Semi-conservative Replication in Yeast Nuclear Extracts ... JMB... · semi-conservative DNA replication, and initiation is both origin-speciﬁc and ORC-dependent. Repli-cation was

Article No. mb981973 J. Mol. Biol. (1998) 281, 631±649

Semi-conservative Replication in Yeast NuclearExtracts Requires Dna2 Helicase andSupercoiled Template

D. Braguglia, P. Heun, P. Pasero, B. P. Duncker and S. M. Gasser*

Swiss Institute forExperimental Cancer Research,Ch. des Boveresses 155, CH-1066, Epalinges/Lausanne,Switzerland

Present address: D. Braguglia, SuCH-8401 Winterthur, Switzerland.

Abbreviations used: sc, supercoilreplicating sequence; ORC, origin r1D, 2D, one and two-dimensional;RI, replication intermediate; BrdUTH-L, heavy-light; L-L, light-light; Emicroscopy; EtBr, ethidium bromidds, double-stranded; ts, temperatur

E-mail address of the [email protected]

0022±2836/98/340631±19 $30.00/0

We describe the preparation of nuclear extracts from yeast cells synchro-nised in S-phase that support the aphidicolin-sensitive, semi-conservativereplication of primer-free, supercoiled plasmid in vitro. This is monitoredby one and two-dimensional gel electrophoresis of replication intermedi-ates that have incorporated [a-32P]dATP, by the conversion of methylatedtemplate DNA into a hemi-methylated or DpnI-resistant form, and bysubstitution of dTTP with the heavy derivative BrdUTP, which results ina shift in density corresponding to complete second strand synthesis. Wedemonstrate dependence on DNA pol d and the pol a/primase complex,and are able to detect putative Okazaki fragments under ATP-limitingconditions. In contrast to the semi-conservative replication of supercoiledplasmid, linear or open-circular templates incorporate labelled nucleo-tides through repair synthesis that produces no signi®cant density shifton CsCl gradients. Consistent with a true replication reaction we ®ndthat semi-conservative replication of plasmid DNA is stimulated in S-phase relative to G1-phase nuclear extracts, and is independent of therecombination-promoting factor Rad52p. Using this novel system wedemonstrate that semi-conservative replication, but not polymeraseactivity per se, requires the activity of the DNA helicase encoded byDNA2.

# 1998 Academic Press

Keywords: in vitro replication; yeast; DNA polymerases; DNA helicase;DNA2
*Corresponding author
Introduction

Replication of the eukaryotic genome must beachieved once per cell cycle, and with high ®delityand speed, to avoid deleterious consequences forthe cell's progeny. Although signi®cant progresshas been made identifying factors that catalyzeeukaryotic DNA replication in vivo (reviewed byCoverley & Laskey, 1994; Dif¯ey, 1996), site-speci®c initiation of replication from a eukaryotic

lzerInnotec Ltd,

ed; ARS, automouslyecognition complex;oc, open circular;P, bromodeoxy UTP;M, electrone; ss, single-stranded;e-sensitive.ing author:

origin has never been successfully reconstituted ina soluble system. In contrast, assays based on frogoocytes, HeLa cell, or yeast nuclear extracts havedemonstrated cell-cycle-dependent replication ofgenomic DNA by using G1-phase nuclei as tem-plate (Blow, 1996; Gilbert et al., 1995; Krude et al.,1997; Pasero et al., 1997). When naked plasmid orphage DNA is introduced into the Xenopus acti-vated egg extract, the template is ®rst organisedinto pseudo-nuclei prior to replication, andinitiation occurs with no detectable sequence-speci-®city (Blow & Laskey, 1986; Gilbert et al., 1995;Hyrien & Mechali, 1992; Mahbubani et al., 1992). Ifextracts are depleted for membranes, nuclear for-mation is impaired and only very low levels ofsupercoiled (sc) plasmid replication can bedetected (Blow & Laskey, 1986; Hyrien & Mechali,1992; Mahbubani et al., 1992). A recent study indi-cates that condensed chromatin can be replicatedin the absence of the nuclear envelope if a subsetof factors within the Xenopus extract are suf®cientlyconcentrated (Walter et al., 1998). It is not known

# 1998 Academic Press

mailto:[email protected]

632 DNA Replication in Vitro

whether the lack of origin speci®city in thesesystems re¯ects a peculiarity of amphibians, ofearly stages of development, or the absence of anelement of subnuclear organisation (reviewed byGilbert, 1998). Indeed, early embryonic cell cycleslack a G1-phase, during which functional pre-repli-cation complexes assemble on origins in yeast andmammalian cells (reviewed by Pasero & Gasser,1998; Gilbert, 1998).

In view of these differences between embryonicand somatic cell cycles, it is important to establisha soluble in vitro replication system from an organ-ism that undergoes normal mitotic division. Sac-charomyces cerevisiae has provided a convenientgenetic model for the characterisation of DNA syn-thesis (reviewed by Dif¯ey, 1996), and conditionalmutations exist in many genes that are essential foreither the initiation or elongation steps of replica-tion. Moreover, two-dimensional gel techniquesallow a precise mapping of sites within the yeastgenome and on plasmids where DNA replicationinitiates (Brewer & Fangman, 1987; Hubermanet al., 1987; Bielinski & Gerbi, 1998). Thus, buddingyeast provides many advantages for the establish-ment of an in vitro replication assay, and shouldallow us to identify missing elements of theeukaryotic replication machinery, such as the repli-cation helicase.

Earlier papers have described whole yeast cellextracts that were purported to support DNAreplication in vitro (Celniker & Campbell, 1982;Jazwinski & Edelman, 1979; Kojo et al., 1981).However, complete complementary strand syn-thesis was never demonstrated and a requirementfor the replicative DNA polymerases, pol a andpol d, was not shown. Subsequent studiessuggested that the labelling of input plasmid inthese systems might have been due to the exten-sion of endogenous RNA or DNA primers in thebacterially prepared template, an event that couldalso be mediated by bacterial DNA Pol I and/orthe Klenow fragment of this enzyme (reviewed byJong & Scott, 1985). Surprisingly, these systemsshowed preferential incorporation of labellednucleotides into ARS-containing plasmids, whichwas later thought to re¯ect plasmid isolation tech-niques and/or selective repair mechanisms (Jong &Scott, 1985).

In recent years the assays for monitoring replica-tion have evolved such that bidirectional semi-con-servative DNA synthesis can be unequivocablydemonstrated. Essential criteria for bona ®de DNAreplication are as follows. (1) Dependence on DNApolymerases d and e, and in particular, pol a/pri-mase, since it is the only polymerase uniquelyimplicated in the de novo initiation of DNA replica-tion (reviewed by Campbell & Newlon, 1991;Foiani et al., 1997). (2) Requirement for correct cellcycle co-ordination (reviewed by Stillman, 1996).(3) Evidence that complete complementary strandsof DNA are synthesised. Since repair synthesisusually results in short stretches of substitutedDNA, the most rigorous distinction between repair

and replicative DNA synthesis relies on densitygradient centrifugation following substitution witha heavy nucleotide analogue.

We have recently demonstrated that nucleartemplates replicate ef®ciently in yeast nuclearextracts (Pasero et al., 1997). During the reaction 10to 15% of the input DNA is used as template forsemi-conservative DNA replication, and initiationis both origin-speci®c and ORC-dependent. Repli-cation was signi®cantly reduced, however, whentemplate nuclei were disrupted by non-ionic deter-gents (Pasero et al., 1997). Here we analyse thesemi-conservative replication of supercoiled plas-mid DNA in S-phase extracts from yeast, whichallows for both a genetic and biochemical charac-terisation of the reaction. In addition to the use oftwo-dimensional (2D) gel electrophoresis, whichpermits visualisation of replication intermediates,we monitor the shift in density of DNA replicatedin yeast extracts after substitution with BrdUTP,show conversion of methylated plasmid to a hemi-methylated form, and demonstrate sensitivity toconditional mutations in subunits of the DNApolymerases. Semi-conservative replication is fullydependent on the integrity of the input plasmidand initiates in the absence of Rad52p-mediatedrecombination events. Additionally, we show thatour system is cell-cycle-dependent, as levels ofreplication are signi®cantly higher in nuclearextracts obtained from cells in S-phase, as com-pared to G1-phase. Using this system we are ableto show that replication in vitro requires Dna2p, aDNA helicase (Budd et al., 1995), with an essentialbut hitherto unde®ned role in S-phase progression.

Results

DNA synthesis in S-phase extracts

Nuclear extracts were prepared as previouslyoptimised for ef®cient RNA pol II transcription,using yeast cells synchronised in S-phase by apheromone block-release protocol (see Materialsand Methods and Verdier et al., 1990). Double-stranded supercoiled (sc) plasmid carrying the384 bp Sau3AI fragment from the yeast H4ARSelement (pH4ARS) was incubated for 90 minuteswith the extract including an ATP regenerationsystem, an excess of deoxy- and ribonucleotides,and [a-32P]dATP. After proteinase K treatment,phenol extraction, and electrophoretic analysis, theplasmid DNA that incorporated radiolabellednucleotide forms a diffuse smear towards the ori-gin of a non-denaturing agarose gel (Figure 1A,lane 1). Its appearance is dependent on theaddition of template and is inhibited by aphidico-lin (Figure 1A, lanes 3 and 2, respectively). If theplasmid is linearised by restriction digestion afterthe incubation, we recover a single band of 3.6 kb,and a smear of larger labelled forms, but no linearproducts are detected by electrophoresis or elec-tron microscopy in the absence of digestion (seebelow). Input sc template DNA and even open cir-

Figure 1. Biochemical require-ments for the incorporation of[a-32P]dATP into supercoiled plas-mid in S-phase nuclear extracts. Inall reactions 300 ng pH4ARS andS-phase extracts with all nucleo-tides and [a-32P]dATP were incu-bated for 90 minutes at 25�C,except where noted otherwise. TheDNA was extracted and separatedon a 0.8% agarose gel containing0.5 mg/ml EtBr. An autoradiogramof the dried gel is shown in eachcase except B, which is an EtBrstained agarose gel. oc indicates theopen circular form of the inputplasmid, while sc indicates thesupercoiled form. A, Lane 1, stan-dard reaction; lane 2, as 1, but withthe addition of 500 mg/ml aphidi-colin; lane 3, no plasmid DNAadded to the reaction. B, There isno modulation in the topology ofthe bulk of input DNA during thethree hour replication reaction.Lane 1, 1 kb ladder; lane 2, inputDNA, lane 3, DNA after incubationin the replication reaction duringthree hours. C, Lane 1, completereaction; lane 2, omission of allfour ribonucleotides (NTPs); lane 3,omission of dTTP, the other threedeoxynucleotides (dA, dC, dG)were present at 20 mM; lane 4,omission of ATP; lane 5, omissionof the ATP regeneration system(creatine phosphate and creatine

kinase). D, Lane 1, complete reaction; lane 2, template plasmid was treated with 1 unit of RNase H prior to the repli-cation reaction; lanes 3 and 4, 100 mM and 200 mM of all four dideoxynucleotides were added, respectively. E, Lane 1,complete reaction; lanes 2 and 3, addition of 1 and 3 mg of monoclonal antibody Mab24D9 (anti-pol a), respectively(Plevani et al., 1985); lane 4, addition of 1 mg of mouse anti-rabbit IgG. In lanes 2 and 3, an additional 32P-labelledband of DNA of about 200 bp appears (see the arrowhead).

DNA Replication in Vitro 633

cular (oc) forms are stable in the extract for up tothree hours at 25�C (Figure 1B, lanes 2 and 3). Thisappearance of larger radiolabelled forms of plas-mid is thus consistent with the presence of replica-tion intermediates (RIs) in the input DNA, incontrast to previous reports in which small, newlysynthesised fragments did not remain template-associated (Jong & Scott, 1985).

The biochemical characteristics of this reactionare consistent with the involvement of DNA pol a,d and/or e. Namely, the reaction is aphidicolin-sensitive (Figure 1A, lane 2), dependent on all fourdeoxynucleotides (Figure 1C, lane 3), and on ribo-nucleotides (Figure 1C, lane 2). Since the reaction isinsensitive to a-amanitin, an inhibitor of RNA polII and III (data not shown), the need for ribonu-cleotides is likely to re¯ect a requirement for RNAprimer synthesis. Consistently, 32P incorporation isinhibited by the addition of anti-pol a antibodies(Figure 1E, lanes 2 and 3; see below). The reactionrequires a creatine kinase-based regeneration sys-tem that converts residual ADP into ATP, but notadditional ATP itself (Figure 1C, lanes 4 and 5).

Finally, we ®nd no inhibition by dideoxy-nucleo-tides at concentrations up to 200 mM, which impairactivity of pol b, but not of pol d, e or a (Figure 1D,lanes 3 and 4; Kornberg & Baker, 1992).

Previous studies have demonstrated that replica-tion systems using bacterial plasmid template riskto simply elongate RNA primers present on theinput plasmid DNA (Jong & Scott, 1985). Our tem-plate plasmids are uniformly supercoiled and areisolated by the alkaline lysis method without chlor-amphenicol ampli®cation (Sambrook et al., 1989).Consistent with an absence of RNA primers on thetemplate, we note that the addition of exogenousRNase H, which degrades the RNA moiety ofRNA-DNA hybrids, does not reduce DNA syn-thesis in vitro (Figure 1D, lane 2). Furthermore,when template alone is incubated with the bac-terial Pol I Klenow fragment, all four dNTPs and[a-32P]dATP, no signi®cant incorporation of radio-activity is observed (data not shown), ruling outthe presence of gaps or preformed primers on thetemplate.


Electron microscopic confirmation of theappearance of replication intermediates

Early analysis of simian virus 40 (SV40) replica-tion in vitro identi®ed structures typical of bi-directional fork movement in a circular templateby electron microscopy (Sundin & Varshavsky,1980). We have visualised the plasmid DNArecovered after a 90 minute yeast replication reac-tion, to score for the presence of replication inter-mediates (Figure 2 and Table 1). If the DNA isanalysed prior to EcoRI linearisation, we observea low but signi®cant frequency of small replica-tion bubbles, which are not present in the inputDNA (Figure 2A and B; 0.7% of 1870 circularmolecules). We also observe a large fraction oflate replication intermediates (plasmids sharingdouble-stranded regions) and interlocked dimers(11.6%; Figure 2C and D). Both this latter popu-lation of interlocked dimers and the presence of

Figure 2. EM analysis of products of the in vitro DNA rep90 minutes in a wild-type S-phase extract at 25�C as describreplication reaction were visualised by transmission electrongested pH4ARS DNA molecule. B, y form.C, late replicationintertwined dimer). D, Catenated dimer. The distribution ofrepresents 200 nm.

multiple interlocked molecules (not shown) couldarise from topoisomerase II-driven catenation(Wang et al., 1990) or from a failure to decatenatefully replicated molecules. Thus, quantitation ofthese catenated forms is not an accurate measureof DNA synthesis. Nonetheless, when early repli-cation intermediates are quanti®ed either with orwithout linearisation of the replication products,we observe comparable levels of molecules con-taining small bubbles (0.7%; Table 1) and roughly3.9% containing fork structures (Table 1). Nosuch structures are present in the input DNA,and their appearance correlates with the smear oflabelled products on one-dimensional agarosegels (Figure 1). Importantly, no structures typicalfor Holliday junctions (i.e. recombination inter-mediates; see Hsu, 1991; Preiser et al., 1996;Malezka et al., 1991), nor rolling circle forms,were observed in over 3500 molecules examined.

lication assay. pH4ARS DNA (300 ng) was incubated fored except with no radioactive label. The products of themicroscopy. A, Replication bubble (�500 bp) in an undi-intermediate (incompletely replicated molecule or highlyindicated structures is summarised in Table 1. The bar

Table 1. Quantitation of in vitro DNA replication

Method

CsCl density gradients (H-L) % of input plasmid used as templateExp. 1: 0.3% (1.8 ng)Exp. 2: 0.3% (1.8 ng)Exp. 3: 0.2% (1.2 ng)Exp. 4: 0.4% (2.4 ng)

Electron microscopy % of recovered plasmid with RIsExp. 1: 1086 linearised molecules analysed:

3.9% forks0.7% bubbles0.1% bubbles within fork

Exp. 2: 1870 circular molecules analysed:0.7% bubbles

11.6% late RIa and interlocked dimersb

The standard replication reaction was performed with 300 ng of supercoiled pH4ARS inwild-type S-phase extracts with BrdUTP added for the experiments analysed by CsCl gradi-ents (see Materials and Methods). DNA was either spread for electron microscopic evalua-tion (see Figure 2), or loaded on CsCl gradients (see Figure 3). Quantitation of the heavy-light peak recovered from CsCl gradients was determined by scintillation counting of theincorporated [a-32P]dATP, calculation of the moles of plasmid synthesized assuming 25%dA content, and dividing this value by two. To determine the percentage of input DNAused as template this value was divided by the moles of input plasmid. In parentheses weindicate total DNA synthesized in the H-L peak. These are minimal estimates of optimalreplication ef®ciency since BrdUTP incorporation is at least ®vefold less ef®cient than dTTP(see the text).

For analysis by electron microscopy, reaction products from two independent experimentsperformed in the absence of derivatised nucleotides were either linearised by digestion withEcoRI, or were spread as circular molecules. Random ®elds were photographed, and allclearly distinguishable replication intermediates (bubbles, forks, etc.) were scored among1086 linearised molecules and 1870 circular molecules.

a Late replication intermediates (RIs) refers to large bubbles or clearly interlocked circlesusually sharing non-replicated domains (e.g. see Figure 2C).

b Interlocked dimers represent catenated DNA.


BrdUTP-substitution and DpnI resistanceconfirm semi-conservative DNA synthesis

To provide conclusive evidence for completecomplementary strand synthesis, and to accu-rately quantify replication, we have analysed thereaction products on CsCl density gradients aftersubstitution of dTTP with the heavy analogue,bromo-deoxyuracil triphosphate (BrdUTP). Sedi-mentation of the fully substituted DNA (heavy-light or H-L) and non-substituted DNA (light-light, or L-L) was standardised by priming singlestranded M13mp18 DNA with the M13 universalprimer and elongating with phage T7 DNA poly-merase in the presence or absence of BrdUTPusing [a-32P]dATP as a tracer (see Materials andMethods). Samples were mixed, separated onCsCl gradients and the refractive indices (Z) forthe substituted and non-substituted bands weredetermined (Figure 3C). Arrowheads indicate thecorrect refractive indices for H-L and L-L DNAfor each gradient.

When BrdUTP is incorporated into supercoiledpH4ARS, we observe a density shift on CsCl gradi-ents to a position of H-L DNA of up to 81% of theradioactivity that is incorporated during the repli-cation reaction (H-L; Figure 3A, open circles). Theef®ciency of the shift to H-L DNA depends bothon the integrity of the sc template and the quality

of the extract, but is highly reproducible whenthese parameters are constant. No H-L peak isobserved in the presence of aphidicolin (Figure 3A,®lled circles). If any fraction of the template DNAis nicked, the peak of H-L DNA drops dramaticallyand a majority of label is incorporated in the L-Lpeak, indicative of repair DNA synthesis, which islargely aphidicolin-resistant (Figure 3B; compareopen circles with ®lled squares). We con®rm thatthe peak at the H-L position re¯ects the substi-tution of dTTP by BrdUTP, since the inclusion of®vefold less dTTP ef®ciently competes for the H-Lpeak, shifting it to a L-L position (Figure 3B, greytriangles).

Integration of the amount of radioactivityrecovered in the fully substituted H-L DNA peak,and its conversion into moles of dATP, allows usto estimate the amount of template that is semi-conservatively replicated in vitro (see Materialsand Methods). Results from four independentreactions indicate that between 0.2% and 0.4% ofthe input plasmid is semi-conservatively repli-cated during a three hour incubation in the pre-sence of BrdUTP (1.2 to 2.4 ng DNA synthesized;Table 1). Unfortunately, this method underesti-mates replication in two ways. First, partiallyreplicated templates and RIs do not migrate inthe H-L peak and are not taken into consider-ation. Second, the reactions have to be done in

Figure 3. CsCl density gradient and DpnI digest ana-lyses indicate semi-conservative replication of super-coiled plasmid DNA. A, sc pH4ARS (300 ng) wasincubated in a GA-59 S-phase nuclear extract with200 mM BrdUTP either with (®lled circles) or without(open circles) 50 mg/ml aphidicolin. DNA was extractedand analysed on CsCl gradients as described inMaterials and Methods. Refractive index readings (Z)and cpm were determined for gradient fractions. TheH-L peak corresponds to Z � 1.4040 (indicated by H-Larrowhead). Lower levels of incorporation are observedin the L-L region of the gradient (L-L arrowhead,Z � 1.4000). B, As A, except that a different preparationof pH 4ARS was`incubated in a GA-59 S-phase nuclearextract either with (®lled squares) or without (open cir-


the presence of BrdUTP which inhibits polymer-ase ef®ciency by at least ®vefold (Laird &Bodmer, 1967; Harland & Laskey, 1980; see alsoFigure 3). Indeed, the quantitation of replicationintermediates seen in linearised products by EM(4.7%; Table 1) suggests that template usage inthe absence of BrdUTP may be up to tenfoldhigher.

Independent evidence for complete de novocomplementary strand synthesis was obtained byevaluating the differential sensitivity of the repli-cated DNA to the restriction enzymes, DpnI andNdeII (an isoschizomer of MboI), which cleaveeither fully methylated or entirely unmethylatedGATC sequences, respectively. Another isoschizo-mer, Sau3AI, cuts both. Thus, the conversion offully methylated template to hemi-methylatedDNA by semi-conservative replication correlateswith an acquired resistance to DpnI (while remain-ing sensitive to Sau3AI and resistant to NdeII; seefor example, Sanchez et al., 1992). This allows us tocalculate the relative ef®ciency of semi-conserva-tive replication versus localised patch repair, byquantifying the label incorporated into full-lengthDpnI-resistant forms. In the experiment shown inFigure 3D, using sc plasmid as template, 59% ofthe incorporated label is converted to DNA that isresistant to DpnI digestion (compare lanes 1 and2). Consistent with a hemi-methylated state, thislabelled fraction is resistant to NdeII and sensitiveto Sau3AI (Figure 3D, lanes 3 and 4). All DpnI-resistant incorporation was inhibited by aphidico-lin (Figure 3D, lane 5). The EtBr staining of the gelindicates that DpnI was present in an excess overthe replicated plasmid, since digestion of bulkDNA is complete (Figure 3D, EtBr). Under optimalconditions up to 86% of incorporated [a-32P]dATPcan be shown to be DpnI-resistant, which is ingood agreement with the fraction of label recov-ered in the H-L peak (Figure 4A; Duncker et al.,unpublished results).

cles) 500 mg/ml aphidicolin, or with the addition of40 mM dTTP (®lled circles), which is ®vefold less thanthe molarity of BrdUTP present in the reaction. C, Inseparate reactions, M13mp18 was primed with M13 uni-versal primer and elongated with T7 DNA polymerase(Pharmacia) in the presence of dTTP or BrdUTP (seeMaterials and Methods), producing light-light (L-L) andheavy-light (H-L) DNA, respectively. Equal amounts ofeach reaction were mixed and run on a CsCl gradient.D, sc pH4ARS (300 ng) was incubated in the standardreplication reaction, and the labelled products weredigested with EcoRI and either DpnI (lane 2), NdeII (lane3) and Sau3AI (lane 4). M indicates end-labelled bac-terial pH4ARS after digestion with EcoRI and DpnI, toindicate where complete digestion products migrate. Anautoradiograph of the dried 1% agarose gel is shownalongside the corresponding EtBr-stained image of thegel. In lane 5 the reaction was performed in the presenceof 50 mg/ml aphidicolin.

Figure 4. Replication in nuclear extracts requiressupercoiled DNA as a template. A, SupercoiledpH4ARS plasmid (300 ng) was incubated in GA-59nuclear extracts for three hours, either in the absence(open circles) or presence (®lled circles) of 500 mg/mlaphidicolin. Puri®ed DNA was run on CsCl density gra-dients and fractionated as described in Materials andMethods. Refractive indices corresponding to themigration of heavy-light (H-L) and light-light (L-L)DNA are indicated by arrowheads. Similar reactionswithout aphidicolin were carried out using: B, 300 ng of


Semi-conservative plasmid replication in vitrorequires a supercoiled template

Additional density gradient analyses were per-formed to assess which forms of plasmid DNAserve as ef®cient substrates for semi-conservativereplication. Parallel reactions were conductedusing the same preparation of pH4ARS plasmidDNA that was either left in its original supercoiledform (sc), converted to 100 % open circular (oc)form by incubation with mung bean nuclease (pro-ducing one nick per molecule), or completely line-arised with EcoRI (lin). As previously observed, thereaction using supercoiled template produced aprominent H-L peak, and a small shoulder in theL-L region of the gradient (Figure 4A, open circles).The addition of aphidicolin to the reactionabolishes the H-L peak (Figure 4A, ®lled circles).In contrast to the results obtained with the sc tem-plate, equal quantities of both the oc and lineartemplates incorporate nucleotides ef®ciently, yet allnewly synthesised DNA migrates in the L-L peak(Figure 4B and C). This indicates that the DNAsynthesis of these forms of plasmid is not semi-conservative, producing only short stretches ofBrdUTP substitution. The total nucleotide incorpor-ation into nicked template is very high, presum-ably because a larger fraction of templatemolecules is used. This shows that neither poly-merases nor nucleotides are limiting, and that asecond, non-processive DNA synthesis can occurat nicks and/or DNA ends. It is important to notethat oc and linear plasmid compete ef®ciently forthe use of sc template; no semi-conservative repli-cation is detected when oc and sc templates aremixed in a 1:1 ratio (Figure 4D). Even when ocplasmid represents 10% of the template, we detectno shift of the sc template to the H-L peak (datanot shown), underscoring the importance of havinga uniform population of sc plasmid as template.

The loss of semi-conservative replication in thepresence of oc or linear substrates was con®rmedby monitoring the conversion of pH4ARS to aDpnI-resistant form, using similarly prepared sub-strates. When randomly nicked, open circularDNA, and/or mung bean nuclease-treated plas-mids are used, only 6 to 8% of the labelled DNAbecomes DpnI resistant (Table 2).

RIs detected by 2D gel electrophoresis are notdependent on Rad52p-mediatedrecombination, nick-translation nor strand-invasion events

Although CsCl gradients provide a rigorousmeasure for second strand synthesis, the method isslow, sensitive to BrdUTP inhibition, and does not

mung bean nuclease nicked pH4ARS; C, 300 ng ofpH4ARS linearised with EcoRI; and D, a 1:1 mixture ofsupercoiled (150 ng) and linearised (150 ng) pH4ARS.

Table 2. Supercoiled plasmid is necessary for semi-con-servative DNA replication as monitored by DpnI resist-ance

% of total incorporationTemplate Aphidicolin Resistant Sensitive

pH4ARS sc ÿ 59 41pH4ARS sc � 0 100pH4ARS oc ÿ 8 92pH4ARS oc � 7 93pH4ARS nk ÿ 6 94pH4ARS nk � 4 96

A standard replication reaction was performed with equalquantities of either supercoiled pH4ARS (sc), a randomlynicked oc form (oc), or a form relaxed by treatment with mungbean nuclease, which results in a single nick per plasmid (nk;see Materials and Methods), in either the presence or absenceof 50 mg/ml aphidicolin. Extracted DNA was digested by DpnIas described in Materials and Methods. The labelled DNA spe-cies that migrated above 1600 bp after an exhaustive DpnIdigestion, represented resistant forms of the plasmid, and thosebelow 1100 bp (the largest DpnI fragment of pH4ARS) repre-sent cleaved forms, labelled by repair reactions or incompletereplication. The relative signals were quanti®ed by a MolecularDynamics PhosphorImager.

Figure 5. Appearance of replication intermediates isindependent of RAD52. Supercoiled pH4ARS (300 ng)(A and B) was incubated with a GA-500 (rad52::TRP1) S-phase extract in a standard replication mixture including[a-32P]dATP for 90 minutes at 25�C, following a pre-incubation in the extract (see Materials and Methods).The DNA was puri®ed and separated by 2D gel electro-phoresis either with A, no restriction digestion or B,after digestion with EcoRI. A ``tail'' of small (<1 n)labelled fragments is visible with the digested post-repli-cation DNA products, and likely represents breakage ofreplication intermediates (Brewer et al., 1988). C, Thereplication intermediates were not detected when thereaction was performed with the addition of three unitsof bacterial pol I and 0.5 mg/ml aphidicolin. Themigration of expected RIs is shown schematically in D(circular DNA), and E (linearised DNA). The pattern ofintermediates in A is very similar to that described forin vivo replicating plasmids separated without restrictiondigestion (Brewer et al., 1988), except that our in vitroproducts show signi®cantly less supercoiling, probablydue to the lack of organisation into nucleosomes. Inpanel F-H we show that ssDNA competes ef®ciently forthe appearance of replication intermediates in the stan-dard replication assay. Either 200 ng of denaturedpH4ARS digested with F, HinfI or G, EcoRI or, 400 ngof linear ssM13 (panel H) were used as competitor inthe standard replication reaction containing 300 ngpH4ARS as template. The arrow in H indicates incor-poration into the full length M13 DNA. Equivalentexposure times were used for all panels.


detect intermediates in the replication reaction. Wetherefore used neutral-neutral two-dimensional(2D) gel electrophoresis (Brewer & Fangman, 1987)as a convenient tool for identifying RIs in the[a-32P]dATP-labelled products recovered after incu-bation of sc pH4ARS in S-phase extracts. The 2Dgel analyses were performed on radiolabelled pro-ducts with (Figure 5B) and without (Figure 5A),linearisation at a unique restriction site oppositethe ARS element after replication. After digestionwe observe typical Y and double Y arcs, inaddition to a large amount of labelled full-lengthplasmid (compare Figure 5B with 5E). The amountof termination intermediates recovered variesamong extracts, and appears to re¯ect the levels oftopoisomerase II activity in the extract (Verdieret al., 1990; data not shown). After restriction diges-tion, we rarely detect the bubble arc that is repre-sentative of a centrally located bidirectional fork inthe linearised plasmid (Duncker et al., unpublishedresults). This is in contrast to the clear bubble arcpresent in the 2D analysis of undigested DNA (see®lled triangle, Figure 5A), which is consistent withthe migration predicted for a replication bubble ina closed circular plasmid (see Preiser et al., 1996;Brewer et al., 1988). Since the only differencebetween these two analyses is the restriction diges-tion performed after DNA replication, it appearsthat the replication fork is broken either by a nick-ing activity or by physical breakage during therestriction digestion or subsequent plasmid recov-ery, allowing loss of the bubble from linearisedintermediates. This fork instability could re¯ectnon-physiological aspects of the in vitro replicationreaction, although such instability has also beenproposed to explain the loss of bubble arcs in repli-cation intermediates isolated from intact cells(Linskens & Huberman, 1990; Martin-Parras et al.,1992).

Figure 6. DNA pol d and primase subunits arerequired for replication in vitro. A to F, 300 ng ofpH4ARS was incubated for 90 minutes either at 25�C orat 37�C, as indicated, with 60 mg of S-phase extract fromeither GA-59 (WT), the temperature-sensitive mutantspri1-1 (Francesconi et al., 1991), or cdc2-2 (Boulet et al.,1989), as indicated, under otherwise standard con-ditions. The extracts used in A to F were from cellssynchronised in S-phase at permissive growth tempera-tures. After incubation, pH4ARS was linearised withEcoRI prior to 2D gel electrophoresis. Autoradiographsof equal exposure times of the dried gels are shown.Reactions were performed at the either 25�C (permissivetemperature, A ,C and E) or at 37�C (restrictive tempera-ture, B, D and F). The presence or lack of a strong Y arcis indicated by ®lled or open arrowheads, respectively.For G to J, the indicated mutant cells were grown at25�C and shifted to 37�C prior to the isolation of nucleiand nuclear extracts. After four hours at restrictive tem-perature these cells arrest in the dumbbell phenotypetypical for mutants affecting DNA replication (Bouletet al., 1989; Francesconi et al., 1991). The wild-type GA-59 cells were also grown at 37�C, but were synchronisedin S-phase by a-factor block release, as described inMaterials and Methods. pH4ARS (300 ng) was incu-bated at 37�C with (G) 60 mg pri2-1 extract, (H) 60 mgcdc2-2 extract, (I) a mixture of 30 mg each of pri2-1 andcdc2-2 extracts or (J) a mixture of 30 mg each of cdc2-2


As demonstrated in Figure 8, below, initiationevents in our soluble replication system do notrequire an ARS consensus and occur ef®ciently onpGEM3Z(-). This raises the possibility that thereplication intermediates and the semi-conservativeDNA synthesis arise from recombination events,which have been shown to produce double Y arcson 2D gels from linear substrates (Lockshon et al.,1995; Preiser et al., 1996). To rule out that recombi-nation events are responsible for the RIs observedin our S-phase extracts, we prepared extracts froma rad52::TRP1 strain, which is de®cient for hom-ology-driven strand invasion events (Heyer &Kolodner, 1993). The RIs detected in rad52-de®cientstrains are identical to those in wild-type extracts(compare Figures 5B and 6E), indicating that theappearance of single and double Y arcs is not dueto Rad52p-mediated recombination.

To rule out that nick-translation events producethe observed RIs, we incubated plasmid DNA inyeast nuclear extract under standard conditions,but with the additions of both aphidicolin (to blockthe major yeast polymerases) and bacterial DNApol I (to promote nick-translation). Since the extractpromotes some nicking-closing reactions, weobserve a high level of label incorporation, but thisis exclusively in the 1n template (Figure 5C,exposure times are equivalent for all gels). Thus,aphidicolin blocks the formation of RIs and bac-terial DNA pol I can not produce equivalent struc-tures. In addition, single-stranded DNA (ssDNA)is shown to compete for the initiation of replicationon the template, whether it is a substoichiometriclevel of denatured pH4ARS (either linearised withEcoRI or digested with Hinf1) or full length M13ssDNA (Figure 5H). We ®nd that intact M13ssDNA, unlike linear ss pH4ARS, can serve as aweak substrate for second strand synthesis (seearrow, Figure 5H), although all forms of ssDNAef®ciently inhibit ds plasmid replication. This inhi-bition may re¯ect competition for limiting amountsof replication factor A (RPA), the eukaryoticssDNA binding factor, although in reactions with-out ssDNA competitor, the addition of bacterialSSB did not stimulate ds plasmid replication (datanot shown).

Replication in vitro requires DNA pol ddd and theDNA pol aaa/primase complex

The data presented up to this point indicate thatwe can detect semi-conservative DNA replicationin vitro, and that the biochemical properties of this

and GA-59 S-phase extracts. In all experiments, reactionswere performed at 37�C, with template and reactionconditions identical. G, H and J are two-day exposures,while I is a ®ve-day exposure, compensating for the factthat the levels of complementing enzymes are at halfthe levels found in the reactions shown in G and H. TheY arc present in the complementation experiment is vis-ible after a two-day exposure as well (data not shown).


reaction are consistent with a role for DNA pol a,d, and/or e. To demonstrate a requirement for thepol a/primase complex conclusively, we haveused two approaches. First, as mentioned above,we show that the incorporation of 32P is inhibitedby a monoclonal antibody speci®c for the largesubunit of DNA pol a (MAb24D9; Plevani et al.,1985; Figure 1E). An equivalent aliquot of a controlmouse anti-rabbit IgG has no effect (Figure 1E,lane 4). Coincident with the block, a short(�200 bp) radiolabelled DNA product accumu-lates, possibly representing abortive DNA syn-thesis. Control Western blots con®rm the stabilityof DNA pol a in our nuclear extracts for 90 min-utes at 25�C or 37�C, as well as that of DNA topoi-somerase II, Orc2p, and the 86 kDa subunit of thepol a complex under in vitro replication conditions(data not shown).

As a second, independent test of pol a/primase-dependence we prepared S-phase nuclear extractsfrom strains carrying temperature sensitive (ts)mutations in PRI1 and PRI2, the catalytic and regu-latory subunits of the yeast primase. In vivo thepri1-1 and pri2-1 mutations block the synthesis ofthe RNA primers used in leading and laggingstrand elongation (Boulet et al., 1989; Francesconiet al., 1991), arresting cells in S-phase at 37�C. Forthe comparison of such extracts, we ®rst optimisethe concentration of nuclear proteins used in eachreaction, which are performed in parallel usingidentical template, isotope dilutions and standar-dised exposure times for auto-radiograms. In con-trast to the wild-type extract in whichintermediates appear to accumulate at 37�C(Figure 6F), 2D-analysis of plasmid labelled in thepri1-1 extract shows a reduced level of RIs at per-missive temperature (Figure 6A) and an overalldrop in both ®nal product and intermediates atrestrictive temperature (Figure 6B). Similar resultswere obtained for the pri2-1 extract (Braguglia,1995).

To demonstrate involvement of the leadingstrand DNA pol d, encoded by the CDC2 gene, weprepared S-phase extracts from the ts cdc2-2 strainat permissive temperature, and used these for thereplication assay. In contrast to wild-type extracts,overall incorporation is higher at 25�C than at37�C, and RIs are no longer detectable at restrictivetemperature (Figure 6C and D). Since non-speci®cartefacts can arise from performing reactions at37�C, we demonstrate complementation of thede®ciencies by mixing two mutant extracts atrestrictive temperature in vitro. In this case, extractswere prepared from pri2-1 and cdc2-2 cells thatwere grown at 37 oC for four hours prior to nuclearisolation. Indeed, extracts from wild-type cellsgrown at elevated temperatures support replicationin vitro at elevated temperatures more reproducibly(Fig 6F; P.H. & B.P.D., unpublished observations).In reactions performed at 37�C with pri2-1 andcdc2-2 extracts (Figure 6G and H, respectively)incorporation into both RIs and the 1n fragment isreduced. Importantly, the appearance of RIs can be

recovered at 37�C by mixing these same extracts ina 1:1 ratio (Figure 6I; the exposure time is extendedto account for 50% pri2p and pol d protein concen-trations, see Figure legend). Complementationcould also be demonstrated between mutantextracts and a wild-type extract isolated aftergrowth at 37�C (shown for cdc2-2 and WT;Figure 6J). The fact that the temperature-induceddefect in the primase ts extract can be partiallycompensated by the pol d temperature-sensitiveextract validates the use of temperature-shiftassays and strongly implicates both pol a/primaseand pol d in the soluble replication reaction.

In vitro conversion of short DNA and RNA-containing fragments into long DNA

Lagging strand synthesis is achieved in vivo bythe ligation of short Okazaki fragments, whicheach initiate from an RNA primer (reviewed byKornberg & Baker, 1992). Okazaki fragments nor-mally have an estimated half-life of only two tofour seconds in vivo (Edenberg & Huberman,1975), but by reducing the ef®ciency of either theelongation and/or ligation steps under ATP-limit-ing conditions they can be accumulated and visual-ised (Friedman, 1974). To obtain further evidencethat the pol a/primase complex is active in ourin vitro reaction, we screened for the expectedradiolabelled 75 to 200 nt sized fragments by anal-ysis on a denaturing polyacrylamide/urea gel.

In replication reactions labelled with[a-32P]dATP and with limiting amounts of ATP,we detect labelled fragments on denaturing gelsbetween 75 and 150 nucleotides in length, whichare absent when either an excess of ATP or aphidi-colin is added to the reaction (Figure 7A, lanes 1 to3). We do not know if the limiting ATP concen-tration favours the accumulation of short frag-ments or simply enhances the speci®c activity ofthe [a-32P]dATP, increasing detection ef®ciency. Insome extracts we detect a second labelled speciesthat migrates at 75 nt that is neither aphidicolinnor DNase I-sensitive. This additional band comi-grates with radiolabelled tRNA, which has beenshown to become artefactually labelled by nucleo-tide transfer reactions in vitro (Manley, 1984).Importantly, the DNase I-sensitive band at 150 ntis also labelled in reactions containing [a-32P]UTPas the radioactive nucleotide (Figure 7A, lane 4),arguing that this fragment contains RNA, as wellas DNA. Consistently, the [a -32P]UTP labelling ofthe short fragments is sensitive to aphidicolin(Figure 7, lane 5), but insensitive to a-amanitin,which is added to all [a-32P]UTP-containing reac-tions. Other high molecular mass bands labelledwith [a-32P]UTP co-migrate with the L-species ofdsRNA (�4400 nucleotides, known as killer RNA;Hopper et al., 1977), and re¯ect an uncharacterised,aphidicolin and a-amanitin-resistant modi®cationof this molecule.

To test whether the short labelled fragments inour standard replication reaction are precursors to

Figure 7. Initiation starts from short primers whichcan be detected at early time points of the replicationreaction. A, 300 ng of pGEM3Z(-) was incubated at 25�Cin nuclear extract from GA-266 cells in the presence or


longer DNA chains, we used a pulse-chase label-ling protocol during the in vitro replication reaction(Figure 7B). During replication of pH 4ARS in thecontinuous presence of [a-32P]dATP we detectincorporation into small fragments at the earliesttime points (30 seconds; one, two, three minutes,etc.). On this denaturing gel, these fragments forma smear between 110 nt and 250 nt (trace shownFigure 7B). To see a precursor-product relationship,we added an excess of cold dATP after 30 secondsof labelling with radioactive dATP, and followedthe pulse-labelled DNA on denaturing gels. In thiscase the bands detected between 100 nt and 300 ntlessen in intensity with time, while longer mol-ecules accumulate, suggesting a precursor-productrelationship (Figure 7B). This is consistent with theinterpretation that the low molecular mass bandsrepresent Okazaki fragments.

Plasmid replication in yeast nuclear extracts iscell-cycle dependent

To test whether our soluble replication systemmaintains the S-phase dependence of replicationobserved in vivo, we prepared extracts from cdc4-1cells arrested in G1 at restrictive temperature, andcompared these with extracts from cells traversingS-phase. Cdc4p is required for the degradation ofthe Cdk1 inhibitor p40SIC1 which prevents the G1-S

absence of 1 mM ATP and 0.5 mg/ml aphidicolin(Aph), as indicated (see Materials and Methods). Thearrowhead indicates the accumulation of short DNAmolecules around 150 nucleotides in reactions contain-ing either [a-32P]dATP or [a-32P]UTP as radioactivelabel. In all reactions containing [a-32P]UTP unlabelledUTP was omitted and 10 mg/ml a-amanitin was addedto eliminate fragments arising from transcription.Extracted DNA was subjected to gel electrophoresis ona 10% acylamide, 7 M urea gel. An autoradiograph ofthe dried gel is shown. The star indicates a second frag-ment at 75 nt that behaves like the 150 nt fragment.Some extracts contain tRNAs that run at 75 nt whichbecome labelled by a nucleotide transfer reaction that isinsensitive to aphidicolin and ATP concentration(Manley, 1984), rendering conclusions about this frag-ment dif®cult. B, 300 ng of pGEM3Z(-) was preincu-bated for 90 minutes at 25�C in a GA-59 S-phase extract,as described in Materials and Methods. At the zero timepoint (00) [a-32P]dATP was added with all otherunlabelled nucleotides to both reactions. For the pulse-chase experiment, 2 mM unlabelled dATP was added 30seconds after this. Aliquots were taken from both reac-tions at the indicated time points, the DNA wasextracted and subjected to gel electrophoresis asdescribed in A. The intensity of label incorporationalong each lane was quantitated on a MolecularDynamics Phosphorimager; 100 is an arbitrary unit andis not 100%. Less than 5% of labelled species migratewith a fragment size <154 nucleotides at the 30 second(3000) timepoint of the pulse-chase experiment. The insetindicates the total [a-32P]dATP incorporation during thetime course experiment for either the continuous label-ling or the pulse-chase experiment.


transition (Schwob et al., 1994). CsCl gradient anal-ysis of DNA isolated following standard replica-tion reactions, demonstrates that S-phase extractssupport much higher levels of replication than G1-phase extracts (Figure 8, left panels). To demon-

Figure 8. DNA replication is stimulated by S-phase nuclreplication reactions including BrdUTP and [a-32P]dCTP ausing 40 mg of either an S-phase nuclear extract (wild-type GS-phase, labelled S) or a G1 nuclear extract from GA-851labelled G1). A third reaction was performed using a mixturanalysis was as described (see Materials and Methods) aheavy-light (H-L) and light-light (L-L) DNA are indicated bhand panels) 300 ng of pH4ARS was incubated for 90 minua-factor-blocked (G1) extracts, both isolated from wild-type30 mg of each extract. The DNA was puri®ed and separatAutoradiograms of the dried gels, exposed for two days, apresence of RIs, respectively.

strate that this is not simply due to degradative orinhibitory factors in the G1 extract, a reaction wascarried out with a 1:1 mixture of the two extracts(same total amount of protein as for single extractreactions). The combination resulted in replication

ear extracts. In the left-hand panels standard three hours a tracer, were carried out with 300 ng of pGEM3Z(-)

A-360 cells synchronised with a-factor and released into(cdc4-1) cells arrested at restrictive temperature (37�C;

e of 20 mg of each extract (S � G1). CsCl density gradientnd refractive indices corresponding to the migration ofy the arrowheads. In a separate series of reactions (right-tes with [a-32P]dATP, and 60 mg of either S-phase (S) or

GA-59 cells. (S � G1) indicates a reaction performed withed by 2D gel electrophoresis after digestion with EcoRI.re shown. Open and ®lled triangles indicate absence or


levels higher than for either extract alone, furtherunderscoring the stimulatory effects of S-phaseenriched factors. Among these is the Cdc28 kinasecomplexed to a B-type cyclin (Duncker et al.,unpublished results).

We next monitored the presence of replicationintermediates in reactions carried out in nuclearextracts from S-phase and G1-phase cells. No repli-cation intermediates are detected in the supercoiledtemplate after incubation in G1 extracts, even afterextended autoradiography (Figure 8, middle rightpanel; see Figure 5E for an explanation of DNAforms observed on 2D gels). As with the fully repli-cated products observed by gradient analysis, theabsence of replication intermediates is not due tofactors in the G1-phase extract that repress replica-tion in trans, since a mixture of S and G1-phaseextracts produces RIs identical to those in S-phase

Figure 9. Nuclear extracts prepared from dna2-1 cells at rreplication. A, Nuclear extracts were prepared from GA-81(23�C) or after two hours incubation at restrictive temperatperformed at 23�C for three hours, using 300 ng of pH4ARS23�C (left panel) or 54 mg of nuclear extract from cells at 37on CsCl density gradients; fractions containing fully substituDNA by L-L. Even less replication was observed with 30 m(dna2-1) and a congenic wild-type strain, GA-811 (WT), weformed at 23�C with 54 mg GA-810, 54 mg GA-811 or a mipeaks in the CsCl gradient are indicated by arrowheads. C,polymerase activity by measuring the incorporation of [a-32Ptemplate) at 30�C for 90 minutes as described in Materials anwith (light grey bars) or without (dark grey bars) the inclusadjusting the reaction to 0.5 M NaOH, incorporated [a-32P]dquanti®ed using Fuji®lm BAS Phosphor Imaging Plates. InAIDA 1.2.

extracts alone (Figure 8, compare top and bottomright panels).

In vitro replication requires the Dna2p helicase

DNA helicases perform essential functionsduring DNA replication by advancing the replica-tion fork through unwinding of DNA duplexes,and aiding in both Okazaki fragment processingand the repair of replicative errors. To date, noneof the known eukaryotic helicases have beenunambiguously assigned to promote any of theseprocesses, and the only yeast helicase that has beenshown to be essential for DNA replication is theDna2 protein (Budd et al., 1995). This enzyme hasbeen characterized as a 30 ! 50 DNA helicasespeci®c for forked substrates. Originally proposedto be the replication fork helicase, recent results

estrictive temperature are de®cient for semi-conservative0 (dna2-1) cells grown at either permissive temperatureure (37�C) prior to nuclear isolation. All reactions were

DNA, and either 30 mg of nuclear extract from cells at�C (right panel). BrdUTP-substituted DNA was analysedted DNA are indicated by H-L, and short patch repairedg of the 37�C extract. B, Nuclear extracts from GA-810re prepared at 37�C and replication reactions were per-xture of both extracts (27 mg each). H-L and L-L DNAThe nuclear extracts used in A and B were assayed for]dCTP into DNase I-treated calf thymus DNA (activatedd Methods. The indicated extract (40 mg) was used eitherion of 50 mg/ml aphidicolin. Reactions were stopped byCTP was bound to Hybond-N� nylon membranes, andcorporation of [a-32P]dCTP in pmol was quanti®ed by


suggest that Dna2p is more likely to be involved inlagging strand synthesis (Budd & Campbell, 1997;Fiorentino & Crabtree, 1997).

To study the effect of Dna2p on the in vitro repli-cation reaction, S-phase nuclear extracts from astrain (GA-810) with the ts dna2-1 allele were pre-pared at either permissive (23�C) or restrictive tem-perature (37�C). These were then used inreplication reactions carried out at permissive tem-perature, and DNA puri®ed following the reac-tions was analysed by means of CsCl densitygradients. While the dna2-1 extract prepared at23�C supports relatively high levels of semi-conser-vative DNA replication, the extract prepared fromthe same strain at 37�C reveals no detectable H-Lreplication peak (Figure 9A). The fact that nuclearextract from the congenic wild-type (WT) strain,prepared at 37�C, promotes DNA replication(Figure 9B) suggests a speci®c inactivation of themutant Dna2p in the dna2-37�C extract, rather thanlability of other factor(s) at the higher preparationtemperature. The presence of a non-speci®c domi-nant negative inhibitor could be ruled out be mix-ing equal amounts of dna2-37�C and WT-37�Cextracts, keeping the total protein amount thesame. Importantly, replication of the plasmid DNAwas even more ef®cient in the mixture than theWT-37�C extract alone (Figure 9B), which indicatesthat the dna2-37�C extract is not generally inactivefor replication. Rather, this result suggests that afunctional Dna2p is limiting in dna2-37�C extracts.As con®rmation that the dna2-37�C extract is notdefective for polymerase activity, assays were car-ried out using activated DNA (calf thymus DNAtreated with DNase I; Weiser et al., 1991) as a tem-plate (Figure 9C). dna2-1 extracts prepared at either23�C or 37�C reveal polymerase activity levelssimilar to that observed for a wild-type extract pre-pared at 37�C. In each case the nucleotide incor-poration could be inhibited by aphidicolin,indicating that we are monitoring levels of DNApol a, d, and/or e activity.

Discussion

As the ®rst step towards the complete reconstitu-tion of in vitro DNA replication, we describe ayeast system based on S-phase nuclear extractsthat replicates supercoiled plasmids in a semi-con-servative manner. We detect the aphidicolin-sensi-tive incorporation of [a-32P]dATP into non-linearreplication intermediates on 1D and 2D neutral-neutral gels. Complete second strand synthesis ismonitored by a density shift on CsCl gradientsafter substitution with the heavy nucleotide ana-logue BrdUTP, and conversion of methylated bac-terial plasmid DNA to a hemi-methylated form.Both the presence of replication intermediates anda lack of rolling circle or recombination intermedi-ates is con®rmed by electron microscopy. Finally,we demonstrate that both the pol a/primasecomplex and pol d are involved in the replication

reaction by a variety of methods, including tem-perature-sensitive replication in extracts fromstrains carrying conditional mutations in catalyticsubunits of these enzymes. The complementationof conditional replication mutations is shown usingnuclear extracts, but these studies can be extendedto test for complementation by puri®ed proteins.Here we use this system and a conditional mutantin DNA2 to demonstrate a direct role for this DNAhelicase in the semi-conservative replication of sctemplate. In contrast, DNA synthesis on plasmidscontaining nicks and/or gaps is not Dna2p-depen-dent. This is consistent with the S-phase arrest ofthe dna2-1 mutant, and implicates the Dna2 heli-case directly in DNA replication.

S-phase extracts stimulate replication, butprevent origin-dependent initiation

Semi-conservative replication of sc template iscompletely dependent on the addition of a nuclearextract from cells synchronised in S-phase, whileextracts from cells arrested in G1 are nearly inactivefor replication. Western blot analysis shows thatmost replicative enzymes are present in equalamounts in both extracts, but that the G1-phaseextract lacks an active cyclin B/Cdc28 complex,which is necessary to promote DNA replication ofsc template molecules (Duncker et al., unpublishedresults). In the S-phase extracts Cdc28p is active,and when p40SIC1, a G1-speci®c inhibitor of theClb/Cdc28 kinase complex, is added to an S-phaseextract, it becomes modi®ed. This may explain theef®cient replication observed in mixtures of S andG1-phase extracts (Figure 7).

Although we see a physiological S-phase depen-dence of the replication reaction, the initiationevent in vitro does not require the presence of anARS consensus, which is necessary for replicationin intact cells (see replication of pGEM3Z(-) inFigures 7 and 8). Indeed, the absence of a preciselypositioned bubble arc on 2D gels suggests thatinitiation is not precisely localised. Elsewhere wedemonstrate that initiation of replication on thesenaked templates occurs independently of ORC andCdc6p-activities, indicating that a pre-replicationcomplex (pre-RC) is not formed on our templateDNA prior to replication (Duncker et al., unpub-lished results). Ironically, the high Cdc28p kinaseactivity required for ef®cient replication may at thesame time prevent assembly of an initiation com-petent pre-RC at the ARS on naked plasmid DNA,consistent with genetic evidence showing that theactivation of the cyclin B/Cdk1 complex at theG1/S boundary both triggers initiation, and inacti-vates pre-RC formation (Piatti et al., 1996; Tanakaet al., 1997). To establish ORC and ARS-dependentreplication, current studies aim at assembling bothnucleosomes and the pre-RC on plasmids inG1-phase extracts, and subsequently activatingreplication with S-phase kinases.


Semi-conservative replication requires asupercoiled template

We show here that nicked or linear plasmidsserve as templates for highly ef®cient DNA syn-thesis in nuclear extracts, but they do not supportthe synthesis of a complete second strand. Indeed,the highest ef®ciency achieved in this in vitro reac-tion requires intact, supercoiled template. Eventhen, less than 1% of the input plasmid is con-verted to H-L DNA by semi-conservative incorpor-ation of BrdUTP. While low, this ef®ciency is notunlike that observed for plasmid templates inXenopus extracts (Blow & Laskey, 1986), in HeLacell extracts in the absence of T-antigen (Berberichet al., 1995), or the template usage observed inin vitro transcription assays (e.g. see Verdier et al.,1990). Although the BrdUTP substitution methodmay underestimate the optimal replication ef®-ciency, the high levels of incorporation on nickedor linear DNA indicates that DNA polymerases arenot limiting in our extracts. Rather, competition bysmall amounts of nicked or linear plasmid, as wellas by ssDNA, suggests that the successful initiationon sc template, which occurs preferentially atregions of helical instability (DNA unwindingelements or DUE, Umek & Kowalski, 1988;Braguglia, 1995), is the least ef®cient step in thisreaction. Consistently, we have observed that notonly the integrity of the DNA, but the topologicalstate of closed circular DNA, alters the ef®ciency oftemplate usage (Braguglia, 1995).

Our data suggest that the template DNA is criti-cal for determining the type of DNA synthesisachieved in vitro. This may re¯ect a qualitativedifference among the polymerases and/or helicasesthat assemble onto and replicate sc templates asopposed to those that repair nicked or linear DNA.Since both reactions are aphidicolin-sensitive, wepropose that the nature of the DNA helicase tar-geted to the initiation site is crucial for nucleatingeither repair or replicative DNA synthesis. Forinstance, proteins that recognise nicks or DNAends, such as the Ku70/Ku86 complex (reviewedby Lieber et al., 1997), may recruit a helicase thatcan only unwind the helix for short distances,resulting in short patch repair, while true replica-tion helicase(s) might interact with replication pro-tein A (RPA) and pol a at the A � T-rich regionsthat favour local strand separation, promotingsemi-conservative replication. We assume that acorrectly assembled pre-RC would be much moreef®cient at nucleating or promoting exclusively thislatter event.

Dna2 helicase is required for semi-conservative DNA replication in vitro

The results shown here using a ts dna2-1 extractimplicate the Dna2p helicase in semi-conservativeDNA replication, and its association with theFen1p (Rad27p) endonuclease suggests that Dna2pis primarily involved in lagging strand synthesis or

the processing of Okazaki fragments (Budd &Campbell, 1997). The fact that we observe a com-plete loss of semi-conservative DNA replication inthe dna2-1 mutant extract, rather than a residualleading strand synthesis, suggests that semi-conser-vative DNA replication requires a co-ordinationbetween leading and lagging strand polymerasecomplexes (Baker & Bell, 1998). This ``replisome''complex may be unable to form in the absence of acritical component, such as Dna2p, resulting in acomplete loss of plasmid replication, althoughpolymerases are active on nicked or gapped tem-plates in the dna2-1 extract (Figure 8). AlthoughDNA2 is an essential gene, it is unlikely to be theonly DNA helicase involved in genomic DNAreplication, and both the fractionation of theextracts described here, and the testing of strainsde®cient for other candidates (Ishimi, 1997; Luet al., 1996), should allow us to identify other heli-cases and con®rm their role in semi-conservativeDNA replication.

Bubble and Y arcs are thought to re¯ect replica-tion forks, rather than intermediates of recombina-tion (Lockshon et al., 1995), yet it remainedpossible that homologous recombination eventswere responsible for the initiation observed in thissystem. We therefore performed replication assaysin extracts from cells lacking Rad52p, a proteinessential for most homologous recombinationevents in yeast. Rad52p-de®cient extracts behaveidentically to those from RAD52� cells with respectto the appearance of replication intermediates andplasmid multimers during the reaction. Homolo-gous recombination, in any case, could not accountfor the bubble arc observed within a single circulartemplate (see Figure 5A). Finally, we also showthat incorporation into nicked template moleculesin the nuclear extract with or without Escherichiacoli DNA Pol I, fails to produce these intermedi-ates, and does not achieve semi-conservative repli-cation, as detected by BrdUTP substitution.

The significance of ARS-independent replication

The ARS-independent replication of naked plas-mid DNA in S-phase extracts is fully consistentwith the origin-independent replication observedwhen vectors are introduced into Xenopus eggs orinto human cultured cells (Mechali & Kearsey,1984). In contrast, when intact G1-phase yeastnuclei are introduced into these same extracts, weobserve origin-speci®c and ORC-dependent semi-conservative DNA replication in vitro (Pasero et al.,1997), consistent with genetic results showing thatORC and accessory factors must assemble ontotemplate DNA prior to S-phase. DUE sites mayfunction by promoting the binding of the single-strand DNA binding complex, RPA, which in turnmight target the pol a/primase complex(Dornreiter et al., 1992). Such a Cdk-dependentinitiation event, occurring after the release of ORCand Cdc6p, has been demonstrated in reconstituted


nuclei in Xenopus extracts (Hua & Newport, 1998).The events we observe in vitro, even if origin-inde-pendent, may thus re¯ect partial steps in the phys-iological initiation of DNA replication whichusually proceed from modi®cation of the pre-repli-cation complex at origins. Our in vitro replicationreaction, which is similarly stimulated by theCdc28 kinase (Duncker et al., unpublished results),should allow a convenient means to characterise itstargets among components of the replicationmachinery.

The soluble yeast replication system provides aquantitative assay for the enhancement of semi-conservative DNA replication, which may beachieved by purifying and adding back limitingcomponents, such as the replication helicases. Bytaking advantage of other ts mutants in genes withpoorly de®ned roles in replication (e.g. CDC45,CDC46, CDC47 or CDC7), we should be able todetermine the precise functions of these gene pro-ducts. Most importantly, this current system pro-vides a reproducible assay for monitoring thefunctional reconstitution of a pre-replication com-plex in vitro, such that the minimal elementsrequired for a controlled, origin-dependentinitiation event can be determined.

Materials and Methods

Yeast strains

The S. cerevisiae strain GA-59 (MATa, leu2, trp1, ura3-52, prb1-1122, pep4-3, prc1-407, gal2), lacking all the threemajor vacuolar proteases, was used as our standardwild-type strain. If not indicated otherwise, ``wild-type''refers to this strain. Additional wild-type strains usedwere GA-360 (MATa, trp1-1, ade2-1, ura3, leu2-3,112,HIS�, pep4::URA3) and GA-811 (MATa, prb1-1122, pep4-3, prc1-407, ura3-52, trp1-1), which is congenic to GA-810.For temperature-sensitive assays we used the relatedstrains GA-264 (MATa, pri2-1, ura3-52, ino1, can1,pep4::URA3), GA-266 (MATa, pri1-1, ura3-52, ino1, can1,pep4::URA3), GA-280 (MATa, cdc2-2, ura3-52, can1,pep4::URA3), kindly provided by P. Plevani (Universityof Milan) and GA-810 (MATa, prb1-1122, pep4-3, prc1-407, ura3-52, trp1-1, dna2-1). GA-851 (MATa, cdc4-1,bar1::LEU2, pep4::TRP1, ura3-52, leu2-3,112, trp1-289), wasused to obtain G1-phase extracts, and GA-500 (MATa,rad52::TRP1, bar1::LEU2, ade2-1, can1-100, his3-11,15, leu2-3,112, trp1-1, ura3-1), provided by W. Heyer (Universityof Bern), was used for rad52 de®cient extracts. The pep4yeast strains were regularly tested for carboxypeptidaseY activity (Jones, 1977).

bbb-1,3 Glucanase purification

b-1,3 Glucanase (lyticase) was puri®ed from E. coliharbouring the pUV5-G1S plasmid which contains the b-1,3 glucanase gene under the control of the lacUV5 pro-moter (Shen et al., 1991). Puri®cation is carried outaccording to Shen et al. (1991), with the following modi®-cations: osmotic shock is performed in 300 ml distilledwater for one hour on ice under constant stirring. Aftersedimentation of the cells at 14000 g for 15 minutes in aGS3 rotor, the released enzyme is recovered by precipi-tation with 40% (w/v) ammonium sulfate. These steps

are repeated twice. After 20 hours dialysis in 10 mMsodium-acetate (pH 5.0), units are estimated (Scott &Schekman, 1980), and the puri®ed enzyme is stored with0.005% (v/v) NaN3 at 4�C.

Synchronisation, isolation and extraction ofyeast nuclei

Yeast cells were grown in YPAD at 30�C to a densityof 0.5 to 1 � 107 cells/ml. Temperature-sensitive strainswere grown at 25�C (23�C where indicated) or 37�C forgrowth or arrest. For S-phase extracts, cells were har-vested and washed by centrifugation at 2000 g for ®veminutes at room temperature (rt), and were resuspendedin 0.25 the initial volume using YPAD (pH 5.0). a-Factorwas added to a ®nal concentration of 1.5 � 10ÿ7 M (forGA-59; optimal a factor concentrations must be deter-mined for each strain). Cells were incubated at 30�C (orat the appropriate permissive temperature) about 90minutes until no small buds were visible, indicating ablock at START (late G1). The cells were harvested asbefore and resuspended in the same volume of YPADpre-warmed to growth temperature. Release from theblock was checked by the appearance of budded cells.As soon as 80% of the cells displayed small buds (about45 minutes at 30�C), the cells were centrifuged andwashed in distilled water. Spheroplasting was describedby Verdier et al. (1990), and was normally completedwithin 15 minutes. It is critical that spheroplasting beperformed quickly to keep cells from advancing throughthe cell cycle.

For G1-phase extracts from GA-59 cells, a-factor waspresent during spheroplasting and no subsequent releasewas done. For G1-phase extracts from the cdc4-1 mutantstrain GA-851, cells were arrested in G1 for four hours at37�C, and were spheroplasted at 37�C. Where indicatedextracts from GA-810 and GA-811 were prepared at37�C, by shifting cells to 37�C for two hours includingthe time of a-factor release and spheroplasting.

For all extracts, crude nuclei were prepared andextracted by the addition of (NH4)2SO4 as described byVerdier et al. (1990). The extracted proteins were solubil-ized in a minimal volume of 20 mM Hepes-NaOH(pH 7.6), 10 mM MgSO4, 10 mM EGTA, 20% (v/v) gly-cerol, 3 mM DTT, 0.5 mM phenylmethylsulfonyl ¯uorideand dialyzed against the same buffer for 15 hours. Theextracts were clari®ed by centrifugation at 4�C for 15minutes at 14,000 rpm in an Eppendorf 5415C centrifuge,and were subsequently stored in liquid nitrogen.

In vitro replication reaction

Replication was carried out using either pGEM3Z(-) orpH4ARS (384 bp H4ARS Sau3AI fragment cloned intothe SmaI site of the polylinker) prepared according to astandard alkaline lysis method and puri®ed either overQiagen columns (Qiagen Inc.) or by banding on CsClgradients (Sambrook et al., 1989). The puri®ed plasmidswere at least 90% pure in their supercoiled form. The25 ml standard reaction mixture contained 2.5 ml of 10�replication buffer (120 mM Hepes-NaOH (pH 7.6),48 mM MgCl2, 3 mM EDTA-NaOH (pH 7.6), 6 mMDTT), 300 ng of supercoiled plasmid DNA, an energyregeneration system composed 40 mM creatine phos-phate, 0.125 mg/ml creatine kinase, 20 mM of dCTP,dGTP, dTTP, 8 mM dATP, 5 mCi/ml [a-32P]dATP, 20 mMof each ATP, CTP, GTP, UTP and 40 to 60 mg of nuclearprotein extract. Optimal protein concentration was deter-


mined by titration with each extract. In early reactions5% (w/v) polyethylene glycol 6000 was included, but asit did not improve replication, it was later routinelyomitted. Where indicated 1 mM ATP or either 500 or 50mg/ml aphidicolin dissolved in dimethyl sulfoxide(DMSO) was added. The same amount of DMSO wasadded to the control reactions. The replication reactionwas stopped after 90 minutes at 25�C (or the indicatedtemperature) by adjustment to 0.3 M sodium acetate(pH 5.2), 0.1% (w/v) SDS, 5 mM EDTA and 100 mg/mlProteinase K, followed by 30 minutes at 37�C. After add-ing tRNA carrier, the DNA was phenol extracted, preci-pitated, washed with 1 ml of 70% (v/v) ethanol, driedand resuspended in 20 ml of distilled water. A modi®edversion of this protocol was used for replication reactionsthat were subsequently analysed by CsCl density gradi-ents (see below). All other methods were performedaccording to standard protocols (Sambrook et al., 1989).

EM analysis of replicating DNA

pH4ARS DNA (300 ng) was incubated for 90 minutesat 25�C in a wild-type S-phase extract, as describedabove, but with the omission of [a-32P]dATP. Afterextraction with phenol/chloroform, the DNA was eitherlinearised with EcoRI or not, and was spread on a dis-tilled water hypophase (Davis et al., 1971) and trans-ferred on EM grids. The contrast was enhanced byplatinum rotary shadowing under a 6 to 8� angle vapori-sation using a Balzers apparatus. Approximately 3000molecules were visualised and photographed on a Phi-lips CM10 microscope and quanti®ed with a HP ScanJetscanner. Contour lengths of all replicating moleculeswere measured and analysed statistically to con®rm thecorrect size of the plasmid.

Cesium chloride density gradient centrifugation

For CsCl gradient analysis, replication reactions con-tained 200 mM of the heavy base analogue BrdUTP inplace of dTTP, along with 100 mM each of dCTP anddGTP, 2 mM dATP and [a-32P]dATP as a tracer. Whereindicated 40 mM dTTP was added to compete for theBrdUTP incorporation. Reactions were for three hours.Following phenol/chloroform extraction and removal ofunincorporated label using either Nuc-Trap (Stratagene)or Qiaquick (Qiagen) columns, 0.75 of the isolated DNAwas mixed with a CsCl solution to a ®nal density of1.7176 g/ml at 25�C (Z � 1.7176). The gradient was gen-erated at 35,000 rpm in a ®xed-angle T1270 rotor (Sor-vall) for 40 hours at 20�C. Fractions of 250 to 300 ml werecollected from the bottom of the gradient, counted in aPackard liquid scintillation counter and the refractiveindex of every second fraction was read. The gradientstypically span densities from 1.6742 g/ml (Z � 1.3970) to1.7936 (Z � 1.4080), and H-L DNA is usually recoveredat refractive index values between 1.4035 and 1.4040,while the unsubstituted DNA peaks at values between1.4000 and 1.4010. The remaining 0.25 of DNA isolatedfor each reaction was run on a 0.8% (w/v) agarose geland stained with ethidium bromide, to verify equalrecovery between samples.

DpnI resistance assays

DNA was extracted following a standard replicationreaction, linearised with EcoRI, which is not methylation-sensitive, and subsequently digested with either NdeII,

Sau3AI or DpnI for one hour at 37�C prior to gel electro-phoresis on a 1% agarose gel (Sanchez et al., 1992). NdeIIand Sau3AI digests were performed in the recommendedBoehringer Mannheim buffers, while the DpnI digestused 50 mM Tris-Cl, pH 7.5, 10 mM MgCl2, 1 mM DTTand 150 mM NaCl. Gels were dried and the label incor-poration was quanti®ed on a Molecular Dynamics Phos-phorimager.

Two-dimensional gel electrophoresis

Two-dimensional gel electrophoresis was essentiallycarried out as described by Brewer & Fangman (1987).Before loading the gel, the plasmid DNA was linearisedwith the indicated restriction enzyme. The ®rst dimen-sion gel was run in 0.4% agarose in 1� TBE (87 mMTris-borate, 2 mM EDTA) without ethidium bromide(EtBr) at 0.5 V/cm for 20 hours at room temperature.The gel slice was excised and a second gel (1% agarosein 1� TBE with 0.5 mg/ml EtBr) was poured around it.Migration was at 5 V/cm for seven hours at 4�C. The gelwas photographed, dried on Whatman 3MM paper andexposed by standard methods.

Detection and analysis of putative Okazaki fragments

All samples contained 300 ng of pGEM3Z(-), and thereactions containing [a-32P]dATP were performed understandard conditions at 25�C, in either the presence orabsence of an additional 1 mM ATP or 0.5 mg/ml aphi-dicolin. In reactions with [a-32P]UTP, unlabelled UTPwas omitted and all reactions contained 10 mg/ml a-amanitin. For the pulse-chase and time course exper-iments the reactions were carried out with a pre-incu-bation step of 90 minutes in an S-phase extract prior toaddition of nucleotides. At time zero [a-32P]dATP wasadded with other nucleotides, including 20 mM ATP, butlacking unlabelled dATP in both reactions and sampleswere taken at indicated times. For the pulse-chase exper-iment 2 mM of unlabelled dATP was added at one min-ute. The extracted DNA of the replication reaction wasdissolved in 10 ml of 90% (v/v) formamide, 20 mMEDTA, 10% glycerol, heated for ®ve minutes at 95�C andloaded on 10% polyacrylamide, 7 M urea gels and run at20 V/cm for 50 minutes in 1 � TBE. Gels were dried andthe label incorporation along each line was quanti®ed ona Molecular Dynamics Phosphorimager.

Polymerase assays on activated template

Polymerase assays were carried out essentially asdescribed for standard replication reactions, except that300 ng of DNase I-treated (activated) calf thymus DNAwas used as template (HuÈ bscher & Kornberg, 1979).40 mg extract was used in 90 minute reactions at 30�C.Polymerase activity was stopped by adjusting to 0.5 MNaOH and 0.2 of the volume was blotted on a chargednylon membrane (Hybond-N�) with a capillary slot blot.Radioactive blots were exposed on Fuji®lm BAS Phos-phorimaging Plates and incorporation of [a-32P]dCTPwas quantitated by the Fuji program AIDA 1.2.

Acknowledgements

We thank P. Plevani, W.-D. Heyer, K. Nasmyth, J.Campbell and L. Hartwell for yeast strains, P. Plevani


for MAb24D9 antibodies, and S. N. Slilaty for the pUV5-G1S E. coli plasmid expressing b-1,3,-glucanase. Discus-sions with Drs R. Laskey, P. Plevani, and U. HuÈ bscher,and critical reading of the manuscript by J. Lingner andE. Roberts are gratefully acknowledged. P.P. thanksARC, EMBO and the Roche Foundation, P.H. thanksBoehringer Ingelheim Fonds, and B.P.D. thanks ISRECfor support. Research in the Gasser laboratory is fundedby Swiss National Science Foundation, Human FrontiersScience Program and the Swiss Cancer League.

References

Baker, T. A. & Bell, S. (1998). Polymerases and the repli-some: machines within machines. Cell, 92, 295±305.

Berberich, S., Trivedi, A., Daniel, D. C., Johnson, E. M.& Leffak, M. (1995). In vitro replication of plasmidscontaining human c-myc DNA. J. Mol. Biol. 245,92±109.

Bielinsky, A. K. & Gerbi, S. A. (1998). Discrete start sitesfor DNA synthesis in the yeast ARS1 origin. Science,279, 95±98.

Blow, J. J. (1996). Chromosome Replication in Xenopusegg extracts. In Eukaryotic DNA Replication (Blow,J. J., ed.), pp. 143±165, IRL Press, New York.

Blow, J. J. & Laskey, R. A. (1986). Initiation of DNAreplication in nuclei and puri®ed DNA by a cell-free extract of Xenopus eggs. Cell, 47, 577±587.

Boulet, A., Simon, M., Faye, G., Bauer, G. A. & Burgers,P. M. (1989). Structure and function of the Saccharo-myces cerevisiae CDC2 gene encoding the large sub-unit of DNA polymerase III. EMBO J. 8, 1849±1854.

Braguglia, D. (1995). Development of an in vitro DNAreplication system in the yeast Saccharomyces cerevi-siae. PhD thesis, University of Lausanne.

Brewer, B. J. & Fangman, W. L. (1987). The localizationof replication origins on ARS plasmids in S. cerevi-siae. Cell, 51, 463±471.

Brewer, B. J., Sena, E. P. & Fangman, W. L. (1988). Anal-ysis of replication intermediates by two-dimensionalagarose gel electrophoresis. In Cancer Cells 6/Eukary-otic DNA Replication, pp. 229±234, Cold Spring Har-bor Laboratory Press, Cold Spring Harbor, NY.

Budd, M. E. & Campbell, J. L. (1997). A yeast replicativehelicase, Dna2 helicase, interacts with yeast FEN-1nuclease in carrying out its essential function. Mol.Cell. Biol. 17, 2136±2142.

Budd, M. E., Choe, W. C. & Campbell, J. L. (1995).DNA2 encodes a DNA helicase essential for replica-tion of eukaryotic chromosomes. J. Biol. Chem. 270,26766±26769.

Campbell, J. L. & Newlon, C. S. (1991). ChromosomalDNA Replication. In The Molecular and CellularBiology of the Yeast Saccharomyces cerevisiae: Gen-ome Dynamics, Protein Synthesis, and Energetics(Broach, J. R., Pringle, J. R. & Jones, E. W., eds),pp. 41±146, Cold Spring Harbor Laboratory Press,Cold Spring Harbor, NY.

Celniker, S. E. & Campbell, J. L. (1982). Yeast DNAreplication in vitro: initiation and elongation eventsmimic in vivo processes. Cell, 31, 201±213.

Coverley, D. & Laskey, R. A. (1994). Regulation ofeukaryotic DNA replication. Annu. Rev. Biochem. 63,745±776.

Davis, R. W., Simon, M. & Davidson, N. (1971). Electronmicroscope heteroduplex methods for mappingregions of base sequence homology in nucleic acid.Methods Enzymol. 21, 413±428.

Dif¯ey, J. F. (1996). Once and only once upon a time:specifying and regulating origins of DNA replica-tion in eukaryotic cells. Genes Dev. 10, 2819±2830.

Dornreiter, I., Erdile, L. F., Gilbert, I. U., von Winkler,D., Kelly, T. J. & Fanning, E. (1992). Interaction ofDNA polymerase alpha-primase with cellular repli-cation protein A and SV40 T antigen. EMBO J. 11,769±776.

Edenberg, H. J. & Huberman, J. A. (1975). Eukaryoticchromosome replication. Annu. Rev. Genet. 9, 245±284.

Fiorentino, D. F. & Crabtree, G. R. (1997). Characteriz-ation of Saccharomyces cerevisiae dna2 mutantssuggests a role for the helicase late in S phase. Mol.Biol. Cell, 8, 2519±2537.

Foiani, M., Lucchini, G. & Plevani, P. (1997). The DNApolymerase alpha-primase complex couples DNAreplication, cell-cycle progression and DNA-damageresponse. Trends Biochem. Sci. 22, 424±427.

Francesconi, S., Longhese, M. P., Piseri, A., Santocanale,C., Lucchini, G. & Plevani, P. (1991). Mutations inconserved yeast DNA primase domains impairDNA replication in vivo. Proc. Natl Acad. Sci. USA,88, 3877±3881.

Friedman, D. L. (1974). On the mechanism of DNAreplication in isolated nuclei from HeLa cells. Bio-chim. Biophys. Acta, 353, 447±462.

Gilbert, D. M. (1998). Replication origins in yeast versusmetazoa: separation of the haves and the have nots.Curr. Opin. Genet. Dev. 8, 194±199.

Gilbert, D. M., Miyazawa, H. & DePamphilis, M. L.(1995). Site-speci®c initiation of DNA replication inXenopus egg extract requires nuclear structure. Mol.Cell Biol. 15, 2942±2954.

Harland, R. M. & Laskey, R. A. (1980). Regulated repli-cation of DNA microinjected into eggs of Xenopuslaevis. Cell, 21, 761±771.

Heyer, W. D. & Kolodner, R. D. (1993). Enzymology ofhomologous recombination in Saccharomyces cerevi-siae. Prog. Nucl. Acid Res. Mol. Biol. 46, 221±271.

Hopper, J. E., Bostian, K. A., Rowe, L. B. & Tipper, D. J.(1977). Translation of the L-species dsRNA genomeof the killer-associated virus-like particles of Sac-charomyces cerevisiae. J. Biol. Chem. 252, 9010±9017.

Hsu, M. T. (1991). Visualization of novel simian virus 40DNA recombination intermediates induced byultraviolet light irradiation. Nucl. Acids Res. 19,7193±7199.

Hua, X. H. & Newport, J. (1998). Identi®cation of a prei-nitiation step in DNA replication that is indepen-dent of origin recognition complex and Cdc6, butdependent on Cdk2. J. Cell Biol. 140, 271±281.

Huberman, J. A., Nawotka, K. A., el-Assouli, S. M. &Davis, L. R. (1987). The in vivo replication origin ofthe yeast 2 micron plasmid. Cell, 51, 473±481.

HuÈ bscher, U. & Kornberg, A. (1979). The delta subunitof E. coli DNA pol III holoenzyme is the dnaX geneproduct. Proc. Natl Acad. Sci. USA, 76, 6284±6288.

Hyrien, O. & Mechali, M. (1992). Plasmid replication inXenopus eggs and egg extracts: a 2D gel electro-phoretic analysis. Nucl. Acids Res. 20, 1463±1469.

Ishimi, Y. (1997). A DNA helicase activity is associatedwith an MCM4, -6, and -7 protein complex. J. Biol.Chem. 272, 24508±24513.

Jazwinski, S. M. & Edelman, G. M. (1979). Replicationin vitro of the 2-micrometer DNA plasmid of yeast.Proc. Natl Acad. Sci. USA, 76, 1223±1227.

Jones, E. W. (1977). Proteinase mutants of Saccharomycescerevisiae. Genetics, 85, 23±33.


Jong, A. Y. & Scott, J. F. (1985). DNA synthesis in yeastcell-free extracts dependent on recombinant DNAplasmids puri®ed from E. coli. Nucl. Acids Res. 13,2943±2958.

Kojo, H., Greenberg, B. D. & Sugino, A. (1981). Yeast 2-micrometer plasmid DNA replication in vitro: originand direction. Proc. Natl Acad. Sci. USA, 78, 7261±7265.

Kornberg, A. & Baker, T. A. (1992). DNA Replication,2nd edit., W. H. Freeman and Company, NewYork.

Krude, T., Jackman, M., Pines, J. & Laskey, R. A. (1997).Cyclin/Cdk-dependent initiation of DNA replica-tion in a human cell-free system. Cell, 88, 109±119.

Laird, C. D. & Bodmer, W. F. (1967). 5-Bromouracil util-ization by Bacillus subtilis. J. Bacteriol. 94, 1277±1278.

Lieber, M. R., Grawunder, U., Wu, X. T. & Yaneva, M.(1997). Tying loose ends: roles of Ku and DNA-dependent protein kinase in the repair of double-strand breaks. Curr. Opin. Gen. Dev. 7, 99±104.

Linskens, M. H. & Huberman, J. A. (1990). Ambiguitiesin results obtained with 2D gel replicon mappingtechniques. Nucl. Acids Res. 18, 647±652.

Lockshon, D., Zweifel, S. G., Freeman-Cook, L. L.,Lorimer, H. E., Brewer, B. J. & Fangman, W. L.(1995). A role for recombination junctions in thesegregation of mitochondrial DNA in yeast. Cell, 81,947±955.

Lu, J. A., Mullen, J. R., Brill, S. J., Kleff, S., Romeo, A. M.& Sternglanz, R. (1996). Human homologues ofyeast helicase. Nature, 383, 678±679.

Mahbubani, H. M., Paull, T., Elder, J. K. & Blow, J. J.(1992). DNA replication initiates at multiple sites onplasmid DNA in Xenopus egg extracts. Nucl. AcidsRes. 20, 1457±1462.

Malezka, R., Skelly, P. J. & Clark-Walker, G. D. (1991).Rolling circle replication of DNA in yeast mitocho-dria. EMBO J. 10, 3923±3929.

Manley, J. L. (1984). Transcription and Translation: A Prac-tical Approach (Hamer, B. D. & Higgins, S. J., eds),pp. 71±88, IRL, Washington, DC.

Martin-Parras, L., Hernandez, P., Martinez-Robles, M. L.& Schvartzman, J. B. (1992). Initiation of DNA repli-cation in ColE1 plasmids containing multiple poten-tial origins of replication. J. Biol. Chem. 267, 22496±22505.

Mechali, M. & Kearsey, S. (1984). Lack of speci®csequence requirement for DNA replication in Xeno-pus eggs compared with high sequence speci®cityin yeast. Cell, 38, 55±64.

Pasero, P. & Gasser, S. M. (1998). New systems for repli-cating DNA in vitro. Curr. Opin. Cell Biol. 10, 304±310.

Pasero, P., Braguglia, D. & Gasser, S. M. (1997). ORC-dependent and origin-speci®c initiation of DNAreplication at de®ned foci in isolated yeast nuclei.Genes Dev. 11, 1504±1518.

Piatti, S., BoÈhm, T., Cocker, J. H., Dif¯ey, J. F. X. &Nasmyth, K. (1996). Activation of S-phase-promot-ing CDKs in late G1 de®nes a ``point of no return''

after which Cdc6 synthesis cannot promote DNAreplication in yeast. Genes Dev. 10, 1516±1526.

Plevani, P., Foiani, M., Valsasnini, P., Badaracco, G.,Cheriathundam, E. & Chang, L. M. (1985). Polypep-tide structure of DNA primase from a yeast DNApolymerase-primase complex. J. Biol. Chem. 260,7102±7107.

Preiser, P. R., Wilson, R. J., Moore, P. W., McCready, S.,Hajibagheri, M. A., Blight, K. J., Strath, M. &Williamson, D. H. (1996). Recombination associatedwith replication of malarial mitochondrial DNA.EMBO J. 15, 684±693.

Sambrook, J., Fritsch, E. F. & Maniatis, T. (1989). Molecu-lar Cloning, A Laboratory Manual, 2nd edit., ColdSpring Harbor Laboratory Press, Cold Spring Har-bor, NY.

Sanchez, J. A., Marek, D. & Wangh, L. J. (1992). The ef®-ciency and timing of plasmid DNA replication inXenopus eggs: correlations to the extent of priorchromatin assembly. J. Biol. Chem. 267, 25786±25791.

Schwob, E., Bohm, T., Mendenhall, M. D. & Nasmyth,K. (1994). The B-type cyclin kinase inhibitor p40SIC1

controls the G1 to S transition in S. cerevisiae. Cell,79, 233±244.

Scott, J. H. & Schekman, R. (1980). Lyticase: endogluca-nase and protease activities that act together inyeast cell lysis. J. Bacteriol. 142, 414±423.

Shen, S. H., Chretien, P., Bastien, L. & Slilaty, S. N.(1991). Primary sequence of the glucanase genefrom Oerskovia xanthineolytica. Expression and puri-®cation of the enzyme from Escherichia coli. J. Biol.Chem. 266, 1058±1063.

Stillman, B. (1996). Cell cycle control of DNA replica-tion. Science, 274, 1659±64.

Sundin, O. & Varshavsky, A. (1980). Terminal stages ofSV40 DNA replication proceed via multiply inter-twined catenated dimers. Cell, 21, 103±114.

Tanaka, T., Knapp, D. & Nasmyth, K. (1997). Loading ofan Mcm protein onto DNA replication origins isregulated by Cdc6p and CDKs. Cell, 90, 649±660.

Umek, R. M. & Kowalski, D. (1988). The ease of DNAunwinding as a determinant of initiation at yeastreplication origins. Cell, 52, 559±567.

Verdier, J. M., Stalder, R., Roberge, M., Amati, B.,Sentenac, A. & Gasser, S. M. (1990). Preparationand characterization of yeast nuclear extracts foref®cient RNA polymerase B (II)-dependent tran-scription in vitro. Nucl. Acids Res. 18, 7033±7039.

Walter, J., Sun, L. & Newport, J. (1998). Regulated chro-mosomal DNA replication in the absence of anucleus. Mol. Cell, 1, 519±529.

Wang, J. C., Caron, P. R. & Kim, R. A. (1990). The roleof DNA topoisomerases in recombination and gen-ome stability: a double-edged sword? Cell, 62, 403±406.

Weiser, T., Gassmann, M., Thommes, P., Ferrari, E.,Hafkemeyer, P. & Hubscher, U. (1991). Biochemicaland functional comparison of DNA polymerases a,d, and e from calf thymus. J. Biol. Chem. 266,10420±10428.

Edited by M. Yaniv

(Received 5 January 1998; received in revised form 18 May 1998; accepted 27 May 1998)

Documents

Semi-conservative Replication in Yeast Nuclear Extracts ... JMB... · semi-conservative DNA replication, and initiation is both origin-speciﬁc and ORC-dependent. Repli-cation was