(Received for publication, October 11, 1996, and in revised form, January 16, 1997)
From the Laboratory of Molecular Genetics, Box 174, The Rockefeller University, New York, New York 10021
We have discovered that T7 RNA polymerase,
purified to apparent homogeneity from overexpressing Escherichia
coli cells, possesses a DNase and an RNase activity. Mutations in
the active center of T7 RNA polymerase abolished or greatly decreased
the nuclease activity. This nuclease activity is specific for
single-stranded DNA and RNA oligonucleotides and does not manifest on
double-stranded DNAs. Under the conditions of promoter-driven
transcription on double-stranded DNA, no nuclease activity was
observed. The nuclease attacks DNA oligonucleotides in mono- or
dinucleotide steps. The nuclease is a 3 to 5
exonuclease leaving a
3
-OH end, and it degrades DNA oligonucleotides to a minimum size of 3 to 5 nucleotides. It is completely dependent on Mg2+. The
T7 RNA polymerase-nuclease is inhibited by T7 lysozyme and heparin,
although not completely. In the presence of rNTPs, the nuclease
activity is suppressed but an unusual 3
-end-initiated polymerase
activity is unmasked. RNA from isolated pre-elongation and elongation
complexes arrested by a psoralen roadblock or naturally paused at the
3
-end of an oligonucleotide template exhibited evidence of nuclease
activity. The nuclease activity of T7 RNA polymerase is unrelated to
pyrophosphorolysis. We propose that the nuclease of T7 RNA polymerase
acts only in arrested or paused elongation complexes, and that in
combination with the unusual 3
-end polymerizing activity, causes
heterogeneity in elongation complexes. Additionally, during normal
transcription elongation, the kinetic balance between nuclease and
polymerase is shifted in favor of polymerase.
Transcription by procaryotic DNA-dependent RNA polymerases can be represented as follows (Scheme I).
![]() |
![]() |
![]() |
![]() |
![]() |
Initiation involves binding of RNAP holoenzyme to promoter DNA (RPc) and the isomerization to open complexes (RPo) (for a review, see Ref. 1). Open complexes synthesize short RNAs (up to 10 nts) during abortive initiation. After clearing the promoter, RNAP enters the elongation phase. Termination occurs at certain DNA sequences in either a factor-dependent or a factor-independent manner (2). The current view of transcription elongation has been possible because of our ability to arrest elongation at specific sites on DNA templates, partial purification of arrested complexes, and the enzymatic and chemical probing of their structures (3-17).
The unexpected discovery of an RNA cleavage reaction in arrested Escherichia coli complexes (18) was followed by documentation of analogous cleavage reactions in other RNAPs (19-23). E. coli RNAP and eucaryotic RNAP II are capable of RNA cleavage in binary and ternary complexes (24-26). GreA and GreB of E. coli enhance the intrinsic cleavage by E. coli RNAP (25, 27, 28). GreA and GreB (and a eucaryotic counterpart, SII) prevent elongation arrest (29). GreA may participate in the fidelity of RNA synthesis (30).
T7 RNAP (98.8 kDa) belongs to a class of single-subunit RNAPs that
includes T3 and SP6 phage RNAPs (31-33). The three-dimensional structure of T7 RNAP shows high -helicity with a deep cleft. Using a
novel photochemical cross-linking technique, we have identified the T7
RNAP cleft as the site of promoter binding (34). The polymerase shows a
striking structural similarity to E. coli DNA polymerase
Klenow fragment and HIV reverse transcriptase (35). In
vitro, T7 RNAP transcribes DNA without additional protein factors. Promoter complexes have been characterized by footprinting and low
resolution NMR (36, 37). T7 RNAP has a higher rate of elongation
(~250 nts/s) than E. coli RNAP (40-50 nts/s). Elongation complexes are remarkably resilient in their elongating phase. T7 RNAP
can transcribe past noncovalently bound proteins (38), DNA triple
helices (39), and some base-specific adducts (40), whereas other
roadblocks such as psoralens (3, 4), acetoaminofluorine (40), and
Z-DNA antibodies (41) can temporarily block elongation. RNA cleavage by
bacteriophage T7-like RNAPs has not been demonstrated so far. Part of
the problem has been the difficulty of isolating partially purified T7
RNAP elongation complexes in a stable register.
Here, we show that T7 RNAP also exhibits a novel DNA and RNA cleaving
property. RNase activity occurs in ternary elongation complexes and in
binary complexes. In combination with an unusual 3-end-initiated
polymerase activity, the RNase activity produces heterogeneity in
stalled/arrested T7 RNAP elongation complexes.
DNA and RNA oligonucleotides
were commercially synthesized by automated solid phase procedures. DNAs
were purchased from Midland Certified Reagent Co. (Midland, TX), and
RNAs were purchased from the Rockefeller University biotechnology
facility. The plasmids pBluescript and pBR322 were prepared using
standard protocols (42). The concentrations of nucleic acids were
calculated from their respective molar extinction coefficients at 260 nm (~104
M1·cm
1/nt). rNTPs and dNTPs
were purchased from Pharmacia Biotech Inc. at a concentration of 100 mM. T4 polynucleotide kinase and calf terminal transferase
were from New England Biolabs and Boerhinger Mannheim, respectively.
Heparin sulfate was from Sigma. T7 lysozyme and
anti-T7 RNAP antiserum were a kind gift from Dr. F. W. Studier (Brookhaven National Laboratory, Upton, NY).
T7 RNAP was prepared locally according to
the procedure of Grodberg and Dunn (43) and also by a procedure that
was modified from Ref. 44. We also used preparations of T7 RNAP that
were prepared by a procedure from the Uhlenbeck laboratory (45). His-tagged T7 RNAP was overexpressed from a plasmid (pBH161) in E. coli cells (BL21). The enzyme was purified using a
Ni2+-agarose column. The bacterial strains overexpressing
the His-tagged T7 RNAP, the strains containing the insertion mutations
in the T7 RNAP structural gene, and the procedure for purification of His-tagged T7 RNAP were a generous gift from Dr. Bill McAllister (State
University of New York at Brooklyn). Other sources of T7 RNAP are
mentioned under "Results." The concentration of the purified enzyme
was determined using 280 = 1.4 ± 0.1 × 105 (43). For nuclease assays, 2 pmol of 5
- or
3
-end-labeled DNAs or RNAs were mixed with 5-10 pmol of T7 RNAP in a
25-µl reaction containing transcription buffer (50 mM
Tris-HCl, pH 8.0, 10 mM MgCl2, 1 mM
dithiothreitol, 1 mM spermidine, 5% (v/v) glycerol). The
reaction mixture was incubated at 37 °C for various lengths of time
during which 2- or 4-µl aliquots of the reaction were taken out at
different times and mixed with 4 or 10 µl of 8 M urea-Tris borate, 20 mM EDTA plus tracking dyes and heated
in a boiling water bath for 5 min and run on a 24% acrylamide, 8 M urea gel (19:1 acrylamide:bisacrylamide). The cleavage
reaction was visualized by autoradiography with an x-ray film or
phosphor screen. Where necessary, the kinetics of the cleavage reaction was measured by quantitation of the residual full-length 23-mer DNA
band or 20-mer RNA band at specific intervals relative to the band
representing the starting material, i.e. the 0-min time point. Semiquantitative data were obtained using the ImageQuant program
of a PhosphorImager.
Two templates were
prepared, one unmodified and the other with a psoralen cross-link
(between T at the +36 position on the top strand and T at the +37
position of the bottom strand (66XL); see Fig. 7). The 66XL template
was prepared as described previously (Ref. 3; see Ref. 46 for methods
of adduct preparation). Briefly, a 14-mer furanside monoadduct
(CGAAGCTACGAGCA) was ligated to a synthetic, kinased 52-mer
(5-GAGGCCATCGATAAAGGTCTAGATCTCCCTATAGTGAGTCGTATTAATTAGC-3
) and a
13-mer (5
-GGCCTCTGCTCGT-3
). The 13-mer DNA served as a ligation
"bridge" between the 14-mer and the 52-mer. An equimolar amount of
a synthetic 66-mer nontemplate strand was then added and the 66-mer
furanside monoadduct bottom strand (the ligation product) was
cross-linked to the nontemplate strand with 320-380 nm light. The 66XL
was then purified from preparative denaturing gels.
Analysis of Elongation Complexes
Unmodified 66-mer ds
template was constructed from two complementary strands that were
synthesized individually and purified by high performance liquid
chromatography. The sequence of the template is the same as the one
previously used (see Fig. 7 for sequence) (3, 4). To construct an
unmodified 66-mer template equimolar amounts of a mixture of the two
complementary strands (1 µM each) were heated at 70 °C
for 10 min and slowly cooled to room temperature over a period of
2 h. When needed, the template strand was 32P-labeled
with the aid of [-32P]ATP and T4 polynucleotide kinase
(42). Psoralen cross-linked templates were constructed as described
above. The psoralen site specifically cross-linked the T at the +36
position on the top strand with the T at the +37 position on the bottom
strand (see Fig. 7). Typical transcription reactions were carried out
by first mixing ~80 pmol of template DNA (either
32P-labeled at the 5
-end of the template strand or
unlabeled) in transcription buffer containing 1 mM each of
cold ATP, GTP, and UTP, cold CTP at 50 µM,
[
-32P]CTP (specific activity, 3000 Ci/mmol) at 0.3 µM, and 60 units of human placental RNase inhibitor.
Transcription was initiated by the addition of 400 pmol of T7 RNAP and
incubation at 37 °C. After 15 min of transcription, 300 mM NaCl or 90 µg/ml heparin was added. When necessary,
sodium pyrophosphate (Sigma) was added at a final concentration of 2 mM after the addition of NaCl. After 3 min at room
temperature, the entire reaction was passed through a Sephadex G-25
spin column. The flow-through was concentrated by SpeedVac
centrifugation at room temperature (~10 min) to about 50-100 µl,
and glycerol was added to 5% final concentration. The reaction was
loaded on an 8% acrylamide nondenaturing gel (14 cm × 16 cm) and
run at 10 V/cm. For some reactions, the spin column step was eliminated
to achieve a better yield of the complexes. The gel was run until the
blank xylene cyanol dye was at the bottom of the gel. C1 and C2 bands
were identified by autoradiography of the wet gel, and the respective
complexes were eluted from gel pieces by electroelution. The nucleic
acids were recovered by precipitation with EtOH and run on a 24%
acrylamide, 8 M urea-TBE gel. The sizes of the RNAs were
measured using markers generated by diethyl pyrocarbonate plus
piperidine (purines) cleavage of DNA oligonucleotides whose sequences
are known. Additional markers were a synthetic DNA ladder ranging from
8 to 75 nts. RNA migrated about 2-2.5 nts slower than DNA of the same
length on 24% denaturing acrylamide gels.
Fig. 1
shows that purified T7 RNAP from overexpressing E. coli
cells cleaved a 5-32P end-labeled 23-mer ss DNA
(5
-TAATACGACTCACTATAGGGAAG-3
, promoter top strand) and a 20-mer RNA
(5
-UUUUUUUUUUCUGACUUAGC-3
). The cleavage occurred from the 3
-end
(see below) in steps of mono- and dinucleotides (Fig. 1, lanes
2-8). After 60 min of DNA digestion, major products of 2 or 3 nts
were seen. The pattern of DNA and RNA fragments was the same, except
that the DNA oligonucleotide was cleaved to smaller products faster
than the RNA oligonucleotide. The 20-mer RNA migrates about 2-2.5 nts
slower compared with the same-sized DNA. Hence, in Fig. 1, the RNA and
DNA bands appear to migrate approximately at the same positions on the
high resolution gel (24% acrylamide). To rule out the possibility that
the nuclease activity was due to an adventitious contaminant in the
enzyme preparations, we purified mutant T7 RNAPs containing insertion and point mutations in the structural gene of T7 RNAP. These mutants were cloned in Dr. McAllister's laboratory (see Refs. 47 and 48 for
details). We prepared T7 RNAPs from four mutant strains containing
plasmids pLG12, pLLG22, pWJC22, and pCAR27. All four mutant T7 RNAPs
were defective in DNA and RNA cleavage to different extents. In Fig. 1
we show an example of a cleavage pattern that was observed with the
pLG12 T7 RNAP insertion mutant (MUT). This mutant had a
6-base pair linker insertion within or immediately after codon 566 in
bacteriophage T7 gene 1 (the structural gene for T7 RNAP) (47).
This mutant was defective in promoter binding and catalysis (47). In
our assay, the mutant enzyme was defective in both DNA and RNA cleavage
(Fig. 1). Surprisingly, the mutant T7 RNAP behaved somewhat differently
with DNA and RNA oligonucleotides. Whereas DNA cleavage occurred to
some extent, albeit at a slower rate than with wild type T7 RNAP, RNA
cleavage was almost completely absent (Fig. 1). For example, with the
mutant T7 RNAP, after 60 min of incubation with ss DNA, bands extending
up to 17-mer were visible, the most intense bands being the 11-mer and
12-mer (Fig. 1, MUT, lane 8). Whereas with wild
type T7 RNAP, at 60 min a much smaller distribution was observed (Fig.
1, WT, lane 8). This experiment showed that 1)
the nuclease activity is indeed due to the polymerase itself; 2)
because the overall pattern of cleavage of both ss RNA and ss DNA
appears to be the same, a similar mechanism of cleavage probably
occurred; 3) the cleavage rates are different for ss DNA
versus ss RNA for both wild type and mutant, perhaps due to
differential affinities of the polymerase for ss DNA versus ss RNA (indeed, using gel-shift assays, we found that T7 RNAP binds ss
DNA better than ss RNA; not shown); and 4) because the mutant T7 RNAP
is defective in promoter binding and catalysis (47) and is also
defective in our cleavage assay here, cleavage is probably catalyzed by
the active center of the enzyme.
Four other T7 RNAP preparations were also tested for nuclease (data not
shown): T7 RNAP that was prepared using the procedure described in Ref.
45; His-tagged T7 RNAP that was prepared using a
Ni2+-agarose column (bacterial strains and procedure from
Dr. W. T. McAllister); T7 RNAP that was purchased from Epicenter
Technologies (Madison, WI) (43) and modified by Dr. J. J. Jendrisak
(Epicenter Technologies); and T7 RNAP from Dr. J. J. Dunn (Brookhaven
National Laboratory, Upton, NY) (43). All these purified preparations displayed the same patterns of cleavage as in Fig. 1. Cleavage activity
also occurred on a 66-mer ss DNA or a 66-mer ds DNA with a 3
nucleotide overhang (not shown). All these oligonucleotides contained a
T7 RNAP promoter sequence. Nuclease assays using a 5
32P-end-labeled 34-mer ss DNA without promoter sequence
(5
-CGAAGCTACGAGCGGTAGCCATCGATAAATAGCT-3
) gave the same result
(not shown). Therefore, promoter sequence is not required for T7 RNAP
nuclease activity.
To further rule out the possibility of contaminating nuclease in our T7
RNAP preparations we carried out the following experiments. SDS-polyacrylamide gel electrophoresis showed the presence of a single
band migrating at ~99 kDa after staining with Coomassie Brilliant
Blue (Fig. 2A) or with silver. We estimated
that these preparations were >90-95% pure T7 RNAP. When T7 RNAP was
run on a native (i.e. without SDS) polyacrylamide gel only
two distinct bands were seen after Coomassie Brilliant Blue staining
(Fig. 2B). A multimeric complex (MC) was observed. Since T7
RNAP is known to form aggregates in low
salt,2 it is possible that a large fraction
of T7 RNAP migrates close to a 669-kDa marker apparently as a multimer
in native gels containing low concentrations of salt. However, it is
difficult to assign a mass to this multimer based solely on native
gels. Hence, we shall simply refer to this as a MC. A second, fainter
band, apparently corresponding to a monomer (Fig. 2B,
M), based on its electrophoretic mobility, was also
observed. Because this was a native gel (i.e. without SDS),
we are not certain whether band M represented the uncleaved ~99-kDa
T7 RNAP molecule or the "nicked" 80-kDa fragment that is often seen
in polymerase preparations (43). There was no visible amount of the
80-kDa fragment in our T7 RNAP in SDS gels (Fig. 2A). This
is because we prepared T7 RNAPs from E. coli strain BL21
(without Omp T protease; see Ref. 43). Since the same T7 RNAP
preparation was run on SDS and native gels, and because the ~20 kDa
(~99-80 kDa) was not seen on native gels, we believe that the lower
band probably represents a monomer of T7 RNAP. However, for the
interpretation of the data below, whether band M represents the
full-length monomer or the nicked polymerase is not relevant because
the material in band M did not bind DNA.
To assay for DNA binding we soaked a gel identical to that in Fig.
2B in nuclease assay buffer containing 5
32P-end-labeled 23-mer DNA. Only the MC band showed a
prominent 32P signal (Fig. 2C). Overexposed
autoradiograms (not shown) of the gel did not reveal DNA binding to
band M. The band representing MC was excised from the gel, and the DNA
was eluted and subsequently run on a denaturing 24% acrylamide, 8 M urea gel (Fig. 2D, lane 2). It is
clear that cleavage of the 23-mer DNA occurred in the T7 RNAP complex.
The in situ cleavage pattern is qualitatively similar to the
cleavage pattern in solution (compare Fig. 1, WT, lanes 2 and 3, with Fig. 2D,
lane 2). These experiments show that T7 RNAP may exist as a
multimer that cleaves 32P ss DNA. Next, we did a corollary
experiment to demonstrate that T7 RNAP indeed binds to the ss DNA
oligonucleotide in stoichiometric amounts. Fig. 3 is a
gel-shift assay demonstrating that increasing amounts of T7 RNAP bind
to the 23-nt ss DNA. A single major upper band (U) and two
faster migrating minor bands (L) are seen (Fig. 3). The
faster migrating shifted bands may represent the 5
cleavage products
that are still bound to the enzyme, consistent with the results from
the in situ cleavage experiment. Alternatively, they may be
faster migrating conformational isomers of the enzyme-ss DNA complexes.
These experiments demonstrate that T7 RNAP binds and cleaves ss DNA in
a stepwise manner reminiscent of a 3
to 5
exonuclease. Gel-shift
assays with 5
32P-end-labeled 20-mer RNA also showed that
T7 RNAP was bound to RNA in binary complexes (not shown).
Additional criteria were used to show that the cleavage activity is a
property of T7 RNAP. 1) The kinetics of heat inactivation of T7 RNAP at
70 °C indicated that the destruction of nuclease activity paralleled
the destruction of transcription activity (not shown). 2) Amino acid
composition analyses of two of our preparations matched the known amino
acid composition of T7 RNAP. 3) Western blotting of gel shifts of
binary complexes of T7 RNAP and 23-nt DNA (such as those in Fig. 3) and
transcription complexes (see Fig. 8) with anti-T7 RNAP antibody
indicated that these complexes contained T7 RNAP. 4) DNA cleavage
activity during T7 RNAP purification coincided with T7 RNAP peak
fractions from the cation-exchange column.
ss DNA Cleavage Activity Starts at the 3
When the 23-mer ss DNA was
32P-labeled at the 3-end, the nuclease immediately
released a mononucleotide (Fig. 4). This was deduced
using 32P-labeled dinucleotide ([
-32P]GpG)
and 32P-labeled ATP as markers on a 24% acrylamide
denaturing gel. The product size remained the same even after prolonged
incubation. This indicated that the initial product of the cleavage is
probably a mononucleotide. (Within the scope of the present paper we do not think it is necessary to unambiguously identify this product.) The
same result was seen with 3
32P-end-labeled 34-mer
nonpromoter ss oligonucleotide DNA (not shown). Putting together the
5
- and 3
-end labeling experiments, T7 RNAP nuclease acts by a
stepwise removal of nucleotides starting from the 3
-end and going
toward the 5
-end with the limit product at the 5
-end being 3-5
nucleotides. That cleavage left a 3
-OH end was verified by
32P labeling the cleaved unlabeled 3-mer and 20-mer with 3
terminal transferase and [
-32P]ddATP after the nuclease
assay. Omission of MgCl2 or inclusion of molar equivalents
of EDTA with Mg2+ blocked cleavage, indicating complete
Mg2+ dependence of the nuclease (not shown).
Nuclease Activity Is Inhibited by T7 Lysozyme or Heparin
A
20× excess (w/w) of T7 lysozyme over T7 RNAP or heparin (250 µg/ml)
inhibited the rate of nuclease activity (Fig. 5). After 5-30 min of incubation with T7 lysozyme, 21-mer and 20-mer or 10-12-mers were still seen, whereas without lysozyme the DNA was cleaved faster to much shorter products. Prolonged incubation with
these reagents showed smaller products, indicating that these reagents
did not completely block nuclease activity. The inhibition by T7
lysozyme or heparin was concentration-dependent (not shown). Half-maximal inhibition by heparin was at 50 µg/ml. T7 lysozyme is a
specific inhibitor of T7 RNAP (49, 50), again indicating that T7 RNAP
is the source of nuclease activity. Because the polyanion heparin is a
competitive inhibitor of DNA binding, DNA binding is a prerequisite for
T7 RNAP nuclease activity.
Polymerase Activity Suppresses Nuclease Activity
Addition of
rNTPs suppresses the nuclease activity while unmasking an unusual
polymerase activity (compare lanes without NTPs and with rNTPs, Fig.
6). The 3-end extension of DNA (or RNA, not shown) by
T7 RNAP with rNTPs has been observed previously (51-53). Prolonged
incubation in the presence of rNTPs shows evidence of nuclease activity
on the 40-mer DNA-RNA chimeras (e.g. see 30-min lane of Fig.
6). Consistent with T7 RNAP being the nuclease, dNTPs do not elicit the
same effect as rNTPs (Fig. 6). This showed that the active site of T7
RNAP is involved in nuclease activity, consistent with our earlier
results with mutant T7 RNAPs (Fig. 1). Fig. 6 shows that nuclease
activity and polymerase activity can manifest independently. In the
presence of NTPs, polymerase activity predominates over nuclease
activity (see below).
Nuclease Activity Is Absent during Normal Transcription
The
plasmid pBluescript (Stratagene Corp., San Diego, CA) contains a class
III promoter but not a T7 terminator. In the presence of T7 RNAP and
rNTPs, a large amount of RNA was produced by multiple rounds of
transcription, indicating that our T7 RNAP was transcriptionally active. No small cleavage products originating from the DNA were seen.
In the absence of NTPs, no cleavage of the plasmid was seen either
(data not shown). To confirm that no cleavage of the plasmid occurred,
we isolated the DNA from gel bands from agarose gel and restricted the
DNA with HindIII and PstI separately.
Re-examining the restricted DNA separately on agarose and denaturing
polyacrylamide gels revealed no indication that T7 RNAP cleaved the
plasmid with or without NTPs. We observed similar results with the
plasmid pBR322 that has no consensus T7 promoter, although in this case very little RNA was made. We examined transcription initiation on a
short oligonucleotide template containing a 23-base pair core promoter
sequence (5-TAATACGACTCACTATAGGGAAG-3
). In the presence of GTP as the
sole NTP, T7 RNAP produces a G-ladder that extends to ~14 nts and
then tapers off (71). The G-ladder is thought to be produced by
multiple rounds of slippage synthesis at the three cytidines in the
promoter (54). The G-ladder accumulated in a time-dependent
manner, indicating that the ds template was not destroyed and the RNA
was not degraded (71). We also examined transcription on an unmodified
66-mer DNA template (71). During abortive transcription, there was a
time-dependent accumulation of transcripts but no evidence
of template cleavage (71). These observations indicate that in the
presence of NTPs during normal transcription no T7 RNAP DNase activity
is seen.
Next, we examined RNA
cleavage activity in isolated ternary elongation complexes. Because T7
RNAP elongation complexes are relatively unstable (dwell times of
~5-15 min) we were not able to "walk" discrete and isolated T7
RNAP elongation complexes in a manner that was done with E. coli and eucaryotic RNAP elongation complexes. To circumvent these
problems, an alternative protocol (albeit a less "clean" strategy
compared with traditional walking methods) to examine the RNA component
in elongation complexes was developed. Transcription was initiated by
the addition of T7 RNAP to a synthetic oligonucleotide template in the
presence of NTPs. Following 15 min of transcription, NaCl (to 300 mM) or heparin was added. Addition of high salt or heparin
disrupted all existing promoter complexes and inhibited new rounds of
initiation. Footprinting and other experiments have shown that 300 mM NaCl disrupted promoter-initiation complexes but not
elongation complexes (3, 4, 55-57). To isolate stable complexes, we
first passed the transcription reactions through a Sephadex G-25 spin
column to remove NTPs and other low molecular weight components. The radiolabeled complexes were resolved on a low percentage nondenaturing acrylamide gel. The 32P label was present either in the
nascent RNA or on the 5-end of the template strand. To ensure that the
resolved complexes seen on nondenaturing gels contained T7 RNAP, we
probed the complexes with anti-T7 RNAP antiserum using Western blotting
techniques (not shown). The same bands that were visible in
autoradiograms when the RNA or the DNA templates were labeled with
32P were also recognized by anti-T7 RNAP antibody,
attesting that the complexes indeed contained T7 RNAP. Omission of the
spin column step increased the yield of the complexes without changing
the pattern of bands on the nondenaturing gel. Following
autoradiography of the native gels, the radioactive RNAs were recovered
and subsequently resolved on a 24% denaturing acrylamide gel (see
"Experimental Procedures"). This protocol selected high salt or
heparin-resistant stable elongation complexes.
We examined the RNA composition in two different T7 RNAP ternary
complexes: 1) elongation complexes apparently paused at the end of an
unmodified 66-mer ds DNA template; and 2) elongation complexes arrested
by a psoralen cross-link site specifically placed toward the 3-end of
the same 66-mer (66XL) template (3).
Fig. 7 shows the sequence of the 66-mer DNA for examining end-paused and psoralen-arrested complexes. Footprinting demonstrated that elongating T7 RNAP, as indicated by the box in Fig. 7, was blocked by the psoralen cross-link (3). Here, we made the unmodified as well as the psoralen cross-linked template to study cleavage and end-polymerization reactions. On the unmodified 66-mer template, the elongation complexes paused at the end of the template (end-paused complexes). The position of the polymerase molecule in end-paused complexes is suggested by the box in Fig. 7. Although we have not footprinted the end-paused elongation complexes, we assume that these complexes occupied a similar overall position at the end of the template as the psoralen-arrested complex (Fig. 7). This assumption is justified because the majority of the RNA in the isolated complexes is a runoff transcript (43 nts; Fig. 8C).
Using the protocol described in the previous section, we examined the 32P-labeled RNA in end-paused and psoralen-blocked complexes. Fig. 8A (with unmodified template) and 8B (with cross-linked template) show complex 1 and complex 2 with different mobilities. Identical bands were seen when only the DNA was labeled (not shown). From the specific radioactivity of [32P]RNA or [32P]DNA in gel bands seen in the absence of high salt, we estimated that complexes contained approximately an equimolar ratio of DNA and RNA. Two bands were seen after addition of 300 mM NaCl, which was added after 15 min of transcription. Only elongation complexes were stable during gel electrophoresis, and they have different mobilities on nondenaturing gels. When heparin was added, only complex 2 was visible. This is in agreement with previous observations that heparin caused disruption of some elongation complexes (4, 58). The ratio of [32P]CMP-labeled C1 to C2 with 66XL was almost equal (Fig. 8B), whereas with unmodified 66-mer ds DNA template, C1 accounted for >90% of the total radioactivity (Fig. 8A). Clearly, the presence of the psoralen cross-link influenced the relative populations of C1 and C2. The basis for this difference is unknown.
Examination of 32P-labeled RNA in NaCl- (C1) or
heparin-resistant (C2) complexes revealed a heterogeneous pattern of
transcripts (Fig. 8C). In C1, the major cleavage products
were of 12, 17, 20, 26, and 31 nts, etc., and were interspersed with
minor products with 1- or 2-nt differences. Longer than +1-initiated
transcripts (up to ~60-mers) were also seen. These transcripts were
not due to end initiation (or internal initiations at sites other than +1) because they were observed even when [-32P]GpG was
used, which forces T7 RNAP to correctly initiate at +1 (not shown). The
12-nt transcripts are not abortive transcripts because the latter are
usually only up to 10 nts and are generally more abundant relative to
the longer RNAs. Moreover, just after addition of NaCl or heparin and
prior to passage through the spin column, most of the complexes were
end-paused, as indicated by the abundance of the 43-nt runoff species
(Fig. 8C, lane RF). We believe that the
heterogeneity is due to RNA cleavage and end-polymerization reactions
analogous to those seen earlier (see Figs. 1 and 6). Heterogeneity
cannot be due simply to the trapping of polymerase complexes in various
registers of elongation by the spin column/gel. This would have
produced a more or less stochastic distribution of RNAs. There are
other compelling reasons to eliminate the latter explanation. 1)
Addition of 300 mM NaCl or heparin precluded any further
polymerase binding after 15 min of transcription. 2) Even in
heparin-resistant complexes (C2 in Fig. 8C) the cleavage and end-extended RNAs are observed. In fact, more or less the same pattern
(down to 22 nts) is seen with both NaCl and heparin (because of the
lower yield of heparin-resistant complexes, the band intensities were lower). 3) Furthermore, examination of the transcription products
before spin column/gel purification (Fig. 8C, lane
RF) showed RNAs that were overwhelmingly in the larger sizes
(35-43 nts), indicating that just before spin column passage most of the complexes in solution were close to the end of the template. Because the NTPs were removed after the passage through the spin column
gel, RNA cleavage and end polymerization must have occurred prior to
the spin column step and/or before native gel electrophoresis. The high
salt- or heparin-resistant heterogeneous complexes were trapped by the
gel. We believe the heterogeneity of transcripts is due to a fraction
of the end-paused complexes undergoing end-polymerization and cleavage
reactions.
Next, we examined the RNA composition of T7 RNAP
complexes arrested by an authentic roadblock (psoralen cross-link) (3, 56). Fig. 8D shows that the patterns of cleavage of nascent RNA transcripts from C1 and C2 complexes is quite different from the
end-paused ones on noncross-linked template (Fig. 8C).
Surprisingly, RNA transcripts were extremely heterogeneous in both C1
and C2, occurring in a wide range in 1- or 2-nt steps (11-28 nts; Fig. 8D). The pattern of cleavage is reminiscent of that in Fig.
1. Transcripts extending up to ~75 nts were seen in abundance in C2.
The 36-mers expected from transcription up to the psoralen cross-link
were only a minor species. Longer RNAs than those specified by +1
initiation were due to RNA 3-end polymerization, whereas the shorter
RNAs were due to cleavage. Pyrophosphorolysis is the chemical reversal
of nucleotide incorporation by nucleic acid polymerases.
Pyrophosphorolysis is usually accomplished in vitro by
millimolar concentrations of PPi (13, 23, 59-61). We did not observe any significant differences in the pattern of transcripts from psoralen-arrested elongation C1 and C2 complexes following addition of PPi, suggesting that PPi has no
effect on RNA cleavage. In general the yield of complexes was lower in
the presence of NaPPi (data not shown).
We have discovered that T7 RNAP
prepared from overexpressing E. coli cells exhibits DNase
and RNase activity. This is a new enzymatic activity of T7 RNAP. Many
experiments have ruled out the probability of other contaminating
nucleases in our T7 RNAP preparations (see "Results"). Moreover, T7
RNAP prepared using different methods and from different sources
exhibited nuclease activity. We cannot completely rule out the
possibility that a low molecular weight protein factor (analogous to
GreA) is tightly bound to T7 RNAP. It is, however, highly unlikely that
there was an adventitious contaminant nuclease unrelated to T7 RNAP.
Hydrolytic cleavage of RNAs has been documented with all tested RNAPs.
These include E. coli RNAP, higher eucaryotic pol II and pol
III, vaccinia RNAP, and yeast RNAP (see Introduction for Refs.). In at
least two cases (E. coli and yeast pol II) the cleavage of
RNA has been demonstrated in binary complexes as well as in ternary
complexes. Here, both binary and ternary complexes of T7 RNAP exhibit
nuclease activity. The DNase activity of T7 RNAP leaves a 3-OH end and a 5
-PO4 and is absolutely dependent on Mg2+.
The nuclease appears to be a 3
to 5
exonuclease on single-stranded oligonucleotide. Although we have not isolated and analyzed the first
cleavage product to the 3
-end, based on the pattern of cleavage of
5
-end- or 3
-end-labeled DNA and RNA we can only surmise that the
cleavage occurs in mono- or dinucleotide steps (Figs. 1 and 4). The 5
cleavage products appear to be bound to the enzyme as indicated by the
results from the in situ cleavage assay (Fig. 2). Multiple
T7 RNAP molecules may be involved in the cleavage of each DNA
(distributive mechanism) as indicated by the binding stoichiometry
(Fig. 3) and by the fact that only the multimeric form of the T7 RNAP
showed binding and cleavage of oligonucleotide DNA (Fig. 2). Since T7
RNAP nuclease cannot act on covalently closed circular plasmid DNA or
linear oligonucleotide ds DNA, it is not an endonuclease. We could not
detect T7 RNAP nuclease activity on linear blunt ended ds DNA
oligonucleotides in the absence of transcription. The time course of T7
RNAP RNase activity is slower than that of the DNase. We believe that
DNase and RNase activities are carried out by the same catalytic pocket because the cleavage ladders from a 5
-end-labeled DNA or RNA are of
the same pattern, in both cases polymerase activity competes with and
suppresses nuclease activity, and T7 RNAP mutants are defective in
cleavage. It is possible that the same binding site can accommodate
both ss RNA and ss DNA. This proposal is consistent with the ability of
T7 RNAP to strongly bind ss DNA and ss RNA oligonucleotides (5, 62)
(Fig. 3). Recently, one of us (S. S.) has shown by photochemical
cross-linking that the ss DNA (and also perhaps ss RNA) binding
activity lies in the fingers (or fingers-palm junction) domain
(34).
Previous reports suggest that 3-end NMP addition to RNA was due to the
presence of a self-complementary secondary structure or partial
intermolecular complementarity in the nascent RNAs (51, 53). These
authors suggested that T7 RNAP binds to free RNA in transcription
reactions to form binary complexes in which end polymerization
occurred. In our experiments, we have eliminated this possibility by a
multistep procedure. 1) We passed the transcription reaction through a
Sephadex spin column to remove NTPs. 2) We ran the complexes on native
polyacrylamide gels. In this step, we have carefully identified ternary
complexes by 32P labeling either the DNA or RNA and
ensuring that the complexes contained T7 RNAP using Western blotting
with anti-T7 RNAP antibody. We also made sure that when the template
DNA or RNA were independently 32P-labeled, the reactions
were run on the same gel and bands with identical mobilities were
recovered from the gels. 3) We recovered the complexes from gel pieces
by electroelution and then ran the radiolabeled RNAs from disrupted
complexes on high resolution denaturing acrylamide gels. These
procedures assured us that the complexes isolated from native gels were
indeed ternary complexes. Some elongation complexes are resistant to
heparin, whereas others are not (4, 55, 58, 63). Therefore, we believe
that the observed cleavage and end-polymerization reactions occurred in ternary transcription complexes and not simply in binary complexes of
free RNA and DNA in transcription reactions. During elongation, ternary
complexes of T7 RNAP are highly heterogeneous. This heterogeneity is
clearly seen in isolated elongation complexes paused/arrested by a
psoralen roadblock or at the end of a template (Fig. 8). End pausing
has been observed with eucaryotic pol II during transcription on
blunt-ended templates (64).
Fig.
9A is a diagrammatic representation of an
elongation complex paused at the end of the template or arrested by a
psoralen roadblock. The RNAP encloses a "bubble" with a 3-base pair
(or up to 7-base pair) RNA-DNA hybrid (3, 37, 55, 65, 71). The size of
the bubble and hybrid in elongation complexes of T7 RNAP have not been
directly measured. However, they can be inferred from published reports
(55, 71). A fraction of the paused/arrested polymerases may "back
up" (Fig. 8B). Backup of the polymerase was probably
facilitated by the inhibition of enzyme turnover by high salt/heparin.
Backing up may be associated with conformational rearrangements in the
catalytic core of RNAP. The 3 nascent RNA product binding site in the
arrested/paused RNAP may be reorganized such that 3
-end RNA-DNA
contacts are lost but the 3
distal contacts are maintained (Fig.
8B). A part of the 3
-end is now partially single-stranded
and is held in a new RNAP site. The binding of the RNA in the new RNAP
site may facilitate or effect cleavage (Fig. 8B). Recent
work with E. coli RNAP indicated that RNA product binding
site reorientation and conformational changes occur in paused and
arrested complexes (67). We visualize a stalled/paused T7 RNAP as
having two activities, viz. 3
-end-initiated polymerization (Refs. 51, 53, and 66, and this work) and 3
to 5
RNase activity (this
work). These are opposing activities that generate heterogeneity in
paused/arrested complexes. Although the catalytic pocket is probably
the site of both nuclease and polymerase activities, in the presence of
NTPs the polymerase activity is kinetically dominant. This explains why
addition of NTP to the ss DNA oligonucleotide suppresses the nuclease
(Fig. 6) and why during normal processive transcription the nuclease
activity is not observed. The nuclease activity is only kinetically
unmasked when the polymerase is paused or arrested by a roadblock.
Why does cleavage of ss RNA oligonucleotide occur with higher than a 1:1 molar ratio of enzyme to RNA, whereas cleavage in ternary elongation complex occurs at an equimolar ratio? The two events, binding and catalytic cleavage, occur in that sequence. In paused/arrested ternary elongation complexes, the RNA is held very tightly by the enzyme. The RNA and RNAP in these complexes are not in equilibrium with free RNA and free polymerase, unlike in binary complexes, where they are in equilibrium with free RNA and free RNAP. Hence, higher concentrations of polymerase are required to drive by mass action the equilibrium toward binding and cleavage in binary complexes.
Role of Nuclease ActivityComparison of the T7 RNAP nuclease
activity with the known nuclease activity of other RNAPs or HIV reverse
transcriptase reveals many differences and some similarities. At first
glance, the ss DNA-ss RNA nuclease activity of T7 RNAP appears to be
somewhat different from the known GreB-enhanced intrinsic cleavage
activity of E. coli RNAP. The GreB-enhanced E. coli RNAP cleaves the nascent RNA in arrested/paused ternary
complexes into fragments larger than T7 RNAP (24, 25). The T7 RNAP
nuclease pattern resembles that of GreA-enhanced E. coli
RNAP, eucaryotic pol II, or vaccinia RNAP in that these enzymes seem to
cut in mono- or dinucleotide steps (23, 61). T7 RNAP cleavage is
reminiscent of SII-mediated "minor" mode cleavage in yeast pol II
binary complexes (26) because cleavage starts at the 3-end releasing
1-3 nts at a time. It is not known if E. coli RNAP or the
eucaryotic RNAPs have a single-stranded DNase activity. T7 RNAP
nuclease activity resembles the HIV reverse transcriptase nuclease in
that arrest of template-specified polymerization is needed (72). The
dual activities, viz. the 3
-end-initiated nontemplated
polymerization and the nuclease activity, are common to T7, E. coli, and pol II RNAPs and to HIV reverse transcriptase (73).
Other common features are the complete requirement for a divalent
cation and the apparent independence from pyrophosphorolysis. Although
we have not demonstrated that cleavage is accompanied by a backward
movement of the T7 RNAP from the end of the template or from a stalled
site, the patterns of the cleavage sites (Fig. 8) are consistent with
this possibility. Because of the relative instability (dwell times of
<5-10 min) and heterogeneity of T7 RNAP elongation complexes, it is
technically difficult or perhaps impossible to walk T7 RNAP from one
register to another in isolated homogeneous complexes as was
done with E. coli RNAP. In the case of E. coli
RNAP and eucaryotic pol II, it has been speculated that hydrolytic
nuclease activity may be an error-correcting mechanism or a way of
rescuing paused/arrested complexes. Similar mechanisms may operate
during T7 RNAP transcription. What role might the ss DNase activity
play in vivo? During T7 phage DNA replication in
vivo, linear T7 DNA concatemers are formed through base pairing of
terminal repeats. This is accompanied by digestion of surplus
single-stranded DNA (68). The mechanisms of these processes are not
clearly understood. We speculate that digestion of surplus
single-stranded DNA serves two functions: 1) production of dNTPs to
augment cellular pools that are channeled for rapid T7 DNA replication;
and 2) reduction of the competition from ss DNA for T7 DNA polymerase
activity (69). There are three known nucleases produced by the T7
phage. 1) Gene 6 exonuclease (5
to 3
), which is normally involved in
removal of primers after lagging strand synthesis, in recombination,
and in the destruction of bacterial DNA. This is a ds DNA-specific
exonuclease (70). 2) Gene 3 endonuclease, which is specific to
single-stranded DNA and may be a resolvase involved in homologous
recombination. 3) 3
to 5
exonuclease of T7 DNA polymerase, which is
an error-correcting activity during T7 DNA synthesis. T7 RNAP nuclease,
the fourth nuclease activity, may serve as an additional nuclease in
the digestion of the surplus single-stranded DNA during T7 DNA
replication and maturation. Since T7 RNAP nuclease is a single-stranded
3
to 5
exonuclease and is suppressed during normal transcription elongation, the double-stranded form of T7 DNA would not be its target.
We thank Prof. F. W. Studier for the kind gifts of anti-T7 RNAP antisera and T7 lysozyme and Dr. Bill McAllister for bacterial strains and suggestions. We appreciate the very helpful suggestions from Profs. Mike Chamberlin and Charles Richardson. We also thank Dr. Stewart Shuman for critical comments on the manuscript and Prof. Joshua Lederberg for interest in the project.