Loss of the mammalian APC/C activator FZR1 shortens G1 and lengthens S phase but has little effect on exit from mitosis

The anaphase-promoting complex/cyclosome (APC/C) is a large multisubunit ubiquitin protein ligase that is essential for sister chromatid separation and exit from mitosis. Because these events are crucial for equational chromosome segregation, the activity of the APC/C is tightly controlled by several mechanisms, including cyclin-dependent kinase (CDK)-dependent phosphorylation and the binding of cell division control protein 20 (CDC20). Once activated, the APC/C recognises its substrates via specific motifs, such as the destruction boxes (D-box) present in cyclin B1 and securin. Its activity against these essential substrates is, however, constrained by the mitotic checkpoint to prevent premature sister chromatid separation and exit from mitosis, which can generate aneuploid progeny (reviewed by Peters, 2006). In contrast to the ubiquitylation of securin and cyclin B, which depends on the completion of chromosome bi-orientation, the APC/C-dependent degradation of cyclin A occurs as soon as cells enter mitosis, despite the activity of the mitotic checkpoint (den Elzen and Pines, 2001; Geley et al., 2001). These differences are poorly understood, but recent evidence points to significant differences in the molecular recognition of different substrates (Kimata et al., 2008; Matyskiela and Morgan, 2009; Wolthuis et al., 2008).

As cyclins are degraded, CDK activity declines and phosphatases convert the APC/C back to its hypophosphorylated interphase state, which is activated by Fizzy-related (FZR1, also known as CDH1), a CDC20-related protein (Kramer et al., 1998). Similarly to CDC20, FZR1 contains seven WD40 repeats, which form a seven-bladed propeller that is involved in substrate and APC/C binding (Kraft et al., 2005). In contrast to APC/C-CDC20, the FZR1-associated form of the APC/C has a different substrate spectrum. In addition to D-box-containing proteins, the APC/C-FZR1 also targets other proteins with different degrons, such as the KEN box present in CDC20 (Pfleger and Kirschner, 2000). The switch from CDC20- to FZR1-activated APC/C occurs during anaphase B (Hagting et al., 2002) and it is believed that FZR1, by targeting Aurora kinase A and PLK1, contributes significantly to APC/C-dependent proteolysis during exit from mitosis (Pines, 2006).

After completion of mitosis, the APC/C remains active until the next S phase (Brandeis and Hunt, 1996) and several APC/C substrates either remain or become unstable in G1 phase. Degradation of some of these substrates is important for maintenance of G1 phase and avoidance of premature entry into S phase. These include the mitotic cyclins, which continue to be translated after exit from mitosis, but are unstable in G1 phase (Almeida et al., 2005), and the F-box protein SKP2, which controls the abundance of the CDK inhibitor p27^Kip1 (Wei et al., 2004). Recently, it has been shown that the tumour suppressor RB1 (pRB) cooperates with FZR1 in targeting SKP2 for polyubiquitylation (Binne et al., 2007). Thus, in addition to its role in destabilising late mitotic substrates, FZR1 seems to have an important role in controlling G1 phase. By maintaining the APC/C activity during exit from mitosis and in early G1 phase, FZR1 could also contribute to the establishment of pre-replicative complexes (pre-RCs) on origins of DNA replication by providing a time window with low CDK activity (reviewed by Diffley, 2004).

At the end of G1 phase, the APC/C is inactivated by FBXO5 (Emi1), which binds to and inactivates FZR1, allowing cyclin A to accumulate and activate CDK2. Active CDK2, in turn, contributes to the inactivation of the APC/C by inhibitory phosphorylation of FZR1 (Kramer et al., 2000). In addition, APC/C activity in G1 phase is self-limited by autoubiquitylation and degradation of UBCH10, one of the E2 enzymes of APC/C (Rape and Kirschner, 2004). Knockdown of UBCH10 has recently been shown to shorten G1 phase by increasing cyclin A levels, suggesting that UBCH10 is the essential E2 for FZR1-APC/C in G1 phase (Rape and Kirschner, 2004; Walker et al., 2008).

FZR1 and its orthologues have been studied in several organisms and were found to be important for establishing or maintaining G1 phase, but are dispensable for mitosis (Blanco et al., 2000; Jacobs et al., 2002; Sudo et al., 2001; Visintin et al., 1997). By contrast, analysis of mouse embryonic fibroblasts (MEFs) derived from FZR1-deficient knockout mice has provided evidence that FZR1 is involved in regulating exit from mitosis (Garcia-Higuera et al., 2008; Li et al., 2008), whereas FZR1 RNAi-based studies in human cells have been controversial (Floyd et al., 2008; Engelbert et al., 2008).

To clarify the role of FZR1 during exit from mitosis, we used two experimental systems using RNAi in human cells and conditional gene ablation in the mouse. In both systems, we found that cells had no obvious problems in returning to interphase after the end of mitosis. Cells lacking FZR1, however, had an unusually short G1 phase and proliferated poorly.

To study whether FZR1 is essential for exit from mitosis, we used transient RNAi to knockdown FZR1 in human U2OS osteosarcoma cells, which reduced FZR1 levels to more than 95% as determined by immunoblotting and densitometry scanning. Fig. 1A shows that knockdown of FZR1 stabilised several known FZR1 substrates, such as hKid (KIF22), Aurora kinase A (AURKA) and B (AURKB). However, CDC20 and PLK1 were only slightly stabilised in FZR1 RNAi cells released from a nocodazole (noc) block. Their overall levels in RNAi cells were higher and both proteins re-accumulated faster in late-G1 phase than in controls. In FZR1 RNAi cells, the early-G1 phase decline of CDC25A and UBCH10 was prevented, but as cells progressed towards S phase, the protein levels of both proteins dropped. SKP2 declined in G1-S phase independently of FZR1 levels (Fig. 1Ac,1B) but the overall amount of SKP2 was about ten times higher in FZR1 RNAi cells.

In contrast to the above FZR1 substrates, the degradation of cyclin A and cyclin B1, geminin and securin was not affected in mitosis (Fig. 1B), but these proteins re-accumulated ∼2 hours sooner in G1 phase in FZR1-knockdown cells (Fig. 1B). In summary, knockdown of FZR1 only affected the degradation of late mitotic and G1-phase APC/C substrates. But although late mitotic substrates, such as AURKA and PLK1 were stabilised, the length of mitosis was not significantly prolonged, as judged as the time from cell rounding to initiation of cytokinesis by live-cell phase-contrast time-lapse microscopy (Fig. 1C). To monitor cyclin degradation with higher resolution, U2OS cells with tetracycline-inducible expression of cyclin-B1-GFP (Wolf et al., 2006) were grown in the presence of 1 μg/ml doxycyline (dox), transfected with 50 nM FZR1 siRNA for 24 hours and analysed by fluorescence time-lapse imaging. As Fig. 1D shows, cyclin B1-GFP was efficiently degraded during mitosis in the absence of FZR1 but re-accumulated faster in G1 phase [2.4±0.46 hours after cytokinesis in RNAi cells (n=8) compared with 5.5±1.3 hours in control cells (n=15)]. Similar effects of transient FZR1 knockdown on APC/C substrate stabilisation and exit from mitosis were obtained in HeLa cervical carcinoma cells (data not shown).

To determine the long-term effects of FZR1 knockdown, we established a line of U2OS cells with dox-inducible expression of a short hairpin RNA (shRNA) directed against FZR1. Induction of this shRNA caused a more than 95% reduction of FZR1 protein levels (supplementary material Fig. S1) and inhibited the degradation of AURKA, AURKB and hKid (not shown). As in the above experiments performed with transient transfection of siRNAs, the degradation of cyclin B1 during mitosis was not affected in induced cells released from a mitotic arrest (Fig. 1E). Consistently, the length of mitosis (mean duration 32±5.5 minutes, n=107, in non-induced controls and 37±17.8 minutes, n=161, in FZR1-knockdown cells) was not altered. In addition, using immunofluorescence microscopy, we could not detect a higher frequency of cytokinesis failure in FZR1-knockdown cells, in which AURKB was stabilised (Fig. 1F). In summary, FZR1 RNAi, induced by transient siRNA transfection or inducible expression of shRNAs in stable cell lines, did not hamper progression through mitosis.

When conditional FZR1 RNAi U2OS cells were treated for prolonged periods with 1 μg/ml dox they were found to grow more slowly than controls (Fig. 2A). To confirm that knockdown of FZR1 impairs proliferation, we mixed GFP-expressing FZR1 shRNA cells with their parental cell line (U2OS) in the presence or absence of dox and measured the proportion of GFP-positive RNAi cells by flow cytometry. As can be seen in Fig. 2B, RNAi induction with dox caused a proliferation disadvantage and percentage loss of GFP-positive FZR1 RNAi cells (MIX+dox). When stable FZR1-knockdown cells were cultivated for up to 11 days and analysed by immunoblotting, we noted that FZR1 RNAi led to an increase of cyclin B1 levels (Fig. 2C) and strong induction of the CDK inhibitor p21^CIP1, starting at day 3 of shRNA induction.

To determine the cause underlying the reduced growth rate, we monitored conditional FZR1 RNAi cells in the presence or absence of dox by time-lapse phase-contrast microscopy. As can be seen in Fig. 2D, the generation time increased from 19.6±2.27 hours in control cells to 23.2±4.05 hours in FZR1-knockdown cells treated for 6 days with dox. In addition, in response to FZR1 knockdown, up to 13% of cells failed to divide within 31 hours [= mean generation time (μ) + 2 s.d.], suggesting that they might have withdrawn from the cell cycle. By contrast, in non-induced control cells only 1.9% of cells failed to divide within the observation period.

Because mitotic cyclins, in particular cyclin A, re-accumulated faster in the G1 phase of FZR1 RNAi cells, we hypothesised that FZR1 RNAi might induce premature entry into S phase, which has been suggested to induce replication stress and DNA-damage responses (Bartek et al., 2007), leading to p21^CIP1 induction or cell death. To have a more synchronous population of FZR1 RNAi cells, we transiently transfected U2OS cells with 60 nM control or FZR1 siRNAs for 24 hours before addition of 250 nM noc for 14 hours. Cells were then synchronously released from the mitotic block and stained for BrdU incorporation every hour. This revealed that half of the FZR1 RNAi cells were already BrdU positive at 4 hours after release, whereas it took control cells 7-8 hours to reach the same number of positive nuclei (Fig. 2E). Similarly, when conditional FZR1 U2OS cells were induced with dox and analysed for cell cycle progression by DNA staining and flow cytometry after release from a mitotic block, a faster accumulation of S-phase cells could be observed (Fig. 2F).

Next, we asked whether FZR1 is required for establishing G1 phase in serum-starved cells. Induced and non-induced U2OS FZR1 RNAi cells were cultivated in the absence of growth factors and pulse labelled with BrdU. FZR1-knockdown cells were found to contain a higher number of BrdU-positive cells (51.6%) than controls (27.7%) and serum-restimulated cells entered S phase faster (supplementary material Fig. S2). Thus, FZR1 is required for establishing G1 phase and determines its length.

To analyse whether the faster onset of DNA replication was due to premature CDK activity, cyclin A was immunoprecipitated from control and FZR1 RNAi cells released from mitotic block. Associated kinase activity started to increase at 2 hours after release in FZR1 RNAi but not in control cells (supplementary material Fig. S3), suggesting that precocious activation of cyclin A or CDK2 could trigger the onset of S phase. To test this hypothesis directly, we measured the onset of DNA replication in cell lines expressing a mutant of cyclin A (lacking amino acid residues 47-83, encompassing the D-box) (Geley et al., 2001). When cells induced for expression of cyclin A Δ47-83 were released from a noc block, ten times more BrdU-positive nuclei (30.2%; n=192) could be detected at 7 hours than in non-induced control cells (3.2%; n=94). Thus, premature accumulation of cyclin A is sufficient to trigger DNA replication, which establishes cyclin A as an essential target of FZR1-APC/C in G1 phase, consistent with previous findings using cells in which levels of UBCH10 were reduced by RNAi (Walker et al., 2008).

The above experiments revealed that FZR1 knockdown not only caused a shortening of G1 phase but also a gradual increase in generation time. Because DNA content analysis by flow cytometry did not reveal an increase in G2-M cells in FZR1 RNAi cells (Fig. 2F, supplementary material Fig. S1B, Fig. S4A), we determined the length of S phase in FZR1 RNAi cells. Mock-transfected and FZR1 siRNA-transfected U2OS cells were filmed for 16 hours by time-lapse microscopy and pulse-labelled with BrdU for 1 hour before fixation and staining. Analysis of movies allowed us to correlate the age of individual cells (with respect to their previous division) with the onset of DNA replication. This analysis revealed that in some FZR1 RNAi cells, DNA synthesis was initiated within the first hour after completion of cytokinesis, and more than 50% of the cells were BrdU positive after ∼3 hours. In control cells (expressing FZR1), the first BrdU-positive cells appeared more than 4 hours after cytokinesis and 50% of the cells replicated their DNA only after 8 hours. In FZR1 RNAi cells the number of BrdU-positive nuclei older than 4 hours was high and, as in control cells, only declined in cells that had undergone cytokinesis more than 14 hours earlier. Thus, although DNA replication was initiated earlier, FZR1 RNAi cells spent more time in S phase. An increase in S-phase duration was also seen when asynchronous cells were pulse-labelled with BrdU, which revealed that a higher fraction (56%) of FZR1-knockdown cells were in S phase compared with 45% in controls (n>1000 in each experiment).

The prolongation of S phase and the induction of p21^CIP1 in FZR1 RNAi cells prompted us to investigate whether knockdown of FZR1 activated the DNA-damage response pathway. As can be seen in Fig. 3, U2OS cells γ-irradiated with 8Gy rapidly produced phosphorylated histone H2AX and CHK2, substrates of the ATM/ATR kinases. In addition, p53 exhibited retarded mobility, suggesting activation by phosphorylation. Consistently with p53 activation, the CDK inhibitor p21^CIP1 was strongly induced in γ-irradiated cells, whereas the levels of phospho-histone H3 declined because of inhibition of mitosis. When dox-induced FZR1 RNAi cells were analysed, it became evident that several of the DNA-damage response markers, such as phospho-H2AX, p53 and p21^CIP1, were already expressed, even without γ-irradiation. In contrast to irradiated cells, however, FZR1 RNAi U2OS cells did not suppress phospho-histone H3 and continued to proliferate (see also supplementary material Fig. S1B).

The above data suggested that loss of FZR1 function activated p53 and subsequent induction of p21^CIP1 and cell death, at least in some FZR1 RNAi U2OS cells. To investigate the role of p53, we first analysed the effect of FZR1 RNAi in different human cell lines with wild-type or altered p53 function, including HeLa (p53 inactivated) and Hct116 (colon carcinoma, p53 wild-type) tumour cells, p53 wild-type non-transformed MRC5 fibroblast and RPE1 retinal pigment epithelial cell lines. In contrast to U2OS cells, which mainly delayed progression through S phase upon reduction of FZR1 levels, MRC5 and RPE1 cells responded with a strong G1 arrest after 6 days of dox treatment. Long-term cultured HeLa and Hct116 FZR1 RNAi cells also accumulated in G1 phase, but in addition underwent cell death and displayed a sub-G1 DNA content that was characteristic of apoptosis (supplementary material Fig. S4). In all conditional FZR1 RNAi cell lines, p21^CIP1 and p27^KIP1 were strongly induced, as was phospho-H2AX. Although FZR1 RNAi impaired proliferation in all cellular model systems, the different cellular responses might be due to cellular differences in p53 function or their ability to activate cell cycle checkpoints in response to a premature entry into S phase.

To further investigate the role of p53 in the antiproliferative effect of FZR1 RNAi, we stably expressed a p53-targeting shRNA in conditional FZR1 U2OS cells. Induction of FZR1 RNAi in these cells failed to induce p21^CIP1 (Fig. 3B), demonstrating that p53 was effectively silenced. Although stabilisation of cyclins, or phospho-H2AX induction were not affected (not shown), p53 knockdown could not rescue the proliferation disadvantage of FZR1 RNAi cells (Fig. 3C). Thus, the p53 response is a consequence of the premature entry into, and impaired progression through, S phase, which is caused by knockdown of FZR1.

To determine the role of FZR1 in mammalian cells without the potential ambiguities associated with RNAi approaches, we generated a mouse strain harbouring a conditional Fzr1 allele (Fig. 4A) based on the loxP/Cre recombination system. FZR1 is encoded by 14 exons located in the C1 region of chromosome 10. By using a `recombineering' strategy (Zhang et al., 2002), we generated a gene-targeting vector (Fig. 4A) to introduce loxP-site-specific recombination sequences into intron 2 and intron 8, respectively. Deletion of exon 2 to exon 8 by Cre recombinase removes the translation start codon as well as the coding sequences for the conserved C-box region (Yu, 2007) and the first two WD40 domains of FZR1. Correctly targeted R1 embryonic stem (ES) cell clones were identified by Southern blotting (Fig. 4B), and clones 3A3 and 4F1 used for blastocyst microinjection to obtain chimeric mice with germline transmission. After establishing Fzr1^flNeo/flNeo mice (Fig. 4C), the neomycin resistance cassette was genetically deleted to obtain Fzr1^fl/fl mice, which were found to be healthy and fertile. Deletion of Fzr1 exon 2 to exon 8, by crossing to PGK-Cre transgenic mice, however, confirmed the previously reported embryonic lethality (Garcia-Higuera et al., 2008; Li et al., 2008) and no homozygous Fzr1-deficient mice could be detected among 73 (27 Fzr1^+/+, 46 Fzr1^+/–, 0 Fzr1^–/–) offspring from pairings of heterozygous Fzr1^+/– mice.

In order to determine whether Fzr1 is essential at the cellular level, we obtained mouse embryonic fibroblasts (MEFs) from embryonic day 12 embryos of homozygous Fzr1^flNeo/flNeo mice and immortalised them at passage 6 by transduction with a SV40 large T-expressing retrovirus. After expansion, cells were infected with a Cre-recombinase expressing adenovirus to delete Fzr1 exon 2 to exon 8, which was verified by PCR and Southern blotting. The established Frz1-proficient and Frz1-deficient cell lines are from now on referred to as Fzr1^fl/fl (control) and Fzr1^–/– cells, respectively. In comparison to control cells (infected with a YFP-expressing adenovirus), FZR1 was not detectable by immunoblotting in Fzr1^–/– cells (Fig. 5A). Fzr1^fl/fl and Fzr1^–/– SV40-large-T-immortalised MEFs proliferated with similar kinetics (Fig. 4D), demonstrating that Fzr1 is not essential for proliferation, at least in SV40-large-T-immortalised cells. In addition, by monitoring the length of mitosis in Fzr1^fl/fl and Fzr1^–/– cells, we found that exit from mitosis was not significantly delayed in cells lacking FZR1 (Fig. 4E).

Consistent with the normal length of mitosis, the degradation of mitotic cyclin B1 and PLK1 was not delayed by the absence of FZR1 (Fig. 5A,B). Similar results were obtained in cyclin B1 immunostaining experiments, which revealed that cyclin B1 levels were similar at various stages of mitosis, regardless of the Fzr1 genotype (Fig. 5C,D). Because cytokinesis is highly sensitive to even low amounts of any remaining cyclin B1 (Wolf et al., 2006), we stained cells for AURKB, microtubules and DNA (Fig. 5E) and counted bi-nucleated cells after release from a nocodazole block. As can be seen in Fig. 5F, no increase in binucleated cells could be detected in Fzr1-deficient cells, which lacked APC/C activity, as shown by the stabilisation of AURKB. Thus, our data demonstrate that, as in human RNAi cells, FZR1 is not required for exit from mitosis in immortalised mouse embryonic fibroblasts.

Next, we analysed the stability of APC/C substrates by immunoblotting cell extracts prepared from synchronised cell cultures of control or Fzr1-deficient MEFs. As can be seen in Fig. 6A, FZR1 substrates such as KID (KIF22), AURKB, CDC20 and B99 were stabilised. Similarly to human RNAi cells, PLK1 was only slightly stabilised in the absence of FZR1 and similarly to CDC20, reaccumulated faster in G1 phase. In contrast to human U2OS cells, however, PLK1 levels dropped more sharply during exit from mitosis in murine cells, and this decline was not affected by loss of FZR1 function. Mitotic cyclins were degraded with similar kinetics in wild-type and Fzr1-deficient MEFs (Fig. 5A, Fig. 6B). When APC/C was immunoprecipitated from nocodazole-treated arrested and released cells using a CDC27-specific antibody, the amount of co-immunoprecipitated CDC20 decreased in control cells but was maintained for longer in Fzr1-deficient cells (Fig. 6C, arrows). Thus, in the absence of FZR1, more CDC20 appeared to be bound to the APC/C, possibly explaining why the degradation of late mitotic substrates, such as PLK1, occurred normally in cells lacking Fzr1.

As we observed in human cells depleted of FZR1 by RNAi, mitotic cyclin A and cyclin B reaccumulated earlier during G1 phase in cells of Fzr1-deficient mice. In addition, the oscillation of Skp2 was prevented and Fzr1-deficient cells exhibited a premature decline of p27^Kip1 (Fig. 6B). When control and Fzr1-deficient MEFs were stained for cyclin A, 56% and 74% of cells were positive, respectively, suggesting an increase of Fzr1-deficient cells in S phase. BrdU incorporation in synchronised cells released from a nocodazole block confirmed this, and demonstrated a considerable shortening of G1 phase in Fzr1-deficient MEFs, as well as an increase in the duration of S phase (Fig. 6D). When MEFs were starved of growth factors for 3 days and analysed for S-phase cells, Fzr1-deficient cells contained more BrdU-positive nuclei (45.1%; n=226) than controls (21.8%; n=220), indicating a reduced capacity to arrest in G1 phase. In summary, our data show that FZR1 is not required for mitosis, but for establishing G1 phase and efficient progression through S phase.

APC/C-dependent proteolysis is essential for progression through mitosis. In all organisms studied, CDC20 and core APC/C subunit genes are essential genes and cells lacking APC/C activity arrest at metaphase (Nasmyth, 2002; Wirth et al., 2004). CDC20 is, however, not the only activator of the APC/C and several CDC20-related proteins have been described. One of these conserved proteins is FZR1, which is kept inactive by CDK-dependent phosphorylation but becomes active after the initiation of cyclin degradation during anaphase. Although FZR1 has been implicated in the regulation of mitosis in budding yeast (Schwab et al., 1997) and Drosophila (Raff et al., 2002), it is not essential for exit from mitosis in these organisms.

As cells exit mitosis, proteins such as AURKA, AURKB and PLK1 (Pines, 2006) are degraded in a FZR1-APC/C-dependent manner, but the significance of this degradation is only poorly understood. When cells are blocked at metaphase by inhibition of the proteasome, exit from mitosis can be triggered by the sole inhibition of CDK activity (Potapova et al., 2006). Thus, it was unclear whether degradation of any substrates, except securin and the mitotic cyclins, is required for mitosis and this motivated our study. We have approached this problem by targeting FZR1 in mammalian cells by using RNAi in human cells as well as by genetic deletion in the mouse, and show that FZR1 is not required for completion of mitosis. We have carefully analysed the effect of FZR1 depletion on the degradation of cyclin B1, which needs to be degraded completely for cells to be able to leave mitosis (Wolf et al., 2006). In both human RNAi cells and Fzr1-deficient MEFs, we could not detect an effect of FZR1 depletion on cyclin degradation in mitosis. Our results agree with that of others (Floyd et al., 2008), who have also shown that FZR1 is not essential for mitosis.

Other reports, however, described that loss of FZR1 function delayed exit from mitosis and induced mitotic defects (Engelbert et al., 2008; Garcia-Higuera et al., 2008). We believe that these effects are consequences of the genetic damage caused by FZR1 depletion. In contrast to the lack of an effect on mitosis, loss of FZR1 severely affected G1 phase. The shortening of G1 phase might be caused by the premature accumulation of cyclin A, stabilisation of SKP2 and CDC25A, all of which are known to be involved in the regulation of the G1-S transition (Pagano, 2004). Enforced entry into S phase might adversely affect the mechanics or fidelity of DNA replication and cause replication stress, which has been shown to induce DNA-damage responses (reviewed by Branzei and Foiani, 2009). Thus, the reported increase in binucleated and abnormal mitoses in FZR1-deficient cells (Engelbert et al., 2008; Garcia-Higuera et al., 2008) might be caused by lagging or broken chromosomes that have interfered with the completion of cytokinesis. In addition, a lack of FZR1 might also weaken the DNA-damage and DNA-replication checkpoints by stabilisation of PLK1 (Bassermann et al., 2008), which would further promote genetic instability by facilitating entry into mitosis (Garcia-Higuera et al., 2008).

Fzr1-deficient cells prematurely initiate DNA replication and spend a longer time in S phase. In addition, they fail to arrest in G1 phase after serum starvation, suggesting that G1 phase APC/C activity sets a threshold for growth factor stimulation, as has also been reported in primary cells obtained from Apc2-deficient mice (Wirth et al., 2004). The prolongation of S phase might have trivial causes, such as the lack of sufficient factors required for replication, such as nucleotides, owing to the shortening of a cellular growth period during G1 phase. Shortening of one cell cycle phase is known to cause compensatory lengthening of other phases (Reis and Edgar, 2004). The increase in the duration of S phase is, however, unlikely to be just a compensatory mechanism for the reduction of G1 phase, because all the conditional FZR1 RNAi cell lines displayed signs of severe cellular stress and either slowed down proliferation or even induced cell death.

Shortening of G1 phase and forced entry into S phase might limit the number of functional pre-RCs or might lead to replication fork collapse, as suggested previously (Bartkova et al., 2006; Di Micco et al., 2006). Because neither mitotic cyclins nor geminin oscillations were affected by FZR1 RNAi or Fzr1 deletion, the replication licensing cycle did not seem to have been affected. Levels of CDC6, which is reported to be an APC/C substrate (Mailand and Diffley, 2005), did not oscillate during the cell cycle in the cell systems used in this study. Thus, the re-setting of the cell cycle clock was not obviously abrogated by loss of FZR1 and a short (but perhaps not quite adequate) time window for establishing pre-RCs was generated. It is not currently known whether it is the number, activation or set-up of pre-RCs, or whether it is replication fork movement that is perturbed in cells lacking FZR1 function. The cellular model systems that we have established, however, should be useful in determining whether and how FZR1 affects origin licensing (Sivaprasad et al., 2007) and DNA replication.

In Fzr1-deficient MEFs transduced with SV40-LT, S phase is also prolonged, but cellular proliferation is not affected. By contrast, primary MEFs lacking FZR1 (Garcia-Higuera et al., 2008; Li et al., 2008) proliferate poorly and undergo premature senescence. This suggests that LT interferes with a cellular response, such as activation of p53, which impairs proliferation in the absence of FZR1 function. Depletion of FZR1 by RNAi in human cells also activates a p53 response, which could explain the lengthening of the generation time and permanent withdrawal from the cell cycle, as well as the induction of cell death, possibly explaining the different cellular phenotypes observed in FZR1 RNAi cells. The failure of p53 depletion to restore the proliferative capacity in FZR1 knockdown U2OS cells, however, demonstrates that loss of FZR1 function has a more profound effect on cell cycle progression, most likely by inducing genetic alterations that are incompatible with efficient proliferation.

In summary, by analysing the degradation of a large set of known APC/C substrates in cells lacking FZR1 function, we found that proteolysis was only affected for proteins that are normally degraded in late mitosis or in G1 phase. Failure to degrade these late mitotic substrates, however, did not interfere with completion of mitosis. In contrast to CDC20, there appears to be no essential FZR1 substrate that has to be degraded during each and every cellular division. FZR1, however, controls the abundance of important G1-phase regulators, including SKP2, CDC25A and cyclin A, and is needed for establishing G1 phase and the timing of DNA replication. Thus, the ability of FZR1 to flush remaining mitotic proteins from the cell is less important than its function in controlling their reappearance in the daughter cells, and it will be important to determine this role in vivo using tissue-specific deletion of Fzr1.

Chemical reagents were obtained from Sigma, enzymes from Promega (Mannheim, Germany) and DNA oligonucleotides from MWG Biotech (Ebersberg, Germany), unless stated otherwise. Antibodies used in this study were AURKA (rb, 12875), AURKB (rb, 2354), CDC25A (m, 2357), PLK1 (m, 14210), Securin (m, 3305), p21Cip1 (rb, Ab7960) from Abcam (Cambridge, UK); B99 (rb, from Claudio Schneider, LNCIB, Trieste, Italy), CDC20 (rb, Sah107, from Jan-Michael Peters, Institute for Molecular Pathology, Vienna, Austria), CDC27 (m, AF3.1, from Julian Gannon, CRUK, Clare Hall Laboratories, South Mimms, UK and BD Transduction Labs, Schwechat, Austria), Cyclin A [m, E23 (Julian Gannon), SG20 (raised against the N-terminal 170 amino acids of human cyclin A)], Cyclin B1 (m, V152, Julian Gannon), GAPDH (m, 6C5, HighTest), Geminin (r, 8B1, from Aloys Schepers, Helmholtz Zentrum, Munich, Germany), KID [m, 8C12 and rb CW3, raised against the C-terminal 250 amino acid residues of hKID (Wandke and Geley, 2006)], p27KIP1 (rb, Santa Cruz Biotechnology, Heidelberg, Germany), SKP2 (m, Zymed), UbcH10 (rb, Boston Biochem, Cambridge, MA), phospho-histone H3 (rb, Upstate, Millipore, Vienna), p-Chk2 (rb, PA1-14093, Affinity Bio Reagents, Thermo Scientific), p53 (m, pAb1801, Oncogene Science), p-H2AX (rb, 9718, Cell Signaling, NEB). FZR1 (m, AR38, from Julian Gannon, raised against recombinant FZR1), FZR1 (Rb, raised against an N-terminal peptide, affinity purified). To obtain monoclonal FZR1 antibodies, full-length human FZR1 coding sequence was subcloned into bacterial expression vector pET28a (Novagen, Merck Chemical, Nottingham, UK) and the recombinant C-terminally hexa-histidine-tagged FZR1 expressed in BL21[DE3]pLysS, purified from inclusion bodies under denaturing conditions and used to immunise Balb/c mice as described (Wandke and Geley, 2006). Spleen cells were used for fusion with Sp2 myeloma cells and clones screened for antibodies by Elisa. Clone AR38 was expanded and found suitable for immunoblotting but not for immunostaining or immunoprecipitation. For production of polyclonal rabbit anti-FZR1 antibodies, anti-peptide antibodies (using the N-terminal 18 residues as antigen) were produced by Gramsch Laboratories (Schwabhausen, Germany) and affinity purified using the immunogen coupled to thiopropyl-Sepharose-6 beads (GE-Healthcare, Vienna, Austria).

The inducible lentiviral FZR1 RNAi vectors were generated by LR-clonase (Invitrogen, Lofer, Austria) mediated transfer of a short-hairpin RNA (shRNA) expression cassette targeting human FZR1 (Brummelkamp et al., 2002) from pENTR-THT into the lentiviral vectors pHR-DEST-SFFV-GFP and pHR-DEST-SFFV-PURO (Ploner et al., 2008). For constitutive expression of shRNAs, the H1 promoter from pRETRO-Super (kindly provided by Reuven Agami, The Netherlands Cancer Institute, Amsterdam, The Netherlands) was subcloned into pENTR-1A (Invitrogen) using EcoRI and SalI. A 64bp p53 targeting (underlined sequence) shRNA encoding oligonucleotide (5′-GATCCCCAGTAGATTACCACTGGAGTCttcaagagaGACTCCAGTGGTAATCTACTTTTTTGGAAA-3′) was subcloned using BglII and HindIII and sequenced. Lentiviral constructs for delivery of the shRNA expression cassette were done using GATEWAY technology. Adenoviral constructs were created by shuttling the iCRE- or Venus-encoding DNA fragments from corresponding ENTR vectors into pAD/CMV-DEST (Invitrogen).

U2OS (ATTC:HTB-96), HEK293T, HEK293A (Invitrogen), RPE1 (kindly provided by Erich A. Nigg, Max-Planck-Institute of Biochemistry, Martinsried, Germany), MRC-5 (ECACC, Salisbury, UK) and HCT116 (ATTC:CCL-247) cells were grown in DMEM supplemented with 10% FCS, 100 μg/ml streptomycin and 100 U/ml penicillin in saturated humidity at 37°C, 5% CO₂. For synchronisation, cells were treated for 12-14 hours with 250 nM noc and harvested by mitotic shake-off, washed three times in medium and replated. For BrdU labelling, cells were incubated for 1 hour in 10 μM BrdU. Cell lines (U2OS, HeLa, Hct116, MRC-5, RPE1) expressing TetR (U2OS-TR) were generated by lentiviral transduction using pLENTI6-TR (Invitrogen). MEFs were obtained from fetuses at E12 according to standard protocols. MEFs were grown in 50:50 DMEM Ham's F-12 medium supplemented as above and immortalised by transduction with a SV40-TAg-expressing retrovirus obtained from Ψ2-865 cells. Production of retrovirus, lentivirus and adenovirus was performed under standard Biological Safety 2 conditions as described (Ploner et al., 2008; Wolf et al., 2006). To generate the inducible FZR1 RNAi cell lines, TetR-expressing cell lines were sequentially transduced with two shRNA-expressing lentiviral constructs conferring puromycin resistance (2.5 μg/ml) and GFP expression, respectively. To generate stable p53-knockdown cells, conditional FZR1 U2OS cell were transduced with a lentiviral shRNA vector coexpressing RFP (kindly provided by Christian Ploner, Medical University of Innsbruck, Innsbruck, Austria). Infectious adenoviral particles were generated in 293A cells, purified using Vivapure Adenopack 20 (Sartorius, Goettingen, Germany) and titres determined by quantitative PCR (Ma et al., 2001) and correlation to Venus-positive cells by flow cytometry. For transient RNAi, U2OS cells were transfected at 50% confluency with 60 nM siRNA targeting FZR1 (5′-UGAGAAGUCUCCCAGUCAGTT-3′) or Luciferase (control) for 24 hours using Lipofectamine 2000 (Invitrogen).

Live-cell microscopy was performed on an Axiovert 200M microscope (Carl Zeiss, Jena, Germany) as described (Wolf et al., 2006). Phase-contrast movies were obtained using a 10× PLAN NeoFluar 0.3 NA objective with images taken every 5 minutes. Duration of mitosis was determined by counting the frames between cell rounding and completion of cytokinesis by phase-contrast microscopy. Images and movies were analysed using Metamorph 7.0 (Molecular Devices, Downingtown, PA), converted to eight-bit TIFF format and further processed using Adobe Photoshop CS2 and Adobe Illustrator CS2.

Vector construction was carried out using a ∼14 kb genomic FZR1 fragment cloned from a λ-KO2 library (Zhang et al., 2002) by targeted insertion of a loxP-site-flanked TcR cassette into intron 8. After RS-recombinase-mediated release of the plasmid, the TcR cassette was removed and a second loxP site introduced along with the eukaryotic NeoR selection marker into intron 2. The resulting conditional gene-targeting plasmid pKO2-FZR1-KO-II was partially sequenced, linearised using SalI and used to electroporate R1 ES cells. G418-resistant cells were screened by Southern blotting and clones 3A3 and 4F1 used for blastocyst microinjection. After germline transmission, mice were backcrossed to C57Bl/6. Exon 2 to exon 8 of NeoR and Fzr1 were sequentially deleted using transgenic Flpe and Cre mice, respectively. All mouse breeding was performed according to the requirements of the animal protection act.

Immunoblotting and immunostaining were performed as described (Wolf et al., 2006). BrdU staining on fixed and acid-treated cells was done using the FLUOS in situ proliferation kit (Roche Biochemicals, Vienna, Austria) and nuclei were counterstained with 2 μg/ml propidium iodide (PI). For CDC27 immunoprecipitation, AF3.1 antibodies were bound to Affiprep protein-A beads (Bio-Rad, Vienna, Austria) and crosslinked using 40 mM dimethyl pimedilate in 0.1 M borate (pH 9). Control beads were prepared using 8C12 antibody (Wandke and Geley, 2006). MEFs were treated with 500 nM noc for 12 hours, harvested and released. The first time point was taken immediately after the first washing, whereas the remaining cells were plated at ∼4×10⁶ cells per 10-cm dish and incubated before being scraped off the plate in cold PBS for pelleting and lysis in 300 μl of 20 mM HEPES-KOH pH 7.4, 150 mM NaCl, 0.1% Triton X-100, 1 mM EGTA, PhosphoStop (Roche), Complete (Roche) protease inhibitors and 50 μg/ml PMSF for 60 minutes at 4°C. After centrifugation for 30 minutes, 350 μg protein was mixed with 30 μl equilibrated antibody beads for 1 hour, followed by three washes in lysis buffer, boiling in 20 μl SDS-sample buffer and analysis of 10 μl by immunoblotting.

For immunoprecipitation of cyclin A, conditional FZR1 RNAi U2OS cells were treated for 3 days in the presence or absence of 1 μg/ml dox before addition of 250 nM noc for 12 hours. After mitotic shake-off, cells were replated at 4×10⁶ cell per 10 cm dish and harvested at 2, 4 and 6 hours, whereas the mitotic sample was taken immediately after the first wash. Lysis was carried out in 250 μl lysis buffer (150 mM NaCl, 50 mM Tris-HCl, pH 7.4, 1 mM EGTA, 0.1% NP40, Complete protease inhibitors, PhosphoStop and 50 μg/ml PMSF) for 1 hour on ice, before clearing for 30 minutes by centrifugation at 4°C. Protein (700 μg) was incubated for 45 minutes at 4°C with SG20 antibodies covalently coupled to Affiprep protein-A beads. After three washes in lysis buffer, beads were split for immunoblotting and kinase reactions. For kinase reactions, beads were washed once in kinase buffer (3 mM EGTA, 5 mM NaF, 50 mM β-glycerophosphate, 1 mM DTT and 15 mM magnesium acetate) and incubated with 250 μg/ml histone H1 and 5 μCi [γ-³²P]ATP for 20 minutes at 30°C. Kinase reaction were stopped on dry ice and analysed by denaturing gel electrophoresis and autoradiography. Quantification was carried out by densitometric scanning of exposed films and analysis by ImageJ software (Abramoff et al., 2004).