|Home | About | Journals | Submit | Contact Us | Français|
Trophoblast stem (TS) cells proliferate in the presence of fibroblast growth factor 4, but in its absence, they differentiate into polyploid trophoblast giant (TG) cells that remain viable but nonproliferative. Differentiation is coincident with expression of the cyclin-dependent kinase (CDK)-specific inhibitors p21 and p57, of which p57 is essential for switching from mitotic cell cycles to endocycles. Here, we show that, in the absence of induced DNA damage, checkpoint kinase-1 (CHK1), an enzyme essential for preventing mitosis in response to DNA damage, functions as a mitogen-dependent protein kinase that prevents premature differentiation of TS cells into TG cells by suppressing expression of p21 and p57, but not p27, the CDK inhibitor that regulates mitotic cell cycles. CHK1 phosphorylates p21 and p57 proteins at specific sites, thereby targeting them for degradation by the 26S proteasome. TG cells lack CHK1, and restoring CHK1 activity in TG cells suppresses expression of p57 and restores mitosis. Thus, CHK1 is part of a “G2 restriction point” that prevents premature cell cycle exit in cells programmed for terminal differentiation, a role that CHK2 cannot play.
Mammalian cells contain two checkpoint kinases that regulate cell cycle progression and ensure genome integrity. Checkpoint kinase-1 (CHK1) is essential for mammalian cell proliferation and embryonic development (22, 26, 32, 47). It is the effector kinase in the ATR-CHK1-CDC25 DNA damage response pathway that senses single-strand DNA breaks, bulky lesions, and stalled replication forks (38, 44). Checkpoint kinase-2 (CHK2) is the effector kinase in the ATM-CHK2-CDC25 pathway that senses double-strand DNA breaks. CHK2 is not an essential gene in mammals, apparently because CHK1 can substitute for CHK2. The function of these pathways is to prevent cells from entering mitosis until they have completed genome duplication. Results presented here reveal a novel function for CHK1 that is independent of its role in the DNA damage response. CHK1 serves as a mitogen-dependent protein kinase that prevents premature exit from the cell division cycle in cells that are developmentally programmed to terminally differentiate during tissue development or regeneration.
Mammals contain at least 12 examples of terminal cell differentiation, all of which involve expression of p57 and/or p21, and at least eight of which result in formation of polyploid cells (49). In each case, the progenitor cell exits its mitotic cell cycle in response to an environmental signal and differentiates into a unique cell type that is viable but not proliferative. Polyploid cells result either from fusion of G1 phase cells (e.g., myoblasts, monocytes, and syncytiotrophoblasts) to form multinucleated cells or from multiple S phases in the absence of cytokinesis to form cells with a single nucleus containing multiple copies of the genome. The latter occur via endoreduplication, endomitosis, or acytokinetic mitosis (49). One extensively characterized example is the differentiation of trophoblast stem (TS) cells into nonproliferating, polyploid, mononucleated trophoblast giant (TG) cells that are required for implantation and placentation. TS cells are derived from the trophectoderm of the blastocyst and give rise exclusively to all of the trophoblast lineages in the placenta (34, 39, 46). When cultured in medium conditioned by primary embryonic fibroblasts and supplemented with fibroblast growth factor 4 (FGF4), a mitogen prominently involved in mammalian embryogenesis (2, 7), TS cells proliferate as tightly packed colonies. When cultured in the absence of FGF4 and conditioned medium (referred to as FGF4-deprivation), TS cells differentiate into TG cells (16).
TS cells proliferating in the presence of FGF4 and conditioned medium can be induced to differentiate into TG cells by selective inhibition of CDK1, the cyclin-dependent kinase (CDK) required for entrance into mitosis (48). Multiple rounds of endoreduplication (termed endocycles) require oscillation of anaphase-promoting complex (APC) activity, which in the absence of CDK1 activity requires activation by CcnA-Cdk2 (27). FGF4 deprivation of TS cells rapidly induces expression of Cdkn1c/p57/Kip2 (p57) and Cdkn1a/p21/Cip1 (p21), two CDK-specific inhibitors that target CDK1 and CDK2 (48). The third member of this gene family, Cdkn1b/p27/Kip1 (p27), remains constant. Apparently, p27 is required to maintain cell proliferation by preventing premature entrance into S phase and M phase (36), whereas p21 is linked to the suppression of CHK1 and apoptosis in TG cells (10, 48), and p57 is essential for switching from mitotic cell cycles in TS cells to endocycles in TG by preventing entrance into mitosis through direct inhibition of CDK1 activity (14, 48). In the absence of a p57 gene, FGF4 deprivation of TS cells results in several rounds of cell division followed by formation of multinucleated TG cells (48), consistent with the observed association between reduced p57 expression, placentamegaly, and preeclampsia in mice and humans (references 19 and 48 and references therein). Sustaining endocycles in wild-type (wt) TG cells requires that p57 levels oscillate because assembly of prereplication complexes occurs only in the absence of CDK activity, whereas S phase requires CDK activity (6). Thus, p57 is expressed during G phase but not during S phase (14, 48).
What links FGF4 deprivation in TS cells to the expression of p57 and p21? Here, we show that the p21, p27, and p57 genes are expressed in proliferating TS cells but that CHK1 phosphorylates the p21 and p57 proteins at specific sites, thereby selectively targeting them for degradation by the 26S proteasome. Thus, in the absence of induced DNA damage, CHK1 acts as a mitogen-dependent protein kinase that prevents proliferating TS cells from exiting their mitotic cell cycle and differentiating into TG cells. Moreover, this novel role for CHK1 in the absence of induced DNA damage was not restricted to TS cells, and it could not be carried out by CHK2. Since TS cells appear to exit their mitotic cell cycle in response to FGF4 deprivation during the transition from G2 to M phase (48), we refer to it as the “G2 restriction point,” a novel checkpoint that presumably exists in all cells that are developmentally programmed for terminal cell differentiation by upregulation of p57 and/or p21.
Mouse TS and TG cells were obtained and cultured as described previously (53). Mouse embryonic stem (ES) cells (line CCE; Chemicon) were cultured per the supplier's instructions. Mouse NIH 3T3 fibroblasts (CRL-1658; ATCC) were cultured in Dulbecco's modified Eagle's medium (DMEM; Invitrogen) supplemented with 10% fetal bovine serum. Chk1 (sc-8408; Santa Cruz), p21 (sc-6246 and sc-397; Santa Cruz), p27 (sc-776; Santa Cruz), p57 (AB4058 and AB75974; Abcam), tubulin (Iowa E7; Developmental Studies Hybridoma Bank [DSHB]), actin (sc-1616; Santa Cruz), and Myc tag (2276; Cell Signaling) antibodies were used. Inhibitors were MG132 (C2211; Sigma), UCN01 (U6508; Sigma), and AZD7762 (1399; Axon).
Recombinant CHK1 and CHK2 proteins produced in SF9 insect cells (Sigma) were assayed for protein kinase activity using CHKtide, a synthetic 23-amino-acid peptide with a CHK1 and CHK2 phosphorylation site at S216 in CDC25C (Sigma), at 30°C, per the supplier's instructions. p57 peptide was synthesized by CHI Scientific. Preparation of recombinant p57 is described below.
Flow cytometry and immunofluorescence microscopy were as previously described (48), except that after fixation with 3% paraformaldehyde for 15 to 20 min, cells were rinsed briefly with phosphate-buffered saline (PBS) and permeabilized for 10 min with PBS containing 0.1% Triton X-100 and 0.02% SDS. Western immunoblotting and reverse transcription-PCR (RT-PCR) were as previously described (48). To remove protein phosphates, total cell extract containing 2 μg of protein was treated with 3,000 units of lambda phosphatase (P0753S; NEB Biolabs) for 40 min at 30°C according to the manufacturer's instructions. Sodium fluoride (50 mM) was used to inhibit phosphatase activity.
Plasmid DNAs containing mouse yellow fluorescent protein (YFP)-tagged Chk1/Chek1 (MGC-3717; ATCC) or Chk2/Chek2 (MGC-4081; ATCC) cDNA driven by the EF1 promoter was used to transfect TG cells. The K38R mutation in CHK1 [CHK1(K38R)]was constructed by site-direct mutagenesis using PCR. Mouse YFP-tagged Chk1/Chek1 cDNA was amplified by PCR using primers Chk1(Fwd) and Chk1(Rev) (Table 1). The products were digested by BamHI and XhoI and then cloned into a tetracycline (Tet)-inducible expression vector (pcDNA4/TO/myc-HisA) driven by the CMV promoter (T-Rex; Invitrogen). The CHK1(K38R) mutation was introduced using the oligonucleotide pair Mut1oligo and Chk1(Fwd) and the pair Mut2oligo and Chk1(Rev). The two PCR products were digested either by PvuI and BamHI [Mut1oligo-Chk1(Fwd)] or by PvuI and XhoI [Mut2oligo-Chk1(Rev)], ligated together into pcDNA4/TO/myc-HisA that was digested by BamHI and XhoI. All constructs were confirmed by sequencing.
Mouse Cdkn1c/p57/Kip2 cDNA (MGC-5658; ATCC) was amplified by PCR using the primers in Table 1 (psDNA-p57 primers) and cloned into pcDNA4/TO/myc-HisA to produce recombinant p57 proteins tagged with cMyc and His6 at their C termini. NIH 3T3 cells were then cotransfected with pcDNA/6TR (the tetracycline repressor) and the pcDNA-TO-p57-myc-His plasmid (ratio, 4:1) as indicated in the figure legends. Tetracycline (1 μg/ml) was added at 24 h posttransfection, and the cells were cultured for 18 h to allow p57 expression before they were transferred to tetracycline-free medium to assay p57 degradation. Mouse Cdkn1a/p21/Cip1 cDNA (MGC-6025; ATCC) was cloned into pcDNA4/TO/myc-HisA using primers in Table 2 (pcDNA-p21 primers).
Alternatively, mouse p57 and p21 cDNAs were cloned into pET-15b (Novagen) to produce N-terminal His6-p57 and His6-p21 proteins in Escherichia coli (Tables 1 and and2,2, pET-p57 and pET-p21 primers). Recombinant proteins were purified under native conditions using Ni-Sepharose Beads (GE Health Care) and eluting with 500 mM imidazole.
The human CDK1 gene driven by the cytomegalovirus (CMV) promoter was expressed from plasmid 1888 (Addgene).
Control (sc-108060) and CHK1 (sc-29270-SH) lentiviral short hairpin RNA (shRNA; Santa Cruz Biotechnology) constructs containing a puromycin resistance gene and packaging plasmids 8454 (pCMV-VSV-G, where VSV-G is vesicular stomatitis virus G protein) and 8455 (pCMV-dR8.2 dvpr) (Addgene) were used to construct lentivirus particles (45). p57 shRNA oligonucleotides were designed against DNA sequences CAGGTCCCTGAGCAGGTCT and CAAGGCGTCGAACGACGTC and cloned into the Addgene plasmid (pLKO1.TRC; plasmid number 10878). 293T cells at 75 to 80% confluence in T75 flasks were transfected with 3 μg of shRNA plasmid, 0.750 μg of pCMV-VSV-G, 2.3 μg of pCMV-dR8.2, and Lipofectamine 2000 (Invitrogen) according to the manufacturer's instructions. For p57 shRNA, 1.5 μg of each plasmid was used to transfect 293T cells. The cell culture medium was changed the following morning, and lentivirus particles were collected 24 h later. The supernatant was titrated on TS cells to determine the optimum concentration of lentivirus for suppressing p57 expression. Proliferating TS cells in six-well plates were infected with 0.5 ml of supernatant per well in the presence of 8 μg/ml of Polybrene. The culture medium was changed at 18 h postinfection, and 2 μg/ml puromycin was added 24 h later. Cells were then cultured for 3 days to eliminate noninfected cells.
Consistent with previous results (48), FGF4 deprivation of TS cells suppressed expression of CHK1 protein with concomitant induction of both p57 and p21 gene expression, but FGF4 deprivation did not affect the levels of either CHK2 or p27 protein (Fig. 1 A). Therefore, to determine whether CHK1 prevents expression of p57 and p21 in TS cells, TS cells were infected with lentiviral particles that expressed a set of shRNAs targeted against mouse CHK1 mRNA (shRNACHK1) as well as a puromycin resistance gene. Cells expressing shRNA were isolated by their resistance to puromycin.
The results revealed that shRNACHK1 suppressed expression of CHK1 RNA and protein without affecting the levels of CHK2 RNA and protein (Fig. 1B and C). Neither CHK1 nor CHK2 expression was affected by control shRNA (shRNActrl). Downregulation of CHK1 protein induced accumulation of p21 and p57 proteins (Fig. 1B), with p57 localized to the nucleus (Fig. 1D). In contrast, when TG cells (which do not express CHK1) were transfected either with shRNActrl or shRNACHK1, the levels of both p57 and p21 proteins remained unchanged, consistent with the fact that endogenous CHK1 protein is already suppressed in TG cells. The level of CHK2 protein, on the other hand, remained comparatively constant during differentiation of TS cells into TG cells, and it was unaffected by expression of shRNACHK1 (Fig. 1A, B, and C). Thus, CHK1 protein, but not CHK2 protein, was linked to the suppression of both the p57 and p21 genes in TS cells and their expression in TG cells.
To determine whether CHK1 protein itself or CHK1 protein kinase activity was required to prevent the expression and nuclear localization of p57 protein, TS cells were cultured in the presence of FGF4, conditioned medium, and either UCN01 or AZD7762, two potent inhibitors of CHK1 kinase activity with different target selectivity profiles (11, 17, 56). About 5% of proliferating TS cells spontaneously differentiated into TG cells and expressed p57 (Fig. 2 B). When these cells were treated with UCN01, p57 protein rapidly appeared (Fig. 2A, TS) and localized to the nucleus in about 80% of the cells (Fig. 1E and and2B).2B). In contrast, the level of p27 protein was not affected. TG cells which expressed both p57 and p27 were not affected by UCN01. The same results were obtained with AZD7762 (Fig. 1E; also data not shown). In contrast, within 1 to 2 days of FGF4 deprivation, TS cells differentiated into TG cells, but only about 40% of them expressed p57 (Fig. 1E and and2B).2B). The difference between the number of cells with TG morphology and the number of cells that expressed p57 reflects the fact that p57 expression occurs only during G phase of each endocycle because it is targeted for degradation by CDK-dependent phosphorylation during S-phase (14, 48). As with p57, inhibition of CHK1 kinase by either UCN01 or AZD7762 also induced expression of p21 (see Fig. 4C; also data not shown). Therefore, CHK1 kinase activity, either directly or indirectly, prevents expression of both p57 and p21 genes in TS cells.
Differentiation of TS cells into TG cells involves upregulation of p57 and p21 gene expression, selective inhibition of CDK1 activity, endocycles driven by CDK2 activity, and formation of giant cells with a characteristic pattern of gene expression (reference 48 and references therein). Consistent with these studies, suppression of CHK1 protein by shRNACHK1 suppressed TS-specific genes such as Chk1, ErrB, and Fgfr2 and upregulated TG-specific genes such as Pl1, Pl2, p57, and p21 (Fig. 1B and C). These changes in gene expression were accompanied by the appearance of giant cell morphology (Fig. 3 A) and the arrest of cell proliferation (Fig. 3B). Thus, inhibition of CHK1 kinase activity inhibited TS mitotic cell cycles.
Inhibition of CHK1 kinase activity by either shRNACHK1, UCN01, or AZD7762 induced p57 expression in most of the TS cells (Fig. 1E), suggesting that S phase as well as mitosis was inhibited. In fact, both CDK1 and CDK2 kinase activities were inhibited (Fig. 3C). Since S phase in both mitotic cell cycles and endocycles requires either CDK2 or CDK1 kinase activity (see reference 48 and references therein), p57 expression resulting from inhibition of CHK1 prevented induction of endoreduplication in TS cells (Fig. 3D) as well as cell proliferation (Fig. 3B). TG cells lack CHK1, and neither UCN01 nor AZD7762 nor shRNACHK1 affected their DNA content (Fig. 3E; also data not shown). These results were consistent with earlier observations that ectopic expression of p57 in proliferating Rcho-1 stem cells (a rat cell model for mouse TS cells) blocks DNA synthesis but not formation of giant cells (14) and that endogenous p57 in TG cells is expressed only during the G phase of endocycles, not during S phase (48). Other studies have also shown that TS cells can differentiate into giant cells under conditions where they are unable to undergo multiple rounds of DNA replication (37, 48). Therefore, direct inhibition of CHK1 activity resulted in overexpression of p57, thereby inhibiting both DNA replication and mitosis in TS cells, although not the changes in gene expression and cell growth associated with their differentiation into TG cells.
The results described above clearly link CHK1 activity in TS cells to the absence of p57 and p21 proteins in TS cells and the loss of CHK1 protein in TG cells to the appearance of p57 and p21 proteins as TS cells differentiate into TG cells. However, the mechanism by which this occurred remained to be determined. To this end, we compared expression of these genes in mouse embryonic stem (ES) cells, mouse 3T3 fibroblasts, TS cells, and TG cells. Whereas p27 RNA was detected in all four cell lines, p21 RNA was detected in only 3T3, TS, and TG cells, and p57 RNA was detected in only TS and TG cells (Fig. 4 A). This was consistent with the fact that all proliferating cells express p27 protein, that the p21 gene is expressed in most of the embryonic and adult tissues, and that most tissues do not express the p57 gene (29). However, the relationship between protein expression, mRNA expression, and CHK1 kinase activity was more complex.
Consistent with the absence of p57 mRNA in ES and 3T3 cells, p57 protein was not detected in these cells, and inhibition of CHK1 kinase activity with UCN01 or AZD7762 did not induce expression of p57 protein in these cells (Fig. 2A; also data not shown). In contrast, TS cells contained p57 mRNA but not p57 protein; p57 protein was detected only when CHK1 kinase activity was inhibited with either UCN01, AZD7762, or shRNACHK1 (Fig. 1A, 2A and B, and and4C).4C). Therefore, CHK1 regulated either p57 mRNA translation or p57 protein stability. To distinguish between these two possibilities, cells were treated with the proteasome-specific inhibitor MG132 (23). As expected, p57 protein did not accumulate in cells that did not contain p57 mRNA, but it did accumulate in TS cells (Fig. 4B). Therefore, p57 protein is synthesized in TS cells, but it is rapidly degraded by the 26S proteasome.
Consistent with the fact that TG cells contained p57 mRNA (Fig. 3A) but lacked CHK1 protein (Fig. 1A and and2A),2A), TG cells expressed the same amount of p57 protein regardless of the absence or presence of either UCN01 or AZD7762 (Fig. 2A; also data not shown). Since CHK2 was as abundant in TG cells as in TS cells, any effect these inhibitors may have had on CHK2 activity did not affect p57 expression. Treatment of TG cells with MG132, however, did increase the amount of p57 (Fig. 4B), consistent with the observations that p57 protein is degraded during S phase in TG cells as a result of its CDK-dependent degradation by the 26S proteasome (14, 18, 48).
Expression of p21 also was regulated by CHK1 at the level of protein stability. As with p57, TS cells transcribed the p21 gene (Fig. 4A), but they did not express p21 protein except under one of the following conditions: they were deprived of FGF4 (Fig. 1A); CHK1 was inhibited with shRNA (Fig. 1B), AZD7762 (Fig. 4C), or UCN01 (data not shown); or the 26S proteasome was inhibited with MG132 (Fig. 4D). These treatments did not induce p21 protein in ES cells, which lacked p21 RNA, or alter the level of p21 in TG cells, which lack CHK1. However, they did induce p21 protein in 3T3 cells and primary mouse fibroblasts, both of which produced p21 mRNA.
To determine whether CHK1 kinase alone was capable of regulating expression of p57 protein, recombinant CHK1 was expressed constitutively in TG cells by transfection with a plasmid encoding the Chk1 gene linked to yellow fluorescent protein (YFP). Within 48 h posttransfection, the level of p57 protein was reduced in the total cell population about 3-fold, whereas p27 protein levels remained unaffected (Fig. 5 A). In contrast, p57 protein levels in TG cells were unaffected by expressing either YFP alone, CHK2, or a kinase-inactive form of CHK1 with a mutation of K38 to R38 (CHK1*). To confirm the status of CHK1 kinase activity in these cells, YFP protein was immunoprecipitated from cell lysates and tested for its ability to phosphorylate CHKtide, a CDC25C peptide containing a phosphorylation site recognized by both CHK1 and CHK2 (Fig. 5B). As expected, only CHK1-YFP and CHK2-YFP were enzymatically active.
To determine whether the cells that expressed recombinant CHK1 kinase did not express p57, the cells were stained for DNA (blue), YFP protein (green), and p57 protein (red). The results revealed that cells expressing active CHK1-YFP kinase did not express p57 (Fig. 5C). About 30% of all TG cells expressed p57 prior to transfection. Transfection with a plasmid that expressed either YFP alone or the kinase-inactive CHK1*-YFP did not alter the fraction of cells expressing p57 (Fig. 5D). In contrast, transfection with a plasmid that expressed active CHK1-YFP kinase reduced the number of cells expressing p57 to about 10%; those cells that did express active CHK1-YFP did not express p57. Thus, ectopic expression of an active CHK1 kinase in TG cells suppressed endogenous p57 protein.
Notable was the fact that overexpression of CHK2 kinase activity in TG cells by transfecting TG cells with recombinant CHK2-YFP did not suppress endogenous p57 levels (Fig. 5C and D). The results were indistinguishable from those in which TG cells were transfected with either YFP or CHK1*-YFP, thereby confirming that CHK2 was not involved in regulating expression of p57.
TS cells nullizygous for p57 continue to proliferate when deprived of FGF4 for several generations before becoming multinucleated giant cells as a consequence of mitosis in the absence of cytokinesis (48). Therefore, we reasoned that ectopic expression of recombinant CHK1-YFP in TG cells would target p57 for degradation, thereby allowing mitosis to occur after the completion of S phase during endoreduplication but in the absence of cytokinesis. In fact, about 60% of the cells that expressed active CHK1-YFP kinase had two or more nuclei (Fig. 5C and and6A)6A) compared to only 5% of the cells transfected either with YFP alone, with kinase-inactive CHK1*-YFP, or with kinase-active CHK2-YFP (Fig. 5C and D). This was consistent with a restoration of CDK1 activity in the absence of p57. This conclusion was confirmed in two ways. First, TG cells transfected with a plasmid expressing recombinant CDK1 resulted in 65% of the cells becoming multinucleated (Fig. 6 B and D). This was consistent with excess CDK1 protein removing endogenous p57 through formation of stable CDK1-p57 protein complexes (48), thereby allowing the remaining CDK1 to trigger mitosis. Second, TG cells were infected with lentivirus-expressing shRNA targeted against p57 mRNA. This treatment suppressed p57 protein levels at least 10-fold without affecting either p21 or p27 protein levels (data not shown) and resulted in 63% of the cells becoming multinucleated (Fig. 6C and D). Thus, reestablishing CHK1 activity in TG cells suppresses expression of endogenous p57 and allows mitosis to occur following successive rounds of S phase.
To determine whether CHK1-dependent degradation of p57 occurred through phosphorylation of p57, TS cells were treated with AZD7762 to induce the appearance of p57, as well as with MG132 to prevent degradation of a putative p57 phosphorylated intermediate. Indeed, MG132 induced the appearance of an additional form of p57 protein that migrated slightly more slowly than p57, consistent with a phosphorylated form of p57 (Fig. 7 A). Treatment of the cell lysate with lambda protein phosphatase eliminated this new form of p57 but not in the presence of the phosphatase inhibitor NaF. Therefore, CHK1 was required to phosphorylate p57 protein in vivo, a posttranslational modification that frequently targets proteins for degradation by the 26S proteasome.
To determine whether CHK1 kinase can phosphorylate p57 and p21 directly, the ability of recombinant CHK1 and CHK2 proteins to phosphorylate these proteins in vitro was investigated. Of the three Cdkn1 proteins, only p57 and p21 contain putative CHK1 phosphorylation sites [RXX(S/T)] (35). p57 contains one site at threonine 12 and one at serine 19 (Fig. 7B). A 25-amino-acid p57 peptide (Fig. 7B) containing the putative CHK phosphorylation sites was synthesized and compared with a 23-amino-acid peptide (CHKtide) containing the established CHK1 and CHK2 phosphorylation site at S216 in CDC25C. Whereas CHKtide was efficiently phosphorylated by both CHK1 and CHK2 (Fig. 7C and D), the p57 peptide (Fig. 7C and E) and full-length recombinant p57 protein (Fig. 7C and F) were phosphorylated preferentially by CHK1. The same UCN01-sensitive CHK1 kinase activity also phosphorylated endogenous p57 immunoprecipitated from TG cell lysates (Fig. 7G).
To determine the extent to which the CHK1 phosphorylation was site specific, mutations were constructed in the recombinant p57 protein. Changing serine 19 to alanine in p57 abolished phosphorylation of p57 by CHK1 (Fig. 7H and I), whereas changing threonine 12 to alanine did not (Fig. 7H). Thus, CHK1 can phosphorylate p57 protein specifically, if not exclusively, at the CHK consensus phosphorylation site at S19.
The p21 protein contains a putative CHK1 phosphorylation site at 137-RRQTS-141 (Fig. 8 A). CHK1 phosphorylated p21, but CHK2 activity on p21 was below detection (Fig. 8B). To determine the extent to which CHK1 phosphorylation of p21 occurred at the putative CHK1 phosphorylation site, equivalent amounts (Fig. 8C) of wild-type recombinant p21 protein [p21(wt)], p21 with the mutation T140V [p21(T140V)], p21(S141A), and p21 containing both mutations [p21(TV+SA)] were tested in parallel as CHK1 substrates. T140 and S141 were phosphorylated by CHK1 at equal frequencies, which together accounted for the bulk of p21 phosphorylation by CHK1 and for the absence of phosphorylation in the p21 double mutant (Fig. 8D and E). Phosphorylation at the same site in human p21 has been shown to prevent p21 from binding to and inhibiting PCNA and CDK2, two proteins required for DNA replication (40, 43). Thus, phosphorylation of p21 at the CHK1 consensus site inactivates p21 and, in the case of TS cells, leads to its destruction.
The results described above suggest that site-specific phosphorylation of p57 and p21 by CHK1 targets these proteins for degradation by the 26S proteasome. To test this hypothesis, the stability of p57 protein was analyzed in 3T3 cells because the transfection efficiency in 3T3 cells was significantly greater than that in TS cells. Both cell types expressed CHK1, and the essential results obtained in 3T3 cells were subsequently confirmed in TS cells.
Cells were cotransfected with a tetracycline-inducible mammalian expression vector and a plasmid expressing the Tet repressor protein (Fig. 9 A and B). The Tet-inducible vector produced either the wild-type p57 gene or p57 with serine 19 changed either to aspartic acid (S19D) in order to mimic the CHK1 phosphorylated form or to alanine (S19A) in order to prevent p57 phosphorylation by CHK1. The E. coli LacZ gene served as a negative control. At 30 h posttransfection, transcription of the gene carried by the plasmid was induced by addition of tetracycline, and the steady-state levels of each p57 protein and mRNA were assayed 18 h later. Under conditions where recombinant p57 proteins (Fig. 9, lanes wt, S19D, and S19A) were overexpressed relative to endogenous p57 protein (LacZ lane), changing S19 to alanine increased expression, whereas changing S19 to aspartate decreased expression, consistent with their predicted effects on p57 protein stability in TS cells.
Strikingly, the phosphomimetic p57(S19D) was virtually undetectable in 3T3 cells and barely detectable in TS cells, whereas the phosphorylation-resistant p57(S19A) was about 50% more stable than wild-type p57 (Fig. 9A, B, E, and F). The absence of p57(S19D) protein did not result from a failure to express this gene but, rather, from its rapid degradation by the 26S proteasome. All three p57 mRNAs were present at equivalent levels (Fig. 9A), and all three proteins were expressed to comparable levels in the presence of MG132 (Fig. 9C). Therefore, the CHK1 phosphomimetic form of p57 protein was targeted rapidly for degradation by the 26S proteasome. Moreover, the appearance of high-molecular-weight forms of p57 in the presence of MG132 revealed that all three proteins were polyubiquitinated, thereby converting them into substrates for the 26S proteasome. This was consistent with the fact that p57 is targeted for degradation by at least two different protein kinases, CHK1 during G phase (this report) and CDK during S phase (14, 48).
The relative stability of wild-type and mutant p57 proteins over time was determined in proliferating cells by transferring transfected cells to tetracycline-free medium. As expected, p57(S19D) remained virtually undetectable, whereas p57(wt) protein was targeted immediately for degradation, and p57(S19A) degradation was delayed for about one cell cycle, and then the protein was degraded (Fig. 9D and E). The relative stability of wild-type and mutant p57 proteins over time was determined in nonproliferating cells by addition of cycloheximide to arrest protein synthesis. Cells were transfected with the same plasmids but in the absence of the tetracycline suppressor protein. When cycloheximide was added at 24 h posttransfection, p57(wt) protein was immediately subjected to degradation, whereas degradation of p57(S19A) was delayed by 6 to 8 h, and degradation of p57(S19D) was much more rapid than that of the wild-type protein (Fig. 9F). Thus, the inability of the cell to phosphorylate Ser19 significantly increased the stability of p57 protein in vivo, whereas a phosphomimetic substitution at Ser19 significantly decreased the stability of p57. Taken together, these results confirmed that CHK1 phosphorylation at S19 targets p57 for ubiquitin-dependent degradation by the 26S proteasome and that this pathway exists in most, if not all, cells.
CHK1 is essential for proliferation of mammalian cells (26) where depletion of Chk1 leads to premature activation of CcnB-CDK1 activity and mitotic catastrophe (32). In contrast, CHK1 is dispensable in TG cells, and TG cells are resistant to genotoxic stress, presumably because they do not undergo mitosis (48). The role of CHK1 in preventing cells with DNA damage or stalled replication forks from entering mitosis appears to have been taken over by p57 inhibition of CDK1. To test this hypothesis, cells that cannot express the p57 gene were cultured in the presence or absence of AZD7762 in order to determine whether they become sensitized to genotoxic stress. Remarkably, ES cells and 3T3 cells, which do not contain p57 mRNA (Fig. 4A), as well as TS cells that are nullizygous for the p57 gene, ceased proliferation and died in the presence of AZD7762, as evidenced by the appearance of cells in fluorescence-activated cell sorting (FACS) analysis with less than 2N DNA content as well as by a visible loss of cells attached to the culture dish (data not shown). In contrast, only about 25% of TS cells died when treated with either UCN01 or AZD7762; the remainder differentiated into stable TG cells. These results confirm that CHK1 acts upstream of p57 during TS cell differentiation and demonstrate that CHK1 prevents apoptosis by preventing cells that have not completed genome duplication from entering mitosis.
The results presented here reveal a novel role for CHK1 in preventing progenitor cells that are developmentally programmed to differentiate into viable but nonproliferating polyploid cells from exiting their mitotic cell cycle prematurely. In mammals, most, if not all, such progenitor cells express p57 and/or p21 (49). The importance of these two CDK-specific inhibitors in triggering terminal cell differentiation has been established in the case of TS cell differentiation into TG cells (14, 48), an event that occurs during peri-implantation development. Therefore, the studies presented here were undertaken in an effort to identify the link between FGF4 deprivation and expression of the p57 and p21 genes. The results revealed the surprising conclusion that CHK1, the same effector kinase that arrests cell proliferation in response to DNA damage and stalled replication forks (DNA replication checkpoint), prevents the arrest of cell proliferation in the absence of induced DNA damage (Fig. 10, G2 restriction point). These results further account for the fact that the Chk1 gene is essential for mammalian cell proliferation and that CHK1 deficiency results in peri-implantation embryonic lethality (22, 26, 32, 47).
Lack of nutrients or growth factors signals cells that external conditions are not appropriate for S phase and forces them to exit the mitotic cell cycle in G1 phase and enter a quiescent stage termed G0 (57). This cell cycle checkpoint, termed the restriction point, is based on inactivation of the retinoblastoma protein that activates the E2F transcription factors, which in turn activate genes essential for S phase. When external conditions permit, cells in G0 phase can reenter the mitotic cell cycle.
TS cells are subject to a premitosis version of the restriction point (a G2 restriction point) that is not reversible. In response to the absence of a mitogenic factor, TS cells exit their mitotic cell cycle in G2 phase, thereby allowing them to endoreduplicate their genome and differentiate into TG cells. TS cells proliferate as long as the mitogen FGF4 is present. Loss of mitogen activation in TS cells results in loss of CHK1 protein concomitant with the appearance of p57 and p21 proteins, exit from the mitotic cell cycle, and terminal cell differentiation. Here, we show that the p57 and p21 genes are transcribed in proliferating TS cells, but the proteins are phosphorylated at a specific site by CHK1, an event that targets them for polyubiquitination and subsequent degradation by the 26S proteasome (Fig. 10). In this manner, inhibition of CDK1 by p57 and, thus, exit from the mitotic cell cycle are prevented until the FGF4 mitogen is removed. This means that pluripotent cells in preimplantation embryos could begin transcribing p57 and p21 genes as soon as the trophectoderm lineage is specified at the 8-cell stage (33, 53), but CHK1 would prevent differentiation into giant cells until the mural trophectoderm is deprived of FGF4 during the peri-implantation period (16). In contrast, suppressing CHK1 activity in cells that do not express their p57 gene resulted in apoptosis, a consequence of mitotic catastrophe due to accumulated DNA damage and stalled replication forks (22). Mitotic catastrophe is averted in TS cells because p57 inhibition of CDK1 activity prevents cells from entering mitosis.
Of the two checkpoint protein kinases, only CHK1 activity was essential for preventing expression of p57 and p21 proteins in TS cells, and only the absence of CHK1 activity was essential for expression of p57 during the G phase of endocycles in TG cells. These properties were not shared by CHK2. The cellular level of CHK2 remained constant during differentiation of TS cells into TG cells, and it was unaffected by shRNACHK1 suppression of CHK1 in TS cells. The level of p57 in TG cells was unresponsive to UCN01 and AZD7762, despite the presence of CHK2, and ectopic expression of recombinant CHK2 in TG cells did not suppress p57 expression. Moreover, recombinant CHK2 was ineffective at phosphorylating p21 and p57 proteins in vitro.
The mechanism by which CHK1 expression is suppressed in response to FGF4 deprivation of TS cells remains to be elucidated although it likely involves one of the ubiquitin ligases that target this enzyme (24, 59). In addition, expression of p21 in response to FGF4 deprivation may act as a feedback loop in the G2 restriction point by suppressing expression of CHK1 (Fig. 10). CHK1 has been reported to mediate a block to transcription of the p21 gene in response to DNA replication stress in cancer cells (3). As the level of CHK1 protein diminishes during FGF4 deprivation, the level of p21 protein increases, and this increase in p21 can further reduce CHK1 mRNA levels (10).
The G2 restriction point likely operates in tissues other than the trophectoderm. All proliferating cells express CHK1, but not all cells transcribe their p57 gene. Nevertheless, cells such as mouse 3T3 fibroblasts that do not transcribe their p57 gene are capable of targeting p57 protein for degradation (Fig. 10). Many tissues contain p57 RNA (52) although p57 protein appears to be restricted to terminally differentiated cells such as TG cells, myotubes, retinal fiber cells, osteoblasts, and keratinocytes (49). Furthermore, FGF4 is expressed not only during mammalian embryogenesis (2, 7); it is also a potent inducer of angiogenesis (55) and platelet production from megakaryocytes (1, 20, 41). FGF4 also appears to regulate neural progenitor cell proliferation and differentiation (21). Mitogens other than FGF4 may also prevent p57-dependent cell differentiation. Therefore, the G2 restriction point likely operates throughout mammalian development to prevent premature cell cycle exit in progenitor cells destined for terminal differentiation.
How might CHK1 carry out two seemingly contradictory functions, one that prevents mitosis in most, if not all, proliferating mammalian cells in response to stalled replication forks and DNA damage and one that promotes mitosis in TS cells, and presumably other cells developmentally programmed for terminal differentiation, by preventing expression of the CDK-specific inhibitors p21 and p57? The resolution to this paradox rests on three points. The first is that p57 is not part of the DNA damage response; the p57 gene is transcribed primarily, if not exclusively, in cells programmed for terminal differentiation (49). Thus, suppression of p57 by CHK1 does not interfere with the cell's DNA damage response. However, p21 is part of the DNA damage response. DNA damage induces the transcription factor TP53/p53 to upregulate transcription of the p21 gene, and the resulting increase in p21 protein prevents entry into S phase, prevents DNA rereplication, and sustains the G2 phase arrest checkpoint (9). Hence, CHK1-dependent downregulation of p21 expression appears to oppose upregulation of p21 expression by DNA damage response pathways. Therefore, the second point is that the G2 restriction point and the DNA damage response are likely mediated by two different forms of CHK1 (Fig. 10).
CHK1 protein is present throughout interphase and prometaphase, but its activity is regulated by phosphorylation and ubiquitin-mediated proteolysis. In the DNA damage response, the ATR kinase “activates” CHK1 by phosphorylating it at specific sites in response to the accumulation of single-strand DNA coated with replication protein A (RPA). This activated form of CHK1, however, is unstable. ATR-dependent phosphorylation of CHK1 delivers a signal that both activates CHK1 and marks it for ubiquitin-dependent degradation (8, 59, 60), an event that exists in unperturbed proliferating cells as well as in cells responding to replicative stress (24). Furthermore, site-specific phosphorylation of CHK1 is essential for the DNA replication checkpoint, but it is not essential for CHK1 kinase activity (31, 51). Therefore, although ATR activation of CHK1 is required for CHK1 to phosphorylate its targets in the DNA damage response, it is not required for CHK1 to phosphorylate p57 and p21. All of the cellular studies reported here were carried out in the absence of induced DNA damage. Furthermore, CHK1 phosphorylated at S345 (the primary ATR phosphorylation site) was not detected in TS cell extracts (data not shown), and recombinant CHK1 phosphorylated both p57 and p21 in vitro and suppressed expression of endogenous p57 in TG cells. Recombinant CHK1 exhibits only basal activity when expressed in cultured cells (30). We suggest that the basal level of CHK1 phosphorylates p57 and p21, whereas the ATR-activated form of CHK1 does not.
The third point is that p21 is not essential for either cell proliferation or viability, despite the fact that primary mouse embryonic fibroblast cells lacking p21 fail to arrest in G1 phase in response to induced DNA damage (4, 5, 42). Therefore, low cellular levels of p21 may be sufficient to facilitate mitotic cell cycles, whereas high levels of p21 may facilitate terminal cell differentiation.
In Drosophila, cells restricted to mitotic cell cycles undergo apoptosis when induced to rereplicate their DNA, whereas cells programmed to endoreduplicate their genomes do not (28). The TS-to-TG cell paradigm suggests that the same may be true in mammals. Cells undergoing endocycles will have a proportionately greater probability of DNA damage and stalled replication forks. In Drosophila, for example, S phase in endocycles is often truncated, resulting in incomplete DNA replication. This premature entry into the Gap phase prior to the completion of genome duplication results in underrepresentation of late replicating heterochromatic sequences in many types of polyploid cells (12, 13, 25). Normally, incomplete DNA replication triggers apoptosis when the cell attempts to transit mitosis, but this problem is avoided in TG cells by suppressing CHK1 to allow expression of p57 which blocks mitosis.
Selective inhibition of CDK1 activity in TS cells induces endocycles and differentiation into TG cells, but the same treatment in ES cells results in aborted endoreduplication and apoptosis (48). Similarly, suppression of CHK1 activity triggered differentiation in TS cells but only when p57 was present to prevent cells from undergoing mitosis. Inhibition of CHK1 activity in TS p57−/− cells, ES cells, or 3T3 fibroblasts, none of which expressed the p57 gene, induced apoptosis. Similarly, ablation of the p57 gene in mice delays differentiation during mouse development and increases the frequency of apoptosis (54, 58). Cells with DNA damage or incomplete genome duplication cannot proceed into mitosis without inducing apoptosis. The CHK1-dependent phosphorylation of CDC25 prevents mitotic catastrophe during cell proliferation. Suppression of CHK1 accomplishes the same during endocycles.
CHK1 targets p57 for degradation in TS cells, thereby preventing TS cells from exiting their cell cycle. When CHK1 is downregulated in response to FGF4 deprivation, p57 selectively inhibits CDK1 to arrest cells in G2 phase. However, endocycles in TG cells require that p57 levels oscillate (14, 48) so that prereplication complexes can be assembled in the absence of CDK activity (G phase) and then subsequently converted in the presence of CDK activity into preinitiation complexes that initiate DNA replication (S phase). This oscillation in TG cells appears to depend on the absence of CHK1 (this report) and the presence of CDK2 (48). FGF4 deprivation of TS cells results in suppression of CHK1 expression with concomitant expression of p57, arresting the cell in G phase by inhibition of CDK1 (Fig. 10). CcnE accumulates and activates CDK2 (50). CCNE-CDK2 then phosphorylates p57 near its C terminus, thereby targeting it for ubiquitination by SCFSkp2 and subsequent degradation by the 26S proteasome (14, 18). As CcnE is degraded, the level of p57 again rises due to the absence of CHK1 and the loss of CDK2-CcnE activity, S phase terminates, and the cell is again in G phase.
We thank Alex Vassilev for his technical advice.
We thank the intramural research program of the National Institute of Child Health and Human Development for its financial support.
Published ahead of print on 26 July 2011.