DNA Bubble Formation in Transcription Initiation
†
Vladimir Tchernaenko,
‡
Herbert R. Halvorson,
‡
Mikhail Kashlev,
§
and Leonard C. Lutter*
,‡
Molecular Biology Section, Bone and Joint Center, Henry Ford Hospital, Detroit, Michigan 48202, and Molecular
Mechanisms of Transcription Section, NCI Center for Cancer Research, Frederick Cancer Research and
DeVelopment Center, Frederick, Maryland 27102
ReceiVed June 29, 2007; ReVised Manuscript ReceiVed NoVember 15, 2007
ABSTRACT: The properties of the DNA bubble in the transcription open complex have been characterized
by topological analysis of DNA circles containing the lac UV5 promoter or the PR promoter from
bacteriophage lambda. Topological analysis is particularly well suited to this purpose since it quantifies
the changes in DNA duplex geometry caused by bubble formation as well as by superhelical DNA wrapping.
The duplex unwinding that results from bubble formation is detected as a reduction in topological linking
number of the DNA circle, and the precision of this measurement has been enhanced in the current study
through the use of 8 or 10 promoter copies per circle. Several lines of evidence indicate that the linking
number change induced by open complex formation is essentially all due to bubble generation, with very
little derived from superhelical wrapping. Accordingly, the linking number change of -1.17 measured
for the lac UV5 promoter indicates that the size of the lac UV5 bubble is about 12.3 base pairs, while the
change of -0.98 measured for the lambda PR promoter indicates that the lambda PR bubble is 10.3 base
pairs. It was also found that the presence or absence of magnesium ion had little effect on the value of
the linking number change, a result that resolves the uncertainty associated with use of chemical probes
to study the effect of magnesium on bubble size. Finally, the magnitude of linking number change increases
progressively when the 3′ end of a transcript is extended to +2 and +3 in an abortive initiation complex.
This indicates that the transcription bubble expands at its leading edge in the abortive complex, results
that confirm and extend the proposal of a DNA “scrunching” mechanism at the onset of transcription.
These results are relevant to several models for the structure of DNA in the functional open complex in
solution, and provide an important complement to the structural information available from recent crystal
structures.
Transcription initiation is a process that occurs via multiple
steps as illustrated by the minimal scheme
The process begins with RNA polymerase (R) binding to a
promoter (P) to form a “closed” complex (RPc, also termed
I
1
), after which it converts to an intermediate complex (RPi,
also termed I
2
) and finally to an open complex (RPo)
[reviewed in (1, 2)]. The open complex is the active complex
for initiation of transcription, and its formation is ac-
companied by the denaturation of a discrete section of
promoter to form a transcription “bubble”. The various
intermediates in the process have been characterized in
solution by kinetic analysis as well as by structural ap-
proaches such as chemical and enzymatic footprinting. The
most detailed structural information for several of these
complexes has come from the impressive progress made in
the characterization of the crystal structures of RNA poly-
merase alone and in the elongation complex containing
template and short nascent RNA [reviewed recently (3)],
although a crystal structure of the open promoter complex
has yet to be described. However, while the crystal structures
of elongation complexes reveal unprecedented detail, the
extent to which they represent the structures of active
complexes in solution remains unknown, and they as yet do
not provide a full description of the complex DNA (4, 5).
For example, the cocrystals to date contain a length of DNA
that is considerably less than that which footprinting studies
indicate is associated with the polymerase in the open
complex in solution. Furthermore, none of the crystal
structures contain a complete transcription bubble. Finally,
none of the crystallography data provides information of the
dynamics of the bubble structure exhibited in solution.
This issue has been addressed by numerous complemen-
tary methods that have been used to characterize the solution
structures of complete transcription complexes. One such
solution method that has been widely used to analyze the
intact transcription bubble is chemical probe analysis. Thus
direct evidence for the opening of 10-14 bp of promoter
DNA during open complex formation comes from the use
of chemical reagents that react more readily with single
stranded DNA than with duplex DNA (6-11). These studies
find that open complex formation causes enhanced reactivity
to agents such as dimethyl sulfate and KMnO
4
in the region
between -12 and +2 at several promoters. In addition, the
†
This work was supported in part by grants from the National
Institutes of Health (GM49988 and GM56216).
* Corresponding author. E-mail: llutter1@hfhs.org. Tel: 313-916-
8681. Fax: 313-916-8064.
‡
Henry Ford Hospital.
§
Frederick Cancer Research and Development Center.
R + P rfRPc rfRPi rfRPo
1871Biochemistry 2008, 47, 1871-1884
10.1021/bi701289g CCC: $40.75 2008 American Chemical Society
Published on Web 01/19/2008

region of reactivity in the absence of magnesium was
observed to be about 8-10 bp and not encompass the start
site, while in the presence of magnesium the reactivity
expanded to at least 15 bp and encompassed the start site
(10-12). It was concluded that there are two forms of open
complex, with the one in the presence of magnesium
competent for transcription initiation.
However, while the chemical probe approach has proven
very useful in determining the location and estimating the
size of the bubble, there are a number of uncertainties
associated with the interpretation of the results: (1) Although
the chemical probes used do react preferentially with
denatured DNA, other distortions of the duplex can also
result in reactivity. Thus the method is not strictly specific
for a transcription bubble (11, 13, 14). (2) The reacted bases
represent only a small fraction [e.g., <5% (15)] of the
complexes, raising an uncertainty about whether this small
fraction is representative of the total population of complexes
[this common problem is discussed in (16)]. (3) The reagents
used can react with residues on the polymerase, leading to
its inactivation (13, 17). This raises a question about whether
the results represent the active complex or an inactive, and
perhaps distorted, complex (17, 18). (4) Portions of poly-
merase can protect nucleotides that are part of the bubble
from reaction with the probe (19-22), leading to incorrect
conclusions about the size, existence, or location of the
bubble. (5) Incomplete sampling of the sequence due to the
base specificity of many reagents makes a precise determi-
nation of bubble size difficult. (6) Different reagents can give
different results on the same promoter, leading to opposite
conclusions about the presence of a bubble (23). (7) If the
bubble is dynamic, i.e., “breathes”, then reactivity will tend
to represent the most open extent of the breathing. This leads
to an overestimate of the average bubble size. (8) Factors
such as Mg
2+
can increase reactivity of negatively charged
chemical probes such as permanganate ion by reducing ionic
repulsion of the negative phosphate groups of DNA (2, 10,
24-26). This can lead to the incorrect conclusion that Mg
2+
causes an increase in the bubble size. (9) If the nucleotide
bases in the single stranded DNA are involved in stacking
interaction, it may affect their reactivity to the chemical
probes.
These limitations of the use of chemical probe analysis
can be addressed through the use of other, complementary
methods for structural characterization of the open complex
bubble in solution. DNA topological analysis is one such
method that has been used to provide an independent
assessment of bubble size. It is a “noninvasive” method: it
characterizes the structure of an active transcription complex
in solution without perturbation or inactivation of the
complex. It measures the polymerase-induced change in the
topological linking number, which is the number of times
one strand of the double helix crosses over the other in
covalently closed circular DNA [reviewed in (27-30)]. For
example, if open complex formation induces a bubble of 10.5
bases, this will cause an unwinding of one turn of DNA
duplex, i.e., a loss of one strand crossing. If the complex is
formed on a promoter in circular DNA which is then treated
with topoisomerase, the linking number of the complex DNA
will be reduced by one relative to that of similarly treated
bare DNA. In addition, wrapping of the DNA in a super-
helical path (chiral wrapping) will also cause a linking
number change, e.g., a single left-handed superhelical turn
also induces a linking number change of -1.0. The linking
number difference is readily quantified by electrophoresis
of the two DNA samples in an agarose gel.
Early topological studies of polymerase complexes did not
analyze specific known prokaryotic promoters, leading to
an uncertainty about polymerase occupancy due to the
inability to demonstrate promoter saturation by titration
analysis (31-34). Such saturation was demonstrated in an
analysis of open complex formation at the lac UV5 promoter,
where Amouyal and Buc (35) concluded that complex
formation induced a linking number change of -1.7.
Chemical probe results (7, 8, 21, 22) predicted that bubble
formation would induce a change of only ∼-1. Following
from the nucleosome example in which the linking number
change contains contributions from both duplex winding
change and superhelical wrapping (36), Amouyal and Buc
proposed that the excess -0.7 linking number change
represents a left-handed superhelical wrap of about 0.7 turn
(35). This has been widely cited as some of the earliest
evidence for wrapping of DNA on the polymerase surface
[reviewed in (37)].
There are some difficulties with this interpretation of these
results. First, the presence of several additional promoters
in the plasmids meant that the lac UV5 contribution was
only a minor portion of the total polymerase-induced linking
number change, i.e., there was a substantial “background”
due to additional promoters. This, coupled with the use of
only a single copy of the lac UV5 promoter, limited the
precision of the measurement. Second, similar studies of
other promoters (38, 39) obtained smaller changes that were
closer to the -1 predicted by the chemical probe studies,
i.e., values that do not represent an excess of -0.7 linking
number difference that leads to invoking superhelical DNA
wrapping on the polymerase surface.
To address these various issues, we have developed a
modified topological characterization of the transcription
complex. The analysis is performed on DNA circles that
contain multiple copies (8 or 10) of a single promoter only,
resulting in an amplified linking number change signal. This
amplification of the linking number change as well as the
absence of other promoters means that the linking number
change per promoter can be measured with high precision
and specificity. Both the lac UV5 promoter and bacterioph-
age lambda PR promoter (λP
R
1
) have been analyzed. Several
types of experimental results indicate that the linking number
change represents bubble formation and not superhelical
wrapping of DNA around the open complex. Accordingly,
the value -1.17 measured for the lac UV5 complex indicates
a bubble of 12.3 bases, while the value of -0.98 measured
for λP
R
indicates a bubble of 10.3 bp. In addition, analysis
of λP
R
abortive initiation complexes containing RNA tran-
scripts extending to positions +2 and +3 indicates that the
bubble expands with increasing transcript length. Finally, it
was found that the bubble size was essentially the same in
the presence or absence of Mg
2+
, a finding that contrasts
with the conclusions of chemical probe studies (10, 11). The
relevance of these results to structures of intermediates in
the process of transcription initiation as well as the use of
1
Abbreviations: λP
R
, the bacteriophage lambda PR promoter; ∆L,
DNA topological linking number change; bp, base pair.
1872 Biochemistry, Vol. 47, No. 7, 2008 Tchernaenko et al.