Initiation of Human DNA Replication in Vitro Using Nuclei from Cells Arrested at an Initiation-competent State*

Initiation of human DNA replication is investigated in vitro , using initiation-competent nuclei isolated from cells arrested in late G 1 phase by a 24-h treatment with 0.5 m M mimosine (Krude, T. (1999) Exp. Cell Res . 247, 148–159). Nuclei isolated from mimosine-arrested HeLa cells initiate semiconservative DNA replication upon incubation in cytosolic extracts from proliferating human cells. Initiation occurs in the absence and presence of a nuclear membrane. The cyclin-dependent kinase (Cdk) inhibitors roscovitine and olomoucine inhibit initiation of DNA replication, indicating a dependence of initiation on Cdk activity. Cell fractionation shows that cyclins A, E, and Cdk2 are bound to nuclei from mimosine-arrested cells. Exogenously added human cyclin A z Cdk2 and cyclin E z Cdk2 complexes, but not cyclin B1/Cdk1 or cyclin D2/Cdk6, can overcome inhibition of initiation by roscovitine in vitro . Depleting Cdk2 from cytosolic extract does not prevent initiation, demonstrating that cyclin z Cdk2 complexes are not required in the soluble extract, but are provided by the nuclei. Initiation depends further on an essential and soluble activity present in cytosolic extracts from proliferating cells, but not from mimosine-arrested cells, acting together with nuclear cyclin/Cdk2 activity. The of the G 1 to S phase tran-sition cell division in Once and after and subsequently analyzed by Western blotting using monoclonal antibody HE12. Depletion of Cdk1 and from Cytosolic Extract— Sepharose beads coupled to human p9 Cks1 protein (37) (a gift of M. Jackman, Wellcome/ CRC Institute, Cambridge) equilibrated and washed three times in replication buffer and concentrated by gravity sedimentation. Depletion of interphase cytosol was achieved by three successive rounds of (i) adding a fifth volume of the p9 Cks1 beads, (ii) incubating the slurry for 20 min at 4 °C in a rotator, and (iii) removal of the beads by pelleting at 13,000 rpm in an Eppendorf 5415C centrifuge for 5 min. Mock deple-tions were performed in parallel in the absence of p9 Cks1 -Sepharose.

Initiation of human DNA replication is investigated in vitro, using initiation-competent nuclei isolated from cells arrested in late G 1 phase by a 24-h treatment with 0.5 mM mimosine (Krude, T. (1999) Exp. Cell Res. 247, 148 -159). Nuclei isolated from mimosine-arrested HeLa cells initiate semiconservative DNA replication upon incubation in cytosolic extracts from proliferating human cells. Initiation occurs in the absence and presence of a nuclear membrane. The cyclin-dependent kinase (Cdk) inhibitors roscovitine and olomoucine inhibit initiation of DNA replication, indicating a dependence of initiation on Cdk activity. Cell fractionation shows that cyclins A, E, and Cdk2 are bound to nuclei from mimosinearrested cells. Exogenously added human cyclin A⅐Cdk2 and cyclin E⅐Cdk2 complexes, but not cyclin B1/Cdk1 or cyclin D2/Cdk6, can overcome inhibition of initiation by roscovitine in vitro. Depleting Cdk2 from cytosolic extract does not prevent initiation, demonstrating that cyclin⅐Cdk2 complexes are not required in the soluble extract, but are provided by the nuclei. Initiation depends further on an essential and soluble activity present in cytosolic extracts from proliferating cells, but not from mimosine-arrested cells, acting together with nuclear cyclin/Cdk2 activity.
The initiation of DNA replication at the G 1 to S phase transition is a key regulatory step of the cell division cycle in eukaryotic cells. Once DNA replication has initiated, control mechanisms ensure that the entire genomic DNA is replicated precisely once, and after completion, one replicated genome segregates to each of the two daughter cells during mitosis (for reviews, see Refs. [1][2][3][4][5][6]. Cell fusion experiments in mammalian somatic cells established that G 1 , but not G 2 phase nuclei, initiate DNA replication prematurely when exposed to an S phase cytosolic environment (7). Key regulators of initiation were identified in genetic and cytological experiments in vivo. Roles for cyclin-dependent protein kinases (Cdks) 1 and their G 1 and S phasespecific regulatory subunits cyclin D, E, and A in inducing DNA replication have been documented (8 -16). The analysis of initiation of DNA replication in vivo was recently complemented by a biochemical approach through the establishment of cellfree systems from human and mammalian cells (17)(18)(19). DNA replication is initiated in nuclei isolated from human G 1 , but not G 2 phase cells when incubated in S phase cytosolic extract and S phase-specific nuclear factors. A nuclear extract could be substituted by purified recombinant cyclins A and E, complexed to their kinase partner Cdk2, to initiate DNA replication, directly demonstrating functional roles for these nuclear cyclin⅐Cdk complexes (17). These cyclin/Cdks were essential, but not sufficient, as nuclei also required soluble factors present in a cytosolic extract from S phase cells to initiate replication (17).
For an assembly of replication-competent chromatin in eukaryotic G 1 phase nuclei, an evolutionarily conserved series of molecular events is required involving the origin-recognition complex (ORC), Cdc6 protein, and the mini-chromosome maintenance (MCM) proteins (reviewed in Refs. 2-4, 20, and 21). In mammalian cell-free systems, competence to initiate DNA replication in S phase cytosol is observed when template nuclei are prepared from late G 1 phase cells after release from a block either in mitosis (17) or quiescence (18), followed by transit through early G 1 phase in vivo. Upon release from quiescence, competence of G 1 phase nuclei to initiate in vitro coincides with maximum expression of Cdc6 protein, and addition of recombinant Cdc6 protein advances the onset of replication competence in vitro (18). However, human cells undergoing mitotic proliferation contain Cdc6 protein at all stages of the cell cycle (22)(23)(24) and the state of DNA replication competence therefore depends on other factors as well.
Late G 1 phase nuclei from cells synchronized by release from either mitosis or quiescence are relatively undefined heterogeneous and dynamic populations as a result of cells passing the state of competence at the time of preparation (17,18). A significant step toward molecular and temporal dissection of the establishment of DNA replication forks in human somatic cells would therefore be the availability of defined populations of homogeneous template nuclei reversibly arrested in a state of replication initiation competence. A recently established cellfree system for the initiation of nuclear DNA replication in the yeast Saccharomyces cerevisiae made use of cells reversibly synchronized at defined points in the cell division cycle (25). Template nuclei from yeast mutants blocked at defined points in late G 1 phase initiated DNA replication at high percentages upon incubation in S phase nuclear extracts (25). Because such cell cycle mutants are currently not available as synchronization tools for human cells, I sought to achieve analogous arrest of human cells in late G 1 phase by chemical synchronization.
The plant amino acid mimosine is a versatile agent for blocking DNA replication in proliferating eukaryotic somatic cells but the literature on its mode of action is controversial. Depending on cell type and concentration of mimosine, evidence for blocking both initiation and elongation steps of DNA replication has been reported (26 -34). In the case of human cultured cells, however, this controversy has been clarified because the different effects of mimosine depend on its concentration in the cell culture medium during treatment (34). At concentrations below 0.5 mM, mimosine interferes with elongation steps of DNA replication, and treatment results in populations of cells enriched in S phase after establishment of replication forks (34). In contrast, at concentrations of 0.5 mM and above, mimosine additionally prevents entry into S phase, resulting in a population of cells arrested in late G 1 phase, before establishment of active DNA replication forks in vivo (34). Importantly, the late G 1 phase arrest in vivo is fully reversible and cells enter S phase upon removal of mimosine (27,34).
In this paper, I characterize the ability of nuclei from mimosine-arrested human cells to initiate DNA replication in human cell extracts. Nuclei prepared from these cells efficiently initiate semiconservative DNA replication upon incubation in cytosolic extracts from proliferating cells, even in the absence of a nuclear membrane. Initiation depends on a cytosolic extract from proliferating cells and, furthermore, on cyclin E/Cdk2 and/or cyclin A/Cdk2 activity which are present in the nuclei from mimosine-arrested cells. The data suggest that the competence of nuclei from mimosine-arrested cells to initiate replication is characterized by the presence of the cell cycleregulated cyclin/Cdk2 proteins in cis and initiation is triggered by a different activity in trans.

MATERIALS AND METHODS
Cell Culture and Synchronization-HeLa-S3 cells were cultured as monolayers and synchronized in G 1 , S, and G 2 phase exactly as described (17). Synchronization in mitosis was performed as described (35). Cells were arrested in late G 1 phase by adding 0.5 mM mimosine (Sigma) from a 10 mM stock solution to proliferating cells for 24 h (34).
Cell synchronization in interphase was determined by flow cytometry of isolated nuclei. One million nuclei were directly stained with propidium iodide (5 g/ml in phosphate-buffered saline containing 0.4% Triton X-100) and analyzed by FACScan (Becton Dickinson) using the Lysis II-software. Data are presented as histograms showing relative DNA content (x axis) and cell number (y axis).
Preparation of Nuclei and Cell Extracts-Nuclei from mimosinearrested cells were prepared by hypotonic treatment, followed by Dounce homogenization and centrifugation exactly as described (34). Concentrations of nuclei were determined with a hemocytometer. Nuclei were stored at Ϫ80°C for up to 3 months without loss of DNA replication competence. For permeabilization, nuclei were incubated in 0.1% Triton X-100 in SuNaSpBSA (250 mM sucrose, 75 mM NaCl, 0.5 mM spermine trihydrochloride, 0.15 mM spermidine tetrahydrochloride, 3% bovine serum albumin) at 4°C in a rotator for 20 min and washed two times in SuNaSpBSA without Triton X-100.
Cytosolic and nuclear extracts from asynchronously proliferating and synchronized cells were prepared exactly as described (17,35). Protein concentrations were determined with the Bio-Rad Protein assay using bovine serum albumin as standard. Cytosolic extracts were frozen in liquid nitrogen and stored up to 4 months at Ϫ80°C without loss of replication initiation activity.
DNA Synthesis Reactions and Analysis of Reaction Products-DNA replication initiation reactions (17) contained the following components: HeLa cell cytosolic extract (100 g of protein, unless indicated otherwise); a buffered mixture of rNTPs and dNTPs including either biotin-16-dUTP (Roche Molecular Biochemicals) or [␣-32 P]dATP as tracers (17); and 2-5 ϫ 10 5 nuclei from mimosine-arrested HeLa cells (34). The final reaction volume of 50 l was adjusted with replication buffer (20 mM K-HEPES, pH 7.8, 100 mM potassium acetate, 1 mM MgCl 2 , 0.1 mM dithiothreitol). Where indicated, reactions were also supplemented with S phase nuclear extract (50 g of protein). Incubation time was 3 h, unless indicated otherwise.
The Cdk-inhibitors roscovitine and olomoucine (both Calbiochem) were dissolved in dimethyl sulfoxide at 50 mM. When used, they were added to replication reactions at the final concentrations specified in the figure legends. Control reactions contained an equivalent volume of dimethyl sulfoxide.
Recombinant cyclin A/Cdk2 and cyclin E/Cdk2 were prepared from SF9 cells infected with recombinant baculovirus expression vectors (gifts of W. Krek, Friederich-Miescher-Institute, Basel). Purified human cyclin B1/Cdc2 was a gift of M. Jackman (Wellcome/CRC Institute, Cambridge) and cyclin D2/Cdk6 a gift of E. Laue and W. Zhang (Department of Biochemistry, University of Cambridge).
For analysis by confocal fluorescence microscopy, nuclei were fixed and spun onto coverslips. Total genomic and replicating DNA were visualized and analyzed exactly as described in Refs. 18 and 34. Analysis of radioactively labeled replication products by acid precipitation and by density substitution were performed exactly as detailed before (Refs. 17, 18, and 34, and references therein).
Immunoprecipitation of cyclin E from human cell extracts was performed with polyclonal antibody sc-198 (Santa Cruz) from a total of 150 g of extract protein diluted in 1 ml of phosphate-buffered saline (36). The immunoprecipitate was washed in phosphate-buffered saline and subsequently analyzed by Western blotting using monoclonal antibody HE12.
Depletion of Cdk1 and 2 from Cytosolic Extract-Sepharose beads coupled to human p9 Cks1 protein (37) (a gift of M. Jackman, Wellcome/ CRC Institute, Cambridge) were equilibrated and washed three times in replication buffer and concentrated by gravity sedimentation. Depletion of interphase cytosol was achieved by three successive rounds of (i) adding a fifth volume of the p9 Cks1 beads, (ii) incubating the slurry for 20 min at 4°C in a rotator, and (iii) removal of the beads by pelleting at 13,000 rpm in an Eppendorf 5415C centrifuge for 5 min. Mock depletions were performed in parallel in the absence of p9 Cks1 -Sepharose.

Reversible G 1 Phase Arrest of Human Cells by Mimosine-
Asynchronously proliferating human cells arrest in late G 1 phase when a 0.5 mM concentration of the plant amino acid mimosine is added to the culture medium for 24 h (34). Removal of mimosine from the culture medium in vivo results in a synchronous entry into S phase in about 50 -70% of the cells (27,34). Our previous work has established that a proportion of late G 1 phase nuclei prepared from cells released either from mitosis (17) or quiescence (18) can serve as templates for initiation of DNA replication upon incubation in human S phase extracts. Therefore, I investigated whether the mimosine-dependent reversible arrest point of human cells before entry into S phase could be exploited for a preparation of more homogeneous populations of competent and defined template nuclei for efficient initiation of DNA replication in vitro.
Initiation of DNA Replication in S Phase Extracts-Nuclei were isolated from mimosine-arrested human cells and used as templates for DNA replication reactions in vitro ( Fig. 1). DNA synthesis was detected by incorporation of biotin-dUTP into genomic DNA and confocal fluorescence microscopy. Incubation of nuclei from mimosine-arrested cells in elongation buffer resulted in about 10% of the nuclei incorporating biotin-dUTP in vitro (Fig. 1A). These nuclei represent a small proportion of contaminating true S phase nuclei present in the preparation that continue semiconservative DNA replication at sites established prior to preparation in vivo (34). In contrast, addition of S phase cytosol to the reaction supported DNA synthesis in about 50 -55% of the nuclei (Fig. 1B). Because 10% of the nuclei had initiated DNA replication in vivo before preparation (cf. Fig. 1A and Ref. 34), this result demonstrates that S phase cytosol triggers initiation in about of 40 -45% of the nuclei that had been arrested by mimosine before establishment of DNA replication forks in vivo. The percentage of nuclei synthesizing genomic DNA depended on the amount of S phase cytosolic protein added and reached saturation at 100 g (Fig. 1C, columns 2, 5, and 6). Addition of nuclear extract from S phase cells also increased the percentage of nuclei synthesizing DNA (Fig.  1C, column 3). Addition of subsaturating amounts of both extracts together increased the percentage of nuclei synthesizing DNA in vitro in an additive fashion (Fig. 1C, column 4). The maximum percentage of nuclei initiating DNA synthesis in vitro was about 60 in most preparations, over and above the contaminating proportion of S phase nuclei, corresponding to the percentage of cells entering S phase in vivo upon removal of mimosine.
The fluorescent signal of DNA synthesis in nuclei from mimosine-arrested cells was not homogeneous within the nuclei (Fig. 1B). Replicating nuclei were therefore analyzed by confocal microscopy at higher magnification ( Fig. 1, D-F). Against the background staining of nuclear DNA (Fig. 1, D and F, red signal), a clear pattern of many very small discrete intranuclear sites of DNA synthesis was detected ( Fig. 1, E and F, green signal). This pattern resembles the small replication foci found in very early S phase, which are located in the euchromatic regions of the nucleus ( Fig. 1F; see Refs. 38 -40, for reference). These data suggest that nuclei from mimosine-arrested cells initiate DNA synthesis in S phase extracts in vitro at sites used in early S phase in vivo.
In other eukaryotic DNA replication initiation systems, access of soluble initiation factors to the genomic DNA is regulated by the integrity of the nuclear membrane (1,18). Therefore, I asked next whether removal of the nuclear membrane influences initiation of DNA synthesis in this system. Nuclei from mimosine-arrested HeLa cells were treated with 0.1% of the non-ionic detergent Triton X-100 to remove the nuclear membrane and used as templates for DNA replication in S phase cytosolic extracts (Fig. 2). Nuclear membrane permeability was measured by exclusion of fluorescent dextran. About 90% of the template nuclei were permeable, and remained permeable during the replication reaction in vitro ( Fig. 2A). In comparison, without detergent treatment, typically 5-30% of the nuclei were permeable in a standard preparation by Dounce homogenization (data not shown). Importantly, about 40% of the permeable nuclei initiated DNA synthesis upon addition of S phase cytosol (Fig. 2B), which is the same per-centage as compared with untreated nuclei (cf. Fig. 1C). Therefore, access of soluble initiation activity from S phase cytosolic extract to nuclei of mimosine-arrested cells is not influenced by the presence or absence of an intact nuclear membrane. These data establish that the terminal stages of initiation of human DNA synthesis in vitro as observed in nuclei from mimosinearrested cells do not require the integrity of the nuclear membrane, consistent with observations made in extracts from Xenopus eggs (41).
These data so far do not allow a distinction between initia- tion of semiconservative DNA replication or DNA repair synthesis in vitro. Therefore, initiation of DNA synthesis was analyzed by density substitution (Fig. 3). Nuclei from mimosine-arrested cells synthesized only background amounts of DNA upon incubation in elongation buffer (34), migrating between hemi-and unsubstituted densities (Fig. 3A). In contrast, incubation in S phase cytosol resulted in the synthesis of hemisubstituted DNA products (Fig. 3B). During this 2-h incubation in S phase cytosol, about 65 pmol of dNMP were incorporated into 10 5 replicating nuclei (data not shown), indicating that about 4 -5% of the genomic DNA was replicated. These data directly demonstrate that semiconservative DNA replication is initiated in mimosine-arrested nuclei by a soluble activity present in cytosolic S phase extract.
These data raise the possibility that this soluble initiation activity is inhibited by 0.5 mM mimosine in vivo, causing cellcycle arrest before onset of S phase (34) and allowing efficient initiation of DNA replication in isolated nuclei upon incubation in S phase extracts. In the next experiments, I therefore characterized the competence of cytosolic extracts from mimosinearrested cells to allow DNA replication in vitro.
Lack of DNA Replication Initiation Activity in Cytosolic Extract from Mimosine-arrested Cells-First, S phase nuclei were used as control templates for elongation of DNA replication in vitro (Fig. 4A). As demonstrated before (34), S phase nuclei elongate DNA replication at pre-existing replication forks in either elongation buffer or in S phase cytosol (Fig. 4A, white and black columns). Cytosolic extract from mimosine-arrested cells allowed elongation in the same percentage of S phase nuclei (Fig. 4A, gray column), following the same incorporation kinetics as in S phase cytosol (data not shown). Density substitution confirmed synthesis of hemisubstituted DNA products in all of these three incubations (data not shown). These results demonstrate that cytosol from mimosine-arrested cells efficiently allows elongation of semiconservative DNA replication at established active replication forks in vitro.
In contrast, cytosol from mimosine-arrested cells allowed initiation of DNA replication only in 15-20% of the nuclei from mimosine-arrested cells (Fig. 4B, gray column), over and above the background of S phase contaminants (Fig. 4B, white column). As S phase cytosol triggered initiation in about 50% of these nuclei, the ability of cytosol from mimosine-arrested cells was significantly, but not entirely inhibited. This inhibition can either be explained by the presence of dominant inhibitors of initiation (but not of elongation), or the lack of soluble initiation activity in cytosol from mimosine-arrested cells. To discriminate between these possibilities, I supplemented cytosolic extract from mimosine-arrested cells with cytosol from G 1 and S phase cells (Fig. 4C). Addition of subsaturating amounts of either G 1 or S phase cytosol fully restored the initiation of DNA replication (Fig. 4C), demonstrating that cytosol from mimosine-arrested cells lacks soluble initiation activity.
The data of Fig. 4C suggest that cytosol from G 1 phase cells may also contain replication initiation activity. This would point toward a requirement for a different soluble initiation activity than the S phase-specific cyclin⅐Cdk complexes observed before, using nuclei from cells synchronized in late G 1 phase by release from mitosis or quiescence (17,18). I therefore analyzed the initiation activity of cytosolic extracts from cells arrested at all stages of the cell division cycle in relation to the presence of cyclin/Cdk proteins in these extracts.
Cell Cycle Specificity of Soluble DNA Replication Initiation Activity-HeLa cells were synchronized in early, mid, and late G 1 phase, and in S, G 2 , and M phase (Fig. 5). The presence of cell cycle-regulated cyclins and their kinase partners in cytosolic extracts from these cells was analyzed by Western blotting (Fig. 5A). Cyclin E protein was not detectable in early G 1 phase cytosol, but was present in maximum amounts in mid and late G 1 phase, and in lower amounts in S, G 2 , and M phase cytosol. Small amounts were also present in a salt extract from S phase nuclei (Fig. 5A, S N ). Cyclin A protein was barely detectable in cytosol from G 1 phase cells, but was present in large amounts in S and G 2 phase cytosol. S phase nuclear extract contained Template nuclei were prepared from S phase HeLa cells and incubated in the absence of cytosolic extracts (white column, labeled "buffer") or in cytosolic extract (100 g of protein) from mimosine-arrested cells (gray column, "mim") and S phase cells (black column, "S"). The percentage of nuclei replicating was quantitated as specified in Fig. 1C. B, DNA replication in nuclei from mimosine-arrested cells in vitro. Nuclei from mimosine-arrested HeLa cells were incubated in vitro as specified in panel A. C, complementation of replication initiation activity in cytosol from mimosine-arrested cells by G 1 and S phase cytosolic extracts. Nuclei from mimosine-arrested HeLa cells were incubated in cytosolic extract from mimosine-arrested cells, that was supplemented either with buffer only (white column, buffer), or with cytosolic extract (50 g of protein) from mimosine-arrested (gray column, mim), S phase (black column, S), and G 1 phase cells (black column, G1). large amounts of cyclin A protein. Cyclin B1 protein peaked in G 2 cytosol and was barely detectable in the other extracts. Cyclin D1 and the kinases Cdk1, Cdk2, and Cdk4 were clearly detectable in cytosolic extracts from all stages of the cell cycle. Additionally, Cdk2 was most abundant in S phase nuclear extract, similar to cyclin A (Fig. 5A, S N ).
Next, the DNA replication initiation activity of these extracts was tested in vitro using nuclei from mimosine-arrested cells (Fig. 5B). Surprisingly, cytosolic extracts from all stages of G 1 phase cells triggered initiation of DNA replication most efficiently in up to 60% of the nuclei. S and G 2 phase cytosolic extracts were less active, triggering initiation in about 25-40% of the nuclei and even cytosol of mitotic cells, arrested in metaphase by nocodazole, triggered initiation in about 25% of the nuclei. Initiation activity of S phase cytosol was stimulated by addition of S phase nuclear extract to the maximal efficiency observed in late G 1 phase cytosol alone (Fig. 5B).
These data clearly establish that the cytosolic DNA replication initiation activity for nuclei from mimosine-arrested cells is not strictly restricted to a particular phase of the cell cycle, however, it peaks in G 1 phase and is partially inhibited in mitosis. Most importantly, replication initiation activity does not simply correlate with the protein levels of the G 1 /S phasespecific cyclin E⅐Cdk2 and cyclin A⅐Cdk2 complexes present in these extracts.
These unexpected data raise two important questions, which I will address in the remaining experiments successively. (i) Does cytosol from untreated, asynchronously proliferating cells also trigger initiation in nuclei from mimosine-arrested cells? If so, this would provide an enormous simplification in the experimental protocol to study the initiation of human DNA replication in vitro by dismissing the requirement to synchronize cells for preparation of initiating extracts. (ii) Is there a requirement for the G 1 /S phase-specific cyclin⅐Cdk complexes to initiate DNA replication in this system?
Interphase Cytosol Triggers Initiation of DNA Replication After a Short Lag Period-Nuclei from mimosine-arrested cells were incubated in cytosolic extract from asynchronously proliferating cells and the time course of DNA replication was followed in vitro (Fig. 6). After an initial lag of 15 min, cytosol from interphase cells triggered efficient initiation of replication in about 30% of the nuclei during the following 45 min (Fig. 6A,  closed symbols). Then, the rate of initiation dropped to about 7% of nuclei initiating per hour for the remaining incubation. The amount of replicated DNA accumulated after an initial lag for at least 2 h (Fig. 6B, closed symbols). In control incubations in the absence of cytosolic extract, the background percentage of nuclei replicating did not change over the 3-h incubation period (Fig. 6A, open symbols) and only small amounts of DNA were synthesized (Fig. 6B, open symbols). These data demonstrate that interphase cytosol triggers initiation of DNA replication in nuclei from mimosine-arrested cells efficiently after a short lag in the initial hour of the in vitro incubation.

FIG. 5. Cytosol from all phases of the cell cycle allows initiation of DNA replication in nuclei from mimosine-arrested cells.
HeLa cells were synchronized in early G 1 (G1 e , 4 h post-release from nocodazole-arrest), mid G 1 (G1 m , 6 h post-release), late G 1 (G1 l , 8 h post-release), S (2 h post-release from thymidine block) and G 2 phase (9 h post-release from thymidine block), and in mitotic metaphase (M, nocodazole-arrest) (17). Cytosolic extracts containing cytoplasmic and nucleosolic proteins were prepared from these cells. A high-salt extract from S phase nuclei (S N ) was also prepared. A, Western blot analysis of these extracts. For a detection of cyclin E protein, it was first immunoprecipitated from 150 g of total extract protein and immunoprecipitates were analyzed by Western blotting. For the other proteins, identical amounts of each extract (50 g of protein) were directly analyzed by Western blotting using antibodies against the indicated human cyclin/Cdk proteins (see "Materials and Methods"). B, initiation of DNA synthesis in nuclei from mimosine-arrested cells in these extracts. Reactions contained identical amounts (100 g) of the indicated cytosolic extracts, and control reactions contained no cytosol (buffer, white column) or both, S phase cytosol and 50 g of protein of S phase nuclear extract (SϩS N ). The percentages of nuclei replicating were quantitated as described in the legend to Fig. 1. Initiation Depends on Cyclin/Cdk Activity-To address an involvement of cyclin⅐Cdk complexes in the initiation of DNA replication in nuclei from mimosine-arrested cells, the influence of the specific Cdk inhibitors roscovitine and olomoucine (42,43) was first determined (Fig. 7). Concentrations above 0.5 and 2 mM roscovitine and olomoucine, respectively, inhibited replication to a background of about 10% of the nuclei (Fig. 7,  A and B). An inhibition to 50% of the maximal number of replicating nuclei was observed at about 20 M roscovitine and 0.6 mM olomoucine (Fig. 7A), reproducing the 20-fold difference of the half-maximal inhibition by these two inhibitors of purified Cdks 1, 2, and 5 but not of other kinases (42). In density substitution experiments, only small amounts of DNA synthesis products of intermediate densities between LL and HL were formed in the presence of 0.5 mM roscovitine (Fig. 7C). Together, these data strongly suggest that initiation of DNA replication in nuclei from mimosine-arrested cells is inhibited by roscovitine and olomoucine, and elongation of DNA replication occurs only for relatively short distances in the 10% of contaminating S phase nuclei present in the preparation. The identity of the cyclin⅐Cdk complexes required for initiation was analyzed by adding recombinant human cyclin⅐Cdk complexes to reactions in the presence of roscovitine. Human cyclin A/Cdk2 and cyclin E/Cdk2, but not cyclin D2/Cdk6 or cyclin B1/Cdk1 could fully overcome the inhibition of initiation by roscovitine (Fig. 7D). These results strongly suggest an essential and specific role for cyclin A/Cdk2 and/or cyclin E/Cdk2 activity in the initiation of DNA replication in this system. As the cytosolic initiation activity does not correlate with either cyclin A or E protein levels (Fig. 5), I therefore investigated the contribution of the kinase Cdk2 and the intracellular localization of the endogenous cyclin/Cdks to the initiation of DNA replication in this system.
Cyclin/Cdk2 Is Present in Cis and Is Not Required in Trans for Initiation-The contribution of the kinase Cdk2 from either soluble extract in trans or nuclei in cis during the initiation reaction in vitro was first analyzed by depleting Cdks 1 and 2 from initiating interphase cytosol using Sepharose beads coated with protein p9 Cks1 (Fig. 8). Western blot analysis confirmed depletion of Cdk1 and Cdk2 from the extract (Fig. 8A). As control, Cdk4 was not depleted (Fig. 8A). The depleted cytosol triggered initiation of DNA replication in nuclei from mimosine-arrested cells as efficiently as mock-depleted cytosol (Fig. 8B), indicating that Cdk2 (and Cdk1), and hence its kinase activity, is not supplied by the cytosol in trans in this system. Finally, the relative localization of these cyclin/Cdks in either cytosol or nuclei from mimosine-arrested cells was analyzed by Western blotting (Fig. 8C). Cyclins A and E, and the kinase Cdk2 were clearly and predominantly present in salt extracts of nuclei from mimosine-arrested cells (Fig. 8C). Protein levels of cyclin A and Cdk2 in cytosolic and nuclear extracts from mimosine-arrested cells were similar to the protein levels found in cytosolic and nuclear extracts from S phase cells (data not shown). Taken together, these data demonstrate that cyclin A/Cdk2 and cyclin E/Cdk2 are provided by the nuclei in cis. Therefore, initiation of DNA replication in nuclei from mimosine-arrested cells requires nuclear cyclin/Cdk activity and an additional activity present in cytosolic extracts from untreated cells. DISCUSSION The experiments reported here demonstrated that 0.5 mM mimosine arrests human tissue culture cells in late G 1 phase at a state of competence to initiate DNA replication. Nuclei isolated from mimosine-arrested cells serve as efficient templates for initiation of DNA replication in soluble extracts from proliferating human cells. The state of competence to initiate DNA replication correlates with a nuclear localization of cyclin A, E, and Cdk2 proteins. Initiation of DNA replication in vitro depends on nuclear cyclin/Cdk2 activity and an additional, essential and soluble initiation activity present in the cytosolic extracts. Replication Initiation Competent Template Nuclei-Nuclei from mammalian cells synchronized in the G 1 phase of the cell division cycle can initiate DNA replication upon incubation in human S phase extracts (17,18). The competence of isolated nuclei to initiate DNA replication in vitro arises in late G 1 phase shortly before onset of S phase in vivo. Template nuclei were previously obtained from cells that are progressing through G 1 phase after a release from a block in either mitosis (17) or quiescence (18). As a result of the inherent degree of asynchrony of these dynamic cell populations, only limited and varying percentages of nuclei in a preparation initiate DNA replication in S phase extracts. However, using release from quiescence and addition of exogenous Cdc6 protein significantly increased the percentage of nuclei initiating in this approach (18), but the identification of endogenous markers for initiation competent nuclei is hampered by the relative asynchrony of the nuclear preparation.
Here, I report on the use of template nuclei prepared from human cells reversibly arrested in the cell cycle by 0.5 mM of the plant amino acid mimosine. These nuclei have a G 1 phase DNA content and do not contain active DNA replication forks capable of elongating DNA replication in elongation buffer in vitro (Ref. 34 and this paper). As control, true S phase nuclei do elongate DNA replication at existing forks under these conditions (Ref. 34 and this paper). Nuclei from mimosine-arrested cells initiate semiconservative DNA replication reproducibly with high efficiency upon incubation in cytosolic extracts from interphase cells (this paper).
Initiation competence of nuclei from mimosine-arrested cells correlates with high nuclear cyclin A/Cdk2 protein levels and initiation of DNA replication depends on nuclear, but not on cytosolic cyclin/Cdk2 activity (Fig. 8). In proliferating cells, cyclin A accumulates in the nucleus from S phase onwards until degradation in mitosis (44). Because of the nuclear localization of cyclin A, nuclei from mimosine-arrested cells could be considered S phase in character. However, active DNA replication forks clearly have not been established in these nuclei (34), and by the stringent criterion of not synthesizing DNA, they have to be considered pre-S phase, or late G 1 phase in nature. It can therefore be concluded that mimosine blocks proliferating cells in a state of initiation competence before the actual establishment of active DNA replication forks. It is also conceivable, that mimosine arrests human cells by preventing establishment of replication forks involving a late G 1 phase checkpoint, but allowing continued cyclin A synthesis and nuclear accumulation.
Nuclear membrane integrity is neither required for, nor inhibits initiation of DNA replication in this system. This observation is consistent with work on Xenopus egg extracts, where initiation of DNA replication can be observed in the absence of a nuclear membrane when chromatin is first incubated in cytosolic egg extract, followed by addition of a highly concentrated nucleosolic extract (41). In this Xenopus system, replication competent chromatin is assembled by the cytosolic egg extract and initiation is subsequently triggered by a high concentration of soluble nuclear factors in the absence of an intact nuclear structure, mimicking a nuclear environment (41). In human cell extracts, it was demonstrated that an intact nuclear membrane prevents exogenous Xenopus Cdc6 protein from binding to chromatin and establishing premature initiation of DNA replication in mouse nuclei (18). However, initiation of DNA replication in the absence of exogenous Cdc6 protein was not increased or inhibited by permeabilizing the nuclear membrane at the beginning of the incubation in a subpopulation of template nuclei (18). The data reported here (Fig. 2) show that human nuclei from mimosine-arrested cells do not require, and are not inhibited by the nuclear membrane for initiation of DNA replication in human cell extracts. Together, these data suggest that G 1 phase events of establishing the competence of nuclei to initiate DNA replication may re-

FIG. 8. Cis-requirement for cyclin A/Cdk2 and E/Cdk2 for the initiation of DNA replication in nuclei from mimosine-arrested cells.
A and B, cytosolic extract from interphase cells was depleted with p9 Cks1 beads as specified under "Materials and Methods." A, Western blot analysis of mock-treated (mock) and depleted cytosol (⌬). Identical protein amounts of each extract (50 g) were separated on polyacrylamide gels, blotted, and probed with antibodies against the indicated proteins as specified under "Materials and Methods." B, DNA replication in nuclei from mimosine-arrested cells upon incubation in mocktreated and depleted cytosolic extracts. Nuclei from mimosine-arrested HeLa cells were incubated in vitro in the absence of cytosol (white column, buffer) and in mock-treated and depleted cytosol (black columns, as indicated). The percentages of nuclei replicating were determined by confocal fluorescence microscopy as detailed in the legend to Fig. 1. C, localization of cyclins and Cdks in mimosine-arrested cells. Western blots of cytosolic and nuclear extracts (50 g of protein each) from mimosine-arrested cells using antibodies against the indicated proteins as specified under "Materials and Methods." quire the assistance of selective transport across the nuclear membrane, however, initiation per se does not depend on it after competence for initiation is established.
The experimental approach of using template nuclei from mimosine-arrested cells provides an extensive and robust simplification in the experimental protocols to study human initiation of DNA replication in vitro. Preparation of competent template nuclei involves a single synchronization step of proliferating human cells and trans-acting initiating extracts are obtained from unsynchronized proliferating human cells. This approach will allow future analysis of the molecular events during establishment of DNA replication forks in a variety of human cell types. Furthermore, it allows establishment of screening tests for novel inhibitors of the initiation of human DNA replication. This system, however, is limited in the analysis of earlier G 1 phase events during the establishment of competence to initiate DNA replication which lie before the arrest point of mimosine.
Involvement of Cyclin⅐Cdk Complexes in the Initiation of DNA Replication-Initiation of DNA replication in nuclei of mimosine-arrested cells in vitro depends on the addition of cytosolic extract from interphase cells and requires cyclin A/Cdk2 or E/Cdk2 activity. The evidence for Cdk dependence stems from inhibition and rescue experiments using the specific inhibitors roscovitine and olomoucine and is further supported by correlating initiation activity with endogenous cyclin and Cdk proteins in the materials used.
Initiation of DNA replication in nuclei from mimosine-arrested cells was inhibited by roscovitine and olomoucine (Fig.  7). In human fibroblasts, both compounds arrest the cell cycle in G 1 phase by inhibiting Cdk2, but not Cdk4 kinase (45). Specifically, both compounds inhibit purified Cdk1, 2, and 5 by competing with ATP binding at the active center of the kinases (42,43,46). The half-maximal inhibition (IC 50 ) of purified protein kinase activity differs by a factor of about 20 between roscovitine and olomoucine (42). This relative difference in the IC 50 is also observed for the inhibition of DNA replication in nuclei from mimosine-arrested cells (Fig. 7), strongly arguing toward an essential requirement of cyclin⅐Cdk1/2 complexes in the initiation reaction. However, the absolute values of the IC 50 differ between the two types of assay and can be explained by the presence of excess free ATP/Mg 2ϩ in the crude replication reactions. Lowering free ATP/Mg 2ϩ in replication assays to levels used in the kinase assays with purified proteins (42,43,46) did not allow DNA replication to occur in vitro (data not shown). Therefore, the requirement for high ATP/Mg 2ϩ precluded determination of the IC 50 under ATP/Mg 2ϩ concentrations of the kinase assays.
Specificity with respect to the kinase and its cyclin partner was analyzed by adding recombinant cyclin⅐Cdk complexes to in vitro replication reactions in the presence of roscovitine (Fig.  7D). These experiments demonstrated that Cdk2 complexed to cyclin A and/or E could rescue the inhibition, supporting roles for one or both of these two kinases in initiating DNA replication in human cell extracts (17).
However, the soluble initiation activity of cytosolic extracts from synchronized cells did not correlate with the endogenous protein levels of either cyclin A/Cdk2 or E/Cdk2 (Fig. 4). Cytosolic cyclin A protein was present only in background amounts throughout G 1 phase and accumulated in S and G 2 phase (Fig.  4A, cf. Ref. 44). Cyclin E protein was absent in early G 1 but was induced maximally in mid/late G 1 phase and persisted at gradually decreasing concentrations through S and G 2 phase (Fig.  4A, cf. Refs. 13 and 47). Initiation activity was greatest in cytosolic extracts throughout G 1 phase (Fig. 4B), indicating that protein levels of either or both cyclins in these extracts cannot constitute the initiation activity of the cytosolic extracts. The kinase Cdk2 was present in cytosolic extracts throughout the cell cycle and would therefore be available for association with the cyclin subunits to constitute protein kinase activity. However, a functional role for cytosolic cyclin A⅐Cdk2 and cyclin E⅐Cdk2 complexes in triggering initiation of DNA replication in nuclei from mimosine-arrested cells was directly excluded by depleting Cdk2 from interphase cytosol without loss of initiation activity (Fig. 8). Furthermore, addition of purified recombinant human cyclin A/Cdk2 and cyclin E/Cdk2 to replication reactions in the absence of cytosolic extract did not initiate DNA replication in nuclei from mimosinearrested cells (data not shown). In any case, cyclin A/Cdk2, and to a lesser extent cyclin E, are provided in cis by the template nuclei from mimosine-arrested cells and could therefore constitute the roscovitine-sensitive initiation activity in vitro (Figs. 7 and 8). This nuclear localization of cyclins A and E, and Cdk2 in nuclei from mimosine-arrested cells may therefore also explain the lack of dependence on S phase-specific soluble extracts to initiate DNA replication in this system.
In nuclei from human cells released from mitosis, cyclin A/Cdk2 and E/Cdk2 triggered initiation synergistically (17), whereas in nuclei from mimosine-arrested cells, both could overcome inhibition of initiation by roscovitine independently from each other (Fig. 7D). These data may indicate that a synergistic effect of both kinases is required for triggering initiation in nuclei from proliferating human synchronized at earlier stages of G 1 phase before the mimosine-arrest point, and both kinases may act redundantly at later stages. However, when nuclei from mouse cells released from quiescence were used as templates, only recombinant cyclin E/Cdk2, but not cyclin A/Cdk2 could overcome a block of initiation by olomoucine (18). This suggests that the requirement for either cyclin⅐Cdk2 complex may actually vary with and depend on the synchronization procedures and sources of template nuclei used.
Mitotic cyclin B1 protein was enriched in G 2 phase cytosol and was only present in background quantities in the other extracts (Fig. 4A), consistent with the intracellular localization of cyclin B1/Cdk1 in the human cell cycle (44,48,49). Cyclin B1/Cdk1 protein levels in the cytosolic extracts did not correlate with the DNA replication initiation activity of these extracts (Fig. 4) and cyclin B1/Cdk1 does not rescue inhibition of initiation of DNA replication by roscovitine. These data exclude a functional role for cyclin B1/Cdk1 in triggering initiation of DNA replication in human cells, consistent with previous data (17).
D type cyclins are expressed in response to mitogen stimulation (50) and, consequently, cyclin D1 and its kinase partner Cdk4 were found in all extracts of synchronized proliferating HeLa cells used here (Fig. 4A). This correlates with, and therefore does not formally exclude an involvement of this soluble protein complex in triggering initiation of DNA replication. A direct functional role is, however, unlikely for the following reasons. Cyclin D/Cdk4 or 6 complexes are inhibited a 1000fold less specifically by roscovitine or olomoucine than cyclin A or E⅐Cdk2 and cyclin B1⅐Cdk1 complexes (42,43,46). Furthermore, recombinant human cyclin D2/Cdk6 did not rescue the inhibition of initiation of DNA replication by roscovitine (Fig.  7D). However, these data do not exclude the possibility that cyclin D complexes contribute indirectly to the initiation activity present in interphase cytosol.
A Novel Soluble Initiation Activity-Taken together, the Cdk dependence of initiation in nuclei from mimosine-arrested cells supports a model, where cyclin A/Cdk2 or E/Cdk2 are essential, but not sufficient for initiation of DNA replication. They are conferred by the template nuclei in cis. Initiation, furthermore, depends on an additional activity present in cytosolic extracts from untreated cells. This model is supported by the observation that cytosol from mimosine-arrested cells lacks this soluble initiation activity (Fig. 5). However, this lack of initiation activity is not complete because initiation still occurs in about 10 -15% of the nuclei. This partial initiation deficiency is fully overridden by addition of small amounts of interphase cytosol from cells that are not arrested by mimosine, restoring full initiation activity (Fig. 5). This restoration of initiation activity can also explain the partial nature of initiation deficiency found in cytosol from mimosine-arrested cells by postulating the presence of residual, subsaturating initiation activity in the extract from mimosinearrested cells. This residual activity can derive from the proportion of cells in the preparation, which are not at the arrest point, but in early/mid G 1 or S phase (34).
The results also suggest that the in vivo target of 0.5 mM mimosine preventing entry into S phase (34), could be identical to the soluble initiation activity found in the interphase cytosol from cells which are not treated with mimosine. Importantly, this initiation activity is not dramatically regulated throughout the cell cycle, however, it accumulates through G 1 phase and peaks before onset of S phase. The identity of this activity is currently unknown, but candidates may include one or more of activities like cyclin/Cdk activating factors, substrates for S phase-specific cyclin A/Cdk and E/Cdk kinases, or activities mediating the assembly of replication forks from DNA replication proteins. We are currently purifying this soluble activity from cytosolic extract in order to identify factors that link cyclin/Cdk activity to the establishment of active DNA replication forks in human cell nuclei.