ABSTRACT
The
[pi]
protein of plasmid R6K functions in both replication and transcription.
[pi]
autoregulates its own synthesis and is required for replication of the R6K
[gamma]
origin.
[pi]
performs these functions by binding to specific DNA sites arranged as pairs of
6-10 bp inverted repeats (IRs) or as a cluster of seven tandem 22 bp direct
repeats (DRs) which lack symmetry. The sites share the TGAGRG nucleotide motif
(where R is A or G). The DRs and IRs flank the central A+T-rich segment of the
[gamma]
origin. In this work we carried out DNase I and hydroxyl radical protection
experiments on various deletion derivatives of the
[gamma]
origin complexed with
[pi]
protein. These experiments revealed binding of
[pi]
to a novel site embedded within the A+T-rich segment. This interaction manifests primarily by the appearance of
the enhanced scissions of DNA by DNase I and hydroxyl radicals.
[pi]
interaction with the A+T-rich site is independent of
[pi]
binding to the DRs and IRs. We propose that
[pi]
protein can recognize distinct families of DNA sequences in the
[gamma]
origin.
For many replicons, initiation of DNA replication depends on replicon-specific initiator proteins. These proteins localize the origin and allow
an ordered series of protein-protein interactions, culminating in the development of the replication
fork. Initiator proteins also frequently function as transcription factors
negatively regulating expression of their own genes. Localization of the origin
and repression of transcription depend on sequence-specific binding of the initiator to DNA. These sequences are frequently
called iterons.
This laboratory studies a basic replicon derived from plasmid R6K because it is
stably maintained at a defined copy number in logarithmically growing cultures
(
1
). Hence, it has all the features needed to elucidate the various aspects of
regulated initiation of replication and plasmid stability (
1
,
2
). As shown in Figure
1
, such a replicon is composed of a
cis
-acting [gamma] origin sequence and [pi] protein encoded by the R6K
pir
gene which can function either in
cis
or in
trans
(
3
,
4
). The [pi] protein is bifunctional in replication; it activates the origin at low a
level and inhibits replication at a high level (
5
,
6
). A balance between these two activities is believed to influence the frequency
of initiation of DNA replication and consequently the plasmid copy number. The
mechanisms allowing [pi] protein to either activate or inhibit replication are not known. In
addition to its function in replication, [pi] acts as a transcription factor negatively regulating the expression of its
own gene,
pir
(
7
-
9
). The [pi] possesses sequence-specific DNA binding activity; it recognizes sites of different
sequence composition and geometry. However, all sequences known to bind [pi] share the TGAGRG nucleotide motif (where R is A or G) (
2
,
10
).
As demonstrated by several different assays, [pi] protein binds the seven 22 bp direct repeats (DRs) in the [gamma] origin (
8
,
11
-
13
). More importantly, a direct relationship between [gamma] origin function and [pi] protein binding to the DRs has been also demonstrated (
14
). It was found that G to A transitions at the first and second Gs in the TGAGRG
motif of either the 1st or 6th DR resulted in the inability of [pi] to bind to the mutated repeat
in vitro
and the concomitant loss of [gamma] origin activity
in vivo
(
14
)
.
We have recently shown that these same Gs are protected by [pi] protein against methylation
in vivo
(
15
). A positive role for the seven 22 bp DRs in replication was also demonstrated
with deletion mutants of the [gamma] origin that precisely removed some of the DRs (
16
). Although a [gamma] origin lacking one or two DRs remains functional, deletion of three or
more of them inactivated the origin (
16
).
Yet another [pi] binding site has been recently discovered in the enhancer segment of the [gamma] origin (
1
). The enhancer [pi] site contains a pair of inverted half repeats (IRs). Mutational analysis
revealed that each half site is required for [pi] binding, but the site has no known function (
1
). Another IR forms a core of the
pir
gene operator to which [pi] binds, thereby regulating its own expression (
13
). Interestingly, the pattern of protection and cleavage of IRs in the enhancer
and
pir
gene operator are very similar (
1
,
8
,
9
,
13
); thus [pi] probably binds both pairs of IRs similarly.
A+T-rich segments are common among many origins replicating via the Cairns
mode (
17
,
18
). Their thermodynamical instability is expected to aid in the development of
the replication fork. It is of fundamental interest, therefore, to determine
which proteins can bind to such segments and, by doing so, may perhaps alter
their stability. DNase I and hydroxyl radical footprinting data presented in this paper, clearly
show that [pi] protein can bind to the A+T-rich segment independently on binding to any other site within the
boundaries of [gamma] origin (IRs and DRs). The protected region contains a very poor match to
the known consensus DNA sequences of [pi] binding sites. Therefore, we propose that [pi] binds to the two families of DNA sequence.
The construction of plasmids pMF36 (which carries seven 22 bp DRs), and pMF239
(which carries one of these 22 bp DRs) has been described (
5
,
19
). The pMF240 (which carries one mutated 22 bp DR; Fig.
1
) was made using a mutant which has three base changes; two indicated in Figure
1
at 7th, 9th, a third at 21st position (not indicated) in the first 22 bp DR
creates
Sna
B1 site. Steps for constructing pMF240 involving a deletion of
Sna
B1 fragment containing six DRs were identical to those described for the
construction of pMF239 except mutant [gamma]117 instead of [gamma]134 was used (
19
,
20
). For footprinting analysis, these plasmids were 3'- or 5'-end-labeled at the artificial
Eco
RI site (coordinate -106), the naturally occurring
Hin
dIII site (coordinate +1) or
Ase
I site (coordinate +78) as described (
11
), followed by digestion with another restriction enzyme indicated in the figure
legend.
DNase I footprinting was performed as described (
11
,
13
). Samples (5-10 ng) of
32
P-end-labeled DNA were incubated at 25oC for 10 min in 20 [mu]l of the following buffer (20 mM Tris-HCl pH 7.5, 6 mM MgCl
2
, 1 mM EDTA, 166 mM K-glutamate) containing varying amounts of purified [pi] protein. Samples were treated with 1 U DNase I (Boehringer Mannheim), freshly diluted from stock with the
reaction buffer supplemented with 50 mg BSA/ml. Digestion was allowed to
proceed for 30 s at 25oC. The reaction was stopped by the addition of 50 [mu]l 20 mM Tris-HCl pH 7.5, 20 mM EDTA, 0.1% SDS. Samples were treated with
phenol and DNA was ethanol precipitated. DNase I cleavage products were analyzed on 7 M urea-8% polyacrylamide gels.
Hydroxyl radical footprinting was performed according to the published protocol
(
21
). A 50 [mu]l reaction mix contained the same reaction buffer which was used for DNase I footprinting (see above). To each reaction sample 0.6 [mu]l 100 mM sodium ascorbate (Sigma) was added, followed by 4 [mu]l 0.15% (v/v) H
2
O
2
(J. T. Baker Inc.) and 6 [mu]l of a solution of freshly mixed 50 mM Fe(NH
4
)
2
(SO
4
)
2
(Sigma) and 100 mM Na
2
EDTA (Sigma). Cleavage was allowed to proceed for 2 min at 37oC, after which 80 [mu]l 20 mM thiourea (Sigma) and 60 [mu]l 0.3 M NaCl were added to quench the reaction. The samples were
then precipitated with ethanol using 1 [mu]g glycogen (Boehringer Mannheim) as carrier. Cleavage products were analyzed
as described for DNase I footprinting.
The [pi] protein was purified according to the previously published procedure (
11
). Briefly,
Escherichia coli
frozen cell paste, containing overproduced [pi] protein from vector pPT39 (
11
) was thawed (4oC) and suspension adjusted to 100 mM KCl, 5 mM DTT and 200 [mu]g/ml lysozyme. The lysis procedure was as described (
11
). The lysate was centrifuged at 30 000
g
for 30 min and the supernatant was loaded onto a heparin-Sepharose column (0.9 * 5 cm), equilibrated with the buffer containing 20 mM Tris-HCl pH 7.5, 1 mM EDTA, 1 mM DTT with 100 mM KCl. Proteins
were eluted with 75 ml of a linear gradient of KCl (0.1-1.0 M). Fractions containing [pi] protein, identified by SDS-PAGE, were pooled and loaded onto a hydroxylapatite column
(0.9 * 3.0 cm) equilibrated with the buffer containing 20 mM KPO
4
pH 7.0, 50 mM KCl and 1 mM DTT. Proteins were eluted with 30 ml of a linear KPO
4
gradient (0.02-0.50 M). Fractions containing the [pi] protein, which eluted at ~0.4 M KPO
4
,
were aliquoted and stored frozen at -70oC. Purified [pi] protein was at least 95% pure as determined by Comassie blue
staining of polyacrylamide gels obtained by of SDS-PAGE analysis. In some experiments we used [pi] protein extracted from exclusion bodies and purified by an
alternative protocol which was described elsewhere (
22
).
It has been originally observed in DNase I footprinting assays that the area of strong protection at coordinates +93
to +246 and periodic enhancements, each in seven DRs, are produced by [pi] binding to a DNA fragment containing the entire [gamma] origin (
11
). Moreover, another `footprint' noted in the presence of [pi] seems to lie outside the DRs (
11
). This second footprint has been localized within the segment adjacent to DRs
that is A+T-rich; the enhanced DNase I cleavage sites at coordinates +15 and +51 form the boundary of the
protected area. The altered susceptibility of the A+T rich segment to DNase I
cleavage in the presence of [pi] can be caused by one of the several factors or their combination: (i) [pi] protein may bind to this site independently of binding to the other
sites. (ii) [pi] binding to the A+T-rich segment may require prior occupancy of DRs and/or IRs (positive cooperativity). (iii) [pi] binding to the seven DRs and/or IRs may induce a change in the
conformation of the A+T-rich segment altering its susceptibility to the DNase I cleavage. Because previous experiments showing
enhancements in the A+T-rich region used DNA fragments carrying other [pi] sites, we could not distinguish between these possibilities (
1
,
11
). We describe in this paper the exploration of these various possibilities.
Our reference sample was prepared with the
Eco
RI-
Pst
I fragment obtained from plasmid pMF36. This plasmid contains the entire [gamma] origin sequence [nucleotides (nt) -106 to +280; see Fig.
1
]. As shown in Figure
2
, there are several prominent features in the resulting digestion patterns.
First, the area containing cluster of seven 22 bp DRs is protected. A specific
phosphodiester bond (the first A in the TGAGRG motif) is hypersensitive to
DNase I cleavage in each DRs unit forming a characteristic ladder of
enhancements. Secondly, the two positions (+15 and +51) in the A+T-rich segment are hypersensitive to cleavage. Thirdly, in the enhancer
segment, protection is evident from nt -71 to -46, with enhancement at positions -46 and -68. This region contains IRs motif (Figs
1
and
7
). Binding of [pi] to this site has been recently described and it is known that mutating one
or both halves of the IRs abolished [pi] binding (
1
).
The interaction of [pi] with the [gamma] origin can be monitored by areas of protection from and enhancements
of DNase I cleavage (
11
,
12
). We have originally reported that such signals span a large area of the [gamma] origin core encompassing the cluster of seven DRs and the adjacent A+T-rich segment (
1
,
11
). Our present investigations of [pi] interactions with the latter segment have been facilitated by the general
interest in elucidating the function of the A+T-rich segments that are common to many origins replicating via Cairns mode
(
17
,
18
).
In this work we employed the wild-type [gamma] origin and various deletion derivatives, dissecting these
templates with combinations of restriction enzymes. Such a series of fragments
was employed in sensitive footprinting assays. DNase I and hydroxyl radical footprinting data provide compelling evidence for [pi] binding to a site in the A+T-rich [gamma] origin segment. At least
in vitro
, this binding is independent of [pi] binding to previously characterized sites in the core (DRs) and enhancer
segments of the [gamma] origin (IRs). It is noteworthy that [pi] interactions with the A+T-rich segment are manifested primarily by the appearance of the
enhanced cleavages by DNase I and hydroxyl radicals. This is an unusual
feature, particularly for the latter reagent. The enhanced DNase I and hydroxyl
radicals cleavages suggest that [pi] binding causes conformational changes in the minor groove of the A+T-rich segment (
33
). The best of our knowledge enhanced strand scission of specific deoxyriboses
by hydroxyl radicals has been observed only in footprints generated by the
binding of a reconstituted RNA polymerase (containing the [alpha]235 mutant subunits) to the
lac
UV5 promoter (W. Ross, personal communication). The mechanism of the enhanced
susceptibility of DNA to the specific cleavage reagents is not known.
The A+T-rich segment contains at least two known sites for the DNA binding
proteins: the consensus -35 and -10 hexamers [P1 promoter (
23
) and Fig.
1
] and
ihf
1 site can bind
in vitro
purified RNA polymerase (M.F., data not shown) and IHF protein (
19
,
24
,
25
) respectively. The smallest A+T-rich segment tested here and shown to bind [pi] contains the P1 promoter and
ihf
1. However, several facts argue against the possibility that binding of either
of these two proteins could be responsible for the footprinting signals we
report here. (i) The footprinting pattern produced by IHF and/or RNA polymerase
binding are very clearly defined and the cleavage patterns differ from the
pattern observed in the presence of highly purified [pi] protein (
19
,
24
,
25
and data not shown). (ii) Both RNA polymerase and IHF protein interactions with
the A+T-rich segment can be monitored by the gel retardation assay, whereas [pi] binding cannot (
19
,
24
,
25
and data not shown). (iii) The enhancements in the DRs and A+T-rich segment are produced at similar concentration of [pi] (even though those at positions +15 and +51 are weaker). (iv) Similar
footprinting data were obtained in the presence of [pi] protein that was extracted from exclusion bodies and purified by an
alternative protocol (
22
and data not shown). The aforementioned observations argue against the
possibility that the footprint in the A+T-rich segment is produced by binding of RNA polymerase, IHF or a
contaminating protein. It should be also mentioned that the entire A+T-rich origin segment or the part of it to which [pi] binds cannot be deleted/substituted without the loss of the [gamma] origin function (
16
,
26
). This observation remains to be explored to elucidate whether [pi] binding to the A+T-rich segment is essential for the [gamma] origin activity and whether it facilitates activator and/or
inhibitor function of [pi]. However, our own data suggest that [pi] may inhibit replication through binding to the A+T-rich segment of the [gamma] origin. The following facts appear to support this notion.
(i) IHF protein permits [gamma] origin replication at otherwise inhibitory [pi] levels (
6
). (ii) IHF reverses the inhibitory effect of [pi] by binding to
ihf
1 site in the A+T-rich segment of the [gamma] origin and not by binding to the other site (
ihf
2) that lies to the left the DRs cluster (
32
and Fig.
1
). (iii) IHF is not required for
in vivo
and
in vitro
replication dependent on mutant variants of [pi] protein that inhibit replication less than wt [pi] (
6
,
32
). (iv) In the neocarzinostatin protection studies, the IHF protein can bind to
the
ihf
1 site only when added first; when [pi] protein was added first IHF could not bind to
ihf
1 site (
24
). Thus, the IHF binding to the
ihf
1 site is occluded by [pi] protein. This latter result is also consistent with the ability of [pi] protein to bind the A+T-rich segment. Additional experimentation will be required to
support the aforementioned model.
What might be the DNA sequence in the A+T-rich segment to which [pi] binds? To answer this question, we first briefly review some facts
about other well characterized sequences to which [pi] binds (Fig.
7
). It was found that enhanced DNase I cleavage sites occur in each of seven 22
bp DRs at A
8.
Secondly, double mutant [gamma]-117 at positions G
7
, G
9
abolishes [pi] binding to a DR unit when it is either isolated or flanked by wt copies of
DRs (
14
and Fig.
3
). G
7
and G
9
are protected against methylation both
in vitro
and
in vivo
(
15
,
27
). Because these protected/mutated bases are within the TGAGRG motif present in
both DRs and IRs it is reasonable to assume that a single domain of [pi] protein confers binding to these sites. Although the remaining sequences
flanking the TGAGRG core are required for the [pi] protein binding to the 22 bp DRs (Fig.
3
and data not shown), it is not known which of those bp are essential. However,
18 out of 22 bp are fully conserved among all 22 bp repeats (Fig.
7
). Six of those 18 bp are not shared with the IRs in the
pir
gene operator or with the IRs in the enhancer. At least for the IRs in the
enhancer, the major
cis
-acting stabilizing factor appears to be a second half of IRs. This
conclusion is supported by a recent finding that mutating one half-site from TGAGAA to GGGGAA, while leaving the other half site intact,
severely decreased the affinity of wt [pi] protein for both halves of the enhancer IRs (
1
). Since [pi] binds to sites arranged as IRs (symmetric site) or DRs (asymmetric sites),
we have proposed that [pi] protomers must be able to assume different conformations to recognize two
types of information, sequence and geometry of two types of binding sites (
2
,
13
). This hypothesis is strongly supported by the properties of [pi] mutant ([pi]S87N) that can bind DRs but cannot bind IRs (
13
). However, the [pi]S87N can bind, although somewhat weaker than wt [pi], to the A+T-rich in the presence of the enhancer and DRs (data not shown).
We searched for a similarity between the consensus sequence of the 22 bp DRs
unit or IRs unit and the A+T-rich segment to which [pi] binds. The best match found (13 out of 18 bp, does not include nt G
7
whose role in binding of [pi] to a 22 bp DRs unit has been established. Moreover, we could not detect any
significant similarity between the two known pairs of IRs and A+T-rich segment. For this reason it is tantalizing to speculate that [pi] may possess domains recognizing distinct families of DNA sequences.
Another possibility is that the overall A+T richness may relax the stringency
of the sequence to which [pi] binds. This seems unlikely since the aforementioned degenerated DR sequence
is far from the enhancement produced by [pi] binding at the nt +15.
It should be mentioned that the ability of proteins to bind two families of DNA
sequence is not unusual: a bipartite organization was originally discovered for
the Int and A proteins controlling site-specific recombination of [lambda] and Mu phages respectively (
28
,
29
). The DNA boxes and 13mers present in A+T-rich segment of the
oriC
of
E.coli
seem to independently bind purified DnaA protein, even though there is no
sequence similarity between these targets (
30
). Genetic and biochemical data indicate that the Rep IB protein of R1162
plasmid can bind the origin's repeats as well as the A+T-rich segment nearby (
31
).
We thank Wilma Ross for advice in performing hydroxyl radical footprinting and
Frank Wu for help in manuscript preparation. This work was supported by Grant
from National Institutes of Health (GM40314) to M.F.
REFERENCES
Return