(Received for publication, August 2, 1996, and in revised form, December 4, 1996)
From the Division of Biology and Biomedical Sciences,
Washington University, St. Louis, Missouri 63130 and
¶ Department of Bacteriology, University of Wisconsin,
Madison, Wisconsin 53706
We tested the susceptibility of nucleic acid strands in a halted transcription elongation complex to digestion by micrococcal nuclease (MN). The 16-nucleotide nascent RNA was protected within RNA polymerase. A 27-28-nucleotide template strand DNA fragment also was resistant to MN digestion. However, the upstream half of the nontemplate DNA within this region was digested rapidly by MN, suggesting that the nontemplate strand emerges from the RNA polymerase near the middle of the melted transcription bubble with the bases oriented away from the enzyme surface. MN cleavage of the exposed nontemplate DNA shifted polymerase backward, making it unable to extend the RNA chain. However, the MN-trimmed G16 complexes could be reactivated by GreB-stimulated cleavage of the nascent RNA. These results favor a model of transcriptional arrest involving upstream slippage of RNA polymerase along the RNA and DNA chains. They also suggest that the exposed segment of nontemplate DNA may directly or indirectly stabilize the lateral position of the transcription complex along the DNA.
Transcribing RNA polymerase contacts ~30
bp1 of DNA, although the length of DNA
protected from cleavage agents can vary by as much as ~10 bp in
transcription complexes halted near initiation, pause, termination, or
arrest sites (1-9). An ~18-bp segment of DNA from near the upstream
edge to the middle of this footprint is melted to form the
transcription bubble, as indicated by both topological and chemical
reactivity assays (7, 9-12). The reactive region also can vary by ±5
bp in some halted complexes (7, 9, 11), although this has yet to be
tested by topological measurement. To maintain the transcription
bubble, the nontemplate DNA strand must be separated from the template
strand as polymerase moves forward and then reanneal after the template
strand is transcribed. Previous findings led us to suggest that,
whereas strand separation occurs within polymerase, ~10-15 bp
upstream from its forward edge on DNA, the strands reanneal passively
as the template DNA exits upstream (Refs. 11 and 13; see Fig. 1).
This hypothesis predicts that a part of the nontemplate DNA may be exposed on the outside of the transcription complex, rather than protected within it. The chemical reactivity of the nontemplate DNA is consistent with this view. Within the transcription bubble, bases on the nontemplate strand are highly reactive with single strand-specific reagents such as KMnO4, dimethylsulfate, or diethylpyrocarbonate (7, 9, 11, 12), whereas the phosphodiester backbone is mostly protected from cleavage by HO· (hydroxyl radical) (3, 7). The template DNA in contact with RNA polymerase also is protected from HO· generated in solution; however, it is readily cleaved by HO· generated from the intercalating reagent 1,10-phenanthroline copper, whereas the corresponding nontemplate DNA is not (14-16). This suggests that interactions of the template strand bases with polymerase or with RNA mediate high-affinity binding of 1,10-phenanthroline copper, but that nontemplate strand bases either are loosely associated with polymerase or held in a channel accessible to small reagents such as KMnO4 but not to the larger 1,10-phenanthroline copper. Although these findings are consistent with the idea that part of the nontemplate strand lies on the surface of RNA polymerase, they fall short of establishing it unambiguously or delineating which part is exposed. Furthermore, although the major sources of transcription complex stability recently were localized to contacts to the nascent RNA, the template DNA strand near the active site, and ~10 bp of duplex DNA in front of the transcription bubble (17), the possible role of the nontemplate DNA in stabilizing the position of polymerase along the DNA is unknown.
To address these questions, we examined the susceptibility of the
nontemplate DNA to cleavage by micrococcal nuclease (MN). MN cleaves
both RNA and DNA endonucleolytically by Ca2+-mediated
attack of OH at the 5
-position of the phosphodiester
bond, yielding 3
-mononucleotides or -oligonucleotides (18-20). Bases
are bound without sequence specificity within a large (~10 Å wide)
hydrophobic pocket, whereas the scissile and 3
-phosphates make
hydrogen bond contacts to MN near the rim of the pocket. These
properties make MN an ideal probe for exposure of bases in the
transcription complex.
We found that rapid MN cleavage of nontemplate DNA within the
transcription bubble causes RNA polymerase to slide upstream along the
DNA and RNA chains into an arrested state. An arrested RNA polymerase
remains bound to the DNA and nascent transcript but is unable to extend
the RNA chain because its 3-OH has relocated outside of the active
site (4, 8, 21-27). Either active site slippage relative to the RNA or
RNA slippage relative to the active site generates complexes in which
the antiarrest factors GreA or GreB can stimulate internal cleavage of
the RNA chain (28), thus restoring the ability of polymerase to extend
it. We consider the implications of our findings for the mechanism
of arrest under "Discussion."
ApU was purchased from
Sigma; polynucleotide kinase from New England Biolabs;
micrococcal nuclease from Boehringer Mannheim; purified NTPs from
Pharmacia Biotech Inc.; and [-32P]GTP and
[
-32P]ATP from Amersham Corp. RNA polymerase was
purified by the method of Burgess and Jendrisak (29) with the
modifications described previously (28). GreA and GreB were purified
after overproduction from plasmids pDNL279 and pGF103 as described
previously (6, 28).
DNA templates for
transcription reactions were synthesized by polymerase chain reaction
amplification from plasmid pCL185 (28) with primers that hybridized
upstream and downstream from the T7A1 promoter-leader segment: primer
947, 5-GAGAGACAACTTAAAGAGAC (91 bp upstream from the transcription
start site); and primer 1498, 5
-GGTGTTTAAATTTGAACGC (50 bp
downstream). The polymerase chain reaction product is 142 bp long and
yields a 50-nt-long runoff transcript (see Fig. 1).
G16 transcription elongation complexes (halted prior to
addition of U17) were formed by incubation of 25 pmol of
Escherichia coli RNA polymerase holoenzyme
(E70) and 20 pmol of DNA in 50 µl of 20 mM
Tris acetate, pH 8.0, 20 mM NaCl, 20 mM
MgCl2, 14 mM
-mercaptoethanol, 20 µg of
acetylated bovine serum albumin/ml 200 µM ApU, and 10 µM ATP, CTP, and GTP (or [
-32P]GTP for
studying RNA) for 15 min at 37 °C (30, 31).
The G16 complex was purified and exchanged to MN buffer by passing
through a 500-µl Sepharose G-50 spin column equilibrated with MN
buffer (10 mM Tris-HCl, pH 7.5, 20 mM ammonium
acetate, 0.2 mM EDTA, 2 mM -mercaptoethanol,
20 µg of acetylated bovine serum albumin/ml, 2% (v/v) glycerol.
After exchanging to MN buffer, CaCl2 was added to G16
complex to a final concentration of 1 mM. G16 complexes
were then digested with 10 units of MN/ml (final) at 37 °C for the
time desired. Digestions were stopped by adding 3 mM EGTA
(final) to the mixture.
G16 complexes before and after MN digestion were electrophoresed through a 50 × 75-mm 4% NuSieve agarose gel (FMC) cast on the hydrophilic side of Gelbond film (FMC; ~12 ml of 4% agarose in 0.5 × TBE (1 × TBE is 89 mM Tris borate, pH 8.3, 2.5 mM EDTA)). Surface tension suffices to adhere the gel to the plastic. Samples were prepared by addition of glycerol to 10%, heparin to 100 µg/ml, and xylene cyanol and bromphenol blue to 0.05% each. Electrophoresis was performed with the gel submerged in 0.5 × TBE buffer at room temperature and 5 V/cm. After migration of the bromphenol blue to the bottom of the gel (75 mm), the gel was soaked in 0.5 µg of ethidium bromide/ml in water to stain DNA. After photography, the gel was soaked in 0.5% Coomassie Brilliant Blue R (Sigma) in 50% methanol, 10% acetic acid for 60 min and destained by gentle shaking overnight submerged in a covered tray of 20% methanol, 10% acetic acid, also containing a small piece of open cell polyurethane foam. Adherence to Gelbond film is essential to prevent disintegration of the agarose gel, which becomes brittle in the stain and destain solutions.
Recovery, End Labeling, and Sequencing of the MN-digested DNATo recover the MN-digested DNA, the digested G16 complexes were extracted once with phenol-chloroform, the aqueous layer was then extracted once with chloroform, and the DNA was recovered by ethanol precipitation.
To end label the recovered DNA products, 1 pmol of MN-digested DNA (MN
digestion generates 5-OH) was incubated with 15 units of
polynucleotide kinase (New England Biolabs) in buffer provided by the
manufacturer and 2 µl of [
-32P]ATP in a final volume
of 15 µl at 37 °C for 30 min. The end-labeled DNA was purified
away from the free [
-32P]ATP by centrifugation through
a pre-equilibrated 150-µl Sepharose G-10 spin column (Pharmacia) at
500 × g for 5 min. The end-labeled DNA then was mixed
with an equal volume of loading buffer (2 × TBE, 50% (w/v) urea,
0.25% (w/v) each xylene cyanol and bromphenol blue), heated at
95 °C for 5 min, and resolved on a 20% polyacrylamide (19:1
acrylamide:bisacrylamide), 7 M urea, 0.5 × TBE gel.
The gel was then exposed to Kodak XAR-5 film to locate the labeled single strand DNA. The radioactive bands were then excised as small gel
slices and soaked in 1% SDS, 0.5 M ammonium acetate overnight at 37 °C for elution. The supernatants were separated from
the gel slices, extracted with phenol-chloroform, and then extracted to
dryness with 1-butanol. The pellets containing the labeled, single
strand DNAs were washed once with 70% ethanol, dried under vacuum, and
resuspended in TE buffer (10 mM Tris-HCl, pH 8.0, 1 mM EDTA).
Chemical sequencing of purified 5-end-labeled single strand DNAs was
performed as described by Maxam and Gilbert (32). The sequencing
products were then resolved on a 20% polyacrylamide sequencing gel
(19:1 acrylamide:bisacrylamide) and visualized by exposing on Kodak
XAR-5 films.
After MN digestion, MgCl2 was added to the trimmed G16 complex to a final concentration of 7 mM. Then the trimmed G16 complex was elongated in the presence of 600 µM all four NTPs for 1, 5, and 30 min at 37 °C with or without 0.12 µM GreB.
To test whether
the G16 complex is stable after MN digestion, we performed a digestion
time course (Fig. 2). After successive intervals, an aliquot
of the reaction was removed and stopped by addition of EGTA to 3 mM. We then electrophoresed the samples on a 4% agarose
gel (see "Materials and Methods"). After electrophoresis, we
stained the gel successively with ethidium bromide to detect the DNA
(Fig. 2A) and with Coomassie Brilliant Blue to detect the
protein (Fig. 2B). Greater than 90% of the DNA template was incorporated into G16 complex (Fig. 2A, lane 1). A small
amount of the DNA was associated with a slower moving complex, probably the result of loading a second polymerase onto the promoter once the
G16 complex formed (33). The second polymerase was released from the
DNA by 2 min of MN digestion, since only a single MN-trimmed complex
was evident at that point (Fig. 2, A and B, lane
3). MN digestion was complete after 8 min when all the G16 complex
was converted to a faster moving band and most of the ethidium
bromide-stainable DNA was lost (Fig. 2, A and B,
compare lanes 5 and 1).
In contrast, the 16-mer nascent RNA in G16 complexes was completely
resistant to digestion by MN (Fig. 3A, lanes
2-7) but not after it was artificially dissociated
from the complexes and gel-purified (Fig. 3B, lanes 2-9).
Presumably the 16-nt nascent RNA remained within RNA polymerase and was
protected from MN attack. Longer nascent RNAs also were cleaved by MN
and shortened to a 20-24-nt 3-protected fragment (data not shown).
However, the G16 nascent RNA no longer could be extended even after 2 min of MN digestion (Fig. 3A, lanes 8-13), which was well
before the DNA digestion was complete (~8 min, see above). We
conclude that MN digestion trims excess DNA from transcription
complexes and efficiently converts them to arrested complexes.
Fate of nascent RNA in the MN-trimmed G16
complex. A, G16 RNA during and after MN treatment. The RNA
was labeled with [-32P]GTP during formation of the G16
complexes (see "Materials and Methods") and then exposed to MN in
the same samples used to generate the results shown in Fig. 2 (lanes 2-7). After
addition of EGTA to stop MN, a portion of the samples was incubated
with all four NTPs at 400 µM (lanes 8-14).
B, digestion of free G16 RNA with MN. RNA was isolated by
phenol extraction of G16 complexes and ethanol precipitation. The
isolated G16 RNA was incubated for the times indicated without MN
(lane 1) or with MN at 10 units/ml (lanes 2-5)
or 100 units/ml (lanes 6-9). C, treatment of
MN-trimmed complexes with GreB. Complexes were incubated with
increasing concentrations of GreB for 30 min at 37 °C in the absence
(lanes 2-5) or presence (lanes 6-8) of 150 µM NTPs. The assignments of bands to individual RNA
species was made by comparison with GreA-generated cleavage products
(not shown), the sequences of which were reported previously (28). The
presence of one terminal phosphate on the 3
-proximal released RNA
products causes their greater mobility relative to the 5
-proximal
nascent RNAs, which contain OH groups at both ends (11, 28).
To test this interpretation, we treated the G16-arrested complexes with
the transcript cleavage factor GreB (Fig. 3C). At low molar
ratio (Fig. 3C, lanes 2 and 3), GreB principally
cleaved off a 6-nt fragment (Fig. 3C, b), leaving a 10-nt
nascent RNA (Fig. 3C, a). A weaker cleavage released a 5-nt
fragment, producing an 11-nt nascent RNA. At a higher concentration
(Fig. 3C, lanes 4 and 5), GreB shortened these
further to a 7-nt nascent RNA (Fig. 3C, c) with release of
dinucleotides or trinucleotides (Fig. 3C, d). When GreB
treatment was conducted in the presence of NTPs, the same initial
cleavage product was evident, but the nascent RNA was rapidly elongated
to an ~24-nt runoff RNA (Fig. 3C, lanes 6-8). Treatment
of MN-trimmed G16 complexes with GreA yielded similar results, although
large 3-cleavage products were not detected (data not shown). We
conclude that G16 complexes were inactivated, but not dissociated, by
MN digestion.
The inactivated complexes had features typical of arrested
transcription complexes; they could resume elongation only after GreB-induced nascent RNA cleavage, which produced large 3-RNA fragments indicative of relocation of the active site over an internal
phosphodiester bond of the nascent RNA. The 10- and 6-nt fragments
produced suggest that the active site was positioned over the G10-A11
phosphodiester bond after MN treatment. Interestingly, prolonged
incubation at 4 °C of G16 complexes that have not been treated with
MN also leads to arrest and slow, spontaneous cleavage between A11 and
C12 on shifting to 37 °C (28). The correlation of large fragment
cleavage, arrest, and dwell time was discovered subsequent to
publication (data not shown) and has been described by several other
groups (5, 8, 21, 23-25, 36). The 1-nt difference in cleavage
position suggests that MN treatment causes RNA polymerase to slide
backward further than occurs spontaneously.
In addition to differences
in elongation potential and GreB cleavage, MN-trimmed complexes
elongated only to position +24, 6 nt short of the boundary of G16
complexes previously defined with ExoIII (Ref. 1; see Fig. 1). To
determine whether this reflected differences in the protected DNA
strands that remained in the MN-trimmed elongation complex, we analyzed
their sequences. After extraction and recovery, the fragments were
5-end-labeled and resolved on a 20% denaturing gel (see "Materials
and Methods"). We observed nine bands ranging from ~12 to ~28 nt
in size (Fig. 4A, bands a-i). We recovered all
the radioactive bands from the gel and subjected them to chemical
sequencing (see "Materials and Methods"). Comparison of these
sequences with that of the 142-bp DNA template revealed that bands a-e
and g were fragments protected from MN digestion by the G16
transcription complex (Fig. 4B). Minor bands f, h, and i
yielded degenerate sequences that could not be matched to DNA within
the G16 complex footprint (data not shown).
Comparison of the sequences of bands a-e and g with that of DNA within the G16 complex led to four main observations (Fig. 4B). First, bands a-c corresponded to the template strand DNA, and bands d, e, and g corresponded to the nontemplate strand. Second, all the bands shared the same downstream edge (Fig. 4B), indicating that the downstream DNA in the MN-trimmed G16 complex is tightly and uniquely associated with RNA polymerase. Third, the upstream half of the nontemplate DNA in the G16 complex was much more accessible to MN digestion than was the template strand DNA. Cleavage sites in the nontemplate strand DNA were within the transcription bubble (compare Figs. 1 and 4B), suggesting that the upstream half of the nontemplate strand in the bubble is exposed. Fourth, the upstream half of the nontemplate DNA in the transcription bubble could not be recovered from the MN-trimmed G16 complex. It is likely that this part of the DNA was released from the polymerase after MN cleavage and reduced to mononucleotides and dinucleotides that were lost during EtOH precipitation (see "Materials and Methods"). We conclude that the upstream portion of the nontemplate DNA strand is unnecessary for transcription complex stability, in agreement with the findings of Nudler et al. (17) that only downstream duplex DNA, the nascent RNA, and a short segment of the template DNA near the active site are required for stability of the transcription complex.
MN Digestion Caused Upstream Slippage of the G16 Complex on DNA and Transcriptional ArrestA comparison of the G16 complex footprint
determined previously with ExoIII (Fig. 1) with that determined with MN
(Fig. 4B) revealed an upstream shift of the contact of the
polymerase to DNA by 5 or 6 bp with MN treatment. Since MN prefers
single strand DNA to double strand DNA (20), we reasoned that the
exposed upstream half of the nontemplate strand DNA in the
transcription bubble was quickly cleaved by MN. After losing part of
the transcription bubble, the G16 complex collapsed and slid backward
on DNA (Fig. 5). This interpretation is consistent with
backward movement of the active site by 6 nt on the RNA transcript and
with formation of the arrested complex before the flanking DNA is
completely digested around RNA polymerase (compare Figs. 2 and 3). It
also explains why the upstream end of the dominant protected species on
template DNA (Fig. 4, A and B, a) is located ~5
nt upstream from the ExoIII footprint boundary. Polymerase must shift
upstream fast enough to protect these 5 nt. Likewise, the dominant
nontemplate species (Fig. 3, A and B, e) extends
5 nt further upstream than band g, which could also reflect protection
upon rapid upstream movement of the complexes.
We report here that the upstream half of the nontemplate DNA in the transcription bubble of a G16 halted elongation complex is exposed to attack by MN. The RNA transcript, the template strand DNA, and the other half of the nontemplate strand DNA, however, are relatively protected. MN digestion of the exposed part of the nontemplate strand caused the G16 complex to collapse backward by ~6 bp into an arrested state. A 6-nt segment of nascent RNA also appeared to be reverse-threaded through the active site during the collapse, since it was removed on incubation with the GreA or GreB transcript cleavage factors.
The conclusion that the entire RNA polymerase molecule moves backward
relies on an assumption that ExoIII and MN would yield similar
footprints if the complex did not move. We favor this interpretation,
because both edges of the MN footprint are offset upstream by ~6 bp,
so that both the ExoIII and MN footprints are ~28 bp (Fig. 4). It
seems likely that this change reflects polymerase movement, rather than
differential nuclease specificity, because ExoIII, which is double
strand-specific and thus would stop when it encounters the edge of the
single-stranded bubble, digests to +1 of the template strand, whereas
MN, which should attack exposed single strand DNA rapidly,
predominantly cleaves the template strand DNA to only 5.
We will discuss the implications of these findings for three topics: the structure and stability of the transcription complex, the mechanism of transcriptional arrest, and the potential for the nontemplate DNA strand to be targeted by regulators of RNA chain elongation.
Transcription Complex Structure and StabilityConsistent with previous ideas (11, 13), our results suggest that the upstream portion of the nontemplate strand lies on the outside of RNA polymerase and reanneals passively with the template strand as the latter emerges from the transcription complex (Fig. 5). This view also is supported by the weak reactivity of the upstream part of the nontemplate DNA with HO· within the generally protected footprint (7). It seems likely that the phosphate backbone of the nontemplate strand is most closely associated with RNA polymerase, since the bases throughout the OH-protected region are reactive to chemical probes such as KMnO4, dimethylsulfate, and diethylpyrocarbonate (7, 9, 11-13). Most likely, the phosphate backbone anchors the downstream portion of the melted nontemplate strand in a channel large enough to let small molecules attack the bases but that is not accessible to MN (Fig. 5). Rapid attack of the unprotected nontemplate strand by MN, which binds a base within a hydrophobic pocket, suggests that the bases in this segment are directly exposed to solvent, perhaps with the phosphate backbone interacting with positively charged amino acid side chains on the surface of polymerase.
What contribution does the nontemplate strand make to stability of the
transcription complex? Nudler and co-workers (17) recently established
that elongation complexes require contact to only the downstream DNA
duplex and single-stranded template from -1 to +1 to retain function
and salt-stable binding. Hence the nontemplate strand cannot be
essential to prevent dissociation of the complex. However, our results
suggest it may be critical for the lateral stability of the
transcription complex along the RNA and DNA chains. Since melting of a
base pair at one end of the transcription bubble is accompanied by
reannealing of another at the other end, the transcription complex may
be able to occupy multiple positions along the RNA chain, one of which
is competent for elongation and others of which would locate the active
site over an internal phosphodiester bond in the RNA. Based on the ability of both transcript cleavage and pyrophosphorolysis to drive
reverse translocation of RNA polymerase, Feng et al. (28) suggested that the distribution among these possible states would be
determined by their relative stabilities and the kinetic barriers to
interconversion, with the active state normally dominant. The slipped
states would arise depending on sequences of the chains and dwell time
at a particular RNA 3-end. This idea is consistent with the
observation that RNA polymerase sometimes enters unactivated states in
kinetic analyses of chain elongation (34) and misincorporation (35). In
this view, rapid reverse slipping of polymerase upon removal of the
exposed nontemplate DNA strand by MN is most easily explained because
the remaining nontemplate strand can re-pair with the template strand
and increase the thermodynamic stability of the slipped state relative
to the original G16 position (Fig. 5).
A current model for transcriptional arrest is that relocation of the active site over an internal phosphodiester bond in the nascent RNA arises when part or all of the RNA polymerase slips backward along the DNA template, blocking subsequent nucleotide addition until transcript cleavage occurs (4, 8, 21-24, 26, 27). Reports that the RNA cleavage interval increases on arrest (5, 8, 21, 23-25) and that cleavage causes upstream shifting of the footprint of RNA polymerase on DNA in a nonarrested complex (13), but not in an arrested transcription complex (22), all favor this model. Here we find that an upstream shift in both edges of the DNA footprint of the polymerase is associated with arrest. Our results thus strongly support the idea that backward translocation of RNA polymerase along DNA and RNA chains can trigger transcriptional arrest. In this case slippage and arrest are caused by nontemplate strand removal, but similar backward slippage on transcriptional arrest also has been observed without MN treatment for both E. coli RNA polymerase (36) and mammalian RNA polymerase II (37).
A key question now is how these slippage events are related to the
apparent discontinuous changes in the RNA and DNA footprints of RNA
polymerase (so-called "inchworming") that have been found to be
associated with pausing, arrest, and termination (5, 6, 8). Flexibility
in RNA polymerase on encountering certain DNA or RNA sequences could
lead to dislocation of the RNA 3-end from the active site of the
polymerase and subsequent sliding of the enzyme. Alternatively, sliding
of RNA polymerase along the RNA and DNA chains when complexes are
halted for study near regulatory sites may generate the inchworm
footprints (36, 37), reflecting a change in the energetic state of the
transcription complex that foreshadows pausing, arrest, or
termination.
The
upstream half of the nontemplate strand recently has been found to be
the target of a noncanonical activity of the E. coli
70 initiation factor (38): induction of pausing at
position 16 of the
late transcription unit (39). At least in these
complexes, the nontemplate strand is exposed in a way that allows
sequence-specific recognition by
70. Thus, as in our
finding that MN can digest this portion of nontemplate DNA (which
appears to require outward orientation of the bases),
70-induced pausing suggests that nontemplate bases are
exposed to potential interactions with regulatory factors. These
findings raise the speculative possibility that some factors that
regulate transcript elongation could target the nontemplate DNA strand, inhibiting or promoting either forward or reverse translocation of the
enzyme, and thus favoring pausing, arrest, termination, or rapid
transcription, by influencing availability of the nontemplate DNA
strand for re-pairing with the template DNA strand upon
translocation.
We thank Irina Artsimovitch for helpful comments on the manuscript and Mikhail Kashlev and Diane Hawley for sharing results prior to publication.