Protein–DNA complexes are the primary sourcesof replication fork pausing in Escherichia coliMilind K. Guptaa, Colin P. Guya, Joseph T. P. Yeelesb, John Atkinsona, Hazel Bella, Robert G. Lloydc,Kenneth J. Mariansb, and Peter McGlynnd,1

aSchool of Medical Sciences, Institute of Medical Sciences, University of Aberdeen, Aberdeen AB25 2ZD, United Kingdom; bMolecular Biology Program,Memorial Sloan–Kettering Cancer Center, New York, NY 10065; cCentre for Genetics and Genomics, Queen’s Medical Centre, University of Nottingham,Nottingham NG7 2UH, United Kingdom; and dDepartment of Biology, University of York, York YO10 5DD, United Kingdom

Edited by Mike E. O’Donnell, The Rockefeller University, Howard Hughes Medical Institute, New York, NY, and approved March 25, 2013 (received for reviewFebruary 28, 2013)

Replication fork pausing drives genome instability, because anyloss of paused replisome activity creates a requirement forreloading of the replication machinery, a potentially mutagenicprocess. Despite this importance, the relative contributions tofork pausing of different replicative barriers remain unknown.We show here that Deinococcus radiodurans RecD2 helicase inac-tivates Escherichia coli replisomes that are paused but still func-tional in vitro, preventing continued fork movement upon barrierremoval or bypass, but does not inactivate elongating forks. Us-ing RecD2 to probe replisome pausing in vivo, we demonstratethat most pausing events do not lead to replisome inactivation,that transcription complexes are the primary sources of thispausing, and that an accessory replicative helicase is critical forminimizing the frequency and/or duration of replisome pauses.These findings reveal the hidden potential for replisome inacti-vation, and hence genome instability, inside cells. They alsodemonstrate that efficient chromosome duplication requiresmechanisms that aid resumption of replication by paused repli-somes, especially those halted by protein–DNA barriers such astranscription complexes.

DNA repair | genome stability | Rep | RNA polymerase | recombination

Faithful duplication of the genome is a key challenge to allorganisms, and overcoming barriers to the timely progression

of replication forks is a major part of this challenge. Templatedamage, proteins bound to the DNA, and non–B-form DNAstructures all have the potential to halt movement of the repli-cation machinery (1–3). If forks pause at such barriers and losefunction, then replisome reloading, often via blocked fork pro-cessing by recombination enzymes, is required to resume genomeduplication (4), an error-prone process associated with grosschromosomal rearrangements (5–7). However, pausing of repli-somes does not necessarily lead to fork breakdown, becausepaused replisomes can continue duplication upon removal orbypass of the block (8–11). The balance between resumption ofreplication versus breakdown of paused forks is therefore a crit-ical factor in the maintenance of genome stability.Forks halted by nucleoprotein complexes can resume repli-

cation upon spontaneous dissociation of the nucleoproteincomplex or active removal of the block by the replisome itself oraccessory motors acting at the fork (9, 10, 12–14). Similarly,a replisome encountering a lesion that inhibits synthesis by oneof the polymerases may continue replication downstream of thelesion by repriming DNA synthesis. Such repriming occurs re-gardless of whether the damage is on the leading or the laggingstrand template, although recombination may be triggered at thegap left in the nascent strand (11, 15, 16).Whether a replisome pauses at a barrier before resuming

replication or loses activity is determined by the rates of barrierclearance/bypass and blocked replisome stability (17, 18). How-ever, despite the critical importance of fork pausing for genomestability, the contributions of different types of barrier to replisome

pausing remain unknown. In particular, although both transcrip-tion complexes and DNA damage present known challenges tofork movement in bacteria and eukaryotes (15, 19–21), the rel-ative frequencies with which they affect fork progression in vivois not known. This is in part because physical detection of forkpausing is only possible when such pausing occurs at specificlocations in the genome and is sufficiently frequent and/or pro-longed to facilitate detection. However, many pausing eventsmight not meet these criteria. Furthermore, if a paused forksubsequently resumes translocation, either upon removal or by-pass of the barrier, there would be no easily detectable phenotypicconsequences.Here we demonstrate that Deinococcus radiodurans RecD2

helicase inactivates paused but not elongating Escherichia colireplisomes in vitro. The basis of this inactivation is unknown, butthis specificity provides a tool to probe the relative frequencies ofreplisome pausing in vivo. Wild-type Escherichia coli can surviveexpression of RecD2 but chromosomal DNA content is per-turbed significantly, indicating that replisomes do pause fre-quently in vivo. Cells lacking a helicase, Rep, that clears protein–DNA barriers ahead of forks (10, 22) are hypersensitive toRecD2 expression. In contrast, defects in base or nucleotideexcision repair do not render RecD2 toxic. These data indicatethat protein–DNA complexes, not template damage, are theprimary sources of replisome pausing in nonstressed cells andthat most replicative barriers result in fork pausing but not in-activation. Thus, although there is a very considerable potentialfor triggering genomic instability during every cell cycle, thispotential is only rarely realized because of the intrinsic stabilityof the replisome and the ability of a secondary motor to helpdrive forks along protein-bound DNA. Together, these twofactors minimize the likelihood of fork pausing leading toreplisome inactivation and the need to restart replication.

ResultsRecD2 Inhibits Resumption of Replication by Paused Replisomes.Superfamily 1 helicases that translocate 3′-5′ along ssDNA(E. coli Rep and UvrD and Bacillus stearothermophilus PcrA)promote movement of E. coli replisomes through nucleoproteincomplexes, whereas D. radiodurans RecD2 and bacteriophage T4Dda, 5′-3′ superfamily 1 helicases do not (10). Indeed, additionof RecD2 results in an apparent increase in the degree of rep-lication blockage at protein–DNA complexes in vitro (10). We

Author contributions: C.P.G., J.T.P.Y., K.J.M., and P.M. designed research; M.K.G., C.P.G.,J.T.P.Y., J.A., H.B., and P.M. performed research; R.G.L. and K.J.M. contributed new re-agents/analytic tools; M.K.G., C.P.G., J.T.P.Y., J.A., K.J.M., and P.M. analyzed data; andP.M. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.1To whom correspondence should be addressed. E-mail: peter.mcglynn@york.ac.uk.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1303890110/-/DCSupplemental.

7252–7257 | PNAS | April 30, 2013 | vol. 110 | no. 18 www.pnas.org/cgi/doi/10.1073/pnas.1303890110

Dow

nloa

ded

by g

uest

on

Aug

ust 9

, 202

0

investigated the basis for this apparent increase in fork blockageby RecD2. We used a system in which replisomes could beblocked completely by a large array of lac repressor–operatorcomplexes and then the block relieved by subsequent addition ofisopropyl β-D-1-thiogalactopyranoside (IPTG), allowing moni-toring of the ability of blocked replisomes to resume replication(18)(Fig. 1A and Fig. S1). Replisomes were reconstituted onplasmid templates bearing oriC and 22 tandem lac operators. Inthe absence of a topoisomerase, replisomes could proceed onlya limited distance along the template owing to accumulation ofpositive supercoiling. Continued fork movement depended oncleavage of the template by a restriction enzyme to relieve thetopological strain (17) (Fig. 1A, ii and iii and Fig. S1) with thesite of cleavage allowing only one of the two forks to progresstoward the lac operators (Fig. 1A, iii and iv).In the absence of lac repressor, replication generated a pop-

ulation of lagging strands of ∼0.5 kb and leading strands of 1.3and 5.2 kb (Fig. 1A, vi and B, lane 1). In the presence of re-pressor, 3.5-kb, rather than 5.2-kb, leading strands were gener-ated (Fig. 1A, vi and B, lane 2 and Fig. S1), as expected ifmovement of replisomes was inhibited within the repressor–operator array. Subsequent addition of IPTG to these blockedforks relieved this inhibition (Fig. 1B, lane 3), reflecting theinitial retention of activity of replisomes blocked at lac repressor–operator complexes (18).The impact of helicases on the ability of blocked replisomes to

resume replication upon addition of IPTG was analyzed. Rep,UvrD, and PcrA did not promote replication through the 22repressor–operator complexes in the absence of IPTG (Fig. 1B,lanes 5, 7, and 9). Addition of IPTG allowed resumption ofreplication even in the presence of these helicases, indicatingthat Rep, UvrD, and PcrA did not inhibit resumption of repli-cation upon removal of a block (Fig. 1 B, i, lanes 4, 6, and 8 and1 B, ii). In contrast, RecD2 reduced severely the ability ofreplisomes to resume replication after IPTG addition (Fig. 1 B, i,compare lanes 10 and 11 with lanes 3 and 2, and 1 B, ii).However, a mutant RecD2 protein, pinless RecD2, that retainsDNA binding but not helicase activity (23) failed to inhibit re-sumption of replication (Fig. 1C). The helicase activity of RecD2thus inhibits continued movement of paused replisomes upondissociation of a blocking nucleoprotein complex. T4 Dda, an-other superfamily 1 5′-3′ helicase, also inhibited resumption offork movement (Fig. S2), demonstrating that fork inactivation is

a general feature of this class of helicases rather than an activityassociated specifically with RecD2.

RecD2 Inactivates Paused but Not Elongating Replisomes. Inhibitionof resumption of fork movement upon dissociation of the pro-tein–DNA block could be explained by RecD2-catalyzed in-activation of paused replisomes, of elongating replisomes, orboth. To distinguish between these possibilities we assessed theimpact of RecD2 on the activity of elongating and of pausedforks in two parallel experiments. First, the effects of wild-typeand pinless RecD2 were analyzed on elongating replication forkswhose movement around a supercoiled plasmid template lackingengineered barriers was sustained by DNA gyrase (Fig. 2 A, i).Addition of either wild-type or pinless RecD2 alongside thereplication initiator DnaA had no significant impact on levels ofDNA synthesis by elongating forks (Fig. 2 A, ii). Second, repli-cation of the same supercoiled plasmid template was initiated byDnaA in the absence of a topoisomerase as in Fig. 1, resulting infork stalling owing to topological strain (Fig. 2 B, i). Subsequentaddition of a restriction enzyme relieved the strain and allowedpaused forks that retained function to continue (Fig. S1).However, addition of wild-type, but not pinless, RecD2 alongsidethe restriction enzyme inhibited subsequent DNA synthesis (Fig.2 B, ii). RecD2 helicase activity results, therefore, in inactivationof forks paused by positive supercoiling. Taken together, thedata in Fig. 2 indicate that RecD2 inactivates paused but notelongating replisomes.

RecD2 Inactivates Forks Paused by a Variety of Replicative Barriers.Different obstacles have different impacts on replisome move-ment. Nucleoprotein complexes and topological strain requireremoval of the original block for resumption of replication (Figs.1 and 2). In contrast, DNA lesions pause forks but can eventuallybe bypassed. A cyclobutane pyrimidine dimer within the leadingstrand template pauses fork progression, but leading strandsynthesis can be reinitiated downstream of the lesion over thecourse of several minutes (11). We tested, therefore, whetherRecD2 inhibited the ability of replisomes paused at a pyrimidinedimer to bypass the lesion and continue replication.Reactions were again initiated in the absence of a topoisom-

erase followed by restriction enzyme cleavage in the presence oflabeled deoxynucleotide to allow forks to progress. The plasmidtemplate contained a single pyrimidine dimer within the leadingstrand template (11). Following restriction enzyme cleavage for

No

helic

ase

Rep

Uvr

DP

crA

Rec

D2

0.6

2.02.3

4.4

6.6

0.13

9.4

+- + ++++ ++++

++++- ----+-

LacIIPTG

0

0.8

0.2

0.6

0.4

No heli

case

RepUvrDPcrA

RecD2

Rel

ativ

e le

vels

of

5.2

kb p

rodu

ct

1.0

α32P dCTP

A B Ci ii

0.6

2.02.3

4.46.6

0.13

9.4

- + +++LacIIPTG - + ++-

wild type RecD2 - +- --pin-less RecD2 - +- --

1 5

lacO22 ii iii

iv

EagI

3.0 - 3.8 kb

i

oriC

1.3 kb1.3 kb

5.2 kb

UnlabelleddCTP

Helicase

+

IPTG

vi

Active replisomesInactive replisomes 1 11

1 kb

v

Fig. 1. RecD2 inactivates replication forks blocked by nucleoprotein complexes in vitro. (A) reaction scheme to monitor the ability of replisomes halted atLacI–lacO complexes to continue replication upon IPTG-induced dissociation of the barrier. (B, i) Denaturing agarose gel of replication products formed withpPM561 in the absence or the presence of 400 nM LacI with or without subsequent addition of 1 mM IPTG. Helicases added to 100 nM final concentration atstep iv/v (A) are indicated. Sizes of markers in kilobases are indicated. (B, ii) Levels of the 5.2-kb leading strand products formed in the presence of repressorafter subsequent addition of IPTG relative to those obtained in the absence of repressor (see lane 1 in B). (C) The effects of wild-type and pinless RecD2 on theability of forks paused at LacI–lacO complexes to continue upon addition of IPTG.

Gupta et al. PNAS | April 30, 2013 | vol. 110 | no. 18 | 7253

BIOCH

EMISTR

Y

Dow

nloa

ded

by g

uest

on

Aug

ust 9

, 202

0

50 s, excess unlabeled deoxynucleotide was added with or with-out RecD2 or pinless RecD2 and incubation continued. Oneminute after addition of the restriction enzyme all reactions con-tained predominantly stalled replication forks (Fig. 3B, lanes 1, 5,and 9) (11). In the absence of RecD2 the majority of stalled forksgenerated full-length products after 8 min (Fig. 3B, lanes 1–4).Wild-type, but not pinless, RecD2 inhibited this conversion ofstalled forks into full-length products (Fig. 3B, compare lanes 5–8with 9–12, and 3C). RecD2 helicase therefore prevents repli-somes paused by a DNA lesion from bypassing the DNA damage.The data in Figs. 1–3 indicate that RecD2 helicase activity

results in loss of function of paused, but not elongating, repli-somes regardless of the nature of the replicative barrier.

Absence of Rep Helicase Hypersensitizes Cells to recD2 Expression.Inactivation of paused but not elongating replication forks byRecD2 in vitro provides a potential tool to probe fork pausing invivo. Conversion of paused forks that retain the ability to con-tinue replication into inactivated forks that cannot continueupon removal or bypass of blocks might present viability prob-lems, the severity of which would depend upon the frequencyand duration of fork pausing. However, induction of recD2 ex-pression from a plasmid-based arabinose-inducible promoterhad no significant impact on viability as monitored by colony-

forming ability (10) (Fig. 4 A and B, i). Chromosomal DNAcontent of wild-type cells induced for recD2 expression was af-fected significantly, however, as monitored by flow cytometryunder run-out conditions (Fig. 4E, i). Thus, RecD2 did havean impact upon chromosome duplication. This inhibition ofchromosomal duplication required RecD2 helicase activity, notjust DNA binding, because recD2pinless had no impact on DNAcontent (Fig. 4E, ii). Sensitivity to wild-type but not pinlessRecD2 in vivo correlates with inactivation of paused forks bywild-type but not pinless RecD2 in vitro (Figs. 1C, 2B, and 3).The ability to survive recD2 expression could be due to repair

of RecD2-inactivated forks providing an efficient means of sur-viving such inactivation. Recombination enzymes provide mul-tiple pathways to repair damaged replication forks (24, 25).However, no significant impact on viability was observed in theabsence of any recombination gene tested, suggesting that forkprocessing is not critical for survival of cells expressing recD2(Fig. 4A; ΔrecA, ΔrecB, and ΔrecF) (Table S1).Alternatively, survival could reflect a low frequency and/or

short duration of replication fork pauses in wild-type cells. Lowlevels of replisome pausing might be due to an inherently lowfrequency of pausing but might also be a consequence of multipleenzyme systems in wild-type cells that minimize fork pausing. Wetherefore analyzed the impact of recD2 expression in strains witha decreased ability to remove potential obstacles to replication.Lesions within the leading strand template cause replisomepausing before bypass of the lesion via repriming (11) (Fig. 3).However, absence of nucleotide excision repair did not rendercells sensitive to RecD2 (Fig. 4A, ΔuvrA, ΔuvrB, and ΔuvrC).Absence of Mfd, a dsDNA translocase that promotes transcription-coupled repair of DNA lesions via UvrABC (26), also did notrender cells sensitive to RecD2 (Fig. 4A). Thus, removal of bulkylesions from either nontranscribed or transcribed DNA was notrequired for tolerance of recD2 expression. Similarly, absence of

A

oriC

Replisome components+ DNA gyrase+ DnaA RecD2/Pin-less RecD2

0

10

20

30

40

50

60

70

DN

A sy

nthe

sis

(pm

ol d

CTP

)

Replisome components- DNA gyrase+DnaA

Positively supercoiledintermediate

EagI RecD2/Pin-less RecD2

DN

A sy

nthe

sis

(pm

ol d

CTP

)

No Rec

D20

10

8

6

12

2

4

Pin-les

s Rec

D2

RecD2

B

No Rec

D2

Pin-les

s Rec

D2

RecD2

-+

-+

i ii

i ii

+

oriC

1 kb

Fig. 2. RecD2 inactivates blocked but not elongating replication forks. (A, i)Method to analyze the impact of RecD2 on elongating replication forks. (A,ii) DNA synthesis in the presence of DNA gyrase upon simultaneous additionof DnaA with or without either wild-type or pinless RecD2, each present at100 nM final concentration. (B, i) Method to monitor the impact of RecD2 onthe ability of forks stalled by positive supercoiling to continue DNA synthesisupon relief of topological strain. (B, ii) Levels of DNA synthesis upon additionof a restriction enzyme with or without either wild-type or pinless RecD2(each at 100 nM).

A BTime

RecD2

23.1

9.46.6

4.4

StalledforkFull

length

No RecD2

0

1

654321 87Time (mins)

No RecD2RecD2Pin-less RecD2

1 2 3 4 5 6 7 8 9 10 11 12

Pin-lessRecD2

EcoRI [α32P]dATP

ii

iii

iv

i CPD

C

+

Unlabelled dATPRecD2+-

v

0.4

0.2

0.6

0.8

(Ful

l len

gth)

(Ful

l len

gth

+ fo

rk)

oriC

1 kb

Fig. 3. RecD2 inactivates forks paused by template lesions. (A) reactionscheme to monitor replication fork pausing at, and bypass of, a cyclobutanepyrimidine dimer (CPD) within the leading strand template. (B) Native aga-rose gel of replication products formed on plasmid DNA harboring a pyrim-idine dimer within the leading strand template. Time points were taken 1, 2,4, and 8 min after addition of restriction enzyme and radiolabel. Positions ofreplication forks stalled at the pyrimidine dimer and full-length productsgenerated by bypass of the lesion are indicated. Wild-type and pinless RecD2were present at 100 nM, as indicated. (C) Accumulation of full-lengthproducts as a function of time.

7254 | www.pnas.org/cgi/doi/10.1073/pnas.1303890110 Gupta et al.

Dow

nloa

ded

by g

uest

on

Aug

ust 9

, 202

0

one or both of the apurinic/apyrimidinic endonucleases requiredfor base excision repair did not result in hypersensitivity toRecD2 (Fig. 4A, ΔxthA and Δnfo). Efficient removal of DNAlesions, and thus a reduction in their potential to pause forks,was not required therefore for tolerance of recD2 expression.Rep helicase provides another means to reduce fork pausing

by acting at the replisome to disrupt nucleoprotein complexesahead of the replication fork (10, 22, 27). Wild-type but notpinless RecD2 had a severe impact on colony-forming ability inthe absence of Rep (Fig. 4 A and B). Absence of Rep also causedhypersensitivity to T4 Dda (Fig. S3A). The toxicity of bothRecD2 and Dda in Δrep cells correlated, therefore, with theability of both helicases to inactivate replisomes in vitro. Therequirement for Rep to ameliorate the impact of RecD2 oncolony-forming ability was dependent on Rep helicase activity,not just DNA binding, because a rep allele bearing a mutation inthe Walker A motif was sensitive to RecD2 (Fig. 4C, iii). ThusRep does not merely block access of RecD2 to a DNA substrate.Chromosomal DNA content upon recD2 induction in Δrep cellswas also reproducibly more severely affected compared with rep+

cells (Figs. 4E, i and iii and 5E, i and F, i).Taken together, these data indicate that inactivation of paused

but otherwise functional replisomes by RecD2 results in chro-mosomal defects that, although exhibited by wild-type cells, areexacerbated in the absence of Rep. The corollary of these

observations is that many fork pausing events do not lead toreplisome inactivation under normal circumstances in either rep+

or Δrep cells and that conversion of these paused forks into in-active forks by RecD2 is deleterious.

Clearance of Nucleoprotein Barriers Ahead of Forks Protects AgainstRecD2-Induced Cell Death. Rep facilitates both clearance of nu-cleoprotein complexes ahead of forks (10, 22) and PriC-directedreloading of the replication apparatus (28, 29). However, thecolony-forming ability of a strain lacking PriC was not reduced byRecD2 (Fig. 4A) or T4 Dda (Fig. S3B), indicating that failure ofPriC-directed replisome reloading is not responsible for the in-viability of Δrep cells expressing RecD2.Efficient clearance of nucleoprotein complexes ahead of forks

by Rep is dependent upon not only Rep helicase activity but alsoa physical interaction between the C terminus of Rep and thereplicative helicase DnaB (10, 30). A rep allele lacking the DnaBinteraction domain but retaining helicase activity (repΔC33) wassensitive to RecD2 expression (Fig. 4D), consonant with efficientclearance of protein–DNA complexes ahead of forks beingneeded to minimize RecD2-induced viability defects. Both UvrDand DinG helicases have also been implicated in promotingreplication of protein-bound DNA (10, 22), but neither helicaseis known to associate physically with the replisome (31). Over-expression of recD2 had no impact on colony-forming ability in

Wild

type

Δrep

1

0.1

0.01

0.001

0.0001

Rel

ativ

e su

rviv

al

ΔdinG

A

B

(ii) Δrep(iii) repK28R

(i) rep+

recD2

D

C

(i) rep+

(ii) repΔC33recD2

10

ΔuvrD

ΔrecAΔrec

BΔrec

FΔuv

rAΔuv

rCΔmfd

ΔxthA ΔnfoΔnfo

ΔxthA

ΔuvrB

ΔpriC

- arabinose + arabinose

(ii) ΔreprecD2

(i) rep+

(iv) Δrep

(iii) rep+

recD2pinless

841 2 2

(iii) Δrep / recD2

(i) rep+ / recD2 (ii) rep+ / recD2pinless

(iv) Δrep / recD2pinless

841

E

Fig. 4. Expression of recD2 is toxic in strains lacking Rep. (A) Colony-form-ing ability of wild-type E. coli (BW25113) and otherwise isogenic strainsbearing single gene deletions upon expression of recD2 using a plasmid-based arabinose-inducible system (pMG31). Survival is represented by thenumber of colonies formed upon induction of recD2 expression relative tothe number of colonies formed in the absence of induction. Table S2 givesstrain numbers. (B) Colony-forming ability of rep+ (TB28) and Δrep (N6577)strains harboring pBADrecD2 and pBADrecD2pinless. (C) Colony-formingability of rep+ (MG1655), rep (N4982), and repK28R (SS1076) containingpBADrecD2. (D) Colony-forming ability of rep+ (MKG08) and repΔC33(MKG10) upon recD2 overexpression (pMG31). (E) DNA content of (i and ii)wild-type (TB28) and (iii and iv) Δrep (N6577) cells without and with in-duction of expression (unshaded and shaded histograms, respectively) frompBADrecD2 and pBADrecD2pinless as monitored by flow cytometry. DNAcontent with respect to number of chromosome equivalents per cell is in-dicated below. Note that Δrep cells have a higher mean number of chro-mosomes per cell compared with rep+ (compare the uninduced samples in iand ii with iii and iv), as noted previously (39).

A

(ii) Δrep rpo+

(iii) rep+ rpoB[H1244Q]

(vii) rep+ rpoB[H551P]

(v) rep+ rpoB[G1260D]

(i) rep+ rpo

+

(iv) Δrep rpoB[H1244Q]

(viii) Δrep rpoB[H551P]

(vi) Δrep rpoB[G1260D]

10-1 10-5- arabinose

B

841 2

(i) Δrep rpo+

(ii) rep+

rpoB[H1244Q]

(iii) rep+

rpoB[G1260D]

(i) rep+ rpo

+

(ii) Δrep

rpoB[H1244Q]

(iii) Δrep

rpoB[G1260D]

841 2 841 2

C

rep+ r

poB

+

Δrep rpoB

+

Rel

ativ

e su

rviv

al

Δrep rpoB[H

1244Q]

rep+ r

poB[H

1244Q]

100

10-4

10-3

10-2

10-1

101

0.02% arabinose 0.2% arabinose

D E

F

10-1 10-5 10-1 10-5

Fig. 5. RecD2-directed lethality is associated with transcription complexes.(A–C) Colony-forming ability of rep+ and Δrep strains bearing either wild-type or mutant rpoB alleles upon no, low-, or high-level induction frompBADrecD2 (0%, 0.02% and 0.2% arabinose, respectively). Strains (i–viii) areTB28, N6577, PM486, N7604, AM2158, HB278, N7616, and HB280. (D) Col-ony-forming ability of rep+ and Δrep strains bearing rpoB+ or rpoB[H1244Q]upon induction of recD2 expression with 0.2% arabinose. Strain numbers areas in A–C. (E and F) DNA content of rep+ and Δrep strains bearing rpoB+,rpoB[H1244Q], or rpoB[G1260D] without and with induction of expression ofrecD2 with 0.2% arabinose (unshaded and shaded histograms, respectively).The number of chromosome equivalents per cell is indicated below. Notethat both rpoB[H1244Q] and rpoB[G1260D] also reduced the median num-ber of chromosome equivalents in Δrep cells in the absence of recD2 ex-pression from eight to four, the same number as seen in rep+ cells (F,compare i with ii and iii). This may reflect suppression of the reduced rate ofgenome duplication in Δrep cells (30). Strain numbers are as in A–C.

Gupta et al. PNAS | April 30, 2013 | vol. 110 | no. 18 | 7255

BIOCH

EMISTR

Y

Dow

nloa

ded

by g

uest

on

Aug

ust 9

, 202

0

the absence of either UvrD or DinG (Fig. 4A), supporting theconclusion that this toxicity is specific for the replisome-associ-ated activity of Rep.These data indicate that promotion of replisome movement

by Rep along protein-coated DNA may be essential to avoidRecD2-directed cell death. A major source of nucleoproteinreplicative barriers is transcription, with mutations within RNApolymerase subunits being able to suppress multiple defects ingenome duplication (20, 32–34). One such mutation, rpoB[H1244Q], suppresses defects in genome duplication by reducingthe stability of stalled transcription complexes (20) and inhibitingbacktracking of RNA polymerase (35). We found that this mu-tation suppressed the RecD2-dependent killing of Δrep cells asevinced by colony-forming ability (Fig. 5 A–C, compare i and iiwith iii and iv, and 5D). Flow cytometry was also used to analyzeDNA content. Although these analyses used rifampicin, an in-hibitor of transcription initiation, to prevent replication reinitia-tion and simplify DNA content analysis, comparison of rpoB+ withrpoB[H1244Q] revealed that rpoB[H1244Q] reduced the impact ofRecD2 on chromosomal DNA content in both rep+ and Δrep cells(compare i and ii in Fig. 5 E and F). However, suppression by rpoB[H1244Q] in Δrep cells was only partial, with both colony sizesreduced and DNA content still perturbed significantly by highlevel expression of recD2 (Fig. 5C, iii and iv, and F, ii).Other mutations in rpo genes with the ability to suppress ge-

nome duplication defects also restored colony-forming ability toΔrep cells, confirming that this suppression was not specific torpoB[H1244Q] (Fig. 5B, v–viii). Indeed, rpoB[G1260D] (33)resulted in effective suppression of RecD2 toxicity even withhigh-level recD2 expression, as indicated by colony sizes andDNA content in Δrep cells (Fig. 5C, compare v and vi, and 5F,compare i and iii), suppression that was also apparent with ddaexpression in Δrep cells (Fig. S3C). This suppression of toxicitysuggested that transcription elongation factors might also pro-vide a key means of ameliorating the consequences of pausedfork inactivation. However, the simultaneous absence of fourfactors known to reduce RNA polymerase pausing and back-tracking (26) did not render cells sensitive to recD2 expression(Fig. S4).These data indicate that transcription complexes are the pri-

mary cause of RecD2-directed DNA content defects in both rep+

and Δrep cells and of lethality in Δrep cells. Together with ourdemonstration that RecD2 inactivates paused forks in vitro,these findings indicate that the most significant cause of pausedreplisomes in otherwise unperturbed cells in vivo are tran-scription complexes. Moreover, the primary means of resumingreplication from these paused replisomes is via an accessoryreplicative helicase.

DiscussionWe have demonstrated that inactivation of paused but notelongating E. coli replisomes is effected in vitro by a heterolo-gous helicase, RecD2. Using this as an in vivo probe we haveshown that many pausing events do not lead to fork inactivation,that protein–DNA complexes, not DNA damage, are the mainsources of these transient pauses in unperturbed cells, and thataccessory replicative helicases are the primary means by whichthe frequency and/or the duration of such pauses are reduced.Although frequent replisome pausing and subsequent re-

sumption of replication by the same fork is often assumed forboth prokaryotes and eukaryotes (14, 24), there is little directevidence to support such views. Indeed, fork pausing is not aninherent property of the replisome itself (36). Use of RecD2 todeactivate paused but still active replisomes has allowed us toprobe such pausing events in vivo. Expression of this helicase inwild-type cells, although not lethal, did have a dramatic impacton chromosomal DNA content (Fig. 4E). Thus, replisomes dopause and subsequently continue duplication at a frequency that

has the potential to perturb chromosome metabolism signifi-cantly. Although in eukaryotes the multiple origins per chro-mosome together with dormant but activatable origins providemechanisms not available in bacteria to rescue inactivated forks(25), it is becoming apparent that inactivation of paused forkscontributes to gross chromosomal rearrangements in higherorganisms (7). Our data provide a direct demonstration of themajor hidden potential for this initiation of genome instabilityduring the course of chromosome duplication.The survival of wild-type cells upon exposure to RecD2

allowed screening for mechanisms that ameliorate the con-sequences of paused fork inactivation. Repair of RecD2-inacti-vated forks by recombination enzymes was not critical forsurvival (Fig. 4 and Table S1), suggesting that mechanisms thatact before fork inactivation are important. Removal of DNAlesions was not required to survive recD2 expression, but Rep-promoted duplication of protein-bound DNA was critical forsurvival (Fig. 4 A, D, and E). Forks paused at both templatedamage and protein–DNA complexes were inactivated byRecD2 in vitro (Figs. 1 and 3), and so an inability to inactivateforks paused at DNA lesions could not explain this differentialRecD2 toxicity. Protein–DNA complexes, rather than DNAlesions, therefore dominate the formation of paused but activereplication forks in otherwise unperturbed cells. This conclusionis supported by the suppression of RecD2 toxicity in both wild-type and Δrep cells by mutations in RNA polymerase (Fig. 5),a known source of nucleoprotein barriers to replication in bothprokaryotes and eukaryotes (19, 32). Moreover, cells lacking fourfactors known to underpin RNA polymerase movement did notdisplay hypersensitivity to recD2 expression, in contrast to cellslacking Rep (Fig. S4). An accessory replicative helicase, ratherthan transcription elongation factors, provides therefore a keymechanism for minimizing replisome pausing.The lack of requirement for transcription elongation factors to

minimize formation of paused but still active forks is in apparentconflict with the requirement for such factors to minimize theneed for repair of stalled forks by recombination enzymes (20).However, recombination enzymes such as RuvABC act on forksthat no longer possess an active replisome (24) rather than onthose that are paused but still retain function. The implication isthat transcription factors are important within the context ofinactive forks that require replisome reloading via recombinationenzymes. Perhaps this requirement reflects the critical impor-tance of such factors in minimizing the formation of backed-uparrays of stalled transcription complexes, formidable obstacles toreplication that might not be surmountable by paused, activereplisomes (20).There is cross-talk between DNA damage, transcription com-

plexes, and replication fork pausing. Lesions are potent blocks toRNA as well as DNA polymerases and could contribute to thegeneration of paused replisomes. However, the ability of cellswith defects in excision repair, including transcription-coupledrepair, to survive expression of recD2 indicates that transcriptioncomplexes stalled at DNA lesions are not major causes of pausedbut active replisomes (Fig. 4A). Other factors such as tem-plate-directed RNA polymerase pausing, a ubiquitous featureof gene expression, must therefore dominate the generation oftranscriptional barriers to replication (35).Our data indicate that protein–DNA complexes are the most

frequently encountered sources of active replisome pausing inthe absence of elevated levels of DNA-damaging agents. How-ever, the relative importance of nucleoprotein complexes versusDNA lesions will be affected by environmental sources of DNAdamage as well as intracellular factors. The duration as well asthe frequency of replisome pausing at leading strand lesions mustalso be considered with respect to genome stability. Pausing atnucleoprotein complexes is dictated by the rate of dissociation ofthe protein from the duplex template, with accessory replicative

7256 | www.pnas.org/cgi/doi/10.1073/pnas.1303890110 Gupta et al.

Dow

nloa

ded

by g

uest

on

Aug

ust 9

, 202

0

helicases effectively reducing the half-life of the barrier (10). It isnot known whether excision repair enzymes can access DNAlesions within the context of a paused replisome to similarly re-duce the barrier half-life. Lesion-induced pausing of replisomesmight therefore be less frequent but of greater duration thanpausing at protein–DNA complexes in vivo.Bacterial superfamily 1 helicases with a 3′-5′ polarity of

translocation can promote fork movement along protein-boundDNA (10). Here we have shown that superfamily 1 5′-3′ helicasesnot only lack accessory helicase function in bacteria but alsocatalyze inactivation of paused replisomes. Although we havebeen unable to identify the mechanism by which RecD2 inacti-vates paused E. coli replisomes, our data provide a dramaticdemonstration of the critical importance of regulating helicaseactivity. If RecD2 has the capacity to inactivate paused repli-somes in D. radiodurans as well as E. coli, then RecD2 activitymust be regulated stringently in this organism. Indeed, the ex-treme toxicity of RecD2 in E. coli suggests that multiple levels ofregulation may be needed to ensure helicase activities areunleashed only where and when required. This challenge may becompounded in more complex organisms by the greater numbersand diversity of these potentially toxic motors needed to sustainnucleic acid metabolism.

MethodsPlasmids, Proteins, and Strains. Construction of pPM561, pBADrecD2,pBADrecD2pinless, pBADdda, and pMG31 are described in SI Methods.

LacI, Rep, UvrD, β, HU, DnaA, DnaB, DnaC, DnaG, single-stranded bindingprotein (SSB), and DNA polymerase III α, e, θ, τ, χ, ψ, δ, and δ′ subunits werepurified as described (10, 13, 37, 38). B. stearothermophilus PcrA, D. radio-durans wild-type and pinless RecD2, E. coli DNA gyrase, and T4 Dda werekind gifts of Panos Soultanas (University of Nottingham, Nottingham, UK),

Dale Wigley (Chester Beatty Laboratories, Institute of Cancer Research,London, UK), Tony Maxwell (John Innes Centre, Norwich, UK), and KevinRaney (University of Arkansas for Medical Sciences, Little Rock, AR). EagI wassupplied by New England Biolabs. All protein concentrations refer tothe monomer.

Strains are listed in Table S2.

In Vitro Replication Assays. Assays to monitor the functionality of forksblocked by nucleoprotein complexes or positive supercoiling are described inSI Methods. To analyze the impact of test helicases on the ability of repli-somes to replicate plasmid DNA in the absence of a replicative block (Fig.2A), reactions containing pPM561 were assembled containing DNA gyrase(140 nM) but without LacI and replication was initiated as outlined in SIMethods. For analysis of replisome stability at a cyclobutane pyrimidine di-mer, proteins and DNA template were prepared as described (11) andreactions performed as described in SI Methods.

Colony Formation Assays and Flow Cytometry. Strains were transformed withthe indicated plasmids and colonies selected on LB agar containing either100 μg·mL−1 ampicillin (pMG31) or 30 μg·mL−1 kanamycin (pBADrecD2,pBADrecD2pinless, and pBADdda) at 37 °C overnight. Single colonies weregrown subsequently in LB broth plus either ampicillin or kanamycin at37 °C to an A650 of 0.4 before serial 10-fold dilutions were made with 56/2salts on ice. Five microliters of the dilutions were then spotted onto LB agarcontaining ampicillin or kanamycin plus the indicated concentrations ofarabinose. Plates were incubated at 37 °C for 16 h. Flow cytometry wasperformed on midlog phase cultures after treatment with rifampicin andcephalexin as described in SI Methods.

ACKNOWLEDGMENTS.We thank Panos Soultanas, Dale Wigley, Kevin Raney,and Tony Maxwell for supplying proteins; Steve Sandler for supplying SS1076;and Bénédicte Michel for supplying JJC735. This work was supported by Bio-technology and Biological Sciences Research Council Grants BB/G005915/1and BB/I001859/1 (to P.M.), National Institutes of Health Grant GM34557 (toK.J.M.), and Medical Research Council Grant G0800970 (to R.G.L.).

1. French S (1992) Consequences of replication fork movement through transcriptionunits in vivo. Science 258(5086):1362–1365.

2. Lindahl T, Wood RD (1999) Quality control by DNA repair. Science 286(5446):1897–1905.3. Paeschke K, Capra JA, Zakian VA (2011) DNA replication through G-quadruplex motifs

is promoted by the Saccharomyces cerevisiae Pif1 DNA helicase. Cell 145(5):678–691.4. Heller RC, Marians KJ (2006) Replisome assembly and the direct restart of stalled

replication forks. Nat Rev Mol Cell Biol 7(12):932–943.5. Horiuchi T, Fujimura Y (1995) Recombinational rescue of the stalled DNA replication

fork: A model based on analysis of an Escherichia coli strain with a chromosome re-gion difficult to replicate. J Bacteriol 177(3):783–791.

6. Lambert S, Watson A, Sheedy DM, Martin B, Carr AM (2005) Gross chromosomal re-arrangements and elevated recombination at an inducible site-specific replicationfork barrier. Cell 121(5):689–702.

7. Mizuno K, Miyabe I, Schalbetter SA, Carr AM, Murray JM (2013) Recombination-re-started replication makes inverted chromosome fusions at inverted repeats. Nature493(7431):246–249.

8. Pomerantz RT, O’Donnell M (2008) The replisome uses mRNA as a primer after col-liding with RNA polymerase. Nature 456(7223):762–766.

9. Pomerantz RT, O’Donnell M (2010) Direct restart of a replication fork stalled bya head-on RNA polymerase. Science 327(5965):590–592.

10. Guy CP, et al. (2009) Rep provides a second motor at the replisome to promote du-plication of protein-bound DNA. Mol Cell 36(4):654–666.

11. Yeeles JT, Marians KJ (2011) The Escherichia coli replisome is inherently DNA damagetolerant. Science 334(6053):235–238.

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

13. Payne BT, et al. (2006) Replication fork blockage by transcription factor-DNA com-plexes in Escherichia coli. Nucleic Acids Res 34(18):5194–5202.

14. Alzu A, et al. (2012) Senataxin associates with replication forks to protect fork in-tegrity across RNA-polymerase-II-transcribed genes. Cell 151(4):835–846.

15. Rupp WD, Howard-Flanders P (1968) Discontinuities in the DNA synthesized in anexcision-defective strain of Escherichia coli following ultraviolet irradiation. J Mol Biol31(2):291–304.

16. McInerney P, O’Donnell M (2004) Functional uncoupling of twin polymerases:mechanism of polymerase dissociation from a lagging-strand block. J Biol Chem279(20):21543–21551.

17. Marians KJ, Hiasa H, Kim DR, McHenry CS (1998) Role of the core DNA polymerase IIIsubunits at the replication fork. α is the only subunit required for processive repli-cation. J Biol Chem 273(4):2452–2457.

18. McGlynn P, Guy CP (2008) Replication forks blocked by protein-DNA complexes havelimited stability in vitro. J Mol Biol 381(2):249–255.

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

20. Trautinger BW, Jaktaji RP, Rusakova E, Lloyd RG (2005) RNA polymerase modulatorsand DNA repair activities resolve conflicts between DNA replication and transcription.Mol Cell 19(2):247–258.

21. Lehmann AR (1972) Postreplication repair of DNA in ultraviolet-irradiated mamma-lian cells. J Mol Biol 66(3):319–337.

22. Boubakri H, de Septenville AL, Viguera E, Michel B (2010) The helicases DinG, Rep andUvrD cooperate to promote replication across transcription units in vivo. EMBO J29(1):145–157.

23. Saikrishnan K, Griffiths SP, Cook N, Court R, Wigley DB (2008) DNA binding to RecD:Role of the 1B domain in SF1B helicase activity. EMBO J 27(16):2222–2229.

24. Atkinson J, McGlynn P (2009) Replication fork reversal and the maintenance of ge-nome stability. Nucleic Acids Res 37(11):3475–3492.

25. Petermann E, Helleday T (2010) Pathways of mammalian replication fork restart. NatRev Mol Cell Biol 11(10):683–687.

26. McGlynn P, Savery NJ, Dillingham MS (2012) The conflict between DNA replicationand transcription. Mol Microbiol 85(1):12–20.

27. Atkinson J, Gupta MK, McGlynn P (2011) Interaction of Rep and DnaB on DNA. NucleicAcids Res 39(4):1351–1359.

28. Sandler SJ (2000) Multiple genetic pathways for restarting DNA replication forks inEscherichia coli K-12. Genetics 155(2):487–497.

29. Heller RC, Marians KJ (2005) Unwinding of the nascent lagging strand by Rep and PriAenables the direct restart of stalled replication forks. J Biol Chem 280(40):34143–34151.

30. Atkinson J, et al. (2011) Localization of an accessory helicase at the replisome is criticalin sustaining efficient genome duplication. Nucleic Acids Res 39(3):949–957.

31. McGlynn P (2013) Helicases at the replication fork. Adv Exp Med Biol 767:97–121.32. McGlynn P, Lloyd RG (2000) Modulation of RNA polymerase by (p)ppGpp reveals

a RecG-dependent mechanism for replication fork progression. Cell 101(1):35–45.33. Trautinger BW, Lloyd RG (2002) Modulation of DNA repair by mutations flanking the

DNA channel through RNA polymerase. EMBO J 21(24):6944–6953.34. Baharoglu Z, Lestini R, Duigou S, Michel B (2010) RNA polymerase mutations that

facilitate replication progression in the rep uvrD recF mutant lacking two accessoryreplicative helicases. Mol Microbiol 77(2):324–336.

35. Dutta D, Shatalin K, Epshtein V, Gottesman ME, Nudler E (2011) Linking RNA poly-merase backtracking to genome instability in E. coli. Cell 146(4):533–543.

36. Yao NY, Georgescu RE, Finkelstein J, O’Donnell ME (2009) Single-molecule analysisreveals that the lagging strand increases replisome processivity but slows replicationfork progression. Proc Natl Acad Sci USA 106(32):13236–13241.

37. Atkinson J, et al. (2009) Stimulation of UvrD helicase by UvrAB. J Biol Chem 284(14):9612–9623.

38. Marians KJ (1995) ϕ X174-type primosomal proteins: Purification and assay. MethodsEnzymol 262:507–521.

39. Lane HE, Denhardt DT (1974) The rep mutation. III. Altered structure of the repli-cating Escherichia coli chromosome. J Bacteriol 120(2):805–814.

Gupta et al. PNAS | April 30, 2013 | vol. 110 | no. 18 | 7257

BIOCH

EMISTR

Y

Dow

nloa

ded

by g

uest

on

Aug

ust 9

, 202

0