|Home | About | Journals | Submit | Contact Us | Français|
DNA mismatch repair (MMR) maintains genomic integrity by correction of mispaired bases and insertion-deletion loops. The MMR pathway can also trigger a DNA damage response upon binding of MutSα to specific DNA lesions such as O6methylguanine (O6meG). Limited information is available regarding cellular regulation of these two different pathways. Within this report, we demonstrate that phosphorylated hMSH6 increases in concentration in the presence of a G:T mismatch, as compared to an O6meG:T lesion. TPA, a kinase activator, enhances the phosphorylation of hMSH6 and binding of hMutSα to a G:T mismatch, though not to O6meG:T. UCN-01, a kinase inhibitor, decreases both phosphorylation of hMSH6 and binding of hMutSα to G:T and O6meG:T. HeLa MR cells, pretreated with UCN-01 and exposed to MNNG, undergo activation of Cdk1 and mitosis despite phosphorylation of Chk1 and inactivating phosphorylation of Cdc25c. These results indicate that UCN-01 may inhibit an alternative cell cycle arrest pathway associated with the MMR pathway that does not involve Cdc25c. In addition, recombinant hMutSα containing hMSH6 mutated at an N-terminal cluster of 4 phosphoserines exhibits decreased phosphorylation and decreased binding of hMutSα to G:T and O6meG:T. Taken together, these results suggest a model in which the amount of phosphorylated hMSH6 bound to DNA is dependent on the presence of either a DNA mismatch or DNA alkylation damage. We hypothesize that both phosphorylation of hMSH6 and total concentration of bound hMutSα are involved in cellular signaling of either DNA mismatch repair or MMR-dependent damage recognition activities.
DNA mismatch repair (MMR) plays a crucial role in maintaining genomic integrity in all living cells. The MMR pathway corrects mispairs and insertion-deletion loops (IDLs), resulting from errors during replication, recombination and other DNA metabolic processes. Loss of MMR activity is associated with hereditary non-polyposis colorectal cancer and several sporadic cancers [1–5].
Repair of a DNA mismatch within human cells is initiated when hMutSα (hMSH2-hMSH6) binds to the mismatch [6–9]. hMutLα (hMLH1-hPMS2) then acts as a “matchmaker” between the hMutSα-DNA complex and downstream enzymatic components responsible for excision and resynthesis of the DNA strand containing the incorrect base [3,10,11]. The regulation of overall MMR activity within the cell is not as well understood as binding interactions of purified hMutSα with various DNA constructs. Highest MMR activity occurs during S phase of the cell cycle . The localization of hMutSα within DNA replication factories during DNA synthesis likely contributes to the high fidelity of MMR during S phase [12,13]. Upon exposure of the cell to low, chemotherapuetic levels of alkylating agents however, there is a rapid and prolonged relocation of hMutSα from active replication factories to other areas of the chromosome [14,15]. Thus, MMR activity is regulated differently during chromosomal replication as opposed to a response to DNA damage. Monofunctional alkylating agents such N-methyl-N’-nitro-N-nitrosoguanidine (MNNG) produce several alkylated DNA adducts, the majority of which are repaired efficiently by the base excision repair pathway (BER) [16–18]. The O6meG modification is directly repaired by the covalent transfer of O6methyl to Methylguanine Methyltransferase (MGMT) . If MGMT is not expressed or DNA replication occurs before repair of O6meG, thymidine is frequently misinserted opposite the damaged guanine [20,21]. The hMutSα complex binds to O6meG:T, initiating a DNA damage response, rather than effective repair [4,22–25]. This binding is required for recruitment and activation of ATR-ATRIP, phosphorylation of Chk1 and subsequent activation of the DNA damage-signaling pathway . Alkylation-induced DNA damage checkpoint arrest in G2 and apoptosis is decreased in MMR-deficient cells [4,26–29]. In addition, murine Msh2 and Msh6 have been mutated in their ATPase domains to demonstrate genetic separation-of-function. Mouse cells harboring these altered proteins have deficient MMR but retain an intact DNA damage-signaling response to alkylating agents [22,23].
Twenty-two distinct phosphorylation sites have been identified in hMSH6 by phospho-proteome mass spectrometry. The majority of these sites are located in six clustered regions within the N-terminus of this protein [30–37]. The functional significance of these clustered phosphorylation sites is unknown, although both increased MMR activity and hMSH6 phosphorylation have been demonstrated after kinase activation or alkylation treatment [38–41]. In this study, we demonstrate that MSH6, but not MSH2, is phosphorylated in nuclear extracts and in intact cells. The phorbol ester, TPA, a classical kinase activator, increases hMSH6 phosphorylation and hMutSα binding to a G:T mismatch, although not to an O6meG:T lesion. The kinase inhibitor, UCN-01, decreases hMSH6 phosphorylation and hMutSα activity for both G:T and O6meG:T. Further, less hMutSα is bound to G:T than is bound to O6meG:T. Therefore phosphorylation of hMSH6 and the quantity of hMutSα bound to DNA may be different cellular mechanisms that regulate activity between mismatch repair and DNA damage-signaling activity.
Monoclonal antibodies against hMSH6 (BD Biosciences), hMSH2 (Calbiochem), phospho-serine (Calbiochem), Chk1 (Santa Cruz Biotechnology), phospho-Chk1 Ser345 and Ser317 (Cell Signaling), phospho-Cdc25c Ser216 (Santa Cruz Biotechnology), Cdc2 (Cdk1) (Cell Signaling), and phospho-Cdc2 (pCdk1) Tyr15 (Cell Signaling) were purchased from the indicated suppliers. Protease inhibitor cocktails were purchased from Calbiochem. E 64 protease inhibitor, N-methyl-N'-nitro-N-nitrosoguanidine (MNNG), 7-hydroxystaurosporine (UCN-01), and 12-O-tetradecanoylphorbol 13-acetate (TPA) were from Sigma. [32P]-orthophosphoric acid ([32P]-H3PO4) was purchased from MP Biomedicals and lambda protein phosphatase (λPPase; serine-threonine phosphatase) from New England Biolabs. [α-32P]-dATP and [γ-32P]-ATP were purchased from Amersham. Unlabeled adenosine 5’-triphosphate (ATP) was from Calbiochem. Protein A/G beads and dialysis cassette were obtained from Pierce Biotechnology and PVDF membrane from Millipore. All other reagents were purchased from Sigma unless otherwise noted. Synthetic oligomers of 69 bases containing the coding strand sequence 5’-AATTCACGGAATATAAGCTGGTGGTGGTGGGCGCCGGCGGTTGGGCAAGAGTGCGCTGACCATCCAGG-3’, and complementary noncoding oligomers were obtained from Operon. The above 69 base-pair oligomer is a portion of the coding sequence of human H-ras DNA . The bolded G represents H-ras codon 12 middle G position. Complementary strands containing either C or T opposite codon 12 middle base G, were annealed in equimolar ratio at a final concentration of 0.2 µg per µL 1 mM Tris-HCL and 1 mM MgCl2, pH 7.5. For O6meG oligomer, bold G in the coding strand was replaced with O6-meG. hMSH6 cDNA in Bluescript KS plasmid and hMSH2 cDNA in pCITE3b plasmid were kind gifts from Dr. Josef Jiricny. The cDNAs were excised and cloned into a pFastBacDual vector (Invitrogen) for stable co-expression of the proteins as hMutSα. The KpnI and XhoI restriction enzyme sites flanking hMSH2 cDNA and the BamHI and NotI restriction enzyme sites flanking hMSH6 cDNA were used for cloning into pFast-BacDual vector. Recombinant baculoviruses with hMSH2 and hMSH6 cDNAs were generated following the manufacturer’s protocol (Invitrogen). The presence and orientation of each gene was confirmed by polymerase chain reaction and DNA sequencing. Quick-change II XL (Stratagene) was used to make phosphorylation site-specific mutants following the manufacturer’s protocol. Baculoviruses with empty pFastBacDual plasmids (control) also were generated.
Sf9 insect cells were infected with either control baculovirus or MSH2/MSH6 recombinant baculovirus (5:1 multiplicity of infection). hMutSα was purified as follows. On day 1 after infection, E 64 protease inhibitor was added to the culture (1 µL per ml medium). On day 3, the cells were harvested, and whole cell extracts prepared as described below. Cell extract was dialyzed overnight in Buffer A (25 mM HEPES, 1 mM DTT, 0.1 mM EDTA, 200 mM NaCl and 20% glycerol). The dialyzed extract was centrifuged at 10,000 rpm for 10 min. The resulting supernatant was diluted to 150 mM NaCl in Buffer A and passed through a pre-equilibrated phosphocellulose column (Whatman). Proteins were eluted with buffer A in a linear salt gradient ranging from 0.2 M to 1 M NaCl. The peak phosphocellulose elution fractions were pooled, diluted to 250 mM NaCl, and passed through a pre-equilibrated heparin-sepharose column (GE healthcare). Proteins were eluted with Buffer A using a linear salt gradient ranging from 0.3 M to 1 M NaCl. The peak fractions from the heparin column were diluted to 250 mM NaCl and passed through a PBE 94 column (Pharmacia). Proteins were eluted with buffer A with a linear salt gradient ranging from 0.25 M to 1 M NaCl. Peak protein fractions were pooled, diluted to 100 mM NaCl, and passed through a pre-equilibrated S-column (Biorad). Proteins were eluted with 10 ml of a linear salt gradient of 0.15 M – 0.5 M NaCl in Buffer A. The fractions containing the pure proteins (as determined by SDS-PAGE with Commassie staining) were pooled, dialyzed extensively against Buffer A, and stored in aliquots at −80°C. All buffers for protein preparation contained protease inhibitors; Pepstatin (1mg/ml), Leupeptin (1mg/ml) and 0.1M phenylmethanesulfonyl fluoride. Protein concentrations in eluted fractions were determined by Bradford protein analysis. Supplement Figure 3 is a demonstration of the purity and activity of recombinant hMutSα used for the experiments described below.
NIH 3T3 cells (ATCC) were grown in Dulbecco's Modified Eagle's Medium (DMEM) with 10% fetal bovine serum (FBS). HeLa S3 (ATCC) and Hec59 cells (hMutSα−/−; ATCC) were grown in Iscove’s Dulbecco’s Medium with 10% FBS. HeLa MR cells (kind gift from Dr. Sankar Mitra) were grown in DMEM/Ham's F12 50/50 mix with 10% FBS. HeLa MR cells do not express MGMT . All cultures were incubated at 37 °C in a 5% CO2 humidified atmosphere.
Nuclear extracts, whole cell lysates and cross-linked chromatin were prepared as previously described [14,15,43]. Protein concentrations were determined by Bradford protein analysis. DNA concentrations were determined by the ratio of absorbance at 260/280 nM. One hundred µL aliquots were snap-frozen in pre-chilled 1.5 ml tubes on dry ice and stored at −80°C.
Immunoblots of proteins separated by SDS or nondenaturing polyacrylamide gel electrophoresis (PAGE) were performed as described [14,15]. Immunoreactive proteins were visualized by enhanced chemiluminescence (ECL) reagent (0.1M Tris HCl, p-Coumaric acid, 3-Aminophthalhydrazide, 30% H202) and exposure to X-ray film. Protein bands were quantified from ECL activity by Alpha Innotech FluorChem HD2. Each immunoblot experiment was repeated at least three times.
During early proliferation (50–60% confluent), the medium was replaced with phosphate- and serum-free DMEM for 5 h to deplete endogenous phosphate. The medium was then replaced with fresh phosphate-free DMEM supplemented with 1% dialyzed FBS and [32P]-H3PO4 (specific activity: 285 Ci/mmol) at a final concentration of 0.5 mCi/ml. After 15 h, cells were harvested in 400 mM NaCl, 1mM EDTA, 0.1% Triton X 100 and 10 mM Tris, pH 8.0 with phosphatase and protease inhibitors. The extracts were centrifuged briefly, hMSH6 was immunoprecipitated and subjected to SDS-PAGE . The dried gel was exposed to X-ray film, and phosphorylation of immunoprecipitated hMSH6 was quantified with a phosphorimager. Each cellular radioactive labeling experiment was performed in triplicate.
Recombinant hMuSα was added to MMR-proficient nuclear extracts or MMR-deficient cell lysates and incubated with 50 µCi of [γ-32P]-ATP and 250 µM unlabeled ATP at 30°C for 30 min. Equimolar concentrations of unlabeled G:C, G:T or O6meG:T 69mers were added to reactions as noted. For dephosphorylation measurements, phosphorylated solutions were incubated with λPPase for 20 min at 30°C. All reactions were carried out in 40 mM Tris pH 7.6, 10 mM Mg Cl2, 0.2 M KCl, 100 µg/ml bovine serum albumin, phosphatase and protease inhibitors. After incubation, hMSH6 was immunoprecipitated and subjected to SDS-PAGE . The dried gel was exposed to X-ray film and phosphorylation of immunoprecipitated hMSH6 was quantified with a phosphorimager. Nuclear extract radioactive labeling experiments were repeated a minimum of three times.
Protein-DNA binding assays were performed using 0.8 – 1.0 × 105 cpm of 69mer and 2–5 µg nuclear extract in an equal volume of 2 X electrophoretic mobility shift reaction buffer (12% glycerol, 2 mM MgCl2, 10 µM EDTA, 10 µM DTT, 0.1 M NaCl, 20 mM Tris-HCL pH 7.5, 0.25 mg/ml PolydI-dC, 1 mg/ml BSA). Nuclear extracts were pre-incubated with 100 X molar excess non-radioactive homoduplex (G:C) 69mer in the same buffer for 20 min on ice. Electrophoretic mobility shifts were performed using nuclear extracts or recombinant hMutSα. Annealed oligomers were radioactively labeled by a fill-in reaction, using [α-32P]-dATP and Klenow polymerase, per manufacturer’s protocol (Invitrogen). 32P-labeled 69mer duplexes containing either a G:C, G:T, or O6meG:T were added to the reactions, and incubation continued for 15 min at room temperature. Each experiment also included reactions without oligomers, with 100X unlabeled G:T oligomer, or with antibody to hMSH6. Final reactions contained 1 mM MgCl2, 5 µM EDTA, 5 µM DTT, 50 mM NaCl, 10 mM Tris-HCL, pH 7.5, 0.125 mg/ml poly[dI:dC], 0.5 mg/ml bovine serum albumin and 6% glycerol in a volume of 20 µl. The reactions were subjected to electrophoresis in a 6% non-denaturing polyacrylamide gel using 0.5 X TBE buffer at 30 mA for 2–3 h at 4°C. Radioactivity was measured using a phosphorimager, and gel-shifted bands were visualized by exposure to X-ray film. Each experiment using radioactively labeled oligomers was performed a minimum of three times. Protein-DNA binding reactions for electrophoretic mobility shift-immunoblots (EMSA-immunoblot) analyses were performed as described above but with nonradioactive oligomers. After nondenaturing PAGE, proteins were transferred to a PVDF membrane and subjected to immunoblot analysis of hMSH6. Protein bands were quantified from ECL activity by Alpha Innotech FluorChem HD2. Each protein immunoblot experiment using unlabeled oligomers was performed a minimum of three times at different protein concentrations.
To determine if either kinase activator or inhibitor alters the DNA binding activity of endogenous hMutSα, proliferating HeLa MR cells were pre-incubated for 12 h with either 200 nM TPA or 200 nM UCN-01. Nuclear extracts were then incubated with radioactively labeled G:T oligomers and subjected to EMSA using nondenaturing polyacrylamide gels. Each experiment involving TPA or UCN-01 pre-incubation was performed at least three times.
Cells were synchronized at late G1/early S by double thymidine block (DTB) . When present, 2 µM MNNG was added to the medium immediately after release from DTB. Distribution of cells in different phases of the cycle was monitored by staining nuclei with propidium iodide followed by analysis of DNA concentration in a Beckman/Coulter EPICS Elite flow cytometer . The data were analyzed by Multicycle software (Phoenix Flow Systems) and reported as the percentage of cells in G1, S, or G2 phase. Cell cycle synchronization experiments by DTB have been performed numerous times in our laboratory and are highly reproducible[14,15].
Chromatin was cross-linked by fixing cells within the culture plates, as described previously . Purification of chromatin and immunoprecipitation with anti-hMSH2 antibody, SDS-PAGE and immunoblot analysis with anti-hMSH6 was also described previously .
Phosphorylation of hMSH6 occurs in whole cells incubated with [32P]-H3PO4 and nuclear extracts incubated with [γ-32P]-ATP. Phosphorylation of hMSH2 was undetectable (Supplement Figure 1). Twenty-two distinct phosphorylated sites have been identified within hMSH6 by mass spectrometry studies of HeLa nuclear extracts [30–37]. Twenty of these sites are clustered into six groups of 2–5 amino acids within the first 348 residues of the unstructured N-terminal region of hMSH6 (Supplement Figure 2). We therefore asked if the cellular environment could affect hMSH6 phosphorylation and, in turn, if alterations in hMSH6 phosphorylation could affect hMutSα mismatch repair or activities that respond to alkylation-induced DNA damage. We hypothesized that mismatched (G:T), alkylated (O6meG:T) oligomers, kinase activator (TPA), and kinase inhibitor (UCN-01) would affect hMSH6 phosphorylation and activities of hMutSα in different manners.
Using recombinant hMutSα (Supplement Figure 3), we found that a G:T mismatch stimulates recombinant hMSH6 phosphorylation in the nuclear environment. Specifically, in Hec59 (hMutSα−/−) cell lysates the amount [32P]-labeled recombinant hMSH6 immunoprecipitated in the presence of the G:T mismatch was three times the levels of radioactively-labeled hMSH6 immunoprecipitated in the absence of oligomer or in the presence of the G:C oligomer (Supplement Figure 3D).
In HeLa MR cells TPA causes a significant increase in binding of hMutSα to G:T relative to that in untreated cells whereas UCN-01 decreases hMutSα G:T binding activity significantly below that of untreated cells (Figure 1A). Amount of hMSH6 and hMSH2 protein was not significantly different in nuclear extracts from control or treated HeLa cell populations indicating that neither TPA-provoked increase nor UCN-01-provoked decrease in mismatch binding activity was due to altered concentrations of hMutSα (Figure 1B). When average hMutSα binding activity was normalized to hMSH6 protein concentration, TPA increased hMutSα mismatch binding activity (ratio of 1.21) and UCN-01 decreased binding by over one-half (ratio of 0.48) (Figure 1C).
Is phosphorylation of endogenous hMSH6 altered by the presence of a mismatch or correlated to changes in kinase activity? Incubation of HeLa nuclear extracts with a G:T oligomer increased phosphorylation of endogenous hMSH6 significantly (Figure 2A). TPA further increased phosphorylation of hMSH6 significantly in the presence of both the G:C and G:T oligomer (Figure 2A). Treatment of HeLa cells with UCN-01 significantly decreased phosphorylation of endogenous hMSH6 by an average of 55% (Figure 2B), consistent with its effect on mismatch binding activity (Figure 1A). Addition of G:C or G:T oligomers did not have an effect on this result (results not shown).
The hMutSα heterodimer binds to mismatches and triggers activation of the MMR pathway. This heterodimer also binds to O6meG:T and triggers the DNA damage-signaling pathway . We hypothesized that phosphorylation of hMSH6 is affected by the type of DNA lesion, and is involved in signaling either activation of MMR or the DNA damage-signaling pathway. To determine if endogenous kinase activity affects the amount of O6meG:T oligomer bound by endogenous hMutSα, we determined DNA binding activity using HeLa MR nuclear extracts purified from cells incubated with either TPA or UCN-01. The amount of O6meG:T shifted by nuclear extract is not significantly different in HeLa MR cells incubated with or without TPA (Figure 3A). However, treatment of cells with UCN-01 significantly decreased the amount of O6meG:T shifted (Figure 3A). TPA had no significant effect on the increased phosphorylation of hMSH6 in the presence of O6meG:T (Figure 3B). In contrast TPA increased phosphorylation of hMSH6 with G:C or G:T oligomer present (Figures 2A and and3B).3B). These results indicate that TPA does not increase phosphorylation of endogenous hMSH6 in the presence of O6meG:T nor does hMutSα have increased binding to alkylated lesions in the presence of TPA (Figure 3A). However, UCN-01 significantly decreased binding of hMutSα to this alkylated lesion (Figure 3A), consistent with the decreased electrophoretic mobility shift of the G:T oligomer (Figure 1A) and decreased phosphorylation of hMSH6 (Figure 2B) caused by UCN-01.
EMSA demonstrates that more G:T than O6meG:T radioactive oligomer is bound (shifted) in the presence of equal amounts of hMutSα (Figure 4A). These results are in agreement with others [44,45]). An EMSA-immunoblot measures the amount of hMSH6 protein bound to unlabeled oligomer (rather than the amount of shifted radioactive oligomer). In contrast to the results with labeled oligomers, we consistently observe significantly more hMSH6 bound to the shifted O6meG:T oligomer as compared to the shifted G:T (Figure 4B). Therefore, in the presence of equimolar concentrations of G:T and O6meG:T, our results agree with others in that hMutSα binds less O6meG:T oligomer than G:T oligomer (Figure 4A). In contrast, the amount of hMSH6 bound to O6meG:T from equal amounts of total nuclear protein is statistically higher (P=0.04) than to G:T (Figure 4B).
To determine if decreased phosphorylation of endogenous hMSH6 alters the amount of hMutSα bound to either the G:T or O6meG:T oligomer, we used HeLa MR nuclear extracts from the cells incubated with UCN-01 to perform additional EMSA-immunoblots. UCN-01 decreased the amount of HeLa hMSH6 protein bound to both types of oligomers, in the presence of equal hMSH6 protein concentrations in treated and untreated nuclear extracts (Figure 5A). Therefore, UCN-01 decreases the amount of hMutSα bound to both types of oligomers, in agreement with less DNA shifted (Figure 3A), and less phosphorylation of hMSH6 after UCN-01 treatment of cells (Figure 2B). To determine if UCN-01 inhibits binding of hMSh6 to O6meG within chromosomes, synchronized HeLa MR cells were pre-incubated with UCN-01 and exposed to MNNG 2 h later. Protein-bound chromosomal DNA was purified 5 h later (late S, early G2 phase ). Similar to inhibition of hMSH6 binding to oligomers containing G:T or O6meG:T, UCN-01 decreases binding of hMSH6 to alkylation-treated chromosomes, as well as decreasing constitutive binding to untreated chromosomes (Figure 5B).
The UCN-01-induced decrease in DNA binding activity for both G:T and O6meG:T appeared similar to the effects of ATP in DNA binding activity experiments. DNA-bound hMutSα releases the mismatched oligomer in an “ATP-dependent” manner (Supplementary Figure 4A and references [46,47]). However UCN-01 does not cause the same decrease in hMSH6 as that caused by ATP. ATP causes release of hMSH6 and hMSH2, which migrate as intact hMutSα below the gel-shifted oligomer in the nondenaturing polyacrylamide gel (Supplementary Figure 4B). In UCN-01-treated HeLa MR cells, a much smaller amount of hMSH6 is initially bound to either G:T or O6meG:T, and unbound hMSH6 migrating below the DNA-bound protein is undetectable (Figure 5C). Therefore, a different mechanism, in addition to dephosphorylation of hMSH6, appears to be inhibiting the binding of hMutSα to both oligomers, rather than ATP-dependent release.
UCN-01 directly inhibits activation of Chk1 kinase that, in the presence of sufficient DNA damage, normally triggers a DNA damage response and cell cycle arrest [15,48,49]. We therefore assessed the effect of UCN-01 on activation of Chk1 kinase in the MNNG-induced DNA damage response pathway. After synchronization of HeLa MR cells by double thymidine block (DTB), we exposed the cells to 2 µM MNNG and 12 h later either added UCN-01 or left the cells untreated. This concentration of MNNG results in phosphorylated Chk1 at 24 – 32 h after alkylation damage (Figure 6A and reference ). At 12 h – 24 h after additional UCN-01 treatment (24 h – 32 h after MNNG), the phosphorylation of chromatin-crosslinked Chk1 was not inhibited, in agreement with the literature that UCN-01 does not inhibit ATR activity . Also, UCN-01 does not appear to inhibit MNNG-induced Chk1 kinase activity, as evidenced by persistence of phosphorylated Cdc25c in cytoplasmic extract. This is not consistent with in vitro experiments reported by others . Several other kinases are known to phosphorylate Cdc25c at S216 as well, therefore inhibition of Chk1 cannot be ruled out by these results . However UCN-01 does inhibit phosphorylation of Cdk1 at tyrosine 15 within cells exposed to MNNG, therefore allowing the cell to escape the G2 arrest occuring after MNNG treatment alone (Figure 6A & 6B and reference ). The combined MNNG and UCN-01 treatment subsequently resulted in 100% cell death (Figure 6B at 48 h, and 0% colony survival, results not shown). Flow cytometric analysis of the cell cycle demonstrated that UCN-01 treatment alone did not alter cell cycle kinetics (Figure 6B) and colony survival was 100% (results not shown). Taken together, these results suggest that activation of Chk1 may not play a primary role in the inhibition of hMSH6 phosphorylation and decrease in activity caused by UCN-01. Further experiments will be required to fully elucidate kinase pathway activation/inhibition by UCN-01 after DNA alkylation damage. It is possible that inhibition of hMSH6 phosphorylation by UCN-01 may inhibit the activation of a different DNA damage-signaling pathway. This notion agrees with the fact that only one of the 22 identified phosphorylation sites on hMSH6 is an ATM/ATR recognition motif (Supplement Figure 2).
Several of the phosphorylation sites located in the N-terminus of hMSH6 contain CK2 recognition motifs. We therefore generated recombinant hMutSα with a mutant hMSH6 containing a cluster of four site-specific mutations located at S252A, S254A, S256A, and S261A, all of which have recently been identified by phospho-proteome studies as CK2 phosphorylation motifs (Supplementary Figure 2 and references [30–37]). The mutated heterodimer binds to both [32P]-G:T and [32P]-O6meG:T oligomers more poorly than the wildtype heterodimer (Figure 7A). Mutation of the four serines inhibits binding activity to [32P]-O6meG:T less than to [32P]-G:T. Similar to endogenous hMSH6 (Figure 4), recombinant wildtype hMutSα shifts approximately 2/3rd of the amount of [32P]-O6meG:T compared to its shift of [32P]-G:T, even at twice the recombinant protein concentration (20 nM for O6meG:T vs 10 nM for G:T). In addition, there is a decreased amount of phosphoserine within the recombinant mutant hMSH6 protein, as compared to an equal concentration of the recombinant WT hMSH6 (Figure 7B).
There remains much to learn about regulation of the activity of hMutSα in the cell. Technical difficulties have inhibited efforts to express exogenous hMSH6 in mammalian cells, therefore there are few reports describing phosphorylation of hMutSα and associated cellular activities. Increased phosphorylation, nuclear translocation, and mismatch and chromosomal binding activity is stimulated after cells are exposed to MNNG [13,14,39]. Increased G:T binding activity is associated with increased phosphorylation of hMutSα within nuclear extracts and, additionally, phosphorylation of recombinant hMutSα can be stimulated by PKC and CK2 in vitro . MMR protein expression, repair activity and sensitivity to 6-thioguanine are increased by TPA . Further, hMutSα phosphorylation by the PKCζ pathway protects against ubiquitin-dependent proteasomal degradation within human cell lines . As well, endogenous hMSH6 undergoes decreased phosphorylation, and MMR proteins are decreased in expression after cellular exposure to different kinase inhibitors having a wide range of activity [39,41].
We show here, by several different approaches, that hMSH6, but not hMSH2, phosphorylation occurs within both intact cells and nuclear extracts. Our results agree with data from several large studies that have collectively identified 22 distinct phosphorylated sites within the N-terminus of hMSH6 by phospho-proteome mass spectrometry. Eighteen of these phosphorylation sites are clustered into six distinct clusters of amino acids within the first 348 residues of the unstructured N-terminal region of hMSH6 (Supplementary Figure 1 and references [30–37]. The majority of phosphorylation sites contain CDK, MAPK, PKC or CK2 recognition motifs . Serine 348 is the only residue that has been reported as differentially phosphorylated (after gamma irradiation), and is also the only phosphorylation site that is an ATM/ATR recognition motif . CK2 has the highest number of potential recognition motifs within the N-terminus of hMSH6 (8 sites; Supplement Figure 2). CK2 is constitutively active in human cells and has been called the master regulator of cellular function, with over 300 substrates identified to date [53,54]. We have created a recombinant mutant hMSH6 containing four S→A alterations in the fifth cluster of phosphorylation sites (S252–S261) within the N-terminus. This cluster contains only CK2 recognition motifs. This alteration caused decreased binding of both G:T and O6meG:T-containing DNA suggesting a role for CK2-catalyzed phosphorylation of these sites in post-translational control of hMutSα activity. Much work remains to be accomplished with eighteen phosphorylation sites yet to be investigated for their roles in differential signaling of hMutSα. All of the above kinases are potential effectors of hMutSα activity, and they could operate in a number of different kinase signaling pathways. Taken together, these observations suggest a complicated set of cellular signaling pathways that control both quantitative and qualitative hMSH6 phosphorylation patterns, in turn affecting hMutSα activity.
The N-terminal region is a proteolytically-sensitive unstructured domain that tethers hMSH6 to PCNA [55,56]. Upon deletion of the PCNA recognition motif (PIP box), deletion-mutant hMutSα still exhibit in vitro MMR activity, indicating that PCNA may be used to tether hMutSα to the DNA during replication . The N-terminal portion of the yeast Msh6 confers sensitivity to MNNG; however this purified N-terminal region does not specifically recognize O6meG . Human MSH6 that lacks the N-terminal 340 amino acids has been co-crystallized with hMSH2 as the hMutSα heterodimer. This truncated heterodimer was also crystallized bound to several heteroduplex DNA structures (G:T, O6meG:T, G:U) without notable structural differences . Each of these three DNA structures elicits activation of different cellular pathways (MMR, DNA damage-signaling, and somatic immunoglobulin hypermutation, respectively). These diverse functions activated by DNA-bound hMutSα require different signaling pathways. Phosphorylation/dephosphorylation of binding proteins is a well known mechanism of cellular signaling. Differential phosphorylation at multiple sites within a protein is now recognized as an elegant and essential mechanism for modulation of activity, structure, and interaction of cellular substrates within a large variety of metabolic pathways [59–63].
In this report, we have investigated how phosphorylation of hMSH6 is regulated by binding of hMutSα to G:T and O6meG:T lesions that activate MMR or DNA damage-signaling, respectively. The degree of phosphorylation of hMSH6 and DNA binding activity of hMutSα is different in the presence of G:T than in the presence of O6meG:T. A G:T mismatch elicits increased phosphorylation of hMSH6, whereas O6meG:T does not. More G:T than O6meG:T radioactive oligomer is shifted after incubation with equal amounts of nuclear extract, in agreement with other reports [44,45]. However, comparison of hMSH6 protein bound to each shifted oligomer indicates more protein bound to O6meG:T than to the G:T oligomer. These oligomers are identical, except for either G or O6meG opposite T, and relatively short at 69 bp, therefore the observed effects may be minimized in our system. These results may indicate that both the degree of hMSH6 phosphorylation and differential binding activity of hMutSα may be required to trigger specific cellular pathway responses.
We also probed the relationship between changes in phosphorylation of hMSH6 and changes in the binding of hMutSα to G:T and O6meG:T. TPA is a phorbol ester that activates all PKC isoenzymes. The majority of PKC isoenzymes are located at the plasma membrane or within the cytosol . PKC phosphorylates hMutSα and stimulates MMR functions . The atypical PKC ζ isoenzyme phosphorylates hMutSα and inhibits degradation of this heterodimeric protein . Incubation of HeLa MR cells with TPA increases hMSH6 phosphorylation in the presence of an oligomer containing a G:T mismatch, concomitant with a small increase in binding of hMutSα to G:T. However, TPA does not stimulate binding activity to O6meG:T, nor is hMSH6 phosphorylated to an increased extent by TPA in the presence of O6meG:T. Based on these results, one or more of the PKC isoenzymes may operate within a cellular signaling network that increases phosphorylation of hMSH6 and activates hMutSα in the presence of a DNA mismatch but not in the presence of DNA alkylation damage. These results point toward differential kinase regulation of the mismatch repair system that is dependent on the type of lesion bound by hMutSα.
Conversely, the staurosporine analogue UCN-01 inhibits hMSH6 phosphorylation and hMutSα binding to both G:T and O6meG:T. This effect is not due to decreased amounts of hMSH6 in nuclear extracts of these cells. UCN-01 is a broad range kinase inhibitor that has been reported, by in vitro methods, to be a potent inhibitor of a number of different cell cycle-dependent kinases, including Chk1, at the concentration used in this study [49,65]. Treatment with 200 nM UCN-01 alone is nontoxic to HeLa MR cells. However when UCN-01 is used in combination with exposure to a low level of MNNG, the MNNG-induced second cell-cycle arrest does not occur and the cell population dies after undergoing mitosis. Phosphorylation of Chk1 is not inhibited by UCN-01 after MNNG exposure, demonstrating as previously reported that UCN-01 does not inhibit upstream ATR activity [50,66]. Phosphorylation of Cdk1 at tyrosine 15 is inhibited by UCN-01 after MNNG treatment, therefore allowing the cell to escape G2 arrest after alkylation damage, despite that Cdc25c is still phosphorylated at S216, rendering this cell cycle kinase inactive . Kinases other than activated Chk1 phosphorylate Cdc25c at S216, such as Chk2, C-TAK1, and Plk3, therefore inhibition of Chk1 activity cannot be ruled out in these studies . Chk2 and C-TAK1 have been reported to be sensitive to this level of UCN-01 within cells, it is not known if Plk3 activity is affected by UCN-01 within the cell [48,49]. Overall, this suggests that another DNA damage-signaling pathway, other than ATR-Chk1, may be induced within the cell by activation of hMutSα in the presence of alkylation damage. Further investigations within this area are needed. Several clinical trials with UCN-01 in combination with chemotherapeutic DNA damaging agents are currently ongoing. These trials are based on the assumption that UCN-01 inhibits ATR-Chk1 check point signaling pathway, therefore driving the damaged tumor cell to continue cycling and subsequent cell death.
Based on our current results and all other evidence to date, we propose a mechanistic model for differential activation of hMutSα to trigger either the DNA mismatch repair pathway or the DNA damage-signaling pathway (Figure 8). According to this model, hMutSα recognizes both undamaged mismatched bases (G:T) and specifically damaged mismatches (O6meG:T). If a mismatch occurs during replication, hMutSα binds and triggers the mismatch repair pathway and the mismatch is rapidly and correctly repaired. This pathway requires less bound hMutSα containing more highly phosphorylated hMSH6. Cells exposed to monofunctional alkylators contain O6methylG that is frequently replicated incorrectly with insertion of thymidine opposite the damaged guanine, thereby creating O6meG:T. This mismatched lesion is bound by more hMutSα molecules, and does not trigger hMSH6 phosphorylation above constitutive levels. These two effects initiate the MMR-induced DNA damage-signaling pathway. Differentially phosphorylated hMSH6 within the hMutsα population may bind to different lesions or differential phosphorylation of hMSH6 may take place after binding to specific lesions. We hypothesize that the cell controls signaling for MMR versus DNA damage response by hMSH6 phosphorylation (higher for G:T versus O6meG:T) and binding stoichiometry of hMutSα to different DNA constructs (lower for G:T versus O6meG:T), providing an elegant mechanism for the cell to differentiate between mismatch repair and the DNA damage-signaling pathway.
A. For in vivo phosphorylation, HeLa and NIH 3T3 cells were pre-incubated in phosphate-free media for 5 h and subsequently labeled with [32P]-H3PO4 for 15 h. Whole cell extracts from each cell line were prepared, MSH6 was immunoprecipitated, separated on SDS-PAGE, and transferred to PVDF membrane for autoradiography. B. The PVDF membrane was then immunoblotted for both MSH6 and MSH2 proteins. C. For in vitro phosphorylation, HeLa nuclear extracts were incubated with [γ-32P]-ATP and either immediately immunoprecipitated using antibody to hMSH6 (lane 2) or dephosphorylated by λPPase (lane 3) before immunoprecipitation. Proteins were separated on SDS-PAGE and the dried gel was subjected to autoradiography.
Twenty-two phosphorylated sites have been identified within hMSH6 [30–37]. Twenty of these sites are clustered into six groups of 2–5 amino acids within the first 348 residues of the unstructured N-terminal region of hMSH6. Phosphorylation sites are updated at: http://www.uniprot.org/uniprot/P52701. Software for prediction of kinase recognition motifs at each phosphorylation site can be found at: http://networkin.info/search.php .
A. Commassie Blue stained SDS-PAGE of S column fractions. Lanes 3–6 were pooled, dialyzed in storage buffer and frozen at −80°C after determination of protein concentration. B. hMSH6 immunoblot of increasing concentrations of purified recombinant hMutSα from above pooled fractions. C. DNA binding activity was determined by electrophoretic mobility shift (EMSA) after incubation of different concentrations (10–20 nM) of recombinant hMutSα with 32P-G:T oligomer (plus 100X unlabeled G:C oligomer). Right lane of the autoradiograph demonstrates a super-shifted radioactive band after incubation with antibody to hMSH6. These experiments have been repeated several times using the same purified fraction (manuscript in progress). D. Alteration of recombinant hMSH6 phosphorylation activity. 200 µg of total cell lysate from Hec59 cells was incubated with 1.2 pmol (300 ng) recombinant hMutSα and 50 µCi of [γ-32P]-ATP, with or without 200 ng non-radioactive G:C or G:T oligomer for 30 min at 30°C. The concentration of recombinant hMutSα is equivalent to that measured previously within the human nucleus . Recombinant hMSH6 was then immunoprecipitated, subjected to SDS PAGE, and subsequent autoradiography of the dried gel. The bar graph represents phosphorimager quantification of each phosphorylated hMSH6 protein band.
A. Electrophoretic mobility shifts (EMSA) were performed using 2 µg of HeLa nuclear extract and 5 ng of 32P-G:T oligomer (plus 100X unlabeled G:C). The nondenaturing polyacrylamide gel was dried and exposed to film. B. EMSAs were performed after incubating 10 µg HeLa nuclear extracts with 5 ng of G:T oligomer (plus 20X G:C) and increasing concentrations of ATP, as indicated. Proteins from nondenaturing polyacrylamide gels were transferred to PVDF and blotted with antibody to hMSH6 (upper) or hMSH2 (lower).
National Institutes of Health Grants CA84412 (K.J.W), CA106575 (K.J.W) and American Cancer Society RSG-06-163-01 GMC (S.M.P.) supported this work. We thank Allen Schroering for accomplishing the experiments in Supplement Figure 4 and Tim Mueser for helpful discussions and proof reading the manuscript.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.