Dynamic Compartmentalization of Base Excision Repair Proteins in Response to Nuclear and Mitochondrial Oxidative Stress

Abstract

DNAs harbored in both nuclei and mitochondria of eukaryotic cells are subject to continuous oxidative damage resulting from normal metabolic activities or environmental insults. Oxidative DNA damage is primarily reversed by the base excision repair (BER) pathway, initiated by N-glycosylase apurinic/apyrimidinic (AP) lyase proteins. To execute an appropriate repair response, BER components must be distributed to accommodate levels of genotoxic stress that may vary considerably between nuclei and mitochondria, depending on the growth state and stress environment of the cell. Numerous examples exist where cells respond to signals, resulting in relocalization of proteins involved in key biological transactions. To address whether such dynamic localization contributes to efficient organelle-specific DNA repair, we determined the intracellular localization of the Saccharomyces cerevisiae N-glycosylase/AP lyases, Ntg1 and Ntg2, in response to nuclear and mitochondrial oxidative stress. Fluorescence microscopy revealed that Ntg1 is differentially localized to nuclei and mitochondria, likely in response to the oxidative DNA damage status of the organelle. Sumoylation is associated with targeting of Ntg1 to nuclei containing oxidative DNA damage. These studies demonstrate that trafficking of DNA repair proteins to organelles containing high levels of oxidative DNA damage may be a central point for regulating BER in response to oxidative stress.

Oxidative DNA damage, which occurs frequently in all cells, is linked to aging and human disease, such as cancer and various degenerative disorders (6, 13, 45, 81, 83). Reactive oxygen species (ROS) are a by-product of normal cellular metabolic processes that can cause oxidative damage to DNA, lipids, and proteins (79). Unrepaired oxidative DNA lesions can result in mutations and lead to arrest of both DNA replication and transcription (34). In order to combat such continuous insults to the genome, cells have evolved DNA repair and DNA damage tolerance pathways (2).

Base excision repair (BER) is the primary process by which oxidative DNA damage is repaired (74, 90). BER is initiated by the recognition and excision of a base lesion by an N-glycosylase, resulting in an apurinic/apyrimidinic (AP) site (47, 48). The resulting AP site is processed by an AP endonuclease or an AP lyase, which cleaves the sugar-phosphate DNA backbone on the 5′ side or 3′ side of the AP site, respectively (5). Subsequent processing involving DNA repair polymerases replaces the excised nucleotides, and DNA ligase completes the repair process (8).

Very little is known about how eukaryotic cells regulate events that initiate BER in response to oxidative stress. Deleterious oxidative DNA damage can occur in both nuclear and mitochondrial genomes, adding a level of complexity to this cellular response. In this case, the intracellular localization of BER proteins would be regulated dynamically in response to the introduction of either nuclear or mitochondrial DNA damage. Controlled protein localization has been implicated in the regulation of a number of critical cellular processes (24, 29, 37, 58). For example, under normal growth conditions the human c-Abl protein tyrosine kinase is cytoplasmic, but in response to cellular stress that results in DNA damage, c-Abl translocates into the nucleus, where it induces apoptosis (88). Yap1 is a critical transcription factor in the oxidative stress response in budding yeast (Saccharomyces cerevisiae) that is imported from the cytoplasm into the nucleus, where it regulates many stress response genes in response to oxidative stress (43). The human DJ-1 protein, mutations of which are implicated in Parkinson's disease (9), translocates to mitochondria following oxidative stress in order to protect against cytotoxicity (11, 46). Since subcellular localization is a regulatory component of numerous non-DNA repair pathways, it is possible that DNA repair is regulated in a similar manner.

If subcellular localization of BER proteins is regulated, then such events might be modulated through posttranslational modification. Phosphorylation, myristoylation, and numerous other modifications affect nuclear localization of certain proteins, such as c-Abl, FoxO proteins, and p53 proteins (10, 71, 88). Another posttranslational modification that has been implicated in modulation of intracellular localization, especially that of nuclear proteins, is modification by the ubiquitin-like protein SUMO (22, 30). Several proteins involved in DNA repair and maintenance are sumoylated, conferring a range of functions (55). For example, sumoylation of human thymine DNA glycosylase affects its glycosylase activity and localization to subnuclear regions (28, 54). SUMO modification also affects the nuclear localization of proteins such as mammalian heat shock transcription factor and the repressor of transcription, TEL, in response to environmental stress (27, 33, 42, 82). Sumoylated heat shock transcription factor colocalizes with nuclear stress granules, facilitating transcription of specific heat shock genes (33). SUMO modification of TEL is required for TEL export from the nucleus in response to cellular stresses, such as heat shock and exposure to UV radiation (27). Thus, SUMO modification is a major mechanism for regulation of subcellular protein localization.

Saccharomyces cerevisiae has been utilized extensively to investigate the mechanisms that underlie DNA repair, as the DNA damage management pathways are conserved between yeast and humans (18, 53). To determine whether targeting of BER proteins to the appropriate organelle harboring oxidative DNA damage is likely to represent a general regulatory component of DNA repair, we evaluated the localization of BER proteins in response to oxidative stress. This study focused on the S. cerevisiae BER proteins, Ntg1 and Ntg2, which are both homologs of Escherichia coli endonuclease III, possessing N-glycosylase/AP lyase activity that allows recognition and repair of oxidative base damage (primarily of pyrimidines) as well as abasic sites (3, 25, 73, 89). Because Ntg1 and Ntg2 play an important role in the repair of oxidative DNA damage in S. cerevisiae, the aim of this study was to determine how oxidative stress and sumoylation influence subcellular localization of these proteins. Consistent with the presence of predicted nuclear localization signals and a mitochondrial targeting sequence (4, 89), Ntg1 is found in both the nucleus and mitochondria (1, 90). In contrast, Ntg2, which contains only a putative nuclear localization signal but no mitochondrial targeting sequence, is localized exclusively to the nucleus (1, 90).

In this study, we evaluated the localization of Ntg1 and Ntg2 following exposure to nuclear and/or mitochondrial oxidative stress. The results show that the localization of Ntg1 is dynamically regulated in response to nuclear and mitochondrial oxidative stress. However, Ntg2 remains nuclear regardless of the oxidative stress state of the cell. Importantly, we provide evidence that dynamic localization of Ntg1 is a response to DNA damage rather than a general response to ROS. Additionally, sumoylation is associated with nuclear localization of Ntg1 that occurs in response to oxidative stress. These results indicate that the localization of BER proteins can likely be regulated by the introduction of nuclear and mitochondrial oxidative DNA damage and suggest that sumoylation plays a role in modulating the localization of BER proteins.

MATERIALS AND METHODS

Strains, media, and growth conditions.

Haploid S. cerevisiae strains and all plasmids used in this study are listed in Table 1. Yeast cells were cultured at 30°C in rich YPD medium (1% yeast extract, 2% peptone, 2% dextrose, 0.005% adenine sulfate, and 2% agar for plates) or YPGal medium (1% yeast extract, 2% peptone, 2% galactose, 0.005% adenine sulfate, and 2% agar for plates). In order to introduce plasmids or integrated chromosomal gene modifications, yeast cells were transformed by a modified lithium acetate method (35).

TABLE 1.

Strains and plasmids used in this study

Strain or plasmid	Description or genotype	Reference(s)
Strains
FY86 (ACY193)	MATaura3-52 leu2Δ1 his3Δ200	90
ACY737	MATaulp1ts ulp2Δ lys2 trp1	70
SJR751 (DSC0025)	MATα ade2-101oc his3Δ200 ura3ΔNco lys2ΔBgl leu2-R	75
SJR1101 (DSC0051)	MATα ade2-101oc his3Δ200 ura3ΔNco lys2ΔBgl leu2-R ntg1Δ::LEU2 ntg2Δ::hisG apn1Δ::HIS3 rad1Δ::hisG	75
DSC0221	MATaulp1ts ulp2Δ lys2 trp1 GAL-HA-SMT3 GAL-GST-NTG1	This study
DSC0222	MATaulp1ts ulp2Δ lys2 trp1 GAL-HA-SMT3 GAL-GST-NTG2	This study
DSC0282	MATaura3-52 leu2Δ1 his3Δ200 ntg1::KAN bar1::HYG	This study
DSC0283	MATaura3-52 leu2Δ1 his3Δ200 ntg2::blasticidin bar1::HYG	This study
DSC0291	MATaleu2Δ1 his3Δ200 ntg1::blasticidin bar1::HYG pNTG1-GFP [rho0]	This study
DSC0295	MATahis3Δ1 leu2Δ0 met 15Δ0 ura3Δ0; TET-repressible C-terminally TAP-tagged Ntg1	This study
DSC0296	MATahis3Δ1 leu2Δ0 met 15Δ0 ura3Δ0; TET-repressible C-terminally TAP-tagged Ntg2	This study
YSC1178-7499106 (DSC0297)	MATahis3Δ1 leu2Δ0 met15Δ0 ura3Δ0; C-terminally TAP-tagged Ntg1	Open Biosystems
YSC1178-7502650 (DSC0298)	MATahis3Δ1 leu2Δ0 met15Δ0 ura3Δ0; C-terminally TAP-tagged Ntg2	Open Biosystems
BY4147 (DSC0313)	MATahis3Δ1 leu2Δ0 met15Δ0 ura3Δ0	Open Biosystems
Plasmids
pNTG1-GFP	NTG1-GFP; 2μm URA3 Ampr	41, 84, 90
pNTG2-GFP	NTG2-GFP; 2μm URA3 Ampr	41, 84, 90
pNTG1K364R-GFP	NTG1 K364R-GFP; 2μm URA3 Ampr	This study
pFA-KMX4	KanrCEN Ampr	80
pAG32	Hygromycin B phosphotransferase MX4r Ampr	23
pCM225	tet07CEN KanMX4 Ampr	7
p1375	3HA CEN TRP1 Ampr	51
p2245	pGAL1-GST; CEN TRP1 Ampr	51

The pPS904 green fluorescent protein (GFP) expression vector (2μm URA3) was employed for generation of C-terminally tagged Ntg1-GFP and Ntg2-GFP fusion proteins (41). The S. cerevisiae haploid strain FY86 was utilized for all localization studies (84). ΔNTG1 and ΔNTG2 strains (DSC0282 and DSC0283, respectively) were generated by precisely replacing the NTG1 and NTG2 open reading frames in FY86 with the kanamycin antibiotic resistance gene (pFA-KMX4 [80]; selected with 150 mg/liter G418 [U.S. Biologicals]) and blasticidin antibiotic resistance gene (BsdCassette vector pTEF1/Bsd 3.6 kb [Invitrogen]; selected with 100 mg/liter blasticidin S HCl [Invitrogen]), respectively. Plasmids encoding Ntg1-GFP or Ntg2-GFP were introduced into ΔNTG1 or ΔNTG2 cells. Plasmid mutagenesis of Ntg1-GFP to create Ntg1 K364R-GFP was performed using a QuikChange II site-directed mutagenesis kit (Stratagene), and these plasmids were introduced into ΔNTG1 cells.

For studies of cells lacking mitochondrial DNA, a [rho⁰] yeast strain (DSC0291) was generated by incubating 4 × 10⁶ ΔNTG1 cells in ethidium bromide as previously described (17). Following this incubation, cells were stained with 4′,6-diamidino-2-phenylindole dihydrochloride (DAPI; Sigma) and MitoTracker Red CMXRos stain (Invitrogen) and evaluated via fluorescence microscopy in order to verify that no mitochondrial DNA was present.

Haploid yeast strains expressing integrated genomic copies of C-terminally tandem affinity purification (TAP)-tagged Ntg1 and Ntg2 were obtained from Open Biosystems (Ntg1-TAP [DSC0297] and Ntg2-TAP [DSC0298]). A tetracycline-repressible promoter (Tet off) from the plasmid pCM225 was integrated at the N termini of the NTG1 and NTG2 products by use of the kanamycin resistance gene to generate tetracycline-repressible Ntg1-TAP and tetracycline-repressible Ntg2-TAP strains (DSC0295 and DSC0296, respectively) as previously described (7). Cells expressing galactose-inducible Smt3-HA and Ntg1-GST (DSC0221) or Smt3-HA and Ntg2-GST (DSC0222) were generated by integrating the hemagglutinin (HA) tag from the vector p1375 and the GAL promoter and glutathione S-transferase (GST) tag from the vector p2245 (51) at the C termini of the SMT3 and NTG1 or NTG2 products in the haploid strain ACY737 (70). ACY737 contains mutations in the sumoylation-deconjugating enzymes, Ulp1 and Ulp2, which can aid in the isolation of sumoylated proteins (70).

Exposure to DNA damaging agents.

Cells were grown in 5 ml YPD to a density of 5 × 10⁷ cells/ml, centrifuged, and washed with water. Cells were then resuspended in 5 ml water containing the appropriate DNA damaging agent, i.e., 2 to 20 mM H₂O₂ (Sigma), 25 to 55 mM methyl methanesulfonate (MMS; Sigma), or 10 μg/ml antimycin A (Sigma). Cells were exposed to an agent(s) for 1 hour at 30°C. Cytotoxicities of agents were evaluated by growing cells in agent, plating cells, and counting colonies.

Fluorescence microscopy.

For all experiments, cells (grown and treated as described above) were treated as follows: no treatment, 5 mM H₂O₂, 10 mM H₂O₂, 20 mM H₂O₂, 25 mM MMS, 55 mM MMS, 10 μg/ml antimycin, 5 mM H₂O₂ plus 10 μg/ml antimycin, 10 mM H₂O₂ plus 10 μg/ml antimycin, or 20 mM H₂O₂ plus 10 μg/ml antimycin. During exposure to DNA damaging agents, cultures were also incubated with 25 nM MitoTracker in order to visualize mitochondria. Following washes, cells were placed in 1 ml of water containing 1 μg DAPI to visualize DNA and then incubated for 5 min at room temperature. Cells were washed and analyzed by direct fluorescence confocal microscopy, employing a Zeiss LSM510 Meta microscope. Images were analyzed using Carl Zeiss LSM image browser software, and cells were evaluated for nuclear only or nuclear plus mitochondrial Ntg1-GFP or Ntg2-GFP localization. Mitochondrion-only localization was negligible. At least 200 cells were counted for each strain and treatment condition, and each microscopic evaluation was repeated at least twice. Standard deviations were calculated for each strain and treatment condition. The image analysis software program Metamorph 6.2 was utilized in order to quantify the intensities of GFP in nuclei and mitochondria of individual cells. Mitochondrial GFP intensities were determined by subtracting nuclear GFP intensity from the total cellular GFP intensity. The fraction of cells with a mitochondrial GFP intensity score of >500 was determined for cells exposed to H₂O₂ and H₂O₂ plus antimycin, and the t test was employed to determine P values.

Measurement of ROS levels by flow cytometry.

For all experiments, cells were grown and treated as described above. Following exposure to DNA damaging agents, cells were washed with water and resuspended in YPD at a density of 2 × 10⁷ cells/ml. Dihydroethidium (DHEt) was added to the YPD to a concentration of 160 μM to detect cellular superoxide (67), MitoSox (Molecular Probes) was added to the YPD to a concentration of 5 μM per the manufacturer's instructions to detect mitochondrial superoxide, or cells were left untreated. Cells were incubated for 45 min in the fluorescent dye, washed, and resuspended in 2 ml phosphate-buffered saline (PBS). The fluorescence intensity of 10,000 cells for each strain and condition was assessed by employing a BD LSR II flow cytometer (BD Biosciences). Excitation and emission wavelengths employed to evaluate cells were 488 nm and 595 nm, respectively, for DHEt and 488 nm and 575 nm, respectively, for MitoSox.

Ntg1 and Ntg2 analysis.

Purification of TAP-tagged Ntg1 and Ntg2 was achieved as follows. Four liters of tetracycline-repressible Ntg1-TAP (DSC0295) or Ntg2-TAP (DSC0296) was grown in YPD to a density of 5 × 10⁷ cells/ml without tetracycline in order to overproduce Ntg1-TAP and Ntg2-TAP. Cells were then pelleted and washed with water. Cell pellets were frozen at −80°C. A version of the previously published TAP method was utilized (63), with the following modifications. Cell pellets were crushed with a mortar and pestle, and powdered yeast lysate was suspended in 10 ml buffer A (10 mM K-HEPES, pH 7.9, 10 mM KCl, 1.5 mM MgCl₂, 0.5 mM dithiothreitol, 0.5 mM phenylmethylsulfonyl fluoride [PMSF], 2 mM benzamidine, 1 μM leupeptin, 2.6 μM aprotinin) with 5 mM N-ethylmaleimide (Sigma) and 10 mM iodoacetamide (Sigma). Lysate was centrifuged at 3,000 × g for 10 min, and supernatant was recentrifuged at 88,000 × g for 1 1/2 h. Dialysis, incubation with immunoglobulin G (IgG) beads, and incubation with tobacco etch virus (AcTEV; Invitrogen) protease were performed as instructed. Western analysis was performed on 50 μl of eluate. Anti-TAP antibody (1:3,333 dilution; Open Biosystems) was employed for Western analysis. Ntg2-TAP migrates as a slightly smaller species in Western analysis than the predicted TAP-tagged Ntg2 (∼53 kDa), while Ntg1-TAP migrates at its expected size of 55 kDa.

In order to purify GST-tagged Ntg1 and Ntg2, 1 liter of cells expressing galactose-inducible Smt3-HA and Ntg1-GST (DSC0221) or galactose-inducible Smt3-HA and Ntg2-GST (DSC0222) was grown in YPGal to a density of 5 × 10⁷ cells/ml. Cells were centrifuged, and pellets were washed and frozen at −80°C. Cell pellets were crushed with a mortar, and powdered lysate was suspended in 500 μl PBS with 0.5 mM PMSF and 3 μg/ml (each) leupeptin and aprotinin. Lysate was centrifuged at 3,000 × g for 10 min, and supernatant was applied to 150 μl washed glutathione-Sepharose 4 Fast Flow beads (Amersham Biosciences). Beads and lysate were incubated at 4°C overnight. Beads were washed three times with 1 ml PBS, and then 50 μl of 1× sodium dodecyl sulfate-polyacrylamide gel electrophoresis loading buffer (50% [wt/vol] glycerol, 10% [wt/vol] sodium dodecyl sulfate, 1% [vol/vol] 2-mercaptoethanol, 0.1 M Tris-HCl, pH 6.8, 0.1% bromophenol blue) was applied to beads to elute the bound fraction. Western analysis was performed on 20 μl of eluate. Anti-HA (1:1,000 dilution; Covance) and anti-GST (1:1,000 dilution; Oncogene) antibodies were employed for immunoblotting.

Sucrose gradient subcellular fractionation.

In order to fractionate yeast cells into nuclear and mitochondrial preparations, 1 liter of cells expressing tetracycline-repressible Ntg1-TAP (DSC0295) or tetracycline-repressible Ntg2-TAP (DSC0296) was grown in YPD to a density of 5 × 10⁷ cells/ml without tetracycline in order to overproduce Ntg1-TAP or Ntg2-TAP, respectively. Crude mitochondrial and nuclear protein lysate fractions were generated using a differential centrifugation protocol as described previously (14). Following this procedure, mitochondrial fractions were further purified using sucrose gradient centrifugation (61). Solutions of 20%, 40%, and 60% (wt/vol) sucrose in 10 mM Tris-HCl, pH 7.5, 0.1 mM EDTA, and 1 mM PMSF were prepared. The crude mitochondrial pellet was resuspended in the 20% sucrose solution. The 60% sucrose solution was placed in the bottom of a Beckman Ultraclear centrifuge tube, followed by the 40% sucrose solution and the 20% sucrose solution containing mitochondria. The tubes were centrifuged at 100,000 × g for 1 h at 4°C. The mitochondria were removed from the 40%-60% interface, concentrated, resuspended in storage buffer (0.6 M sorbitol, 20 mM HEPES-KOH, pH 7.4, 20% glycerol, 5 mM N-ethylmaleimide, 10 mM iodacetamide, 5 mg/ml aprotinin, 5 mg/ml leupeptin, and 0.1 M PMSF), and stored at −80°C.

The crude nuclear pellets were further purified using sucrose step gradient purification as previously described (66). The step gradient contained solutions of 58.2%, 68.8%, 71.9%, and 78.7% sucrose in sucrose buffer (8% polyvinylpyrrolidone 40, 11.5 mM KH₂PO₄, 8.4 mM K₂HPO₄, 0.75 mM MgCl₂, pH 6.53). The nuclei were removed from the 71.9%-78.7% sucrose interface, concentrated, and resuspended in storage buffer. Western analysis was performed using 10 μg of nuclear or mitochondrial protein lysate. Anti-TAP, anti-Por1 (1:25,000 dilution; MitoSciences), and anti-Nop1 (1:25,000 dilution; EnCor) antibodies were employed for Western analysis. Anti-Nop1 and anti-Por1 antibodies were used to ensure enrichment of nuclear (Nop1) and mitochondrial (Por1) fractions (62). To optimize visualization, Western blot exposures were variable for each protein analyzed. Analysis of Western blots by chemiluminescence was employed in order to determine the change in sumoylated Ntg1 in the nuclear fraction. The ratio of modified to unmodified Ntg1 in the nuclear fraction was determined, and values for each condition were normalized to that for the no-treatment condition. The standard error of the mean was calculated for each strain and treatment condition, and the t test was employed to determine P values.

Functional analysis of Ntg1 in vivo.

To assess the function of Ntg1 in vivo, we utilized BER⁻/nucleotide excision repair-negative (NER⁻) (ntg1 ntg2 apn1 rad1) cells (SJR1101/DSC0051), which are highly sensitive to oxidative stress (75). BER⁻/NER⁻ cells containing Ntg1-GFP or Ntg1 K364R-GFP plasmid were assessed for the ability of the episomal Ntg1 protein to function in vivo and decrease the sensitivity of these cells to treatment with H₂O₂. Cytotoxicity assays were carried out as described above for exposure to DNA damaging agents. The steady-state level of each GFP fusion protein was assessed by immunoblotting whole-cell lysates with a polyclonal anti-GFP antibody (1:5,000 dilution) (72). Anti-phosphoglycerate kinase antibody (1:5,000 dilution; Invitrogen) was utilized to determine the relative levels of protein loaded per lane.

RESULTS

Ntg1 and Ntg2 localization under normal growth and oxygen conditions.

To determine whether changes in the subcellular distribution of Ntg1 and Ntg2 occur in response to oxidative stress-induced DNA damage, the localization of Ntg1-GFP and Ntg2-GFP was evaluated in live yeast cells. Under normal growth conditions, Ntg1 is localized to both nuclei and mitochondria, while Ntg2 localization is exclusively nuclear (1, 90). This localization pattern was verified by analyzing cells expressing either Ntg1-GFP or Ntg2-GFP by direct fluorescence microscopy. As expected, Ntg1-GFP was localized to both nuclei and mitochondria (Fig. 1A), whereas Ntg2-GFP localization was strictly nuclear (Fig. 1B). To biochemically confirm the organellar distribution of Ntg1 and Ntg2, sucrose gradient subcellular fractionation (see Materials and Methods) was performed on lysates from cells expressing Ntg1-TAP or Ntg2-TAP to separate nuclear and mitochondrial fractions. Nuclear and mitochondrial protein lysate fractions were evaluated for purity by use of antibodies against a nuclear protein, Nop1, and a mitochondrial membrane protein, Por1 (62). Mitochondrial fractions were free of nuclear proteins, as indicated by the detection of Por1 but not Nop1 (Fig. 1C and D). Nuclear fractions were enriched for nuclear proteins, with some mitochondrial contamination (Fig. 1C and D). These results were expected, as cytoplasmic contaminants have routinely been documented in conjunction with nuclear fractionation of S. cerevisiae (57, 65, 91). The localization of Ntg1-TAP and Ntg2-TAP was determined by probing nuclear and mitochondrial fractions. Sucrose gradient subcellular fractionation verified that Ntg1 was present in both nuclear and mitochondrial fractions (Fig. 1C), and Ntg2 was detected only in nuclear fractions (Fig. 1D).

FIG. 1.

Subcellular localization of Ntg1 and Ntg2 under normal growth conditions. (A and B) Localization of GFP-tagged proteins was assessed via direct fluorescence microscopy. GFP, DAPI, MitoTracker (MT), and merged images of cells expressing Ntg1-GFP or Ntg2-GFP are displayed. (C and D) Sucrose gradient subcellular fractionation (see Materials and Methods) and Western analysis were performed on Ntg1-TAP and Ntg2-TAP cells. Antibodies to Nop1 (nuclear marker protein), Por1 (mitochondrial marker protein), and the calmodulin domain of the TAP tag (to detect Ntg1 or Ntg2) were utilized to detect proteins present in total cell lysate (TCL), nuclear (Nuc), and mitochondrial (Mito) fractions.

Nuclear and mitochondrial oxidative stress induction by H₂O₂, antimycin, and MMS.

ROS levels increase in response to DNA damage in cells exposed to genotoxic agents, including MMS, UV light, and H₂O₂ (19, 67, 68). In order to determine whether increased ROS levels influence the localization of Ntg1 and Ntg2, wild-type cells were exposed to H₂O₂ to directly increase oxidative stress or to the DNA-alkylating agent MMS to indirectly increase ROS levels in response to DNA damage (68). In addition, cells were exposed to H₂O₂ plus antimycin to increase mitochondrial oxidative stress. Antimycin blocks oxidative phosphorylation (60), and exposure of cells to H₂O₂ plus antimycin increases oxidative stress, leading to induced oxidative DNA damage in yeast mitochondria (17). The relative levels of cellular ROS in different cellular compartments were determined following exposure to H₂O₂, H₂O₂ plus antimycin, and MMS by use of the fluorescent probes DHEt and MitoSox. DHEt is a general cellular superoxide probe (78), whereas MitoSox accumulates in the mitochondrial matrix, allowing determination of superoxide levels specifically in mitochondria (40). Analysis of cells by flow cytometry revealed that H₂O₂ exposure resulted in elevated levels of cellular superoxide compared to those in unexposed cells (Fig. 2A) but did not increase levels of mitochondrial superoxide (Fig. 2B). Flow cytometry analysis also revealed that H₂O₂ plus antimycin exposure resulted in a general cellular increase in superoxide levels, including an increase in mitochondrial superoxide, revealed by both the DHEt and MitoSox fluorescent probes (Fig. 2A and B). Exposure of cells to nonoxidative DNA damaging agents, such as MMS and UV light, can also increase cellular ROS levels (19, 67, 68). Consistent with this observation, exposure of cells to MMS resulted in a substantial elevation in both total cellular and mitochondrial superoxide levels compared to those in untreated controls (Fig. 2C and D). Evaluation of mitochondrial superoxide levels following treatment with MMS revealed two subpopulations of cells, each containing levels of mitochondrial superoxide higher than those observed with no treatment (Fig. 2D). These two subpopulations may represent cell stress and death response groups. Collectively, these results demonstrate that oxidative stress can be targeted to nuclei or mitochondria by exposure to specific agents. Importantly, a combination of H₂O₂ and antimycin or MMS exposure induces mitochondrial oxidative stress in a manner that is distinct from the primarily nuclear oxidative stress that results from exposure to H₂O₂ alone.

FIG. 2.

Flow cytometric analysis of cells to determine intracellular ROS levels following nuclear or mitochondrial oxidative stress. (A and B) Cells were left untreated (NT) (red) or exposed to 20 mM H₂O₂ (green) or 20 mM H₂O₂ plus 10 μg/ml antimycin (blue) and incubated with DHEt or MitoSox to assess relative levels of total cellular superoxide (DHEt) or mitochondrial superoxide (MitoSox). (C and D) Cells were left untreated (red) or exposed to 55 mM MMS (green) and incubated with DHEt or MitoSox.

Relocalization of Ntg1 in response to increased nuclear and mitochondrial oxidative stress.

In order to assess whether the steady-state localization of Ntg1 is altered in response to nuclear oxidative stress, Ntg1-GFP localization was evaluated before and after a 1-hour induction of oxidative stress with various concentrations of H₂O₂ (Fig. 3A and B). The cytotoxicities for H₂O₂ exposures of 0 mM, 5 mM, 10 mM, and 20 mM were 0%, 64%, 68%, and 75%, respectively (data not shown). The localization of Ntg1-GFP was assessed by direct fluorescence microscopy. As shown in Fig. 3B, Ntg1-GFP appeared more enriched in nuclei upon exposure to H₂O₂. In order to provide a quantitative measure of Ntg1-GFP localization, the subcellular localization of Ntg1-GFP was designated nuclear only or nuclear plus mitochondrial, based on colocalization with nuclear DAPI, mitochondrial DAPI, and MitoTracker staining in all cells displaying a GFP signal. The percentage of cells with nuclear only or nuclear plus mitochondrial localization of Ntg1-GFP was determined for several hundred cells for each treatment group. A dose-dependent increase in nuclear only Ntg1-GFP was observed following H₂O₂-induced nuclear oxidative stress (Fig. 3D). This result correlated with a dose-dependent decrease in the number of cells with a nuclear plus mitochondrial distribution of Ntg1-GFP, reflecting a decrease in mitochondrial localization of Ntg1. These results suggest that Ntg1 can be targeted to nuclei in response to nuclear oxidative stress.

FIG. 3.

Subcellular localization of Ntg1 following exposure to nuclear and mitochondrial oxidative stress. Localization of GFP-tagged Ntg1 was assessed via direct fluorescence microscopy following exposure to the indicated oxidative stress agent for 1 h. (A) GFP images of untreated cells expressing Ntg1-GFP. (B) GFP images of cells exposed to 20 mM H₂O₂. (C) GFP images of cells exposed to 20 mM H₂O₂ plus 10 μg/ml antimycin A. Arrows indicate increased mitochondrial Ntg1 localization observed by Metamorph image analysis. (D) Ntg1-GFP localization analysis. Cells were left untreated (NT) or were exposed to H₂O₂, MMS, and/or antimycin (Ant) as indicated (see Materials and Methods). Localization of Ntg1-GFP to nuclei only (nuclear) or nuclei plus mitochondria (nuc + mito) was determined for each cell and plotted as a percentage of the total cells evaluated. Error bars represent standard deviations.

Oxidative stress can be induced in mitochondria by exposing cells to H₂O₂ in combination with antimycin (Fig. 2B), resulting in increased mitochondrial oxidative DNA damage (17). We treated cells with H₂O₂ plus antimycin to determine whether elevated mitochondrial ROS trigger increased localization of Ntg1 to mitochondria. As shown in Fig. 3C, localization of Ntg1-GFP to mitochondria increased following H₂O₂-plus-antimycin-induced mitochondrial oxidative stress. The intensity of the GFP signal located in mitochondria of cells exposed to H₂O₂ plus antimycin was statistically greater than the intensity of the GFP signal located in mitochondria of H₂O₂-induced cells, as determined via image analysis using the software program Metamorph 6.2. Specifically, the fraction of cells containing a mitochondrial GFP intensity score of >500 was significantly greater for cells exposed to H₂O₂ plus antimycin (0.78 ± 0.08) than for cells exposed to H₂O₂ (0.63 ± 0.09) (P value = 0.04). These data indicate that Ntg1 localization is influenced by mitochondrial ROS. In addition to increased mitochondrial localization, Ntg1 nuclear localization was increased following exposure to low doses of H₂O₂ plus antimycin. H₂O₂ plus antimycin induces oxidative stress not only in mitochondria but also in nuclei, thus increasing Ntg1 nuclear localization as well (Fig. 2A and B).

Under normal growth conditions, Ntg2 is localized exclusively to nuclei (1, 90). Therefore, it was of interest to determine whether Ntg2 localization was affected by nuclear or mitochondrial oxidative stress. The localization of Ntg2-GFP was examined following exposure to nuclear (H₂O₂) or mitochondrial (H₂O₂ plus antimycin) oxidative stress. Ntg2-GFP localization remained exclusively nuclear following either nuclear or mitochondrial oxidative stress (data not shown), indicating that Ntg2 is not responsive to changes in either nuclear or mitochondrial oxidative stress.

Relocalization of Ntg1 in response to MMS exposure.

To determine whether other DNA damaging agents that do not directly cause oxidative DNA damage are also capable of inducing a change in the localization of Ntg1, cells were exposed to the DNA-alkylating agent MMS, resulting in an increase in intracellular ROS (68) (Fig. 2C and D). Survival of cells treated with 0 mM, 25 mM, and 55 mM MMS was 100%, 30%, and 3%, respectively (data not shown). An increase in nuclear only localization of Ntg1-GFP was observed following exposure to MMS (Fig. 3D) (P values of <0.04 for comparing nuclear only localization for no-treatment and MMS exposures). This result indicates that in addition to the ability of Ntg1 to respond to oxidative stress caused by H₂O₂ exposure, Ntg1 also responds to oxidative stress caused by DNA damaging agents, such as MMS, that do not directly introduce oxidative DNA damage.

Oxidative stress-induced relocalization of Ntg1 to mitochondria is due to a DNA damage response.

Oxidative stress could provoke a change in localization of Ntg1 via a direct response to elevated levels of ROS or in response to the presence of oxidative DNA damage. In order to distinguish between these possibilities, [rho⁰] cells were generated as described in Materials and Methods. [rho⁰] mitochondria do not contain DNA, whereas [rho⁺] mitochondria contain intact DNA (20). The absence of mitochondrial DNA in [rho⁰] cells was confirmed by direct fluorescence microscopy, as evidenced by the absence of any extranuclear DAPI staining (Fig. 4A). In [rho⁰] cells, a lack of Ntg1 mitochondrial localization following increased mitochondrial oxidative stress (exposure to H₂O₂ plus antimycin) would indicate that Ntg1 responds to the presence of mitochondrial oxidative DNA damage rather than to ROS. Flow cytometric analysis of [rho⁰] cells revealed that mitochondrial superoxide levels increased in response to H₂O₂ plus antimycin exposure (Fig. 4B). Regardless of exposure to ROS-generating agents, fewer [rho⁰] mitochondria than [rho⁺] mitochondria contained Ntg1-GFP, as determined by colocalization of GFP with MitoTracker (Fig. 4A) and quantification of cells with nuclear or nuclear plus mitochondrial GFP-Ntg1 localization (Fig. 4C). The results indicate that [rho⁰] cells subjected to increasing levels of mitochondrial oxidative stress did not exhibit a change in Ntg1 localization (Fig. 4C). In contrast, [rho⁺] cells subjected to the same mitochondrial oxidative stress conditions displayed a significant increase in mitochondrial localization of Ntg1. Exposure of [rho⁺] cells to H₂O₂ plus antimycin results in increased mitochondrial oxidative DNA damage (17) caused by increased mitochondrial ROS (Fig. 2). The difference in Ntg1 localization observed between [rho⁰] and [rho⁺] cells indicates that mitochondrial oxidative stress induces DNA damage that results in the relocalization of Ntg1 to mitochondria. Importantly, these data suggest that the mitochondrial localization of Ntg1 is directed by the presence of mitochondrial oxidative DNA damage and not simply by elevated levels of mitochondrial ROS.

FIG. 4.

Mitochondrial localization of Ntg1 is influenced by mitochondrial oxidative DNA damage. [rho⁺] cells and [rho⁰] cells were analyzed in order to assess the change in localization of Ntg1-GFP in the presence and absence of mitochondrial DNA in response to mitochondrial oxidative stress. (A) Fluorescence microscopy was performed in order to confirm the [rho] status of the cells. Panels from left to right show GFP (Ntg1-GFP), DAPI (DNA), MitoTracker, and merged images. (B) Flow cytometry analysis of ROS levels in [rho⁰] cells. Cells were left untreated (NT) (red) or were exposed to 20 mM H₂O₂ plus 10 μg/ml antimycin (Ant) (green) and incubated with MitoSox to assess the levels of mitochondrial superoxide. (C) Quantification of Ntg1-GFP localization in [rho⁺] or [rho⁰] cells. Cells were left untreated (NT) or were exposed to H₂O₂ and/or antimycin (Ant) as indicated (see Materials and Methods). Localization of Ntg1-GFP to nuclei only (nuclear) or nuclei plus mitochondria (nuc + mito) was determined for each cell and plotted as a percentage of the total cells evaluated. Error bars represent standard deviations in the data. The gray line references an overall higher percent localization of Ntg1-GFP to mitochondria in [rho⁺] cells than in [rho⁰] cells (H₂O₂ plus antimycin).

Ntg1 and Ntg2 are posttranslationally modified by sumoylation.

Posttranslational modification of various proteins via sumoylation can direct subcellular localization in response to environmental signals (27, 33, 42, 82). Several lines of evidence indicate that Ntg1 and Ntg2 may be modified posttranslationally by SUMO. Ntg1 and Ntg2 contain seven and one putative sumoylation site (Fig. 5), respectively, as predicted using the SUMO prediction program SUMOsp 1.0 (64, 85). Cell lysates from yeast that express Ntg1-GFP and Ntg2-GFP reveal a major species corresponding to the size of the fusion protein and a second, minor species of higher molecular size corresponding to the predicted size for monosumoylated Ntg1 and Ntg2 (90). In addition, a recent study cataloging sumoylated yeast proteins reported that Ntg1 interacts with the Smt3 gene (26), which encodes yeast SUMO (16, 39, 52); however, covalent modification of Ntg1 by Smt3 was not assessed in that study.

FIG. 5.

Amino acid sequences of Ntg1 and Ntg2. The amino acid sequences of Ntg1 (top) and Ntg2 (bottom) are shown. The following domain structures are indicated: potentially sumoylated lysines with a [hydrophobic]-K-X-[ED] motif (red) predicted by the SUMOsp 1.0 program (85), predicted nuclear localization sequences (green) determined with the NUCDISC subprogram of PSORTII (56), a predicted mitochondrial targeting sequence in Ntg1 (bold and italicized) determined with the MITDISC subprogram of PSORTII (56), predicted helix-hairpin-helix active-site regions (underlined) and active-site lysine (blue) determined due to significant homology with endonuclease III and its homologs (4, 31), and the [4Fe-2S] cluster (brown) characterized by the sequence C-X₆-C-X-X-C-X₅-C (44).

In order to test for sumoylation of Ntg1 and Ntg2, we looked for the presence of high-molecular-weight forms of Ntg1 and Ntg2. Western analysis of TAP-purified Ntg1 and Ntg2 revealed species corresponding to the sizes of Ntg1-TAP (55 kDa) and monosumoylated Ntg1-TAP (70 kDa) as well as Ntg2-TAP (46 kDa) and monosumoylated Ntg2 (58 kDa) (Fig. 6A and B). The sizes of the higher-molecular-size species correspond to the sizes predicted for addition of a single SUMO moiety (12 kDa) to both Ntg1 and Ntg2. To determine whether Smt3 is covalently attached to Ntg1 and Ntg2, GST-tagged Ntg1 and Ntg2 were purified from cells expressing both GST-tagged Ntg proteins and HA-tagged Smt3. Detection of the same high-molecular-weight species with both GST and HA antibodies would reveal covalent modification of Ntg1 and Ntg2. Western analysis confirmed the covalent modification of Ntg1 and Ntg2 by SUMO, as indicated by the codetection of a high-molecular-weight species by both GST and HA antibodies (Fig. 6C and D). Collectively, these results are consistent with the conclusion that both Ntg1 and Ntg2 are posttranslationally modified by sumoylation.

FIG. 6.

Posttranslational modification of Ntg1 and Ntg2 by SUMO. (A and B) Western analysis of TAP-purified Ntg1-TAP and Ntg2-TAP utilizing an anti-calmodulin TAP antibody. Nonsumoylated and sumoylated species of Ntg1 and Ntg2 are indicated. Protein sizes are indicated in the right margins. (C and D) Western analysis of purified Ntg1-GST and Ntg2-GST detected with antibodies to GST (Ntg1 or Ntg2) and HA (Smt3). Nonsumoylated and sumoylated species of Ntg1 and Ntg2 are indicated. Protein sizes are indicated in the right margin. Double-headed arrows indicate sumoylated Ntg1 and Ntg2 detected simultaneously with GST and HA antibodies.

Sumoylated Ntg1 accumulates in the nucleus following oxidative stress.

To address whether sumoylation could play a role in the subcellular localization of Ntg1, cells expressing TAP-tagged Ntg1 were exposed to H₂O₂ (nuclear oxidative stress) or H₂O₂ plus antimycin (mitochondrial oxidative stress) and subjected to sucrose gradient subcellular fractionation (see Materials and Methods). Mitochondrial fractions were free of nuclear proteins, as determined by Western analysis using Por1 and Nop1 as mitochondrial and nuclear protein markers (62), respectively, whereas nuclear fractions were enriched for nuclear proteins (Fig. 7A and B). Sumoylated Ntg1 was detected in nuclei and increased in amount relative to unmodified Ntg1-TAP following both nuclear and mitochondrial oxidative stress (Fig. 7A and B). Analysis of sumoylated and nonsumoylated Ntg1-TAP in nuclear fractions by chemiluminescence revealed that exposure to oxidative stress results in an approximately fivefold increase in nuclear sumoylated Ntg1 (Fig. 7C). These results suggest that sumoylation of Ntg1 is associated with the nuclear localization of Ntg1 in response to oxidative stress.

FIG. 7.

Sumoylation of nuclear Ntg1 increases in response to oxidative stress. Cells were exposed to no treatment, 10 mM H₂O₂, 10 μg/ml antimycin, or 10 mM H₂O₂ plus 10 μg/ml antimycin. Sucrose gradient subcellular fractionation (see Materials and Methods) was employed to assess the localization of Ntg1 to nuclei and mitochondria following exposure to nuclear (H₂O₂) or mitochondrial (H₂O₂ plus antimycin) stress-inducing agents. (A) Western analysis of nuclear fractions, utilizing antibodies to Nop1 (nuclear marker), Por1 (mitochondrial marker), and the calmodulin domain of TAP to detect Ntg1 and Smt3-Ntg1. (B) Western analysis of mitochondrial fractions. (C) Levels of nuclear sumoylated Ntg1 species detected by chemiluminescence in response to oxidative stress. Nucleus-enriched and mitochondrial subcellular fractions were generated (see Materials and Methods) and evaluated by Western analysis. Following chemiluminescence evaluation of nuclear Ntg1, unmodified and sumoylated Ntg1 proteins were quantified, and the change in percent sumoylated Ntg1 was calculated. Error bars represent standard errors of the means. The asterisk indicates statistical significance (P < 0.005).

Subcellular localization and function of mutant Ntg1 lacking a predicted SUMO site.

The sumoylation prediction program SUMOsp 1.0 (85) was utilized to determine lysine residues where Ntg1 is most likely to be sumoylated. SUMOsp 1.0 predicted that lysine 364 within the sequence KREL was most likely to be sumoylated among the 36 lysines present in Ntg1. To assess the possible requirement for lysine 364 in Ntg1 function, lysine 364 was replaced with arginine. This amino acid substitution retains the positive charge of the residue but blocks sumoylation (32). In order to determine whether sumoylation could affect the subcellular distribution of Ntg1, the intracellular localization of Ntg1 K364R-GFP was compared to that of wild-type Ntg1-GFP. Ntg1 K364R-GFP was localized to both nuclei and mitochondria (Fig. 8A); however, the relocalization of Ntg1 K364R-GFP in response to H₂O₂ exposure (nuclear oxidative stress) or MMS exposure (Fig. 8B) was markedly different from that of wild-type Ntg1-GFP (Fig. 3). Specifically, the fraction of nuclear only Ntg1 K364R-GFP decreased in response to either H₂O₂ or MMS exposure, while the nuclear only localization of wild-type Ntg1-GFP increased in response to both agents (compare Fig. 3D and 8B). These results indicate that the predicted Ntg1 sumoylation site, K364, is important for relocalization of Ntg1 in response to nuclear oxidative stress, likely resulting in oxidative DNA damage, and provide further evidence of a role for SUMO in the dynamic localization of Ntg1.

FIG. 8.

Subcellular localization and function of the Ntg1 K364R mutant. (A) GFP image of cells expressing Ntg1 K364R-GFP. (B) Quantification of Ntg1 K364R-GFP localization. Cells were not treated (NT) or were exposed to the indicated oxidative stress-inducing agent for 1 h (see Materials and Methods). Localization of Ntg1 K364R-GFP to nuclei only (nuclear) or nuclei plus mitochondria (nuc + mito) was determined for each cell and plotted as a percentage of the total cells evaluated. Error bars represent standard deviations. Refer to Fig. 3D for localization of wild-type Ntg1. (C) Functional analysis of Ntg1 K364R. The H₂O₂ sensitivity of wild-type (WT) cells, BER⁻/NER⁻ deficient cells, and BER⁻/NER⁻ deficient cells containing an episomal copy of wild-type Ntg1-GFP or Ntg1 K364R-GFP was assessed. Cells were exposed to 0, 2, 4, or 6 mM H₂O₂. The percent survival was set to 100% for untreated samples. Error bars indicate standard deviations in the data. (D) Steady-state expression levels of wild-type Ntg1-GFP and mutant Ntg1 K364R-GFP in BER⁻/NER⁻ deficient cells. Western analysis of whole-cell lysates from BER⁻/NER⁻ deficient (ntg1 ntg2 apn1 rad1) cells (Ctrl) and BER⁻/NER⁻ deficient cells containing an episomal copy of wild-type Ntg1-GFP (WT) or mutant Ntg1 K364R-GFP (K364R) was performed utilizing an anti-GFP antibody (to detect Ntg1) and anti-phosphoglycerate kinase antibody (to determine relative levels of protein loaded per lane).

In order to assess the function of Ntg1 K364R, which cannot properly relocalize in response to oxidative stress, we exploited BER⁻/NER⁻ (ntg1 ntg2 apn1 rad1) defective cells (75). These BER⁻/NER⁻ defective cells are severely compromised in the repair of oxidative DNA damage and are highly sensitive to H₂O₂ (75). Importantly, these cells lack endogenous Ntg1, so the function of Ntg1 K364R could be assessed as the only cellular copy of Ntg1. For this experiment, plasmids encoding wild-type Ntg1-GFP or Ntg1 K364R-GFP were transformed into BER⁻/NER⁻ cells, and the sensitivity of these cells to H₂O₂ was determined (Fig. 8C). An episomal copy of wild-type Ntg1 substantially increased cell survival following H₂O₂ exposure compared to that of BER⁻/NER⁻ cells. In contrast, following H₂O₂ treatment, the survival of BER⁻/NER⁻ cells expressing Ntg1 K364R was comparable to or less than the survival of the control BER⁻/NER⁻ cells, demonstrating that Ntg1 K364R is not properly localized in vivo to mediate its DNA repair function in the nucleus. To ensure that the compromised function of Ntg1 K364R was not due to decreased expression of the mutant protein, we assessed the steady-state levels of both Ntg1-GFP and Ntg1 K364R-GFP in cell lysates by immunoblotting with an anti-GFP antibody (Fig. 8D). This analysis revealed that the level of Ntg1 K364R was equivalent to that of wild-type Ntg1. These results suggest that the predicted Ntg1 sumoylation site, K364, is important for the function of Ntg1 in conferring cellular survival following oxidative stress.

DISCUSSION

To gain insight into the regulation of BER in response to oxidative stress-induced DNA damage, the localization and posttranslational modification of S. cerevisiae Ntg1 and Ntg2 were evaluated. We demonstrate that Ntg1 relocalizes in response to both nuclear and mitochondrial oxidative DNA damage. In contrast, Ntg2 is exclusively nuclear, and this localization does not change in response to oxidative DNA damage. Furthermore, sumoylation of Ntg1 is associated with nuclear localization in response to nuclear oxidative stress.

ROS are a by-product of environmental factors and important cellular processes, including oxidative phosphorylation. Nuclear and mitochondrial oxidative stress occurs due to inefficiencies and malfunctions of these processes. Furthermore, increased nuclear and mitochondrial oxidative stress has been observed in cells with compromised nuclear and mitochondrial ROS scavenging systems (36, 50). Mitochondrial oxidative stress is increased when cellular oxidative phosphorylation activity is particularly high or disrupted (79). Furthermore, aging and various diseases have been associated with increased nuclear and mitochondrial ROS levels (13, 45, 81, 83). Under conditions where nuclear oxidative stress is high, nuclear oxidative DNA damage is elevated (19, 86). Likewise, conditions that increase mitochondrial oxidative stress are associated with high levels of mitochondrial oxidative DNA damage (17, 79). When oxidative stress is increased, it is essential for the cell to respond to oxidative DNA damage rapidly in order to prevent the detrimental consequences of unrepaired DNA, and a rapid response to oxidative DNA damage requires explicit regulation of BER components.

Regulating the localization of proteins is a way for cells to respond quickly to a stimulus without having to produce more protein. Because localization is a significant component of regulation for many processes (24, 29, 37, 58), we evaluated the localization of the BER proteins Ntg1 and Ntg2 in response to oxidative stress and determined that the localization of Ntg1 is influenced by nuclear and mitochondrial oxidative stress, whether the stress is caused by an oxidizing agent (H₂O₂) or indirectly by MMS, a nonoxidative DNA-alkylating agent (Fig. 3). In addition to Ntg1, the human transcription factor/AP endonuclease Ref1/hAPE was previously reported to translocate to nuclei and mitochondria following exposure to a DNA damaging agent (15, 21, 76, 77), adding further credibility to our claim that dynamic localization is a mechanism for regulation of BER in eukaryotic systems in general. In addition to demonstrating the dynamic localization of Ntg1, we were able to delineate the origin of the signal that results in targeting of Ntg1 to mitochondria by utilizing [rho⁰] yeast cells. Specifically, we demonstrated that Ntg1 responds to mitochondrial oxidative DNA damage and not simply to elevated levels of ROS (Fig. 4). To our knowledge, this is the first experimental strategy that has been able to distinguish between protein localization caused by ROS and its DNA damage products. We hypothesize that the nuclear localization of Ntg1 is similarly controlled by high levels of nuclear oxidative DNA damage. Because the localization of several human DNA repair proteins is affected by oxidative stress (15, 21, 49, 76, 77, 87, 88) and because BER is highly conserved between S. cerevisiae and humans, we suggest that modulation of DNA repair protein localization is a general mechanism by which BER is regulated in eukaryotic cells. We propose a model in which BER proteins such as Ntg1 are located in nuclei and mitochondria in cells under normal growth and oxygen conditions (Fig. 9). When nuclear and mitochondrial oxidative stress occurs, nuclear and mitochondrial oxidative DNA damages result. We hypothesize that specific signals are generated in response to oxidative DNA damage to target BER proteins such as Ntg1 to nuclei and mitochondria in order to increase the capacity to repair these lesions rapidly.

FIG. 9.

Proposed model for regulation of BER proteins in response to oxidative stress. The oxidative stress state of a cell is affected by its environment and metabolic processes. Nuclear and mitochondrial oxidative DNA damage occurs as a result of nuclear and mitochondrial oxidative stress. Nuclear and mitochondrial oxidative DNA damage initiates signaling of BER proteins, such as Ntg1, to sites of damage. NODDS and MODDS are responsible for recruiting BER proteins to nuclei and mitochondria, respectively. NODDS likely include the sumoylation machinery and influence nuclear protein localization. SUMO modification (triangles) of Ntg1 concentrates Ntg1 in the nucleus following oxidative stress. When the target BER proteins are contacted by NODDS or MODDS, relocalization to nuclei and/or mitochondria occurs, depending on the level of oxidative DNA damage present in each organelle. Following recruitment of BER proteins into the nucleus and mitochondria, the capacity for repair of oxidative DNA damage increases accordingly. In order to maintain a steady-state (baseline) level of BER proteins in nuclei, BER proteins such as Ntg2 do not relocalize in response to oxidative DNA damage.

Nuclear oxidative DNA damage signals (NODDS) and mitochondrial oxidative DNA damage signals (MODDS) are likely to involve various proteins and pathways, including components of the BER pathway, components of other DNA damage management pathways, and other molecules that are involved in cellular stress responses. Since oxidative DNA damage can be produced spontaneously, other types of spontaneous DNA damage, such as alkylation, methylation, deamination, and depurination, could alter the subcellular localization of BER proteins through signals similar to NODDS and MODDS. Our observation that Ntg1 relocalizes in response to MMS exposure supports the idea that a variety of spontaneous DNA damages can trigger recruitment of BER proteins. The signals from nonoxidative species of spontaneous DNA damage could recruit BER proteins directly or indirectly. Using alkylation as an example, abasic sites generated during repair of the alkylation damage may signal for recruitment of BER proteins directly, or the ROS produced as a result of the alkylation damage may cause oxidative DNA damage, which in turn recruits BER proteins through NODDS and MODDS.

We hypothesize that components of the sumoylation pathway are NODDS molecules. Several DNA repair and other DNA maintenance proteins are sumoylated (55), and sumoylation has been implicated in the nuclear localization of a number of proteins (59). Our results indicate that Ntg1 and Ntg2 are modified posttranslationally by sumoylation and are consistent with a model where sumoylation plays a role in the localization of Ntg1 to the nucleus in response to increased nuclear oxidative stress (Fig. 6 and 7). A fivefold increase in nuclear sumoylated Ntg1 was observed following oxidative stress. We found that 1% of the Ntg1 pool is sumoylated in cells under normal conditions, increasing to 5% in cells exposed to oxidative stress. These results are consistent with data describing other sumoylated proteins, where often less than 1% of the substrate can be detected as sumoylated at any given time (38). Sumoylation of a human BER N-glycosylase, thymine-DNA glycosylase, has been hypothesized to occur in a cyclical pattern of sumoylation and desumoylation (28, 38). In this case, sumoylated thymine-DNA glycosylase promotes a single event whose consequences persist after desumoylation (28). We hypothesize that sumoylation of Ntg1 occurs in order to concentrate Ntg1 within nuclei but that desumoylation occurs very quickly, making it very difficult to detect the pool of sumoylated Ntg1. Furthermore, we provide evidence that Ntg1 K364 is a potential target site of sumoylation that may contribute to the nuclear localization and function of Ntg1, although further experimentation is necessary to confirm this notion (Fig. 8). We predict that other nuclear BER proteins may be sumoylated in order to allow intricate regulation of DNA repair protein localization. The function of sumoylated Ntg2 is unknown, but it is possible that SUMO plays a role in modulating the intranuclear localization of Ntg2.

SUMO could contribute to the localization of BER proteins in several ways. Sumoylation of Ntg1 and other BER proteins could modulate interactions with nuclear transport receptors, as sumoylation modulates interaction with the nuclear transport receptors for various proteins (59). Sumoylation could also block BER proteins from exiting the nucleus in the event of oxidative DNA damage (Fig. 9). Sumoylation is also implicated in the regulation of subnuclear localization of numerous proteins. Localization of proteins to nucleoli, promyelocytic leukemia nuclear bodies, and other subnuclear locations is associated with sumoylation (22, 30). Therefore, sumoylation could allow Ntg1 and other BER proteins to accumulate in certain subnuclear, subgenomic regions containing oxidative DNA damage.

As Ntg1 localizes to both nuclei and mitochondria, the proportion of the pool of Ntg1 that localizes to each organelle must be adjusted so that some level of repair is maintained in nuclei and mitochondria at all times. Various factors are likely to influence the localization of Ntg1 to nuclei or mitochondria. Yeast mitochondrial DNA contains two to three times more oxidative lesions than nuclear DNA following oxidative stress induced by various agents (69). When more Ntg1 is needed in mitochondria, relocalization diminishes nuclear Ntg1 pools. Nuclear DNA will not accumulate significant DNA damage in the absence of Ntg1 because nuclear Ntg2, NER proteins, and Apn1 are available to repair baseline levels of oxidative DNA damage (75). Yeast mitochondria do not contain Ntg2 or NER proteins, leaving mitochondrial DNA vulnerable in the absence of Ntg1 (12). Because of the numerous factors influencing Ntg1 localization, a careful balance of NODDS and MODDS is required in order to increase repair capacity in one organelle without diminishing repair in the other. Such a balance of NODDS and MODDS is illustrated in [rho⁰] cells, as unexposed [rho⁰] cells display increased nuclear Ntg1 localization compared to unexposed [rho⁺] cells (Fig. 4C). We speculate that the increased nuclear localization results from elimination of MODDS-mediated recruitment of Ntg1 to mitochondria that results in enhanced recruitment of Ntg1 to nuclei by NODDS.

Very few investigations have addressed the issue of dynamic localization of BER proteins in the process of initiating BER in response to oxidative stress. Our studies have uncovered what is likely to be a major component of the regulation of BER. By controlling the subcellular localization of BER proteins, cells can rapidly mobilize repair machinery to sites of oxidative DNA damage.