Dividing a supercoiled DNA molecule into
two independent topological domains
Fenfei Lenga,1, Bo Chena, and David D. Dunlapb
a
Department of Chemistry and Biochemistry, Florida International University, 11200 SW 8th Street, Miami, FL 33199; and bDepartment of Cell Biology,
Emory University, Atlanta, GA 30322
Both prokaryotic and eukaryotic chromosomes are organized
into many independent topological domains. These topological
domains may be formed through constraining each DNA end from
rotating by interacting with nuclear proteins; i.e., DNA-binding
proteins. However, so far, evidence to support this hypothesis is
still elusive. Here we developed two biochemical methods; i.e.,
DNA-nicking and DNA-gyrase methods to examine whether certain
sequence-specific DNA-binding proteins are capable of separating
a supercoiled DNA molecule into distinct topological domains. Our
approach is based on the successful construction of a series of
plasmid DNA templates that contain many tandem copies of one
or two DNA-binding sites in two different locations. With these
approaches and atomic force microscopy, we discovered that several sequence-specific DNA-binding proteins; i.e., lac repressor,
gal repressor, and λ O protein, are able to divide a supercoiled
DNA molecule into two independent topological domains. These
topological domains are stable under our experimental conditions.
Our results can be explained by a topological barrier model in
which nucleoprotein complexes confine DNA supercoils to localized
regions. We propose that DNA topological barriers are certain
nucleoprotein complexes that contain stable toroidal supercoils
assembled from DNA-looping or tightly wrapping DNA around
DNA-binding proteins. The DNA topological barrier model may be
a general mechanism for certain DNA-binding proteins, such as
histone or histone-like proteins, to modulate topology of chromosome DNA in vivo.
T
he Escherichia coli chromosome is comprised of a 4.6 Mb
circular, negatively supercoiled DNA molecule. A singlestranded nick or double-stranded break should release all superhelical tension and therefore relax the circular DNA molecule.
However, early studies showed that multiple single-stranded
nicks are required to fully relax the E. coli DNA molecule (1, 2).
These studies also suggested that the E. coli chromosome consists
of 40 to 100 independent topological domains in vivo (2). More
recently, Postow, et al. (3) reassessed the size of the topological
domains and demonstrated that the E. coli chromosome is segregated into 400 to 500 different topological domains; the sizes
of the topological domains are dynamic and variable ranging
from 2 to 66 kb. These studies coupled with genetic studies (4)
strongly support the existence of topological barriers that divide
the E. coli chromosome into different topological domains (5).
One question arises from these studies: What forms topological
barriers in DNA? Several models have been proposed to explain
the DNA topological barriers (5, 6). For instance, because transcription by an RNA polymerase generates positive and negative
supercoils (7), transcription of a gene, especially a gene producing a membrane insertion protein can induce the formation
of a topological domain barrier in vivo (8). Another interesting
model is that certain DNA-binding proteins especially DNAlooping proteins may constrain DNA loops to serve as topological
barriers (5, 9–11). To support this model, we previously showed
that certain nucleoprotein complexes, resulting from the binding
of several sequence-specific DNA-binding proteins to their recognition sites, could form topological barriers that impede the
diffusion and merger of independent chromosomal supercoil
www.pnas.org/cgi/doi/10.1073/pnas.1109854108
domains (12, 13). Nevertheless, conclusive evidence to support
these hypothetical models is still required.
Although several attempts have been made to decipher the
mechanism by which the E. coli chromosome is divided into
independent topological domains (1–3), the nature of the topological barriers is still a mystery. A primary difficulty in determining the identity of the topological barriers in DNA is the lack of a
simple, effective system to examine what property or properties
of DNA or protein-DNA complexes can serve as topological
barriers to divide a DNA molecule into different topological
domains. A simple barrier might divide a small supercoiled DNA
molecule into two independent topological domains. In this
framework, it would be feasible to test whether certain nucleoprotein complexes function as topological barriers and divide a
DNA molecule into distinct topological domains. In this report,
we present our efforts to establish a simple, in vitro system to
examine which nucleoprotein complexes are capable of serving
as topological barriers to confine free DNA supercoils within
a defined region. With this unique approach, we discovered that
certain sequence-specific DNA-binding proteins, such as lac
repressor, gal repressor, and λ O protein, are able to act as topological barriers that prevent supercoil diffusion.
Results
A Unique Strategy to Study DNA Topological Barriers In Vitro. In this
study, we developed a unique strategy to examine whether certain
sequence-specific DNA-binding proteins can block supercoil
diffusion along DNA. Our first step was to construct a series of
plasmids that contain one copy or several tandem copies of one or
two distinct DNA-binding sites in one or two different locations
(Fig. 1, Fig. S1, and Table S1). The DNA-binding sites at two
locations divide the plasmid into two regions of different sizes,
∼2.9 and ∼1.2 kb. We also placed nicking restriction endonuclease recognition sites for Nt.BbvC1 and Nb.BtsI into the plasmid, such that the Nt.BbvCI site resides in the 1.2 kb region and
the Nb.BtsI site in the 2.9 kb region (Fig. 1, Fig. S1). For this
report, we made DNA templates that contain multiple DNAbinding sites for lac repressor (LacI), gal repressor (GalR), or λ
O protein (Table S1). Our next step was to examine whether these
sequence-specific DNA-binding proteins can divide a supercoiled
DNA molecule into two independent topological domains. For
this purpose, we developed two methods: the DNA-nicking and
gyrase methods (Fig. 2). In the DNA-nicking method (Fig. 2A),
one DNA-binding protein; e.g., LacI, will bind to the two groups
of DNA-binding sites on the supercoiled DNA template. If the
DNA-binding protein stably blocks supercoil diffusion, a nick
generated by either Nt.BbvCI or Nb.BtsI should not fully release
the superhelical stress of the DNA molecule. After the single nick
Author contributions: F.L. and D.D.D. designed research; F.L., B.C., and D.D.D. performed
research; F.L. analyzed data; and F.L. wrote the paper.
The authors declare no conflict of interest.
This article is a PNAS Direct Submission.
1
To whom correspondence should be addressed. E-mail: [email protected].
This article contains supporting information online at www.pnas.org/lookup/suppl/
doi:10.1073/pnas.1109854108/-/DCSupplemental.
PNAS ∣ December 13, 2011 ∣ vol. 108 ∣ no. 50 ∣ 19973–19978
BIOCHEMISTRY
Edited by Sankar Adhya, National Institutes of Health, NCI, Bethesda, MD, and approved October 12, 2011 (received for review June 27, 2011)
Fig. 1. Plasmids containing tandem copies of one DNA-binding site; i.e., lac
O1 operators or λ O binding sites in one location or two different locations.
Plasmids pCB112, pCB115, pCB138, and pCB144 were constructed as detailed
under Materials and Methods. The restriction enzyme sites for Nt.BbvCI and
Nb.BtsI are shown. Each closed or open rectangle represents a lac O1 operator or λ DNA replication origin, respectively. Each λ DNA replication origin
contains four λ O binding sites.
is sealed by T4 DNA ligase, a partially supercoiled DNA molecule will be generated. In the DNA-gyrase assay (Fig. 2B), the
DNA template has a nick in either the 1.2 kb or 2.9 kb region.
Upon binding by a sequence-specific DNA-binding protein, the
plasmid DNA molecule is divided into two regions. If the DNAbinding protein prevents supercoil diffusion, DNA gyrase should
be able to supercoil the region without the nick. After ligation,
(−) supercoiled DNA templates should result. Agarose gel electrophoresis and atomic force microscopy (AFM) were used to
examine the topological status of our DNA molecules.
E. coli LacI Blocked Supercoil Diffusion and Divided a Supercoiled DNA
Molecule into Two Independent Topological Domains. E. coli LacI, a
homotetramer (14) was chosen as the first DNA-binding protein
for this study. As demonstrated previously, LacI simultaneously
binds to two of three potential chromosomal sites; i.e., lac O1, O2,
or O3 operators to form a DNA loop (15). A DNA template,
pCB115 (Fig. 1) that contains two pairs of lac O1 operators in
two different locations was used in our assays. The space between
the two lac O1 operators in each location is 25 bp (Table S1).
Because lac O1 site is a 21 bp DNA sequence and the headto-tail distance of the lac O1 operators is 46 bp, it is reasonable
to assume that each LacI tetramer cannot simultaneously bind to
the neighboring lac O1 operators of pCB115 which face in opposite directions. Instead, two LacI tetramers are able to simultaneously bind to the four lac O1 operators to form two highly
stable LacI-lac O1 nucleoprotein complexes. In this case, two
stable DNA loops are formed. Fig. 3 shows results of the DNAnicking and gyrase assays, which clearly demonstrate that LacI
successfully blocked supercoil diffusion and divided the DNA
molecule into two independent topological domains. In the absence of LacI, Nt.BbvCI, or Nb.BtsI fully relaxed the plasmid
pCB115 (Fig. 3A, lanes 1, 4, and 7). In the presence of LacI, however, neither Nt.BbvCI nor Nb.BtsI alone could remove all supercoils from the (−) supercoiled DNA molecule (Fig. 3A, compare
lane 2 to lane 1 and lane 5 to lane 4). Nt.BbvCI removed ∼7 (−)
supercoils that equal to supercoils constrained in the 1.2 kb
region where Nt.BbvCI recognition site is located [Fig. 1; the
supercoiling density of pCB115 was determined to be ∼ − 0.06;
i.e., it has ∼26 (−) supercoils. Fig. S2A was also used to calculate
the constrained supercoils]. Nb.BtsI removed ∼17 (−) supercoils
19974 ∣
www.pnas.org/cgi/doi/10.1073/pnas.1109854108
Fig. 2. The experimental strategy to examine whether a site-specific DNAbinding protein blocks DNA supercoil diffusion. (A) The DNA-nicking method.
A (−) supercoiled DNA molecule (a) contains a nicking endonuclease Nt.BbvCI
recognition site and several DNA-binding sites (green dots) of a site-specific
DNA-binding protein in two different locations. The site-specific DNA-binding protein (red cylinder) binds to the DNA-binding sites to divide the DNA
molecule into two DNA loops (b). After digestion by Nt.BbvCI, a DNA nick is
formed. If the DNA-binding protein blocks DNA supercoil diffusion, two independent topological domains are formed (c). A large excess of an oligonucleotide containing an Nt.BbvCI recognition site is then added into the
reaction mixtures to inhibit Nt.BbvCI activities. After ligation by T4 DNA ligase (d) and phenol extraction, a partially (−) supercoiled DNA molecule is
produced (e). (B) The DNA-gyrase method. A nicked DNA molecule (a) containing several DNA-binding sites (green dots) of a site-specific DNA-binding
protein in two different locations is used to bind to the DNA-binding protein
(red cylinder). In this case, two DNA-loops are formed (b). E.coli DNA gyrase
should be able to introduce supercoils to the DNA loop without the DNA nick
(c). If the DNA-binding protein blocks supercoil diffusion, after the inhibition
of DNA gyrase by novobiocin, the DNA supercoils should stay with the loop
without the DNA nick (d). After ligation by T4 DNA ligase and phenol extraction, a (−) supercoiled DNA template should be generated (e).
that correspond to those constrained in the 2.9 kb region in which
it nicks the plasmid. As expected, Nt.BbvCI and Nb.BtsI together
removed all (−) supercoils from the DNA template (Fig. 3A,
lane 8). Interestingly and also as expected, the constraining of
DNA supercoils in defined regions by LacI is sensitive to the presence of its inducer, isopropyl-β-D-thiogalactoside (IPTG), which
lowers the affinity of LacI for its operators (Fig. 3A, lanes 3 and
6). Similar results were obtained when the DNA-gyrase method
was used (Fig. 3B, Fig. S2B). In this assay, we first digested the
DNA template using Nt. BbvCI to yield a nicked plasmid for
the following reactions. In the absence of LacI and DNA gyrase,
the DNA template was fully relaxed (Fig. 3B, lane 1). DNA gyrase
significantly supercoiled the DNA template (Fig. 3B, lane 2) and
novobiocin was able to completely inhibit the gyrase activities
(Fig. 3B, lane 3). Regardless, DNA gyrase was able to supercoil
the DNA region confined by LacI (Fig. 3B, lane 6), which is also
sensitive to the presence of IPTG (Fig. 3B, lane 7). These results
demonstrated that LacI, upon binding to lac O1 operators,
formed a topological barrier to block supercoil diffusion and
divide the DNA molecule into two independent topological
domains. Control experiments indicated that the division of plasmid DNA templates into two topological domains requires the
presence of lac O1 operators in two different locations. LacI cannot divide the DNA molecule into two topologically independent
Leng et al.
In this article, we also demonstrated that the LacI-mediated
DNA looping is required for the formation of topological barriers. For this purpose, we constructed a similar plasmid DNA
template, pCB152 that contains four lac O1 operators equally
distributed between two different locations. Because the space
between the two lac O1 operators in each location is 20 bp
(Table S1) and the head-to-tail distance of the lac O1 operators
is 41 bp, each LacI tetramer is able to simultaneously bind to the
neighboring lac O1 operators of pCB152 due to these DNA-binding sites locating on the same side. In this case, LacI is not capable
of mediating the formation of DNA loops. This arrangement
of lac O1 operators should not support the formation of the
topological barriers on pCB152. Indeed, our results showed that
LacI did not block supercoil diffusion and divide the plasmid into
two stable topological domains (Fig. S3 C and D), suggesting that
DNA looping is required for the formation of LacI-mediated
topological barriers. Furthermore, our results show that the protein-DNA-looping complex resulting from one LacI tetramer
binding to two lac O1 operators is sufficient to form a topological
barrier to block supercoil diffusion, although the topological barrier is much less stable comparing with those containing multiple
LacI-lac O1 nucleoprotein complexes.
domains when plasmid pCB112 with four adjacent lac O1 sites
was used as the DNA template (Fig. 1, Fig. S3 A and B).
Next, we performed a time course of the DNA-nicking assay
of pCB115 to examine whether the LacI-mediated topological
barrier is kinetically stable. In this assay, significantly more
amount of Nt.BbvCI was used, which was able to nick pCB115
within 2 min (Fig. S4A). We then incubated the reaction mixture
at 37 °C for various time before adding T4 DNA ligase to seal the
DNA nick. Our results are shown in Fig. 3 C and D. At 5 min,
about 91% of topoisomers were (−) supercoiled and after
120 min incubation, about 47% of topoisomers were still (−)
supercoiled, indicating that the LacI-mediated DNA topological
barrier was quite stable. These kinetic data were also fitted to a
first-order rate equation (Fig. 3D), producing a first-order rate
constant (kA ) of 0.0062 min−1 and a half-life (t1∕2 ) of 112 min.
Leng et al.
DNA-Wrapping Proteins, such as λ O Protein and E. coli GalR also Divided a Supercoiled DNA Molecule into Two Independent Topological
Domains. Due to availability and the biological importance of
λ O protein and GalR, we decided to examine whether these
DNA-wrapping proteins are also able to divide a supercoiled
DNA molecule into two independent topological domains. As
demonstrated previously (16), λ O protein specifically binds to
the four repeating sequences (iterons) of λ DNA replication origin and forms a unique nucleoprotein complex, the “O-some” to
initiate DNA replication. GalR is a dimer and specifically binds
to the OE and OI operators of E. coli galactose operon to form
a loop in the presence of HU protein (17), which inhibits transcription from two gal promoters P1 and P2 (18). Both proteins
are capable of inducing DNA wrapping upon binding to their
recognition sites (19). In order to test whether λ O protein is able
to divide a supercoiled DNA molecule into two distinct topological domains, we made a DNA template, pCB138 containing
eight λ DNA replication origins (a total of 32 λ O-binding sites)
equally distributed between two locations (Fig. 1). In this scenario, the nucleoprotein complexes, which built from the DNA
replication origins wrapping around λ O proteins, should divide
the plasmid into two independent topological domains. Indeed,
our results shown in Fig. 4 A and B, Fig. S4C demonstrated that λ
O protein was able to divide the plasmid into two distinct, stable
topological domains. In the absence of λ O protein, a nick introduced by Nt.BbvCI or Nb.BtsI fully relaxed the DNA template
(Fig. 4A, lanes 1, 3, and 5). In the presence of λ O protein,
Nt.BbvCI or Nb.BtsI alone could not completely remove all (−)
supercoils from the DNA template (Fig. 4A, lanes 2 and 4). These
results suggested that λ O protein upon binding to its recognition
sites served as DNA topological barriers to block supercoil diffusion. Consistent with our previous results (19), the binding of λ O
protein to the mutiple λ O-binding sites on pCB138 caused DNA
wrapping and introduced about eight (−) supercoils into the
DNA template (Fig. 4A, lane 6). Nevertheless, our control experiments showed that two barriers are required to divide the
DNA template into two independent topological domains (Fig. 1,
Fig. S4B). Similar results were also obtained by using the DNAgyrase assay (Fig. 4B, lane 6).
We also cloned a few plasmids to test whether GalR is capable
of dividing a supercoiled DNA molecule into two independent
topological domains. Among them are pCB132 and pCB155
that carry 36 gal OE operators equally distributed between two
locations (Fig. S1, Table S1). As described under Materials and
Methods, the difference between these two plasmids is the space
PNAS ∣ December 13, 2011 ∣
vol. 108 ∣
no. 50 ∣
19975
BIOCHEMISTRY
Fig. 3. LacI divided a supercoiled DNA molecule, plasmid pCB115 into two
independent topological domains. (A) The DNA-nicking assays were performed as described under Materials and Methods and Fig. 2. In addition
to 0.156 nM of pCB115, as indicated at the top of the image, the reaction
mixtures also contained LacI (2.5 nM), IPTG, Nt.BbvCI (4 units), and Nb.BtsI
(4 units). The DNA molecules (topoisomers) were isolated and subjected to
agarose gel electrophoresis in the absence of chloroquine as detailed under
Materials and Methods. (B) The DNA-gyrase assays were performed as described under Materials and Methods and Fig. 2. In addition to 0.156 nM
of Nt.BbvCI-nicked pCB115, as specified at the top of the image, the reaction
mixtures also contained LacI (2.5 nM), IPTG, E. coli DNA gyrase (5 units), and
novobiocin (3 μM). The DNA molecules (topoisomers) were isolated and subjected to agarose gel electrophoresis in the absence of chloroquine. Lanes 1
of (A) and (B) contain DNA relaxed at 37 °C [it is slightly (þ) supercoiled because the gels were run at 24 °C]. (C) Time course of DNA supercoiling diffusion in the presence of LacI. The DNA-nicking assays were performed as
described under Materials and Methods. Each reaction mixture (320 μL) contained 0.156 nM of pCB115, 2.5 nM of LacI, and 12 units of Nt.BbvCI. The
reaction mixtures were incubated at 37 °C for the time indicated. Then, a
large excess of a double-stranded oligonucleotide containing an Nt.BbvCI recognition site were added to the reaction mixtures to inhibit the restriction
enzyme activities. The nicked DNA templates were ligated by T4 DNA ligase
in the presence of 1 mM of ATP at 37 °C for 30 min and the reactions were
terminated by extraction with an equal volume of phenol. The DNA molecules were isolated and subjected to agarose gel electrophoresis. (D) Quantification analysis of the time course. The percentage of supercoiled DNA was
plotted against the reaction time. The curve was generated by fitting the
data to a first-order rate equation to yield a first-order rate constant of
0.0062 min−1 and a half-life of 112 min.
Fig. 4. DNA-wrapping proteins λ O protein and GalR divided supercoiled
DNA molecules, pCB138, and pCB155 into two independent topological domains, respectively. (A) The DNA-nicking assays were performed as described
under Materials and Methods and Fig. 2. In addition to 0.156 nM of plasmid
pCB138, as indicated at the top of the image, the reaction mixtures also contained λ O protein (20 nM), Nt.BbvCI (4 units), and Nb.BtsI (4 units). After the
assay, the DNA molecules (topoisomers) were isolated and subjected to agarose gel electrophoresis in the absence of chloroquine as described under Materials and Methods. (B) The DNA-gyrase assays were performed as described
under Materials and Methods and Fig. 2. In addition to 0.156 nM of Nt.BbvCInicked pCB138, as specified at the top of the image, the reaction mixtures
also contained λ O protein (20 nM), E. coli DNA gyrase (5 units), and novobiocin (3 μM). After the assay, the DNA molecules (topoisomers) were isolated
and subjected to agarose gel electrophoresis. (C) The DNA-nicking assays
were performed as described under Materials and Methods and Fig. 2. In addition to 0.156 nM of plasmid pCB155, as indicated at the top of the image,
the reaction mixtures also contained GalR (22.5 nM), Nt.BbvCI (4 units),
Nb.BtsI (4 units), and galactose. After the assay, the DNA molecules (topoisomers) were isolated and subjected to agarose gel electrophoresis. (D) The
DNA-gyrase assays were performed as described under Materials and Methods and Fig. 2. In addition to 0.156 nM of Nt.BbvCI-nicked pCB155, as specified at the top of the image, the reaction mixtures also contained GalR
(22.5 nM), E. coli DNA gyrase (5 units), novobiocin (3 μM), and galactose. After
the assay, the DNA molecules (topoisomers) were isolated and subjected to
agarose gel electrophoresis. Lanes 1 of (A), (B), (C), and (D) contain DNA relaxed at 37 °C [it is slightly (þ) supercoiled because the gels were run at 24 °C].
separating the neighboring OE operators: pCB155, 20 bp and
pCB132, 25 bp (the head-to-tail distance of the OE operators are
36 and 41 bp, respectively). Because the OE operator is a 16 bp
DNA sequence, it is reasonable to assume that GalR binds to the
neighboring gal OE operators of pCB155 on the opposite side; in
contrast, GalR binds to the neighboring gal OE operators of
pCB132 on the same side. In this case, we were able to examine
whether the operator phasing affected the ability of GalR to induce the formation of DNA topological barriers. Our results are
summarized in Fig. 4 C and D, Fig. S5. These results clearly
demonstrated that GalR is capable of dividing a supercoiled
DNA molecule into two distinct topological domains and the
operator phasing does not affect GalR as a topological barrier.
Similar to λ O protein, the binding of GalR protein to the OE
sites on pCB155 and pCB132 also caused DNA wrapping and
introduced a few (−) supercoils into the DNA template, which
is consistent with our previous results (19). Intriguingly, galactose
did not completely abolish GalR as a topological barrier (Fig. 4 C
and D, Fig. S5) although it showed some inhibitory effects on the
formation of the DNA topological barriers in the DNA-nicking
19976 ∣
www.pnas.org/cgi/doi/10.1073/pnas.1109854108
assay. Control experiments indicated that two barriers of GalROE complexes are required to divide the plasmid DNA template
into two independent topological domains (Fig. S6).
In this article, we made three additional plasmids, pCB160,
pCB162, and pCB163 that carry multiple tandem copies of two
types of DNA-binding sites in two different locations (Table S1).
pCB160 contains 18 gal OE operators in one location and 16 lac
O1 operators in another location; pCB162 carries 16 λ O-binding
sites in one location and another 16 lac O1 sites in a different
location; and pCB163 has 16 λ O-binding sites in one location
and 18 gal OE operators in another location. These plasmids were
used to test whether two topological barriers deriving from
unrelated DNA-binding proteins were able to divide a DNA
molecule into two distinct topological domains. Our results summarized in Fig. S6 unambiguously demonstrated that two topological barriers resulting from different DNA-binding proteins
are able to divide a supercoiled DNA molecule into two independent topological domains. In the absence or presence of only one
DNA-binding protein; i.e., either λ O protein or GalR or LacI, a
nick introduced by Nt.BbvCI fully relaxed the DNA templates
(Fig. S6, lanes 1–3). However, in the presence of a combination
of two DNA-binding proteins; i.e., λ O protein and GalR for
pCB163, λ O protein and LacI for pCB162, and GalR and LacI
for pCB160, Nt.BbvCI could not completely remove all (−)
supercoils from these DNA templates (Fig S6, lanes 4). These
results suggest that the topological barriers derived from the
two unrelated DNA-binding proteins are able to confine free
supercoils in a defined region and separate the supercoiled
DNA molecules into two independent topological domains. As
expected, galactose or IPTG had some inhibitory effects on
the formation of the topological barriers (Fig. S6, lanes 5 and 6).
Atomic Force Microscope Images Revealed that LacI Divided a Supercoiled DNA Molecule into Two Distinct Topological Domains. AFM is a
powerful technique that has provided high-resolution images for
a variety of protein-DNA complexes (20–25). Here we used AFM
to visualize whether LacI divides a supercoiled DNA molecule
into distinct topological domains. Two plasmids, pCB115 and
pCB 109 were used. As described under Materials and Methods
and also in Table S1, pCB115 and pCB109, respectively, contain
4 and 32 lac O1 operators equally distributed between two locations (Fig. 1). In addition, the neighboring lac O1 operators were
cloned on the opposite directions such that LacI cannot simultaneously binds to the neighboring lac O1 sites. Instead, LacI tetramer binds to the lac O1 sites of the two different locations and
divides the plasmids into two loops. We used the DNA-nicking
method for our AFM imaging studies (Fig. 2). After supercoiled
pCB115 and pCB109 were digested by Nt.BbvCI in the presence
of LacI (step c of Fig. 2A), the LacI-plasmid complexes were
deposited on freshly cleaved mica surface and visualized using an
AFM microscope. Our results are summarized in Fig. 5, Fig. S7.
These results clearly demonstrated that LacI divided a supercoiled DNA molecule into two distinct topological domains. For
plasmid pCB115, in the absence of LacI, the average contour
length of the DNA molecules was measured to be 1;437.6
42.4 nm (Table S2). For B-form DNA with 0.34 nm per base pair,
this length was calculated to be 4;228 125 bp, which is almost
equivalent to the plasmid sequence length, 4,350 bp. Fig. 5 also
shows that two LacI molecules bound to the specific DNA-binding sites and separated the plasmid into one relaxed and one
supercoiled domain. Interestingly, the contour lengths of the
relaxed and supercoiled domains were measured to be 423.9
18.1 nm (1;247 53 bp) and 983.1 44.5 nm (2;891 131 bp),
respectively. These lengths are consistent with the DNA sequence
lengths of the two topological domains (Table S2). For pCB109,
LacI also divided the plasmid into one relaxed and one supercoiled domain (Fig. S7). The measured contour lengths of the
relaxed and supercoiled domains are 413.5 30.4 nm (1;216
Leng et al.
89 bp) and 986.5 66.7 nm (2;901 196 bp), respectively
(Table S2). These lengths are also consistent with the DNA sequence lengths of the two topological domains calculated from
the DNA map (Table S2). Intriguingly, our AFM images show
that multiple LacI tetramers (up to 16 molecules) bound to the
DNA molecule in a zigzag manner and formed a long filament
between two DNA domains (Fig. S7). It is likely that this long
filament represents LacI binding to the 32 lac O1 operators of
the two locations on the plasmid. The length of the LacI-DNA
filament was measured to be 196.7 22.0 nm, significantly
shorter than the length of the 16 lac O1 operators cloned in one
location of pCB109. These results indicate that LacI binding to
the lac O1 operators caused the wrapping of lac O1 operators
around the LacI molecules. These results are consistent with
our results of gel electrophoresis and also with our previous interpretation of LacI-induced ΔLk (model D of Fig. S8 of ref. 19).
Discussion
In this article, we demonstrated that the binding of a DNA-binding protein to its recognition sites in two different locations on
a supercoiled DNA molecule can confine free supercoils to a
defined region and divide the DNA molecule into two distinct
topological domains: a relaxed and a supercoiled domain. We
also showed that two types of DNA-binding proteins, DNA-looping and -wrapping proteins have this functionality. For DNAlooping proteins, we used LacI, the best-characterized DNAlooping protein as a model protein for our studies. Our results
show that LacI upon binding to appropriate-spaced lac O1 operators formed highly stable nucleoprotein complexes and separated
the plasmid DNA molecules into two loops (Fig. 5, Fig. S7).
These two DNA loops are topologically independent (Fig. 5,
Fig. S7). For DNA-wrapping proteins, we tested two proteins, λ
O protein and GalR. Both are dimers and induce the wrapping
of their DNA-binding sequences around themselves (19). Our
results clearly showed that these two DNA-wrapping proteins
restricted free supercoils to a defined region and divided the
supercoiled DNA molecules into two independent topological
domains (Fig. 4). We believe that DNA wrapping is the main
cause for these two proteins to confine supercoils within defined
regions. First, there is no strong evidence to suggest that these
Leng et al.
Fig. 6. The DNA topological barrier models. (I) A DNA-looping protein, such
as LacI, upon binding to its recognition sites in two different locations, forms
a DNA topological barrier to block DNA supercoil diffusion and therefore divide the circular DNA molecule into two independent topological domains.
(II) A DNA-wrapping protein, such as λ O protein and GalR, wraps DNA
around itself in two different locations. In this case, two topological barriers
are formed to divide the circular DNA molecule into two independent topological domains. Blue circle and red cylinder represent, respectively, the DNA
recognition sequence of a site-specific DNA-binding protein and the sitespecific DNA-binding protein.
PNAS ∣ December 13, 2011 ∣
vol. 108 ∣
no. 50 ∣
19977
BIOCHEMISTRY
Fig. 5. AFM images to demonstrate that LacI divided a supercoiled DNA molecule, plasmid pCB115 into two independent topological domains: a relaxed
and a supercoiled domain. The AFM imaging experiments were performed as
described under Materials and Methods. Arrows indicate two LacI tetramers
binding to four lac O1 operators in two different locations and dividing the
DNA molecule into two loops and also into two topological domains. (Scale
bar represents 100 nm).
two proteins are able to form distinct, stable DNA loops under
our experimental conditions. Second, our results show that a
combination of two unrelated DNA-binding proteins; e.g., λ O
protein and GalR also confined free supercoils to a defined
region and separated the DNA molecules into two distinct topological domains (Fig. S6). These results suggest that DNA wrapping, rather than looping, is the main reason for these two DNAwrapping proteins to divide the DNA molecules into different
topological domains. Nevertheless, although our AFM images
showed that λ O protein and GalR are able to divide the supercoiled plasmids into distinct topological domains (Fig. S8), we
cannot fully exclude the role of protein-protein interactions of
λ O-DNA complexes and GalR-DNA complexes in the formation
of the two distinct topological domains.
We favor models depicted in Fig. 6 to explain our results. Model (I) is for DNA-looping proteins that are able to bring two
or two groups of the DNA-binding sites together to fold into a
topologically constrained nucleoprotein complex. This nucleoprotein complex serves as a DNA topological barrier or divider
to block supercoil diffusion. This model represents the most likely
way for LacI to divide a supercoiled DNA molecule into two
independent topological domains. Our AFM images strongly
support this interpretation (Fig. 5, Fig. S7). Model (II) is for
DNA-wrapping proteins, such as λ O protein and GalR. Specific
DNA sequences wrap around these DNA-wrapping proteins to
form a unique nucleoprotein structure, such as the O-some (26).
These nucleoprotein structures form a topological barrier that
slow or prevent diffusion of supercoils past the nucleoprotein
complex. In this scenario, it requires two such nucleoprotein complexes to divide a circular DNA molecule into two topological
domains. This model also provides a reasonable explanation for
a transcribing RNA polymerase to serve as a topological barrier
(8, 27). First, RNA polymerases cause DNA wrapping (28). In
addition, a transcribing RNA polymerase generates a (þ) supercoil domain in front of the RNA polymerase and a (−) supercoil
domain behind it (7). These topological structures should be able
to block supercoil diffusion along DNA.
The discoveries presented here have great biological ramifications. Previously we showed that certain sequence-specific DNAbinding proteins strongly stimulate transcription-coupled DNA
supercoiling (12, 13). We used the “twin-supercoiled-domain”
model to explain these results where nucleoprotein complexes,
especially those containing stable toroidal supercoils assembled
from DNA looping or tightly wrapping DNA around these DNAbinding proteins, can form topological barriers that impede the
diffusion and merger of independent chromosomal supercoil
domains (12). Our results in this report demonstrated that these
nucleoprotein complexes are indeed able to form topological
barriers to block supercoil diffusion (Figs. 3, 4, 5, Figs. S5 and S7).
In addition, our results can be used to explain transcription
activation of bacterial phage λ, a hallmark of λ DNA replication
control in vivo (29). Our recent results showed that transcriptioncoupled DNA supercoiling is responsible for the activation of λ
DNA replication. Specifically, the O-some (26) assembled from
wrapping DNA around O protein in the replication origin blocks,
confines, and captures transcription-coupled DNA supercoiling,
which causes structural changes in λ DNA replication origin
(30). In this case, the DNA replication origin is unwound and
DNA replication is initiated. This mechanism can also be used to
explain transcriptional dependence of E. coli DNA replication
initiation (31). In this scenario, DnaA, the E. coli replication
initiator coupled with HU protein, binds to specific DnaA boxes
in the oriC replication origin to form a nucleoprotein complex to
unwind the AT-rich DNA sequence in the origin. Transcriptioncoupled DNA supercoiling should greatly stimulate this process.
Our topological barrier model also provides unique insights
into how the E. coli chromosome is divided into many different
topological domains. As demonstrated previously (32, 33), the
E. coli chromosomal DNA is associated with several abundant
histone-like proteins, such as HU, H-NS, FIS, and IHF and
folded into a compact nucleoid structure (1, 34). These histonelike DNA-binding proteins also wrap and loop DNA, which constrain DNA supercoils in vitro and in vivo (35–38). It is possible
1. Worcel A, Burgi E (1972) On the structure of the folded chromosome of Escherichia
coli. J Mol Biol 71:127–147.
2. Sinden RR, Pettijohn DE (1981) Chromosomes in living Escherichia coli cells are segregated into domains of supercoiling. Proc Natl Acad Sci USA 78:224–228.
3. Postow L, Hardy CD, Arsuaga J, Cozzarelli NR (2004) Topological domain structure of
the Escherichia coli chromosome. Genes Dev 18:1766–1779.
4. Valens M, et al. (2004) Macrodomain organization of the Escherichia coli chromosome.
EMBO J 23:4330–4341.
5. Travers A, Muskhelishvili G (2005) Bacterial chromatin. Curr Opin Genet Dev
15:507–514.
6. Thanbichler M, Viollier PH, Shapiro L (2005) The structure and function of the bacterial
chromosome. Curr Opin Genet Dev 15:153–162.
7. Liu LF, Wang JC (1987) Supercoiling of the DNA template during transcription. Proc
Natl Acad Sci USA 84:7024–7027.
8. Deng S, Stein RA, Higgins NP (2004) Transcription-induced barriers to supercoil
diffusion in the Salmonella typhimurium chromosome. Proc Natl Acad Sci USA
101:3398–3403.
9. Deng S, Stein RA, Higgins NP (2005) Organization of supercoil domains and their
reorganization by transcription. Mol Microbiol 57:1511–1521.
10. Wu HY, Liu LF (1991) DNA looping alters local DNA conformation during transcription.
J Mol Biol 219:615–622.
11. Saldanha R, Flanagan P, Fennewald M (1987) Recombination by resolvase is inhibited
by lac repressor simultaneously binding operators between res sites. J Mol Biol
196:505–516.
12. Leng F, McMacken R (2002) Potent stimulation of transcription-coupled DNA supercoiling by sequence-specific DNA-binding proteins. Proc Natl Acad Sci USA 99:9139–9144.
13. Leng F, Amado L, McMacken R (2004) Coupling DNA supercoiling to transcription in
defined protein systems. J Biol Chem 279:47564–47571.
14. Matthews KS, Nichols JC (1998) Lactose repressor protein: functional properties and
structure. Prog Nucleic Acid Res Mol Biol 58:127–164.
15. Oehler S, Eismann ER, Kramer H, Muller-Hill B (1990) The three operators of the lac
operon cooperate in repression. EMBO J 9:973–979.
16. Dodson M, et al. (1986) Specialized nucleoprotein structures at the origin of replication of bacteriophage lambda: localized unwinding of duplex DNA by a six- protein
reaction. Proc Natl Acad Sci USA 83:7638–7642.
17. Lia G, et al. (2003) Supercoiling and denaturation in Gal repressor/heat unstable nucleoid protein (HU)-mediated DNA looping. Proc Natl Acad Sci USA 100:11373–11377.
18. Lewis DE, Adhya S (2002) In vitro repression of the gal promoters by GalR and HU
depends on the proper helical phasing of the two operators. J Biol Chem
277:2498–2504.
19. Chen B, et al. (2010) DNA linking number change induced by sequence-specific
DNA-binding proteins. Nucleic Acids Res 38:3643–3654.
20. Finzi L, Dunlap DD (2010) Single-molecule approaches to probe the structure, kinetics,
and thermodynamics of nucleoprotein complexes that regulate transcription. J Biol
Chem 285:18973–18978.
19978 ∣
www.pnas.org/cgi/doi/10.1073/pnas.1109854108
that the nucleoprotein complexes generated from these histonelike proteins serve as general DNA topological barriers to modulate localized DNA supercoiling. Indeed, our recent results
showed that HU and H-NS are able to confine supercoils to a
defined region. Nevertheless, the topological barriers stemming
from nonspecific binding of HU or H-NS to DNA were less
stable, which is consistent with the dynamic nature of E. coli chromosome topological domains (3, 39). Furthermore, nucleosomes
may use the same mechanism to modulate DNA topology of
eukaryotic chromosome (40).
Materials and Methods
Details of the preparation of purified proteins and the construction of
plasmid DNA templates are described in SI Materials and Methods. The
DNA-nicking and DNA-gyrase methods are summarized in Fig. 2. Details of
the procedures are also described in SI Materials and Methods. For AFM, the
LacI-DNA samples were prepared according to the DNA-nicking method.
After the supercoiled DNA templates were digested by Nt.BbvCI, the LacIDNA complexes were deposited on freshly cleaved mica and visualized with
a NanoScope MultiMode AFM microscope. Details of the AFM procedure are
described in SI Materials and Methods.
ACKNOWLEDGMENTS. We thank Drs. Roger McMacken and Sankar Adhya
for providing us with λ O protein and E. coli GalR, respectively. We thank
Kathleen S. Matthews for providing us with an E. coli strain overexpressing
E. coli LacI. We also thank Drs. James C. Wang, Roger McMacken, W. David
Wilson, Yanbin Zhang, and Geraldine Fulcrand for critically reading the
article before submission and for helpful discussion. We thank Dr. Wilma
K. Olson for suggestions and encouragement. This work was supported by
National Institutes of Health Grant 5SC1HD063059-02 (to F.L.) and Human
Frontier Science Program Grant RGP0051 (to D.D.D.).
21. Virnik K, et al. (2003) “Antiparallel” DNA loop in gal repressosome visualized by
atomic force microscopy. J Mol Biol 334:53–63.
22. Gaczynska M, et al. (2004) Atomic force microscopic analysis of the binding of the
Schizosaccharomyces pombe origin recognition complex and the spOrc4 protein with
origin DNA. Proc Natl Acad Sci USA 101:17952–17957.
23. Heddle JG, Mitelheiser S, Maxwell A, Thomson NH (2004) Nucleotide binding to DNA
gyrase causes loss of DNA wrap. J Mol Biol 337:597–610.
24. Lyubchenko YL (2011) Preparation of DNA and nucleoprotein samples for AFM
imaging. Micron 42:196–206.
25. Wang H, Finzi L, Lewis DE, Dunlap D (2009) AFM studies of lambda repressor oligomers
securing DNA loops. Curr Pharm Biotechno 10:494–501.
26. Dodson M, Roberts J, McMacken R, Echols H (1985) Specialized nucleoprotein structures at the origin of replication of bacteriophage lambda: complexes with lambda O
protein and with lambda O, lambda P, and Escherichia coli DnaB proteins. Proc Natl
Acad Sci USA 82:4678–4682.
27. Scheirer KE, Higgins NP (2001) Transcription induces a supercoil domain barrier in
bacteriophage Mu. Biochimie 83:155–159.
28. Gamper HB, Hearst JE (1982) A topological model for transcription based on unwinding angle analysis of E. coli RNA polymerase binary, initiation and ternary complexes.
Cell 29:81–90.
29. Hase T, Nakai M, Masamune Y (1989) Transcription of a region downstream from
lambda ori is required for replication of plasmids derived from coliphage lambda.
Mol Gen Genet 216:120–125.
30. Sei-Hoon Chung (1996) Transcriptional activation of bacteriophage lambda DNA
replication. Ph.D. Dissertation (Johns Hopkins University, Baltimore, MD), pp 1–185.
31. Baker TA, Kornberg A (1988) Transcriptional activation of initiation of replication
from the E. coli chromosomal origin: an RNA-DNA hybrid near oriC. Cell 55:113–123.
32. Pettijohn DE (1988) Histone-like proteins and bacterial chromosome structure. J Biol
Chem 263:12793–12796.
33. Schmid MB (1990) More than just “histone-like” proteins. Cell 63:451–453.
34. Stonington OG, Pettijohn DE (1971) The folded genome of Escherichia coli isolated in
a protein-DNA-RNA complex. Proc Natl Acad Sci USA 68:6–9.
35. Rouviere-Yaniv J, Yaniv M, Germond JE (1979) E. coli DNA binding protein HU forms
nucleosomelike structure with circular double-stranded DNA. Cell 17:265–274.
36. Swinger KK, Lemberg KM, Zhang Y, Rice PA (2003) Flexible DNA bending in HU-DNA
cocrystal structures. EMBO J 22:3749–3760.
37. Tupper AE, et al. (1994) The chromatin-associated protein H-NS alters DNA topology in
vitro. EMBO J 13:258–268.
38. Mojica FJ, Higgins CF (1997) In vivo supercoiling of plasmid and chromosomal DNA in
an Escherichia coli hns mutant. J Bacteriol 179:3528–3533.
39. Higgins NP, Yang X, Fu Q, Roth JR (1996) Surveying a supercoil domain by using the
gamma delta resolution system in Salmonella typhimurium. J Bacteriol 178:2825–2835.
40. Luger K, et al. (1997) Crystal structure of the nucleosome core particle at 2.8 A resolution. Nature 389:251–260.
Leng et al.
Supporting Information
Leng et al. 10.1073/pnas.1109854108
SI Materials and Methods
Purified Proteins. Bacteriophage λ DNA replication O protein
was a generous gift of R. McMacken (Johns Hopkins University).
Escherichia coli gal repressor (GalR) was kindly provided by
S. Adhya (National Institutes of Health). E. coli LacI was purified
by the method of Chen and Matthews [(1); E. coli strains containing the plasmid overexpressing LacI was kindly provided
by K. S. Matthews at Rice University]. Restriction enzymes
Nt.BbvCI, Nb.BtsI, T4 DNA ligase, and E. coli DNA gyrase were
purchased from New England Biolabs (Beverly, MA, USA).
Plasmid DNA Templates. All plasmids are derived from the low copy
number plasmid pACYC184. Construction of plasmid DNA templates sometimes required DNA fusions between noncomplementary cohesive termini. In this scenario, cohesive ends were
converted before ligation to blunt ends by incubation of the DNA
fragments with T4 DNA polymerase in the presence of dNTPs.
Plasmid pYZX43F was constructed by the insertion of a 24-bp
synthetic DNA fragment, containing a BglII site, into the unique
BspHI site of pACYC184. A 46 bp synthetic DNA fragment
containing one lac O1 operator was then cloned into the BamHIBglII sites of pYZX43F to create pCB67. The same 46 bp synthetic DNA fragment was inserted into the unique BglII site of
pCB67 to yield pCB69 that contains two head-to-tail tandem
copies of lac O1 operators. Plasmid pCB71 was constructed by
the insertion of the 92 bp BamHI-BglII fragment of pCB69 into
the unique BglII site of pCB69. In this case, pCB71 contains four
tandem copies of lac O1 operators. Plasmid pCB73 that contains
eight tandem copies of lac O1 operators was made by the cloning
of the 184 bp BamHI-BglII fragment of pCB71 into the unique
BglII site of pCB71. Plasmid pCB74 carrying 16 tandem copies
of lac O1 operators was created by the insertion of the 368 bp
BamHI-BglII fragment of pCB73 into the unique BglII site of
pCB73. Plasmid pCB107 was constructed by the insertion of a
19 bp synthetic DNA fragment containing a nicking enzyme
Nt.BbvCI recognition site into the unique EagI site of pCB74.
A 746 bp BamHI-KpnI fragment of pCB107 carrying 16 tandem
copies of lac O1 operators was then cloned into the unique AhdI
site of pCB107 to yield plasmid pCB109. In this case, pCB109
contains 32 lac O1 operators equally distributed between two
locations. Plasmid pCB112 was created by the insertion of the
19 bp synthetic DNA fragment as described above into the unique
EagI site of pCB71. Because pACYC184 also contains an Nb.BtsI
recognition site, pCB112 contains four lac O1 operators, one
Nt.BbvCI site and one Nb.BtsI site. Plasmid pCB111 was constructed by the insertion of the 19 bp synthetic DNA fragment
as described above into the unique EagI site of pCB69. Next,
a 102 bp BamHI-KpnI fragment of pCB111 containing two
head-to-tail tandem copies of lac O1 operators was inserted into
the unique AhdI site of pCB111 to generate pCB115 that contains four copies of lac O1 operators equally distributed between
two different locations (Fig. 1). The space between the neighboring lac O1 operators is 25 bp for all plasmids as described above.
Using a similar approach, plasmid pCB152 that carries two pairs
of head-to-tail tandem copies of lac O1 operators equally distributed between two locations was constructed. However, the space
between the neighboring lac O1 operators is 20 bp for pCB152.
Plasmids pCB137 and pCB138 were derived from pCB2 and
pCB5, respectively (2). A 334 bp BamHI-SacI fragment of pCB2,
which contains a λ DNA replication origin (four λ O-binding
sites), was inserted into the unique AdhI site of pCB2 to generate
plasmid pCB135. Then, the 19 bp synthetic DNA oligomer conLeng et al. www.pnas.org/cgi/doi/10.1073/pnas.1109854108
taining an Nt.BbvCI recognition site as described above was
cloned into the unique EagI site of pCB136 to yield plasmid
pCB137. Plasmid pCB138 was also constructed in two steps. First,
a 1,306 bp BamHI-SacI fragment of pCB5, which contains four
tandem copies of λ DNA replication origins (16 λ O-binding
sites), was inserted into the unique AdhI site of pCB5 to produce
pCB136. Then, the 19 bp synthetic DNA oligomer containing an
Nt.BbvCI recognition site as described above was cloned into the
unique EagI site of pCB136 to yield pCB138 (Fig. 1). Plasmid
pCB144 was constructed by the insertion of the 19 bp synthetic
DNA oligomer containing an Nt.BbvCI recognition site into the
unique EagI site of pCB5.
Plasmids pCB132 and pCB155 both carrying 36 head-to-tail
tandem copies of gal OE operators equally distributed between
two locations were derived from plasmids pCB46 and pCB42,
respectively (2). First, the 19 bp synthetic DNA fragment containing a nicking enzyme Nt.BbvCI recognition site as described
above was inserted into the unique EagI site of pCB46 and
pCB42 to yield plasmids pCB131 and pCB154, respectively.
Then, the 748 bp BamHI-KpnI fragment containing 18 headto-tail tandem copies of gal OE operators of pCB131 was cloned
into the unique AhdI site of pCB131 to produce pCB132. Similarly, the 658 bp BamHI-KpnI fragment containing 18 headto-tail tandem copies of gal OE operators of pCB154 was inserted
into the unique AhdI site of pCB154 to generate pCB155. The
difference between pCB132 and pCB155 is the space between
each neighboring gal OE operators, which is 25 and 20 bp, respectively (the head-to-tail lengths for each repeating sequence are 41
and 36 bp, respectively).
Plasmids pCB160, pCB162, and pCB163 contain multiple tandem copies of two types of DNA-binding sites in two different
locations. In each location, they carry either lac O1 operators,
gal OE operators, or λ O-binding sites (iteron III of λ replication
origin). To clone plasmid pCB160, the 846 bp EcoRV-KpnI fragment containing 18 head-to-tail tandem copies of gal OE operators of pCB42 was inserted into the EcoRV-KpnI sites of pCB109.
In this case, pCB160 carries 16 lac O1 operators in one location
and 18 gal OE operators in a different location. Plasmids pCB162
and pCB163 were derived from pCB37 (2). Plasmid pCB129 was
constructed by the insertion of the 19 bp synthetic DNA fragment
containing an Nt.BbvCI recognition site as described above into
the unique EagI site of pCB37. Plasmid pCB130 was created
by cloning the 714 bp SacI-BamHI fragment of pCB129 into the
unique AhdI site of pCB129. In this case, pCB130 carries 32 tandem copies of λ O-binding sites equally distributed between two
locations. Plasmid pCB162 was made by cloning the 1,013 bp
EcoRV-SphI fragment of pCB109 containing 16 tandem copies
of lac O1 operators into the EcoRV-SphI sites of pCB130. As a
result, pCB162 contains 16 λ O-binding sites in one location and
16 lac O1 operators in another location. Plasmid pCB163 was
constructed by inserting the 925 bp EcoRV-SphI fragment of
pCB42 carrying 18 gal OE operators into the EcoRV-SphI sites
of pCB130. In this scenario, pCB163 carries 16 λ O-binding sites
in one location and 18 gal OE operators in a different location.
The DNA-Nicking Method. A typical DNA-nicking reaction mixture
(320 μL) contained 20 mM Tris-acetate (pH 7.9 at 25 °C), 10 mM
magnesium acetate, 1 mM DTT, a negatively supercoiled DNA
template, and a site-specific DNA-binding protein, such as LacI
or λ O protein or GalR. Where specified, 1 mM of IPTG or
galactose was also added to the DNA-nicking assays. All components were assembled on ice and incubated for 30 min at 37 °C.
1 of 8
After the incubation, the supercoiled DNA templates were digested by either Nt.BbvCI or Nb.BtsI or both nicking enzymes
at 37 °C for 30 min. Then, a large excess of a double-stranded
oligonucleotide containing either Nt.BbvCI or Nb.BtsI recognition site were added to the reaction mixtures to inhibit the restriction enzyme activities. The nicked DNA templates were ligated
by T4 DNA ligase in the presence of 1 mM of ATP at 37 °C for
30 min and the reactions were terminated by extraction with an
equal volume of phenol. The DNA samples were precipitated
with ethanol and dissolved in 25 μL of 10 mM Tris-HCl buffer
(pH 8.5). The linking number of the ligated DNA products was
determined with 1% agarose gel electrophoresis in the absence
or presence of 0.5 μg∕mL of chloroquine and calculated from
the gel images stained with SYBR Gold nucleic acids stain using
KODAK 1D Image Analysis Software.
The DNA-Gyrase Method. In this method, the supercoiled plasmid
DNA templates were nicked using one of the nicking restriction
enzymes, Nt.BbvCI or Nb.BtsI. A standard DNA-gyrase assay
mixture (360 μL) contained 35 mM Tris-HCl, pH 7.5, 24 mM KCl,
4 mM MgCl2 , 2 mM DTT, 1 mM spermidine, 1.75 mM ATP, 5%
glycerol, and a nicked circular plasmid DNA. Where indicated, a
site-specific DNA-binding protein, such as LacI or λ O or GalR,
IPTG or galactose (1 mM), E. coli DNA gyrase, and novobiocin
(3 μM) were added to the reaction mixtures. All components
were assembled on ice and incubated 30 min at 37 °C. E. coli
DNA gyrase (5 units) was then added into the reaction mixtures
1. Chen J, Matthews KS (1992) Deletion of lactose repressor carboxyl-terminal domain
affects tetramer formation. J Biol Chem 267:13843–13850.
2. Chen B, et al. (2010) DNA linking number change induced by sequence-specific
DNA-binding proteins. Nucleic Acids Res 38:3643–3654.
for 30 min at 37 °C. After novobiocin (3 μM) was added to the
mixtures to inhibit the gyrase activities, the nicked DNA templates were ligated by T4 DNA ligase at 37 °C for 15 min. The
ligation reactions were terminated by adding 1% SDS and
50 μg∕mL of proteinase K for 30 min at 37 °C. Each DNA sample
was extracted once with phenol, precipitated with ethanol, and
dissolved in 25 μL of 10 mM Tris-HCl, pH 8.5. The linking number of the ligated DNA products was determined with 1% agarose gel electrophoresis in the absence or presence of 2.5 μg∕mL
of chloroquine and calculated from the gel images stained with
SYBR Gold nucleic acids stain using KODAK 1D Image Analysis
Software.
Atomic Force Microscopy. The LacI-DNA samples were prepared
according to the DNA-nicking method as described above. After
the supercoiled DNA templates were digested by Nt.BbvCI at
37 °C for 30 min, the LacI-DNA complexes (10 μL) were deposited on the poly-L-ornithine-coated mica and incubated for 2 min
at room temp. The droplet was rinsed away with 0.4 mL HPLC
water and dried gently with compressed air. Images were acquired with a NanoScope MultiMode AFM microscope (Digital
Instrument, Santa Barbara, CA) operated in tapping mode using
a 50–60 mV oscillation amplitude of uncoated, etched silicon
tips with a resonance frequency of 75 kHz [NSC18, MirkoMasch,
San Jose, CA (3)]. Areas of 1 × 1 μm2 were scanned at a rate of
1.2 Hz and a resolution of 512 × 512 pixels. The DNA contour
lengths were estimated by using Image Analysis Software ImageJ.
3. Wang H, Finzi L, Lewis DE, Dunlap D (2009) AFM studies of lambda repressor oligomers
securing DNA loops. Curr Pharm Biotechnol 10:494–501.
Fig. S1. Plasmids containing tandem copies of one DNA-binding site; i.e., lac O1 operators or gal OE operators in two different locations. Plasmids pCB109,
pCB132, and pCB155, derivatives of pACYC184, were constructed as described under Materials and Methods in the main text. The restriction enzyme sites for
Nt.BbvCI and Nb.BtsI are shown. Each closed rectangle represents a lac O1 operator for plasmid pCB109 or gal OE operator for plasmids pCB132 and pCB155.
Leng et al. www.pnas.org/cgi/doi/10.1073/pnas.1109854108
2 of 8
Fig. S2. LacI divided supercoiled plasmid pCB115 into two independent topological domains. (A) The DNA-nicking assays were performed as described in
Fig. 3A. In addition to 0.156 nM of plasmid pCB115 (a total of 0.625 nM of lac O1 sites), as indicated at the top of the image, the reaction mixtures also
contained LacI (2.5 nM), IPTG (1 mM), Nt.BbvCI (4 units), and Nb.BtsI (4 units). The plasmid DNA molecules were isolated and subjected to agarose gel electrophoresis in the presence of 0.5 μg∕mL of chloroquine as detailed under Materials and Methods in the main text. (B) The DNA-gyrase assays were performed
as described in Fig. 3B. In addition to 0.156 nM of Nt.BbvCI-nicked plasmid pCB115, as specified at the top of the image, the reaction mixtures also contained
LacI (2.5 nM), IPTG (1 mM), E. coli DNA-gyrase (5 units), and novobiocin (3 μM). The plasmid DNA molecules were isolated and subjected to agarose gel electrophoresis in the presence of 2.5 μg∕m of chloroquine. (C) and (D) The DNA-gyrase assays using Nb.BtsI-nicked plasmid pCB115 were performed as described
under Materials and Methods in the main text. In addition to 0.156 nM of Nb.BtsI-nicked plasmid pCB115, as specified at the top of the image, the reaction
mixtures also contained LacI (2.5 nM), IPTG (1 mM), E. coli DNA-gyrase (5 units), and novobiocin (3 μM). The plasmid DNA molecules were isolated and subjected
to agarose gel electrophoresis in the absence (C) or presence of 2.5 μg∕mL chloroquine (D).
Leng et al. www.pnas.org/cgi/doi/10.1073/pnas.1109854108
3 of 8
Fig. S3. LacI cannot divide plasmids pCB112 and pCB152 into two independent topological domains. (A) The DNA-nicking assays were performed as described
under Materials and Methods in the main text. In addition to 0.156 nM of plasmid pCB112 (a total of 0.625 nM lac O1 operators), as indicated at the top of the
image, the reaction mixtures also contained LacI (2.5 nM), IPTG (3 mM), Nt.BbvCI (4 units), and Nb.BtsI (4 units). The plasmid DNA molecules were isolated and
subjected to agarose gel electrophoresis in the absence of chloroquine as detailed under Materials and Methods in the main text. (B) The DNA-gyrase assays
were performed as described under Materials and Methods in the main text. In addition to 0.156 nM of Nt.BbvCI-nicked plasmid pCB112, as specified at the top
of the image, the reaction mixtures also contained LacI (2.5 nM), IPTG (3 mM), E. coli DNA gyrase (5 units), and novobiocin (3 μM). The plasmid DNA molecules
were isolated and subjected to agarose gel electrophoresis in the absence of chloroquine as detailed under Materials and Methods in the main text. (C) The
DNA-nicking assays were performed as described under Materials and Methods in the main text. In addition to 0.156 nM of plasmid pCB152 (a total of 0.625 nM
lac O1 operators), as indicated at the top of the image, the reaction mixtures also contained LacI (2.5 nM), IPTG (3 mM), Nt.BbvCI (4 units), and Nb.BtsI (4 units).
The plasmid DNA molecules were isolated and subjected to agarose gel electrophoresis in the absence of chloroquine as detailed under Materials and Methods
in the main text. (D) The DNA-gyrase assays were performed as described under Materials and Methods in the main text. In addition to 0.156 nM of Nt.BbvCInicked plasmid pCB152, as specified at the top of the image, the reaction mixtures also contained LacI (2.5 nM), IPTG (3 mM), E. coli DNA-gyrase (5 units), and
novobiocin (3 μM). The plasmid DNA molecules were isolated and subjected to agarose gel electrophoresis in the absence of chloroquine as detailed under
Materials and Methods in the main text.
Leng et al. www.pnas.org/cgi/doi/10.1073/pnas.1109854108
4 of 8
Fig. S4. (A) The nicking endonuclease Nt.BbvCI was able to completely digest plasmid pCB115 in two minutes. A time course of enzyme digestion of pCB115
using 12 units of Nt.BbvCI in 320 μL of 1 × NEB buffer 4. The restriction enzyme digestion assays were performed as described above. Lanes 1 to 4 contained DNA
samples isolated from the reaction mixtures digested by Nt.BbvCI for 1, 2, 5, and 10 min, respectively. Lane 5 contained the supercoiled pCB115. (B) λ O protein
cannot divide plasmid pCB144 into two independent topological domains. The DNA-nicking assays were performed as described under Materials and Methods
in the main text. In addition to 0.156 nM of plasmid pCB144 (a total of 2.5 nM of O-binding sites), as indicated at the top of the image, the reaction mixtures also
contained λ O protein (10 nM), Nt.BbvCI (7.5 units), and Nb.BtsI (7.5 units). The plasmid DNA molecules were isolated and subjected to agarose gel electrophoresis in the absence of chloroquine as detailed under Materials and Methods in the main text. (C) Time course of DNA supercoiling diffusion in the presence
of λ O protein for plasmid pCB138. The DNA-nicking assays were performed as described under Materials and Methods in the main text. Each reaction mixture
(320 μL) contained 0.156 nM of plasmid pCB138, 20 nM of λ O protein, and 12 units of Nt.BbvCI. The reaction mixtures were incubated at 37 °C for the time
indicated. Then, a large excess of a double-stranded oligonucleotide containing an Nt.BbvCI recognition site were added to the reaction mixtures to inhibit the
restriction enzyme activities. The nicked DNA templates were ligated by T4 DNA ligase in the presence of 1 mM of ATP at 37 °C for 30 min and the reactions
were terminated by extraction with an equal volume of phenol. The plasmid DNA molecules were isolated and subjected to agarose gel electrophoresis in the
absence of chloroquine. The rate of supercoil diffusion was estimated to be 0.0091 min−1 (the half-life is about 76 min). (D) Time course of DNA supercoiling
diffusion in the presence of λ O protein for plasmid pCB137. The DNA-nicking assays were performed as described under Materials and Methods in the main
text. Each reaction mixture (320 μL) contained 0.156 nM of plasmid pCB137, 6.25 nM of λ O protein, and 12 units of Nt.BbvCI. The reaction mixtures were
incubated at 37 °C for the time indicated. Then, a large excess of a double-stranded oligonucleotide containing an Nt.BbvCI recognition site were added to the
reaction mixtures to inhibit the restriction enzyme activities. The nicked DNA templates were ligated by T4 DNA ligase in the presence of 1 mM of ATP at 37 °C
for 30 min and the reactions were terminated by extraction with an equal volume of phenol. The plasmid DNA molecules were isolated and subjected to
agarose gel electrophoresis in the absence of chloroquine.
Fig. S5. DNA-wrapping protein gal repressor divided supercoiled plasmid pCB132 into two independent topological domains. (A) The DNA-nicking assays
were performed as described under Materials and Methods in the main text. In addition to 0.156 nM of plasmid pCB132 (a total of 5.62 nM gal OE operators), as
indicated at the top of the image, the reaction mixtures also contained GalR (22.5 nM), Nt.BbvCI (4 units), Nb.BtsI (4 units), and galactose (1 mM). After the
assay, the plasmid DNA molecules were isolated and subjected to agarose gel electrophoresis in the absence of chloroquine as described under Materials and
Methods in the main text. (B) The DNA-gyrase assays were performed as described under Materials and Methods in the main text. In addition to 0.156 nM of
Nt.BbvCI-nicked plasmid pCB132, as specified at the top of the image, the reaction mixtures also contained GalR (22.5 nM), E. coli DNA-gyrase (5 units), novobiocin (3 μM), and galactose (1 mM). After the assay, the plasmid DNA molecules were isolated and subjected to agarose gel electrophoresis in the absence of
chloroquine as described under Materials and Methods in the main text.
Leng et al. www.pnas.org/cgi/doi/10.1073/pnas.1109854108
5 of 8
Fig. S6. The division of a supercoiled DNA molecule into two distinct topological domains by two unrelated DNA-binding proteins. The DNA-nicking assays in
the presence of two unrelated DNA-binding proteins; i.e., λ O protein and GalR, λ O protein and LacI, or GalR and LacI, were performed as described under
Materials and Methods in the main text. After the DNA-nicking assay, the plasmid DNA molecules were isolated and subjected to agarose gel electrophoresis in
the absence of chloroquine as detailed under Materials and Methods in the main text. (A) In addition to 0.156 nM of supercoiled plasmid pCB163 (a total
of 2.81 nM of gal OE operators and 2.5 nM of λ O-binding sites), the reaction mixtures also contained, as indicated at the top of the image, GalR (22.5 nM),
λ O protein (20 nM), and galactose (1 mM). (B) In addition to 0.156 nM of supercoiled plasmid pCB162 (a total of 2.5 nM of lac O1 operators and 2.5 nM of
λ O-binding sites), the reaction mixtures also contained, as indicated at the top of the image, LacI (10 nM), λ O protein (20 nM), and IPTG (2 mM). (C) In addition
to 0.156 nM of supercoiled plasmid pCB160 (a total of 2.81 nM of gal OE operators and 2.5 nM of lac O1 operators), the reaction mixtures also contained, as
indicated at the top of the image, GalR (22.5 nM), LacI (10 nM), galactose (1 mM), and IPTG (2 mM).
Fig. S7. AFM images to demonstrate that LacI divided a supercoiled DNA molecule, plasmid pCB109 into two independent topological domains: a relaxed and
a supercoiled domain. The AFM imaging experiments were performed as described under Materials and Methods in the main text. LacI tetramers bound to 32
tandem copies of lac O1 operators that equally distributed between two locations and formed a zigzag filament. This filamentary nucleoprotein complex
divided the plasmid DNA molecules into a relaxed and a supercoiled domain. (Scale bar represents 200 nm).
Leng et al. www.pnas.org/cgi/doi/10.1073/pnas.1109854108
6 of 8
Fig. S8. (A) AFM images of λ O protein binding to plasmid pCB137. The AFM imaging experiments were performed as described under Materials and Methods
in the main text. (Scale bar represents 200 nm). (B) AFM images to demonstrate that GalR divided a supercoiled DNA molecule, plasmid pCB132 into distinct
topological domains. The AFM imaging experiments were performed as described under Materials and Methods in the main text. (Scale bar represents
200 nm).
Table S1. Plasmid containing one copy or several tandem copies of one or two DNA-binding sites in two locations
Plasmid
DNA-binding site
Space between each DNA-binding site
Copy number of the DNA-binding site in each location
pCB115
pCB152
pCB109
pCB137*
pCB138*
pCB132
pCB155
pCB160
pCB162
pCB163
lacO1
lacO1
lacO1
O-binding site
O-binding site
gal OE
gal OE
gal OE & lacO1
O-binding site & lacO1
O-binding site & gal OE
25 bp
20 bp
25 bp
4 bp
4 bp
25 bp
20 bp
20 or 25 bp
25 bp
25 or 20 bp
2
2
16
4
16
18
18
18 or 16
16
16 or 18
*Plasmids pCB137 and pCB138, respectively, carry 2 and 8 λ DNA replication origins in which each λ DNA origin contains four DNA-binding sites for λ O
protein. In this case, pCB137 and pCB138 carries 8 and 32 λ O-binding sites, respectively; i.e., in each location these two plasmids contain 4 and 16 λ
O-binding sites, respectively.
Leng et al. www.pnas.org/cgi/doi/10.1073/pnas.1109854108
7 of 8
Table S2. DNA contour lengths of pCB115 and pCB109
Plasmid
DNA domain
Measured DNA contour length
nm
bp
DNA sequence length
bp
pCB115
full length
Rx domain†
Sc domain‡
1,437.6 ± 42.4
423.9 ± 18.1
983.1 ± 44.5
4,228 ± 125*
1,247 ± 53*
2,891 ± 131*
4,350
1,280
2,936
full length
Rx domain†
Sc domain‡
LacI-lac O1 complex
1,865.1 ± 99.6
413.5 ± 30.4
986.5 ± 66.7
196.7 ± 22.0
5,486 ± 293*
1,216 ± 89*
2,901 ± 196*
N/A
5,638
1,280
2,936
N/A
pCB109
*The measured DNA contour lengths in bp were calculated with the assumption of the standard
B-form DNA for the plasmids; i.e., using a rise of 0.34 nm per base pair.
†
Rx domain represents the relaxed DNA domain.
‡
Sc domain represents the supercoiled domain.
Leng et al. www.pnas.org/cgi/doi/10.1073/pnas.1109854108
8 of 8