Regulation of Human MutYH DNA Glycosylase by the E3 Ubiquitin Ligase Mule

Background: Regulation of MutYH DNA glycosylase is essential for genomic integrity. Results: MutYH is ubiquitinated by the E3 ligase Mule changing its protein levels and recruitment to chromatin. MutYH levels in cancer cells influence the mutagenic potential of reactive oxygen species. Conclusion: Ubiquitination of MutYH is an important regulatory step. Significance: We identified the residues that are targets for ubiquitination by Mule. Oxidation of DNA is a frequent and constantly occurring event. One of the best characterized oxidative DNA lesions is 7,8-dihydro-8-oxoguanine (8-oxo-G). It instructs most DNA polymerases to preferentially insert an adenine (A) opposite 8-oxo-G instead of the appropriate cytosine (C) thus showing miscoding potential. The MutY DNA glycosylase homologue (MutYH) recognizes A:8-oxo-G mispairs and removes the mispaired A giving way to the canonical base excision repair that ultimately restores undamaged guanine (G). Here we characterize for the first time in detail a posttranslational modification of the human MutYH DNA glycosylase. We show that MutYH is ubiquitinated in vitro and in vivo by the E3 ligase Mule between amino acids 475 and 535. Mutation of five lysine residues in this region significantly stabilizes MutYH, suggesting that these are the target sites for ubiquitination. The endogenous MutYH protein levels depend on the amount of expressed Mule. Furthermore, MutYH and Mule physically interact. We found that a ubiquitination-deficient MutYH mutant shows enhanced binding to chromatin. The mutation frequency of the ovarian cancer cell line A2780, analyzed at the HPRT locus can be increased upon oxidative stress and depends on the MutYH levels that are regulated by Mule. This reflects the importance of tightly regulated MutYH levels in the cell. In summary our data show that ubiquitination is an important regulatory mechanism for the essential MutYH DNA glycosylase in human cells.

Every organism is exposed to high levels of oxidative stress every day from a variety of endogenous and exogenous sources. One of the most often observed consequences is the formation of the highly mutagenic lesion 7,8-dihydro-8-oxoguanine (8-oxo-G), 3 which arises ϳ10 3 times per cell and day in normal tissue and up to 10 5 times in cancer tissues (1). The 8-oxo-G lesion is potentially highly mutagenic, because replicative pols tend to incorporate the incorrect A rather than the correct C opposite the lesion, leading to the formation of an A:8-oxo-G mispair. If these mispairs are not repaired before the cell undergoes replication, GC3 TA transversion mutations occur. Such mutations were found to be frequently present in different types of cancer tissues (2), demonstrating the paramount importance of a mechanism ensuring the correct repair of 8-oxo-G lesions. As we have shown previously (3,4), such a mechanism indeed exists: a specialized base excision repair (BER) pathway coordinated by the DNA glycosylase MutYH and DNA pol (5). MutYH recognizes a mispaired A and excises it, resulting in the formation of a one nucleotide gap, still bearing the 8-oxo G lesion on the template strand, that is subsequently filled by pol . In contrast to other pols, pol and to some extent pol , both in the presence of proliferating cell nuclear antigen and replication protein A, are very efficient in the correct bypass of an 8-oxo-G lesion thereby preventing the generation of mutations (4).
In agreement with the important role of MutYH in the repair of oxidative DNA damage, mouse embryonic fibroblasts as well as embryonic stem cells derived from MutYH Ϫ/Ϫ mice show an increase in mutation frequency (6,7). Furthermore, knock-out mice have a higher incidence of tumor formation compared with the wild type mice. This effect can even be increased by exposing the mice to an oxidant showing that the absence of MutYH drastically impairs the repair of oxidative damage (8). Intriguingly, MutYH transcript levels are inversely correlated with the survival outcome of patients suffering from gastric cancer (9).
The BER machinery needs to be tightly controlled to ensure an efficient and correct repair of damaged DNA. Already small differences in expression of individual BER proteins can disturb the entire pathway and thus lead to a reduced repair capacity (10). Recent data provide evidence that regulation of BER is mainly achieved through various posttranslational modifications, such as phosphorylation and ubiquitination (11)(12)(13)(14). Even if it was shown that the MutYH expression levels fluctuate during the cell cycle (15) and can be increased upon exposure of cells to reactive oxygen species (16), no mechanism elucidating the regulation of MutYH on protein level has been identified so far.
In the present paper we provide the first evidence that MutYH is a target for ubiquitination, and the executing E3 ligase was identified to be Mule. Ubiquitinated MutYH is marked for proteasomal degradation; thereby not only the protein levels are modulated but also its recruitment to chromatin. We furthermore show that a tight regulation of MutYH is of great importance, because altered protein levels lead to an increase in mutation frequency at the HPRT locus of the ovarian cancer cells A2780.
KBrO 3 Treatment and Cell Based Mutagenesis Assay-The cell-based mutagenesis assays were performed using the A2780 cell line as described in Ref. 17. Briefly, the cells were transfected with plasmid or siRNA as indicated in Fig. 6; 24 h later the cells were seeded for treatment with KBrO 3 (concentrations as indicated, 30 min). After culturing the treated cells for 7 days, they were reseeded in five plates/cell line (1 ϫ 10 5 cells/ 10-cm plate), and cells harboring a mutation in the HPRT gene were selected by adding 6-thioguanine (5 g/ml; Sigma). In parallel each cell line was plated (150 cells/6-well plate) in triplicate without the addition of 6-thioguanine to determine the seeding efficiency. The mutation rate was calculated after 14 days under selection, as the ratio of the number of mutant colonies to the number of clone-forming cells, as also described in Ref. 17. Three independent experiments including always the control, Mule knockdown (KD), and Mule overexpression cells were performed.
GST Pulldown Assay-The GST pulldown assay was performed using recombinant and purified GST-tagged MutYH (2 g) coupled to GST-Sepharose beads (GE Healthcare) and incubated with 800 ng of recombinant HECT domain of Mule. The assay was performed as described in detail in Ref. 16.
Pulldown Assay-Cell extract containing overexpressed FLAG-tagged HECT domain of Mule was incubated with FLAG AB (1 g) and 25 l of protein G-Sepharose beads (GE Healthcare) for 2 h. After washing three times with a buffer containing PBS with 0.2% Triton X-100, 1 mM DTT, 1 g/ml leupeptin, 1 g/ml pepstatin, 1 g/ml bestatin, and 1 mM PMSF 800 ng of recombinant purified protein was added to the beads. After washing three times, the beads were finally resuspended in Laemmli buffer and the samples analyzed by Western blotting.
Detection of Protein Ubiquitination-The protocol described in detail in Ref. 18 was used to eliminate all noncovalent protein binding.
Co-immunoprecipitation-Cell extracts were prepared as described, diluted 1:2 in HNTG buffer (20 mM Hepes (pH 7.5), 150 mM NaCl, 10% glycerol, 0.1% Triton-X-100, 10 mM NaF, 1 mM Na 3 VO 4 ) and incubated with 2 g of AB as indicated and 25 l of protein G-Sepharose beads. After rotating for 2 h at 4°C the beads were washed three times with HNTG buffer, resus-pended in Laemmli buffer, and finally analyzed by Western blotting.
DNA Isolation and 8-Oxo-G Determination-Approximately 2 ϫ 10 7 cells were harvested and immediately stored at Ϫ80°C. The DNA was isolated with the DNA Isolation kit for Cells and Tissues (Roche Applied Science) according to the manufacturer's instructions. The 8-oxo-G determination was carried out following the protocol depicted in Ref. 19.
Microscopy-HeLa cells were transfected with plasmids as indicated using Lipofectamine 2000 (Invitrogen) according to the manufacturer's instructions. The cells were seeded on glass coverslips and fixed using methanol for 20 min at Ϫ20°C. After blocking with PBS/10% FCS the cells were incubated with the indicated primary AB (1:300 in PBS/10% FCS) for 2 h. The coverslips were washed three times with PBS/10% FCS (30 min), incubated with the secondary Cy3 (1:300 in PBS/10% FCS) for 1 h, washed three times, and fixed on object slides using Vectashield (Vector Laboratories). The pictures were taken using the Leica CTR 6000.
Statistical Analysis-For all statistical analysis the program GraphPad Prism was used. All experiments shown were repeated at least three times, if not stated differently in the figure legends. The shown error bars correspond to the S.D. Student's t test was used to evaluate the significance of at least three (otherwise as stated in the figure legends) independent experiments.

MutYH Is Ubiquitinated in Vitro and in Vivo by the E3 Ubiquitin Ligase
Mule-Former studies identified the E3 ubiquitin ligase Mule as a regulator of the steady state levels of different proteins involved in BER (12,14,20). Here, we were aiming at the identification of the regulatory mechanisms controlling the MutYH protein levels in the cell. Using an in vitro ubiquitination approach we tested whether MutYH was a substrate for the E3 ubiquitin ligase Mule. Mule is a 482-kDa protein whose catalytic center lies within the HECT domain, mapped to the C-terminal 370 amino acids (21). Using the recombinant, purified HECT domain in an in vitro ubiquitination assay (Fig. 1A), we found that MutYH was ubiquitinated by the HECT domain in presence of the E2 conjugating enzymes H5b, H5c, and H7. To get more distinct bands a mutant ubiquitin, not able to form polyubiquitin chains, was used in all in vitro assays.
We next addressed the question of whether Mule also plays a role in the regulation of MutYH in vivo. We transfected an N-terminal deletion construct of Mule containing the full HECT domain (Mule ⌬N2474) together with MutYH for overexpression into HEK293T cells. As shown in Fig. 1B the co-transfection led to the appearance of higher migrating MutYH bands (lanes 2 and 3) in contrast to the control (lane 1) in which just one form of MutYH was present. The fact that addition of the proteasomal inhibitor MG132 (lane 3) to the transfected cells prior to harvest resulted in an increased intensity in the higher migrating bands of MutYH further suggested that these bands represented an ubiquitinated form of the protein. To ensure that the catalytic activity of Mule was required for the modification of MutYH and to strengthen the result shown in Fig. 1B, an additional experiment was performed in parallel using a catalytic dead mutant of Mule (Mule C4341A). In contrast to the experiment using the Mule WT, the co-transfection of the Mule C4341A mutant together with MutYH (lanes 2 and 3) did not result in any higher migrating forms of MutYH (Fig.  1C). These results clearly showed that MutYH is modified by the catalytic activity of Mule.
We performed a transient KD of Mule in HEK293T cells (Fig.  1D) that overexpressed MutYH and ubiquitin. After inhibition of the proteasomal degradation by adding MG132 we purified MutYH by binding to beads and analyzed the amount of ubiquitinated MutYH (Fig. 1E). The KD of Mule resulted in a 40% reduction of modified MutYH, if normalized to the amount of unmodified MutYH bound to the beads (Fig. 1F), indicating that Mule is the enzyme responsible for ubiquitination of MutYH. The observed decrease in ubiquitinated MutYH upon KD of Mule further supports its role as the E3 ligase for MutYH ubiquitination.
MutYH Protein Levels Depend on the Amount of the E3 Ubiquitin Ligase Mule-To address the relevance of regulation by Mule in vivo, we performed a transient KD of Mule using siRNA to see an effect on the endogenous protein levels of MutYH. As shown in Fig. 2A and quantified in Fig. 2B the KD of Mule led to a significant increase in MutYH protein levels, whereas neither the treatment with Lipofectamine nor with scrambled siRNA showed this effect. Unchanged mRNA levels of MutYH in these cells (Fig. 2C) were indicative that the regulation of MutYH (Fig. 1, A and B) took place at the protein level. Consistently we observed the opposite effect upon Mule overexpression. The transfection of HEK293T cells with a Mule construct caused a dramatic decrease in the endogenous protein levels of MutYH compared with the empty vector control, as shown in Fig. 2D and quantified in Fig. 2E. As expected, the MutYH mRNA levels did not change in these cells upon overexpression of Mule (Fig.  2F). The results showing that the amount of Mule in the cells directly influences the protein levels of MutYH led us to conclude that indeed Mule is the E3 ubiquitin ligase responsible for the regulation of MutYH on protein level in vivo.
MutYH and Mule Physically Interact-We questioned whether MutYH and Mule directly interact. In the first approach we could confirm the physical interaction between the two proteins by performing pulldown assays using tagged, recombinant and purified proteins. The pulldown of GSTtagged MutYH also enriched the HECT domain, demonstrating its direct interaction with MutYH (Fig. 3A). Moreover, the pulldown of the HECT domain using FLAG-beads confirmed the interaction between the two proteins (Fig. 3B).
Finally, we showed that the interaction could be also recapitulated in cells. We performed co-immunoprecipitation exper-iments using whole cell extract containing overexpressed FLAG-tagged Mule and HA-tagged MutYH. By binding of Mule to FLAG-beads we could show that MutYH co-immunoprecipitated (Fig. 3C). The interaction could be also confirmed vice versa by coupling MutYH to HA-beads and analyzing the samples for co-immunoprecipitated Mule (Fig. 3D) Fig. 4A). To narrow the region we generated further deletion constructs of the MutT-like domain. Once the amino acids 475-535 were deleted the MutYH construct was not ubiquitinated any more (lanes 12 and 14) leading finally to the conclusion that the C-terminal 60 amino acids of MutYH contain the ubiquitination sites. The sequence of MutYH (Fig. 4B) contains five lysine residues within this region as potential targets for ubiquitination (Fig. 4B). To determine which function these lysine residues have for the regulation of MutYH we generated ubiquitination-deficient, lysine to arginine point mutants: KK477RR, K495R, KK506RR, and one construct containing all five lysine residues mutated to arginine (5R) (Fig. 4B). Consistent with the idea that ubiquitination of MutYH affects the proteasomal degradation of MutYH all lysine point mutants showed stabilization on the protein level if transfected in HEK293T cells (Fig. 4C). The mRNA amount of the transfected mutants was analyzed, and the protein levels were normalized to that (Fig. 4D) to show that the increase in MutYH was caused by a stabilization of the mutants on protein level rather than by differences in the transfection efficiency. The 5R mutant showed a more pronounced increase in protein level (4-fold compared with WT) than the single mutants (1.5-2.5-fold) (Fig. 4, C and D), leading to the assumption that all five lysine residues appear to contribute to the ubiquitination of MutYH. Strikingly, the treatment of cells with the proteasome inhibitor MG132 caused a dramatic increase in the protein level of the MutYH WT compared with the ubiquitination-deficient mutant 5R that is just slightly stabilized (Fig. 4, E and F). This result further underlines the fact that the MutYH 5R mutant is protected against ubiquitination and proteasomal degradation.
Ubiquitination-deficient MutYH Is Preferentially Bound to Chromatin-Besides protein stability, ubiquitination can also influence the intracellular localization of a protein (22). To figure out whether ubiquitination of MutYH determines its localization within the cell, cell fractionation experiments were performed (Fig. 5A). HEK293T cells transfected with either the MutYH WT or the 5R mutant were fractionated into cytoplasmic, soluble nuclear and chromatin-bound fractions. We observed an enrichment of the ubiquitination-deficient MutYH mutant in the chromatin-bound fraction with respect to the WT (Fig. 5B), whereas the amount of MutYH in the cytoplasm remained comparable (Fig. 5C). In addition to the cellular fractionation, this effect could be further confirmed by using an immunocytochemistry approach. HeLa cells were transfected with the HA-tagged MutYH WT or the 5R mutant and stained for HA and the nuclear marker DAPI. The result clearly showed that the MutYH WT was exclusively localized in the cytoplasm whereas the 5R mutant was also present in the nucleus to a substantial amount (Fig. 5D). Taken together, these results indicated that ubiquitination of MutYH not only influences its turnover but also changes its recruitment to chromatin.

Reactive Oxygen Species Treatment Increases the 8-Oxo-G Levels and the Mutation Frequency in A2780 Ovarian Cancer
Cells-MutYH plays an important role in the repair of 8-oxo-G thereby preventing the onset of deleterious mutations (23,24). Consequently, we expected changes in the mutation frequency according to the amount of MutYH present in the cells. A2780 cells, either transfected with a Mule construct (Mule ⌬N2474) or Mule siRNA, were treated with KBrO 3 to stimulate the formation of 8-oxo-G lesions. From Fig. 6A it is evident that the 8-oxo-G levels in the cells increased upon treatment with increasing amounts of KBrO 3 , as determined in a mass spectrometry-based assay. The observed levels of 8-oxo-G are in the range of 1-5.5 8-oxo-G/10 6 nucleotides for the untreated cells (data not shown) and are in accordance with the numbers reported in literature (25). The results indicate a clear tendency of a higher oxidative burden for the cell upon treatment with KBrO 3 .
We have already shown that the MutYH levels depend on the amount of Mule (Fig. 2). Such a relationship was also confirmed in the A2780 cell system. Fig. 6B clearly shows an increase of endogenous MutYH protein levels upon KD of Mule and, conversely, a decrease upon Mule overexpression. The mutation frequency at the HPRT locus in these cells was analyzed by selection with 6-thioguanine, a toxic guanine analogue that is not incorporated in the DNA by cells harboring a mutation at the HPRT locus. Mule KD cells had a higher MutYH level (Fig.  6B) and a lower incidence of mutations in the HPRT locus (Fig.  6C). The opposite was observed for the Mule (Mule ⌬N2474)overexpressing cells, showing a decrease in MutYH level with respect to the control (Fig. 6, B and C).
These data confirmed that Mule-dependent regulation of MutYH influences the susceptibility of cells to mutagenesis caused by exposure to oxidative stress. Thus, the regulation of MutYH has a direct impact on the mutation frequency in cells, underlining again the importance of a tight and controlled regulation of the BER machinery.

DISCUSSION
In this work we provide evidence that ubiquitination regulates not only the steady state levels of MutYH, but also its subcellular localization. We identified Mule to be the E3 ubiquitin ligase responsible for the modification of MutYH. Mule mainly appears to monoubiquitinate MutYH. Whether other E3 ubiquitin ligases are involved in the polyubiquitination and proteasomal degradation remains so far elusive. Consistent with the predicted role of Mule in the proteasomal degradation of MutYH, we find the MutYH protein levels to be reversely correlated with the levels of Mule present in HEK293T cells. Five lysine residues could be identified as targets for ubiquitination by Mule. The clear stabilization of the ubiquitinationdeficient MutYH mutants confirms their role in the proteasomal degradation of the protein.
The enzymatic activity of MutYH needs to be directed to the nucleus to act on the newly synthesized strand immediately after replication, to prevent the formation of deleterious G:C3 T:A transversion mutations (15,26). MutYH was already shown to be localized in different cellular compartments (27). Our data implicate that the ubiquitination of MutYH plays an important role in enrichment of MutYH on chromatin. We observed that mutation of the ubiquitination sites results in strong accumulation of MutYH in the chromatin-bound cell fraction, respectively, in the nucleus as shown by immunofluorescence. We previously found that the recruitment of pol to chromatin is mediated by interplay of ubiquitination and phosphorylation as well as by formation of active repair complexes on chromatin (12,28,29). Comprising the recent data it seems very likely that MutYH might coordinate the MutYH/pol -dependent LP-BER pathway (16). Whether it is an orchestration of different posttranslational modification adjusting the subcellular localization and protein levels of MutYH remains to be elucidated.
As noted in the literature, a cell contains 1000 -7000 8-oxo-G as a steady state level that can be repaired any time (25). MutYH has an important role in the repair of these lesions, thus misregulation or mutations of the enzyme cause an increased mutation frequency (10,30). In accordance we see differences in the mutation frequency depending on the levels of MutYH.
In general, the mutation frequency at the HPRT locus increases upon KBrO 3 treatment of A2780 cells. Further, the data suggest that cells with elevated MutYH levels, due to KD of Mule, are better able to cope with oxidative stress than cells with a lower amount of MutYH, upon Mule overexpression. These results are underlined by a very recent publication showing that the amount of MutYH influences directly the susceptibility of cells to oxidative damage as well as the amount of 8-oxo-G upon treatment with hydrogen peroxide (31). To elucidate whether this regulation of MutYH has an impact in the context of cancer requires additional research.
In conclusion, the data presented here underline the importance of a tight and controlled regulation of BER, because already small differences in protein levels can have a pivotal effect on the genome integrity. In line with previous results, showing BER (12-14) enzymes to be regulated by posttransla- tional modification to ensure a faithful repair of 8-oxo-G lesions, we observed for the first time a similar regulation for the essential MutYH DNA glycosylase.