ABSTRACT
A key step in the supercoiling reaction is the DNA gyrase-mediated cleavage and religation step of double-stranded DNA. Footprinting studies suggest that the DNA gyrase
binding site is 100-150 bp long and that the DNA is wrapped around the enzyme with the
cleavage site located near the center of the fragment. Subunit A inhibitors
interrupt this cleavage and resealing cycle and result in cleavage occurring at
preferred sites. We have been able to show that even a 30 bp DNA fragment
containing a 20 bp preferred cleavage sequence from the pBR322 plasmid was a
substrate for the DNA gyrase-mediated cleavage reaction in the presence of inhibitors. This DNA
fragment was cleaved, although with reduced efficiency, at the same sites as a
122 bp DNA fragment. A 20 bp DNA fragment was cleaved with low efficiency at
one of these sites and a 10 bp DNA fragment was no longer a substrate. We
therefore propose that subunit A inhibitors interact with DNA at inhibitor-specific positions, thus determining cleavage sites by forming ternary
complexes between DNA, inhibitors and DNA gyrase.
DNA gyrase (EC 5.99.1.3), a prokaryotic topoisomerase II enzyme, consists of two
subunits, A and B, and the active enzyme is an A
2
B
2
tetrameric complex (reviewed in
1
-
5
). The enzyme can introduce negative supercoils into DNA using the free energy
derived from ATP hydrolysis. Footprinting studies have shown that DNA gyrase
protects ~100-150 bp of DNA from nuclease attack, with a most strongly protected
central region of ~40-50 bp (
6
-
9
). The DNA is wrapped around the tetrameric protein in a single positive
supercoil. After binding, DNA gyrase cleaves each strand at sites separated by
4 bp and forms a covalent phosphotyrosine bond between the 5'-phosphate groups of the cleaved DNA and a tyrosine (Tyr122 in
Escherichia coli
) of the A subunits (
10
,
11
). A segment of DNA is translocated through the break and presumably through the
protein complex and the broken phosphodiester bonds are resealed. Binding of
DNA gyrase to DNA is probably sufficiently stable to allow processive
supercoiling before the enzyme dissociates from the DNA (
12
). At some point in the reaction an ATP molecule binds to each B subunit and
hydrolysis of ATP is required for further catalytic cycles (
13
-
15
). Binding of ATP promotes a conformational change of the tetramer and it is
thought that this change brings the DNA segment to be translocated into near
proximity to the double-stranded DNA break (
8
).
A key step in the supercoiling reaction is DNA gyrase-mediated cleavage of DNA and it has been shown that both classes of subunit A
inhibitors, the quinolones and the pyrimido[1,6-
a
] benzimidazoles, interrupt the cleavage and resealing cycle at the cleavage step (
16
-
18
). Cleavage in the presence of these inhibitors does not require ATP, occurs at preferred sites and it is assumed that these
sites represent the physiological sites of action of DNA gyrase (
6
,
19
-
22
).
In vivo
analysis of cleavage sites and their flanking regions in the pBR322 plasmid
generated in the presence of the quinolone oxolinic acid has suggested a
consensus sequence (shown below) where R = purine, Y = pyrimidine, N = any nucleotide; T and G are equally preferred at the position 13 of the consensus
sequence and the G and T in brackets are preferred secondarily to T and G
respectively.
[G] G [T]
5'-RNNNRNRT <=> GRYCTYNYNGNY-3' consensus sequence
5'-GGCTGGAT <=> GGCCTTCCC
CAT-3' preferred cleavage site
-
DNA cleavage occurs at the site indicated by the arrow and also shown is the
major cleavage site (between thymidine and guanosine) at position 990 (black
dot) on this plasmid and the surrounding 20 bp sequence (
19
). The mismatch between the preferred cleavage site and the consensus sequence
is underlined. It has been shown that a 34 bp DNA fragment containing this 20 bp cleavage sequence is not a substrate for the cleavage reaction in the presence
of oxolinic acid (
23
). However, we and others could show that fragments of 70 bp or longer
containing this 20 bp sequence at different positions were accepted as
substrate and that these DNA fragments were positioned onto the enzyme in such a way that cleavage occurred at the expected site (
18
,
23
). Therefore, both the 20 bp cleavage sequence and the length of the flanking DNA on either
one side of the cleavage site seem to be critical for cleavage reactions
carried out in the presence of inhibitors. To further investigate the role of
the 20 bp cleavage sequence and the relevance of the length of the flanking
regions we have examined the DNA gyrase-mediated cleavage reaction in the presence of subunit A inhibitors with
DNA fragments of 10-122 bp each containing the 20 bp cleavage sequence or part of it.
Enzymes for cloning, isolation and labeling of DNA fragments were purchased from
Boehringer and the methods used were essentially as previously described (
24
).
A gel-purified
Bgl
I-
Bst
NI DNA fragment from position 933 to 1061 from the pBR322 plasmid was blunt-ended with the Klenow enzyme,
Eco
RI linkers were attached and the resulting
Eco
RI- digested 140 bp fragment ligated into the desphosphorylated
Eco
RI site of pUC18. The orientation of the inserted fragment was determined by
restriction enzyme analysis and is shown in Figure
1
. For the construction of deletions, the plasmid was digested with
Bam
HI and
Sph
I, the recessed 3'-termini of the
Bam
HI site were removed by incubation of 10 [mu]g DNA with 20 U exonuclease III for 2-5 min at 37oC and the DNA was ethanol precipitated and dissolved in TE
buffer (10 mM Tris-HCl, pH 8.0, 1 mM EDTA). Single-stranded DNA was digested with S1 nuclease, the plasmids were blunt-ended with Klenow enzyme and ligated with T4 DNA ligase.
Competent
E.coli
HB101 cells were prepared by the calcium chloride method and transformed with
the constructs described above. Plasmid DNA from the transformants was prepared and the DNA sequenced using Sequenase (US Biochemicals) and the primer 5'-CAGGAAACAGCTATGAC-3'. The resulting DNA fragments are shown in Figure
1
A.
The A and B subunits of DNA gyrase were expressed and purified as previously
described (
18
). For some experiments the subunits were further purified on a novobiocin-Sepharose column. Novobiocin-Sepharose was prepared as described (
14
). After loading of the subunit A fraction, the column was washed with TGED
buffer [50 mM Tris-HCl, pH 7.5, 1 mM EDTA, 5 mM DTT, 10% (w/v) glycerol] and the protein
eluted with 1 M NaCl in TGED. After loading of the subunit B fraction, the
column was washed with TGED, 1 M NaCl and 2 M urea in TGED and the protein
eluted with 6 M guanidine hydrochloride. The B subunit was then renatured by
dialysis against 50 mM Tris-HCl, pH 8.0, 100 mM KCl, 5 mM DTT, 1 mM EDTA, 10% (w/v) glycerol.
Stock solutions of the subunit A inhibitors (Table
1
) at 1 mM were made as follows. Fleroxacin was dissolved in 25% ethanol and 15 mM NaOH, ciprofloxacin in H
2
O and Ro 46-2825, Ro 46-6962, Ro 46-7864 and Ro 47-3359 in 5 mM HCl. The stock solutions were diluted with
H
2
O and added to the reaction mixtures as indicated in the figures.
Table 1
.
Structures of quinolones and pyrimido[1,6-
a
] benzimidazoles
The DNA fragments, labeled with [[gamma]-
32
P]ATP at the 5'-ends of the top strand, were incubated at an estimated
concentration of 20-40 fmol at 37oC for up to 120 min in a total volume of 20 [mu]l with ~2.5 pmol DNA gyrase in 35 mM Tris-HCl, pH 8.0, 24 mM KCl and 2 mM spermidine. MgCl
2
was added at a concentration of 4 mM and the inhibitor at the concentrations
indicated in the figures (up to 500 [mu]M). Reactions were stopped by the addition of SDS to a final concentration
of 1%. Proteinase K was added to a final concentration of 500 [mu]g/ml and the samples digested for 2 h at 37oC. The DNA was purified by phenol/chloroform extraction and ethanol
precipitation and redissolved in 5 [mu]l TE, pH 8.0. Five microliters loading buffer (50% formamide, 0.05%
bromophenol blue, 0.03% xylene cyanol FF and 5 mM EDTA) were added, samples
were heated for 4 min at 90oC, chilled on ice, loaded on a 12 or 15% sequencing gel containing 7 M urea
and electrophoresed in 90 mM Tris-borate, 2 mM EDTA (1* TBE buffer). Reaction products were visualized by autoradiography and
scanned with a G-700 imaging densitometer (BioRad). Sequencing of the DNA fragments were
carried out by the Maxam-Gilbert sequencing method.
The pBR322 plasmid contains, around position 990, a 20 bp sequence where DNA
gyrase preferentially cleaves in the presence of quinolones (
19
). In an earlier report we demonstrated that DNA fragments of 85 and 71 bp
containing this cleavage sequence at different positions were accepted as
substrate for cleavage reactions in the presence of subunit A inhibitors (
18
). To further explore the role of this 20 bp cleavage sequence and the necessary
length of the flanking regions in more detail, we examined the DNA gyrase-mediated cleavage reaction in the presence of subunit A inhibitors with DNA fragments of 10-122 bp (Fig.
1
A and B). The DNA fragments of 30-122 bp contained the 20 bp cleavage sequence at different positions. The
20 bp fragment consisted of this cleavage sequence and the 10 bp DNA fragment
contained only part of it. Assuming that labeling occurred with similar
efficiency, equal amounts of these DNA fragments were incubated with DNA gyrase
and Mg
2+
in the presence or absence of subunit A inhibitors. After incubation, the
protein was denatured with SDS and digested with proteinase K. The DNA was
purified and loaded onto a denaturing gel. For determination of the positions
of the cleavage sites, the migration of the cleavage products was compared with
the sequenced 122, 111, 105, 96, 30 and 20 bp DNA fragments.
Our earlier results showing that pyrimido[1,6-
a
]benzimidazoles have a mode of action similar to that of quinolones could be confirmed (
18
). Both classes of subunit A inhibitors induced DNA gyrase to cleave at the same
sites and differences between these inhibitors could be observed only in the
preferences of DNA gyrase for these cleavage sites (shown for the 77, 64 and 38
bp DNA fragments in Fig.
2
and Table
2
). However, the results indicate that DNA gyrase preferred the cleavage site
between positions 993 and 994 (TGGC
<=>
CT, cleavage occurs at the site indicated by the arrow) in the presence of the
pyrimido[1,6-
a
] benzimidazoles Ro 46-2825, Ro 46-7864 and Ro 47-3359, but in the presence of Ro 46-6962 the cleavage site between position 990 and 991 (T
<=>
GGCCT) was preferred. It is worthwhile to mention that Ro 46-6962 is one of the pyrimido[1,6-
a
]benzimidazoles that do not chelate divalent cations but nevertheless inhibit the DNA gyrase-promoted supercoiling reaction
as well as the quinolones and the chelating pyrimido[1,6-
a
] benzimidazoles (
18
).
Cleavage generated in the presence of Ca
2+
and the absence of inhibitors occurs at the same sites as those obtained in the
presence of oxolinic acid, but with different relative efficiencies (
3
). However, it seems probable that not only the sequence but also the length of
the DNA fragments is critical for cleavage reactions carried out in the
presence of Ca
2+
(
18
). The results of this work confirmed these observations because the DNA
fragments >= 96 bp were cleaved in the presence of Ca
2+
and absence of inhibitors within the sequence T
<=>
GGCCT but the shorter ones ( <= 77 bp) were not substrates (data not shown).
DNA gyrase cleaves in the presence of quinolones at preferred sites on DNA and
this quinolone-induced cleavage reaction has been taken as a model for the double-stranded cleavage event during supercoiling. Footprinting
experiments have shown that the DNA gyrase binding site is 100-150 bp long, with the cleavage site located near the center of the
fragment (
6
-
9
). Cleavage sequences share homology around the breakage point and, based on
analysis of cleavage sites and their flanking regions in the pBR322 plasmid, a
20 bp consensus sequence has been proposed (
19
). It was also shown that there is a major cleavage site at position 990 on this
plasmid (
19
). However, a 34 bp DNA fragment containing the 20 bp cleavage sequence around
this major cleavage site is not a substrate for the enzyme in the presence of
oxolinic acid and flanking DNA is required for efficient DNA breakage (
23
). In a previous report we used restriction fragments of 85 and 71 bp from the
plasmid pBR322 as model substrate DNA, each containing the preferred 20 bp
cleavage sequence at a different position, and we could show that even these
DNA fragments, despite their length being in principle insufficient to wrap
around the tetrameric protein, were accepted as substrate for cleavage
reactions in the presence of subunit A inhibitors (
18
). The DNA fragments were positioned onto the enzyme in such a way that cleavage
occurred at the predicted site within the 20 bp cleavage sequence. In this
report we show that even a 30 bp DNA fragment containing the 20 bp preferred
cleavage sequence from the pBR322 plasmid is still a substrate for DNA gyrase.
Even a 20 bp DNA fragment was cleaved, but with very low efficiency. Cleavage
within a 10 bp DNA fragment, containing only part of the 20 bp cleavage
sequence, could not be observed, indicating that a fragment of this length is
no longer a substrate for the enzyme. Inefficient cleavage of short DNA
fragments may reflect a lowered binding affinity because a minimum number of
DNA-protein contacts are necessary for efficient cleavage.
Confirming our results obtained earlier with the 85 and 71 bp DNA fragments, DNA
gyrase cleaved all the cleavable DNA fragments in the presence of subunit A inhibitors preferentially at two sites
within the 20 bp cleavage sequence of the pBR322 plasmid.
-
+
-
+
5'-GGCTGGAT[brvbar]GGC[brvbar]CTTCCCCAT-3'
0
3
Bases that form covalent phosphodiester bonds with the enzyme are marked + and
the free 3'-hydroxyl ends at the cleavage site are marked -. The sequencing data shows that cleavage occurred at the site earlier reported within the sequence T
<=>
GGCCT between T at position 990 (numbered 0) and G at position 991, but in
addition 3 bp downstream (TGGC
<=>
CT) between C at position 993 (numbered 3) and C at position 994. Even if one
considers that cleavage occurs on the other strand 4 bp away (
6
,
7
,
25
), no obvious sequence homology can be deduced from these major cleavage sites
and one can only speculate that the quinolones may prefer a guanosine at
position +1 in at least one strand. To confirm this speculation, further strong
cleavage sites would have to be analyzed. However, it is known that some
antitumor drugs target eukaryotic topoisomerase II and that these compounds
also stimulate topoisomerase II-mediated DNA cleavage by interfering with the breakage-religation reaction (
26
-
31
). Different drug families show a variable degree of sequence specificity but
cleavage sites are generally conserved within the same family (
28
,
32
-
35
). Also, quinolone derivatives that have been shown to induce eukaryotic
topoisomerase II to cleave at specific sites prefer a specific base at the
cleavage site (
36
). It is postulated that these antitumor drugs form a ternary complex by binding
to preferred nucleotides adjacent to the cleavage site and to amino acid
residues of the enzyme. It is thus a possibility that DNA gyrase subunit A
inhibitors also prefer specific bases and induce DNA gyrase to cleave at inhibitor-specific sites.
Hence, DNA gyrase cleaved DNA fragments in the presence of subunit A inhibitors
preferentially within the 20 bp cleavage sequence, but the cleavage pattern
also shows cleavage products derived from cleavage reactions at the end of the
DNA fragments. These cleavage products may be the result of a Mg
2+
-dependent 3' -> 5' DNA exonuclease which was co-purified with novobiocin affinity column-purified DNA gyrase subunits. The specific
and very different purification steps for the two subunits are expected to
remove potential exonuclease contamination at least from either one of the two subunits. The purified subunits, however, even after
several purification steps, might still be cross-contaminated with small amounts of the other subunit due to their tight interaction in
the A
2
B
2
tetrameric DNA gyrase complex. Because cleavages at the end of the DNA
fragments also appeared in the presence of Mg
2+
but absence of inhibitors, cleavage at these sites by DNA gyrase, resulting in
DNA fragments shortened by some bases, would not be the effect of specific enzyme-quinolone-DNA complexes but would occur at non-specific sites. However, the mechanism of such a suicide
type of cleavage at the very ends of DNA fragments is not clear.
Based on our results we propose the following model. The enzyme attempts to bind
to DNA by wrapping the DNA around the tetrameric protein. If DNA gyrase
interacts, in the presence of an inhibitor, with DNA fragments which are long
enough to be wrapped around the enzyme (>100 bp), they are preferentially
cleaved at the T
<=>
GGCCT site. Whether DNA gyrase also cleaves at this position in the presence of
Mg
2+
but absence of inhibitors cannot be determined, because DNA gyrase performs the
religation reaction very efficiently in the absence of an inhibitor (
18
). Because DNA fragments shorter than ~100 bp cannot be reasonably positioned onto the enzyme, it may attempt, in
the absence of an inhibitor, to establish a maximum number of DNA-protein contacts by binding such a DNA fragment to at least one side of
the enzyme complex. Thus asymmetrical binding of the DNA fragments occurs and
the enzyme may even bind to the end of the DNA fragments. However, if DNA
gyrase interacts, in the presence of an inhibitor, with DNA fragments of a
length that is in principle insufficient, it cleaves at the same inhibitor-specific sites as with longer fragments but probably with a higher
preference for the TGGC
<=>
CT site. Therefore, the enzyme is trapped at inhibitor-specific positions by forming a ternary complex between DNA, inhibitor and
DNA gyrase. Whether it is the conformation or the sequence which determines
these inhibitor-specific positions is not clear. With the shorter DNA fragments it cannot
be determined whether DNA gyrase also cleaves at these inhibitor-specific positions in the presence of Mg
2+
but absence of inhibitors. However, earlier experiments performed with Ca
2+
instead of Mg
2+
indicated that a 85 bp fragment containing the 20 bp cleavage sequence was not
cleaved within this cleavage sequence but at a site that can be explained by an
asymmetrical wrapping of the DNA fragment around the enzyme. About 70 bp of one
end of the DNA fragment were wrapped around one side of the enzyme and ~15 bp around the other (
18
). We have observed that the 122, 111, 105 and 96 bp DNA fragments were cleaved
in the presence of Ca
2+
within the sequence T
<=>
GGCCT (data not shown) and these results support our hypothesis, because the
distance from the cleavage site to the
Eco
RI site is ~70 bp. DNA fragments <= 77 bp were not substrates for the cleavage reaction in the presence of
Ca
2+
and probably could not be positioned correctly onto the enzyme.
In this work we have shown that even a 20 bp DNA fragment containing a 20 bp
preferred cleavage sequence from plasmid pBR322 was a substrate for the DNA
gyrase-mediated cleavage reaction in the presence of inhibitors. Although such
fragments are too short to be wrapped around the enzyme or at least around one
side, the 30 bp DNA fragment was cleaved, although with reduced efficiency and
with different preferences, at the same sites as the 122 bp DNA fragment. The
20 bp DNA fragment was cleaved with very low efficiency at one of these sites
and only a 10 bp DNA fragment was not a substrate. Whether DNA gyrase in the
presence of subunit A inhibitors either, by analogy with eukaryotic
topoisomerase II inhibitors, interacts with preferred nucleotides adjacent to
the cleavage site or whether it is the DNA conformation which determines the
inhibitor-specific cleavage sites requires further investigation. However, we
propose that the subunit A inhibitors interact with DNA at inhibitor-specific positions thus determining cleavage by forming a ternary complex between DNA, inhibitors and DNA gyrase, but it remains an open question
whether these sites are also the physiological sites.
We thank P.Hartman for carefully reading the manuscript.
*To whom correspondence should be addressed. Tel: +41 61 688 8464; Fax: +41 61
688 2729; Email: hans.gmuender@roche.com
REFERENCES
Return