Single-molecule analysis reveals that the lagging strand increases replisome processivity but slows replication fork progression.

Abstract

Single-molecule techniques are developed to examine mechanistic features of individual E. coli replisomes during synthesis of long DNA molecules. We find that single replisomes exhibit constant rates of fork movement, but the rates of different replisomes vary over a surprisingly wide range. Interestingly, lagging strand synthesis decreases the rate of the leading strand, suggesting that lagging strand operations exert a drag on replication fork progression. The opposite is true for processivity. The lagging strand significantly increases the processivity of the replisome, possibly reflecting the increased grip to DNA provided by 2 DNA polymerases anchored to sliding clamps on both the leading and lagging strands.

Full text

Page 1

Single-molecule analysis reveals that the lagging
strand increases replisome processivity but slows
replication fork progression
Nina Y. Yao1, Roxana E. Georgescu1, Jeff Finkelstein, and Michael E. O’Donnell2
Howard Hughes Medical Institute, Rockefeller University, 1230 York Avenue, New York, NY 10021
Contributed by Michael E. O’Donnell, June 3, 2009 (sent for review May 14, 2009)
Single-molecule techniques are developed to examine mechanistic
features of individual E. coli replisomes during synthesis of long
DNA molecules. We find that single replisomes exhibit constant
rates of fork movement, but the rates of different replisomes vary
over a surprisingly wide range. Interestingly, lagging strand syn-
thesis decreases the rate of the leading strand, suggesting that
lagging strand operations exert a drag on replication fork progres-
sion. The opposite is true for processivity. The lagging strand
significantly increases the processivity of the replisome, possibly
reflecting the increased grip to DNA provided by 2 DNA poly-
merases anchored to sliding clamps on both the leading and
lagging strands.
polymerase ? clamp loader ? replicase ? sliding clamp ? helicase
R
a primase for initiation of lagging strand Okazaki fragments
(1–3). Replisomes also contain ring shaped sliding clamp pro-
teins that tether both polymerases to DNA for high processivity
[reviewed in ref. 2]. Sliding clamp proteins require a clamp
loader machine that couples ATP hydrolysis to open and close
clamps onto primed sites. These several components are pro-
posedtofunctiontogetherinahighlycoordinatedfashionduring
leading and lagging strand replication (1, 2, 4, 5).
The antiparallel structure of DNA requires that 1 strand (the
lagging strand) is synthesized as fragments that are extended in
the opposite direction of the continuous leading strand. This
opposite directionality is resolved by formation of a DNA loop
during extension of each Okazaki fragment [i.e., the ‘‘trombone
model’’ of replication (5)]. Furthermore, each Okazaki fragment
requires an RNA primer and assembly of a sliding clamp to
initiate chain extension. The repeated initiation events and DNA
looping during lagging strand replication may influence the rate
and processivity of the replisome.
Replisome machines are highly processive entities that syn-
thesize extremely long DNA molecules, making their rate and
processivityquitedifficulttostudybyensemblemethodsbecause
of the low resolving limit of most gel electrophoretic techniques.
Thus, ensemble rate studies can only capture the first 10–20 s of
a rapidly moving replisome, and establish a lower limit for
processivity. Furthermore, individual replisomes may move for-
ward in an uneven fashion, and this behavior will be masked in
an ensemble analysis, which quickly loses synchronicity. Single-
molecule studies circumvent these limitations by real-time ob-
servations of individual replication forks over the entire distance
of a highly processive binding event. Long DNA products are
imaged in the microscope for direct measurements of DNA
length.
A further advantage to the single-molecule approach is the
ability to apply a constant flow of buffer during replication. The
use of a flow during replication enables a rigorous test of
replisome processivity, as proteins that dissociate from the
replisome will be quickly carried away from the reaction cham-
ber, preventing them from reassociating with the DNA.
eplisome machines contain a helicase for DNA unwinding,
2 polymerases for replication of both strands of DNA, and
The current report examines the rate and processivity of the
E. coli replisome. The E. coli replisome consists of the ring
shaped hexameric DnaB helicase, DnaG primase, the DNA
polymerase III* replicase (Pol III*), and ? sliding clamps.
Processivity estimates of the E. coli replisome from ensemble
studies indicate a lower limit of 50 kb, and an average bulk rate
of?500–700nt/sat37 °C(6,7).Processivitymeasurementsfrom
other recent single-molecule studies differ over a wide range,
from 3 kb to ?80 kb (8, 9). In each of these previous studies
replication was performed in the presence of excess proteins that
could replace a replisome protein that dissociates from DNA,
and thus the true processivity of the replisome has not been
examined.
DnaB helicase, Pol III*, and the ? clamp are tightly associated
replisome components, while primase is loosely attached and
comes on and off the replisome during synthesis (i.e., it acts
distributively) (10). The Pol III* assembly contains 2–3 Pol III
cores attached to 1 clamp loader apparatus (7). Two Pol III cores
within Pol III* are tethered to the leading and lagging strands by
? clamps, suggesting that Pol III* tightly adheres to the repli-
cation fork for high processivity. However, recent studies show
that translesion DNA polymerases (i.e., Pol II and Pol IV)
efficiently take over ? clamps from Pol III* in a moving
replisome, implying that Pol III* may dissociate and reassociate
with ? during replication (11). Indeed, the lagging strand Pol III
core within Pol III* rapidly dissociates from ? upon finishing an
Okazaki fragment, and it repeatedly cycles to new ? clamps on
upstream RNA primers (12, 13). These findings bring into
question the stability of Pol III* within the moving replisome.
The current study develops single-molecule techniques to
visualize E. coli replisomes in real-time, allowing both their rate
and processivity to be determined with precision. The results
show that DnaB and Pol III*-? are indeed highly processive, and
replicate an average of 86.5 kb of DNA in 1 binding event; some
replisomes are processive for up to 300 kb. Pol III* appears to
release from the fork before DnaB helicase because reactions
that contain an excess of Pol III* result in continued extension
ofDNAtoanaveragelengthof140kb.Wealsodevelopmethods
to study replisomes that only perform leading strand synthesis,
and compare the results to replisomes that replicate both leading
and lagging strands. Interestingly, we find that lagging strand
synthesis has opposite effects on the rate and processivity of the
replisome. However, lagging strand synthesis reduces the rate of
fork progression by ?23%, which may reflect priming or the
strain of DNA looping during Okazaki fragment synthesis.
Author contributions: N.Y.Y., R.E.G., and M.E.O. designed research; N.Y.Y. and R.E.G.
performedresearch;N.Y.Y.,R.E.G.,andJ.F.contributednewreagents/analytictools;N.Y.Y.,
R.E.G., and M.E.O. analyzed data; and N.Y.Y., R.E.G., and M.E.O. wrote the paper.
The authors declare no conflict of interest.
1N.Y.Y. and R.E.G. contributed equally to this work
2To whom correspondence should be addressed. E-mail: odonnel@rockefeller.edu.
This article contains supporting information online at www.pnas.org/cgi/content/full/
0906157106/DCSupplemental.
13236–13241 ?
PNAS ?
August 11, 2009 ?
vol. 106 ?
no. 32www.pnas.org?cgi?doi?10.1073?pnas.0906157106

Page 2

Although, lagging strand replication enhances the processivity of
the replisome by 61%. We propose that either primase or 2 DNA
polymerases attached to sliding clamps on both the leading and
lagging strands, stabilize the replisome, and enhance its proces-
sivity relative to a replisome that operates only on the leading
strand.
Results and Discussion
E. coli replisomes are highly processive and the leading and
lagging strand polymerases are tightly coupled. The E. coli
replisome was assembled onto a 5? biotinylated 100mer mi-
nicircle substrate using DnaB, Pol III*, and ?, and then immo-
bilized on a lipid bilayer through the 5? biotinylated minicircle
tail. Upon including dNTPs and rNTPs in the buffer flow, the
replisome advances around the circle multiple times (i.e., rolling
circle replication). The leading strand product is a continuous
single-strand (ss) DNA ‘‘tail’’ that becomes the template for
lagging strand synthesis, forming a long double-strand (ds) DNA
(see the scheme in Fig. 1A). The force of a hydrodynamic flow
pushes the DNA-lipid complex to a diffusion barrier etched in
the glass surface, concentrating at a defined position within the
flow cell numerous DNA molecules that can be examined
simultaneously in 1 visual field (14). Real-time rolling circle
replication is detected using a fluorescent intercalating dye
(Yo-Pro-1).
A distinctive feature of the assay design is the use of a constant
flow of buffer to deliver dNTPs and other reagents to the
immobilized replisome. We leveraged this feature to determine
the processivity of individual replisomes by omitting Pol III* and
DnaB helicase from the buffer flow. Thus, when either Pol III*
or DnaB dissociate from the replication fork, they cannot rebind
DNA because they will be carried away in the flow, and growth
oftheDNAmoleculewillcometoanabrupthalt.Weusedaflow
of 100 ?L/min, which exerts a force of only 1.45 pN (Fig. S1).
Primase and the ? clamp were included in the buffer flow
because a new molecule of primase and ? are needed to initiate
each Okazaki fragment (10, 12).
AgrowingcurtainofDNAmoleculesisobservedatthediffusion
barrier after replication is initiated (Fig. 1B and Movie S1). Most
DNAmoleculesstopgrowingwithin20min,indicatinglossofeither
DnaBorPolIII*(orboth)fromtheDNA.Thelengthofindividual
DNA molecules varies over a considerable range; some are ?30 kb
while others are ?300 kb. The length distribution of 400 individual
molecules,fittoasingleGaussianfunction,yieldsanaveragelength
of 86.5 ? 5.3 kb (Fig. 1C). DNA molecules that grow outside the
visual field show the same length distribution, and therefore
exposure to the laser does not affect the results. Ensemble exper-
iments indicate that the dye used to stain DNA does not influence
minicirclereplicationusingsimilarconditionsasthesinglemolecule
experiments (Fig. S2).
Formation of long leading/lagging strand duplex DNA prod-
ucts, within a flow cell that continuously removes dissociated
proteins, provides a rigorous test that the helicase, leading/
lagging strand DNA polymerases and clamp loader stay associ-
ated at the moving replication fork. These studies therefore
strongly support the DNA looping ‘‘trombone’’ model of repli-
cation. Although this hypothesis is widely accepted, it has been
difficult to demonstrate unequivocally that the lagging strand
DNA polymerase truly remains bound to the replisome during
replication. In particular, the lagging strand polymerase must
cycle on and off DNA during synthesis of multiple Okazaki
fragments. Elegant electron microscopy studies demonstrate
DNA loops in the T4 and T7 systems, but similar studies have not
been performed in the E. coli system (15–18). The current study
demonstrates that the lagging strand polymerase remains con-
tinuously associated with the moving replisome in the E. coli
system, implying that DNA loops are formed during extension of
each Okazaki fragment.
To estimate the number of Okazaki fragments, and thus
number of DNA loops produced by the replisome during syn-
thesis of an 86 kb duplex, we determined the frequency of
Okazaki fragment synthesis in an ensemble assay under condi-
tions similar to the single-molecule assays (Fig. 1D). The syn-
thetic minicircle contains no dT residues on the inner circle,
which allows the lagging strand to be uniquely labeled using
?32P-dATP; the leading strand can be labeled using ?32P-dTTP.
The alkaline agarose gel analysis gives an average Okazaki
fragment length of 0.65 kb. Therefore, the lagging strand poly-
merase remains continuously attached to the replisome during
production of approximately 130 Okazaki fragments to form an
86 kb duplex. Because the polymerase remains attached to the
replisome, yet comes on and off DNA for each fragment, ?130
DNAloopsmustbeformedduringanaverageprocessivebinding
event. It should be noted that Okazaki fragment size is dictated
by fork speed and primase concentration, and is typically 1–2 kb
underconditionsofthisreportwhenmeasuredat37 °C(10).The
shorter length observed here likely results from the slower rate
of fork progression at the 23 °C temperature used in this report.
Pol III* Often Dissociates from the Replisome Before DnaB. The
replisome will be brought to a sudden halt upon dissociation of
Fig. 1.
DNAcurtainsatadiffusionbarrier.(A)Schemeofsingle-moleculerollingcircle
replication on a lipid bilayer. During leading strand synthesis (blue), DNA is
displaced from the 100mer circle as the replisome ‘rolls’ around the template.
The newly synthesized 5? ssDNA ‘‘tail’’ is converted to dsDNA by lagging-
strandsynthesis(red).(B)GrowthofarollingcircleDNAcurtaininreal-timeas
visualized by Yo-Pro-1 fluorescence. (C) Histogram of observed DNA lengths
and 1-Gaussian fitting analysis (green). (D) Alkaline gel analysis of ensemble
reaction leading and lagging strand products of minicircle replication.
Single-molecule rolling circle replication on a lipid bilayer produces
Yao et al.PNAS ?
August 11, 2009 ?
vol. 106 ?
no. 32 ?
13237
BIOCHEMISTRY

Page 3

either Pol III* or DnaB helicase. To determine the processivity
of DnaB helicase, we included Pol III* in the buffer flow, along
with ? and primase. Thus, if Pol III* dissociates from the
replisome, it should quickly be replaced by Pol III* in the buffer
flow, and fork propagation should continue. Comparison of
DNA curtains produced when Pol III* is present or absent from
the buffer flow shows distinctly longer DNA molecules when Pol
III* is present (compare Fig. 2 A and B). Analysis of 400
molecules, fit to a single Gaussian function, yields an average
length of 140.4 ? 6.8 kb (Fig. 2C). This length of DNA is 62%
longer than observed when Pol III* is omitted from the flow
(compare red and green histograms in Fig. 2C). A simple
explanation for this result is that Pol III* often dissociates from
the replisome before DnaB falls off DNA, and when another Pol
III* is available to associate with DnaB, it reforms the replisome
and the replication fork continues. This observation underscores
the importance of continuously removing dissociated proteins
from a reaction to make accurate measurements of replisome
processivity.
Individual Replisomes Exhibit Different Intrinsic Rates of Fork Move-
ment. Thereplisomemaymoveinacontinuousuniformmotion,or
it may progress at an uneven rate because of periodic pausing.
Real-time observations of individual replisomes show a uniform
rate of fork movement (Fig. 3). However, examination of single
moleculesrevealsaremarkablefinding.Althoughagivenreplisome
moves at a uniform rate, individual replisomes show surprisingly
different rates that vary as much as 5-fold or more, as illustrated by
thekymographsinFig.3A.Thefactthatindividualreplisomesmove
at different rates may explain why ensemble reactions of growing
leading strand products give a broad smear in an alkaline gel (e.g.,
Fig. 1D). One possible explanation for the different rates of
replisome movement is that a given replisome may develop a
unique Okazaki fragment cycle (e.g., Okazaki fragment size) that
is set stochastically by the first few fragments and may result in
different rates of fork movement. Other possible explanations are
discussed later in this report.
Analysis of 112 replisomes, fit to a single Gaussian function,
yields a rate of 246 ? 10 nt/s (Fig. 3B). The assays of this report
are performed at 23 °C. E. coli replisomes move ?2-fold faster
at 37 °C (6, 7, 9). We would like to note that the level of
resolution of the EM-CCD camera is approximately 760 bp/
pixel; therefore, brief pausing events that may occur during
replisome translocation would not be detected under our present
working conditions.
The Lagging Strand Enhances Replisome Processivity.Laggingstrand
specific operations, such as priming or DNA extension in the
opposite direction from fork progression, may exert influence
therateand/orprocessivityofthereplisome[reviewedinref.19].
Indeed, a single-molecule study of T7 replication demonstrates
that primase action causes the replication fork to pause (20).
Additionally, a single-molecule study of the E. coli replisome on
a ? DNA substrate showed that the presence of primase lowers
the processivity of the replisome by ?3-fold, from 10.5 kb to 3
kb (8). The decrease in processivity appeared to be because of
the presence of primase, and not actual priming activity. How-
ever, the previous study used Pol III cores that were not
preattached to the clamp loader, and the ? and ? subunits were
missing in most of the experiments. One or more of these
Fig. 2.
curtains produced when (A) Pol III* is present in the flow, (B) Pol III* is absent
from the flow during replication. (C) Histograms of product length when Pol
III* is present in the flow (green), or absent from the buffer flow (red).
DNA length is enhanced by excess Pol III* in the buffer flow. DNA
Fig. 3.
endpoint trajectories (DNA lengths) plotted versus time defines a curve with a slope that represents the rate of DNA synthesis. (B) Histogram of the rates of
individual replisomes.
Individual replisomes move at different rates. (A) Representative kymographs of DNA molecules from real-time observation of DNA synthesis. The
13238 ?
www.pnas.org?cgi?doi?10.1073?pnas.0906157106Yao et al.

Page 4

differences may explain the low overall processivity observed in
that study. Indeed, the ? subunit is known to enhance the
processivity of Pol III*-? (21). The ? subunit is also known, to
facilitate transfer of an RNA primer from primase to the clamp
loader (22). Furthermore, the ? subunit is required to stabilize
the oligomeric structure of the clamp loader upon which the
architecture of the replisome is built (23–25).
The experiments of Fig. 4 reexamine the influence of lagging
strand synthesis on processivity using replisomes that are recon-
stituted using a fully assembled Pol III*. To study replisomes in
the absence of lagging strand synthesis, we omitted primase from
the buffer flow. The leading strand product of rolling circle
replication is a long ssDNA composed of 100mer direct repeats
of the minicircle sequence. The Yo-Pro-1 intercalating dye does
not detect ssDNA, and therefore we converted the ssDNA to
dsDNA by including in the flow a mixture of 400 nM each of 5
DNA 20mers that anneal end-to-end over the 100mer sequence
(see scheme in Fig. 4A). This allows the observation of leading
strand replication in real-time (see Movie S2). Ensemble reac-
tions demonstrate that the oligonucleotides do not inhibit rep-
lication or slow fork movement (Fig. S3). Analysis of the final
length of 400 molecules yields an average processivity, when fit
to a single Gaussian function, of only 52.3 kb ? 3.8 kb (Fig. 4A).
The experiments of Fig. 1 demonstrate that the leading/lagging
strand replisome has a significantly greater processivity (86 kb)
than observed for the leading strand replisome (52 kb). Taken
together, these results indicate that lagging strand synthesis
enhances replisome processivity by ?61%.
It is possible that the leading strand replisome is less processive
because of the presence of the ssDNA 20mers. Therefore, we
analyzed the processivity of the leading strand replisome by a
multistep method that allowed addition of the DNA 20mers after
replication (see scheme in Fig. 4B). In the first step, rolling circle
replication was allowed to proceed for 30 min in the absence of
primase and DNA 20mers. To ensure that leading strand replica-
tion had occurred, we used fluorescently labeled SSB (SSB labeled
using either Texas Red or Oregon Green 488 retains full activity,
as illustrated in Fig. S4). After replication, SSB was removed from
the DNA by a buffer flow containing 4 M guanidine-HCl, and SSB
removalwasconfirmedbymonitoringlossofthefluorescentsignal.
Then we included the 5 DNA 20mers in the buffer flow (without
guanidine-HCl) to visualize the dsDNA product. Measurement of
400 molecules yields an average length of 52.5 ? 2.4 kb (Fig. 4B).
This length compares well with the reactions performed in the
presence of the ssDNA 20mers.
In summary, the experiments of Fig. 4 demonstrate that lagging
strand synthesis enhances processivity compared with a replisome
that extends only the leading strand. Primase is not an integral part
of the replisome, and thus does not seem a likely candidate as a
stabilizing factor for increased replisome processivity, although this
possibility cannot be excluded. However, a coupled leading/lagging
strand replisome has more contacts to the replication fork com-
paredwithaleadingstrand-onlyreplisome.Specifically,theleading
strand-only replisome has only 1 Pol III core-to-? connection,
whereasthecoupledleadingandlaggingstrandreplisomehas2Pol
III core-to-? connections, 1 on each strand. We propose that these
extra connections may underlie the higher processivity of the
coupled leading/lagging replisome.
Lagging Strand Synthesis Decreases the Rate of Replisome Progres-
sion. Real-time observations of the rate of 112 individual leading
strand replisomes gives a rate, fit to a single Gaussian function,
of 317 ? 4 nt/s (Fig. 5A). Interestingly this rate is approximately
23% faster than the leading/lagging strand replisome. It seems
possible that the DNA 20mers may require some time for
hybridization, and that the rate measurements could be an
underestimate. Therefore, we also examined the rate of fork
movement by ensemble assays. As explained earlier, a limitation
of ensemble studies is the resolving limit of agarose gels, and
therefore only the earliest times during a replication reaction can
be examined before the DNA becomes too long. In the ensemble
experiment of Fig. 5B, replication reactions were performed in
the presence or absence of primase, and aliquots were quenched
at 5 s, 10 s, 15 s, 20 s, 25 s, and 30 s. The result confirms that the
leading strand-only replisome (321 nt/s) moves at a faster rate
than the coupled leading/lagging strand replisome (278 nt/s).
In summary, the experiments of Fig. 5 demonstrate that
lagging strand synthesis decreases the rate of replication fork
progression. The effect of lagging strand synthesis on the rate of
the replisome could be because of primase, or to DNA poly-
merase acting on the lagging strand.
It is interesting to note that the distribution of rates for
individual leading strand replisomes appears more uniform than
the distribution observed for individual leading/lagging strand
replisomes. This result suggests that lagging strand events may
underlie the wide dispersion of rates observed for the leading/
lagging replisome (i.e. Fig. 3). However, there appears to be a
group of leading strand replisomes that move ?100 nt/s slower
than the average (Fig. 5A). Other possibilities that may explain
thepersistentdifferenceintherateofreplicationforkmovement
among individual replisomes include: (i) loss of a nonessential
but important subunit (i.e., ?, ?, ?, ?); (ii) the leading strand
polymerase may bind a ? subunit that occupies any of 3 different
positions in the clamp loader, and different positions may impart
Fig.4.
strand synthesis. (A) (Top) Scheme for observation of leading strand replica-
tion. Five DNA 20mers anneal to the 100mer repeat sequence of the long
leading strand ssDNA product, converting it to dsDNA for visualization using
Yo-Pro-1. (Bottom) Length distribution of replication products. (B) (Top)
Scheme of leading strand replication in presence of E. coli SSB, followed by 4
M Guanidine-HCl to remove SSB, and then addition of the 5 DNA 20mers to
convert the ssDNA to dsDNA. (Bottom) Length distribution of replication
products. The Gaussian fitting analysis is shown in green.
Processivityoftheleadingstrandreplisomeintheabsenceoflagging
Yao et al. PNAS ?
August 11, 2009 ?
vol. 106 ?
no. 32 ?
13239
BIOCHEMISTRY

Page 5

functional consequences; (iii) different replisomes may have
subunits that are partially unfolded, affecting their intrinsic rate;
(iv) some subunits may have undergone oxidative damage during
preparation, and this damage may affect their intrinsic rate.
Single molecule studies on a variety of enzymes have noted
significant differences in the intrinsic catalytic rate of otherwise
identical molecules (26–30). Although the physical basis for these
differencesisnotunderstood,someproposalssuggestthatenzymes
existindifferentstableconformersthathavedistinctcatalyticrates.
However, single molecule studies of enzymes that demonstrate
distinct rates of catalysis are typically performed over much shorter
time scales than those of the current report. Additional studies will
be needed to address this fascinating issue.
Conclusions
The current study outlines single-molecule methods to analyze
the processivity and rate of replisome action in real-time, both
for a coupled leading/lagging strand replisome and for a repli-
some that only operates on the leading strand. Growth of a
curtain of DNA molecules is performed in a flow of buffer that
sweeps dissociated Pol III* and DnaB from the reaction, pre-
venting their reassociation with DNA and providing firm mea-
surements of processivity. Previous ensemble and single-
moleculestudiesofreplisomeprocessivitycontainedanexcessof
unassociated replisome proteins, which may replace a protein
that dissociates from DNA and thereby mask a true measure of
processivity. Indeed, we find here that including excess Pol III*
in the buffer flow results in significantly longer DNA products.
We presume that when Pol III* dissociates from the replisome,
another Pol III* in the flow replaces it and continues fork
movement with DnaB. Interestingly, replisomes that perform
only leading strand synthesis move 23% faster than replisomes
that perform both leading and lagging strand synthesis, indicat-
ing that lagging strand synthesis slows the rate of fork progres-
sion. In contrast, replisome processivity is enhanced 61% by
lagging strand synthesis. It seems likely that the additional
connection to the replication fork, provided by a lagging strand
polymerase held to DNA by the sliding clamp, lends additional
stability to the replisome for enhanced processivity.
Experimental Procedures
Materials. Pol III subunits (?, ?, ?, ?, ?, ??, ?, ?, ?), primase, SSB, and DnaB helicase
werepurifiedasdescribedinref.13.PolIII*(PolIIIcore3?3?????)wasreconstituted
andpurifiedasdescribedinref.7.OligonucleotidesweresynthesizedbyIDTand
gel purified. Glucose oxidase and catalase were from Sigma. Yo-Pro-1 was from
Invitrogen(MolecularProbes).Photoclearsilicone-basedelastomer(Sylgard184)
was from Dow-Corning. Lipids were from Avanti Polar Lipids Inc.
Flow Cell Construction. Flow cells were prepared using photo-clear silicone-
based elastomer poured into a negative lithography mold; the bottom con-
taineda2.5mm?25mmstripof50?mthick3Mtape(i.e.,formingachannel).
After solidification, a rectangular block containing the channel was cut out;
entry and exit ports were punched into either end of the channel. A #1
coverslip was scratched with a diamond-tipped scribe to create diffusion
barriers.Theelastomerblockandthecoverslipweretreatedinaplasmaoven,
then welded together. A lipid bilayer was formed on the glass surface inside
the flow cell using a mixture of freshly sonicated liposomes (DOPC supple-
mented with 8% DOPE-mPEG550 and 0.5% DOPE-biotin) in buffer L1 (10 mM
Tris?HCl, pH 7.8, 100 mM NaCl) as described in ref. 14.
Total Internal Reflectance Fluorescence (TIRF) Microscopy. Microscopy was
performed using an Olympus IX70 inverted microscope fitted with a 60? TIRF
objective (N.A. ? 1.45), a Prior motorized stage, and a motorized shutter.
Yo-Pro-1 was excited using a solid state 488-nm laser (Coherent) at 1.5 mW.
Fluorescence emission was collected back through the objective and filtered
for residual laser light before being captured by a 512 ? 512 back thinned EM
CCD camera (Hamamatsu). The motorized shutter permitted a 100-ms expo-
sure every 1 s. Buffer was driven through the flow cell using a syringe pump
(KD Scientific). Image collection and data work-up were facilitated using the
Slidebook Software suite (Intelligent Imaging Inc.).
Leading and Lagging Strand Replication. A 100mer synthetic DNA minicircle
substrate containing a 40 dT 5? tail was prepared as described in ref. 13. DnaB
helicase (18.2 pmol, 365 nM) was assembled onto the minicircle DNA (655
fmol, 13.1 nM) in 50 ?L Buffer A (20 mM Tris?HCl, pH 7.5, 5 mM DTT, 40 ?g/mL
BSA, and 4% glycerol) containing 8 mM MgOAc2and 1.25 mM ATP, followed
by incubation for 30s at 37 °C. To this solution was added a 25 ?L reaction
containing Pol III* (675 fmol, 27 nM), ?2(1.85 pmol, 74 nM as dimer), 60 ?M
each dCTP and dGTP, and 8 mM Mg(OAc)2in Buffer A. After 6 min at 37 °C, 1
?L of the reaction was added to 1 mL Buffer B (8 mM MgOAc2, 60 ?M each of
dCTP and dGTP, and 50 nm Yo-Pro-1 in Buffer A). The reaction was passed
through the flow cell at 500 ?L/min for 30 s, then at 10 ?L/min for 30 s. DNA
replication was initiated upon flowing (100 ?L/min) Buffer A containing 60
?M of each dNTP, 250 ?M each of CTP, TTP, and UTP, 1 mM ATP, 462 nM SSB4,
300 nM primase, 50 nM ?2, 50 nM Yo-Pro-1, 0.8% glucose, 0.01% ?-mercap-
toethanol, 0.57 U glucose oxidase, and 2.1 U catalase. Control experiments
usingbiotinylated?DNAshowthataflowrateof100?L/minexertsaforceof
1.45pNandstretchesdsDNAto88%ofitsfullcontourlength(Fig.S1).AllDNA
lengths were corrected using this value. To detect dsDNA we used 50 nM
Yo-Pro-1, a single intercalating moiety that binds and dissociates from DNA
much more rapidly than Yo-Yo1.
Ensemble reactions were performed similarly except that after incubation
withDnaB(30s)andPolIII*(6min),replicationwasinitiateduponaddingSSB,
primase, remaining dNTPs and rNTPs, along with either32P-dATP or32P-dTTP
(primase was not present for leading strand only replication).
Leading Strand Replication Reactions. Reactions were performed as described
above except primase, SSB, ?2, CTP, GTP, and UTP were omitted from the flow
topreventlaggingstrandsynthesis.TheleadingssDNAproductwasvisualized
by including in the buffer flow 400 nM each of 5 DNA 20mers that hybridize
to the 100mer sequence. We also performed leading strand synthesis by
Fig. 5.
minicircle leading-only replication (lanes 1–6) and leading/lagging replication (lanes 7–12). Red dotted lines mark the front of the migration bands in
leading-only replication. (B) Histogram of the rates of individual replisomes. The Gaussian fitting analysis is shown in green.
The rate of leading strand replisomes. (A) Alkaline gel analysis of ensemble reaction products in which leading strand synthesis is monitored during
13240 ?
www.pnas.org?cgi?doi?10.1073?pnas.0906157106 Yao et al.