Chk1-mediated Cdc25A degradation as a critical mechanism for normal cell cycle progression Free

DNA damage caused by exogenous and endogenous factors occurs constantly in normal cells. DNA lesion activates a variety of cellular DNA damage response (DDR) mechanisms, including the cell cycle checkpoints, DNA repair, apoptosis and senescence (Bartek and Lukas, 2007; Ciccia and Elledge, 2010; Jackson and Bartek, 2009). The DDR requires the activation of two evolutionarily conserved protein kinase cascades, the ATM–Chk2 and ATR–Chk1 (Chk1 and Chk2 are encoded by CHEK1 and CHEK2, respectively, in mammals) pathways (Antoni et al., 2007; Awasthi et al., 2015; Blackford and Jackson, 2017; Flynn and Zou, 2011; Goto et al., 2015; Medema and Macůrek, 2012; Reinhardt and Yaffe, 2009; Saldivar et al., 2017; Shiloh and Ziv, 2013; Zhang and Hunter, 2014). ATM is primarily activated by DNA double-strand breaks (DSBs). ATM phosphorylates and activates Chk2. Both kinases phosphorylate p53 (encoded by TP53) resulting in p53 stabilization, which promotes p21 (also known as CDKN1A) expression. p21 binds and inhibits the cyclin and cyclin-dependent kinase (CDK) complexes leading to cell cycle arrest. By contrast, ATR is activated by a broader spectrum of DNA-damaging stimuli, such as ultraviolet (UV) radiation, DNA replication stress and inter-strand DNA crosslinking reagents. ATR phosphorylates and activates Chk1, which in turn phosphorylates and inhibits Cdc25 phosphatases. Cdc25 dephosphorylates inhibitory phosphorylation sites on CDK (e.g. CDK1-Tyr15), resulting in CDK activation. Thus, Cdc25 inhibition ends up causing cell cycle arrest (Boutros et al., 2007). There are three Cdc25 paralogs (Cdc25A–Cdc25C) in human cells. Cdc25A phosphorylation by Chk1 triggers its degradation in a ubiquitin/proteasome-dependent manner (Boutros et al., 2007; Mailand et al., 2000). Phosphorylation of Cdc25B and Cdc25C by Chk1 reduces their phosphatase activity (Boutros et al., 2007).

DDR signaling pathways were initially considered to play critical roles mainly in cells under exogenous DNA-damaging stress. However, previous studies showing that complete deficiency of ATR or Chk1 leads to aberrant accumulation of DNA damage with early embryonic lethality (Brown and Baltimore, 2000; Liu et al., 2000; Takai et al., 2000) challenged this general belief. Interestingly, knockout of Cdc25A also results in early embryonic lethality (Ray et al., 2007). In sharp contrast, mice with complete deficiency of Cdc25B or Cdc25C alone or with double knockout of Cdc25B and Cdc25C are viable without any relevant phenotype, except for some meiotic abnormalities during oogenesis in Cdc25B-knockout mice (Chen et al., 2001; Ferguson et al., 2005; Lincoln et al., 2002). Cells derived from Cdc25B and Cdc25C double-knockout mice show no apparent abnormalities in their cell cycle profile or DDR (Ferguson et al., 2005). However, in human somatic cells, Cdc25B reportedly plays critical roles in normal G2/M transition and in the cell cycle checkpoint at the G2 phase, together with Cdc25A (Cazales et al., 2005; Lammer et al., 1998; Timofeev et al., 2010; Vázquez-Novelle et al., 2010).

A previous study has shown that Chk1 inhibition using a Chk1 inhibitor or Chk1-specific siRNA induces accumulation of DNA damage and DDR even in the absence of exogenous DNA insults (Beck et al., 2010; Syljuasen et al., 2005). However, whether the signaling pathway mediated by Chk1 in unperturbed cell cycle progression and in DDR is similar remains unknown. To clarify this, conditional gene-knockout- or RNA interference-mediated knockdown of Chk1 could be used. However, it takes days to observe significant phenotypic changes using these technical approaches, and thus it is extremely difficult to distinguish whether the observed phenotype is due to Chk1 inhibition or to DDR. The use of Chk1 inhibitor is an alternative approach but, in this case, specificity of the inhibitor could become a problem. To overcome these difficulties, in the present study, we established a human colon carcinoma HCT116 cell line carrying endogenous Chk1 that can be rapidly and specifically degraded in an indole-3-acetic acid (IAA; a natural auxin)-dependent manner (Natsume et al., 2016; Nishimura et al., 2009). Rapid Chk1 depletion impeded cell proliferation and caused accumulation of aberrant DNA damage marker proteins, such as γH2AX (H2AX phosphorylated at Ser139; H2AX is encoded by H2AFX). Cdc25A stabilization was observed before the accumulation of DNA damage factors. In addition, the Chk1 depletion-associated phenotype was partially suppressed by depletion of both Chk1 and Cdc25A, suggesting that Chk1 plays a role in the normal cell cycle in the absence of exogenous DNA damage mainly by targeting Cdc25A.

The CRISPR/Cas9 gene editing technology enables insertion of a DNA fragment (e.g. tags and drug-resistant markers) at a desired gene locus by a sequence-specific DSB induction and homologous recombination repair (Cong et al., 2013; Mali et al., 2013; Ran et al., 2013). We combined this gene knock-in technology to generate human conditional cells using a donor harboring a minimum auxin-inducible degron (mAID) tag (Natsume and Kanemaki, 2017; Natsume et al., 2016; Nishimura et al., 2009). As shown in Fig. 1A–C, we established a human near-diploid HCT116 colon carcinoma cell line (Thompson and Compton, 2008) expressing an auxin-inducible ubiquitin ligase component, Oryza sativa (Os)TIR1, and the endogenous Chk1 protein fused with mAID and a monomeric Clover (mAID–mClover; mACl) (Natsume et al., 2016) or with mAID fused to five repeats of the Myc epitope tag (mAID–5Myc; mAM) at its C-terminus. We established two independent clones per each genotype. Even in the absence of IAA, the protein level of mAID-tagged Chk1 was slightly lower than that of the wild-type (WT) Chk1 (Fig. 1C, see the lanes corresponding to DMSO-treated cells, labeled D). This was likely due to weak activation of OsTIR1 in the absence of IAA (Natsume et al., 2016). However, cells expressing mAID-tagged Chk1 (Chk1–mAID cells) showed only a marginal growth defect in the absence of IAA (data not shown; also see Fig. 2A), suggesting that the Chk1 fusions are functional and that the expression level is sufficient to support proliferation.

The addition of IAA induced depletion of Chk1, but not of Chk2 in the established Chk1–mAID cell lines [Fig. 1C, compare IAA-treated cells (labeled I) with DMSO-treated cells (labeled D)]. The mAID-tagged Chk1 was rapidly depleted within 15–30 min after the addition of IAA to the growing medium, as shown for clone 1 and clone 3 in Fig. 1D. We confirmed similar depletion kinetics in clone 2 and clone 4 (data not shown; also see Fig. 2C).

In response to replication stress induced by hydroxyurea (HU) treatment or to DNA damage from UV irradiation, mAID-tagged Chk1 was phosphorylated at Ser317 and Ser345 by ATR (Zhao and Piwnica-Worms, 2001) and at Ser296 through Chk1 auto-phosphorylation (Clarke and Clarke, 2005; Kasahara et al., 2010), in similar manner to that seen for the WT Chk1 protein (Fig. S1A–C). In the absence of IAA, UV irradiation reduced the Cdc25A protein level in Chk1–mAID cell lines, as in their parental cells (Chk1^WT/WT; Fig. S1D). On the other hand, IAA incubation abolished this Cdc25A degradation in Chk1–mAID cell lines, but not in their parental cells (Fig. S1D), consistent with previous reports showing that Chk1 phosphorylates Cdc25A and that this mediates Cdc25A degradation during the DDR (Mailand et al., 2000; Zhao et al., 2002). This Chk1 phosphorylation (Fig. S1A–C) and Cdc25A degradation (Fig. S1D) indicate that the functions of mAID-tagged Chk1 and WT Chk1 protein in the DDR are similar.

We examined the effect of acute Chk1 depletion on cell proliferation in the absence of exogenous DNA insult. As shown in Fig. 2A, each Chk1–mAID cell line showed only marginal changes in cell growth after 1 day in the presence of IAA compared to control DMSO. However, a 3-day culture in the presence of IAA significantly reduced the cell proliferation of four independent Chk1–mAID cell lines but not that of their parental cells (Fig. 2A; also see Fig. 3C). FACS analyses revealed that a 3-day culture in the presence of IAA increases the sub-G1 fraction (a measure of the apoptotic cell fraction) in Chk1^mACl/mACl cell lines (Fig. 2B). However, as judged by the Trypan Blue staining, the percentage of dead cells varied from less than 2% to more than 10% by experiment (data not shown; also see FACS data in Fig. 4A), suggesting that the increase in apoptosis may not be the main cause of growth defect by Chk1 depletion. FACS analyses also indicated that the G2/M fraction was elevated upon IAA treatment for 3 days in Chk1^mACl/mACl cells but not in their parental cells (Fig. 2B). Since only 1–2% mitotic cells were detected in these IAA-treated cells (Fig. S2A), Chk1 depletion induces cell cycle arrest at the G2/M transition; we observed a similar tendency for the Chk1^mAM/mAM cell lines (data not shown). A 3-day culture in the presence of IAA increased the level of p53 and p21 proteins but not the level of CDK1 phosphorylated at Tyr15 (an inactive form of CDK1; Fig. S2B) in Chk1–mAID cell lines. Therefore, Chk1 depletion results in p53 stabilization, which induces cell cycle arrest likely through the induction of CDK inhibitor(s), such as p21.

We analyzed early changes after IAA treatment by immunoblotting (Fig. 2C; Fig. S3). The phosphorylation of H2AX-Ser139 (γH2AX) and Chk2-Thr68 gradually increased after IAA incubation; it peaked at ∼20–24 h in the observed range (from 0 h to 24 h). Since p53 stabilization was detected just after this elevation, DNA damage accumulated upon the rapid Chk1 depletion, which in turn induced p53 stabilization. By contrast, the protein level of Cdc25A started to increase at ∼2–4 h, peaked at ∼8 h and then decreased thereafter (Fig. 2C; Fig. S3; also see Fig. 1D). These observations suggest that acute Chk1 depletion leads to Cdc25A stabilization before accumulation of DNA damage and expression of DDR proteins.

Since DNA damage products such as γH2AX reportedly start to increase 1 h after incubation with the Chk1 inhibitor UCN-01 in U2OS cells (Syljuasen et al., 2005), we used two types of Chk1 inhibitors, MK-8776 (Thompson et al., 2012; Chang and Eastman, 2012) and CHIR-124 (Tao et al., 2009; Tse et al., 2007). In HCT116 cells, incubation with 2 µM MK-8776 stabilized Cdc25A, similar to what was observed with IAA-induced Chk1 depletion (compare Fig. S2C with Figs 1D, 2C and Fig. S3); however, we observed no significant elevation of γH2AX (data not shown). In HCT116 and HeLa cells, Cdc25A stabilized at ∼2–4 h after the treatment with 250 nM CHIR-124 whereas it took ∼8 h to detect γH2AX (Fig. S2D); these kinetics closely resemble data obtained using Chk1–mAID cell lines (Fig. 2C; Fig. S3). We could be observing a time lag (∼2–4 h) between Chk1 inhibition and Cdc25A stabilization because the Cdc25A degradation triggered by Chk1-induced phosphorylation requires multiple steps, including additional Cdc25A phosphorylation and ubiquitylation (Boutros et al., 2007; Goto et al., 2015). On the other hand, the kinetics of γH2AX appears to differ slightly among experimental settings, which may reflect a difference in the degree of Chk1 inhibition and/or off-target effect(s).

In order to further clarify the role of Cdc25A in cell proliferation, we tagged endogenous Cdc25A with mAID and mCherry2 (mACh) at the C-terminus in the Chk1^WT/WT or Chk1^mACl/mACl backgrounds (Fig. S4A; Fig. 3A). Before generating the double mAID cell lines (Chk1^mACl/mACl+Cdc25A^mACh/mACh), we removed the floxed Hyg-resistant cassette from Chk1^mACl/mACl cells (clone 3) with Cre recombinase (see Fig. 1A). From this intermediate clone (referred to as Chk1^mACl/mACl clone 3′), we generated the double mAID cell lines (Chk1^mACl/mACl+Cdc25A^mACh/mACh clones 5 and 6; Fig. 3A). Simultaneous depletion of Chk1 and Cdc25A partially suppressed the cell growth defect observed in Chk1-depleted cells (Fig. 3B,C). Cdc25A single depletion was not associated with a growth advantage (Fig. S4B; Fig. 3C) because the growth defect was prominent in both Cdc25A^mACh/mACh cell lines at late stages (at ∼6 days after the addition of IAA; data not shown). Therefore, Cdc25A counteracts Chk1 even in normal cell cycle progression. Our current observations closely resemble the results reported in a previous study that used Chk1- and/or Cdc25A-specific siRNA(s) in U2OS or TIG-3-tert cells (Beck et al., 2010). Cell growth was impaired by single depletion of Chk1 or Cdc25A but the phenotype was far milder in Cdc25A-depleted cells. The low proliferation rate seen upon Chk1 depletion was partly improved by co-depletion of Cdc25A.

These observations also raised the question of whether prior Cdc25A degradation completely rescues growth defect caused by Chk1 inhibition. To address this question, Cdc25A^mACh/mACh cells (clone 8) were pre-treated with DMSO or IAA for 2 h and then incubated with 0–750 nM CHIR-124 for a further 3 days (Fig. S4C). Prior degradation of Cdc25A did not completely recover the growth inhibition caused by CHIR-124 treatment (Fig. S4D). However, the growth defect caused by Cdc25A depletion was partly attenuated by co-treatment with 250–500 nM CHIR-124 (Fig. S4D,E); we observed a similar tendency using clone 7 (data not shown). These findings support the idea that there is a close relationship between Cdc25A and Chk1.

Finally, we performed the immunoblotting (Fig. 3D) or FACS analyses after pulse BrdU labeling (Fig. 4). A 3-day incubation in the presence of IAA elevated the levels of p53 and p21 not only in Chk1-mACl cells (clone 3′) but also in double mAID cells (clones 5 and 6; Fig. 3D). Both the mitotic index (Fig. S2A) and the FACS data (Fig. 4A,B) revealed arrest of the above cells (clones 3′, 5 and 6) at G2 phase after the 3-day incubation in the presence of IAA. The Chk1-depleted cells also arrested at S phase because the proportion of BrdU-negative S phase cells was elevated among the total S phase cells (Fig. 4A,C); however, the population of cells arrested at G2 phase was much higher (Fig. 4). Since co-depletion of Cdc25A partially attenuated all these parameters linked to the cell cycle arrest caused by Chk1 depletion (Figs 3D and 4), we can infer that Chk1 controls normal cell cycle progression mainly by inducing Cdc25A degradation.

Even in the absence of any exogenous DNA insult, Chk1 is reportedly required to control DNA replication: Chk1 restricts the number of firing replication origins (Beck et al., 2010; Maya-Mendoza et al., 2007; Saldivar et al., 2018; Syljuasen et al., 2005; Zhang and Hunter, 2014). Treatment with Chk1 inhibitor, Chk1-specific siRNA or disruption of the Chk1-gene activates latent DNA replication origins (e.g. late origin firing) leading to depletion of replication factors at each replicon. This results in stalled DNA replication forks that induce DNA breaks. In this study, we established HCT116 cell lines carrying endogenous Chk1 tagged with a mAID tag at its C-terminus. Using these cell lines, we found that Chk1 depletion induces cell cycle arrest at the S or G2 phase but not at the G1 phase or G1/S transition. These results are consistent with the observations showing that Chk1 inhibition mainly disturbs DNA replication during a normal cell cycle progression leading to accumulation of DNA damage specifically at S or G2 phase (Beck et al., 2010; Maya-Mendoza et al., 2007; Saldivar et al., 2018; Syljuasen et al., 2005). In addition, we demonstrated that rapid and specific Chk1 depletion induces Cdc25A stabilization before DNA damage accumulation. This finding is in line with previous reports showing Cdc25A stabilization after treatment with a Chk1 inhibitor (Sørensen et al., 2003) and in Chk1-haploinsuffient mice (Lam et al., 2004). In addition, co-depletion of Chk1 and Cdc25A partially suppressed the growth defect observed after Chk1 single depletion, confirming a previous report using Chk1- and/or Cdc25A-specific siRNA(s) (Beck et al., 2010). These observations suggest that Chk1 controls normal cellular proliferation through a signaling pathway similar to the DDR. The ability of ChK1 to phosphorylate several substrates other than Cdc25A during the DDR (Dai and Grant, 2010; Patil et al., 2013; Zhang and Hunter, 2014) might explain why co-depletion of Chk1 and Cdc25A did not completely suppress the growth defect. These issues will be addressed in future investigations.

Genomic DNA (gDNA) was purified from HCT116 cells stably expressing Oryza sativa (Os)TIR1 (HCT116-OsTIR1s) (Natsume et al., 2016) and their derivatives by using a Wizard^® genomic DNA purification kit (Promega, Madison, WI). Genomic PCR (gPCR) was amplified using Tks Gflex DNA polymerase (Takara Bio, Kusatsu, Shiga, Japan) (Natsume et al., 2016) with the following primers: 5′-AGCATTTGCCGCAGTACTCT-3′ (Chk1 sense primer, denoted S) and 5′-GCTTCGCTTCACAGACTGA-3′ (Chk1 antisense primer, denoted AS) for the Chk1 gene; 5′-GTTTCAGGTTCAGGGGGAGG-3′ (3′UTR antisense primer, denoted 3′ UTR), 5′-TCGATCAGGATGATCTGGAC-3′ (Neo sense primer, denoted Neo) and 5′-CATATGCGCGATTGCTGATC-3′ (Hyg sense primer, denoted Hyg in Fig. 1A) for the inserted DNA cassette described below; and 5′-GGTGCCGGTATGTGAGAGAG-3′ (Cdc25A sense primer) and 5′-GCAACATTCCAGCACTGAGC-3′ (Cdc25A antisense primer) for the Cdc25A gene.

Gene targeting using CRISPR/Cas9 was performed in HCT116-OsTIR1s and their derivatives as described previously (Natsume et al., 2016). In brief, annealed DNA oligonucleotides for each guide RNA (gRNA) was cloned into BbsI sites in pSpCas9 (BB)-2A-Puro V2.0 (pX459, Addgene plasmid ID #62988) (Ran et al., 2013). The gRNA sequences targeting Chk1 and Cdc25A loci were 5′-GAGCCGATGGTCCGATCATG(TGG)-3′ and 5′-GAAGAAGCTCTGAGGGCGGC(AGG)-3′, respectively [protospacer adjacent motif (PAM) sequences are indicated in parentheses]. For the construction of the Chk1-targeting vectors, 5′- and 3′-homology arms were amplified from the Chk1 gPCR product. By using a NEBuilder HiFi DNA assembly master mix (New England Biolabs, Ipswich, MA), these arms were inserted into the pMK289 [mAID-mClover (mACl)-Neo resistance cassette], pMK290 (mACl-Hyg resistance cassette), and their modified vectors [mAID-5Myc (mAM)-Neo and -Hyg resistance cassettes; also see Fig. 1A] as described previously (Natsume et al., 2016). For the generation of Chk1^{mACl (mAM)/mACl (mAM)} cells, HCT116-OsTIR1 s were co-transfected with Chk1-pX459 plasmid and (modified) pMK289- and pMK290-Chk1-targeting vectors (1:1:1 ratio). After selection with 0.4 mg/ml G418 and 0.1 mg/ml Hygromycin B Gold (Nacalai Tesque, Kyoto, Japan), each clone was screened by gPCR and immunoblotting. For the Cdc25A gene, we employed similar strategies, except for the use of pMK292 [mAID–mCherry2 (mACh)-Neo resistance cassette]- and pMK293 (mACh-Hyg resistance cassette)-Cdc25A-targeting vectors (Natsume et al., 2016). For the generation of double mAID (Chk1^mACl/mACl+Cdc25A^mACh/mACh) cell lines, the floxed Hyg resistance cassette (also see Fig. 1A) was removed from Chk1^mACl/mACl cells (clone 3) through infection with adenovirus carrying Cre recombinase. The established cells (clone 3′) were then transfected with Cdc25A-pX459 plasmid and pMK293-Cdc25A-targeting vectors (1:2 ratio).

Original HCT116 or HeLa cells were purchased from the American Type Culture Collection (ATCC CCL-247 or CCL-2, respectively). HCT116 cell lines including HCT116-OsTIR1 and their derivatives were cultured as previously described (Natsume et al., 2016). HeLa cells were grown in Dulbecco's modified Eagle's medium containing 10% fetal bovine serum. For the induction of degradation of each mAID-tagged endogenous protein, a final concentration of 500 μM IAA was added to the growing medium. An equal volume of DMSO was used as a negative control.

All cell lines used in this study tested negative for mycoplasma contamination.

One day before analysis of the cell proliferation rate (day −1), logarithmically growing cells were trypsinized and re-plated at a density of 5×10⁴ cells per dish. One day after re-plating (day 0), cells were incubated in a culture medium containing DMSO or 500 μM IAA for the number of indicated days. Each medium was changed every 2 days. The number of living cells was quantified using a TC20 automated cell counter (Bio-Rad Laboratories, Richmond, CA). In brief, cells were trypsinized and resuspended in the culture medium at each time point. Then, 10 µl of cell suspension was mixed with an equal volume of 0.4% Trypan Blue solution (Bio-Rad Laboratories). The mixture was applied to the cell counter following the manufacturer's protocol.

FACS analyses were performed as described previously (Natsume et al., 2017), except that cells were fixed in 95% ethanol.

Immunoblotting analyses were performed as previously described (Enomoto et al., 2009; Goto et al., 2016). We used the following antibodies: Chk1, 1:2000 [mouse monoclonal antibody (mAb), 2G1D5; cat. no. #2360, Cell Signaling Technology (CST), Beverly, MA]; Chk1-pS296, 1:2000 (rabbit mAb, D309F; cat. no. #90178, CST); Chk1-pS317, 1:2000 (XP^® rabbit mAb, D12H3; cat. no. #12302, CST); Chk1-pS345, 1:2000 (rabbit mAb, 133D3; cat. no. #2348, CST); Chk2, 1:2000 (XP^® rabbit mAb, D9C6; cat. no. #6334, CST); Chk2-pT68, 1:2000 (rabbit mAb, C13C1; cat. no. #2197, CST); p21 Waf1/Cip1, 1:2000 (rabbit mAb, 12D1; cat. no. #2947, CST); γH2AX, 1:2000 (rabbit mAb, 20E3; cat. no. #9718, CST); CDK1-pY15, 1:2000 (rabbit mAb, 10A11; cat. no. #4359, CST); Cdc25A, 1:500 (mouse mAb, F6; cat. no. sc-7389, Santa Cruz Biotechnology, Santa Cruz, CA); p53, 1:2000 (mouse mAb, 80/p53; cat. no. 610184, BD Transduction Laboratories, San Diego, CA); GAPDH, 1:10,000 (mouse mAb, 5A12; cat. no. 010-25521, Wako Pure Chemical Industries, Osaka, Japan); horseradish peroxidase (HRP)-conjugated anti-rabbit-IgG, 1:50,000 [goat polyclonal antibody (pAb); cat. no. ab205718, Abcam, Cambridge, MA]; and HRP-conjugated anti-mouse-IgG, 1:50,000 (rabbit pAb; cat. no. ab97046, Abcam). Except for the immunoblotting with anti-p21, -γH2AX or -GAPDH antibodies, we used immunoreaction enhancer solution (Can Get Signal^®, Toyobo, Osaka, Japan) for the dilution of primary and secondary antibodies. The band intensity in the immunoblotting was quantified using densitometry as described previously (Li et al., 2012).

We would like to thank Y. Hayashi, N. Tanigawa and Y. Nakai (Aichi Cancer Center Research Institute) for technical assistance; M. Fujita (Kyushu University), K. Furuya (Kyoto University) and H. Kosako (Tokushima University) for helpful discussion.