Nuclear Translocation of Mismatch Repair Proteins MSH2 and MSH6 as a Response of Cells to Alkylating Agents*

Mammalian mismatch repair has been implicated in mismatch correction, the prevention of mutagenesis and cancer, and the induction of genotoxicity and apoptosis. Here, we show that treatment of cells specifically with agents inducing O6-methylguanine in DNA, such asN-methyl-N′-nitro-N-nitrosoguanidine and N-methyl-N-nitrosourea, elevates the level of MSH2 and MSH6 and increases GT mismatch binding activity in the nucleus. This inducible response occurs immediately after alkylation, is long-lasting and dose-dependent, and results from translocation of the preformed MutSα complex (composed of MSH2 and MSH6) from the cytoplasm into the nucleus. It is not caused by an increase in MSH2 gene activity. Cells expressing the DNA repair protein O6-methylguanine-DNA methyltransferase (MGMT), thus having the ability to repair O6-methylguanine, showed no translocation of MutSα, whereas inhibition of MGMT by O6-benzylguanine provoked the translocation. The results demonstrate that O6-methylguanine lesions are involved in triggering nuclear accumulation of MSH2 and MSH6. The finding that treatment of cells with O6-methylguanine-generating mutagens results in an increase of MutSα and GT binding activity in the nucleus indicates a novel type of genotoxic stress response.

Mammalian mismatch repair has been implicated in mismatch correction, the prevention of mutagenesis and cancer, and the induction of genotoxicity and apoptosis. Here, we show that treatment of cells specifically with agents inducing O 6  Mammalian DNA mismatch repair is implicated in the repair of DNA base mismatches arising from spontaneous base instability and during replication (1). Mismatch repair is associated with the prevention of mutagenesis and cancer (2,3) and the induction of genotoxicity and apoptosis (4 -6). The mismatch repair system is able not only to detect and repair mismatches derived from spontaneous base modification and replication but can also process chemically induced DNA damage. An important group of chemicals associated with mismatch repair are alkylating mutagens. These agents, such as N-methyl-NЈ-nitro-N-nitrosoguanidine (MNNG) 1 and N-meth-yl-N-nitrosourea (MNU), cause methylation in the O 6 position of guanine. The resulting O 6 -methylguanine (O 6 -MeG) pairs with thymine instead of cytosine, thus leading to GC 3 AT transition mutations (7)(8)(9). O 6 -MeG paired with thymine, as well as various other mutagen-induced lesions such as 1,2intrastrand d(GpG) cross-links generated by cisplatin, are subject of repair by the mismatch repair system (10,11). In the case of O 6 -MeG-generating agents, which are most powerful mutagens and carcinogens, a lack of mismatch repair confers resistance to cytotoxicity and thus increases the mutagenic response of cells (12)(13)(14); this implies that genotoxic and cytotoxic effects of O 6 -MeG are mediated by mismatch repair. This could occur by erroneous mismatch repair cycles because of repeated misincorporation of thymine opposite O 6 -MeG or by direct signaling of apoptotic functions due to faulty mismatch repair activity. Since O 6 -MeG is a highly mutagenic and genotoxic lesion, its repair by the repair protein O 6 -methylguanine-DNA methyltransferase (MGMT) prior to replication is highly important for avoiding mutagenic, carcinogenic, and genotoxic effects. MGMT activity is highly variable among different tissues and cell types, and numerous cases (including cell lines, tumors, and knockout mice) have been reported unable to repair O 6 -MeG (15,16). In these cases, O 6 -MeG-triggered cell death via mismatch repair might be advantageous for the organism as it lowers alkylation-induced mutation pressure.
Mismatch repair of O 6 -methylguanine-thymine base pairs is initiated by the binding of a protein complex (designated as MutS␣) to the mismatch (17). This complex is composed of the bacterial MutS homologous proteins MSH2 and MSH6 (18), which is also known as GT-binding protein. In addition, MSH2 can also form a complex with the mismatch repair protein MSH3, designated MutS␤ (19,20). Depending on the binding partner, the heterodimers have differerent substrate specificities and therefore play different roles in mismatch repair. Thus, the MutS␣ complex (MSH2 and MSH6) is able to bind to base-base mismatches and to insertion/deletion mismatches (21)(22)(23) in contrast to the MutS␤ complex (MSH2 and MSH3), which is capable of binding to insertion/deletion mismatches. Upon binding to the mismatch, MutS␣ associates with another heterodimeric complex (MutL␣), consisting of the MutL homologous mismatch repair proteins MLH1 and PMS2 (24).
Whereas the function of the individual mismatch repair proteins has been studied in great detail, data are scarce regarding the regulation of mismatch repair. It has been shown that regulation of MSH2 is cell cycle-dependent, because MSH2 is expressed at higher level in proliferating than in resting cells (25). Whether MSH2 and/or other mismatch repair proteins are regulated by exogenous stimuli, notably mutagenic treatments, has not yet been analyzed. This subject, however, would be of great interest because mutagens are able to induce various immediate-early and late cellular responses, which are related to gene induction (c-fos, c-jun) and protein stabilization (p53) (for review see Refs. 26 and 27). In this work, we addressed the question of whether DNA-damaging treatments exert an effect on the expression of mismatch repair proteins. Here, we show that treatment of cells specifically with O 6 -MeG-generating agents leads to an increase of mismatch repair proteins and GT binding activity in nuclear extracts due to translocation of MSH2 and MSH6 from the cytoplasm into the nucleus upon mutagen treatment. The data indicate a novel type of inducible response of cells to alkylating agents.
Protein Extracts-For preparing whole cell extracts, cells were washed twice with ice-cold phosphate-buffered saline (PBS), harvested, resuspended in buffer (20 mM Tris-HCL, pH 8.5, 1 mM EDTA, 2 mM DTT, 0.5 mM PMSF, 5% glycerol) and sonicated on ice with a Branson sonifier (Branson, Dunbury, CT) at 30 kHz, three times for 15 s. The homogenates were centrifuged (10,000 ϫ g for 10 min at 4°C), and the clear supernatants were stored frozen at Ϫ80°C. For the preparation of nuclear extracts, cells were suspended in 1 ml of lysis buffer I (10 mM Tris-HCL, pH 7.4, 10 mM NaCl, 3 mM MgCl 2 , 0, 5 mM PMSF, 2 mM DTT) and incubated for 10 min on ice. After the addition of Nonidet P-40 (final concentration 0.5%) the solution was vortexed, incubated on ice for 5 min, and centrifuged (400 ϫ g, 5 min). Thereafter, the pellet was washed again with lysis buffer I. To isolate nuclear extracts suitable for EMSA, pellets were resuspended in 2 volumes of lysis buffer II (20 mM HEPES-KOH, pH 7.4, 600 mM KCl, 0.2 mM EDTA, 0.5 mM PMSF, 2 mM DDT) and incubated for 30 min on ice. After centrifugation, the supernatant was diluted by the addition of 1 volume of lysis buffer III (20 mM HEPES-KOH, pH 7.4, 0.2 mM EDTA, 0.5 mM PMSF, 2 mM DDT). Glycerol was added to a final concentration of 20%, and aliquots were shock-frozen in liquid nitrogen and stored at Ϫ80°C. The amount of protein was determined as described (28). To isolate nuclear extracts for Western blot analysis, pellets were resuspended in 150 l of lysis buffer IV (20 mM Tris-HCL, pH 7.4, 40 mM Na 4 P 2 O 7 , 5 mM MgCl 2 , 50 mM NaF, 100 M Na 3 VO 4 , 10 mM EDTA, 1% Triton X-100, 1% SDS) and sonicated. After centrifugation, the amount of protein in the supernatant was determined. Extracts were stored at Ϫ80°C. For preparing cytoplasmic protein extracts, cells were harvested as described above. The cell pellets were resuspended in 500 l of lysis buffer I (10 mM Tris-HCL, pH 7.4, 10 mM NaCl, 3 mM MgCl 2 , 0, 5 mM PMSF, 2 mM DTT) and incubated for 10 min on ice. After the addition of Nonidet P-40 (final concentration 0.5%), the solution was vortexed, incubated on ice for 2 min, and centrifuged (10,000 ϫ g, 2 min). The supernatant was centrifuged again to remove traces of nuclei. The amount of protein in the supernatant was determined as described, and extracts were stored at Ϫ80°C.
Western Blot Analysis-Samples of 25 g of nuclear or cytoplasmic protein extracts were loaded onto a 10% SDS-polyacrylamide gel and run for 2 h at 35 mA. Separated proteins were transferred onto a 0.2-mm cellulose nitrate membrane (Schleicher & Schü ll) in a Bio-Rad blot cell using buffer consisting of 25 mM Tris-HCl, 86 mM glycine, and 20% methanol. To avoid unspecific binding, the filters were incubated in 5% nonfat dry milk, 0.1% Tween 20 in PBS for 3 h. Thereafter, filters were incubated with monoclonal mouse antibody against hMSH2 (Calbiochem), hMSH6 (Transduction Laboratories), hPMS2 (PharMingen), hMLH1 (PharMingen), or polyclonal antibody against ERK2 (Santa Cruz Biotechnology), diluted in the same solution (1:750) overnight at 4°C, and incubated afterward with donkey anti-mouse or anti-rabbit IgG horseradish peroxidase (Amersham Pharmacia Biotech; dilution 1:4000) for 1 h. The protein-antibody complexes were visualized by ECL (Amersham Pharmacia Biotech) according to the manufacturer's protocol.
CAT Assays-The human MSH2 promoter was cloned from HeLa S3 DNA as described (29). For transient transfection, 10 6 cells were seeded/10-cm dish. Cells were transfected 1 day later with 10 g of hMSH2 promoter-CAT construct, using the calcium phosphate co-precipitation method (30). Cells were incubated 36 h after transfection with mutagens, and 6 h later, cell extracts were prepared and the amount of CAT protein per cell extract protein was determined by the CAT ELISA kit from Roche Molecular Biochemicals.
Immunoprecipitation-For preparing nuclear protein extracts, cells were harvested as described above. The cell pellets were resuspended in 1 ml of RIPA buffer (10 mM Tris-HCL, pH 8.0, 140 mM NaCl, 1 mM PMSF, 2 mM DTT) and incubated for 10 min on ice. After the addition of Nonidet P-40 (final concentration 0.5%), the solution was vortexed, incubated on ice for 5 min, and centrifuged (400 ϫ g, 5 min). The pellet was washed and resuspended in 1 ml of RIPA buffer. Thereafter, Triton X-100 (final concentration 1%) and SDS (final concentration 0.1%) were added, and samples were incubated for 10 min on ice. For co-immunoprecipitation experiments, instead of Triton X-100 and SDS, Nonidet P-40 was added (0.5%) and incubated for 20 min and centrifuged (10,000 ϫ g, 10 min). The supernatant was incubated on ice for 30 min together with 20 l of protein G and centrifuged (400 ϫ g, 2 min) to remove proteins that unspecifically bind protein G. The supernatant was incubated on ice together with 15 l of the specific antibody. 1 h later, 25 l of protein G was added and incubated for an additional 2 h. The protein-antibody complex was isolated by centrifuging it five times (400 ϫ g, 2 min), washing it in 1 ml of RIPA buffer, and subjecting it to Western blot analysis. For preparation of cytoplasmic protein extracts, cells were harvested as described above. The cell pellets were resuspended in 1 ml of RIPA buffer and incubated for 10 min on ice. After the addition of Nonidet P-40 (final concentration 0.5%), the solution was vortexed, incubated on ice for 2 min, and centrifuged (10,000 ϫ g, 2 min). The supernatant was centrifuged to remove traces of nuclei and incubated with protein G together with the specific antibody as described above.
Immunofluorescence-Cells were grown on coated glass slides in 10-cm dishes. 2 h after treatment with MNNG, cells were fixed for 30 min in PBS containing 4% paraformaldehyde and 0.2% Triton X-100. After several rinses in cold PBS, cells were preincubated for 30 min in PBS containing 1% bovine serum albumin. After additional rinses in cold PBS, cells were incubated for 2 h at 4°C with the primary antibody anti-MSH2 or anti-MSH6 (1:500) and thereafter rinsed again in PBS. After incubation with a rhodamine-conjugated anti-mouse IgG (Dako) and washing in PBS, cells were analyzed by fluorescence microscopy.

FIG. 1. Induction of GT binding activity and increase of MSH2 protein level in cells treated with alkylating agents.
A, HeLa MR cells deficient for the repair protein MGMT were incubated with different doses of MNNG for 6 h. Nuclear extracts (4 g each) were prepared as described under "Experimental Procedures" and subjected to EMSA using a radiolabeled GT oligonucleotide. The arrow indicates the GT binding complex (competition experiments not shown). B, HeLa MR cells were incubated for 6 h with different alkylating agents (MNU, ENU, and MNNG), and nuclear proteins were subjected to Western blot analysis. control, nontreated cells. The filter was incubated with MSH2 and, for loading control, ERK2 antibody.

RESULTS
Alkylating Agents Induce Increase in Nuclear Level of MSH2-Initially, we observed that treatment of HeLa MR cells with a methylating agent such as MNNG induces an increase in the level of GT binding activity in nuclear extracts (Fig. 1A). A major component of the GT binding complex is MSH2. To determine whether induction of GT binding activity is caused by an increase in the level of MSH2, we quantified this mismatch repair protein in nontreated and mutagen-exposed cells. As shown in Fig. 1B, treatment of HeLa MR cells with MNNG significantly enhanced the level of MSH2 in nuclear extracts. This was also found to be the case for other highly potent alkylating mutagens, such as MNU and ENU, having in common the capacity to induce O 6 -alkylguanine in DNA. The MSH2 protein level was not enhanced, however, after exposure of cells to MMS, which induces only very low amounts of O 6 -MeG in DNA, nor was the amount enhanced following ultraviolet light and x-ray treatment (data not shown). The doseresponse curve of induction revealed that MSH2 increased in amount linearly with doses up to 25 M MNNG, where the maximum of increase was observed (ϳ8-fold above the control level). With higher doses of MNNG (Ͼ25 M), the expression of MSH2 again declined, approaching the control level ( Fig. 2A,  left panel). We should note that even very low, nontoxic doses of MNNG (0.5 and 1 M) caused an increase in the level of MSH2 (data not shown). Time course experiments revealed an immediate-early increase of MSH2 protein detectable already 30 min after the addition of MNNG to the medium, with a maximum of expression 4 h later. The increased MSH2 level stayed above the control level for at least 24 h after methylating treatment ( Fig. 2A, right panel).
Increase in Nuclear MSH2 Level Is Caused by Translocation-Transcriptional activation of the msh2 gene or mRNA stabilization would be a reasonable explanation for an observed increase in the MSH2 protein level. This, however, was not the case, since MNNG treatment did not increase MSH2 mRNA expression (Fig. 2B). Also, the human MSH2 promoter, which was cloned into an expression plasmid and transiently transfected into HeLa MR cells, was found not to be inducible upon treatment with MNNG or with other alkylating agents (MNU, ENU, MMS) (Fig. 2C). Obviously, the observed increase in MSH2 protein level is due neither to transcriptional activation of MSH2 nor to stabilization of MSH2 mRNA. The MNNGinduced increase of the amount of nuclear MSH2 protein was independent of de novo protein synthesis, since it occurred in the presence of cycloheximide and anisomycin (data not shown). An increase in the MSH2 protein level upon MNNG treatment was observed in different cell lines from human and rodent origin, an example of which is given in Fig. 2D. Remarkably, the increase was not dependent on the p53 status, since it was found in p53 wild-type, p53 mutant, and p53 null (knockout) cells ( Fig. 2D and data not shown).
In the experiments performed so far, nuclear extracts have been used to analyze the level of MSH2 protein upon alkylation. Because MSH2 is located both in the nucleus and in relatively high amounts in the cytoplasm, the possibility arises that the observed MSH2 induction is caused by nuclear translocation of the cytoplasmic protein. This hypothesis was proven by measuring the level of MSH2 in the cytoplasmic and nuclear fraction after MNNG treatment. As shown in Fig. 3A (for a representative Western blot experiment) and Fig. 3B (for quantification), the MSH2 level clearly increased in nuclear extracts, whereas at the same time it gradually decreased in the cytoplasm, as measured at various times after methylation. This demonstrates that after exposure of cells to MNNG, MSH2 is translocated from the cytoplasm into the nucleus, leading to an increased nuclear MSH2 level.
Role of MSH6 in MSH2 Translocation-In view of the fact that MSH2 forms the GT binding complex MutS␣ together with MSH6, the question arises whether the expression level of MSH6 is also altered upon alkylation. Similar to MSH2, MSH6 was found to be present at a high level in the cytoplasm (see Fig. 4A). Upon treatment of cells with MNNG, the overall MSH6 level (determined in total cell extracts, data not shown) did not change, whereas it increased in the nuclear and decreased in the cytoplasmic fraction (Fig. 4, A and B (for quantification)). This indicates that MSH6, like MSH2, is subject to nuclear translocation in response to alkylation. We also observed an increase in the amount of two other MMR proteins, PMS2 and MLH1, within the nuclear fraction after treatment with MNNG. It is noted that we were unable to detect any MLH1 protein in the cytoplasmic fraction, which may indicate that for this protein a mechanism other than nuclear translocation is responsible for the increase in nuclear level (Fig. 4A). Nuclear translocation of MSH2 and MSH6 in MNNG-treated cells was confirmed by immunoprecipitation experiments (data not shown) and by immunofluorescence. As shown in Fig. 4C, both MSH2 and MSH6 were detectable in cells in situ in the cytoplasm, whereas upon MNNG treatment the proteins were no longer detectable there, but significant levels were found in the nucleus.
The complex composed of MSH2 and MSH6 is known to be preformed in total and nuclear extracts, which have been used for investigation in most experiments reported thus far (19,31). To find out whether MutS␣ is cytoplasmically preformed and thus subject to nuclear translocation, we performed coimmunoprecipitation experiments using nuclear and cytoplasmic extracts of HeLa MR cells. In nuclear and cytoplasmic extracts treated with an anti-hMSH6 antibody, the MSH2 protein was found to be co-precipitated. On the other hand, the addition of an anti-hMSH2 antibody to the extracts resulted in co-precipitation of MSH6 (Fig. 5). This shows that formation of the complex of MSH2 and MSH6 had occurred already in the cytoplasm. It also indicates that the preformed MutS␣ complex is translocated from the cytoplasm into the nucleus upon MNNG treatment. If true, this would suggest that lack of either one of the proteins prevents nuclear translocation of the second protein. This is indeed the case, because in DLD1 cells deficient for MSH6 (32), translocation of MSH2 into the nucleus was not found (Fig. 6A). Also, as one would expect, GT binding activity of nuclear extracts did not increase in these cells upon MNNG treatment, in contrast to HeLa MR control cells (Fig. 6B). (At the same time, this experiment confirmed the MSH6 deficiency of the cells we used.) Obviously, nuclear translocation of MSH2 requires MSH6. In this context, it should be noted that computer analysis revealed that MSH6 contains nuclear localization signal (NLS) sequences not found in MSH2 protein (data not shown). It thus appears that the NLS of MSH6 are utilized for transportation of the preformed cytoplasmic MutS␣ complex into the nucleus upon DNA methylation.
Role of O 6 -MeG in Nuclear Translocation of MutS␣-Increases in GT binding activity and nuclear accumulation of MSH2 and MSH6 were observed upon treatment of cells with agents that induce O 6 -MeG in DNA (see Fig. 1). O 6 -MeG is removed from DNA by the repair protein MGMT in a relatively fast methyl group transfer reaction (33). The HeLa cells (strain HeLa MR) utilized for the experiments described above are deficient for MGMT and thus unable to remove O 6 -MeG from DNA (34). If O 6 -MeG provided the signal triggering the response observed, one would expect cells deficient in MGMT to be more sensitive than MGMT proficient cells as regards to MSH2 translocation and the induction of GT binding activity. This indeed was the case. As shown in Fig. 7A, an increase of MSH2 protein also occurred in cells expressing MGMT (HeLa S3 cells, 750 fmol/mg protein MGMT) although at a much higher dosage level of the mutagen (Ͼ25 M MNNG), at which point MGMT repair capacity appears to be saturated. When HeLa S3 cells were preincubated with O 6 -benzylguanine (O6- HeLa MR cells 2 h after mutagen exposure. The filter was reprobed with hERK2 for loading control. B, quantification occurred as described in Fig. 3 by densitometric analysis. Induction factors (in relation to the non-mutagen-treated control) of the individual proteins in the nuclear fraction are given in the column diagram. C, immunofluorescence analysis of HeLa MR cells that were nontreated (control) or treated with MNNG (25 M, 2 h). Cells were fixed by paraformaldehyde, permeabilized by Triton X-100, and incubated with MSH2 and MSH6 monoclonal antibody as described (see "Experimental Procedures"). Thereafter, they were incubated with a rhodamine-conjugated anti-mouse IgG antibody, which was detected by fluorescence microscopy. Note that after MNNG treatment, MSH2 and MSH6 were no longer detectable in the cytoplasm.
FIG. 5. Immunoprecipitation of MSH2 and MSH6. Nuclear and cytoplasmatic extracts were isolated from HeLa MR cells, and MSH2 or MSH6 were co-immunoprecipitated using the corresponding monoclonal antibodies. The precipitated proteins were subjected to Western blot analysis, again using MSH2 and MSH6 antibodies to probe the filter. BG), which specifically inactivates MGMT (35), the same cells clearly became more sensitive, thus resembling MGMT-deficient HeLa MR cells in their response to MNNG (Fig. 7B). On the other hand, when we transfected human MGMT cDNA into HeLa MR cells, resulting in strong MGMT expression in this cell type (1467 fmol/mg of MGMT versus nondetectable levels in HeLa MR cells), MSH2 translocation was completely abrogated (Fig. 7C).
The addition of O6-BG to HeLa S3 cells prior to MNNG treatment not only led to an increase in the nuclear MSH2 protein level but also evoked the appearance of the GT binding complex, which was not seen without the inhibitor (Fig. 8). In the control experiment with HeLa MR cells, O6-BG was unable to enhance further GT binding (Fig. 8), indicating that drug treatment on its own was not responsible for the effect observed. In HeLa MR cells stably transfected with the ada gene of Escherichia coli (36), which is as effective as MGMT in repairing O 6 -MeG in the transgenic cell line (37,38), the increase of GT binding upon MNNG treatment was abrogated; thus, the cells behaved like HeLa S3 cells. In contrast to MGMT, the Ada protein is not subject to inhibition by O6-BG (39). As expected, treatment of HeLa MR ada-transfected cells with O6-BG did not lead to an increase of GT binding as compared with HeLa S3 cells (Fig. 8). Collectively, the data indicate that O 6 -MeG lesions are involved primarily in triggering nuclear translocation of MSH2 and MSH6.

DISCUSSION
Mammalian cells respond on exposure to genotoxic agents with the activation of a variety of functions, because of posttranslational modification of pre-existing proteins and gene activation (for review see Ref. 27). Here, we report an increase in the nuclear level of mismatch repair proteins upon alkylation. We show that the treatment of cells with agents having in common the ability to alkylate DNA in the O 6 position of guanine (e.g. MNNG, MNU, ENU) leads to an increase in the nuclear level of MSH2 protein. MMS, which is a powerful DNA methylating agent as well, was ineffective in eliciting this response. MMS is a S N 2 agent that predominantly induces N-alkylations. As opposed to MNNG and MNU, only a very small portion of DNA alkylations comprises O 6 -MeG (0.3 versus 8% of total alkylations for MMS and MNNG/MNU, respectively) (40). Thus, the inability of MMS to induce significant GT binding (data not shown) and an increase in MSH2 protein may be taken to indicate that neither DNA methylation per se nor methylation of other cellular constituents such as proteins, but O 6 -MeG itself, triggers the response. Interestingly, MMS is highly active in inducing other cellular responses such as epidermal growth factor receptor signaling and fos/jun activation (41,42). Obviously, the response reported here is not related to these immediate-early signal transduction pathways. An increase of nuclear MSH2 and MSH6 as well as GT binding activity was found specifically upon treatment of cells with O 6 -alkylguanine-generating mutagens, as noted above, as well as the anti-neoplastic agent streptozotocin (data not shown); it was not observed after treatment with UV light and x-rays. We also did not observe the phenomenon after treatment with melphalan and the tumor promoter 12-O-tetradecanoylphorbol-13-acetate (TPA; data not shown). Previously, it was shown that treatment of human neuroblastoma cells with doxorubicin leads to an increase of nuclear MSH2 and MSH6 protein level (43). Whether this was related to nuclear translocation or because of another mechanism is unknown.
The increase in nuclear MSH2 protein upon MNNG treatment turned out to be dose-and time-dependent and was associated with a clear increase in the GT binding activity of nuclear extracts. An increase of MSH2 was already detectable after treatment of cells with 0.5 M MNNG, a dose insufficent to cause cell death. Also, with the other treatments applied, immediate cytotoxic effects were not observed, indicating that the response is not a side effect of unspecific cytotoxicity or cell damage. Because already very low doses of a methylating agent were able to trigger increases in nuclear MSH2 and MSH6 expression, the response is likely to be of physiological relevance. The observed increase in MSH2 protein level is not caused via promoter activation and is not a result of mRNA stabilization. By examining the cellular distribution of MSH2 protein both in cell extracts and by immunohistochemistry in situ, we showed that the increase of nuclear MSH2 protein is caused by translocation of MSH2 from the cytoplasm into the nucleus. This translocation is independent of the p53 status of the cells and is a cross-species phenomenon. Furthermore, we demonstrate that not only MSH2 but also MSH6 and PMS2 proteins are translocated from the nucleus into the cytoplasm. Co-immunoprecipitation of MSH2 and MSH6 from cytoplasmic and nuclear extracts revealed that the assembly of the MutS␣ complex has already occurred in the cytoplasm. Therefore, we surmise that, upon treatment of cells with O 6 -MeG generating agents, MSH2 and MSH6 are translocated from the cytoplasm into the nucleus as preformed MutS␣ complex. Using different HeLa strains either lacking the DNA repair protein MGMT (HeLa MR) or expressing it (HeLa S3 and HeLa MR transfected with MGMT cDNA), we showed that cells having the ability to repair O 6 -MeG are less responsive than O 6 -MeG repair-deficient cells in regard to alkylation-induced translocation of MSH2 and MSH6. This finding was confirmed by depletion experiments with O 6 -benzylguanine, which specifically inactivates MGMT. Overall, the data indicate that O 6 -MeG is primarily involved in triggering translocation and the subsequent nuclear accumulation of MSH2 and MSH6.
O 6 -MeG has been shown to result in the formation of distortions in DNA upon pairing with cytosine or thymine, both of which are targets for MutS␣, because purified MutS␣ is able to bind both to O6-MeG/C and O6-MeG/T base pairs (10). This finding is confirmed in our study showing that, in nuclear extracts, MutS␣ is able to bind to an oligonucleotide containing a single O 6 -MeG/C pair; this binding is elevated upon MNNG treatment of cells (data not shown). Because nuclear translocation of MSH2 and MSH6 occurs immediately after exposure of cells to O 6 -MeG-generating agents detectable already 30 min after the beginning of treatment, where only insignificant numbers of O 6 -MeG/T mismatches will be formed in replicating cells, we hypothesize that O 6 -MeG/C base pairs are the major early trigger for nuclear accumulation of MSH2 and MSH6 and the induction of nuclear GT binding activity.
How do O 6 -MeG lesions trigger nuclear translocation of MutS␣? The presence of O 6 -MeG in DNA could provide in itself the signal leading to translocation of MutS␣. Another possibility is that binding of MutS␣ to O 6 -MeG lesions leads to sequestration of free MutS␣ in the nucleus, which finally may provide the signal for transportation of MSH2 and MSH6 from the cytoplasm into the nucleus. Nuclear translocation of MutS␣ is unlikely to be the result of a passive dislocation of the complex into the nucleus because MutS␣ and even the single components MSH2 (102 kDa) and MSH6 (160 kDa) are too big to enter the nucleus through the nuclear pore simply by diffusion. Only proteins up to 40 kDa are able to get through the nuclear pore complex in a passive way (for review see Ref. 44). Therefore, MSH2 and MSH6 must enter the nucleus actively by NLS-mediated transport. Analysis of the amino acid sequence of MSH2 and MSH6 reveals a lack of NLS in MSH2, whereas MSH6 contains several polyoma large T-and SV40 large Tantigen-like NLS (45) as well as several bipartite sequences (46). The presence of NLS sequences within the MSH6 protein may explain the finding that nuclear translocation of MSH2 is related to the translocation of MSH6. Thus, MSH2 alone was not effective in entering the nucleus upon alkylation, which was proven by using DLD1 cells that are deficient in MSH6. Based on these data, we conclude that MSH6 is utilized for the nuclear transportation of MSH2, hooking MSH2 into the nucleus.
The involvement of an active transport would imply that signaling triggered by O 6 -MeG lesions leads to post-translational modification of MMR components that stimulate nuclear translocation. Post-translational protein modification can lead to changes in the NLS. Thus, the nuclear translocation of SV40 T-antigen is regulated by phosporylation of the NLS or NLS flanking sequences within the protein (47). Another possibility is that post-translational modification alters the assembly of the MutS␣ complex. Differences in nuclear transportation between monomeric proteins and dimer complexes have been shown for MAPK (48). This protein is bound to MAPK kinase in the cytoplasm, is released after phosphorylation, and enters FIG. 7. Translocation of MSH2 is triggered by the induction of O 6 -MeG in DNA. A, HeLa S3 cells proficient for MGMT were not incubated (control) or were preincubated with 20 M O6-BG for 1 h and thereafter treated with different doses of MNNG for 2 h. Nuclear extracts were isolated and subjected to Western blot analysis. The filter was probed with MSH2 and reprobed with ERK2 antibody, which served as a loading control. B, MSH2 expression in the nucleus as a function of dose of MNNG in HeLa S3, HeLa MR, and HeLa S3 cells pretreated with O6-BG. B, quantification of data (shown under A and blots not shown) occurred as described in Fig. 3. C, HeLa MR cells transfected with mgmt-cDNA and, for control, with the neo gene only were treated with different doses of MNNG for 2 h. Thereafter, nuclear extracts were prepared and subjected to Western blot analysis. The filter was probed with hMSH2 and hERK2 antibody. MSH2 expression was quantified and set in relation to ERK2. The relative MSH2 expression levels are shown in the column diagram. the nucleus by passive diffusion. Upon treatment with serum, MAPK forms homodimers that can enter the nucleus by the more efficient active transport. Currently, we are analyzing whether MSH2 and MSH6 are subject to post-translational modification. Our preliminary data show that MSH2 and MSH6 can be phosphorylated in vitro and under in vivo conditions. 2 The data reported here indicate that regulation of MMR upon DNA damage occurs largely at the level of post-translational modification (including nuclear transportation) rather than at the level of gene activation. The immediate-early translocation of MMR proteins into the nucleus, triggered very likely by O 6 -MeG/C and O 6 -MeG/T lesions, is supposed to provoke an increase in MMR capacity in the nucleus; this would be important, in view of O6-MeG/C lesions that form during replication highly mutagenic GT mismatches. Defects in MMR are associated with various hereditary forms of cancer (2,18,49); they have also been shown to increase dramatically the resistance of cells to O 6 -MeG-generating agents, notably if MGMT is not expressed or is expressed at a low level (7,8,14,29,50). MMR defects, therefore, have a strong impact on the mutagenic and carcinogenic response of cells exposed to alkylating agents. The data presented here imply that not only a deficiency of MMR proteins but also defects in nuclear translocation could alter the mutagenic and carcinogenic response of cells to endogenously formed methylating species, environmental carcinogens, and methylating drugs used in tumor therapy.