Cross-linking of Promoter DNA to T7 RNA Polymerase Does Not
Prevent Formation of a Stable Elongation Complex*
Received for publication, July 8, 2004, and in revised form, August 5, 2004
Published, JBC Papers in Press, August 10, 2004, DOI 10.1074/jbc.M407688200
Edward A. Esposito‡ and Craig T. Martin§
From the Department of Chemistry, University of Massachusetts at Amherst, Amherst, Massachusetts 01003-9336
T7 RNA polymerase recognizes a small promoter,
binds DNA, and begins the process of transcription by
synthesizing short RNA products without releasing pro-
moter contacts. To determine whether the promoter
contact must be released to make longer RNA products
and at what position the promoter must be released, a
mutant RNA polymerase was designed that allows cross-
linking to a modified promoter via a covalent disulfide
bond. The modifications individually have no measura-
ble effect on transcription. Under oxidizing conditions
that produce the protein-DNA cross-link, the complex is
able to synthesize short RNA products, strongly sup-
porting a model in which promoter contacts are not lost
on translocation through at least position ⴙ6. However,
cross-linked complexes are impaired in promoter escape
in that only about one in four can escape to make full-
length RNA. The remainder release 12- and 13-mer RNA
transcripts, suggesting an increased energetic barrier
in the transition from an initial transcribing complex to
a fully competent elongation complex. The results are
discussed in the context of a model in which promoter
release helps drive initial collapse of the upstream edge
of the bubble, which, in turn, drives initial displacement
of the 5ⴕ-end of the RNA.
T7 RNA polymerase recognizes a relatively small promoter
with near nanomolar affinity (1, 2) and transcribes DNA in a
manner that appears to be mechanistically similar to that of
the more complex, multi-subunit, eukaryotic and prokaryotic
RNA polymerases (3). Because T7 RNA polymerase is a single
subunit enzyme capable of carrying out the complete transcrip-
tion cycle without additional protein cofactors, it is an ideal
enzyme to study as a model. Like other DNA-dependent RNA
polymerases, T7 RNA polymerase recognizes and binds pro-
moter DNA, melts open an initiation bubble downstream of the
promoter, and positions the initial templating bases in the
active site to begin the process of transcription (4, 5). After an
initial abortive cycling phase characteristic of all RNA poly-
merases, the enzyme enters a more stable elongation phase
after transcription of an ⬃10–14-mer RNA product (6 –9). Pre-
vious studies have suggested that promoter release occurs near
the position at which the enzyme switches to the more stable
elongation phase, perhaps simultaneously (10). Details con-
cerning the timing and mechanism of promoter release, how-
ever, remain unclear.
Several studies, including footprinting and fluorescence ap-
proaches (2, 10, 11), have shown that polymerase binds the
promoter DNA, melts open an initiation bubble positioning the
templating (position ⫹1) base in the active site, and then be-
gins transcription, all while maintaining promoter contacts.
During the early abortive cycling phase (at least until the
polymerase reaches position ⫹6), the promoter contacts remain
intact as evidenced by footprinting, although there may be
minor perturbations as evidenced by photo cross-linking stud-
ies (12). It has also been clearly shown that these promoter
contacts are released at some position beyond ⫹6 and prior to
⫹15 in the transcription cycle (2, 10). This phenomenon of
maintaining promoter contacts until the polymerase reaches a
more stable elongation phase is also characteristic of other
polymerases (13, 14).
The timing of promoter release corresponds well to an in-
crease in the overall stability of the ternary complex. There are
fewer abortive products released after translocation to position
⫹10 and lower overall turnover values for complexes artifi-
cially stalled beyond position ⫹8 (6). Alternatively, on con-
structs where promoter release may be inhibited (nicked non-
template strand or partial single-stranded promoters), an
abundance of 12- and 13-mer products relative to runoff (full-
length) products is observed (15–17). Recent studies suggest
that these 12- and 13-mer products may be the result of a failed
bubble collapse at the start site.
1
We expect that bubble col-
lapse and promoter release are coupled; the current study
focuses on the latter.
To determine the timing of promoter release or to ask
whether promoter release is required for the formation of a
stable elongation complex, we have designed a mutant T7 RNA
polymerase (A94C) that allows us to reversibly cross-link the
polymerase to its promoter DNA. The native alanine at position
94, shown in Fig. 1, is unconserved in the phage polymerases
and lies near the 3⬘-hydroxyl of the adenine at position ⫺17 of
the promoter (3, 18–21). Replacement of this template strand
3⬘-hydroxyl by a phosphodiester alkyl thiol allows covalent
cross-linking of promoter DNA to the protein. Under oxidizing
conditions, we have been able to effectively cross-link 3⬘-thiol-
modified promoter DNA to the protein, such that the complete
loss of promoter contacts is impossible. This cross-linking
should not impair the initial events of initiation that do not
require promoter release. The results presented here confirm
that cross-linking promoter DNA to the enzyme in its binding
site has no effect on the synthesis of short products (ⱕ6-mer),
and although the complexes can escape to produce full-length
* This work was supported by National Institutes of Health Grant
1RO1 GM55002. The costs of publication of this article were defrayed in
part by the payment of page charges. This article must therefore be
hereby marked “advertisement” in accordance with 18 U.S.C. Section
1734 solely to indicate this fact.
‡ Supported by National Institutes of Health National Research Ser-
vice Award T32 GM08515.
§ To whom correspondence should be addressed: Dept. of Chemistry,
University of Massachusetts at Amherst, 710 N. Pleasant St., LGRT
701, Amherst, MA 01003-9336. Tel.: 413-545-3299; Fax: 413-545-4490;
E-mail: [email protected].
1
Gong, P., Esposito, E. A., and Martin, C. T. (2004) J. Biol. Chem.
279, 44277–44285.
THE JOURNAL OF BIOLOGICAL CHEMISTRY Vol. 279, No. 43, Issue of October 22, pp. 44270–44276, 2004
© 2004 by The American Society for Biochemistry and Molecular Biology, Inc. Printed in U.S.A.
This paper is available on line at http://www.jbc.org44270
runoff products, cross-linking of the promoter to the enzyme
presents a new (or increased) energetic barrier leading to pre-
mature release of RNA at positions ⫹12 and ⫹13.
MATERIALS AND METHODS
Mutant Construction—An expression vector coding for the mutant
polymerase (His-tagged A94C) was prepared by utilizing the Strat-
agene QuikChange™ site-directed mutagenesis kit. The parental plas-
mid was isolated from cells (pBH161/BL21) generously provided by W.
T. McAllister. The oligonucleotide primers (5⬘-GACTGGTTTGAG-
GAAGTGAAATGTAAGCGCGGCAAGCGCCCG-3⬘ and its comple-
ment) directing the single amino acid mutation were purchased from
Integrated DNA Technologies (Coralville, IA). The underlined region
encodes the alanine (GCT) to cysteine (TGT) mutation. Candidate plas-
mids were sequenced to confirm the mutation and then transformed
into BL21 cells. Mutant polymerase activity, which was the same as
wild type, was assayed using methods described below.
Protein Expression and Purification—His-tagged wild type and mu-
tant T7 RNA polymerase were overexpressed in Escherichia coli strain
BL21 and purified using Qiagen nickel-nitrilotriacetic acid as described
(22). Protein purity (⬎95%) was determined by SDS-PAGE analysis.
The purified protein was dialyzed against storage buffer (20 mM potas-
sium phosphate, pH 7.8, 100 mM NaCl, 50% glycerol, and 1 mM
Na
2
EDTA) and stored at ⫺20 °C in the same buffer containing 1 mM
dithiothreitol (DTT).
2
Concentration was calculated from the measured
absorbance at 280 nm (in the absence of DTT) using the molar extinc-
tion coefficient of 1.4 ⫻ 10
5
M
⫺1
cm
⫺1
(23).
Oligonucleotide Synthesis and Purification—Oligonucleotides were
synthesized trityl-off using an Applied Biosystems Expedite 8909 DNA
synthesizer, gel purified as described previously (24), excised from the
gel, and eluted using an Elu-Trap® device (Schleicher and Schuell Inc.,
Keene, NH). The DNA sequences encoding a 20-mer RNA are 5⬘-TAA-
TACGACTCACTATAGGGAGACCACAACGGTTTCC-3⬘ (nontemplate)
and 3⬘-ATTATGCTGAGTGATATCCCTCTGGTGTTGCCAAAGG-5⬘
(template).
A3⬘ thiol-modified DNA template strand (HS–CH
2
–CH
2
–CH
2
–PO
4
–
3⬘-template) and a 3⬘-biotinylated nontemplate strand were synthe-
sized in a similar manner using either a 3⬘-thiol modifier or a 3⬘-biotin
column (Glen Research, Sterling VA). Purified single strand DNAs were
combined at equimolar concentrations, heated to 90 °C, and then cooled
slowly to room temperature to anneal.
Isolation of Cross-linked Complexes—To form initial cross-linked
complexes by disulfide formation, solutions (200
l) containing equimo-
lar concentrations (1
M) of T7 RNA polymerase (A94C) and 3⬘-thiol-
modified, double-stranded DNA were run through a 1-ml Sephadex
G-25 column equilibrated with 30 mM HEPES, pH 7.8, 0.25 mM EDTA,
25 mM potassium glutamate, and 0.025% Tween 20. To facilitate com-
plete cross-linking, samples were then incubated at 4 °C for 4–10 days
with the tubes opened at least once daily to allow air exchange. In later
experiments that attempted to ensure more complete cross-linking,
protein concentration was increased relative to DNA in ratios of 4:1 and
10:1 prior to treatment with G-25 Sephadex spun columns. Even with
these increased concentrations of protein there was significant loss of
protein to the column material, presumably via nonspecific adsorption.
These samples were used directly in transcription assays described
later.
To better isolate the cross-linked complex, subsequent experiments
utilized DYNAL® Dynabeads® M-280 Streptavidin (Dynal Inc., Lake
Success, NY) and biotinylated DNA. Cross-linked enzyme-DNA com-
plexes, oxidized as described above in 30 mM HEPES, pH 7.8, 0.25 mM
EDTA, 25 mM potassium glutamate, and 0.5% glycerol (glycerol was
substituted for Tween 20 because of the tendency of Tween 20 to form
peroxides), were added to 25
l of beads washed previously and equil-
ibrated into the above buffer as described in the DYNAL protocol. The
samples were then washed three times with 1⫻ DYNAL binding and
wash buffer (10 mM Tris-HCl, 1 mM EDTA, and 1 M NaCl), transferred
to clean Eppendorf tubes between washes to prevent residual protein
contamination, and then re-equilibrated to 30 mM HEPES, pH 7.8, 25
mM potassium glutamate, 0.25 mM EDTA, 0.5% glycerol, and 50 mM
NaCl with two washes (see Fig. 5A for schematic representation of the
isolation protocol). These samples, beads included, were used directly in
the transcription assays described later. Control experiments substi-
tuted native protein for the A94C mutant and/or native DNA for the
3⬘-thiol-modified DNA.
Transcription Assays—Transcription reactions were performed in a
total volume of 20
l at 37 °C for 10 min and quenched with an equal
volume of stop solution (95% formamide, 20 mM EDTA, 0.01% brom-
phenol blue, and 0.01% xylene cyanol). Equimolar concentrations of
double strand DNA and enzyme were used at final concentrations of 0.2
M in a reaction buffer containing 30 mM HEPES, pH,7.8, 25 mM
potassium glutamate, 15 mM magnesium acetate, 0.25 mM EDTA, 0.5%
glycerol, 50 mM NaCl, and 1 mM freshly prepared DTT (absent in
oxidized samples). Reactions were initiated by the addition of nucleo-
side triphosphates to a final concentration of 400
M each and were
labeled with 1
Ci of [
␣
-
32
P]GTP.
Transcription assays with samples on beads or samples run through
the G-25 Sephadex spun columns were performed at 37 °C for 10 min
and then quenched with an equal volume of stop solution. RNA prod-
ucts were separated on a 7
M urea, 20% polyacrylamide Tris-borate
gel and quantified using a Storm 840 PhosphorImager as described
previously (25).
Oxidation by Glutathione or Diamide—In some experiments, oxi-
dized glutathione and 1,1⬘-azobis(N,N-dimethylformamide) (diamide)
were utilized to ensure that any non-cross-linked complex was driven
toward complete oxidation (26, 27). To demonstrate that the disulfide
cross-link was not reversibly exchanging, a 1-
l volume of 10 mM
oxidized glutathione, 50 mM oxidized glutathione, and 10 mM diamide or
double-distilled H
2
O (as a control) was added to 9
l of bead-isolated,
cross-linked complexes to increase the oxidizing strength of the buffer
solution. Each sample was then incubated for1hat4°Cbefore use in
transcription assays as described above.
RESULTS
Recent crystal structures of elongation complex models
strongly support a mechanism for transcription in which en-
zyme-promoter contacts are completely lost on progression to
an elongation complex (2, 10, 28, 29). To determine whether
initial promoter contacts must be released when the enzyme
switches to a more stable elongation complex, a mutant polym-
erase was designed that allows covalent cross-linking of the
promoter DNA to the promoter binding region of the enzyme in
the initially bound complex. There are three regions of the
enzyme that make up the promoter binding domain as revealed
by contacts seen in the DNA-bound crystal structures (3, 21).
As shown in Fig. 1, these regions include the AT-rich recogni-
tion loop centered on Arg-96, the specificity loop (residues
745–759), and an intercalating loop centered on Val-237.
Residue Ala-94, in the first of these regions, was chosen for
mutation based on its close proximity to the 3⬘-hydroxyl of the
adenosine at position ⫺17 of the template strand DNA, its lack
of contact with the DNA or the rest of the protein, its lack of
conservation among the closely related phages T3, SP6, and
K11 (18–20), and biochemical evidence that minor changes in
this region of the promoter DNA are reasonably tolerated (30).
3
Template strand DNA was synthesized with a 3⬘-thiol modifi-
cation so that it could be directly cross-linked via formation of
a disulfide bond with Cys-94. We expected that cross-linking of
the promoter DNA would allow normal initiation and initial
translocation, whereas full run transcription would be elimi-
nated because the enzyme would not be able to escape the
promoter.
Modifications Have No Effect on Transcription in the Absence
of Cross-linking—To ensure that modification of the protein
alone has no direct effect on transcription, we compared tran-
scription assays with mutant A94C and wild type T7 RNA
polymerase. In each case we used double-stranded DNA con-
taining an upstream consensus promoter and a downstream
sequence encoding a 20-mer runoff transcript. Results pre-
sented in lanes 3 and 4 of Fig. 2 show that, under reducing
conditions, transcription from the mutant polymerase is essen-
tially identical to that of wild type polymerase.
Similarly, to ensure that adding a thiol group to the 3⬘-end of
2
The abbreviation used is: DTT, dithiothreitol.
3
F. Tanga, E. A. Esposito, and C. T. Martin, unpublished results.
Restricting Promoter Release via Covalent Cross-linking 44271
the template strand DNA has no detrimental effect on binding
or on the ability of the enzyme to transcribe, we analyzed
transcription by wild type T7 RNA polymerase on an identical
DNA sequence containing a 3⬘-thiol on the template strand (see
Fig. 2). As shown in lane 2 of Fig. 2, the addition of the 3⬘-thiol
to the DNA has no effect on transcription. Finally, comparison
of lanes 1 and 4 in Fig. 2 shows that under the normal reducing
conditions of our assays, which should prevent formation of the
disulfide, the combination of both A94C and the addition of a
3⬘-thiol on the DNA has no effect on transcription. In all cases,
there is no significant difference in the amount of transcribed
product using either enzyme, nor is there a difference in the
product profile.
Covalent Cross-linking Does Not Alter Short Product Synthe-
sis—Because T7 RNA polymerase retains promoter contacts
from binding through initial transcription (at least through
position ⫹6), it is expected that a cross-linked complex should
initiate as well as or better than an uncross-linked complex and
that translocation to at least position ⫹6 should be unimpeded.
To test this hypothesis, DTT was removed from solutions con-
taining DNA and enzyme (as described under “Materials and
Methods”) to drive the formation of a disulfide bond between
the protein and the DNA.
In a transcription assay with only GTP as the substrate on a
promoter encoding GGGA at the start site, T7 RNA polymerase
produces a range of poly(G) products up to ⬃14 bases in length
(8, 25). Because forward translocation is minimal in this sys-
tem, it is a simple measure of initiation. As shown in Fig. 3,
cross-linking the promoter to the enzyme has no effect on the
enzyme’s ability to synthesize poly(G) products. Similarly, in
the presence of GTP and ATP on a promoter encoding the
substrate-limited six-base product GGGAGA, there is no dif-
ference in the product profile (intensity differences between
lanes are attributed to differences in concentrations after proc-
essing). With no change in the product profile (relative
amounts of 3-mer or 4-mer to 6-mer) evident, we conclude that
initiation and early transcription are not adversely affected by
the covalent cross-linking of the promoter DNA to the enzyme.
Cross-linked Complex Can Make Full Run Product—Based
on earlier observations from both footprinting and fluorescence
experiments, we predicted that restricting promoter release
would allow the synthesis of only a 9-mer to 11-mer product.
Footprinting experiments have shown that promoter release
occurs at some time after the synthesis of a 6-mer product and
before the synthesis of a 15-mer (2, 10). Similarly, we have
shown that the initially melted bubble collapses after synthesis
of a 8- to 9-base pair product, presumably concurrent with
promoter release (11). Thus, we hypothesized that locking the
promoter onto the enzyme by covalently cross-linking the up-
stream promoter DNA to the AT-rich recognition loop (a sub-
FIG.1.Schematic highlighting the promoter binding region of T7 RNA polymerase interacting with the AT-rich recognition loop
(AT) containing Ala-94 (A94), the specificity loop (SL), and the intercalating valine loop (IV). The
␣
-carbon of alanine 94 is 5.4 Å from
the 3⬘-hydroxyl of the template strand DNA in the crystal structure (Protein Data Bank code 1QLN) and was chosen for mutation to a cysteine
based on its proximity to the 3⬘-OH and its lack of conservation among the phage polymerases and because changes to the immediately adjacent
promoter DNA are reasonably well tolerated. The 3⬘-OH of a synthetic promoter template strand has been modified to a phosphodiester alkane
thiol (HS-CH
2
–CH
2
–CH
2
–PO
4
–3⬘-template) in order to covalently attach the promoter DNA to the polymerase. A schematic of the promoter DNA
is shown (below) with the specific binding region ⫺17 to ⫺5 highlighted in gray, identifying the initially melted bubble region (gray oval) and noting
the 3⬘-thiol modification.
FIG.2.Comparison of transcription by wild type T7 RNA po-
lymerase and mutant (A94C) on native template DNA and on
3ⴕ-thiol-modified DNA under reducing conditions (1 m
M DTT).
The mutant polymerase has wild type activity on both native and
3⬘-thiol-modified DNA. Neither modification has any apparent effect on
binding or activity. Except where indicated, transcription assays were
performed for 10 min at 37 °C with equimolar enzyme and DNA; NTP
substrates were present at 400
M concentration. Final reaction buffers
are described under “Materials and Methods.”
Restricting Promoter Release via Covalent Cross-linking44272
domain of the promoter binding region) would limit the length
of the final product to an ⬃10-mer RNA. The results presented
in Fig. 4, however, indicate that the cross-linked complex is
indeed able to synthesize a full-length runoff (20-mer) product.
Although a full-length transcript is produced by the cross-
linked complex, there is a substantial increase in the amount of
12- and 13-mer falloff products relative to the 20-mer runoff
product.
Under our current conditions, the control (uncross-linked)
wild type DNA and protein produce 12- and 13-mer products at
⬃8–10% each relative to the amount of 20-mer runoff product
(Fig. 4, lane 3). In contrast, the cross-linked complex produces
12- and 13-mer products that are ⬎200% the amount of the
20-mer product. The finding that these same short products are
seen (at low levels) with the wild type enzyme and DNA sug-
gests that they reflect an intrinsic barrier to escaping the
promoter. Promoter release is likely to be that barrier. Cova-
lent cross-linking of the DNA to the enzyme increases the
probability that proper promoter release will not occur and,
therefore, increases the amount of the short product generated.
In this experiment, the cross-link was allowed to form via
simple air oxidation (4 days at 4 °C) following removal of DTT
with a G-25 Sephadex spun column. As a result, differences in
the overall amounts of transcripts between lanes might arise
from a loss of protein during oxidation and/or gel filtration. In
these experiments we can also not be sure that the complexes
are 100% cross-linked, such that the observed full-length runoff
product might arise from a population of uncross-linked com-
plexes. The following experiments address this uncertainty.
Runoff Product Is Produced by Bona Fide Cross-linked Com-
plexes—To better isolate cross-linked complexes, double-
stranded DNA containing an upstream 3⬘-thiol group on the
template strand and a downstream 3⬘-biotin moiety on the
nontemplate strand was incubated with the mutant enzyme in
the absence of DTT, as described above. The resulting cross-
linked complexes were then captured using streptavidin-coated
paramagnetic beads. The beads were then washed using buffer
containing 1
M NaCl to remove any non-covalently bound pro-
tein. After equilibrating the beads to transcription buffer, tran-
scription assays performed directly from these beads showed
ratios of 12- and 13-mer products relative to 20-mer products
similar to those seen above (Fig. 5B). In a control experiment,
wild type enzyme and DNA containing the downstream biotin
but lacking the upstream 3⬘-thiol were incubated, captured,
washed, and assayed as described above. No transcription was
observed in this control, indicating efficient washing of the
noncovalently bound enzyme from the DNA. The subsequent
addition of free enzyme restored wild type levels of transcrip-
tion, demonstrating that streptavidin-bound DNA was re-
tained (data not shown).
Finally, to ensure that the observed 20-mer transcript does
not arise from complexes in which the cross-link has reversed,
we employed glutathione and diamide to maintain complete
oxidation (26, 27). As shown in Fig. 5C, the addition of gluta-
thione to the bead-isolated cross-linked complex (to a final
concentration of 1 m
M) does not alter the percentage of 12- and
13-mer products relative to 20-mer, demonstrating that the
12-, 13-, and 20-mer products are being synthesized by fully
cross-linked complexes. The addition of glutathione to 5 m
M
(not shown) or the addition of diamide (a stronger oxidizer) to
1m
M (Fig. 5C, lane 2) significantly reduces the overall activity
of the protein. A similar reduction in activity is seen for a
parallel treatment of the wild type enzyme, suggesting that
under these stronger oxidizing conditions the native cysteines
are forming nonnative cross-links, thus reducing the protein
activity (not shown).
FIG.4. Runoff transcription from T7 RNA polymerase-DNA
complexes under oxidizing conditions. Comparison of lane 1 with
the controls in lanes 2– 4 shows that an abundance of 12- and 13-mer
RNA is produced from cross-linked complexes. Surprisingly, cross-
linked complexes also produce the 20-mer runoff transcript. WT, wild
type; 3⬘-SH DNA,3⬘-thiol modified DNA.
FIG.3.Synthesis of short RNA products from T7 RNA polym-
erase-DNA complexes under oxidizing conditions (DTT re-
moved by gel filtration). Indicated above each lane are the enzymes
(mutant or wild type (WT)) and the DNA (3⬘-thiol-modified (3⬘-SH)or
native) used in each experiment. Differences in overall intensities be-
tween lanes are attributed to intermolecular cross-linking as well as to
concentration differences following DTT removal. A, given only GTP as
a substrate, a G-ladder is made by all complexes with no change in
product profile. B, given GTP and ATP as substrate, a 6-mer product is
made, and, similarly, there is no change in product profile.
Restricting Promoter Release via Covalent Cross-linking 44273
Are All Complexes Coupled to Cys-94?—High salt washing of
complexes containing the native alanine at position 94 and
lacking a 3⬘-thiol on the DNA template led to a complete loss of
transcription, demonstrating that noncovalently bound polym-
erase can be efficiently washed from the bead-DNA complex.
This finding does not, however, preclude cross-linking of the
modified DNA to any of the 12 native cysteine residues in T7
RNA polymerase.
Although the native cysteine nearest the 3⬘-thiol is ⬃24 Å
distant (C216), a secondary control was run to address the
possibility that wild type polymerase could be cross-linked to
the thiol-modified DNA through one of these native cysteines.
In the previous transcription assay the wild type enzyme was
not completely washed away, as there was residual activity
resulting in a faint band corresponding to 20-mer RNA (⬃10%
of that seen in the A94C control) and very weak bands corre-
sponding to 12- and 13-mer RNAs (data not shown). Cross-
linking of promoter DNA to any of the native cysteines would
allow the RNA polymerase to survive the high salt challenge,
but function with DNA would almost certainly happen in trans,
as it should not allow correct positioning of the promoter on the
protein to which it is cross-linked.
If one enzyme can utilize DNA tethered to a separately
tethered enzyme, one would expect that complex to be suscep-
tible to free, competing promoter DNA. In contrast, complexes
with DNA bound at Cys-94 can be expected to be resistant to
such competition. Therefore, the addition of a promoter sink
that binds but will not support transcription can distinguish
between complexes cross-linked properly via Cys-94 in the
promoter binding domain and those that are cross-linked to one
of the native cysteines. Complexes cross-linked to Cys 94
should be resistant to the sink (a free sink promoter cannot
effectively compete with a locally tethered promoter), whereas
the complexes containing promoter DNA accessible in trans
should be completely inhibited by the trap. In the bead-isolated
control experiment with wild type enzyme and thiol modified
DNA there is no increase in the ratio of 12- and 13-mer prod-
ucts relative to 20-mer, and transcription from these complexes
is completely inhibited by the addition of the sink, suggesting
that this residual level of transcription is occurring in trans
(data not shown). This result raises the possibility that some of
the cross-linked complexes in the experiment with the A94C
mutant are cross-linked via one of the native cysteines. Tran-
scription in trans would lead to full-length transcripts.
To assess whether some of the 20-mer RNA observed in Fig.
5B is being synthesized by complexes transcribing in trans,
similar assays were run but with increasing concentrations of
promoter sink. The results shown in Fig. 6 show that challeng-
ing the cross-linked, bead-isolated complexes with promoter
sink leads to only a small decrease in 20-mer synthesis. This
small decrease must reflect a similarly small population of
incorrectly cross-linked complexes operating in trans. At the
end point of this titration, the ratio of 12- and 13-mer products
relative to 20-mer increases to 2.9, which should now reflect
only those complexes cross-linked via Cys-94. Titration of sink
into enzyme and thiol-modified DNA incubated previously un-
der the oxidizing conditions described above but not bead-
isolated similarly shows a final maximal ratio of 2.7 (data not
shown). Together, these data show that bona fide cross-linked
complexes terminate transcription at positions 12 and 13 ⬃75%
FIG.5. Runoff product is produced by bona fide cross-linked complexes. A, schematic diagram of the procedure utilized to isolate
cross-linked complex. B, transcription from bead-washed complexes. Wild type enzyme (WT) does not cross-link and, therefore, is efficiently washed
from the beads. C, using a bead-isolated cross-linked complex, buffer, diamide, or oxidized glutathione (lanes 1, 2,or3, respectively) were added
to ensure that the complexes were fully oxidized.
Restricting Promoter Release via Covalent Cross-linking44274
of the time. Only 25% escape to form competent elongation
complexes.
As a final control to show that the effects of the cross-link are
reversible, DTT was added to the transcription assay at a final
concentration of 50 m
M. For bead-isolated cross-linked com-
plexes, the addition of DTT restored the ratios of 12- and
13-mer to 20-mer products to near native levels (data not
shown). The addition of DTT to 10 m
M (or lower) does not fully
reverse the cross-link, as observed previously for DNA disulfide
cross-linked to its native protein binding site (31, 32).
DISCUSSION
Despite dissimilarities in sequence and structure, all RNA
polymerases are remarkably similar in that they transition
through an abortive cycling phase prior to entering the more
stable elongation phase (4, 5, 8, 9, 33). This transition from an
unstable initiation complex to a stable elongation complex oc-
curs after synthesis of ⬃10 bases. Understanding this con-
served transition is critical for understanding the mechanisms
and regulation of transcription.
T7 RNA polymerase, like the multi-subunit bacterial en-
zyme, binds promoter DNA, melts open the initiation region,
and begins transcription while maintaining initial promoter
contacts (13, 14). At some point during the transition from an
unstable initiation complex to a stable elongation complex,
promoter contacts are lost (2, 10, 34).
Recently, two different research groups have published crys-
tal structures of T7 RNA polymerase elongation complexes
derived from synthetic RNA-DNA scaffolds (28, 29). Both crys-
tal structures show a dramatic rearrangement of the N-termi-
nal domain, including a region of the portion of the enzyme
responsible for promoter recognition and binding (3, 21). In the
elongation complex, two of the three promoter binding ele-
ments have moved as a rigid body with respect to the C-
terminal domain. This rigid body movement might allow the
enzyme to bring downstream DNA into the active site (extend-
ing the footprint downstream) while maintaining upstream
promoter contacts (and the upstream footprint). Tahirov et al.
have proposed a model for a late initiation complex in which an
8-mer RNA can be formed while two of the three initial pro-
moter contacts are retained (28). In this model, contacts are
retained between the AT-rich recognition loop and the ⫺17 to
⫺14 region of the promoter DNA, as well as between the
intercalating loop and the promoter DNA at ⫺5. This model
also predicts that the initially melted bubble remains open
when the polymerase reaches position ⫹8 but that the bubble
collapses and the promoter contacts are lost on translocation
beyond positions ⫹9to⫹11. This interpretation is supported
by biochemical studies (11, 35) that additionally suggest that
promoter release may not simply occur at one defined/precise
position. A more recent model suggests that the rigid body
rotation is preceded by a simple back translocation of the
promoter binding element.
4
In this case, promoter contacts can
also be retained during initial translocation.
By monitoring the melted state of the DNA bases at position
⫺2, Liu and Martin (11) showed that bubble collapse begins
when the polymerase translocates to position ⫹9 and is nearly
complete by translocation to position ⫹11. In a similar study
with exonuclease as a footprinting probe, Brieba and Sousa
showed that promoter release may begin as early as position
⫹8 but again suggested that timing of the release is non-
homogeneous (35). When stalled at position ⫹7 and then trans-
located to position ⫹8 by addition of the next incoming NTP,
5–10% of the complexes exhibit a shift in the upstream bound-
ary of exonuclease protection. In complexes walked to position
⫹8, 40% of the complexes show a shift. Taken together, these
results support a model in which promoter release begins when
the polymerase reaches position ⫹8 and is likely complete
when the polymerase translocates to position ⫹11.
To assess the functional importance of promoter release, we
have constructed a covalently cross-linked binary complex be-
tween T7 RNA polymerase and its promoter. Current models
predict that the cross-linked complex should initiate as well as
the corresponding non-cross-linked controls (2, 10, 11). In an
attempt to determine the position at which promoter release
must occur, we allowed the cross-linked complex to transcribe a
20-mer runoff product and compared the products of the cross-
linked complex with those of the non-cross-linked complex.
Cross-linking Does Not Perturb Transcription through Posi-
tion ⫹6—As expected, cross-linking of promoter DNA to the
promoter binding region of T7 RNA polymerase has no effect on
the initiation of transcription. Product profiles from cross-
linked complexes in the presence of GTP only (allowing trans-
location to position ⫹3) or in the presence of GTP and ATP as
the substrate (allowing translocation to position ⫹6) are iden-
tical to those of the uncross-linked control. This shows clearly
that initial promoter contacts need not be released to synthe-
size up to, at least, a 6-mer product.
Cross-linking Perturbs but Does Not Eliminate Escape to an
Elongation Complex—In the cross-linked constructs created in
this study, we expect that the loss of promoter contacts and
subsequent (or concurrent) initial bubble collapse (from posi-
tion ⫺4 downstream) should both be impeded. Thus, some
4
Theis, K., Gong, P., and Martin, C. T. (2004) Biochemistry, in press.
FIG.6. Use of a promoter sink to
“remove” incorrectly cross-linked
complexes. A promoter sink was titrated
into the cross-linked complex. Transcrip-
tion between improperly cross-linked
complexes and naked DNA is efficiently
inhibited by the addition of promoter
sink, whereas the correctly cross-linked
complex is resistant. A, titration curve
showing a maximum of 290% of 12- and
13-mer product to 20-mer product, indicat-
ing that the 20-mer product is indeed pro-
duced by properly cross-linked enzyme-
DNA complexes. B, transcription assay
with increasing sink:DNA concentration.
Restricting Promoter Release via Covalent Cross-linking 44275
complexes do not proceed to become stable elongation com-
plexes and instead release RNA products 12–13 bases in
length. Indeed, ⬃75% of initiated complexes stop at ⫹12 and
⫹13; only 25% successfully pass this barrier to go on to syn-
thesize full-length RNA products. That 25% escape suggests
that the barrier is not absolute.
What is the nature of this barrier and why does transcription
stop at positions ⫹12 and ⫹13 rather than positions ⫹8to
⫹10? A similar increase in 12- and 13-mer transcripts relative
to full-length products is observed in transcription from con-
structs that do not allow normal bubble collapse. An increase in
12- to 13-mer products can be seen in constructs that are nicked
on the nontemplate strand in the region of the initially melted
bubble, constructs that have an artificially melted (noncomple-
mentary) bubble, and partially single-stranded DNA constructs
(15–17). It has been suggested that improper RNA displace-
ment results in a complex that cannot transcribe well beyond
position ⫹13 (16). Artificial bubble scaffolds, such as those that
were utilized to trap the elongation complex conformation for
crystallographic studies, also lack the ability to properly dis-
place the upstream end of the RNA and are similarly unable to
make products longer than a 13-mer with any efficiency (33).
All of these constructs prevent or weaken the collapse of the
initially melted bubble (or of the upstream edge of the bubble in
the case of the scaffold) and therefore weaken the ability of the
complex to competitively displace the 5⬘-end of the nascent
RNA.
We propose that the increase in the amounts of 12- and
13-mer products from our cross-linked constructs similarly
arises from an impairment of bubble collapse, leading to an
impairment in the proper displacement of the 5⬘-end of the
RNA. In the current case, however, bubble collapse is impaired
by maintenance of the promoter contact, suggesting that pro-
moter release contributes directly to bubble collapse. This is to
be expected, because the intercalating loop in promoter-bound
complexes is thought to stabilize the melted bubble (15, 17).
Release of the promoter during promoter clearance therefore
destabilizes the bubble. In either case, incorrect or delayed
bubble collapse prevents proper positioning of the 5⬘-end of the
nascent RNA into the RNA exit channel.
A Model for Promoter Escape—Recent studies provide strong
evidence that the timing of promoter release is simultaneous
with bubble collapse and that a contiguous, complementary,
nontemplate strand is required for native RNA displacement.
1
Based on those results and the results presented herein, we
believe that a critical event in the formation of a stable elon-
gation complex is bubble collapse, driving initial displacement
of the 5⬘-end of the nascent RNA for correct positioning near
the exit channel. Promoter release allows bubble collapse, so
limiting promoter release indirectly limits proper RNA dis-
placement. Either the lack of displacement or translocationally
delayed displacement prevents proper threading of the RNA
into the exit channel. We suspect, therefore, that complexes
that do not properly displace the RNA at position ⫹9 can
continue to elongate only 3–4 bases further, as in the elonga-
tion scaffolds, leading to the production of 12- to 13-mer RNA
transcripts.
Acknowledgments—We thank a devoted group of undergraduates who
participated in this project, namely Carlos J. Lo´pez Colo´n, Shannon
Reilly, Carolyn Robinson, Stefanie Stadnicki, and Alex Yazhbin.
REFERENCES
1. U
´
jva´ri, A., and Martin, C. T. (2000) J. Mol. Biol. 295, 1173–1184
2. Gunderson, S. I., Chapman, K. A., and Burgess, R. R. (1987) Biochemistry 26,
1539–1546
3. Cheetham, G. M., Jeruzalmi, D., and Steitz, T. A. (1999) Nature 399, 80–83
4. Gralla, J. D., Carpousis, A. J., and Stefano, J. E. (1980) Biochemistry 19,
5864–5869
5. Dvir, A., Conaway, J. W., and Conaway, R. C. (2001) Curr. Opin. Genet. Dev.
11, 209–214
6. Mentesana, P. E., Chin-Bow, S. T., Sousa, R., and McAllister, W. T. (2000) J.
Mol. Biol. 302, 1049–1062
7. Temiakov, D., Mentesana, P. E., Ma, K., Mustaev, A., Borukhov, S., and
McAllister, W. T. (2000) Proc. Natl. Acad. Sci. U. S. A. 97, 14109–14114
8. Martin, C. T., Muller, D. K., and Coleman, J. E. (1988) Biochemistry 27,
3966–3974
9. Hsu, L. M. (2002) Biochim. Biophys. Acta 1577, 191–207
10. Ikeda, R. A., and Richardson, C. C. (1986) Proc. Natl. Acad. Sci. U. S. A. 83,
3614–3618
11. Liu, C., and Martin, C. T. (2002) J. Biol. Chem. 277, 2725–2731
12. Place, C., Oddos, J., Buc, H., McAllister, W. T., and Buckle, M. (1999) Bio-
chemistry 38, 4948–4957
13. Carpousis, A. J., and Gralla, J. D. (1985) J. Mol. Biol. 183, 165–177
14. Straney, D. C., and Crothers, D. M. (1985) Cell 43, 449 – 459
15. Stano, N. M., and Patel, S. S. (2002) J. Mol. Biol. 315, 1009–1025
16. Jiang, M., Rong, M., Martin, C., and McAllister, W. T. (2001) J. Mol. Biol. 310,
509–522
17. Brieba, L. G., and Sousa, R. (2001) Biochemistry 40, 3882–3890
18. Dietz, A., Weisser, H. J., Kossel, H., and Hausmann, R. (1990) Mol. Gen. Genet.
221, 283–286
19. Kotani, H., Ishizaki, Y., Hiraoka, N., and Obayashi, A. (1987) Nucleic Acids
Res. 15, 2653–2664
20. McGraw, N. J., Bailey, J. N., Cleaves, G. R., Dembinski, D. R., Gocke, C. R.,
Joliffe, L. K., MacWright, R. S., and McAllister, W. T. (1985) Nucleic Acids
Res. 13, 6753–6766
21. Cheetham, G. M., and Steitz, T. A. (1999) Science 286, 2305–2309
22. He, B., Rong, M., Lyakhov, D., Gartenstein, H., Diaz, G., Castagna, R., McAl-
lister, W. T., and Durbin, R. K. (1997) Protein Expression Purif. 9, 142–151
23. King, G. C., Martin, C. T., Pham, T. T., and Coleman, J. E. (1986) Biochemistry
25, 36–40
24. Perez-Howard, G. M., Weil, P. A., and Beechem, J. M. (1995) Biochemistry 34,
8005–8017
25. Kuzmine, I., Gottlieb, P. A., and Martin, C. T. (2001) Nucleic Acids Res. 29,
2601–2606
26. Schoneich, C. (1995) Methods Enzymol. 251, 45–55
27. Kosower, N. S., and Kosower, E. M. (1995) Methods Enzymol. 251, 123–133
28. Tahirov, T. H., Temiakov, D., Anikin, M., Patlan, V., McAllister, W. T., Vas-
sylyev, D. G., and Yokoyama, S. (2002) Nature 420, 43–50
29. Yin, Y. W., and Steitz, T. A. (2002) Science 298, 1387–1395
30. Imburgio, D., Rong, M., Ma, K., and McAllister, W. T. (2000) Biochemistry 39,
10419–10430
31. Verdine, G. L., and Norman, D. P. (2003) Annu. Rev. Biochem. 72, 337–366
32. Huang, H., Harrison, S. C., and Verdine, G. L. (2000) Chem. Biol. 7, 355–364
33. Temiakov, D., Anikin, M., and McAllister, W. T. (2002) J. Biol. Chem. 277,
47035–47043
34. Huang, J., and Sousa, R. (2000) J. Mol. Biol. 303, 347–358
35. Brieba, L. G., and Sousa, R. (2001) EMBO J. 20, 6826 – 6835
Restricting Promoter Release via Covalent Cross-linking44276
