Correction

BIOCHEMISTRYCorrection for “Phosphorylation of CMG helicase and Tof1 isrequired for programmed fork arrest,” by Deepak Bastia, PankajSrivastava, Shamsu Zaman, Malay Choudhury, Bidyut K. Mohanty,Julien Bacal, Lance D. Langston, Philippe Pasero, and Michael E.O’Donnell, which appeared in issue 26, June 28, 2016, of ProcNatl Acad Sci USA (113:E3639–E3648; first published June 13,2016; 10.1073/pnas.1607552113).The authors note that the affiliation for Julien Bacal and

Philippe Pasero should instead appear as Institute of HumanGenetics, CNRS, 34396 Montpellier, France. The correctedauthor and affiliation lines appear below. The online versionhas been corrected.

Deepak Bastiaa, Pankaj Srivastavaa, Shamsu Zamana, MalayChoudhurya, Bidyut K. Mohantya, Julien Bacalb, Lance D.Langstonc,d, Philippe Paserob, and Michael E. O’Donnellc,d

aDepartment of Biochemistry and Molecular Biology, Medical University ofSouth Carolina, Charleston, SC 29415; bInstitute of Human Genetics, CNRS,University of Montpellier, 34396 Montpellier, France; cHoward HughesMedical Institute, Rockefeller University, New York, NY 10065;and dLaboratory of DNA Replication, Rockefeller University, NewYork, NY 10065

www.pnas.org/cgi/doi/10.1073/pnas.1610262113

www.pnas.org PNAS | August 2, 2016 | vol. 113 | no. 31 | E4577

CORR

ECTION

Dow

nloa

ded

by g

uest

on

Dec

embe

r 14

, 202

0 D

ownl

oade

d by

gue

st o

n D

ecem

ber

14, 2

020

Dow

nloa

ded

by g

uest

on

Dec

embe

r 14

, 202

0 D

ownl

oade

d by

gue

st o

n D

ecem

ber

14, 2

020

Dow

nloa

ded

by g

uest

on

Dec

embe

r 14

, 202

0 D

ownl

oade

d by

gue

st o

n D

ecem

ber

14, 2

020

Dow

nloa

ded

by g

uest

on

Dec

embe

r 14

, 202

0 D

ownl

oade

d by

gue

st o

n D

ecem

ber

14, 2

020

Dow

nloa

ded

by g

uest

on

Dec

embe

r 14

, 202

0 D

ownl

oade

d by

gue

st o

n D

ecem

ber

14, 2

020

Dow

nloa

ded

by g

uest

on

Dec

embe

r 14

, 202

0 D

ownl

oade

d by

gue

st o

n D

ecem

ber

14, 2

020

Phosphorylation of CMG helicase and Tof1 is required forprogrammed fork arrestDeepak Bastiaa,1, Pankaj Srivastavaa,2, Shamsu Zamana,2, Malay Choudhurya,2, Bidyut K. Mohantya,2, Julien Bacalb,3,Lance D. Langstonc,d, Philippe Paserob, and Michael E. O’Donnellc,d,1

aDepartment of Biochemistry and Molecular Biology, Medical University of South Carolina, Charleston, SC 29415; bInstitute of Human Genetics,CNRS, University of Montpellier, 34396 Montpellier, France; cHoward Hughes Medical Institute, Rockefeller University, New York, NY 10065;and dLaboratory of DNA Replication, Rockefeller University, New York, NY 10065

Contributed by Michael E. O’Donnell, May 12, 2016 (sent for review May 9, 2016; reviewed by Satya Prakash and Robert A. Sclafani)

Several important physiological transactions, including control ofreplicative life span (RLS), prevention of collision between replicationand transcription, and cellular differentiation, require programmedreplication fork arrest (PFA). However, a general mechanism of PFAhas remained elusive. We previously showed that the Tof1–Csm3fork protection complex is essential for PFA by antagonizing theRrm3 helicase that displaces nonhistone protein barriers that impedefork progression. Here we show that mutations of Dbf4-dependentkinase (DDK) of Saccharomyces cerevisiae, but not other DNA repli-cation factors, greatly reduced PFA at replication fork barriers in thespacer regions of the ribosomal DNA array. A key target of DDK isthe mini chromosome maintenance (Mcm) 2–7 complex, which isknown to require phosphorylation by DDK to form an active CMG[Cdc45 (cell division cycle gene 45), Mcm2–7, GINS (Go, Ichi, Ni, andSan)] helicase. In vivo experiments showed that mutational inacti-vation of DDK caused release of Tof1 from the chromatin fractions.In vitro binding experiments confirmed that CMG and/or Mcm2–7had to be phosphorylated for binding to phospho-Tof1–Csm3 butnot to its dephosphorylated form. Suppressor mutations that by-pass the requirement for Mcm2–7 phosphorylation by DDK re-stored PFA in the absence of the kinase. Retention of Tof1 in thechromatin fraction and PFA in vivo was promoted by the suppres-sor mcm5-bob1, which bypassed DDK requirement, indicating thatunder this condition a kinase other than DDK catalyzed the phos-phorylation of Tof1. We propose that phosphorylation regulatesthe recruitment and retention of Tof1–Csm3 by the replisome andthat this complex antagonizes the Rrm3 helicase, thereby promot-ing PFA, by preserving the integrity of the Fob1–Ter complex.

DDK | Ter | Fob1 | Tof1 | programmed fork arrest

Replication fork arrest can either occur at random sites (1)or, in some cases, is physiologically programmed to occur at

specific sequences such as the Ter sites of Escherichia coli andSaccharomyces cerevisiae (also called replication fork barriers;RFBs), which bind to specific terminator proteins. The proteinsTus and Fob1 bind to the Ter sites of E. coli and S. cerevisiae,respectively, and cause polar replication fork arrest (2, 3) (re-viewed in ref. 4). The arrested fork, with the help of DNA po-lymerase, helicase, and topoisomerase (5), merges with the forkapproaching Ter from the opposite direction, and the resultingcatenated daughter molecules are resolved by topoisomerase IVin bacteria (6) and topoisomerase II in eukaryotes (7, 8). In bud-ding yeast, the unloading of the replisomal components from thetemplate after replication termination is facilitated by Cdc48 and aubiquitin ligase (9).Eukaryotic cells of diverse types arrest and prevent replication

fork entry into the highly transcribed ribosomal (r)DNA locus(reviewed in ref. 4). In budding yeast, a single replication ter-minator protein called Fob1 binds to the Ter sites (RFBs) lo-cated in nontranscribed spacer 1 (NTS1) of rDNA to cause polarfork arrest (10, 11), which prevents transcription–replicationcollision, fork stalling, and genome instability. Fission yeast has threeknown terminator proteins: Rtf1, which functions at the mating-type

switch locus (12, 13), and Reb1 (Sp. Reb1) and Sap1, which act atthe Ter sites of rDNA spacers (14–18). Sp. Reb1 also functions atTer sites located in the other two chromosomes and promotes“chromosome kissing” (19).Several studies have shown that a complex of three proteins

(Tof1, Csm3, and Mrc1) called the fork protection complex(FPC) acts at replication forks to ensure normal fork progressionand stabilization of forks stalled at nonhistone protein barriersand at sites of DNA damage (20). We and others have shownthat two members of the heterotrimeric FPC are needed forstable fork arrest at Ter sites and at other nonhistone proteinbarriers. All eukaryotic cells contain homologs of these twoproteins, called Tof1 and Csm3 of budding yeast (21–23), Swi1 andSwi3 of fission yeast (17, 24), and TIM and TIPIN of mammalian cells(25, 26). The third member of the complex, called Mrc1 and claspinin budding and fission yeast and mammalian cells, respectively, isdispensable for maintenance of programmed replication fork arrest(PFA) (22, 23). The FPC binds to mini chromosome maintenance(Mcm)2-7 and inhibits its DNA-dependent ATPase activity and alsothe helicase activity of the CMG complex (27). We have previouslyreported that Tof1 and Csm3 promote stable fork arrest by antagonizingthe Rrm3 helicase (23, 28), which removes nonhistone protein

Significance

Programmed replication fork arrest (PFA) at specific terminatorsites and the proteins that bind to these sites functionally in-terconnect replication, transcription, and recombination. PFA pre-vents collision between replication and transcription that cancause genome instability, promotes intrachromatid recombinationat ribosomal (r)DNA that controls replicative life span, and main-tains rDNA homeostasis. This work reveals the mechanism of PFAby showing that the Tof1 protein of budding yeast remains as-sociated with the replication fork only when it is phosphorylated,and uses the CMG helicase that drives the replication fork as alanding pad. Tof1–Csm3 promotes PFA by preventing the termi-nator protein Fob1 from getting displaced by other factors suchas the Rrm3 helicase by counteracting the latter. This importantmechanism appears to be evolutionarily conserved.

Author contributions: D.B., B.K.M., L.D.L., and P.P. designed research; P.S., S.Z., M.C.,B.K.M., and J.B. performed research; L.D.L., P.P., and M.E.O. contributed new reagents/analytic tools; D.B., P.S., S.Z., M.C., B.K.M., J.B., L.D.L., P.P., and M.E.O. analyzed data; andD.B., L.D.L., and M.E.O. wrote the paper.

Reviewers: S.P., University of Texas Medical Branch at Galveston; and R.A.S., University ofColorado School of Medicine.

The authors declare no conflict of interest.1To whom correspondence may be addressed. Email: bastia@musc.edu or odonnel@rockefeller.edu.

2P.S., S.Z., M.C., and B.K.M. contributed equally to this work.3Present address: Department of Chemical and Systems Biology, Stanford UniversitySchool of Medicine, Stanford, CA 94305.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1607552113/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1607552113 PNAS | Published online June 13, 2016 | E3639–E3648

BIOCH

EMISTR

YPN

ASPL

US

barriers from in front of moving forks throughout the budding yeastgenome (29).PFA promotes several important physiological functions. In

E. coli, two sets of Ter sites of opposite polarity are located at theantipode (with respect to the origin) of the circular chromosome,forming a replication trap. Fork convergence occurs within thistrap, at or near the dimer resolution and segregation sites (calleddif) that are necessary for reduction of the daughter chromo-somes to monomers and their proper separation preceding celldivision. Ter sites also prevent a topological shift from a Cairns-type to a rolling-circle replication, thereby contributing to genomestability (reviewed in ref. 4). In eukaryotes, PFA is required for(i) prevention of collision between a replication fork and an ac-tively transcribing RNA polymerase approaching from the op-posite direction, thereby preventing genome instability (30–33),(ii) promoting genetic imprinting and cellular differentiationin fission yeast (12), and (iii) controlling replicative life span inbudding yeast (34, 35).The two cell-cycle kinases called cyclin-dependent kinase (CDK)

and Dbf4-dependent kinase (DDK) are required for replicationinitiation (36–39). CDK (40) but not DDK (41, 42) is involved inregulating the S-phase checkpoint. Replication initiation is a two-step process that involves the assembly of a pre-replication complex(RC) through recruitment of Mcm2–7 by an origin-bound Orc1–6complex with the help of the Cdt1 and Cdc6 proteins. The inactiveMcm2–7 is then phosphorylated by DDK, which initiates a multi-step assembly process involving the heterotetrameric GINS (Go,Ichi, Ni, and San) complex (43) and Cdc45 (cell division cycle gene45) (44, 45) to generate the CMG complex, which has vigoroushelicase activity (46–48). CDK and DDK also restrict DNA repli-cation to only once per cell cycle (49).In budding yeast, the requirement for DDK can be bypassed

by a suppressing point mutation called mcm5-bob1 (50). TheN-terminal region of Mcm4 contains an inhibitory domain that isneutralized by DDK-catalyzed phosphorylation. Consequently,deletion of this domain bypasses the need for DDK for CMGassembly (51–53). Recently, it was shown that RIF1, which controlsthe timing of origin firing in S. cerevisiae and recruitment of thePP1 phosphatase, causes dephosphorylation of the residues ofMcm2–7 that are phosphorylated by DDK. Deletion of RIF1suppresses the cdc7-1 mutation and allows growth to occur atthe semipermissive temperature of 30 °C (54).The present work shows that in vitro phosphorylation of not

only Tof1 but also Mcm2–7 by DDK is essential for recruitment/retention of Tof1 at the replisome, which is obligatory for PFA.Consistent with this observation, mutational inactivation of DDKin vivo caused either the release of Tof1 or its failure to beretained in the chromatin fractions. The apparent need for DDKto cause PFA was bypassed by suppressor mutations that bypassMcm2–7/CMG function. Together, these results indicate thatin addition to the known role of DDK in CMG assembly andhelicase activation, it also contributes to the assembly of theCMG platform that recruits phosphorylated but not dephos-phorylated Tof1 to the replisome. Together with our earlier resultsthat Tof1–Csm3 antagonizes the Rrm3 helicase, which displacesFob1 from Ter sites, we propose that phosphorylation of Tof1 andMcm2–7 promotes stable association of the FPC with the repli-some and is needed for functional PFA.

ResultsInactivation of DDK Caused Significant Reduction of PFA.Does DDK,besides its roles in replication initiation, limiting replicationto once per cell cycle, and checkpoint control, also regulatePFA? We addressed this question by investigating replicationfork arrest at the Ter site of the rDNA of budding yeast intemperature-sensitive mutants of DDK called cdc7-1 and dbf4-1of its two subunits at the permissive (24 °C) and nonpermissive tem-peratures (37 °C). We used Brewer–Fangman 2D gel electrophoresis

(55), as modified (23), to follow fork progression in the spacer re-gions of rDNA. Our stratagem was to let initiation occur at 24 °Cand then allow fork progression at 37 °C, which should prevent anynew origins from firing. Asynchronous cultures were grown at 24 °Cto midlog phase and then shifted to 37 °C for various time periods(1–2 h) and the cells were harvested. This approach, in our hands,resulted in higher yields of scarce replication intermediates, partic-ularly from the mutant cells. Control experiments were similarlyperformed at 24 °C but without a shift to 37 °C. FACS analyses ofthe asynchronous cultures used for the experiments are shown(Fig. S1). It should be noted that although all of the 2D gelsshowed Y-shaped arc (Y-arc), the pattern corresponding to bub-ble arc was not observed because the experiments were notdesigned to capture and display the early replication intermedi-ates. Fig. 1A shows the topography of NTS1 containing a single arsand tandem Ter sites present in each rDNA repeat. Replicationfork arrest at the twin Ter sites, as revealed by five independentsets of 2D gels, as expected showed normal fork arrest at Ter inWT cells at both temperatures (Fig. 1B), but there was a signifi-cant reduction in the relative intensity of the Ter spots at the

Fig. 1. Replication termination in theWT and cells with temperature-sensitivemutants of DDK. (A) Schematic diagram of the rDNA repeats showing thelocations of the Ter (RFBs) and ars sites. Blue arrows show the direction ofrRNA transcripts; red arrows show the direction of replication fork pro-gression. (B–D) Images of Brewer–Fangman 2D gels showing replication forkprogression (Y-arc) and the fork arrest at the twin Ter sites (arrows). The cellswere grown for 4 h at 24 °C to reach early log phase, transferred to 37 °C for2–3 h, and then rapidly chilled and treated with Na azide as described in thetext and replication intermediates were prepared. Note that whereas forkarrest occurred in the WT cells at both temperatures, it was greatly reducedin the cdc7-1 and dbf4-1 cells at 37 °C.

E3640 | www.pnas.org/cgi/doi/10.1073/pnas.1607552113 Bastia et al.

nonpermissive temperature (Fig. 1 C and D) in cdc7-1 and dbf4-1cells, as determined by quantification of the images using theNIH ImageJ program. We concluded from these experimentsthat temperature-dependent inactivation of either the cdc7-1 ordbf4-1 component of DDK significantly reduced the magnitude ofPFA at the Ter sites. Is unhindered fork progression through aprotein barrier a general property of temperature-sensitive (ts)mutations at the nonpermissive temperature in several replicationproteins, or is it specific to DDK? Is this caused by direct action ofDDK on replication termination, or is it an indirect and relativelytrivial consequence of changes in the rate of fork progression as aresult of loss of DDK activity that might have caused fewer forksto reach the Ter sites? Was this possibly caused by an increase inthe rate of fork progression that somehow overpowered theFob1–Ter barriers? Alternatively, could this have been causedby the displacement of Fob1 from Ter sites precipitated by theloss of DDK activity? These questions have been systematicallyaddressed below.

Abolition of PFA Is Not a General Property of Several Mutant Forms ofProteins Known to Be Defective in Replication Initiation. We in-vestigated whether the loss of fork arrest upon inactivation of DDKwas a specific or general defect caused by reduction in the rates ofinitiation and ongoing replication by ts mutations in other repli-cation proteins. Two-dimensional gel analyses of fork movementusing several ts mutants in proteins known to be involved in initi-ation (e.g., psf1-1, psf2-1, sld2-2, sld5-12, and mcm10-1) showedthat PFA at Ter sites in these mutants was not significantly di-minished at 37 °C in comparison with that of the WT cells (Fig. 2Aand Fig. S2). As an example, the psf-1 mutant is known to benonleaky (43). Published 2D gel analyses of its replication inter-mediates clearly show significant reduction of the bubble arc in 2Dgels in comparison with that of the WT, clearly indicative of loss ofreplication initiation at 37 °C (43). Therefore, loss of PFA is not aninevitable consequence of a loss of initiation function.

DNA Combing Showed No Correspondence Between Replication ForkArrest and the Relative Differences in Fork Movement. To investigatewhether a relative difference in the rate of fork progression in thetwo mutants compared with that ofWT cells could cause PFA or itsabolition, we performed DNA combing experiments with the WT,cdc7-1, and psf1-1 cells at the nonpermissive temperature (Fig. 2B).The cells of the three genotypes were first grown at 24 °C, shiftedto 37 °C, and pulse-labeled with BrdU. The DNA was gentlyextracted to minimize hydrodynamic shear and spread on silanizedcoverslips and visualized by antibodies against BrdU (green) andsingle-stranded DNA (red). Several hundred molecules wererecorded under a fluorescence microscope, tract lengths weremeasured, statistical analyses were performed (Table 1), and thedata were plotted as described in Materials and Methods and thelegend to Fig. 2. The relative differences in the rates of forkmovement were small but consistent, with the magnitudes as fol-lows: psf1-1 > cdc7-1 >WT (Fig. 2 B–D). Keeping in mind that forkarrest at Ter occurred in psf1-1 and WT cells but not in cdc7-1 at37 °C, the data suggested that there was no detectable correlationbetween the relative rate of fork progression and fork arrest at Ter.It should be noted that the psf1 and WT tracts are longer in theDNA combing profile because initiation was blocked in cdc7-1 andpsf1-1 cells, and those that initiated, presumably because of lesscompetition for existing initiation factors, cause somewhat fasterfork movement (56).

Failure to Arrest Forks in cdc7-1 Cells at 37 °C Was Not Due toDisplacement and Loss of Fob1 from the Ter Sites. We wished to in-vestigate whether the absence of DDK activity caused Fob1 todissociate from the Ter sites by measuring the magnitudes of Fob1occupancy at Ter in the cdc7-1 strains at 24 °C, 30 °C, and 37 °C for3 h. ChIP analyses of Fob1 binding to Ter in vivo were performed

using a segment of DNA from the 35S transcribing region as thecontrol. The data showed that Fob1 remained bound to Ter sites incdc7-1 cells at all temperatures tested (Fig. 2 E and F).

Key Role for DDK-Catalyzed Phosphorylation in PFA. It is known thatthe requirement for DDK-catalyzed phosphorylation of Mcm2–7for cell survival can be bypassed by several different suppressormutations. For example, cdc7-1 or a cdc7Δ deletion can be sup-pressed by the mcm5-bob1 suppressor, which maintains cell via-bility in the absence of DDK (50). We compared and contrasted

Fig. 2. DNA combing to measure the rate of replication fork progression inthe WT and cdc7-1 and psf1-1 mutants at 37 °C and ChIP analysis of Fob1 oc-cupancy at Ter. (A) Two-dimensional gel analysis of psf1-1 DNA. (B) Represen-tative images of the DNA pulse-labeled with BrdU and combed, showing theextent of replication during the pulse period in WT, cdc7-1, and psf1-1 cells.(C and D) Data analysis of BrdU tract length in WT, cdc7-1, and psf1-1 cells. Thecells were grown overnight at 24 °C and shifted to 37 °C for 15 or 30 min inthe presence of 200 μg/mL BrdU. DNA fibers were stretched on silanizedcoverslips by DNA combing and BrdU incorporation was visualized by immu-nofluorescence. DNA fibers from WT, cdc7-1, and psf1-1 cells after 15 min ofBrdU pulse at 37 °C. Green, anti-BrdU; red, anti-ssDNA. (Scale bar, 20 kb.) Thebottom track of each of the 6 pairs of tracks in Fig. 2B show BrdU green channel.Distribution of BrdU tract length in WT, cdc7-1, and psf1-1 cells after a 15-minBrdU pulse at 37 °C. Boxes represent 25th to 75th percentiles; horizontal linesin the boxes represent median lengths; whiskers span 10th to 90th percentiles.**P < 0.01; ***P < 0.001. Median BrdU tract length inWT, cdc7-1, and psf1-1 cellsafter 15 and 30 min at 37 °C (Table 1). (E and F) ChIP analyses showing that Fob1occupies the Ter sites in the cdc7-1 cells at both the permissive (24 °C) andnonpermissive (37 °C) temperatures. The DNA segment from the transcribedregion of 35S rRNA was used as an internal (loading) control. Error bars re-present standard error.

Bastia et al. PNAS | Published online June 13, 2016 | E3641

BIOCH

EMISTR

YPN

ASPL

US

PFA at the Ter site of rDNA in strains containing cdc7-1 mcm5-bob1 and cdc7Δ mcm5-bob1 and the WT control. The experi-ments were repeated twice. The images of the 2D gels showedthat PFA was restored in cdc7-1 mcm5 bob1 and cdc7Δ mcm5-bob1 cells at 37 °C (Figs. 3A and 4 A, iii) as contrasted with thecdc7-1 single mutant, which failed to arrest forks at Ter at 37 °C(Fig. 1C). The results were consistent with the conclusion that(i) DDK was dispensable for fork arrest in the presence of mcm5-bob1 and (ii) alteration in the Mcm2–7 structure caused by thesuppressors controlled PFA, probably by assembling CMG in-dependent of DDK. Additional pathways of suppression of DDKand their effect on fork arrest are described below. It should benoted that although in some of the 2D gels the sizes of the Y-arcsseem to be different, these are caused by image processing andthere are no real differences.

Deletion of the Inhibitory Domain in the N-Terminal Region of Mcm4Restored Fork Arrest in DDK ts Mutants at 37 °C. The N-terminalregion of Mcm4 contains an anchoring domain for Dbf4 re-cruitment, and residues 2–174 of Mcm4 also contain an inhibitorydomain (52). Because the main function of Mcm4 phosphorylationby DDK is believed to be neutralization of the inhibitory domainof Mcm4, a ts dbf4-1 mutation, as expected, is suppressed by theN-terminal deletion (Δ2–174) of Mcm4 (51–53). If PFA is con-trolled by DDK through phosphorylation of Mcm4, required toform the CMG complex, it should be manifested in stable forkarrest in a dbf4-1 mcm4 (Δ2–274) double mutant at both 24 °C and37 °C for 3 h. Two-dimensional gel analysis confirmed this pre-diction by showing that, whereas the single mutant dbf4-1 showedsignificantly reduced fork arrest at 37 °C (Fig. 1D), a dbf4-1 mcm4(Δ2–174) double mutant showed no defect in replication fork arrestat 37 °C (Fig. 3B). Thus, the results provided additional support forthe conclusion that although Mcm2–7 phosphorylation by DDKwas required for stable fork arrest, the N-terminal deletion ofMcm4 facilitated PFA, most likely by permitting formation of activeCMG complex in the absence of DDK activity.

Deletion of Rif1 Suppressed the cdc7-1 Mutation at 30 °C and RestoredPFA at the Semipermissive Temperature. It was recently reported thatRif1 can erase Mcm2–7 phosphorylation by recruiting the proteinphosphatase PP1 to the replisome. Therefore, rif1Δ suppressed acdc7-1 mutation at 30 °C and restored cell growth at that tem-perature (54, 57, 58). To investigate whether there was a corre-spondence between suppression of DDK mutants by rif1Δ andrestoration of PFA, we first performed growth analysis and con-firmed that rif1Δ could suppress a cdc7-1 mutation at 30 °C butnot at 37 °C (Fig. 3C). We took this result to mean that whereas amoderate defect in cdc7-1 observed at 30 °C was suppressible byrif1Δ, a more severe phenotype manifested at 37 °C was notsuppressed. We then performed 2D gel analyses of replicationintermediates of a rif1Δ cdc7-1 double mutant and the cdc7-1single mutant at all three temperatures and observed fork arrest inthe double mutant at 24 °C and 30 °C but not at 37 °C for 3 h (Fig.3 F–H). In contrast, similar analysis using a cdc7-1 single mutantrevealed, as expected, a clear reduction in fork arrest at both 30 °C(Fig. 3D and E) and 37 °C (Fig. 1C). Together, the data from threeseparate suppressors of DDK mutants support the conclusion thatphosphorylated Mcm2–7, but not its dephosphorylated form, playsa critical role in promoting PFA, and that the requirement for

Mcm2–7 phosphorylation is dispensable in mutants that can ini-tiate replication in the absence of DDK.

DDK-Catalyzed Phosphorylation of the Mcm2–7 Complex Was Necessaryfor the Retention of Tof1 at the Replisome.How does DDK-dependentCMG (Mcm2–7) phosphorylation control fork arrest at Ter? Wepreviously reported that Tof1 and Csm3, but not Mrc1, of the FPCare necessary for stable fork arrest at Ter sites and protein bar-riers elsewhere in the chromosome and that Tof1–Csm3 protectsthe arrested fork by counteracting the Rrm3 helicase, which evictsFob1 protein from the Ter sites in the absence of either Tof1 orCsm3 (23, 28). We hypothesized that a phosphorylated Mcm2–7complex was necessary for the recruitment and/or retention of Tof1at the replisome and that inactivation of DDK by a ts mutation atthe nonpermissive temperature was predicted to evict Tof1 from, orfail to recruit it to, the replisome. We expected this to cause un-hindered fork passage not only through the Fob1–Ter complexes butalso past many other DNA–protein barriers. Furthermore, weexpected a suppressor such as mcm5-bob1 to restore PFA, pre-sumably by blocking removal of Tof1 from the replisome at 37 °C.We tested these predictions as described below.First, we wished to determine whether Mcm4 phosphorylation

was reduced by inactivation of DDK in vivo. We performedWestern blots (WBs) using FLAG-tagged Mcm4 in cdc7-1 rif1Δto monitor dephosphorylation of Mcm4 by gel electrophoresis at24 °C and 37 °C. We used a rif1Δ strain to enhance the yield ofphosphorylated Mcm4 for easier detection, and the data showedthat inactivation of cdc7-1 at 37 °C caused a significant reductionin the ratio of the dephosphorylated to the phosphorylated formof Mcm4 (Fig. 4 A, i and ii).To follow the degrees of retention of Tof1 protein in the

nuclear vs. cytoplasmic fractions in the presence and absence ofactive DDK, we synchronized the cells of the different genotypesby growing them at 24 °C until reaching log phase and thenarrested the cells in G1 phase with α-factor, released the block byproteolyzing the α-factor, and allowed the cells to grow for anadditional 30 min at either 24 °C or 37 °C. Spheroplasts wereprepared and fractionated into soluble and chromatin fractions,and the presence or absence of Tof1-myc in each fraction wasmeasured by WBs using anti-myc antibodies. Measurement of

Fig. 3. PFA in various suppressors of cdc7-1 or dbf4-1 at permissive (24 °C),semipermissive (30 °C), and nonpermissive temperatures (37 °C). (A and B) Two-dimensional gel analysis of (A) mcm5-bob1 cdc7-1 cells at 24 °C and 37 °C and(B) dbf4-1 mcm4 (Δ2–174) cells at 24 °C and 37 °C. (C) Growth patterns of cdc7-1rif1Δ cells at three different temperatures. (D–H) Two-dimensional gel analysisof fork movement in cdc7-1 cells at (D) 24 °C and (E) 30 °C and fork movementand PFA of cdc7-1rif1Δ at (F) 24 °C, (G) 30 °C, and (H) 37 °C.

Table 1. Mean BrdU tract length (kilobases)

Time, min WT cdc7-1 psf1-1

0 0 0 015 28.45 40.21 53.5230 46.19 62.16 79.92

E3642 | www.pnas.org/cgi/doi/10.1073/pnas.1607552113 Bastia et al.

Orc3 from the same fractions, using anti-Orc3 antibodies, was usedas an internal control. In WT cells, Tof1 was associated with thechromatin fraction at both 24 °C and 37 °C (Fig. 4B). By contrast,in cdc7-1 cells, we consistently observed, in six replicates of theexperiment, that Tof1 remained chromatin-bound at 24 °C but wasgreatly reduced in the chromatin fractions at 37 °C (Fig. 4 C and F, i).Similar experiments were performed in a cdc7Δ mcm5-bob1

strain. As predicted, Tof1-myc was retained in the chromatin frac-tion even after incubation at 37 °C (Fig. 4 D and F, ii). Surprisingly,in the congenic cdc7-1 strain a myc-tagged Csm3 was retained in thechromatin fraction (Fig. 4 E and F, iii). We propose that trapping ofCsm3 by secondary interactions with other proteins, for examplewith the replication protein A [single-strand DNA-binding protein(59)], might have prevented its release from the chromatin fraction.In the presence of the mcm5-bob1 suppressor, these cells grewwithout DDK and also displayed replication fork arrest at Ter atboth 24 °C and 37 °C (Figs. 3A and 4 A, iii). The data suggestedthat direct interaction between DDK and Tof1 was not obligatory

for PFA when Mcm2–7 was activated in the suppressor mutant.We therefore investigated why Tof1 was not recruited or retainedin the chromatin fraction upon inactivation of DDK in the bio-chemical experiments described below.

Protein–Protein Interaction Between Tof1 and the Replisome. Wewished to test the hypothesis that, in addition to the known roleof phosphorylation of Mcm2–7 by DDK for CMG assembly,Mcm2–7 phosphorylation also contributes to the function ofCMG as a landing pad for Tof1 recruitment. We performed invitro protein–protein interaction experiments using phosphory-lated and dephosphorylated CMG and Tof1–Csm3, and alsoexamined the Mcm2–7 complex (both the phosphorylated anddephosphorylated forms) and separately Cdc45 and GINS forbinding to Tof1–Csm3. The purity of the various proteins usedwas determined by Coomassie blue-stained SDS/PAGE profiles(Fig. S3). The experiments provided two significant insights intothe regulation of PFA: (i) Tof1 bound to the phosphorylated

Fig. 4. Western blots showing Tof1 retention in the chromatin fraction in cells with the designated genotypes at 24 °C and 37 °C. (A, i) Phosphorylation ofMcm4 in cdc7-1 at 24 °C and 37 °C. (A, ii) Quantification of Mcm4 phosphorylation. (A, iii) Two-dimensional gel showing that PFA occurs in mcm5-bob1 cdc7Δcells at 37 °C. (B) WBs showing the relative proportions of Tof1-myc present in the soluble vs. chromatin fractions ofWT cells at 24 °C and 37 °C; Orc3 was usedas a loading control. (C) The same as B except that cdc7-1 cells were used. (D) The same as C except that cdc7Δ mcm5-bob1 cells were used. (E) Relativedistribution of Csm3-myc in the soluble vs. chromatin fractions of cdc7-1 cells. (F) Quantification of the WBs (replicated four to seven times) and the dataplotted with SE bars. (F, i) Tof1 in chromatin fractions of cdc7-1 at 24 °C and 37 °C. (F, ii) Same as i except that cdc7-1 mcm5-bob1 cells were used. (F, iii)Retention of Csm3-myc in the chromatin fractions at the two temperatures in cdc7-1 cells.

Bastia et al. PNAS | Published online June 13, 2016 | E3643

BIOCH

EMISTR

YPN

ASPL

US

CMG and Mcm2–7 but not to their dephosphorylated forms orto Cdc45 or GINS, and (ii) only phospho-Tof1 was competent tointeract with CMG and Mcm2–7. The evidence for these con-clusions is presented below.We purified the Tof1–13myc–Csm3 complex, taking care to

protect the proteins from possible dephosphorylation duringpurification by using multiple phosphatase inhibitors (SI Mate-rials and Methods). We tested the purified complex to determineits phosphorylated state by Zn2+-Phos-tag PAGE analysis (60),which revealed that ∼50% of the purified Tof1 was phosphorylated,as indicated by the upward mobility shift of the protein bands

and loss of these shifted bands after dephosphorylation (Fig. 5A,Left). A control SDS/PAGE of the same protein without Phos-tag isshown in Fig. 5A, Right, with actin as the loading control. The Tof1protein, before and after dephosphorylation, had almost the samemobility, commensurate with its molecular mass, by SDS/PAGE.The gels were replicated three times with nearly identical results.The data suggested that in vitro dephosphorylation did not causeincidental proteolytic degradation of Tof1. To measure the possibleeffect of dephosphorylation on binding to phospho-CMG, an aliquotof Tof1-myc was pretreated with λ-phosphatase and the enzyme wasinactivated by a mixture of phosphatase inhibitors (PhosSTOP)

Fig. 5. Interactions of phosphorylated and dephosphorylated forms of CMG and Tof1 with each other as revealed by Western blotting. (A, Left) Zn2+ Phos-tag gel of phosphorylated and dephosphorylated Tof1. (A, Right) SDS/PAGE profiles of both untreated Tof1-myc and that treated with λ-phosphatase. Actinloading controls are also shown. (B) Interaction of Tof1-myc–Csm3 complex untreated and treated with λ-phosphatase with immobilized CMG-His6. (C) WBsshowing binding of Tof1-myc untreated (-) and treated with λ-phosphatase (+) to immobilized FLAG-tagged Mcm2–7. (D) Interaction of Tof1-myc withimmobilized untreated CMG-His6 (Left) and dephosphorylated immobilized CMG-His6 (Right). (E) Interaction of Tof1-myc with untreated immobilized Mcm2–7-FLAG and with dephosphorylated and immobilized Mcm2–7-FLAG. (F) Quantification of binding of Tof1-myc untreated (-) and dephosphorylated (+) toimmobilized CMG. (G) Binding of untreated (-) and dephosphorylated Tof1-myc to immobilized Mcm2–7-FLAG (not dephosphorylated with λ-phosphatase).(H) Binding of untreated Tof1-myc to immobilized untreated CMG-His6 and to the same after dephosphorylation with λ-phosphatase. (I) Quantification ofbinding of Tof1-myc to untreated immobilized Mcm2–7-FLAG and to the same after dephosphorylation with λ-phosphatase. (J) Quantification of binding ofCdc45 in solution to untreated immobilized Tof1-myc or to the same after treatment with λ-phosphatase. (K) Quantification of the binding of immobilizedGINS complex to Tof1-myc in solution before (-) and after (+) treatment with λ-phosphatase. Error bars represent standard error.

E3644 | www.pnas.org/cgi/doi/10.1073/pnas.1607552113 Bastia et al.

before loading onto a CMG affinity column. We made sure thatpretreatment of Tof1-myc with the phosphatase followed by additionof inhibitors was effective in inactivating the enzyme by carrying outstandard phosphatase assays, with little or no residual phosphataseactivity detectable by standard colorimetric phosphatase assays (SIMaterials and Methods and Fig. S4). It should be noted that Csm3 isalso a phosphoprotein (PhosphoGRID, www.phosphogrid.org/), andcould also be contributing to Tof1–Csm3 interaction(s) with CMG.Approximately equal picomoles of phospho-Tof1 and the dephos-phorylated form were applied to the CMG affinity columns, eachcontaining the same amount of the affinity matrix. Control bindingto the affinity matrix showed only low background levels of retentionof Tof1 (Fig. 5B). The Tof1-myc that bound to the immobilizedCMG columns (as measured in seven replicates of WBs using threedifferent protein preparations) was quantified as a percentage of theinput protein bound to immobilized CMG. The data consistentlyrevealed an approximately eightfold increase in binding of phos-phorylated (untreated) Tof1 to the CMG matrix compared with thedephosphorylated Tof1 protein (Fig. 5B; see quantification in Fig.5F). Does Tof1–Csm3 in the phosphorylated form bind to the sep-arate components of the CMG complex? We separately purifiedMcm2–7, Cdc45, and GINS as described in SI Materials and Meth-ods. Mcm2–7 included a FLAG-tagged Mcm5 subunit. The complexwas immobilized on anti-FLAG affinity beads and challenged withphosphorylated (untreated with λ-phosphatase) and dephosphory-lated (treated with 1 unit λ-phosphatase at 30 °C for 1 h) Tof1protein. WBs of the bound Tof1 protein visualized with anti-Mycantibodies revealed that whereas phosphorylated Tof1–Csm3 readilybound to the immobilized phospho-Mcm2–7, the dephosphorylatedform of Tof1 showed significantly reduced binding to the affinitymatrix (Fig. 5 C and G; - and + indicate no treatment or de-phosphorylation with λ phosphatase, respectively). Similar bindingexperiments using Cdc45 and GINS complex showed 15–20%binding by either phospho-Tof1–Csm3 or its dephosphorylated form(Fig. 5 J and K). We conclude from the data that phospho-Tof1specifically binds to Mcm2–7 but not to Cdc45 and GINS.Do CMG and Mcm2–7 require phosphorylation to bind to

phospho-Tof1? To investigate this question, purified CMG wasextensively treated with λ-phosphatase followed by removal of thephosphatase by a gel-filtration step. Aliquots of the dephosphory-lated CMG and separately the same amounts of the phosphory-lated form were immobilized on Ni-NTA columns. Equal amountsof Tof1-myc–Csm3 complex were applied to each affinity column,and the proteins were eluted, resolved by SDS/PAGE, and visu-alized by WBs using anti-Myc Ab. The results showed that CMG,after dephosphorylation, showed a significant reduction in bindingto Tof1-GST (Fig. 5 D and H). Therefore, a continued phos-phorylated state of CMG, even after complex formation, is re-quired for optimal binding to and retention of Tof1–Csm3 at thereplisome. Similarly, we determined that immobilized phospho-Mcm2–7 but not its dephosphorylated form was able to bind toTof1-myc–Csm3 complex (Fig. 5 E and I).Independent support for the aforementioned conclusion is

provided by the expression and purification of Mcm2–7 subunits inE. coli reconstitution into an Mcm2–7 complex [that had limitedhelicase activity (61)]. The rationale was that Mcm2–7 expressed inE. coli is not expected to be phosphorylated because of a lack of aDDK or DDK-like kinase. Therefore, Mcm2–7E. coli shouldnot bind to Tof1. We confirmed this expectation by attemptingto bind phosphorylated Tof1 to immobilized Mcm2–7E. coli butobserved only low background levels of binding with bothphosphorylated and dephosphorylated Tof1 (Fig. S5).The proposed mechanism of PFA, supported by these data, is

schematically shown in Fig. 6. The scheme takes into consider-ation our previous discovery that the Tof1–Csm3 complexcounteracts the Rrm3 “sweepase” from displacing Fob1 fromTer, thereby maintaining stable fork arrest (Fig. 6A) (23, 28). Insummary, the model posits that DDK phosphorylates Mcm2–7,

which, along with Cdc45 and GINS, forms the CMG helicase.The CMG complex then recruits the phosphorylated but not thedephosphorylated form of Tof1–Csm3 to the replisome. Al-though Tof1 also requires phosphorylation for its recruitment byCMG, the kinase responsible for Tof1 phosphorylation remainsunclear at this time. We propose that inactivation of DDKprevents the CMG complex from recruiting or retaining Tof1,thereby allowing Rrm3 to displace Fob1, resulting in unhinderedfork progression past Ter (Fig. 6B). Why does PFA still work incdc7-1 mcm5-bob1 cells at the nonpermissive temperature? Wespeculate that in the absence of DDK, CMG assembles in analternative conformation that, as has been suggested, may bedependent on Cdc28, Clb2, and Clb5 (62).

DiscussionThe biochemical mechanism of control of PFA in eukaryotes andits mode of regulation are poorly understood (4). This mechanismis of considerable general interest because of the many importantphysiological transactions mediated by the process. The cell-cyclekinases CDK and DDK are required for replication initiation, andCDK is involved in limiting replication to only once per cell cycleand in S-phase checkpoint control (37, 39, 40, 49, 63). The workreported here demonstrates that DDK is also required for PFA ata nonhistone protein barrier. This mechanism has been illuminatedby our observation that inactivation of cdc7-1 not only causedabolition of PFA at Ter sites of rDNA but that the phosphorylated

Fig. 6. Schematic representations of a model illustrating the mechanism ofprogrammed fork arrest. Cdc7, GINS, and Cdc45 are not shown, to simplifythe figure. (A) DDK phosphorylates Mcm2–7 and, along with Cdc45 andGINS, assembles CMG, which recruits phosphorylated Tof1–Csm3. Tof1–Csm3antagonizes Rrm3 to prevent displacement of Fob1 from Ter. (B) In theabsence of DDK, Tof1 is no longer maintained at Ter, causing unhinderedfork passage. See the text for further details.

Bastia et al. PNAS | Published online June 13, 2016 | E3645

BIOCH

EMISTR

YPN

ASPL

US

form of Mcm2–7, besides its well-known function in the assemblyof the CMG complex, was also necessary for binding to Tof1 invitro and its recruitment in vivo. Furthermore, phospho-Tof1 but notits dephosphorylated form bound to phospho-CMG and phospho-Mcm2–7 but not to the dephosphorylated protein complexes.Consistent with the in vitro binding data, Tof1 in vivo was retainedin the chromatin fraction only in the presence of active DDK in thecell milieu.Which kinase phosphorylates Tof1–Csm3? The data in Fig. 4

show that Tof1 is still recruited to chromatin in a cdc7-1 mutant inthe presence of the mcm5-bob1 suppressor at the nonpermissivetemperature, which suggests that DDK-catalyzed Tof1 phosphor-ylation is not necessary in this genetic background and probablyanother kinase phosphorylates Tof1. It is also possible that a phos-phorylated Tof1 is not needed in this genetic background. Never-theless, several lines of evidence support the possibility that DDK isimportant for Tof1 function. First, the Schizosaccharomyces pombehomolog of Tof1, called Swi1, interacts with the homologous DDKcomplex (Hsk1–Dfp1) in the two-hybrid system and the proteinsare also shown to form a complex by coimmunoprecipitation (64).Second, Tof1–Csm3 and DDK from S. cerevisiae also form a com-plex, as shown by coimmunoprecipitation, and Tof1–Csm3 is believedto recruit DDK to the replication fork during meiotic DNA repli-cation (65). The requirement of Tof1–Csm3 for subsequent meioticdouble-strand-break formation could be bypassed by physically con-necting DDK to Cdc45/CMG within the replication fork, indicatingthat Tof1–Csm3 serves as a platform for recruitment of DDK.Whether this interaction also leads to DDK phosphorylation of Tof1remains to be determined. It is possible that in WT yeast, DDK isrecruited by Mcm2–7 and not only phosphorylates Mcm2–7 but alsoTof1, but that in the absence of DDK the mcm5-bob1 mutant formof Mcm2–7 is able to recruit another kinase, perhaps Cdc28-Clb5(62), that helps carry out PFA and replication. Further work isneeded to resolve this issue.Tof1 along with Csm3 plays a critical role in stabilizing stalled

replication forks at protein barriers even in the absence of S-phasecheckpoint activation, and this is central to its function in PFA,although DDK is not necessary for S-phase checkpoint activa-tions (41, 42). However, unlike most instances, in which repli-cation forks stall transiently while awaiting resolution of animpediment to fork progression, fork stalling in PFA is physio-logically programmed and longer-lasting. This raises the questionof how PFA differs mechanistically from transient fork stalling.We showed previously that Tof1 plays an additional role in PFA bycounteracting the effects of the Rrm3 helicase. Rrm3 is known toact as a sweepase for its function in removing nonhistone DNA-binding proteins, thereby allowing unrestrained replication forkmovement through protein-bound DNA. In the absence of Rrm3in yeast, replication fork stalling is observed at over 1,000 sites thatare not observed in wild-type cells (66). Many of these sites areknown to be densely populated with DNA-binding proteins, in-cluding centromeres and tRNA-coding genes, among many others.However, in contrast with PFA, the data from the rrm3Δ strainsuggest that these DNA-bound proteins are readily removed by theaction of the Rrm3 sweepase, so how does Tof1 antagonize theaction of Rrm3 in the rDNA array? In this context, it should bekept in mind that a deletion of rrm3Δ only partially rescues PFA incells harboring either tof1Δ or csm3Δ (23). This suggests that be-sides Rrm3, other factors are also involved in the abolition of PFAin tof1Δ and csm3Δ cells.As with most regulatory functions in biology, the answer appears

to lie in a balancing act between positive and negative regulatorsand also in the evolutionary tradeoff between genome stability andgenome adaptability. The DNA at stalled replication forks is sub-ject to breakage followed by recombination that can lead to ge-nome rearrangements, particularly in the case of repeated DNAsequences. At the same time, collisions between DNA replicationand DNA transcription can also lead to replication fork collapse.

The Fob1-mediated DNA replication barrier in the rDNA ofS. cerevisiae strikes a balance between these imperatives by pre-venting collisions between RNA and DNA synthesis and by usingthe components of the replication fork protection complex to pre-vent collapse of the replication fork stalled at the barrier. Althoughother organisms, as well as human cells, appear to use variousmechanisms to prevent collisions between replication and tran-scription in the highly transcribed ribosomal RNA genes, they alluse Tof1–Csm3 or the homologous human Tim/Tipin componentsof the fork protection complex to prevent collapse of the pausedreplication forks in the rDNA locus (reviewed in ref. 67).Finally, although it is well-established that DDK-catalyzed

phosphorylation of Mcm2–7 is needed for CMG assembly in WTcells, it is not known whether continued phosphorylation of theCMG complex is necessary for fork progression. Our data showingthat even after a CMG complex has been formed its dephos-phorylation causes failure to recruit the fork protection complexin vitro indicate that continued maintenance of the phosphory-lated state of CMG seems to be necessary for error-free forkprogression. It is known that under genotoxic stress, in the ab-sence of Tof1 or Mrc1, the helicase and polymerase of yeastdissociate from each other in the replisome, causing fork col-lapse (68). Therefore, one would infer that continued mainte-nance of the phosphorylated state of CMG in the postinitiationstep would be critical for genome stability.

Materials and MethodsStrain Construction. Yeast strains used in this study are listed in Table S1. Allstrains are from either the A364a or W303 background unless otherwisenoted. Strains were constructed by standard genetic and molecular ge-netic techniques. Epitope tagging was performed using standard PCR-based insertion procedures (69). Incorporation of BrdU cassettes intostrains was done as described elsewhere using the plasmids p404–BrdU–Incand p405–BrdU–Inc (70). To disrupt the Bar1 locus by Ura3 incorporation,the pbar1::Ura3 plasmid was digested with BamHI and BglII and thentransformed into the appropriate strain and selected on Ura dropout plates.

Oligonucleotides. Oligonucleotides are listed in Table S2.

Analysis of Mcm4 Phosphorylation. Chromatin fractions were used to analyzein vivo Mcm4 phosphorylation (54, 58). Phenylmethylsulfonyl fluoride (PMSF),sodium fluoride (NaF), and sodium orthovanadate (Na3VO4) were addeddirectly to cultures of MCM4-6His-3FLAG cells to a final concentration of 1, 1,and 0.5 mM, respectively. We collected cells by a brief centrifugation (Sorvall5C Plus; rotor# HS-4) at 5,000 rpm for 5 min and washed them with an ice-coldsolution of 1 mM PMSF, 2 mM NaF, and 1 mM Na3VO4. Cells were collected bycentrifugation and protein extracts were prepared as described above for thechromatin fractionation assay [51]. Proteins were analyzed by electrophoresison 6% SDS/PAGE, followed by Western blotting using monoclonal anti-FLAGantibody (M2; Sigma-Aldrich) as the primary antibody.

ChIP Assays. ChIP assays were performed in the cdc7-1 strain (YB556) in whichFob1 was tagged with TAP at the C terminus (primers 2F, 2R, 3F and 3R), usingcells grown at 24 °C and 37 °C (23, 71). Each step has been described before(23, 71), except that cells grown at 37 °C were quickly chilled for 5 min on icebefore adding 1% formaldehyde. PCR was performed using the primers 4F, 4R,5F, and 5R (Table S2).

Two-Dimensional Agarose Gel Electrophoresis. Brewer–Fangman 2D agarosegel electrophoresis was performed to visualize replication intermediateswithin cells obtained from cultures grown at 24 °C and 37 °C as previouslydescribed (55). Primers 1F and 1R were used to amplify the 1.4-kb regionencompassing the rDNA NTS region, which was used as a probe to visualizereplication intermediates.

Overnight starter cultures were inoculated into yeast, peptone, dextrose(YPD) medium at a starting OD600 of 0.2 and then grown for 2 h at 24 °C. Half ofthis culture was poured into another flask and swirled for 5 min in a water bathat 37 °C and then grown a further 2–3 h at 37 °C before being harvested. Thecontrol culture was grown 2 h more at 24 °C after reaching an OD600 of 0.2 andthen harvested. Genomic DNA was prepared and digested with EcoRV and BglIIrestriction enzymes to yield a 4.5-kb region of rDNA that included thereplication origin.

E3646 | www.pnas.org/cgi/doi/10.1073/pnas.1607552113 Bastia et al.

DNA Content Analysis by FACS. FACS analysis was performed on either a syn-chronous or asynchronous culture grown at 24 °C or 37 °C as described (56).

Chromatin Fractionation. Chromatin fractionation was performed to de-termine the level of chromatin-bound Tof1 or Csm3 in cells at 24 °C or 37 °C.The detailed method is described in SI Materials and Methods. Briefly, cellswere harvested after Pronase treatment. Proteins from the chromatin andsoluble fractions were prepared and resolved by SDS/PAGE and blotted toNytran membranes, and the membranes were developed with anti-myc andanti-ORC3 antibodies.

DNA Combing. DNA combing was performed as described (72). BrdU was de-tected with a rat monoclonal antibody (AbCys; clone BU1/75) and goat anti-ratcoupled to Alexa 488 (Invitrogen). DNA fibers were counterstained using anti-mouse MAB 3034 (Millipore) and goat anti-mouse coupled to Alexa 546. Imageswere recorded on a Leica DM6000 microscope equipped with a 40× objectiveand a CoolSNAP HQ CCD camera (Roper Scientific). Images were processed asdescribed (73). BrdU tracts were measured with MetaMorph (Molecular De-vices). Statistical analyses of differences between samples were performed withGraphPad Prism 6.0 using the Mann–Whitney rank-sum test. More details ofdata collection and statistical analysis are provided in the legend to Fig. 2.

Protein Purification. CMG, Mcm2–7, and Cdc45 were purified by expressionin yeast strains containing integrated genes as described in greater detailin SI Materials and Methods. Tof1–Csm3 was purified as follows. Tof1 ORFwith 13c-Myc was cloned into the vector pBJ842 (74). Tof1 was expressed in

S. cerevisiae as a GST fusion protein in this vector and purified on a glutathione-agarose column as described in detail in SI Materials and Methods.

Protein–Protein Interaction in Vitro. Details are described in SI Materialsand Methods.

Phos-Tag Gels. Phosphorylated Tof1 protein was detected by Zn2+ Phos-taggels as described in detail in SI Materials and Methods following a publishedprocedure (60, 75).

Phosphatase Assay. The phosphatase assay was performed as described to notonly detect phosphatase activity of λ-phosphatase at different dilutions butto check the effectiveness of the phosphatase inhibitor (PhosSTOP; Roche). Astock solution (1 tablet per mL) was prepared and 1 μL/50 μL reaction mix-ture was added to stop dephosphorylation. For the colorimetric assay,4-nitrophenyl phosphate was used as substrate and color was detected withan ELISA plate reader (76).

ACKNOWLEDGMENTS. We thank Robert Sclafani, Bruce Stillman, BonitaBrewer, M. K. Raghuraman, and Anne Donaldson for plasmids and strainsand Bruce Stillman for Orc3 antibodies. We thank Ryan Mayle for λ-phos-phatase–treated CMG. We also thank Oscar Aparicio for FACS analyses. Thiswork was supported by National Institute of General Medical Sciences Grant5 R01-GM-098013 (to D.B.), grants from Agence Nationale pour la Rechercheand Ligue contre le Cancer (Équipe Labellisée) (to P.P.), NIH Grant GM-115809 (to M.E.O.), and the Howard Hughes Medical Institute (M.E.O.).

1. Hawkins M, et al. (2013) High-resolution replication profiles define the stochastic

nature of genome replication initiation and termination. Cell Reports 5(4):1132–1141.2. Khatri GS, MacAllister T, Sista PR, Bastia D (1989) The replication terminator protein of

E. coli is a DNA sequence-specific contra-helicase. Cell 59(4):667–674.3. Lee EH, Kornberg A, Hidaka M, Kobayashi T, Horiuchi T (1989) Escherichia coli repli-

cation termination protein impedes the action of helicases. Proc Natl Acad Sci USA

86(23):9104–9108.4. Bastia D, Zaman S (2014) Mechanism and physiological significance of programmed

replication termination. Semin Cell Dev Biol 30:165–173.5. Zechiedrich EL, Cozzarelli NR (1995) Roles of topoisomerase IV and DNA gyrase in

DNA unlinking during replication in Escherichia coli. Genes Dev 9(22):2859–2869.6. Madabhushi R, Marians KJ (2009) Actin homolog MreB affects chromosome segre-

gation by regulating topoisomerase IV in Escherichia coli. Mol Cell 33(2):171–180.7. Baxter J, Diffley JF (2008) Topoisomerase II inactivation prevents the completion of

DNA replication in budding yeast. Mol Cell 30(6):790–802.8. Fachinetti D, et al. (2010) Replication termination at eukaryotic chromosomes is

mediated by Top2 and occurs at genomic loci containing pausing elements. Mol Cell

39(4):595–605.9. Maric M, Maculins T, De Piccoli G, Labib K (2014) Cdc48 and a ubiquitin ligase drive

disassembly of the CMG helicase at the end of DNA replication. Science 346(6208):

1253596.10. Mohanty BK, Bastia D (2004) Binding of the replication terminator protein Fob1p to

the Ter sites of yeast causes polar fork arrest. J Biol Chem 279(3):1932–1941.11. Kobayashi T (2003) The replication fork barrier site forms a unique structure with

Fob1p and inhibits the replication fork. Mol Cell Biol 23(24):9178–9188.12. Dalgaard JZ, Klar AJ (2001) A DNA replication-arrest site RTS1 regulates imprinting by

determining the direction of replication at mat1 in S. pombe. Genes Dev 15(16):

2060–2068.13. Eydmann T, et al. (2008) Rtf1-mediated eukaryotic site-specific replication termina-

tion. Genetics 180(1):27–39.14. Krings G, Bastia D (2005) Sap1p binds to Ter1 at the ribosomal DNA of Schizo-

saccharomyces pombe and causes polar replication fork arrest. J Biol Chem 280(47):

39135–39142.15. Sánchez-Gorostiaga A, López-Estraño C, Krimer DB, Schvartzman JB, Hernández P

(2004) Transcription termination factor reb1p causes two replication fork barriers at

its cognate sites in fission yeast ribosomal DNA in vivo. Mol Cell Biol 24(1):398–406.16. Mejía-Ramírez E, Sánchez-Gorostiaga A, Krimer DB, Schvartzman JB, Hernández P

(2005) The mating type switch-activating protein Sap1 is required for replication fork

arrest at the rRNA genes of fission yeast. Mol Cell Biol 25(19):8755–8761.17. Krings G, Bastia D (2004) swi1- and swi3-dependent and independent replication fork

arrest at the ribosomal DNA of Schizosaccharomyces pombe. Proc Natl Acad Sci USA

101(39):14085–14090.18. Krings G, Bastia D (2006) Molecular architecture of a eukaryotic DNA replication

terminus-terminator protein complex. Mol Cell Biol 26(21):8061–8074.19. Singh SK, Sabatinos S, Forsburg S, Bastia D (2010) Regulation of replication termi-

nation by Reb1 protein-mediated action at a distance. Cell 142(6):868–878.20. Bando M, et al. (2009) Csm3, Tof1, and Mrc1 form a heterotrimeric mediator complex

that associates with DNA replication forks. J Biol Chem 284(49):34355–34365.21. Calzada A, Hodgson B, Kanemaki M, Bueno A, Labib K (2005) Molecular anatomy and

regulation of a stable replisome at a paused eukaryotic DNA replication fork. Genes

Dev 19(16):1905–1919.

22. Tourrière H, Versini G, Cordón-Preciado V, Alabert C, Pasero P (2005) Mrc1 and Tof1promote replication fork progression and recovery independently of Rad53. Mol Cell19(5):699–706.

23. Mohanty BK, Bairwa NK, Bastia D (2006) The Tof1p-Csm3p protein complex coun-teracts the Rrm3p helicase to control replication termination of Saccharomyces cer-evisiae. Proc Natl Acad Sci USA 103(4):897–902.

24. Dalgaard JZ, Klar AJ (2000) swi1 and swi3 perform imprinting, pausing, and termi-nation of DNA replication in S. pombe. Cell 102(6):745–751.

25. Gotter AL, et al. (2000) A time-less function for mouse Timeless. Nat Neurosci 3(8):755–756.

26. Gotter AL (2003) Tipin, a novel Timeless-interacting protein, is developmentally co-expressed with Timeless and disrupts its self-association. J Mol Biol 331(1):167–176.

27. Cho WH, et al. (2013) Human Tim-Tipin complex affects the biochemical properties ofthe replicative DNA helicase and DNA polymerases. Proc Natl Acad Sci USA 110(7):2523–2527.

28. Mohanty BK, Bairwa NK, Bastia D (2009) Contrasting roles of checkpoint proteins asrecombination modulators at Fob1-Ter complexes with or without fork arrest.Eukaryot Cell 8(4):487–495.

29. Ivessa AS, et al. (2003) The Saccharomyces cerevisiae helicase Rrm3p facilitates repli-cation past nonhistone protein-DNA complexes. Mol Cell 12(6):1525–1536.

30. Bermejo R, Lai MS, Foiani M (2012) Preventing replication stress to maintain genomestability: Resolving conflicts between replication and transcription. Mol Cell 45(6):710–718.

31. Brewer BJ (1988) When polymerases collide: Replication and the transcriptional or-ganization of the E. coli chromosome. Cell 53(5):679–686.

32. Deshpande AM, Newlon CS (1996) DNA replication fork pause sites dependent ontranscription. Science 272(5264):1030–1033.

33. Lin YL, Pasero P (2012) Interference between DNA replication and transcription as acause of genomic instability. Curr Genomics 13(1):65–73.

34. Defossez PA, et al. (1999) Elimination of replication block protein Fob1 extends thelife span of yeast mother cells. Mol Cell 3(4):447–455.

35. Choudhury M, Zaman S, Jiang JC, Jazwinski SM, Bastia D (2015) Mechanism of reg-ulation of ‘chromosome kissing’ induced by Fob1 and its physiological significance.Genes Dev 29(11):1188–1201.

36. Sclafani RA, Holzen TM (2007) Cell cycle regulation of DNA replication. Annu RevGenet 41:237–280.

37. Siddiqui K, On KF, Diffley JF (2013) Regulating DNA replication in Eukarya. ColdSpring Harb Perspect Biol 5(9):a012930.

38. Tanaka S, Araki H (2013) Helicase activation and establishment of replication forks atchromosomal origins of replication. Cold Spring Harb Perspect Biol 5(12):a010371.

39. Bell SP, Dutta A (2002) DNA replication in eukaryotic cells. Annu Rev Biochem 71:333–374.

40. Tsuji T, Lau E, Chiang GG, Jiang W (2008) The role of Dbf4/Drf1-dependent kinaseCdc7 in DNA-damage checkpoint control. Mol Cell 32(6):862–869.

41. Pessoa-Brandão L, Sclafani RA (2004) CDC7/DBF4 functions in the translesion synthesisbranch of the RAD6 epistasis group in Saccharomyces cerevisiae. Genetics 167(4):1597–1610.

42. Brandão LN, Ferguson R, Santoro I, Jinks-Robertson S, Sclafani RA (2014) The role ofDbf4-dependent protein kinase in DNA polymerase ζ-dependent mutagenesis inSaccharomyces cerevisiae. Genetics 197(4):1111–1122.

43. Takayama Y, et al. (2003) GINS, a novel multiprotein complex required for chromo-somal DNA replication in budding yeast. Genes Dev 17(9):1153–1165.

Bastia et al. PNAS | Published online June 13, 2016 | E3647

BIOCH

EMISTR

YPN

ASPL

US

44. Zou L, Mitchell J, Stillman B (1997) CDC45, a novel yeast gene that functions with theorigin recognition complex and Mcm proteins in initiation of DNA replication. MolCell Biol 17(2):553–563.

45. Zou L, Stillman B (1998) Formation of a preinitiation complex by S-phase cyclin CDK-dependent loading of Cdc45p onto chromatin. Science 280(5363):593–596.

46. Moyer SE, Lewis PW, Botchan MR (2006) Isolation of the Cdc45/Mcm2-7/GINS (CMG)complex, a candidate for the eukaryotic DNA replication fork helicase. Proc Natl AcadSci USA 103(27):10236–10241.

47. Georgescu RE, et al. (2014) Mechanism of asymmetric polymerase assembly at theeukaryotic replication fork. Nat Struct Mol Biol 21(8):664–670.

48. Langston LD, et al. (2014) CMG helicase and DNA polymerase e form a functional 15-subunit holoenzyme for eukaryotic leading-strand DNA replication. Proc Natl Acad SciUSA 111(43):15390–15395.

49. Nguyen VQ, Co C, Li JJ (2001) Cyclin-dependent kinases prevent DNA re-replicationthrough multiple mechanisms. Nature 411(6841):1068–1073.

50. Hardy CF, Dryga O, Seematter S, Pahl PM, Sclafani RA (1997) mcm5/cdc46-bob1 by-passes the requirement for the S phase activator Cdc7p. Proc Natl Acad Sci USA 94(7):3151–3155.

51. Sheu YJ, Stillman B (2006) Cdc7-Dbf4 phosphorylates MCM proteins via a docking site-mediated mechanism to promote S phase progression. Mol Cell 24(1):101–113.

52. Sheu YJ, Stillman B (2010) The Dbf4-Cdc7 kinase promotes S phase by alleviating aninhibitory activity in Mcm4. Nature 463(7277):113–117.

53. Sheu YJ, Kinney JB, Lengronne A, Pasero P, Stillman B (2014) Domain within thehelicase subunit Mcm4 integrates multiple kinase signals to control DNA replicationinitiation and fork progression. Proc Natl Acad Sci USA 111(18):E1899–E1908.

54. Hiraga S, et al. (2014) Rif1 controls DNA replication by directing protein phosphatase1 to reverse Cdc7-mediated phosphorylation of the MCM complex. Genes Dev 28(4):372–383.

55. Brewer BJ, FangmanWL (1987) The localization of replication origins on ARS plasmidsin S. cerevisiae. Cell 51(3):463–471.

56. Zhong Y, et al. (2013) The level of origin firing inversely affects the rate of replicationfork progression. J Cell Biol 201(3):373–383.

57. Mattarocci S, et al. (2014) Rif1 controls DNA replication timing in yeast through thePP1 phosphatase Glc7. Cell Reports 7(1):62–69.

58. Davé A, Cooley C, Garg M, Bianchi A (2014) Protein phosphatase 1 recruitment by Rif1regulates DNA replication origin firing by counteracting DDK activity. Cell Reports7(1):53–61.

59. Kemp MG, et al. (2010) Tipin-replication protein A interaction mediates Chk1 phos-phorylation by ATR in response to genotoxic stress. J Biol Chem 285(22):16562–16571.

60. Kinoshita E, Kinoshita-Kikuta E (2011) Improved Phos-tag SDS-PAGE under neutral pHconditions for advanced protein phosphorylation profiling. Proteomics 11(2):319–323.

61. Davey MJ, Indiani C, O’Donnell M (2003) Reconstitution of the Mcm2-7p hetero-hexamer, subunit arrangement, and ATP site architecture. J Biol Chem 278(7):4491–4499.

62. Sclafani RA, TecklenburgM, Pierce A (2002) The mcm5-bob1 bypass of Cdc7p/Dbf4p inDNA replication depends on both Cdk1-independent and Cdk1-dependent steps inSaccharomyces cerevisiae. Genetics 161(1):47–57.

63. Bousset K, Diffley JF (1998) The Cdc7 protein kinase is required for origin firing duringS phase. Genes Dev 12(4):480–490.

64. Matsumoto S, Ogino K, Noguchi E, Russell P, Masai H (2005) Hsk1-Dfp1/Him1, theCdc7-Dbf4 kinase in Schizosaccharomyces pombe, associates with Swi1, a componentof the replication fork protection complex. J Biol Chem 280(52):42536–42542.

65. Murakami H, Keeney S (2014) Temporospatial coordination of meiotic DNA replica-tion and recombination via DDK recruitment to replisomes. Cell 158(4):861–873.

66. Ivessa AS, Zhou JQ, Schulz VP, Monson EK, Zakian VA (2002) Saccharomyces Rrm3p, a5′ to 3′ DNA helicase that promotes replication fork progression through telomericand subtelomeric DNA. Genes Dev 16(11):1383–1396.

67. Dalgaard JZ, Godfrey EL, MacFarlane RJ (2011) Eukaryotic replication barriers: How,why and where forks stall. DNA Replication—Current Advances, ed Seligmann H(InTech, Rijeka, Crotia), pp 14–24.

68. Katou Y, et al. (2003) S-phase checkpoint proteins Tof1 and Mrc1 form a stablereplication-pausing complex. Nature 424(6952):1078–1083.

69. Wach A (1996) PCR-synthesis of marker cassettes with long flanking homology re-gions for gene disruptions in S. cerevisiae. Yeast 12(3):259–265.

70. Viggiani CJ, Aparicio OM (2006) New vectors for simplified construction of BrdU-incorporating strains of Saccharomyces cerevisiae. Yeast 23(14–15):1045–1051.

71. Huang J, Moazed D (2003) Association of the RENT complex with nontranscribed andcoding regions of rDNA and a regional requirement for the replication fork blockprotein Fob1 in rDNA silencing. Genes Dev 17(17):2162–2176.

72. Bianco JN, et al. (2012) Analysis of DNA replication profiles in budding yeast andmammalian cells using DNA combing. Methods 57(2):149–157.

73. Pasero P, Bensimon A, Schwob E (2002) Single-molecule analysis reveals clustering andepigenetic regulation of replication origins at the yeast rDNA locus. Genes Dev 16(19):2479–2484.

74. Johnson RE, Prakash L, Prakash S (2006) Yeast and human translesion DNA synthesispolymerases: Expression, purification, and biochemical characterization. MethodsEnzymol 408:390–407.

75. Kinoshita E, Kinoshita-Kikuta E, Koike T (2009) Separation and detection of largephosphoproteins using Phos-tag SDS-PAGE. Nat Protoc 4(10):1513–1521.

76. Zhang ZY, et al. (1992) Expression, purification, and physicochemical characterizationof a recombinant Yersinia protein tyrosine phosphatase. J Biol Chem 267(33):23759–23766.

E3648 | www.pnas.org/cgi/doi/10.1073/pnas.1607552113 Bastia et al.