Human DNA-(Cytosine-5) Methyltransferase-PCNA Complex as a Target for p21WAF1

See allHide authors and affiliations

Science  26 Sep 1997:
Vol. 277, Issue 5334, pp. 1996-2000
DOI: 10.1126/science.277.5334.1996


DNA–(cytosine-5) methyltransferase (MCMT) methylates newly replicated mammalian DNA, but the factors regulating this activity are unknown. Here, MCMT is shown to bind proliferating cell nuclear antigen (PCNA), an auxiliary factor for DNA replication and repair. Binding of PCNA requires amino acids 163 to 174 of MCMT, occurs in intact cells at foci of newly replicated DNA, and does not alter MCMT activity. A peptide derived from the cell cycle regulator p21WAF1 can disrupt the MCMT-PCNA interaction, which suggests that p21WAF1 may regulate methylation by blocking access of MCMT to PCNA. MCMT and p21WAF1 may be linked in a regulatory pathway, because the extents of their expression are inversely related in both SV40-transformed and nontransformed cells.

DNA methylation in mammals is involved in imprinting (1), regulation of transcription (2), and development (3). Various diseases, including cancer (4) and fragile X syndrome (5), are associated with abnormal DNA methylation, which indicates that one or more regulatory mechanisms must exist to ensure the maintenance of precise methylation patterns by MCMT in the mammalian genome.

To investigate whether PCNA, an auxiliary factor for DNA replication and repair, was involved in the regulation of MCMT activity, we first determined whether the two proteins interact in vitro. We incubated human acute lymphoblastic leukemia (CEM) cell extracts with immobilized glutathione-S-transferase (GST) fusion proteins containing fragments of MCMT (Fig. 1, A and B) and analyzed the bound cellular proteins on immunoblots. Results with both cellular and recombinant PCNA (rPCNA, Fig. 1C) indicated that MCMT binds to PCNA directly through amino acids 122 to 322. We refer to this region as hMPBD (human methylase-PCNA binding domain). Further deletion analysis (Fig. 2, A to C) revealed that hMPBD requires only the sequence TRQTTITSHFAKG (6). Comparative studies on vertebrate MPBDs, as well as point-mutation analyses (Fig. 2, D to F), indicated that Arg163, Gln164, Thr166, Ile167, His170, and Phe171 are critical for binding to PCNA.

Figure 1

Binding of PCNA to human recombinant MCMTs. (A) Map of the MCMT fragments used to construct GST fusion proteins. (KG)5, hinge region; PC, active site; fragments (F) are labeled by the start and end codons. (B) SDS–polyacrylamide gel electrophoresis (SDS-PAGE) of fusion proteins detected by Coomassie staining (23). Lanes labeled “crude” contain bacterial lysates; As %, ammonium sulfate precipitates; GSH-Sp, GSH-Sepharose–purified fusion proteins; and MonoQ, the 0.1 M NaCl fraction from MonoQ (HR5/5) chromatography. AcA22, G200, and AcA34 are the gel filtration fractions. (C) Immunoblot of bound PCNA (24,25). CEM cell extracts (200 μg) or rPCNA (100 ng) were incubated with 100 ng of GST-(F122–1616), GST-(F323–1616), or GST-(F122–418). The GST fusion proteins were recovered by GSH-Sepharose and the bound PCNA was detected by PCNA antibodies (PC10, 0.5 μg/ml). Ctr, control input PCNA.

Figure 2

Identification of hMPBD. (A) Map of deletion constructs used to produce GST fusion proteins. F6 (codons 206–418), F5 (122–205), F4 (149–205), F3 (174–205), F2 (162–174), and F1 (164–171) are fragments of F122–418 (6, 23). (B) SDS-PAGE of fusion proteins detected by Coomassie staining (23). (C) Immunoblot of bound PCNA (24, 25). rPCNA (100 ng) was incubated with the GSH-Sepharose–immobilized proteins in (B). (D) SDS-PAGE of vertebrate GST-MPBD derivatives by Coomassie staining [wt h, wt m, and wt c are wild-type human, murine, and chicken MPBDs, respectively; cV190H is a Val190 → His substitution at codon 190 of cMPBD, whereas others are human MPBD mutants (6,23)]. (E) Immunoblot of PCNA bound to the samples in (D). (F) Schematic showing the PCNA-binding properties of human MPDB derivatives from (E). Residues that are critical for PCNA binding (except K173) and conserved in the m (mouse) and c (chicken) MPBDs are in bold (6).

To test whether PCNA binding affected the nuclear localization of MCMT, we transfected SV40-transformed MRC5 cells (MRC5SV) with an expression vector encoding GST fused to codons 122 to 207 of MCMT (which includes both MPBD and the nuclear localization signal) and treated the cells with bromodeoxyuridine (BrdU) to label replicated DNA. In transfected cells costained with a polyclonal antibody to GST (anti-GST) and a monoclonal antibody (mAb) to BrdU, there was precise colocalization of the GST fusion protein with the newly incorporated BrdU in small nuclear speckles (Fig. 3, A, E, and I). By contrast, these fine speckles, which represent early replication foci, were not stained by anti-GST in cells transfected with the H170V null PCNA-binding mutant (Fig. 3, B, F, and J). Similar results were obtained with anti-GST and PCNA mAb (Fig. 3). The binding of PCNA does not alter the ability of MCMT to methylate poly(deoxyinosine-deoxycytidine) [poly(dI-dC)] substrates (Fig.4A). Thus, PCNA binding to MCMT may recruit this DNA modification enzyme to methylate newly replicated DNA. Another replication foci–targeting domain has been mapped to codons 325 to 573 of murine MCMT; however, this sequence targeted the protein to “larger” or late replication foci (7).

Figure 3

Colocalization of MPBD and PCNA with newly replicated DNA. MRC5SV cells were transfected with the GST fusion constructs of wild-type (WT) and H170V mutant inserts in pXJ41neo vector for 24 hours (26) and labeled with BrdU for 20 min. After fixation, cells were stained sequentially with anti-GST (10 μg/ml, for the expressed GST fusion proteins) and BrdU mAb (5 μg/ml) or anti-GST and PCNA mAb (PC10, 10 μg/ml) (24, 26). Left panels, colocalization of MPBD and BrdU (which represents newly replicated DNA); right panels, colocalization of MPBD and PCNA. Nuclear DNA is stained with 4,6-diamidino-2-phenylindole (DAPI). NLS, nuclear localization signal.

Figure 4

Relation among MCMT, p21WAF1, and PCNA. (A) Effect of PCNA on MCMT methylase activity. GST-(F122–1616) (7 μg) was incubated with rPCNA (0.5, 2, or 10 μg), poly(dI-dC) (2 μg), and [3H]SAM (2 μM, 71 Ci/mmol) in assay buffer (27). The graph summarizes the average of two experiments and shows the time course of [3H]CH3 group incorporation into poly(dI-dC). (B) Comparative effect of MPBD and WPBD on binding of GST-MPBD to rPCNA (25). Top panel is the sequence of the peptide used (6). Peptide 2 is the null PCNA binding mutant of wtWPBD (8). In the competition assays, peptides were added to the rPCNA before binding by GST-MPBD proteins (direct) or added to the immobilized GST-MPBD and rPCNA complexes (preformed). The PCNA associated with GST-MPBD was analyzed by immunoblot. C indicates input rPCNA. (C) Immunoblot showing the amount of MCMT and p21WAF1 present in the lysates [200 μg (24)] of SV40-transformed (MRC5SV and WISV) and nontransformed (MRC5 and WI38) cells. Blots were developed with mAb D12 (1 μg/ml) for MCMT and mAb Cip1 (1 μg/ml) for p21WAF1. (D) Immunoblot of PCNA coprecipitated with MCMT from the cell lysates in (C) by anti-MCMT. To balance the small amounts of MCMT present in the nontransformed cell lysates [see (C)], 10-fold excesses of their immunoprecipitates were used for comparison with the SV40-transformed counterparts (24). Top, MCMT detected by mAb D12; bottom, coprecipitated PCNA (by PC10). (E) [3H]5meC in the genomic DNA from NMU-treated and untreated (CTR) cells labeled with [3H]SAM (28). Top left, a chromatogram of standard 2′-deoxynucleosides resolved by reversed-phase HPLC; bottom left, a control experiment where [3H]5meC is detected by scintillation counting of the HPLC fractions from nucleoside analysis of poly(dI-dC) treated with GST-(F122–1616) and [3H]SAM. The bar graph shows the relative amount of [3H]5meC found in the genomic DNAs from NMU-treated and untreated MRC5 and MRC5SV cells, which were labeled with [3H]SAM for 6 or 16 hours.

The hMPBD motif resembles a sequence in the tumor suppressor p21WAF1. This sequence (KRRQTSMTDFYHSKRRLIFS, corresponding to codons 141 to 160 of p21WAF1) binds tightly to PCNA and inhibits the in vitro replication of SV40 DNA (8). We therefore compared the ability of the synthetic peptides corresponding to wtWPBD (wild-type p21WAF1-PCNA binding domain), wtMPBD, and a chimeric MPBD-WPBD (Fig. 4B) to disrupt the MPBD-PCNA interaction. Less rPCNA bound to immobilized GST-MPBD after pretreatment of rPCNA with wtWPBD relative to pretreatment with wtMPBD (Fig. 4B). Similar results were observed when the wtWPBD and wtMPBD peptides were added to preformed GST-MPBD and PCNA complex (Fig. 4B). Because the chimeric peptide failed to compete, this result suggests that residues within the NH2- and COOH-termini of WPBD and MPBD are noninterchangeable and may contain unique PCNA-binding determinants.

What is the relation among MCMT, PCNA, and p21WAF1 in intact cells? Surprisingly, analysis of asynchronous SV40-transformed and nontransformed cells by immunoblot (Fig. 4C) revealed that the extents of expression of MCMT and p21WAF1 were inversely related. The transformed cells had large amounts of MCMT but small amounts of p21WAF1, whereas the nontransformed cells exhibited the reverse pattern. However, in experiments where similar amounts of MCMT were immunoprecipitated for comparison (by using an excess of nontransformed cell extracts), PCNA coprecipitated with MCMT in the SV40-transformed but not the nontransformed cells (Fig. 4D). These results agree with the in vitro observation that WPBD can disrupt the MPBD-PCNA complex and suggest that p21WAF1 may regulate the formation of the MCMT-PCNA complex in vivo. In addition, during the G1-S transition, there must be one or more mechanisms that facilitate MCMT-PCNA interaction in the nontransformed cells.

Because MCMT preferentially methylates certain types of damaged DNA (9), the enzyme must be prevented from contacting damaged DNA sites in order to avoid unscheduled hypermethylation of the genome. In addition to its inhibitory effect on the cyclin-dependent kinases (CDKs), p21WAF1 also plays a role in DNA repair. Upon induction by p53 as a result of DNA damage, p21WAF1colocalizes with PCNA at DNA repair sites (10). Because the MCMT-PCNA complex is fully active in methylation (Fig. 4A), we explored the possibility that p21WAF1 prevents methylation by blocking the access of MCMT to PCNA at the repair sites. To test this, we compared the concentrations of [3H]5meC (5-methyl-2′-deoxycytidine) formed in the genomic DNA of MRC5 and MRC5SV cells (which express extreme amounts of p21WAF1) that had been treated with N-methylnitrosourea [NMU, which produces 06-methylguanine residues (6MeG) in the DNA] and then labeled withS-adenosyl-l-[methyl-3H]methionine ([3H]SAM). 06-Methylguanine–DNA methyltransferase, which repairs 6MeG, was inactivated by 06-benzylguanine treatment before NMU addition (11). Consistently, the incorporated [3H]5meC concentrations were found to be higher in the NMU-treated than in the untreated MRC5SV cells, but the opposite was observed for MRC5 cells (Fig. 4E). Thus, during DNA damage, the p21WAF1 in the nontransformed cells appeared to “delay” methylation, whereas the stable MCMT-PCNA complex in the transformed cells remained “active” in methylation. The extent of MCMT involvement in this induced hypermethylation will remain unresolved until the putative de novo methylase is identified (12).

These results may have implications for DNA methylation in vivo, cellular transformation by MCMT, DNA methylation–associated genomic instability (4), and DNA repair. In mammalian cells, newly replicated DNA is rapidly packaged into nucleosomes to which histone H1 is subsequently added (13). Because H1 inhibits DNA methylation, replicated DNA must be methylated before H1 is incorporated into the nucleosomes (13). This limited time window requires a mechanism to ensure proper coordination of DNA methylation after DNA replication. PCNA is structurally similar to the β subunit of the Escherichia coli DNA polymerase complex, which can remain bound to the replicated DNA during its switch from polymerase III to the loader-unloader γ subunit (14). If PCNA behaves similarly to the β subunit in maintaining its association with the replicated DNA, then these PCNA molecules are potential sites for the loading of MCMT onto the replicated DNA. To accommodate the size of this replicated DNA or to prevent its interference with replication, MCMT may use a second replication foci–targeting domain B1 (downstream of MPBD, residues 325 to 573) (7) that can bind to replicated DNA of a particular size (15). With MPBD and B1 contacting the replicated DNA, the large domain between this region and the active site (the PC dipeptide) could “scan” for the hemimethylated sites for methylation.

Our data suggest that p21WAF1 and MCMT are potential antagonists. The quaternary p21WAF1-PCNA–cyclin D1–CDK4 complex, which regulates G1-S transition of the cell cycle, is stable in normal cell extracts but not in some virus-transformed and tumor cells (16). The cell-transforming activity of overexpressed MCMT and the high MCMT activities often present in tumor cells (17) suggest that MCMT may exert its oncogenic effects by competing with cellular p21WAF1 for PCNA and may thereby perturb the stability of the quaternary complex. The released cyclin D1–CDK4 complex would phosphorylate and inactivate the retinoblastoma protein (Rb), a G1 restriction point regulator, to stimulate cell proliferation (16). This might explain why the two SV40-transformed cells used in this study grow faster than their nontransformed counterparts (18).

The MCMT-PCNA interaction adds complexity to our understanding of DNA repair in vivo, because both nucleotide excision repair and mismatch repair require PCNA (19). However, by its presence at the DNA repair sites, p21WAF1 may sequester the damaged DNA for repair and prevent hypermethylation, because it may inhibit MCMT from contacting PCNA while allowing DNA repair to continue (20). If this model is correct, the genomic instability observed in p53-deficient cells that cannot transactivate p21WAF1 after DNA damage (21) may, in part, be attributable to aberrant methylation patterns in the genome, a characteristic of tumor cells (4).

  • * Present address: Department of Genetics and Development, Columbia University, Hammer 1124, 701 West 168th Street, New York, NY 10032, USA.

  • To whom correspondence should be addressed. E-mail: mcblib{at}


Stay Connected to Science

Navigate This Article