ABSTRACT
Termination of DNA replication in Bacillus subtilis involves the polar arrest of replication forks by a specific complex formed between the replication terminator protein (RTP) and DNA terminator sites. While determination of the crystal structure of RTP has facilitated our understanding of how a single RTP dimer interacts with terminator DNA, additional information is required in order to understand the assembly of a functional fork arrest complex, which requires an interaction between two RTP dimers and the terminator site. In this study, we show that the conformation of the major B.subtilis DNA terminator, >TerI, becomes considerably distorted upon binding RTP. Binding of the first dimer of RTP to the B site of TerI causes the DNA to become slightly unwound and bent by ~40o. Binding of a second dimer of RTP to the A site causes the bend angle to increase to ~60o. We have used this new data to construct two plausible models that might explain how the ternary terminator complex can block DNA replication in a polar manner. In the first model, polarity of action is a consequence of the two RTP-DNA half-sites having different conformations. These different conformations result from different RTP-DNA contacts at each half-site (due to the intrinsic asymmetry of the terminator DNA), as well as interactions (direct or indirect) between the RTP dimers on the DNA. In the second model, polar fork arrest activity is a consequence of the different affinities of RTP for the A and B sites of the terminator DNA, modulated significantly by direct or indirect interactions between the RTP dimers.
Completion of a round of replication of the circular chromosomes of
Bacillus subtilis
and
Escherichia coli
involves arrest of replication forks at specific DNA sequences of <= 30 bp in length that are located in the terminus region approximately
opposite the origin. Arrest at these sites requires the binding of a terminator
protein to form a DNA-protein complex that blocks fork movement in a polar manner (i.e. it blocks forks approaching from one direction
only) (
1
,
2
). This complex has been proposed to cause replication fork arrest by inhibiting the DNA unwinding activity of the
helicase present at the apex of the replication fork (
3
,
4
).
A set of at least six DNA terminators has been identified in the terminus
regions of both bacteria (
Ter
I
-
VI
in
B.subtilis
and
Ter A
-
F
in
E.coli
) (
5
,
6
); at least three of these terminators are orientated to stop the anticlockwise
replication fork, whilst the other three are positioned to stop the clockwise
replication fork. In both organisms, the relative arrangement of the two groups
of terminators is such that the terminus region acts as a replication fork
trap, allowing replication forks to enter the region but not exit (see
7
). Surprisingly, there is no significant sequence or structural homology between
the terminator proteins of the two organisms (RTP in
B.subtilis
and Tus in
E.coli
) nor between the sequences of their DNA terminator sequences (
8
,
9
). Thus, the two organisms may use quite different molecular mechanisms to achieve helicase inhibition.
RTP is a 29 kDa symmetrical dimer (
10
-
12
) which interacts specifically with an ~30 bp region of the major
B.subtilis
terminator
TerI
(originally described as a 47 bp segment of DNA called IRI; see
13
). This binding region (see Fig.
1
) contains two overlapping RTP dimer binding sites, A and B, the latter having a
higher affinity for RTP. The B site is located proximal to the approaching fork
and is pseudosymmetrical in sequence (
11
). When RTP is bound to this site alone, replication fork arrest does not occur;
a second RTP dimer must bind cooperatively to the low affinity A site to
produce an active terminator (
14
,
15
).
All chemicals were of analytical or molecular biology grade. Unlabelled RTP was
prepared as described previously (
11
). The plasmid pBend2 was a generous gift from S.Adhya (Bethesda, MD). The plasmid pPW2 was produced in this laboratory (see below). Restriction endonucleases (EC 3.1.21.4) and mung bean nuclease (EC 3.1.30.1) were obtained from Boehringer Mannheim (Mannheim, Germany). T4 DNA ligase (EC 6.5.1.1) was obtained from New England
Biolabs (Beverly, MA). T4 polynucleotide kinase (EC 2.7.1.78) was obtained from Amersham International (Amersham, UK).
T7 DNA polymerase (EC 2.7.7.7) was obtained from Pharmacia LKB (Uppsala,
Sweden).
Two 20 nt oligonucleotides spanning the 16 nt complementary strands of the B
site from
TerI
(see Fig.
1
) were synthesized by Biotech International (Perth, WA): fragment 1, 5'-CTATGTACCAAATGTTCAGT-3'; fragment 2, 3'-GATACATGGTTTACAAGTCA-5'. Equimolar quantities of
each DNA strand were mixed at room temperature in TE buffer (10 mM Tris, pH
7.5, 1 mM EDTA) and placed in a heating block at 95oC for 5 min. The fragments were then allowed to anneal by slow cooling over 40 min to 37oC, followed by incubation at this temperature for a further 1 h. The
solution was then cooled to room temperature overnight. The duplex DNA was ethanol precipitated, aerated and resuspended in phosphate buffer (5 mM NaH
2
PO
4
, pH 7.8, 1 mM DTT) for CD and band retardation analysis. The concentration of
each duplex was determined from the absorbance at 260 nm; it was assumed that
an A
260
of 1.0 is equivalent to a double-stranded DNA concentration of 50 [mu]g/ml.
All CD spectra were acquired using a Jasco J-720 spectropolarimeter under constant nitrogen flush at 20oC. The spectropolarimeter was calibrated with an aqueous solution of ammonium D-camphor- 10-sulfonic acid (
20
). The temperature of the cell was maintained and adjusted with a Neslab RTE-111 circulating water bath. All spectra were acquired using the following
parameters: four scans, resolution 0.2 nm, bandwidth 1.0 nm, sensitivity 50
mdeg, response time 1 s and scan speed 20 nm/min. Except for the titration
study of B site DNA, all CD studies were performed using a 0.1 cm cell containing 300 [mu]l samples with a buffer of 5 mM NaH
2
PO
4
, pH 7.8, 1 mM DTT. In all cases, prior to spectral acquisition, the sample was
prepared at room temperature and incubated for at least 5 min within the
spectropolarimeter to ensure temperature and solvent equilibration. Spectral signal-to-noise was optimized by multiplying the frequency power spectrum by a
trapezoidal apodization function prior to inverse Fourier transformation to
give the noise-reduced frequency domain spectrum.
CD spectra of B site DNA alone (10 [mu]M) were acquired over the wavelength range 180-320 nm. The stoichiometry of the RTP-B site interaction was monitored by the stepwise addition of 4
[mu]l 0.1625 [mu]M RTP to 2600 [mu]l 0.5 [mu]M B site DNA in 0.5 mM NaH
2
PO
4
, 0.1 mM DTT, pH 7.8, in a 1 cm cell. Previous attempts using a 5 [mu]M B site sample in standard buffer in a 0.1 cm cell were subject to mixing
problems, large dilutions and often transient precipitation of the added
protein: these problems did not occur with the 1 cm cell. Spectra were acquired between 260 and 290 nm at the following ratios of RTP to B site: 0, 0.25, 0.5, 0.75, 1.0, 1.25, 1.5,
1.75, 2.0, 2.25, 2.5, 3.0, 3.5 and 4.0. Prior to spectral acquisition, the
sample was allowed to incubate for 10 min at 20oC to ensure that the binding reaction had reached equilibrium.
These were performed essentially as described previously (
21
). Binding reactions using RTP dimer:B site molar ratios of 0, 0.25, 0.5, 1.0
and 2.5:1 were performed under identical conditions to the CD experiments (i.e.
5 [mu]M B site DNA in 5 mM NaH
2
PO
4
, 1 mM DTT, pH 7.8, 20oC). After allowing 10 min for DNA- protein complexation, reaction mixtures were electophoresed on a 6%
polyacrylamide gel and stained with ethidium bromide.
A double-stranded DNA oligonucleotide containing the 29 bp minimal region of
TerI
plus an additional 3 bp beyond the B site, as well as protruding single-strand
Xho
I ends, was phosphorylated (5') and cloned into the
Sal
I site of pBend2 (
22
) to give the plasmid pPW2. This plasmid was digested with each of the
restriction enzymes
Mlu
I,
Bgl
II,
Nhe
I,
Xho
I,
Pvu
II,
Stu
I and
Bam
HI to yield DNA fragments of approximately the same length (154 bp). The 5'-termini of the resulting fragments were labelled with [[gamma]-
32
P]ATP (7000 Ci/mmol, 5 mCi) using T4 polynucleotide kinase. The labelled
fragment was purified from the unincorporated radionucleotide using a Nick
Columntm (Pharmacia).
Band retardation experiments were performed using a DNA concentration of 6.8 * 10
-13
mol/[mu]l in a total volume of 6 [mu]l TGMK (
21
). An RTP (monomer):DNA ratio of 1:1 was used to enable both of the DNA-RTP complexes, I and II, and the free DNA species to be visualized. After incubation at 25oC for 30 min, bromophenol blue (0.07%) was added and samples were loaded onto a
10% non-denaturing polyacrylamide gel (in 36 mM Tris, 30 mM NaH
2
PO
4
, pH 7.5) and run at 8.3 V/cm for 3-7 h. The gel was fixed in acetic acid (10%), exposed to a Storage
Phosphor Screen and analysed using a PhosphorImager (Molecular Dynamics, Sunnyvale, CA).
Figure
2
(solid line) shows the CD spectrum of a double-stranded oligonucleotide corresponding to the B site of
B.subtilis TerI
DNA. It has a number of features that identify it as being B-form in solution (
23
,
24
): it has a major longwave positive peak centred at 275 nm and the intensity of
this positive peak is similar in magnitude to that of the negative peak centred
at ~240 nm. Addition of 80% trifluoroethanol (TFE) to B-form DNA has been shown to induce a B to A transition, resulting in a
CD spectrum with characteristics totally different from those of B-form DNA; the longwave positive peak is larger with a maximum at ~270 nm and a very large shortwave peak becomes apparent below 200 nm
(
24
-
26
). The changes in the CD spectrum of B site DNA upon addition of 80% TFE (Fig.
2
, dotted line) are consistent with a B to A transition, providing further
evidence that `free' B site DNA is B-form in aqueous solution.
As demonstrated above, the longwave positive CD band of nucleic acids is very
sensitive to the local structure of DNA. This band has been used as an
indicator of conformational changes in DNA upon binding various proteins, as
well as a means of investigating the stoichiometry of the binding reaction;
examples of such studies include those on Tet repressor (
27
),
trp
repressor (
28
) and CAP (
29
).
The only contribution of a protein to the CD signal in the wavelength range 260-290 nm comes from its aromatic residues. For most proteins, this
contribution is small relative to the CD signal from DNA (
30
); indeed, the CD signal of concentrations of free RTP encompassing those used
in the B site titration was found to be within the noise level of the solvent
baseline for the 260-290 nm region. If the substantial increase in ellipticity at 275 nm observed upon RTP binding to B site DNA was due to RTP alone, a radical
reorganization of the protein would be required. While this seems unlikely, it
is not impossible; for example, the DNA binding domains of bZIP transcription factors are disordered in the uncomplexed protein but become [alpha]-helical upon complexation with DNA (
31
).
Bent DNA migrates anomalously in non-denaturing polyacrylamide gels because of its reduced ability to reptate through the pores of the
gel matrix (
31
-
34
). The mobility of the complex depends on the location of the bend within the
DNA fragment; minimal migration occurs when the bend is located near the middle
of the fragment, as this gives the largest effective Stokes radius (
34
). The circular permutation assay of Kim
et al
. (
22
), which takes advantage of the anomalous gel behaviour of bent DNA, was used to
investigate whether the binding of RTP to
TerI
causes bending of the DNA.
A theoretical study of the relationship between nucleic acid conformation and
the resulting circular dichroism found a strong linear correlation between the magnitude of the CD signal at 275 nm and the helix winding angle and twist (
36
), i.e. as the winding angle increases the twist decreases, resulting in an
increase in ellipticity. We know of only three atomic resolution structures
that have been obtained for protein-DNA complexes in which binding of the protein to the DNA causes a
substantial increase in intensity of the longwave CD signal: GCN4-ATF-CREB (
31
,
37
), Lac repressor headpiece-half operator (
38
-
40
) and
trp
repressor-
trpO
operator (
28
,
41
-
43
); in all cases, as expected on the basis of the CD data, the structures show
that the DNA is underwound in the protein-DNA complex. Interestingly, the increases in ellipticity observed for these complexes (41-60%) and the DNA bending angle (15-20o) are similar to those observed in the current work
for RTP binding to a single B site (66% and 40o respectively).
Thus, the increased ellipticity observed at 275 nm when RTP binds to the B site
can be interpreted as an underwinding of B site DNA due to a decrease in twist
angle. This can be crudely interpreted as B site DNA becoming more `A-like' upon binding RTP, since B- and A-form DNA have average twists of 36o (10 bp/turn) and 33o (11 bp turn) respectively (
44
). Interpretation of the circular permutation assays described herein as simply
reflecting DNA bending shows that
TerI
DNA is bent by ~40o in the RTP-
TerI
binary complex and by ~60o in the RTP
2
-
TerI
ternary complex. While in theory such bending could be achieved by either a
roll or tilt between sequential base pairs, analysis of atomic resolution
protein-DNA structures shows that such bends are always produced by alterations
in roll rather than tilt (
45
). An increase in roll angle can, like the decrease in twist discussed above, be
crudely interpreted as
TerI
DNA becoming more `A-like' upon binding RTP (A- and B-form DNA have average roll angles of +20o and 0o respectively; see
44
). Low twist and positive roll are typical of base pair steps with `low twist
profile' (
46
).
Thus it would appear that binding of a single RTP dimer to the B site of
TerI
causes underwinding and bending of the DNA, which will serve to widen the minor
groove and compress and deepen the major groove. It is possible that these
conformational changes could affect the A site and be responsible for the
cooperative
binding of a second dimer at this site (
10
).
It is significant that binding of the second dimer to the A site causes
additional
bending of
TerI
DNA. It has recently been suggested, on the basis of studies with symmetrical
terminators (
19
) and experimentally determined RTP-DNA contacts (
13
), that in forming the ternary complex, the centres of the RTP dimers in the A
and B sites are positioned on almost opposite faces of B-form DNA (
19
). If the binding of an RTP dimer to the B site induces a simple bend and if
this applied also when a second dimer filled the A site, one might have
expected to see a reduction in the overall bend angle in the absence of other
effects. The apparent increase in bend angle may argue for a significant
rearrangement of the ternary terminator complex which places the two dimers on
closer faces of the DNA, thereby promoting protein-protein interactions. Alternatively, it may argue for something other
than a simple bend in the DNA when both dimers are bound. It is possible that
circular permutation assays of RTP binding to `terminator' constructs with
altered spacing between two strong binding sites (so-called `phasing' experiments; see
34
) might provide more definitive information on the orientation of the two RTP
dimers in the ternary complex.
To date, the favoured model for generation of an RTP-terminator complex with polar fork arrest activity has evoked asymmetry
in the ternary complex as a result of the different contacts between RTP and DNA at the A and B sites (which have substantially different sequences),
modulated crucially by interactions between the RTP dimers bound at the two overlapping sites (
19
). We shall refer to this as the `induced conformational change' (ICC) model,
which is illustrated schematically in Figure
6
A. In this model, RTP binds to the high affinity B site and, while there may be
induced rearrangements in the DNA and (to a lesser extent) the protein (as
indicated in the current work), the final conformation of the binary complex is
unsuitable for impeding progression of the replicative helicase. However, it is
proposed that subsequent binding of a second dimer to the A site causes further rearrangements (as evidenced by the increase in the DNA bending angle seen in the current
study), possibly via protein-protein contacts between the two bound RTP dimers, to yield a unique asymmetrical ternary complex. The asymmetry is such that only the B site complex is
appropriately configured for impeding the helicase.
It is worth noting that the DBA model might explain why the RTP-terminator complex blocks RNA chain elongation by T7, SP6 and
E.coli
RNA polymerases with the same polarity as its contrahelicase activity (
49
). The DBA model predicts that the RTP-terminator complex will have similar effects on any protein or protein
complex (such as helicase or RNA polymerase) that requires DNA unwinding to
precede its translocation along the DNA. Of course, this raises the question of
why RTP can block the activity of replicative helicases but appears unable to
inhibit those involved in DNA repair and conjugal DNA transfer, including Rep
helicase, helicase I and helicase II of
E.coli
(
48
,
51
). However, one must be cautious in extending the results of such
in vitro
studies to the
in vivo
situation. For example, it has recently been demonstrated that an A173V mutant
of Tus was only a very weak inhibitor of the DnaB replicative helicase in
in vitro
assays but it halted DNA replication
in vivo
at 75% of the efficiency of wild-type Tus (
52
).
Of course, it is possible that the blocks to replication and transcription by
the RTP-terminator complex occur via quite different molecular mechanisms. For
example, the report that a specific point mutant of RTP has been constructed
which fails to block replication forks but still imposes a polar block on RNA
chain elongation (
49
) argues in favour of specific but different protein-protein or protein-DNA interactions in the inhibition of DNA replication and
transcription. In this respect, it is worth noting that there is a
statistically significant homology between RTP and the [beta]' subunit of the family of DNA-directed RNA polymerases, including that of
E.coli
. Residues 12-77 of RTP (GenBank accession no. S01271) align with residues 217-281 of the [beta]' subunit of
E.coli
RNA polymerase (GenBank accession no. RNECC) with 12 identities and 18
conservative substitutions, giving an overall homology of 45%; there is a
single residue gap in the
E.coli
sequence corresponding to residue 34 of RTP. This may indicate that RTP is
capable of displacing the [beta]' subunit from the core RNA polymerase complex or that it can bind
some critical factor that is normally attached to the [beta]' subunit. Alternatively, this homology, which includes most of the
winged-helix DNA binding motif of RTP (only the final [beta]3 strand is missing: see
12
,
17
), may simply indicate that RTP and the [beta]' subunit of the family of prokaryotic DNA-directed RNA polymerases bind DNA in a similar fashion.
It should be stressed that while the DBA model can adequately explain the polar
fork arrest activity of the RTP-terminator ternary complex without invoking specific RTP-helicase interactions, it does not rule out such a possibility. Furthermore, the model is not applicable to the
E.coli
terminator complex, where a single monomer binds to a smaller DNA terminator sequence; the DBA model
requires two terminator proteins to bind with
differential
affinities in order to generate polarity of action. In the case of the Tus-terminator binary complex, evidence against the `molecular clamp'
hypothesis has already been described in favour of specific interactions
between the terminator complex and DnaB helicase (
52
,
53
). However, given the lack of sequence homology between Tus and RTP and between
their DNA terminator binding sites, as well as the formation of binary
terminator complex in the case of Tus versus a ternary complex in the case of
RTP, it is possible that the molecular mechanism of action of Tus and RTP are
quite different. Indeed, the distinct lack of structural homology between Tus
and RTP (
8
,
9
,
12
) argues in favour of significant differences between their mechanisms of
achieving polar fork arrest.
This work was supported by research grants to G.F.K. and R.G.W. from the
Australian Research Council and by the award of a University of Sydney
Postgraduate Scholarship, an Australian Postgraduate Research Award and a
University of Auckland William Georgetti Scholarship to A.V.K. We would like to
thank Dr Tony Day (Department of Biochemistry, University of Oxford, UK) for alerting us to the homology between RTP and the family of DNA-directed RNA polymerases.
*To whom correspondence should be addressed. Tel: +61 2 9351 3902; Fax: +61 2
9351 4726; Email: g.king@biochem.usyd.edu.au
REFERENCES
Return