PIP-box-mediated degradation prohibits re-accumulation of Cdc6 during S phase Free

Cell division control protein 6 (Cdc6) is an essential factor of the pre-replication complex (preRC). Together with Cdt1, Cdc6 loads the mini-chromosome maintenance (MCM) complex, which consists of the six MCM proteins (numbered 2 to 7, hereafter referred to as MCM2–7) on origins of DNA replication. This process is called DNA replication licensing. Once licensed, replication origins can be activated by cyclin–Cdk2 complexes and Cdc7 kinase. DNA replication initiation factors, such as Cdc45 and GINS complex factors, subsequently facilitate the process of DNA unwinding that is carried out by MCM2–7. Replication begins when the homotrimeric DNA-clamp proliferating cell nuclear antigen (PCNA) is loaded onto the chromatin, enabling recruitment of DNA polymerases (Masai et al., 2010). Altogether, this means that the onset of S phase is largely controlled by regulatory events that take place during the preceding mitosis and G1 phase.

Early in mitosis, cells start preparing for DNA replication licensing: Cdc6 translocates to condensed chromosomes, and the levels of Cdt1 greatly increase owing to formation of a stable complex with its inhibitor geminin (Ballabeni et al., 2004; Clijsters et al., 2013). When cells satisfy the spindle checkpoint and exit mitosis, cyclin B1 is degraded through the E3 ligase anaphase-promoting complex/cyclosome containing Cdc20 (APC/C–Cdc20), which inactivates Cdk1, a key licensing inhibitor (Pines, 2011). Geminin, a second key licensing inhibitor in mammals, is also degraded by APC/C–Cdc20 when the spindle checkpoint is satisfied (Clijsters et al., 2013). This means that, in proliferating cells, the competence to license DNA replication greatly increases at the metaphase-to-anaphase transition. Immediately after anaphase, however, Cdc6 is degraded through APC/C containing Cdh1 (also known as FZR1 in human). This schedule of events suggests that, in rapidly cycling cells, there is a short window of time at the end of mitosis when licensing competence peaks (Clijsters et al., 2013).

Cdt1, after its release from geminin, remains present during mitotic exit and G1 phase but is lost when PCNA is loaded onto DNA (Clijsters et al., 2013). Cdt1 degradation requires a so-called PIP-box motif that is typically recognized by the E3 ligase cullin ring (CRL)4 complex containing Cdt2 (CRL4–Cdt2). This Cdt1 degradation motif is only functional when CRL4–Cdt2 interacts with the DNA-loaded form of PCNA (Havens and Walter, 2009; Havens et al., 2012). Thereby, Cdt1 clearance coincides robustly with the start of S phase, reducing the chance that any newly formed DNA is re-licensed and subsequently re-replicated. Another Cdt1 destruction mechanism might rely on phosphorylation and subsequent destruction via the E3 ligase SCF–Skp2 (Liu et al., 2004; Nishitani et al., 2006). The risk of DNA re-replication is also reduced by the degradation of Cdc6 during mitotic exit. However, by the end of G1 phase, the APC/C switches off and Cdc6 is newly synthesized (Bassermann et al., 2014; Duursma and Agami, 2005; Hsu et al., 2002; Piatti et al., 1996; Rape and Kirschner, 2004). In fact, when quiescent cells re-enter the cell cycle, the first opportunity to license DNA replication would arise at this point, when newly synthesized Cdc6 is protected from degradation by phosphorylation, even though APC/C–Cdh1 is highly active in quiescent cells. (Mailand and Diffley, 2005). By contrast, we found that rapidly proliferating cells have, probably, already licensed their origins during mitotic exit, before APC/C–Cdh1 is activated (Clijsters et al., 2013). In these cells, expression of Cdc6 at the end of G1, when Cdt1 levels are still high, could pose a threat to genomic integrity. Here, we investigated how cycling cells control Cdc6 and prohibit re-licensing as they enter, and progress through, S phase.

Cdc6 degradation starts in anaphase in a manner strictly dependent on APC/C–Cdh1 (Fig. 1A) (Clijsters et al., 2013). Cdc6 is absent from early G1 phase onwards and, in most cells, does not re-accumulate until cells are at least several hours into S phase. Remarkably, this means that in cycling cells, Cdc6 appears later in the cell cycle than other typical APC/C–Cdh1 substrates, such as Aurora A (Clijsters et al., 2013). Here, we investigated whether additional destruction mechanisms might exist that can destabilize Cdc6 from the G1-S transition onwards, when APC/C–Cdh1 is not active.

Firstly, we followed fluorescent Cdc6–Venus by time-lapse fluorescence microscopy in cells treated with siRNA. Cdc6–Venus remained largely stable during and after mitosis, as well as in G1 phase. However, Cdc6 was rapidly degraded several hours later (Fig. 1B,C). In Cdh1 siRNA-treated cells, fluorescent Cdc6 levels declined specifically from within the nucleus (Fig. 1B, 90 and 120 minutes). To further investigate the apparent existence of an APC/C–Cdh1-independent Cdc6 degradation pathway, we used a destruction box mutant of Cdc6, named KEN-Cdc6–Venus, in which Lys81, Glu82 and Asn83 were mutated to Ala. Indeed, KEN-Cdc6–Venus was stable at anaphase but suddenly disappeared later in G1 phase, again, with fluorescence declining specifically in the nucleus (Fig. 1D, 120 minutes). KEN-Cdc6–Venus degradation started in a switch-like fashion (Fig. 1E). A suggested D-box recognition motif comprising Arg56 and Leu59 did not contribute to the degradation pattern of Cdc6 in our live-cell degradation assay (Petersen et al., 2000) (supplementary material Fig. S1). We hypothesize from these experiments that a previously unknown APC/C-independent Cdc6 destruction mechanism is initiated by an event scheduled at a time around the end of G1 phase.

To investigate more precisely when the APC/C-independent Cdc6 destruction pathway is activated, Cdc6–Venus was filmed together with a fluorescent marker of S phase, Cherry-tagged PCNA. In most control cells, Cdc6–Venus was degraded substantially before Cherry–PCNA re-distributed towards discrete nuclear foci (dots), which represent DNA replication factories (Fig. 2A) (Burgess et al., 2012). However, in cells depleted of Cdh1, Cdc6–Venus remained stable until the end of G1 phase. Following the end of G1 phase, Cdc6 levels began to diminish, a process which started in the nucleus, while at the same time, Cherry–PCNA dots formed (Fig. 2A; an example is shown in Fig. 2B, compare 60 minutes to 90 minutes). In the presence of Cdh1, KEN-Cdc6–Venus also started to be degraded when Cherry–PCNA dots appeared (Fig. 2A). This suggested that the APC/C–Cdh1-independent pathway that destabilizes Cdc6 is linked to the timing of PCNA loading. Decrease of Cdc6 levels was not related to a reduction in protein synthesis as treatment of cells with proteasome inhibitor MG-132 led to a rapid accumulation of Cdc6, regardless of whether cells were in G1 or S phase. The increase of Cdc6 in MG-132-treated G1 or S phase cells was reversed by co-treatment with the translation inhibitor cycloheximide (Fig. 2C). Furthermore, fluorescently tagged Cdt1, or Cdc6, accumulated exclusively in the nucleus of S phase cells upon proteasome inhibition (Fig. 2D–F). Therefore, we conclude that from the G1-S transition onwards, and throughout S phase, Cdc6 levels are kept low by a previously unknown degradation pathway that is initiated in the nucleus.

The licensing factor Cdt1 is degraded in S phase by a PCNA-dependent mechanism involving targeted destruction by CRL4–Cdt2 (Havens and Walter, 2009; Havens et al., 2012). Generally, CRL4s consist of cullin 4 and the adaptor DNA damage binding protein 1 (DDB1), binding different substrate receptors called DDB1-CUL4-associated factors (DCAFs), and a ring protein, ROC1 or ROC2, interacting with an E2 enzyme. Degradation of Cdt1 requires binding of an N-terminal PIP-box to DNA-loaded PCNA and recognition of adjacent positively charged residues by DCAF Cdt2 (Havens et al., 2012). Destruction of Cdt1–Venus in live cells starts at the G1-S transition and coincides with the appearance of PCNA dots (Clijsters et al., 2013), similar to the degradation profile of Cdc6–Venus when APC/C–Cdh1 is blocked (Fig. 1C). To test the potential involvement of CRLs in the Cdc6-degradation pathway that we found, cells were synchronized in S phase and treated with the NEDD8-activating-enzyme inhibitor MLN-4924, which inhibits activity of CRLs (Lin et al., 2010). Cdc6 levels clearly increased upon the addition of MLN-4924; however, Cdc6 accumulated to a lesser extent than Cdt1 (Fig. 3A). Furthermore, degradation of KEN-Cdc6–Venus at the G1-S transition was strongly impaired by MLN-4924 treatment (Fig. 3B,C). Therefore, we conclude that degradation of Cdc6, starting at the G1-S transition and continuing throughout S phase, occurs via a CRL.

The DCAF Cdt2 is a well-described inhibitor of over-replication. Cdt2 has been suggested to perform this function mainly through destabilizing Cdt1 in S phase; thus preventing re-licensing of origins (Jin et al., 2006). We hypothesized that Cdc6 is also a target of CRL4–Cdt2. To test this, we investigated the requirement of DCAF Cdt2 for the degradation of Cdc6 by measuring changes in Cdc6 levels in cells that were depleted of Cdt2 by siRNA. The levels of Cdc6 increased substantially after depletion of Cdt2, independently of the levels of Cdt1 (Fig. 4A). As mentioned, depletion of Cdt2 could affect the cell cycle. Indeed, analysis of the cell cycle distribution of the samples in Fig. 4A by flow cytometry (Fig. 4B) revealed that Cdt2 depletion induces re-replication. Interestingly, the re-replication induced by the depletion of Cdt2 is efficiently reversed by co-depletion of either Cdc6 or Cdt1 (Fig. 4B). This further indicated that Cdc6, like Cdt1, is an important target of CRL4–Cdt2. To investigate whether the observed stabilization of Cdc6 occurred independently of the change in cell cycle distribution, we tried to follow the stability of Cdc6 in S-phase-synchronized cells after depletion of Cdt2 (Fig. 4C; see supplementary material Fig. S2 for cell cycle profiles). Because Cdc6 levels are normally kept low during S phase, for this we had to pre-treat these synchronized cells with the proteasome inhibitor MG-132, enabling Cdc6 re-accumulation. The addition of MG-132 caused an increase in the levels of Cdt1 (as a positive control), as well as Cdc6. After MG-132 wash-out, in cells treated with Cdt2 siRNA, the endogenous protein levels of both Cdt1 and Cdc6 remained elevated, as compared with controls (Fig. 4C); the quantification in Fig. 4D confirms that, in S phase, both Cdt1 and Cdc6 are degraded in a Cdt2-dependent manner.

In this artificial situation, after MG-132 treatment of S-phase-synchronized cells, effects on Cdc6 stability could be underestimated. Therefore, we tested the role of Cdt2 in otherwise unperturbed single cells that co-express Cdc6–Venus and Cherry–PCNA by time-lapse fluorescence microscopy. In control cells, Cdc6 disappeared at mitotic exit and did not re-accumulate until cells reached the end of S phase, when Cdc6 was exclusively detected in the cytoplasm (Fig. 1B). In some cells, probably cells that synthesized relatively high amounts of Cdc6–Venus, fluorescence re-accumulated in the nucleus before the end of G1 phase, but this signal rapidly disappeared as cells progressed to S phase (Fig. 4E, top panels; see also the scheme presented in Fig. 7). However, in cells depleted of Cdt2, Cdc6–Venus was first degraded normally, at mitotic exit, but then slowly accumulated in the nucleus from the G1-S transition onwards. We conclude that this expression pattern reflects the switch-off of APC/C–Cdh1 at the end of G1 phase, in combination with the inability of these cells to activate the S phase degradation pathway (Fig. 4E, lower panels, Fig. 4F; and supplementary material Fig. S3A). Additionally, in cells expressing Cdc6–Venus, Cdt2 and Cdc6 interacted, as detected by an immunoprecipitation assay of endogenous Cdt2 in S-phase-synchronized cells (supplementary material Fig. S4A). These results strongly indicate that Cdc6, like Cdt1, is an important target of a Cdt2-driven CRL from the G1-S transition onwards. Furthermore, we conclude that S phase degradation of Cdc6, like that of Cdt1, contributes to the inhibitory effects of the E3 ligase CRL4–Cdt2 on DNA over-replication.

To clarify this new degradation pathway in more detail, we followed several fluorescently-tagged Cdc6 mutants in control cells. Cdt2-dependent degradation typically requires recognition of a PCNA-interacting motif (PIP-box) within the substrate. In the evolutionary-conserved N-terminal region of Cdc6, we discovered a potential PIP-box-like motif, which includes a QxxI core-element (Fig. 5A,B) (Moldovan et al., 2007). This PIP-box-like motif is C-terminally flanked by hydrophobic and positively charged residues, which have been suggested to contribute to recognition by CRL4–Cdt2 (Kim et al., 2010; Michishita et al., 2011; Moldovan et al., 2007). Next, two Cdc6–Venus constructs, mutated in the potential PIP-box element, were followed by time-lapse microscopy. A Gln9, Ile12 and Phe14 to Ala mutant in the core PIP-box sequence, QIF-Cdc6–Venus, was degraded normally by the APC/C–Cdh1 pathway at mitosis exit (Fig. 5D, curve denoted by triangles; supplementary material Fig. S3B). Importantly however, a Cdc6 construct with point mutations in both the PIP-box and the KEN-box, QIF-KEN-Cdc6–Venus, remained fully stable throughout the entire cell cycle (Fig. 5C, compare 120 and 240 minutes; and Fig. 5D, curve denoted by diamonds; Fig. 5E).

We reasoned that this Cdt2-recognition motif prohibits the re-accumulation of Cdc6 at the G1-S transition and during S phase. To test this idea, we first investigated whether the motif resembling the PIP-box of Cdc6 could interact with PCNA in vitro. Therefore, we used a fluorescence-polarization assay that could reveal direct binding between purified peptides of either the wild-type or Ala-mutated (at Gln9, Ile12, Phe14) N-terminal regions of PCNA or Cdc6. We found that the wild-type Cdc6 peptide bound to PCNA; however, binding of PCNA to the PIP-box-mutant versions of Cdc6 was reduced to an undetectable level (supplementary material Fig. S4B,C). Even though wild-type Cdc6 interacted with PCNA with lower affinity than was described previously for a Cdt1 peptide, the specific interaction was found consistently and was independent of the lengths of the peptides used. Therefore, these results support the idea that the N-terminal PIP-box of Cdc6 functions as a previously unrecognized PCNA-directed degradation motif. Next, we aimed to confirm this concept in live cells as they transit the G1-S border and progress through S phase. Indeed, wild-type Cdc6–Venus appeared only when cells reached the end of S phase whereas re-accumulation of QIF-Cdc6–Venus occurred much earlier. This earlier occurrence coincided with the appearance of PCNA dots at the G1-S transition (Fig. 6A,B, see also the scheme presented in Fig. 7). The timing of QIF-Cdc6–Venus re-accumulation was very similar to that observed for wild-type Cdc6 in cells depleted of Cdt2 (Fig. 4E). Apparently, the PIP-box of Cdc6, like the PIP-box of Cdt1, is recognized by a Cdt2-directed CRL and PCNA, targeting Cdc6 for proteasome-dependent degradation from the start of S phase. Without this motif, inhibition of the APC/C at the end of G1 phase would lead to the re-accumulation of Cdc6, because, at least in the cell lines that we investigated, Cdc6 is rapidly synthesized at the end of G1 phase and in S phase. High Cdc6 expression after DNA licensing but before DNA replication would increase the risk of re-licensing and re-replicating DNA as the first forks fire, before the entire pool of cellular Cdt1 is removed. Concomitant PIP-box-mediated degradation of Cdt1 and Cdc6 ensures a much more efficient clearance of licensing factors at the G1-S border, when replication begins, and throughout S phase.

QIF-KEN-Cdc6–Venus remained stable during the cell cycle (Fig. 5E). Interestingly, it showed a rapid nuclear to cytoplasmic translocation that suddenly started around the transition from S phase to G2 phase (Fig. 6C,D; roughly 7.5 hours from the appearance of Cherry–PCNA dots: 446.6 minutes±60.6; n = 8; mean±s.d.). This suggests that, apart from a role for Cdk2 in exporting Cdc6 to the cytoplasm from early S phase onwards (Jiang et al., 1999; Petersen et al., 1999), slow accumulation of Cdk1 activity from the start of G2 phase onwards might also be a significant factor in triggering the export of Cdc6. In part, we conclude this because cyclin-E–Cdk2 is already active much earlier in the cell cycle, in G1 phase, and because cyclin-A–Cdk2 accumulates long before the start of S phase (Lukas et al., 1999). In support of a possible role for Cdk1, cytoplasmic translocation of nondegradable Cdc6 in late S phase was completely prevented by addition of the highly selective Cdk1 inhibitor RO-3306 (Fig. 6E). Confirming other reports, we found that RO-3306, at concentrations up to 9 µM in cell culture medium, inhibits Cdk1 substantially but does not block Cdk2 (Ma et al., 2009; Spencer et al., 2013; Vassilev et al., 2006) (data not shown). In further agreement with the assumption that RO-3306 does not fully abrogate the function of Cdk2, we found that U2OS cells treated with 5 µM RO-3306 for 24 hours progressed through S phase relatively normally, but then arrested in G2 phase, when Cdk1 is otherwise activated to allow mitotic entry. After prolonged incubation with 5 µM RO-3306, the cells arrested in G2 phase started several rounds of re-replication to multiply their DNA, a typical consequence of Cdk1 inhibition. By contrast, cells treated with a higher dose of the drug, 10 µM RO-3306, became delayed in S phase, possibly as a result of partial inhibition of Cdk2 (supplementary material Fig. S5A). Although our results do not formally rule out a role for Cdk2, on the basis of our experiments, we favor the conclusion that the accumulation of Cdk1 activity catalyzes nuclear export of Cdc6 as cells progress from S phase to G2 phase.

Altogether, we conclude that PIP-box-dependent destabilization of Cdc6 provides a mechanism to prevent the re-appearance of nuclear Cdc6 at S phase. When the APC/C is inactive, Cdt2 reduces levels of Cdc6 from the G1-S transition onwards, whereas, in late S phase and during G2 phase, Cdc6 is exported to the cytoplasm by the accumulating activity of Cdk1, possibly resulting from the synthesis of cyclin B1. By these distinct, yet subsequent, pathways, replication-licensing remains prohibited between late G1 and the start of mitosis (Fig. 7) (Hsu et al., 2002; Lukas et al., 1999).

The work described here supports the concept that, in proliferating cells, licensing competence peaks in mitosis, before rapid destruction of Cdc6 is catalyzed by APC/C–Cdh1 activation (Clijsters et al., 2013; Petersen et al., 2000). However, Cdc6 is again expressed before S phase when the APC/C–Cdh1 inhibitor Emi1 is synthesized (Hsu et al., 2002; Lukas et al., 1999; Piatti et al., 1996; Pines, 2006). Here, we describe a new degradation mechanism for Cdc6 that is switched on specifically at the G1-S transition. This pathway depends on a previously unidentified PIP-box-like motif within the evolutionarily conserved N-terminus of Cdc6. The pathway we describe requires Cdt2, which typically cooperates with CRL4, and is different from a role of CUL4B, which regulates Cdc6 by modulating Cdk2 levels under specific conditions in G1 phase (Zou et al., 2013).

This sequence of events might be crucially different in cells that start the cell cycle as they emerge from quiescence and a wave of cyclin-E–Cdk2 activity protects Cdc6 from being recognized by APC/C–Cdh1. Because we observed that Cdc6 is both synthesized and degraded during G1 phase, it is possible that a strong burst of growth-stimulus-induced Cdc6 synthesis might help to raise the level of Cdc6 above a crucial threshold in order to initiate replication licensing, even when Cdc6 is not fully stabilized (Spencer et al., 2013). Rapid synthesis of Cdc6 might be another particular feature of cells entering the cell cycle from quiescence. In cycling cells that coexpress Cdc6 and Cdt1 in mitosis, licensing is independent of cyclin E and Cdk2. However, this implies that Cdc6, when synthesized, re-accumulates at the end of G1 and thus could be available to trigger licensing when S phase starts, before Cdt1 is fully degraded. Here, we show that cells simultaneously reduce the levels of Cdt1 and Cdc6 by an overlapping degradation pathway immediately before the start of S phase. This will greatly reduce the risk that cells re-license any newly formed DNA.

Clearly, Cdc6 levels must be regulated to prevent untimely DNA replication and safeguard genomic stability. In Saccharomyces cerevisiae, the cdc6-3 leucine zipper mutant shows strong re-initiation of DNA replication (Liang and Stillman, 1997). In addition, overexpression of CDC6 phosphorylation mutants strongly induces DNA re-replication (Mimura et al., 2004). Similarly, in Schizosaccharomyces pombe, overexpression of a phosphorylation mutant of the Cdc6 ortholog CDC18 causes over-replication (Jallepalli et al., 1997). However, we have not yet determined to what extent abnormal stabilization of Cdc6 causes DNA re-replication in human cells (see supplementary material Fig. S5B). This could be related to cell cycle control of the intracellular localization of Cdc6, inactivation of a non-degradable form of Cdc6 or the existence of additional Cdc6 inactivation mechanisms. Because small defects in the control of replication lead to the accumulation of DNA mutations, detrimental effects of mutating the PIP-box-like motif in Cdc6 are best studied by detailed analyses of alterations to DNA within the context of a developing organism. Therefore, proper insight into the exact role of this precisely timed Cdc6 degradation control pathway awaits more elaborate follow-up investigations.

U2OS cells were cultured in Dulbecco's modified Eagle's medium (Gibco) with 8% fetal calf serum and antibiotics. Cell synchronization, western blotting, flow cytometry and time-lapse imaging were as described previously (Clijsters et al., 2013). Drugs used were: Nocodazole (Sigma, 830 nM), Cdk1 inhibitor RO-3306 (#217699, Calbiochem, concentration as indicated), proteasome inhibitor MG-132 (#13697, Cayman Chemicals, 5 µM), translation inhibitor cycloheximide (#C6255, Sigma, 10 µM) and neddylation inhibitor MLN-4924 (Millennium Pharmaceuticals, 1 µM).

The Cdc6–Venus and Cdt1–Venus constructs have been described elsewhere (Clijsters et al., 2013). The R54,L59A-Cdc6–GFP (supplementary material Fig. S1) construct has been described previously (Duursma and Agami, 2005). The KEN-Cdc6–Venus construct was synthesized by PCR amplification of KEN-Cdc6 from KEN-Cdc6-GFP-N1 (Duursma and Agami, 2005) and was cloned into the pVenus-N1 vector (NheI/XhoI). The QIF-Cdc6–Venus and QIF-KEN-Cdc6–Venus constructs were synthesized by PCR amplification with the primer FWD#1: 5′-ATATCTCGAGACCATGCCTCAAACCCGATCCCAGGCAGCGGCTACAGCCAGTGCTCCAAAAAGGAAGCCCGGGCCCCCCCGCATCGCGCCGCCCAAGCTG-3′ (mutated codons are underlined) and cloned into pVenus-N1 (XhoI/BamHI). The Δ2-18-Cdc6–Venus construct (supplementary material Fig. S3B) was synthesized by PCR amplification with the primer FWD 5′-ATATCTCGAGACCATGCTGTCTCGGGCATTGAACAAAGC-3′ and was cloned into the vector pVenus-N1 (XhoI/BamHI).

The viral plasmid pLIB-Cherry-PCNA comprises Cherry–PCNA cloned into pLIB-C1 (AgeI/BamHI). pLIB-Cdc6-Venus has been previously described (Clijsters et al., 2013). pLIB-KEN-Cdc6-Venus was constructed by PCR amplification of KEN-Cdc6 from KEN-Cdc6-GFP-N1 (Duursma and Agami, 2005) and cloning into pLIB-Venus-N1 (XhoI/BamHI). The viral plasmids pLIB-QIF-Cdc6-Venus and pLIB-QIF-KEN-Cdc6-Venus were constructed by PCR amplification (primer FWD#1) and were cloned into pLIB-Venus-N1 (XhoI/BamHI). The viral plasmids pLIB-Δ2-18-Cdc6-Venus and pLIB-Δ2-18-KEN-Cdc6-Venus were cloned into pLIB-N1 (XhoI/NotI).

The pSuper plasmids pS-Cdc6#5 and pS-Cdc6#6 both target the 3′UTR of Cdc6 and have the target-sequences Cdc6#5: 5′-GCTGGCATTTAGAGAGCTA-3′; Cdc6#6: 5′-GCCAATGTGCTTGCAAGTG-3′. The sequences of all plasmids were verified. The siRNAs to target Cdc6, Cdt1, Geminin (GMNN), Cdh1 (FZR1) and Cdt2 (DTL) were obtained from Dharmacon as set of four individual ON-TARGET-plus oligonucleotides.

Cells in 9-cm Falcon dishes were transfected with 0.5-2.0 µg expression plasmid or 20 µg pSuper plasmids using the standard calcium-phosphate transfection protocol. Cells were co-transfected with 1 ug pBABE-Puro and selected with puromycin for 24 hours as indicated, or with 40 nM siRNA oligonucleotides using Lipofectamine 2000 (Invitrogen) according to manufacturer's protocol. Ecotropic Phoenix cells were transfected with viral plasmids using the standard calcium-phosphate transfection protocol.

For immunoprecipitations, lysates were incubated with 2 µg of an antibody against Cdt2 and 20 µg protein-G–Sepharose beads (Amersham Biosciences) for 16 hours at 4°C, while tumbling. After incubation, beads were washed four times with ice-cold ELB+ (van Zon et al., 2010) and dissolved in 50 µl sample buffer. The proteins were separated by SDS-PAGE and blotted on nitrocellulose membranes. For membrane blocking and antibody incubation, 4% ELK (van Zon et al., 2010) was used.

The antibodies against the following proteins were used at the indicated dilutions: CDC27 (#610455, mouse, 1∶1000; BD Transduction), Cdc6 (180.2, sc-9964, mouse, 1∶1000; Santa Cruz), Aurora A/AIK (#3092, rabbit, 1∶1000; Cell Signaling), geminin (FL-209, rabbit, 1∶1000; Santa Cruz), Actin (I-19, sc-1616, goat, 1∶1000; Santa Cruz), Cdt1 (ab70829, rabbit, 1∶1000; Abcam), DTL/CDT2 (#A300-948A, rabbit, 1∶1000; Bethyl Laboratories), BrdU (M0744, clone Bu20a, mouse, 1∶40; DAKO). Secondary peroxidase- or FITC-conjugated antibodies were obtained from DAKO, or as indicated (ab49626, 1∶1000; Abcam).

We are grateful to Tassos Perrakis, Tatjana Heidebrecht and Rick Hibbert for designing and performing the in vitro PCNA–Cdc6 interaction experiments. We thank Henk Hilkmann and Dris el Atmioui for synthesizing the Cdc6 peptides. We thank Erik Voets and Michiel Boekhout for critically reading the manuscript.