2019 年 94 巻 3 号 p. 123-132
Cellular aging is characterized by the loss of DNA replication capability and is mainly brought about by various changes in chromatin structure. Here, we examined changes in MCM2–7 proteins, which act as a replicative DNA helicase, during aging of human WI38 fibroblasts at the single-cell level. We used nuclear accumulation of p21 as a marker of senescent cells, and examined changes in MCM2–7 by western blot analysis. First, we found that senescent cells are enriched for cells with a DNA content higher than 4N. Second, the levels of MCM2, MCM3, MCM4 and MCM6 proteins decreased in senescent cells. Third, cytoplasmic localization of MCM2 and MCM7 was observed in senescent cells, from an analysis of MCM2–7 except for MCM5. Consistent with this finding, fragmented MCM2 was predominant in these cells. These age-dependent changes in MCM2–7, a protein complex that directly affects cellular DNA replication, may play a critical role in cellular senescence.
Cellular aging is attributable to various genotoxic stimuli such as telomere shortening, DNA damage, reactive oxygen species and excess growth stimulation by oncogenes (Kuilman et al., 2010), and it is associated with changes in chromatin structure (Ishimi et al., 1987; O’Sullivan and Kariseder, 2012) and in epigenetic regulation (Sidler et al., 2017). Cellular aging plays a role in suppressing immortalization of the aged cells but it also functions in the generation of cancer cells by stimulating the growth of cells surrounding the aged cells through secretion of growth factors. The DNA damage checkpoint system that is activated by shortened telomeres and double-stranded DNA breakage plays a role in preventing cell cycle progression into S phase (d’Adda di Fagagna et al., 2003). This system includes activation of checkpoint kinases such as ATM and thereby activation of transcription factor p53. p21, which is transcribed by p53, binds to CDK to inhibit its activity to phosphorylate Rb; the reaction is required for progression of the cell cycle into S phase. Recently, it has been shown that senescent cells have a DNA content of 4N (Johmura et al., 2014). These data indicate that cells arrested at G2 phase by various stimuli to induce cellular aging bypass M phase to enter G1 phase and then become senescent.
Initiation of DNA replication occurs at DNA replication origins, where the MCM2–7 complex, a replicative DNA helicase, is assembled in the presence of ORC, CDC6 and CDT1 (Masai et al., 2010). Other replication proteins including DNA polymerase ε assemble at the origins in the presence of CDK (Araki, 2011). Several lines of evidence suggest that the CMG complex, which is MCM2–7 complexed with CDC45 and the GINS complex, functions as a replicative DNA helicase (Labib and Gambus, 2007; Ishimi, 2018). Recent reports indicate that the levels of MCM2–7 proteins are down-regulated in senescent cells (Dumit et al., 2014; Flach et al., 2014) and also in quiescent cells (Namdar and Kearsey, 2006). Limited levels of MCM2–7 proteins on the genome may result in the arrest of cell cycle progression at G1 phase and also at S phase. Prolonged arrest of DNA replication forks generates double-stranded DNA breakages at the fork. Thus, down-regulation of MCM2–7 proteins plays a causal role in cellular aging. It has been reported that chronic DNA replication stress reduces the replicative lifespan of mouse cells by p53-dependent microRNA-assisted MCM2–7 down-regulation (Bai et al., 2016). It has been reported that MCM2 is cleaved to produce a 55-kDa protein in senescent keratinocytes (Harada et al., 2008). MCM2–7 proteins are largely localized in nuclei to bind to the replication origins in mammalian cells. Recently, cytoplasmic localization of MCM7 has been reported in aged and systemic sclerosis human dermal fibroblasts (Dumit et al., 2014). Since the MCM2/4/6/7 tetramer may enter the nucleus using nuclear localization signals in the amino-terminal region of MCM2 (Ishimi et al., 1998), the two reactions of cleavage of MCM2 and cytoplasmic localization of MCM7 may be linked.
Here, we examined changes in MCM2–7 proteins during aging of human diploid cells at the single-cell level and by western blot analysis. First, we found that senescent cells contained lower levels of MCM2, MCM3, MCM4 and MCM6 proteins and had higher DNA content. Second, cytoplasmic localization of MCM2 was detected in senescent cells. Biochemical analysis suggested that fragmented MCM2 was predominant in the cells.
WI38 cells were cultured in DMEM containing 10% fetal calf serum and antibiotics. Changes in population doubling level (PDL) of the cells were determined by counting cells at inoculation and at harvest. Under these conditions, the cells became senescent at 40–48 PDL; that is, the final PDL number varies in different culture series. The cells at stages before senescence were used as growing cells.
WI38 cells on a coverslip were cultured in the presence of 20 μM bromodeoxyuridine (BrdU) for 20 min as described before (Tsuji et al., 2018). Anti-p21 mouse antibody (BD Biosciences, Pharmingen, 556430) and anti-BrdU rat antibody (AbD Serotec) were successively used. Fluorescence intensity in the cells was measured using software (area quantification) associated with a fluorescence microscope (BZ9000, KEYENCE). Typically, 200–300 cells on a coverslip were marked and their fluorescence levels were measured by the software. Data in Excel (Microsoft) were used for making graphs. For double-staining of the cells, they were typically incubated first with anti-p21 mouse antibody and then with anti-mouse Cy3-conjugated second antibody, followed by incubation with anti-MCM rabbit antibody and then with anti-rabbit FITC-conjugated second antibody. Detection of senescent cells by staining for β-galactosidase activity was carried out using a SPiDER-β Gal detection kit (Dojin Chemical). Contrast of original pictures was enhanced using Photoshop software (Adobe).
WI38 cells were lysed in modified CSK buffer and fractionated into Triton-soluble S and -insoluble P fractions as described (Tsuji et al., 2018). Proteins were separated by SDS-polyacrylamide gel electrophoresis. Protein concentration in the S fraction was measured, and the amounts loaded onto the gel were adjusted to be equivalent among different samples. Western blotting was performed as described (Tsuji et al., 2018). For detection of p21, the filter after transfer of proteins was incubated for 2 h at 37 ℃ with 10% skim milk in PBS. After washing with PBS containing 0.05% Tween-20, the filter was incubated with anti-p21 mouse antibody (1 μg/ml in 1% skim milk in PBS) for 2 h at 37 ℃ and then with anti-mouse antibody conjugated with horseradish peroxidase for 2 h at 37 ℃. Anti-MCM2 antibody was prepared by immunizing a rabbit with full-length human MCM2. The antibody was fractionated into populations recognizing either the amino-terminal half or the carboxyl-terminal half of MCM2, by affinity purification with antigen beads conjugated with the corresponding regions (amino acid residues 1–448 and 449–904) of mouse MCM2. Affinity-purified anti-MCM3 and anti-MCM4 rabbit antibodies were prepared using full-length proteins as antigens (Kimura et al., 1994, 1995). Anti-MCM6 rabbit antibody was prepared using the full-length protein as an antigen, and anti-MCM7 mouse monoclonal antibody (Santa Cruz Biotechnology, sc-9966) and anti-tubulin mouse antibody (Sigma, T5168) were purchased.
To detect senescent cells in growing WI38 cells (36 PDL) on a coverslip, the cells, pulse-labeled with bromodeoxyuridine (BrdU), were incubated with anti-p21 antibody and then with anti-BrdU antibody (Supplementary Fig. S1A). It is known that p21 accumulated in the nucleus is one of the markers of senescent cells (Kuilman et al., 2010; Johmura et al., 2014). Nuclear staining with anti-p21 antibody was detected in some of the cells. Fluorescence from the second antibodies bound to the anti-p21 and anti-BrdU antibodies was measured and plotted in two dimensions in 218 cells (Supplementary Fig. S1B). All the cells in S phase contain low levels of p21. These results are consistent with the notion that the presence of p21 is inhibitory for the cells to enter S phase, and that nuclear staining with anti-p21 antibody is a good marker of senescent cells. It has been reported recently that cells arrested at G2 phase of the cell cycle bypass M phase and enter G1 phase to become senescent (Johmura et al., 2014). Therefore, the senescent cells arrested in G1 phase may have a DNA content of 4N. In growing WI38 cells (41 PDL), fluorescence from p21 antibody staining and DAPI staining was measured (Fig. 1A). Data in Fig. 1B show the distribution of DNA content of the cells. They show that cells with a DNA content of 2N are dominant and those with 4N are also detected. Only a small proportion of the cells display an intermediate DNA content, which represents cells in S phase. In addition, a very small proportion of the cells have a DNA content higher than 4N. The result in Fig. 1B shows that cells with a DNA content of 8N and 16N contain higher levels of p21. Next, more than 2,000 cells (40–48 PDL) having a DNA content of 2N, 4N and 8N were examined for abundance of p21 (Fig. 1C and 1D). In these experiments, WI38 cells did not proliferate and became senescent at 48 PDL. Cells with a DNA content of 4N showed 4-fold higher abundance of nuclear p21 than cells containing 2N DNA, and those containing 8N DNA showed 11-fold higher abundance of p21. These results suggest that cells with a DNA content higher than 4N tend to become senescent, and are consistent with published data (Johmura et al., 2014). The present data also raise the possibility that a small number of cells with a DNA content of 4N progress through a further S and G2 phase to attain a DNA content of 8N and then become senescent.
DNA content of p21-positive cells. (A) Growing WI38 cells at 41 PDL were stained with anti-p21 antibody and DAPI. Bars, 40 μm. (B) Fluorescence from nuclear p21 and DAPI in the cells in (A) was measured and plotted in two dimensions. Cells numbered 4, 3, 2 and 1 have a DNA content of 2N, 4N, 8N and 16N, respectively. Numbers on the axes are arbitrary units. (C) In randomly selected cells, fluorescence from DAPI was measured and the cells were fractionated by fluorescence level. (D) The levels of fluorescence from second antibody bound to the anti-p21 antibody were averaged for cells with a DNA content of 2N, 4N and 8N. The total number of cells examined was 2,231, and included 1,636 2N, 408 4N and 45 8N cells. Error bars indicate standard errors. A significant difference as determined by the Tukey method is indicated by ** (P < 0.01).
Decreased levels of MCM2–7 expression are detected in aging hematopoietic stem cells in mouse (Flach et al., 2014) and in cultured aged human dermal fibroblasts (Dumit et al., 2014). We examined changes in levels of MCM proteins during cellular aging of WI38 cells (36–40 PDL) by western blotting (Fig. 2). Under this condition, cell proliferation terminated at 40 PDL. WI38 cells with different PDL were fractionated into a Triton-soluble fraction containing proteins in cytoplasm and nucleoplasm, and a Triton-insoluble fraction containing chromatin-bound proteins. The presence of MCM2, MCM4 and MCM6 proteins in these fractions was examined using specific antibodies. The levels of these three MCM proteins drastically decreased at near-senescence. The results were basically reproduced except for MCM2 as shown in Supplementary Fig. S2A, which also reveals that the level of p21 unexpectedly decreased near senescence. Similar changes in the levels of MCM2–7 proteins during cellular aging were also observed in different culture series (Supplementary Fig. S2B, S3 and S4). Data in Supplementary Fig. S4 show that the levels of MCM3 and MCM4 drastically decrease near senescence, and those of ORC2 and ORC3, components of the origin recognition complex, also decrease. Proteins (66–68 kDa) smaller than full-length MCM2 were detected by an anti-MCM2 antibody that recognizes the carboxyl-terminal half region of MCM2 (Fig. 2). Although the reason why p21 level decreases at near-senescence in Supplementary Fig. S2A and S2B remains to be clarified, immunostaining with anti-p21 antibody of the cells from 37, 39 and 40 PDL appears to support the conclusion of the western blotting experiment regarding the number of p21-positive cells and the intensity of their fluorescence (Supplementary Fig. S5).
MCM2–7 protein levels during cellular aging. WI38 cells from the indicated PDL (36–40) were fractionated into Triton-soluble S and -insoluble P fractions. Proteins in these fractions were electrophoresed on an SDS-polyacrylamide gel. MCM2, MCM4, MCM6 and tubulin were detected by western blotting. MCM2 was detected using the antibody that recognizes the carboxyl-terminal half region of MCM2.
We also compared the expression of MCM6 protein between senescent and non-senescent cells by double staining for β-Gal activity and with anti-MCM6 protein antibody in WI38 cells that are almost at the stage of senescence (41 PDL). The data in Supplementary Fig. S6 suggest that senescent cells retain lower levels of nuclear MCM6 protein than non-senescent cells that may undergo cell cycle progression.
WI38 cells at 41 PDL from another culture series were stained with three antibodies against MCM2 (Fig. 3A and 3B). Almost exclusively nuclear staining was observed when antibodies that recognize the entire region of MCM2 and that recognize the amino-terminal half region (N) of MCM2 were used (Fig. 3A). However, cytoplasmic staining was detected in addition to nuclear staining in some of the cells when an antibody that recognizes the carboxyl-terminal half region (C) of MCM2 was used. Data in Fig. 3B suggest that the cells with cytoplasmic staining of MCM2 exhibit nuclear localization of p21 at high frequency. Thus, cytoplasmic localization of MCM2 may be characteristic of senescent cells. Western blotting data indicate that all three antibodies recognize proteins smaller than full-length MCM2; in particular, the antibody recognizing the carboxyl-terminal half region of MCM2 efficiently recognized the smaller proteins (Supplementary Fig. S7). Nuclear localization signals are present in the amino-terminal region of MCM2. Cytoplasmic MCM2 detected by the antibody that recognizes the carboxyl-terminal half region of MCM2 may be the source of the smaller MCM2 fragments of 66 and 68 kDa in the S fraction detected with the MCM2 antibody (Fig. 2 and Supplementary Fig. S7).
Detection of MCM2 in cytoplasm. WI38 cells at 41 PDL were co-stained with anti-p21 antibody and with one of the three MCM2 antibodies that recognize the amino-terminal half region of MCM2 (N), full-length MCM2 (full) or the carboxyl-terminal half region of MCM2 (C). Staining with the anti-MCM2 antibodies alone is shown in (A). Enlarged views of co-staining with p21 and DAPI, and merged views, are shown in (B). Arrows indicate cells showing cytoplasmic localization of MCM2.
To confirm the findings described above, WI38 cells at 43 PDL were stained with anti-p21 antibody and the anti-MCM2 antibody that recognizes the carboxyl-terminal half region (Fig. 4A). The levels of fluorescence in the nucleus from anti-p21 antibody and from DAPI were quantified and plotted in two dimensions (Fig. 4B). The data show that cells with a DNA content of 4N–8N contain higher levels of p21 than 2N cells, although the differences in the level between 2N and 4N cells and between 2N and 8N cells are smaller than those in Fig. 1D. Numbered red dots in Fig. 4B correspond to cells showing distinct cytoplasmic localization of MCM2 in Fig. 4A. They contain relatively high levels of p21 and have a DNA content of 4N–8N. These results suggest a linkage between cytoplasmic localization of MCM2 and senescence.
Relationship between cells showing cytoplasmic localization of MCM2 and senescence. (A) Growing WI38 cells at 43 PDL were stained with p21 antibody and also with an antibody that recognizes the carboxyl-terminal half region of MCM2. The four numbered cells show distinct cytoplasmic localization of MCM2. (B) The levels of fluorescence in the nucleus from the p21 antibody and from DAPI were quantified in about 100 cells and plotted in two dimensions. Numbered red dots correspond to cells showing distinct cytoplasmic localization of MCM2 in (A).
To examine fragmentation and cytoplasmic localization of MCM2–7 except for MCM5, growing WI38 cells at 39 PDL were fractionated into Triton-soluble S and -insoluble P fractions. Proteins of MCM2, MCM3, MCM4, MCM6 and MCM7 in these fractions were examined by western blotting (Fig. 5A). Full-length MCM3, MCM4, MCM6 and MCM7 were clearly detected in both S and P fractions. However, full-size MCM2 was detected as a faint band, and bands that are probably degradation products of MCM2 were prominently detected, similar to the results in Fig. 2. These results indicate that MCM2 is specifically degraded near senescence among MCM2–7 proteins except for MCM5. When WI38 cells at 43 PDL from the same series of culture as in Fig. 5A were reacted with anti-MCM2, -MCM3, -MCM4, -MCM6 or -MCM7 antibodies, distinct cytoplasmic localization was detected in several cells stained with anti-MCM7 antibody, as well as with anti-MCM2 antibody (Fig. 5B). Such a staining pattern may also be observed for the anti-MCM4 antibody. These results suggest that a subset of MCM2–7 proteins is localized in the cytoplasm in senescent cells.
Detection of MCM2–7 proteins in fractions from WI38 cells. (A) WI38 cells at 39 PDL were fractionated into Triton-soluble S and -insoluble P fractions. Proteins in these fractions were examined by western blotting for the presence of MCM2, MCM3, MCM4, MCM6 and MCM7 proteins. Positions where full-length MCM proteins migrate are indicated. For detection of MCM2, an antibody that recognizes full-length MCM2 was used. (B) WI38 cells at 43 PDL were fixed and then incubated with anti-MCM2 rabbit antibody that binds to the carboxyl-terminal half region of MCM2. The cells were also incubated with anti-MCM3 rabbit antibody, anti-MCM4 rabbit antibody, anti-MCM6 rabbit antibody or anti-MCM7 mouse antibody. They were then incubated with Cy3-conjugated anti-rabbit or anti-mouse antibodies. Arrows indicate cells showing distinct cytoplasmic localization of MCM2, MCM4 and MCM7. Bars, 40 μm.
We examined changes in MCM2–7 during cellular aging of WI38 fibroblasts at the single-cell level. We used nuclear localization of p21 protein as a marker of senescence. Cells with a DNA content of 4N had higher levels of p21 on average (Fig. 1D); this finding is consistent with the notion that cells arrested at G2 phase by various stimuli to induce cellular aging bypass M phase to enter G1 phase and then become senescent (Johmura et al., 2014). We also found that cells with a DNA content of 8N had much higher levels of p21 on average. p21-positive cells showing cytoplasmic localization of MCM2 had a DNA content of 6N–8N (Fig. 4B). These findings suggest that cells with a DNA content of 4N progress through the cell cycle to acquire a DNA content of 6N–8N and become senescent. A cell with a DNA content of 4N may be arrested during S phase to generate a cell with a DNA content of 6N. It has been reported that senescent cells display low levels of MCM2–7 expression (Dumit et al., 2014; Flach et al., 2014; Bai et al., 2016). We showed by western blotting analysis that protein levels of MCM2, MCM3, MCM4 and MCM6 decrease at near-senescence (Fig. 2, Supplementary Fig. S2, S3 and S4), although the mode of the change varies significantly among different cell culture series, perhaps due to differences in cell conditions.
MCM2 is reportedly cleaved to generate a 55-kDa fragment lacking the amino terminus and the carboxyl-terminal half region of full-length MCM2 in senescent keratinocytes (Fig. 6) (Harada et al., 2008). Our data show that in WI38 cells, MCM2 is specifically fragmented among MCM2–7, and various MCM2 fragments larger than 55 kDa are produced. MCM2 antibodies reacted with fragments of 66 and 68 kDa in both Triton-soluble and Triton-insoluble chromatin fractions, and the ratio of 66–68 kDa fragments to full-length MCM2 appears to increase near senescence (Fig. 2). Cells with cytoplasmic localization of MCM2 were revealed by the antibody that recognizes the carboxyl-terminal half region of MCM2 (Fig. 3B), and these cells contained higher levels of nuclear p21 and higher DNA content (Fig. 4). MCM2 fragments of 66 and 68 kDa were mainly detected in the S fraction at senescence by the antibody (Fig. 2 and Supplementary Fig. S7). Collectively, these results thus suggest that the MCM2 fragments are localized in the cytoplasm in senescent cells.
Functional motifs and fragments of human MCM2. Nuclear localization signal (NLS), Zn-finger and ATP-binding functional motifs in human MCM2 are shown. The regions used for purification of anti-MCM2 antibodies (MCM2 full, MCM2N and MCM2C) are indicated. An MCM2 fragment that was resistant to digestion with trypsin and an expected MCM2-related fragment from senescent human keratinocytes are also shown. Molecular masses of these fragments estimated from their mobility in SDS-polyacrylamide gels, and estimated amino acid residue numbers, are shown.
In this study, we demonstrated for the first time that MCM2 among MCM2–7 except for MCM5 is specifically degraded, and that MCM2 and MCM7 are localized in the cytoplasm in senescent cells. To examine the relationship between the cytoplasmic localization of MCM proteins (Fig. 5) and cell cycle phase, the distribution of DNA content of cells showing cytoplasmic localization of MCM2 and MCM7 was examined. The results, presented in Supplementary Fig. S8 and S9, show that cells displaying cytoplasmic localization of MCM2 and MCM7 have a DNA content of 4N at high frequency. These results and those in Fig. 4 raise the possibility that the cytoplasmic localization of MCM2 and MCM7 occurs in G2 phase, similar to the case in Saccharomyces cerevisiae (Labib et al., 1999). However, the finding that several cells with cytoplasmic localization of MCM2 and MCM7 have a DNA content of 2N indicates that such localization of MCM is not restricted to G2 phase. Thus, the cytoplasmic localization of MCM2 and MCM7 may be linked to senescence rather than to the G2 phase of the cell cycle.
To examine functional domains of human MCM2, we digested human MCM2 with trypsin and obtained several fragments (Komamura-Kohno et al., 2008). One of the major products is a fragment comprising amino acids 148–676 that contains ATP-binding motifs but lacks two nuclear localization signals (Fig. 6). This trypsin-digested MCM2 fragment, which can interact with MCM4, may be related to the 66–68 kDa fragments. Smaller MCM complexes in the cytoplasm may be inhibitory to DNA replication if the number of functional MCM2–7 complexes assembled at DNA replication origins becomes limited due to their presence. We speculate that structural change of the MCM2–7 complex occurs in senescent cells and makes MCM2 in the complex accessible to digestion with cellular proteases. In addition to the increased fragmentation of MCM2, MCM2–7 expression levels decrease in senescent WI38 cells. These changes in MCM proteins may play a critical role in the loss of DNA replication capability in senescent cells.
Recently, it has been reported that cytoplasmic localization of MCM2 is significantly correlated with increased apoptosis in clear cell carcinoma, resulting in improved prognosis (Aihemaiti et al., 2018). It remains to be determined whether or not this cytoplasmic MCM2 is fragmented. When WI38 cells at 43 PDL were reacted with anti-MCM2, -MCM3, -MCM4, -MCM6 or -MCM7 antibodies, distinct cytoplasmic localization in several cells was detected by staining with the anti-MCM7 antibody as well as the anti-MCM2 antibody (Fig. 5). Such a staining pattern may also be observed for the anti-MCM4 antibody. Our present data and other published data (Dumit et al., 2014; Aihemaiti et al., 2018) suggest that a subset of MCM2–7 proteins can be localized in the cytoplasm during interphase in higher eukaryotes, which leads to inhibition of DNA replication and cell proliferation. In higher eukaryotic cells, it is unclear whether MCM2–7 proteins shuttle between cytoplasm and nucleus during interphase, as they do in S. cerevisiae. However, published data suggest that MCM proteins can shuttle in mouse cells by means of nuclear export signals in the central region of MCM3 (Chuang et al., 2012). In aged cells, cellular localization of MCM complexes may be deregulated by the fragmentation of MCM2, which results in the loss of its nuclear localization signals.
This study was supported by a Grant-in-Aid for Scientific Research from the Japan Society for the Promotion of Science (Ministry of Education, Culture, Sports, Science and Technology of Japan). We thank anonymous reviewers for giving constructive comments on the paper. This paper is dedicated to the memory of Jerard Hurwitz.