Antirepression as a second mechanism of transcriptional activation by a minor groove binding protein.

Abstract

Competence for genetic transformation in the bacterium Bacillus subtilis is a bistable differentiation process governed by the minor groove DNA binding protein ComK. No detectable comK transcription occurs in the absence of an intact comK gene, indicating that ComK has auto-activating properties. ComK auto-stimulation, which is dependent on ComK binding to the comK promoter, is a critical step in competence development, ensuring quick and high-level expression of the late-competence genes. Auto-stimulation is also essential for the bistable expression pattern of competence. Here, we demonstrate that ComK acts as an activator at its own promoter by antagonizing the action of two repressors, Rok and CodY. Importantly, antirepression occurs without preventing binding of the repressing proteins, suggesting that ComK and the repressors might bind at distinct surfaces of the DNA helix. DegU, a DNA binding protein known to increase the affinity of ComK for its own promoter, potentiates the antirepression activity of ComK. We postulate that antirepression is primarily achieved through modulation of DNA topology. Although to our knowledge ComK is the only DNA binding protein shown to act in this novel fashion, other minor groove binding proteins may act similarly.

Full text

Page 1

Antirepression as a second mechanism of
transcriptional activation by a minor groove
binding protein
Wiep Klaas Smits,1Tran Thu Hoa,2†
Leendert W. Hamoen,1‡Oscar P. Kuipers1and
David Dubnau2*
1Department of Genetics, University of Groningen,
Kerklaan 30, 9751NN, Haren, the Netherlands.
2Public Health Research Institute, 225 Warren St,
Newark, NJ 07103-3535, USA.
Summary
Competence for genetic transformation in the bacte-
rium Bacillus subtilis is a bistable differentiation
process governed by the minor groove DNA binding
protein ComK. No detectable comK transcription
occurs in the absence of an intact comK gene, indi-
cating that ComK has auto-activating properties.
ComK auto-stimulation, which is dependent on ComK
binding to the comK promoter, is a critical step in
competence development, ensuring quick and high-
level expression of the late-competence genes.
Auto-stimulation is also essential for the bistable
expression pattern of competence. Here, we demon-
strate that ComK acts as an activator at its own pro-
moter by antagonizing the action of two repressors,
Rok and CodY. Importantly, antirepression occurs
without preventing binding of the repressing pro-
teins, suggesting that ComK and the repressors
might bind at distinct surfaces of the DNA helix.
DegU, a DNA binding protein known to increase the
affinity of ComK for its own promoter, potentiates the
antirepression activity of ComK. We postulate that
antirepression is primarily achieved through modula-
tion of DNA topology. Although to our knowledge
ComK is the only DNA binding protein shown to act in
this novel fashion, other minor groove binding pro-
teins may act similarly.
Introduction
In bacteria, the activation of transcription typically involves
a DNA binding protein that recognizes a sequence
upstream from the RNA polymerase (RNAP) binding site,
resulting in the recruitment of RNAP to the promoter or
stabilization of RNAP binding. Indeed, the latter mecha-
nism obtains in the case of ComK, the master regulator of
competence for genetic transformation (Susanna et al.,
2004).
Competence, the ability to take up DNA from the envi-
ronment and integrate it into the genome, is the end-point
of a complex developmental process resulting in the
expression of ComK (van Sinderen et al., 1995). This
activator drives the expression of a multitude of genes,
including those encoding the DNA uptake and integration
machinery (Berka et al., 2002; Hamoen et al., 2002;
Ogura et al., 2002). In contrast to the majority of DNA
binding regulatory proteins, ComK activates transcription
by binding in the minor groove of the DNA, upstream of its
target genes (Hamoen et al., 1998).
In B. subtilis there is no detectable comK–lacZ expres-
sion in the absence of functional ComK, demonstrating
that ComK has auto-activating properties (van Sinderen
and Venema, 1994). ComK exerts this auto-activation by
direct binding to its own promoter, PcomK (van Sinderen
and Venema, 1994). This binding is stimulated by the
presence of the priming protein DegU, which exerts a
positive effect on competence development (Hamoen
et al., 2000).Although the transcription of the downstream
comG operon does not occur in vitro or in Escherichia coli
unless ComK is provided, the transcription of comK itself
occurs readily under both conditions in the absence of
ComK (van Sinderen et al., 1995; below and Fig. S1).
These apparent discrepancies suggest that ComK acts in
vivo at PcomK by antagonizing the action of one or more
repressor proteins. Three candidate repressors have
been identified that directly repress transcription from the
comK promoter; the transition state regulator AbrB
(Hamoen et al., 2003), the nutritional regulator CodY
(Serror and Sonenshein, 1996) and Rok (Hoa et al.,
2002).
In this study, we demonstrate that ComK acts as an
antirepressor at its own promoter in vivo as well as in vitro,
Accepted 10 February, 2007. *For correspondence. E-mail dubnau@
phri.org; Tel. (+1) 973 854 03400; Fax (+1) 973 854 3401. Present
addresses:†Department of Microbiology, Faculty of Pharmacy, Uni-
versity of Medicine and Pharmacy of Ho Chi Minh City, 41 Dinh Tien
Hoang St., Dist.1, HCMC, Vietnam;‡Institute for Cell and Molecular
Biosciences, The Medical School, University of Newcastle, Framling-
ton Place, Newcastle, NE2 4HH, UK.
Molecular Microbiology (2007) 64(2), 368–381 doi:10.1111/j.1365-2958.2007.05662.x
© 2007 The Authors
Journal compilation © 2007 Blackwell Publishing Ltd

Page 2

relieving Rok- and CodY-mediated repression. Strikingly,
it does so without preventing binding of the repressors.
The antirepression activity of ComK is potentiated by
DegU, which also binds to the promoter of comK. Inter-
estingly, ComK is able to activate transcription of recA by
reversing the repression exerted by LexA (Hamoen et al.,
2001). ComK thus activates transcription in two distinct
manners; as a classical transcriptional activator at down-
stream competence genes such as comG, and as an
antirepressor at PcomK and PrecA. To our knowledge,
ComK is the only DNA binding protein proven to reverse
transcriptional repression without eliminating the binding
of the repressors, although this novel mechanism may
reflect a property of other proteins that bind through the
minor groove of DNA.
Results
Rationale for the antirepression hypothesis
ComK acts as a transcriptional activator at PcomK and at
downstream competence promoters, such as the pro-
moter of comG (PcomG). ComK is required for in vitro
transcription from PcomG (Hamoen et al., 1998; Susanna
et al., 2004) but when PcomK was incubated with nucle-
otides and RNAP in the absence of ComK, strong tran-
scription was observed (see below). Additionally, a fusion
of the lacZ coding sequence to PcomK was readily
expressed in a heterologous host (E. coli), whereas a
PcomG–lacZ fusion was not, unless comK was coex-
pressed (van Sinderen et al., 1995; Susanna et al., 2006).
The observed expression from PcomK in E. coli could
indicate that ComK acts as an antirepressor in B. subtilis,
antagonizing one or more species-specific repressors
(Supplementary Fig. S1). In fact, the E. coli genome
(Blattner et al., 1997) contains no homologues of the pre-
viously identified repressors of comK (CodY, AbrB and
Rok), as judged by a BLAST similarity search (http://www.
ncbi.nlm.nih.gov/blast/). The hypothesis of antirepression
makes a strong prediction; that removal of repressors
would render transcription from PcomK in B. subtilis at
least partially independent of ComK. To test this, we intro-
duced mutations in the genes encoding these repressors,
and evaluated the expression of an ectopic comK–lacZ
reporter fusion.
Repressor mutations result in a partial bypass of the
ComK requirement in vivo
The requirement for ComK was not detectably bypassed
by inactivation of codY or abrB, either individually or in
combination (Fig. 1 and W.K. Smits et al., unpubl. obs.).
Rok was recently identified as a major repressor of com-
petence (Hoa et al., 2002). Therefore, we evaluated the
effect of a rok mutation on PcomK expression in wild type
(Fig. 1A) and comK backgrounds (Fig. 2B). In the pres-
ence of ComK, the deletion of rok leads to high levels of
transcription, as was previously reported (Hoa et al.,
2002). Importantly, there is significant comK–lacZ activity
in the DcomK Drok background, although it is about three-
to fourfold lower than wild type and seven- to eightfold
reduced compared with a comK+background carrying a
rok mutation. This result indicates that a rok mutation
partially bypasses the in vivo requirement of ComK for its
own expression. Interestingly, the bypass is stronger
when both codY and rok are inactivated (Fig. 1B), despite
the fact that a codY mutation alone does not lead to a
detectable bypass.
Taken together, these results indicate that ComK
reverses the action of Rok and strongly suggests a similar
role for ComK towards CodY, thus supporting the hypoth-
esis that ComK acts as antirepressor at its own promoter
in vivo. The partial nature of the bypass suggests either
that ComK plays a role at PcomK as a classical activator
Fig. 1. Mutation of rok and codY results in a bypass of the ComK
requirement. Strains containing the comK–lacZ reporter were either
comK+(A) or DcomK (B). In both panels wild-type expression
levels of comK–lacZ (i.e. in comK+background) are indicated using
solid symbols. Time is given in hours relative to the transition
between exponential and stationary growth phase (T0). A comK
mutation in an otherwise wild-type strain completely abolishes
comK–lacZ expression (not shown). The apparent delay of a rok
codY double mutant in comK+background compared with wild type
was not consistently observed.
ComK reverses Rok and CodY repression 369
© 2007 The Authors
Journal compilation © 2007 Blackwell Publishing Ltd, Molecular Microbiology, 64, 368–381

Page 3

as well as an antirepressor protein or that an additional
repressor is involved. Such an additional repressor is
unlikely to be AbrB, for reasons presented in the
Discussion.
Rok and CodY repress transcription from PcomK in vitro
In order to further analyse the effects of individual tran-
scription factors, we established in vitro transcription
assays for the promoter of comK using purified B. subtilis
RNAP holoenzyme (a kind gift from M. Salas). A template
for these experiments was constructed by cloning a
186 bp fragment derived from sequences upstream of
comK into plasmid pAN583 (Predich et al., 1992), as
described in Experimental procedures. The resulting
plasmid contains the sequence of the comK promoter that
is required for binding of all known transcriptional regula-
tors of the comK gene (Serror and Sonenshein, 1996;
Hamoen et al., 1998; Hamoen et al., 2000; 2003; Hoa
et al., 2002).
In contrast to the results obtained with pAN-G, a
PcomG-containing derivative of pAN583 (Hamoen et al.,
1998), there is a substantial level of transcription from
PcomK under the conditions used (Figs 2 and 3, inset
lanes marked with X). This level of transcription was not
significantly augmented by the addition of ComK or DegU
over a wide range of concentrations, either alone or in
combination (Fig. 2A and B). A small but consistent
increase (around 1.5-fold) was observed when only
ComK was added to the reaction mixture, but this effect
was not observed in the presence of DegU. In both cases,
a decrease in transcription was evident at higher concen-
trations of ComK protein. A similar decrease with high
ComK concentrations was observed when pAN-G was
used as a template for the in vitro transcriptions (data not
shown). The reason for this is not clear, but it may reflect
non-specific inhibition by this DNA binding protein at high
concentrations. These experiments indicate that strong
auto-stimulation from PcomK is absent when ComK is
added to the mixture, either alone or in the presence of
DegU.
Next, we investigated the effects of Rok and CodY on
the in vitro transcription from PcomK. The addition of Rok
leads to strong repression, with no detectable transcrip-
Fig. 2. ComK reverses the repression by Rok in vitro. In vitro
transcription experiments using circular pAN-K plasmid (see
Experimental procedures) and purified B. subtilis RNA polymerase
in the presence of several transcription factors. All experiments
were performed at least in triplicate and representative examples
are shown in the figure. The experiments were performed in the
presence (open symbols) or absence (closed symbols) of DegU
protein.
A. Results from in vitro transcription experiments with ComK
(4–1100 nM). The inset shows signals from reactions containing no
transcription factors (X), 137.5 nM ComK (K) or 137.5 nM
ComK/500 nM DegU (UK).
B. Results from in vitro transcription experiments with Rok
(12–3000 nM) and DegU (14–3500 nM). The inset shows signals
from reactions containing no transcription factors (X), 750 nM Rok
(R), 750 nM Rok/500 nM DegU (RU) or 500 nM DegU (U).
C. Results from in vitro transcription experiments in the presence of
170 nM Rok, various amounts of ComK protein. The dash-dotted
line indicates the level of transcription under these conditions in the
absence of ComK and DegU. The experiments were performed in
the presence (open symbols) or absence (closed symbols) of DegU
protein (500 nM) and ComK. The inset shows signals from
reactions containing no transcription factors (X), 170 nM Rok (R),
137.5 nM ComK/170 nM Rok (KR), 500 nM DegU/170 nM Rok (UR)
or 137.5 nM ComK/500 nM DegU/170 nM Rok (KUR).
370W. K. Smits et al.
© 2007 The Authors
Journal compilation © 2007 Blackwell Publishing Ltd, Molecular Microbiology, 64, 368–381

Page 4

tion at nanomolar concentrations of Rok protein (Fig. 2B).
These values are in good agreement with electrophoretic
mobility shift assay (EMSA) results showing an apparent
KDof around 50 nM for the interaction of Rok with PcomK
(Hoa et al., 2002; Albano et al., 2005). The addition of
DegU slightly enhanced the repression by Rok at lower
concentrations, but had no effect when enough Rok was
added to reduce transcription to less than 20% of the
unrepressed state (Fig. 2B).
The binding of CodY to the promoter of comK has been
described previously (Serror and Sonenshein, 1996).
However, the effects of the recently identified cofactors
guanosine triphosphate (GTP) (Ratnayake-Lecamwasam
et al., 2001) and branched chain amino acids (BCAAs)
(Shivers and Sonenshein, 2004) have not been docu-
mented for PcomK. Although binding of CodY occurs in
the absence of cofactors, despite the presence of an
excess of non-specific competitor (poly-dIdC), we found
that BCAAs stimulated binding of CodY to the ComK
promoter (data not shown), in agreement with results
obtained with other promoters (Shivers and Sonenshein,
2004). In contrast, GTP only stimulated binding detectably
in the presence of BCAAs (data not shown). However,
these effects on binding were relatively minor; affinity in
the presence of both effector molecules increased only
approximately twofold under our experimental conditions.
In spite of this, we observed no transcriptional repression
with CodY in the absence of effectors, under conditions
that resulted in a fully retarded PcomK probe in the EMSA
experiments (Fig. 3A). The addition of a mixture of BCAAs
resulted in strong repression, with a cumulative repres-
sive effect in the presence of GTP (Fig. 3A). We conclude
that the effector molecules have a minor effect on the
affinity of CodY for its target promoter, but are required for
the repressor activity of CodY. The concentration of CodY
required to achieve full repression is in good agreement
with both EMSA experiments and with the concentration
used for previously reported footprinting experiments
(Serror and Sonenshein, 1996). As in the case of Rok, we
observed that the addition of DegU to the in vitro tran-
scription reactions exerted only a mild effect on CodY-
mediated repression under the conditions tested (data not
shown).
Rok reduces the affinity of RNAP for PcomK
Many repressors act by masking the -35 or -10 promoter
elements that constitute the RNAP binding site. Indeed,
DNase I footprinting experiments have revealed thatAbrB
and CodY bind to PcomK at sequences that partially
overlap these elements (Serror and Sonenshein, 1996;
Hamoen et al., 2003). However, the location of the binding
site for Rok is unknown. Despite repeated attempts, we
were unable to further define the binding site for Rok
using DNaseI footprinting experiments or bioinformatic
approaches (Albano et al., 2005). Using Rok and overlap-
ping PcomK-probes in EMSA experiments, we narrowed
the region encompassing the Rok binding site to 130 bp
(Fig. 4). This segment includes the sequences that are
protected by ComK and DegU in hydroxyl radical footprint
experiments (Hamoen et al., 1998; 2000). Surprisingly,
the Rok-binding site appears to lie upstream of the -35
and -10 sequences, making it unlikely that Rok occludes
Fig. 3. ComK reverses the repression by CodY in vitro. In vitro
transcription experiments using circular pAN-K plasmid (see
Experimental procedures) and purified B. subtilis RNA polymerase
in the presence of several transcription factors. All experiments
were performed at least in triplicate and representative examples
are shown in the figure.
A. Results from in vitro transcription experiments with a titration of
CodY protein, in the presence and absence of effector molecules.
GTP, guanosine triphosphate; ILV, a mixture of isoleucine, leucine
and valine. The inset shows signals from reactions containing no
transcription factors (X), 8 mM of CodY (Y), 2.5 mM GTP/8 mM
CodY (GTP Y), 10 mM ILV/8 mM CodY (ILV Y) or 10 mM
ILV/2.5 mM GTP/8 mM CodY (ILV GTP Y).
B. Results from in vitro transcription experiments in the presence of
8 mM CodY, 10 mM of isoleucine, leucine and valine (ILV) and
various amount of ComK protein. The dash-dotted line indicates the
level of transcription under these conditions in the absence of
ComK or ComK and DegU. The experiments were performed in the
presence (open symbols) or absence (closed symbols) of DegU
protein (500 nM) and ComK. The inset shows the signals from
reactions containing no transcription factors (X), 8 mM CodY (Y),
137.5 nM ComK/8 mM CodY (KY), 500 nM DegU/8 mM CodY or
137.5 nM ComK/500 nM DegU/8 mM CodY.
ComK reverses Rok and CodY repression371
© 2007 The Authors
Journal compilation © 2007 Blackwell Publishing Ltd, Molecular Microbiology, 64, 368–381

Page 5

the RNAP binding site in the comK promoter. Considering
its function as a repressor, we were therefore interested to
see whether Rok reduces the affinity of RNAP for PcomK,
despite the location of its binding site.
EMSA experiments using purified RNAP and Rok dem-
onstrated that Rok reduces the binding of RNAP (Fig. 5).
The concentration of Rok required for half-maximal rever-
sal of RNAP binding is comparable to the apparent KDof
Rok-binding to PcomK in the absence of RNAP. Although
this reversal would appear to be consistent with direct
competitionbetweenRNAPandRokforbindingtoPcomK,
we also consistently observed a slightly super-shifted
protein/DNA complex in the presence of both RNAP and
Rok, evident when intermediate concentrations of Rok are
used.Thissuper-shiftedbandmigratesslightlyslowerthan
thecomplexesoftheDNAwithRNAPorRokalone(Fig. 5).
It thus appears possible that, in contrast toAbrB and CodY,
RokdoesnotcompetedirectlywithRNAPforbindingtothe
DNAbut relies on an alternative mechanism to reduce the
affinity of RNAP for PcomK. Although this would be con-
sistent with the unusual location of the Rok repressor
binding site noted above and is not unprecedented (Perez-
Martin et al., 1994; Rojo, 2001), it is a suggestion that
deserves further confirmation.
ComK reverses repression in vitro
These experiments enabled us to directly test the antire-
pression hypothesis by determining whether ComK acts
in vitro to reverse Rok and CodY repression. At 170 nM
Rok, transcription from PcomK was reduced sixfold
(Fig. 2B and C, dash-dotted line). The addition of ComK to
the repressed promoter resulted in a moderate increase
in transcription, although transcription was submaximal
at concentrations greater than ~130 nM ComK protein
(Fig. 2C). The observed decrease in transcription at high
levels of ComK may be non-specific, as noted above, and
was not investigated further. Previously, it was reported
that DegU acts as a priming protein, enhancing the affinity
of ComK for its own promoter (Hamoen et al., 2000).
Interestingly, this phenomenon is independent of protein-
protein contacts, and presumably relies on modulation of
Fig. 4. Schematic representation of the
binding sites of various transcription factors
on the promoter of comK to indicate overlap
and relative positions of the various sites. K,
ComK; U, DegU; Y, CodY; A, AbrB; R, Rok.
-35 and -10 indicate the core promoter
elements. The comK gene is truncated.
Overlapping probes for EMSA experiments
were generated with various combinations of
primers (see Table S2). These probes were
tested for their ability to bind Rok according to
Experimental procedures (indicated in the
column Shift). The inferred area of
Rok-binding is indicated by the shaded box.
Fig. 5. Rok reduces affinity of RNA
polymerase for PcomK. The percentage of
probe in the complex with RNA polymerase
with or without Rok (grey) or Rok only (black)
was determined as described in Experimental
procedures. The lower part of the figure
shows the actual EMSA experiment using a
fragment of the comK promoter. The big
triangle indicates a decrease in Rok
concentration of twofold per lane, ranging
from 1.7 mM to 13 nM. The small triangles on
the image of the gel indicate the shift (closed
symbol) or super-shift (open symbol) caused
by Rok binding. Binding of proteins was
performed as described in Experimental
procedures. Free probe and complex were
resolved on a non-denaturing 6%
polyacrylamide gel. Rp, RNA polymerase
holoenzyme (1.5 mg).
372 W. K. Smits et al.
© 2007 The Authors
Journal compilation © 2007 Blackwell Publishing Ltd, Molecular Microbiology, 64, 368–381