fig. 3) indicates that at least this protein is not necessary for protection from oxidatively induced mtDNA mutations. Together with the fact that CPD are not removed from mtDNA (Clayton et al., 1974), this indicates that there is no NER as such in mt. Whether other NER-associated proteins are present in mt remains to be determined.

Patients suffering from CS show growth and mental retardation and sensitivity of the skin to UV radiation. Interestingly, one of the most prominent clinical progeroid symptoms in CS is loss of subcutaneous fat tissue (Berneburg and Lehmann, 2001; van Hoffen et al., 2003; van der Wees et al., 2007). Particularly, age-associated loss of subcutaneous fat, especially of the buccal region is a prominent feature of normal aging. This is where loss of subcutaneous fat tissue in CS-deficient mice was also most prominent. Extrinsically aged (photoaged) skin shows increased frequencies of functionally relevant mtDNA mutations (Berneburg et al., 1997, 2004; Durham et al., 2003). In line with the notion that mtDNA mutations and age-dependent loss of subcutaneous tissue are causally linked, highly elevated levels of mtDNA mutations were found in reduced subcutaneous fat tissue of aged Csa^−/− and Csb^m/m mice, particularly in the buccal region, where loss of subcutaneous tissue was most prominent. The magnitude of mtDNA mutational levels with more than 40% (Csa^−/−) to 60% (Csb^m/m) of the mitochondrial genome being mutated far exceeded levels previously measured in other tissues under physiological conditions (Berneburg et al., 1997, 1999, 2004, 2005; Durham et al., 2003). This magnitude has only been reported in other genetic diseases caused by mtDNA defects, thus indicating that detected levels of mitochondrial mutations do have direct impact on mitochondrial physiology. Levels of mtDNA mutations in epidermis, dermis, and other organ tissues of aged Csa^−/− and Csb^m/m mice were much lower than those in subcutaneous tissue, indicating that this is a specific effect for aged subcutaneous fat tissue.

Loss of subcutaneous fat tissue could be mediated via simple reduction of fat content in cells. However, selection on the cellular level against cells containing excessive mtDNA mutations have been reported (Lu et al., 2009). Increased lipofuscin staining pointing to increased cellular turnover in vivo in mice, as well as increased apoptosis in vitro in cells with increased mtDNA mutations indicate aging-associated reduction of subcutaneous fat tissue via a mechanism that involves apoptotic cell loss. Although we cannot entirely rule out that this apoptotic cell loss is mediated by nuclear signals, the fact that apoptosis was especially detectable in irradiated fibroblasts with high levels of mtDNA deletions, and because high levels of mtDNA deletions have been shown to induce apoptotic cell death independent from nuclear signals (Bravo-Nuevo et al., 2003; Zanna et al., 2003; Dubec et al., 2008), in our eyes, it is more likely that this was mediated via mitochondrial signals.

In conclusion, these data suggest that the CSA and CSB proteins are not only present in the nucleus but they are also recruited to the mitochondrion upon oxidative stress. Mechanistically, CSA and CSB (a) directly interact with mtDNA and BER-associated mtOGG-1, as well as mtSSBP-1, (b) are involved in protection from oxidatively induced and aging-associated mtDNA mutations, and (c) protect against the age-associated reduction of subcutaneous fat tissue in Csa^−/− and Csb^m/m mice (d) via a mechanism involving increased cell turnover.

This study links a novel role of intramitochondrial CS proteins, namely the protection from oxidative mtDNA damage, to reduced subcutaneous fat tissue as segmental progeroid symptom in CS patients with implications for normal human aging, and this is performed via a mechanism using the NER proteins CSA and CSB, as well as the BER protein mtOGG-1 together with mtSSBP-1. Whether this is performed by a strong link between NER and BER remains to be shown, and the exact mechanistic function of CSA and CSB as well as the specific form of interaction with other proteins in mt still needs to be established.

Human and animal protocols were approved by the local ethics committee of the Eberhard Karls University, Tübingen, Germany.

Primary fibroblasts from normal individuals (F92, FF3, C5ROWT) and CSB patients (CS1TAN, GM739), CSA patients (CS6BR, CS3BE), and XP-D–mutated cells were cultured in Eagle's minimum essential medium (Life Technologies GmbH) containing 15% fetal calf serum (Greiner), 1% l-Glutamine, and 1% Streptomycin/Amphotericin B (Invitrogen) in a humidified atmosphere containing 5% CO₂. In addition, CSB cells (CS1AN SV-40 transformed) complemented with HA-tagged CSB (HACSB) and CSA cells (CS3BE SV-40 transformed) complemented with HA-tagged CSA (HACSA) were used. Cells were kept in 10- or 15-cm culture dishes for culture and irradiation. All cell lines used are listed in Table S1.

All chemicals were purchased from Sigma-Aldrich and applied as described previously (Berneburg et al., 1997, 1999, 2004, 2005) or as indicated.

Mt were isolated as described earlier (Jensen et al., 2003; Schriner et al., 2005). All steps are performed on ice. In brief, 5 × 10⁶ cells were trypsinized and pelleted with subsequent washing in Mito-Isolation buffer (210 mM mannitol, 70 mM sucrose, 1 mM EGTA, and 5 mM Hepes). Cell pellets were broken up by Digitonin (Sigma-Aldrich) and repeated mechanical treatment using a Teflon-homogenizer (Potters). Separation of mt from cell debris was achieved by multiple wash steps with Mito-Isolation buffer, taking mt from the supernatant. Final centrifugation at 10,000 rpm for 20 min in Mito-Isolation buffer yields mitochondrial pellet.

All procedures were performed on ice or at 4°C using ice-cold buffers.

For the preparation of whole-cell extracts, 5 × 10⁶ cells were trypsinized, and washed twice with wash buffer (210 mM mannitol, 70 mM sucrose, 1mM EGTA, 10 mM Hepes-KOH, pH 7.9, and 0.5% BSA). After the first wash, 1/30 of the (cell suspension) volume was pelleted and resuspended in lysis buffer (10 mM Hepes-KOH, pH 7.9, 0.45 M NaCl, 10% glycerol, 210 mM mannitol, 70 mM mucrose, 1 mM EGTA, 1 mM PMSF, and 1 mM DTT), while the rest was used for the isolation of mt extracts. The cell suspension was freeze-thawed 3 times (snap-freeze in liquid nitrogen/thaw at 30°C for 5 min) and the supernatant, obtained after centrifugation (10,000 rpm, 10 min), was used as whole-cell extract.

For the preparation of mitochondrial extracts, the mitochondrial pellet, isolated as described above, was resuspended in 50 µl Mito-Isolation buffer and incubated on ice for 15 min. The mitochondrial extracts with protein concentrations of 8 µg/µl were enriched with 5% BSA solution.

We used two types of antibodies for the immunodetection of the CSB protein (Santa Cruz Biotechnology, Inc.) and two types of antibodies for the immunodetection of the CSA protein (Santa Cruz Biotechnology, Inc. and Biozol). All antibodies used are given in Table S3. The analysis of the different protein extracts for the presence of CSA, CSB and XPA proteins was performed by immunoblotting with chemiluminescent detection (ECL-plus; GE Healthcare). Equal amount of proteins were separated on 8% SDS-PAGE gels for the CSB protein and 12% for the CSA and transferred to Immobilon-P membranes using 1 X Transfer buffer (25 mM Tris, 0.2 M Glycine, 20% Methanol, and 0.1% SDS, pH 8.5). The membranes were incubated overnight with the primary antibodies and for 45 min with the corresponding horseradish peroxidase-conjugated secondary antibodies.

Isolated mt were treated with 25 mg/ml proteinase K in SEM buffer (250mM sucrose, 1 mM EDTA, and 10 mM MOPS, pH 7.2) in the absence or presence of 1% Triton X-100 for 15 min at 4°C. Proteinase K was inactivated by the addition of PMSF to a final concentration of 4 mM. The samples were incubated for 10 min at 4°C. Mt were reisolated and washed with SEM buffer. In the case of Triton lyses, the supernatant was used directly for SDS-PAGE analyzes. Proteins were loaded on a 15% SDS-PAGE and analyzed by Western blotting and immunodecoration.

Procedures were conducted as described in the manufacturer's protocol (Thermo Fisher Scientific). In brief, mitochondrial extracts of oxidatively stressed fibroblasts were incubated with biotin labeled and purified mtDNA fragment (Biotin 3′ End Labeling kit; Thermo Fisher Scientific) for 20 min at 37°C, followed by incubation in the presence or absence of CSA antibody for 10 min at 37°C. Subsequently, the samples were separated by gel electrophoresis using an 8% native acrylamide gel at 120 V for 1 h in TBE buffer. MtDNA–protein complexes were blotted to a nylon membrane (Roth) at 200 mA for 1 h in TBE buffer. Shifted or supershifted biotin-labeled mtDNA was detected using a chemiluminescent nucleic acid detection module (Thermo Fisher Scientific).

CSA protein-containing complexes were isolated from mitochondrial extracts of oxidatively stressed fibroblasts with CSA antibody-associated protein A/G beads (Thermo Fisher Scientific) in PBS. Isolated CSA complexes were washed five times with cold PBS, and coprecipitated mtDNA was labeled with biotin (Biotin 3′ End Labeling kit) and incubated with or without DNase (rDNase; Machery-Nagel) for 15 min at room temperature. Samples were separated by gel electrophoresis in an 8% native acrylamide gel at 120 V for 1 h in TBE buffer. DNA–protein complexes were blotted to a nylon membrane at 200 mA for 1 h in TBE buffer. Biotin-labeled mtDNA associated to the CSA complex was detected using a chemiluminescent nucleic acid detection module.

Repetitive UVA irradiation was performed as described previously (Berneburg et al. 1999, 2005). In brief, for UVA irradiation, medium was replaced by PBS, lids were removed, and cells were exposed to radiation from a Sellamed 24000 (Sellas Medical Devices). The emission was filtered with UVACRYL (Mutzhas) and UG1 (Schott Glaswerke) and consisted of wavelengths >340 nm. The UVA output was determined with a UVAMETER (Mutzhas) and found to be ∼70 mW/cm² UVA at a tube-to-target distance of 30 cm.

To induce mtDNA mutations, cells were irradiated 3 times daily with 8 J/cm² UVA for 4 consecutive days and checked for viability by trypan blue exclusion. Cells were then aliquoted equally with one aliquot stored at −80°C until extraction of mtDNA and another aliquot plated to a 10-cm culture dish for ongoing culture and irradiation.

Total cellular DNA was extracted as described previously (Berneburg et al., 1997, 1999, 2004, 2005).

Detection of the mt common deletion was performed by real-time PCR as described previously (Koch et al., 2001). The normalized level of common deletion in UVA-irradiated cells (ΔCt) was compared with the corresponding normalized common deletion level of (sham-irradiated) unirradiated cells (controls), resulting in ΔΔCt value. Levels of common deletion in UVA-irradiated cells in relation to unirradiated cells are given as base of 2 and the negative power of ΔΔCt values (2^−ΔΔCt), as described previously (Koch et al., 2001; Berneburg et al., 2004).

Whole-skin samples from WT, xpa^−/−, and csb^−/− mice of different age groups (13, 52, and 130 wk) were microdissected under microscopic control to isolate subcutaneous tissue free of other tissue contamination. From these samples, mtDNA was extracted and detection of the mouse mtDNA deletion (D17 and D1) was performed as described previously (Tanhauser and Laipis, 1995). In brief, after DNA extraction and subsequent PCR, levels of mouse mtDNA deletion (D17 and D1) and undeleted mitochondrial DNA sequences were measured densitometrically by phosphorimager analysis. The correct sequence of PCR products was verified by DNA sequencing and digestion with restriction enzymes EcoRV and PacI (New England Biolabs). Values of mouse mtDNA deletion (D17 and D1) were normalized against corresponding values of undeleted mitochondrial DNA sequences.

Cells grown on chamber slides (BD) were treated with or without UVA light and fixed in 4% formalin at 4°C for 5 min and methanol at 0°C for 3 min. After permeabilization with 0.25% Triton X-100 in PBS with 1% donkey serum for 10 min, cells were incubated in blocking solution (0.25% Triton X-100 in PBS with 5% donkey serum). Slides were incubated with primary antibodies: mouse anti ATP-Synthetase-β (mitochondrial staining; BD) and rabbit anti-CSA (Biozol and Santa Cruz Biotechnology, Inc.) or goat anti-CSB (Santa Cruz Biotechnology, Inc.) for 2 h at RT. After washing in 0.25% Triton X-100 in PBS with 1% donkey serum for three times, slides were incubated with secondary antibodies: Cy5-labeled donkey anti–mouse and Cy3-labeled donkey anti–goat (Dianova) or Cy3-labeled donkey anti–rabbit (Dianova). Slides were rinsed (0.25% Triton X-100 in PBS with 1% donkey serum) three times and incubated with YOPRO-1 (Invitrogen) for nuclear staining. Cells were rinsed in PBS and mounted in Vectastain mounting media (Vectastain). Cellular fluorescence and localization of CSA or CSB proteins were analyzed using a Leica confocal laser scanning microscope TCS SP (Leica). Three different fluorochromes for nuclear (YOPRO-1) fluorescence (green), the blue mitochondrial fluorescence (Cy5), and the red fluorescence (Cy3) CSA or CSB proteins were used.

IEM was performed as described previously (Riess et al., 2004). Cells were fixed in periodate-lysine-paraformaldehyde for 4 h at 4°C and centrifuged. The resulting pellet was embedded in 3.5% agarose at 37°C, and then cooled on ice. Small parts of agarose blocks were dehydrated in ethanol and embedded in Lowicryl K4M (Polysciences). The blocks were cut with an ultramicrotome (Ultracut; Reichert). Ultrathin sections (50 nm) were mounted on formvar-coated nickel grids followed by incubation in blocking solution (PBS with 10% goat or donkey serum). Sections were incubated in goat anti-CSA or rabbit anti-CSB antibodies (Table S3) diluted at a ratio of 1:40 in PBS containing 0.5% BSA for 90 min. The sections were then washed in PBS/BSA. CSA-treated sections were incubated with 6-nm gold-conjugated donkey anti–goat IgG (Dianova) and anti-CSB–treated sections in 10-nm gold-conjugated goat anti–rabbit IgG (GE Healthcare). Antibodies were diluted at a ratio of 1:25 in (PBS/BSA/0.5% Tween 20). In control samples, the primary antibody was omitted. Fixation of the grids in glutaraldehyde was followed by counterstaining with uranyl acetate for 2 min and lead citrate for 30 s. Cellular morphology and localization of CSA or CSB proteins were analyzed using a Zeiss EM 109 transmission electron microscope (Carl Zeiss, Inc.) operating at 80 kV.

Average data are presented as mean and SD. Statistical analysis was performed using Excel (Microsoft). For statistical comparison of two groups, we used unpaired, two-tailed Student's t test. Differences were considered highly significant when P < 0.01. In the figures, highly significant P values are indicated by two asterisks.

Fig. S1 presents the levels of CSA and CSB in mt of unstressed cells without green nuclear staining. Fig. S2 presents the levels of CSA and CSB in oxidatively stressed cells without nuclear green staining. Fig. S3 presents the age dependent accumulation of D1 mtDNA mutation in subcutaneous fat tissue of Csa^−/− and Csb^m/m mice. Tables S1–S4 describe cell lines, primer oligonucleotides, and antibodies.

CS cell lines were kindly provided by Alan Lehmann, Brighton, UK. HA-tagged-CS cells were kindly provided by E. Citterio, Amsterdam and Wim Vermeulen, Leiden, NL.

This work was supported by grants from the Deutsche Forschungsgemeinschaft, Emmy-Noether-Programm Be 2005/2-1, 2-2, 2-3, klinische Forschergruppe Alterung KFO 142 Be 2005/3-1, 3-2, Sonderforschungsbereich SFB 728 Project C1, and GK 1033 and EU-IP FP6-512113. M. Schaller was supported by grants from the Deutsche Forschungsgemeinschaft Sch897/3, SFB 773 Project Z2, and graduate college 685, the BMBF (MedSys 0315409B), and by a NIDCR grant R01 DE017514-01. M. Röcken was supported by grants from SFB 685, Wilhelm Sander-Stiftung 205.043.2, Desutsche Krebshilfe 107128

The authors state that they have no conflicting financial interests.