Structure and function of bacterial kid-kis and related toxin-antitoxin systems.

Abstract

Toxin-antitoxin systems were discovered as plasmid auxiliary maintenance cassettes. In recent years, an increasing amount of structural and functional information has become available about the proteins involved, allowing the understanding of bacterial cell growth inhibition by the toxins on a molecular level. A well-studied TA system is formed by the proteins Kid and Kis, encoded by the parD operon of the Escherichia coli plasmid R1. The toxicity of Kid has been related to its endoribonuclease activity, which is counteracted by binding of the antitoxin Kis at the proposed active site. In this review, the structural studies on the Kid-Kis system are compared to those of three related toxin-antitoxin systems: MazF-MazE, CcdB-CcdA and RelE-RelB.

Full text

Page 1

Protein & Peptide Letters, 2007, 14, 113-124
113
0929-8665/07 $50.00+.00 © 2007 Bentham Science Publishers Ltd.
Structure and Function of Bacterial Kid-Kis and Related Toxin-Antitoxin
Systems
Monique B. Kamphuis1, Maria Chiara Monti2,‡, Robert H. H. van den Heuvel2,
Juan López-Villarejo3, Ramón Díaz-Orejas3 and Rolf Boelens1,*
1Bijvoet Center for Biomolecular Research, Department of NMR Spectroscopy, Utrecht University, Padualaan 8, 3584
CH Utrecht, The Netherlands; 2Bijvoet Center for Biomolecular Research, Department of Biomolecular Mass Spec-
trometry, Utrecht University, Sorbonnelaan 16, 3584 CA Utrecht, The Netherlands; 3Centro de Investigaciones Biológi-
cas, Departamento de Microbiología Molecular, Ramiro de Maeztu 9, E-28040 Madrid, Spain
Abstract: Toxin-antitoxin systems were discovered as plasmid auxiliary maintenance cassettes. In recent years, an in-
creasing amount of structural and functional information has become available about the proteins involved, allowing the
understanding of bacterial cell growth inhibition by the toxins on a molecular level. A well-studied TA system is formed
by the proteins Kid and Kis, encoded by the parD operon of the Escherichia coli plasmid R1. The toxicity of Kid has been
related to its endoribonuclease activity, which is counteracted by binding of the antitoxin Kis at the proposed active site.
In this review, the structural studies on the Kid-Kis system are compared to those of three related toxin-antitoxin systems:
MazF-MazE, CcdB-CcdA and RelE-RelB.
Not For Distribution
Keywords: CcdB-CcdA, MazF-MazE, plasmid maintenance, RelE-RelB, ribonuclease.
TOXIN-ANTITOXIN SYSTEMS

and plasmids takes place prior to cell division. Although
plasmids are not crucial for the proliferation of bacteria, the
valuable genetic information they carry can be of great bene-
fit for the bacterial cell line in particular circumstances. An-
tibiotic resistance factors confer to their bacterial hosts the
capacity to grow in the presence of antibiotics, and catabolic
plasmids in Pseudomonas increase the nutritional versatility
of the cells by allowing the utilisation of new substances, for
instance aromatic compounds, as source of carbon and en-
ergy. In spite of the non-essential character of plasmids,
these extrachromosomal genetic elements are maintained
stably in the cells by a variety of auxiliary stability systems.
Some of these systems prevent the growth of plasmid-free
progeny via a programmed cell death mechanism called
post-segregational killing (PSK) [1-4].
During bacterial growth, replication of the chromosome
PSK-Inducing Plasmids

that function in a highly similar way. They contain genes that
encode a stable protein toxin and an unstable antitoxin that
can either be an antisense ribonucleic acid (RNA) or a pro-
tein antitoxin [2,5]. The antitoxin inhibits the activity of the
toxin. However, plasmid loss during cell division is lethal,
since the unstable antitoxin cannot be replenished upon deg-
radation [3,4,6]. In the antisense RNA-regulated killing sys-

*Address correspondence to this author at the Department of NMR Spec-
troscopy, Utrecht University, Padualaan 8, 3584 CH Utrecht, The Nether-
lands; Tel: +31 30 253 4035/2652; Fax: +31 30 253 7623; E-mail:
r.boelens@chem.uu.nl
Two different types of PSK-inducing plasmids are known
‡Present address: Dipartimento di Scienze Farmaceutiche, Universita’ di
Salerno, Via Ponte Don Mellilo, 84084, Fisciano, Salerno, Italy
tems, the toxin is only synthesised when the antitoxin is ab-
sent, because only then translation of the inherited toxin-
encoding messenger RNAs (mRNAs) can take place. Subse-
quently, the produced toxins cause bacterial death by damag-
ing the bacterial cell membrane [4,7]. The protein antitoxin-
regulated systems are called plasmid addiction modules or
more generally toxin-antitoxin (TA) systems. In these sys-
tems, the toxin is already present during normal growth of
bacteria. The antitoxin neutralises the toxicity by forming a
protein complex with the toxin. Upon loss of the plasmid and
therefore a diminished level of antitoxin, the free toxin will
exert its lethal potential [3,5,6,8].
TA Systems Are Also Present on Chromosomes

they are also highly abundant in bacterial as well as archaeal
genomes [6,9]. In fact, a large conservation exists between
several plasmid- and chromosome-encoded systems. As al-
ready described for plasmid-encoded TA systems, also
chromosomal TA systems are involved in cell growth arrest.
However, for the latter systems the toxin activity does not
necessarily result in cell death; the expression of toxicity can
be reversed since new antitoxin can be synthesised [6,10-14].
Several systems are thought to function in quality control of
gene expression and to provide a control mechanism for free-
living prokaryotes to cope with nutritional stress [13,15,16].
Also, a relation between bacterial TA systems and the eu-
karyotic nonsense-mediated RNA decay system has been
suggested [9,17].
Originally, TA systems were discovered in plasmids, but
Applications of TA Systems
Interest in bacterial toxins is given by their potential use
in the development of new antibiotics [6,10,15]. These anti-
biotics could mimic the lethal action of the toxins or trigger

Page 2

114 Protein & Peptide Letters, 2007, Vol. 14, No. 2 Kamphuis et al.
the activation of toxins already present in bacterial cells. The
latter option appears very promising since many prokaryotes
carry multiple TA loci, including pathogens such as Myco-
bacterium tuberculosis and Vibrio cholerae. Furthermore,
several bacterial TA systems were reported to function in
yeast and induce apoptosis in human cells [18-20]. Possibly,
these toxins can be converted into medicines that can prevent
uncontrolled division of malignant cells in humans. Another
valuable application of TA systems is their use in biotechni-
cal selection vectors. Like many antibiotics, the toxins are
used to achieve high selective efficiency of recombinants
[21-23].
Five of the Best-Known TA Systems
Not For Distribution

completely sequenced prokaryotic genomes, 671 loci were
found that could be grouped in seven known TA families
[15]. These are, in order of discovery of the plasmid-encoded
systems: the ccd locus of plasmid F [24], the parD (pem)
locus of R1/R100 [25], the vapBC locus of the Salmonella
Dublin virulence plasmid [26], the phd/doc locus of plasmid
P1 [27], the parDE locus of plasmid RK2 [28], the higBA
locus of plasmid Rts1 [29], and the relBE locus of plasmid
P307 [30]. The chromosomal TA systems mazEF and chpB
are members of the parD/pem family [31,32]. MazEF, identi-
fied being a TA system by the group of Ohtsubo, was ini-
tially called chpA which refers to chromosomal homologue
of pem [32]. Chromosome-encoded relBE, yefM-yoeB and
dinJ-yafQ are chromosomal homologous of the plasmid-
encoded relBE system and therefore members of this family
[9,16,33]. However, it has been discussed that yefM-yoeB
and relBE may belong to different families as their toxins
and antitoxins differ substantially either functionally or
structurally (Cherny and Gazit, 2004). The RelE, ParE and
HighB toxins form a superfamily of toxins [9], and the
Kid/PemK and CcdB toxins form another superfamily [34].
An orphan family of the TA systems that is unrelated to the
seven previously described families was identified in plas-
mid pSM19035 of Streptococcus pyogenes and comprises
the toxin ?, the antitoxin ? and a third component, the ?
regulator [35,36].
This review describes five related TA systems, all found
in E. coli: the plasmid-encoded Kid-Kis system, its chromo-
somal homologue the MazF-MazE system, the CcdB-CcdA
system, the RelE-RelB system, and finally the YoeB-YefM
system. In recent years, the biochemical knowledge about
these TA systems has increased enormously and structural
information about all five toxins and antitoxins has become
available [34,37-43]. Fig. 1 contains the 3D structures of the
toxins Kid and CcdB shown in their unbound state, while
RelE, YoeB, and MazF are depicted in complex with their
antitoxins RelB, YefM, and MazE, respectively.
In an exhaustive and recent search for TA loci in 126
THE Kid-Kis SYSTEM

toxin Kis (Killing suppressor; 9.5 kD) are encoded by the
parD locus of the E. coli plasmid R1 [25]. The proteins are
identical to PemK and PemI, respectively, of the pem (plas-
mid emergency maintenance) locus of plasmid R100 [44].
Another parD-homologous locus has been identified on
The toxin Kid (Killing determinant; 11.9 kD) and anti-
plasmid R466B of Morganella morganii [16]. The sequence
of this toxin is almost identical to that of Kid, since only four
out of 110 amino acids are different: V64A, N99D, E100D,
and S106A.
Common Phenomena

coli and Kis counteracts this effect [45]. The coordinate ac-
tion of Kid and Kis has been shown to autoregulate parD at
the level of transcription [46-49]. This kind of transcriptional
control is generally observed for TA systems. The antitoxin
represses transcription [33,47,49-51] by binding the operator
with the N-terminus [48,52,53]. The toxin acts as a co-
repressor, since the TA complex has a higher DNA-affinity
than the antitoxin alone [33,46,53,54]. Another phenomenon
common for TA systems is the constant degradation of the
unstable antitoxin by proteases. The half-lives of antitoxins
are in the range of 15 to 60 minutes in vivo [55,56]. A mutant
of the Lon protease prevents the activation of the toxin Kid,
which indicates that Kis is degraded by the native Lon pro-
tein [57].
Kid has been demonstrated to inhibit cell growth in E.
Structures and Interactions

symmetric dimer, as shown in Fig. 1. Each spherical mono-
mer consists of a five-stranded twisted antiparallel ?-sheet
and a 12 residue C-terminal ?-helix with a small three-
stranded antiparallel ?-sheet inserted in the main sheet, two
additional ?-helices of seven and three residues, and two
loop regions [34,58]. A salt-bridge was observed between
R85 and E18, linking the two Kid monomers [59]. The sec-
ondary structure elements of both Kid and Kis in solution
were determined using NMR spectroscopy (Fig. 2) [43]. A
high structural similarity exists between the crystal and solu-
tion states of Kid, between Kid and MazF and between Kis
and MazE (Fig. 2). The interactions between Kid and Kis
that are responsible for neutralisation of Kid toxicity and
enhance autoregulation of parD transcription have been in-
vestigated [43]. Native macromolecular mass spectrometry
data have demonstrated that Kid and Kis can form multiple
complexes, depending on their precise molar ratios. At
Kis:Kid ratios equal to or exceeding 1:1, as found in vivo in a
plasmid-containing cell, various complexes are present: from
Kid2-Kis2 tetramer up to Kis2-Kid2-Kis2-Kid2-Kis2 decamer.
When Kid is in excess of Kis, corresponding to an in vivo
situation immediately after loss of the plasmid, the 67 kD
Kid2-Kis2-Kid2 heterohexamer is the most abundant species.
NMR titration experiments showed that the interaction sites
of Kid and Kis resemble those within the MazF2-MazE2-
MazF2 complex, which will be described later [38,43]. Fig. 3
shows the Kid-Kis interactions mapped on a MazF-MazE
based model and the crystal structure of the Kid dimer. It
was reported that the affected Kid residues K83 and A84,
located in the small ?-helix within interaction site 1, might
alternatively be part of the other Kid monomer compared to
the situation shown here [60]. Overproduction of MazE has
been demonstrated to neutralise Kid toxicity to a certain ex-
tent in the absence of native Kis. It was shown that Kid2-
MazE2 tetramers can be formed via weak interactions involv-
ing a limited part of the Kis-binding residues of Kid [43].
The results imply that interactions between Kid and the N-
The crystal structure of the Kid protein reveals a two-fold

Page 3

Structure and Function of Bacterial Kid-Kis Protein & Peptide Letters, 2007, Vol. 14, No. 2 115































Figure 1. Ribbon representations of the crystal structures of the toxins Kid (PDB: 1M1F) and CcdB (PDB: 3VUB) and the toxin-antitoxin
complexes MazF-MazE (PDB:1UB4), aRelE-aRelB (PDB: 1WMI) and YoeB-YefM (PDB: 2A6Q). The dimeric Kid and CcdB proteins are
shown with their secondary structure elements highlighted (?-helices coloured red/yellow and ?-strands cyan). For the heterohexameric
MazF2-MazE2-MazF2 complex, the heterodimeric RelE-RelB complex, and the heterotrimeric YoeB-YefM complex, the toxin monomers are
coloured dark/light blue and the antitoxin monomers dark/light yellow. The figures were generated with the program MOLMOL version 2K.1
[109].
terminal part of Kis, known to be involved in autoregulation
[48], likely induce a proper antitoxin-to-toxin orientation and
antitoxin monomer-monomer stabilisation, which both seem
to be important for effective transcriptional repression of the
parD operon.
Mode of Action of the Kid Toxin

extracts, and recently it was demonstrated that in vitro, the
Kid has been reported to inhibit protein synthesis in cell
toxin acts as an endoribonuclease able to cleave RNA in the
absence of ribosomes. Kid cleaves RNA preferentially at the
5’ side of the A residue in the nucleotide sequence 5’-
UA(A/C)-3’ of single-stranded regions, although cleavage at
the 3’ side of the adenosine has been observed as well
[61,62]. Recent experiments by De la Cueva-Méndez and co-
workers have indicated that the 5’-UUACU-3’ sequence is a
longer, more specific target of Kid [63]. Cleavage of the
dinucleotide UpA and the 5-nucleotide RNA fragment
Not For Distribution

Page 4

116 Protein & Peptide Letters, 2007, Vol. 14, No. 2 Kamphuis et al.



























Figure 2. Sequence alignments and secondary structure elements of (a) the structurally related toxins Kid, MazF and CcdB (PDB: 1M1F,
1UB4 and 3VUB), and (b) the structurally related antitoxins Kis and MazE (PDB: n.a. and 1UB4). ?-helices are shown as light grey cylinders
and ?-sheets as dark grey arrows. Very light grey cylinders represent 310-helices for CcdB and helical turns for Kis. The method used to de-
termine the secondary structure, i.e. NMR, NMR/TALOS or X-ray spectroscopy, is indicated. The sequence alignments are shown according
to Kamada et al. (2003).
AUACA yields two products with an uracyl 2’:3’-cyclic
phosphate at one side and an adenosine with a free 5’-OH
group at the other side [60]. The uracyl 2’-OH group is es-
sential for cleavage [60], like in the cleavage mechanisms of
the rather distinct RNases A and T1. RNA cleavage by these
two ribonucleases is well-studied and illustrated by Fig. 4: In
a transphosphorylation reaction the 2’-hydroxyl group is
deprotonated by a catalytic base to perform a nucleophilic
attack on the electrophilic phosphorus. A catalytic acid do-
nates a hydrogen atom to the 5’-leaving group. In the readily
reversible second step, a 3’-monophosphate nucleotide is
formed by hydrolysis of the 2’:3’-cyclophosphate [64-67].
Inouye and co-workers concluded that Kid cleaves only sin-
gle-stranded cellular mRNAs [62]. However, Díaz-Orejas
and co-workers reported that also efficient cleavage of some
target sequences located in double-stranded RNA regions
and primer RNA takes place, and that Kid likely inhibits
RNA-regulated cellular processes in general [61]. It was
shown that Kid becomes active in plasmid-containing cells
upon the decrease of the R1 copy number, instead of only in
plasmid-free cells as originally thought, and that in this situa-
tion the R1 plasmid copy number is restored [68]. More re-
cently it was found that cleavage by Kid of a 5´-UUACU-3´
sequence located close and 3´ to the CopB mRNA region can
lead to decreased levels of the CopB mRNA and therefore of
the CopB protein that limits initiation of R1 replication, thus
favouring increased replication of this plasmid [63].
RNA Binding and Cleavage by Kid
The mechanism of RNA binding and cleavage by the Kid
toxin has been determined in considerable detail via NMR
titration studies with the uncleavable single-stranded RNA
Not For Distribution

Page 5

Structure and Function of Bacterial Kid-Kis Protein & Peptide Letters, 2007, Vol. 14, No. 2 117
mimic AdUACA [60]. It was demonstrated that residues of
both monomers of the Kid dimer are needed to form a RNA
binding surface. These residues cluster into five sequence-
based groups: cluster #1 contains residues 10-23, cluster #2
residues 36-41, cluster #3 residues 47-59, cluster #4 residues
69-76 and cluster #5 residues 82 and 83. Clusters #1, #3 and
#4 of monomer A cover a concatenated surface area together
with clusters #2 and #5 of monomer B, forming the first
RNA binding site, and vice versa for the second binding site.
This implies that Kid is toxic for bacterial cells only in its
dimeric form, in contrast to RNase T1 and RNase A. The
stoichiometry of the Kid2-RNA complex was established to
be 1:1, although Kid is a symmetric dimer. Comparison of
the RNA and Kis binding sites of Kid (Fig. 3) clearly points
out, that upon binding of the antitoxin, one of the RNA bind-
ing sites will be partly occupied by the C-terminal tail of the
protein [43]. This result is in agreement with mutagenesis
data showing the antidote function of this part of Kis [48].
The associated “opening” of the loop between ?-strands 1
and 2 will disrupt the second RNA binding pocket of Kid
[60], thus explaining the way in which Kis prevents RNA
cleavage by Kid [61,62]. A detailed picture of the position of
the 5-nucleotide RNA fragment AUACA within the nucleo-
tide binding pocket of Kid (Fig. 3) has been obtained via
docking calculations performed
HADDOCK [60,69]. Those calculations were based on the
NMR chemical shifts, the cleavage reaction and mutagenesis
data showing that two clusters of residues of Kid
(V9/E18/G21/V25 and G70/R73/D75/D81/R85/P94) are
important for the toxicity of the protein [59,60]. It was pro-
posed that residues D75, R73 and H17 form the active site of
the Kid toxin (Fig. 4). With help of R73, the catalytic base
D75 deprotonates the 2’-OH group of uracyl, which subse-
quently performs a nucleophilic attack on the electrophilic
phosphorus. The transphosphorylation reaction is complete
after donation of a hydrogen atom by the catalytic acid R73
to the adenosine 5’-O. H17 stabilises the complex, and the
RNA sequence specificity is defined by direct interactions of
Kid residues T46, S47, A55, F57, T69, V71 and R73 with
the RNA nucleotides UA(A/C) [60].
with the program
Kid-Ligand Interactions

AMP, 5’-UMP, 5’-CMP, 5’-GMP, 2’:3’-cUMP, Pi, the
minimal substrate UpA, and Mg2+ were studied using NMR
spectroscopy and mass spectrometry. The RNA cleavage
product 2’:3’-cUMP was found to be the most potent Kid
inhibitor among those ligands [70]. In the presence of 0.5
mM 2’:3’-cUMP, Kid cleaves the RNA oligonucleotide with
a catalytic efficiency of 0.60 M-1 s-1 instead of 0.78 M-1 s-1. If
any inhibition takes place by the other ligands, this effect is
very weak and does not result in a significant decrease of the
Kid activity. The Kid residues involved in protein-ligand
interactions were identified via NMR titration experiments
[70]. Addition of 2’:3’-cUMP resulted in the largest chemi-
cal shift changes, showing a nice correlation between the
size of the NMR chemical shift perturbations and the inhibi-
tory effect of the compounds. The addition of any of the five
mononucleotides resulted in a chemical shift perturbations
pattern that is, although weaker, almost identical to the effect
of AdUACA binding (Fig. 5). All Kid residues suggested
The interactions of Kid with the potential inhibitors 5’-
being involved in interactions with the target sequence UAC
(H17, R23, T37, T46, S47, A55, F57, T69, V71, R73 and
D75) were observed in this Kid-inhibitor interaction study.
Interestingly, the catalytic acid R73 only shows a perturba-
tion upon addition of 2’:3’-cUMP, and the catalytic base
D75 only when 2’:3’-cUMP or 5’-CMP is added. Clearly,
the cleavage product 2’:3’-cUMP shows the strongest inter-
actions with the RNA binding pocket and catalytic residues
of Kid. Since 2’:3’-cUMP also inhibits the endoribonuclease
activity of the toxin, it appears to be the most promising lead
for designing an effective regulator of Kid activity in both
prokaryotic and eukaryotic cells.
THE MazF-MazE SYSTEM

are chromosomal homologues of Kid and Kis, respectively.
They are encoded by the E. coli mazEF locus (‘ma-ze’ is
Hebrew for ‘what is it?’) and are also known as ChpAK re-
spectively ChpAI [6,31,32]. Sixty-six other chromosomal
mazEF loci have been identified in both gram-negative and
gram-positive organisms. Publications include the charac-
terisation of homologous TA systems in Bacillus subtilis
[71,72], Streptococcus mutans [73] and M. tuberculosis [74].
The toxin MazF (12.0 kD) and antitoxin MazE (9.3 kD)
Structures and Interactions

shown in Fig. 1. In this structure, the two proteins form a
linear heterohexamer made up by alternating toxin and anti-
toxin homodimers (MazF2-MazE2-MazF2) [38]. The overall
structure of MazF resembles that of its plasmid-encoded
homologue, Kid. The MazE homodimer contains an N-
terminal intertwined ?-barrel and two mainly unfolded C-
terminal tails that are each bound to a MazF dimer interface,
thereby mediating the main interactions between MazE and
MazF [38,39]. The TA complex has been shown to autoregu-
late mazEF expression via binding of the DNA by the N-
terminal domain of MazE [50,53]. During steady state, the
antitoxin is degraded by the protease ClpAP (caseinolytic
protease AP) [75], while during amino acid starvation Lon-
dependent activation of transcription of mazEF was observed
[76].
The structure of the complex of MazF and MazE is
Mode of Action of the MazF Toxin

mRNAs and blocks protein synthesis, a blockage which is
counteracted by tmRNA [76]. Like Kid, MazF has been
shown to inhibit protein synthesis in cell extracts and act as
an endoribonuclease able to cleave RNA in the absence of
ribosomes in vitro [77-79]. This toxin was reported either to
cleave XACA sequences at the 5’ end of ACA yielding a
2’:3’-cyclic phosphate at one side and a free 5’-OH group at
the other side [77,78], or to cleave at the 5’ end of residue A
in NAC sites (where N is U or A) [79]. The 2’-hydroxyl
group of de nucleotide preceding the adenosine was shown
to be essential for RNA cleavage [78]. As for Kid, the pre-
cise nature of the MazF substrate is subject to discussion. It
could be only single-stranded mRNAs [77,78], or alterna-
tively, all cellular RNAs including both single- and double-
stranded RNA [79]. Ikura and co-workers used a single-
stranded DNA fragment in a NMR titration experiment to
It was demonstrated that MazF cleaves translated
Not For Distribution

Page 6

118 Protein & Peptide Letters, 2007, Vol. 14, No. 2 Kamphuis et al.































Figure 3. Overlapping antitoxin and RNA binding sites of Kid. a) Kid-Kis interactions mapped on a ribbon representations of the hexameric
Kid2-Kis2-Kid2 model. The Kid-Kis hexamer is shown in two shades of grey. Kid residues affected by the addition of Kis are depicted in red,
with light to dark red representing a mild to strong effect. Kid exists as a symmetric dimer and therefore two sets of originally identical resi-
dues can be distinguished. For clarity, however, only one of those sets is coloured red on each dimer. Kis residues affected by Kid binding are
shown in yellow (first monomer) and blue (second monomer). The four interaction sites and the loop between ?-strands 1 and 2, comprising
residues S10 to G21, are indicated. b) Kid residues affected by the addition of Kis mapped in red on a ribbon representation of the crystal
structure of the Kid dimer (PDB: 1M1F). The colour coding is equal to Figure a. c) Kid residues affected by the addition of AdUACA
mapped in red on a ribbon representation of the crystal structure of the Kid dimer (PDB: 1M1F). The two monomers are coloured light and
dark grey. Only one set of affected residues is coloured. d) Model of the Kid-RNA complex obtained with HADDOCK (PDB: 2C06). The
RNA fragment is coloured yellow, the two Kid monomers light and dark grey, and the active Kid residues red (HADDOCK definition [69]).
The other Kid residues affected by the addition of AdUACA are shown in blue, with the loop in cluster three transparent to show the residues
lying behind it. e) Detailed view of the catalytic site of Kid. A part of the RNA fragment is shown in yellow with the phosphate at the cleav-
age site in grey. Kid residues H17, R73 and D75, involved in RNA cleavage, are coloured green. Nitrogen atoms are coloured blue and oxy-
gen atoms red. Hydrogen bonds are shown as purple dashed lines. Figures a, b and c were generated with the program MOLMOL version
2K.1 [109] and Figures d and e with Molscript [110] and Raster3D [111]. Figures d and e have been reprinted from [60], with permission
from Elsevier.
Not For Distribution

Page 7

Structure and Function of Bacterial Kid-Kis Protein & Peptide Letters, 2007, Vol. 14, No. 2 119














Figure 4. Catalyses by ribonucleases RNase A, RNase T1 and Kid. a) Transphosphorylation reaction of RNase A and T1 in which the cata-
lytic base B deprotonates the 2’-OH group of the ribonucleotide that subsequently attacks the electrophilic phosphorus. A hydrogen atom of
the catalytic acid A is transferred to the leaving group and via a triester-like intermediate a 2’:3’-cyclic phosphate and 5’-OH group are
formed. b) Hydrolysis of the 2’:3’-cyclophosphate results in a 3’-monophosphate nucleotide and recycled acid and base residues. c) Sche-
matic representation of the active site of RNase A, RNase T1 and Kid, with RNA atoms in grey and Kid atoms in black. Shown is the pen-
tavalent transition state: the 3’-O and two nonbridging oxygens form a trigonal bipyramid (grey dashed lines), while the attacking 2’-O and
leaving 5’-O take up apical positions [65]. Catalytic hydrogen bonds are depicted as black dashed lines. This Figure has been reprinted from
[60], with permission from Elsevier.
map the substrate-binding face of MazF [80]. Their result
closely resembles the RNA-binding site of Kid in our Kid-
RNA complex [60]. The antitoxin MazE prevents RNA
cleavage by MazF [77,79] by binding to the substrate-
binding site of the toxin [43,80].
Potentially Lethal Conditions
Numerous conditions triggering the toxic action of MazF
have been described: stressful conditions like high tempera-
tures, oxidative stress, and DNA damage by thymine starva-
tion, addition of mitomycin C or nalidixic acid, or UV irra-
diation [81,82]; the presence of protein synthesis inhibiting
antibiotics such as ribfampicine, chloramphenicol and spect-
inomycin [83]; activation of the plasmid-borne toxin Doc
(death on curing), which probably inhibits translation [81];
artificial overproduction of ppGpp that was reported to in-
hibit the expression of mazEF [75], although this could not
be reproduced by other researchers [76]. All these situations
result in the inability of the cell to replenish the antitoxin
MazE upon degradation by ClpAP and thus allow MazF to
express its potentially lethal activity. MazF-induced inhibi-
tion of cell growth and colony formation was shown to be
reversible by overexpression of MazE up to six hours after
MazF induction [14,84]. Whether mazEF functions as a bac-
terial system for programmed cell death [85], or rather has a
bacteriostastic effect and functions in quality control of gene
expression [6], still remains to be elucidated.
THE CcdB-CcdA SYSTEM
The next TA system is another well-studied plasmid en-
coded system, comprising the toxin CcdB (11.7 kD) and its
antidote CcdA (8.3 kD). This TA system involves the plas-
mid F locus ccd [86,87], which stands for coupled cell divi-
sion since it was originally thought to couple plasmid repli-
cation and cell division [24,88]. In addition to its role upon
plasmid loss during cell division, it has recently been sug-
gested that this TA system is also important for bacteria in
the stationary phase, for example during nutritional stress
[89].
Structures and Interactions
The crystal structure of CcdB is shown in Fig. 1 [37]. In
fact, the CcdB dimer was the first toxin structure that was
solved. The crystal structures of Kid and MazF resemble that
of CcdB, despite a low sequence homology between CcdB
and the other two toxins [34]. The similarity of their secon-
dary structure elements is shown in Fig. 2. The small three-
stranded antiparallel ?-sheet of CcdB, inserted in a loop of
the main five-stranded antiparallel ?-sheet, is thought to be
the CcdA binding site [37]. The solution structure of the mu-
tant CcdA-R70K antitoxin has been solved by NMR spec-
troscopy [41]. Remarkably, despite the structural similarities
of the toxins, CcdA has a very different structure compared
to Kis and MazE. Each monomer of the dimeric CcdA pro-
tein consists of an N-terminal domain that contains a ?-sheet
followed by two ?-helices, and an unstructured C-terminal
tail. Its topology and DNA-binding properties classify the N-
terminal domain as a member of the ribbon-helix-helix fold
[41]. The C-terminus of the antitoxin is responsible for the
binding to CcdB: a truncated form of CcdA lacking the first
31 N-terminal residues was still able to neutralise the CcdB
toxicity [90] and an NMR titration experiment showed that
only CcdA residues between V46 and D71 interact with
Not For Distribution

Page 8

120 Protein & Peptide Letters, 2007, Vol. 14, No. 2 Kamphuis et al.































Figure 5. Chemical shift perturbations of Kid residues upon addition of AdUACA and mononucleotide ligands. The perturbations are meas-
ured at a RNA-dU:Kid ratio of 2:1 or ligand:Kid ratio of 38:1. For the RNA-dU addition, the shifts of residues in fast exchange are shown in
grey and those of residues in fast to intermediate exchange in black. The dashed lines indicate disappearing residues in intermediate exchange.
Side-chain NH chemical shift changes of W7/Q19-2x/Q20-2x/R23/R38/N50-2x/R73/Q76-2x/ R78/R85/N99-2x/R104 are depicted as well.
T29/E91 and the side-chains of R3/N34/R35/R53/R67/R89/R92, which could not be assigned unambiguously, are not perturbed by RNA-dU
or any of the mononucleotide ligands. Prolines (residues 13/24/30/40/44/77/94) can not be observed in this experiment and are therefore omit-
ted.
CcdB [41]. Apparently, CcdA binds to CcdB in a similar
way as MazE binds to MazF, via interactions of a carboxy-
terminal tail [38,41]. This is corroborated by the observation
of a CcdB4CcdA2 heterohexamer after mixing CcdB and
CcdA at a 2:1 ratio [91]. Like for Kid and Kis, the CcdB-
CcdA interaction is complex since a molar ratio of the two
proteins of 1:1 resulted in a mixture of complexes with vary-
ing stoichiometry [91]. A (CcdB2CcdA2)n complex with a
CcdB:CcdA ratio of one binds DNA and represses the tran-
scription of ccd [51,52,54,92,93]. When the level of CcdB
exceeds that of CcdA, no DNA binding is observed [54]. In
contrast with other TA systems, CcdA alone is not able to
repress transcription [51,52,94]. The solution structure of
CcdA bound to a 6 bp palindromic DNA sequence within the
operator-promoter region of the ccd operon has been re-
ported [41]. The antitoxin is degraded by the ATP-dependant
Lon protease [52,55].
Not For Distribution

Page 9

Structure and Function of Bacterial Kid-Kis Protein & Peptide Letters, 2007, Vol. 14, No. 2 121
Mode of Action of the CcdB Toxin

rather similar, the toxicity of CcdB and the other two pro-
teins involve completely different actions. CcdB lacks nu-
merous amino acids responsible for the RNase activity of
Kid and MazF. Instead, the target of CcdB is the DNA topoi-
somerase gyrase [95-97]. Gyrase is an indispensable enzyme
during DNA replication and transcription, since it introduces
negative supercoils in circular DNA. The CcdB protein poi-
sons gyrase by binding the GyrA dimer with its three C-
terminal residues W99, G100 and I101 [37,92], causing two
different effects depending on the state GyrA is present.
Binding of CcdB to free GyrA or gyrase in the absence of
DNA results in GyrA/gyrase-CcdB complexes unable to bind
and supercoil DNA [95,98]. When CcdB binds gyrase during
the catalytic cycle, it induces gyrase-mediated DNA cleav-
age resulting in a covalent gyrase-DNA complex that blocks
transcription by RNA polymerase [99]. The latter process,
however, is rather slow and is thought to be possible only
during relocation of the T-segment DNA [97,100]. The most
likely model for the CcdB-GyrA interaction is that CcdB
binds an open conformation of the GyrA dimer. This would
explain both poisoning modes, since the two catalytic tyro-
sine residues of gyrase are separated too far to catalyse DNA
cleavage or religation [37,97]. Addition of the antitoxin
CcdA releases the gyrase from the CcdB toxin [101].
Not For Distribution
Although the structures of CcdB, Kid and MazF are
THE RelE-RelB SYSTEM

toxin RelE (10.3 kD) and antitoxin RelB (8.8 kD), encoded
by the relBE locus of the E. coli chromosome (K-12 strain)
[33]. This system is homologous to the relBE system en-
coded by plasmid P307. The relBE locus is highly abundant
on the chromosomes of very diverse prokaryotes: 129 loci
are identified in both gram-negative and gram-positive bacte-
ria and 27 in archaea [15,30].
The next TA system discussed here is formed by the
Structure of the aRelE-aRelB Complex

logues aRelE (9.7 kD) and aRelB (7.5 kD), expressed by the
hyperthermophilic Pyrococcus horikoshii OT3, is shown in
Fig. 1. The proteins form a heterotetramer (aRelE-aRelB)2,
with the four ?-helices and short ?-strand of aRelB being
wrapped around aRelE. The ellipsoidal aRelE consists of
three ?-helices and a five-stranded ?-sheet, with ?2 to ?5
running antiparallel and ?1 parallel to the neighbouring ?5.
Site-directed mutagenesis has indicated that residues R40,
L48, R58, R65 and especially R85 are important for the toxic
activity of aRelE [40]. RelE is structurally related to the
YoeB toxin and also to the microbial RNases barnase and
RNase Sa [42].
The crystal structure of a complex of the archaeal homo-
Mode of Action of the RelE Toxin

in vitro by promoting the cleavage of mRNAs in the ribo-
somal A-site [102]. Since aRelE has comparable dimensions
and shape as the decoding domain in EF-G, the two proteins
have been suggested to enter the ribosomal A-site in a simi-
lar way [40]. RelE usually promotes cutting between the
second and third base; it prefers the stop codon UAG, but
RelE has been demonstrated to inhibit protein synthesis
also cleaves the other stop codons and the sense codons
UCG and CAG. Furthermore, cleavage of mRNA in the ri-
bosomal E-site was observed after removal of the peptide
from peptidyl-tRNA in the P-site, while degradation of
mRNA bound to free 30S subunits was shown as well [102].
The ribonucleic acid tmRNA is thought to recover RelE-
inhibited protein synthesis in vivo by rescuing ribosomes
stalled on mRNAs truncated by RelE [103]. RelE activity
can be neutralised through the induction of transcription of
the RelB gene [14].
The RelE-RelB System as Stress-Response Element

tein synthesis, as happens when the stringent response takes
place. Subsequently, the existing pool of RelB antitoxin is
degraded by the Lon protease and de-repression of the relBE
locus follows, since the transcription is autoregulated by a
cooperative action of RelB and RelE [33,56]. The resulting
increase of relBE mRNA and the concomitant continued
synthesis of RelB were suggested to allow fine-tuning of the
RelB:RelE ratio [56]. It is thought that during the stringent
response, due to the action of RelE, a fresh supply of amino
acids is obtained via degradation of incomplete nascent pro-
teins recognised by their tmRNA tag. Those amino acids can
be used for the synthesis of proteins essential for the changed
circumstances, like amino acid synthesising enzymes. An
additional function of RelE could be to reduce the number of
ribosomes ‘consuming’ the scarce amino acids in the elonga-
tion process, thereby increasing the speed and accuracy of
the ribosomes that remain active. In this way, RelE would
allow the cell to adjust to the new situation, resolve the nutri-
tional stress and restore normal growth conditions. There-
fore, at present, the RelE-RelB TA system is considered to
be a stress-response element rather than a cell-killing module
[102]. However, it was shown that in the case of the ho-
mologous pneumococcal RelBE2 system, prolonged expo-
sure of the cells to the toxin resulted in the inability to rescue
the cultivability of the cells by subsequent expression of the
antitoxin [104].
RelE toxicity is induced upon a strong inhibition of pro-
THE YoeB-YefM SYSTEM

YoeB (10 kD) and antitoxin YefM (10 kD) [42,105,106].
This TA system, encoded by the yefM-yoeB locus of the E.
coli chromosome, shows similarities to the relBE system.
The last TA system described here comprises the toxin
Structures and Interactions

free YoeB toxin have been determined [42], and the structure
of the heterotrimeric YoeB-YefM2 complex is depicted in
Fig. 1. The compact, globular YoeB protein consists of a
five-stranded ?-sheet and two ?-helices [42]. It is highly
similar to the aRelE monomer, except for an extended loop
in aRelE. YoeB shows also structural similarities to RNase
Sa from Streptomyces aureofaciens and Barnase from Bacil-
lus amyloliquefaciens. The bound and free structures of
YoeB are essentially identical, except for the three most C-
terminal amino acids. In the free form, those residues fold
into a typical conformation seen in the catalytic site of mi-
crobial RNases. The YefM homodimer can be divided in a
Crystal structures of the YoeB-YefM complex and the

Page 10

122 Protein & Peptide Letters, 2007, Vol. 14, No. 2 Kamphuis et al.
globular and symmetrical N-terminal part, and two extended
C-terminal segments. The N-terminal region consists of a
single hydrophobic core of six ?-strands and two ?-helices
from each monomer, plus a third helix from each monomer
that cross each other at an angle of 65°. One of the two C-
terminal regions interacts with YoeB and is ordered into an
?-helix and an extended ?-strand, while the other is disor-
dered [42]. This fits well with the observation that free YefM
is unfolded under native conditions [107]. YefM fragments
including the H4 helix consisting of residues 67-83, but not
the H4 helix alone, where able to bind YoeB and neutralise
its toxicity. Free YoeB forms a dimeric structure in which
the hydrophobic interactions between the two monomers
mimic the hydrophobic interactions within the YoeB-YefM
complex [42].
Not For Distribution
Mode of Action of the YoeB Toxin

translated mRNAs in vivo [108] and that it functions as a
ribosome-independent ribonuclease in vitro [42]. YefM is
able to inhibit the RNase activity of YoeB [42], while Lon
overproduction specifically
mRNA cleavage [108]. In vivo, YoeB preferentially cuts
between the second and third base (AAA, CUG, GCG, and
the stop codon UAA), but also cleaves between codons (be-
tween AAA and GCU, after UAC, and after UUA) [108]. In
vitro, YoeB was shown to be a purine-specific endoribonu-
clease [42].
It was shown that the YoeB toxin induces cleavage of
activates YoeB-dependent
The Catalytic Site of YoeB

the microbial RNase Sa and Barnase, a similar reaction
mechanism has been proposed that involves a histidine and
glutamic acid residue as general acid and base, respectively,
and an arginine for binding the reactive phosphate [42].
Candidates would be the YoeB residues H83, E46 and R65.
In the YoeB-YefM2 complex, the distance between the po-
tentially catalytic groups of H83 and E46 is ~21 Å, three
times larger than generally seen. In the free YoeB protein,
however, where the three C-terminal residues adopt a spe-
cific fold also observed in the catalytic site of microbial
RNases, this distance is significantly smaller. Site-directed
mutagenesis showed that all of those three residues, and
Y84, are essential for the toxicity of YoeB. The role of Y84
would be to stabilise the protein-RNA interaction at the ac-
tive site. It was concluded that YefM inhibits the catalytic
activity of YoeB by stabilising non-optimal conformations of
the involved residues and direct obstruction of the catalytic
site [42].
Because of the structural similarities between YoeB and
CONCLUDING REMARKS

dant in plasmids and genomes of both bacteria and archaea.
During normal growth conditions, the unstable antitoxin is
bound to the toxin neutralising its toxicity. Upon stress,
however, a diminished level of antitoxin allows the activa-
tion of the toxin causing cell growth arrest and possibly cell
death. This review discusses five related, well-studied TA
systems found in E. coli. A high structural similarity exists
between the toxins Kid, MazF and CcdB, respectively the
Toxin-antitoxin systems are highly conserved and abun-
antitoxins Kis and MazE, while the structure of CcdA is
rather different. The toxins RelE and YoeB, respectively the
antitoxins RelB and YefM, are also structurally related, but
their complexes possess different stoichiometries.
CcdB causes gyrase poisoning in contrast to the other
four toxins that are RNases. RelE functions by cleaving
mRNAs in the ribosomal A-site, while Kid, MazF and YoeB
seem more effective in ribosome-independent RNA cleav-
age. A model for RNA binding and the catalytic site of Kid
has been determined and it was suggested that D75 and R73
are the catalytic base and acid, respectively. Sequence
alignments show that D75 of Kid is conserved in MazF
(D76) and most related toxins (D or E), but that R73 is not.
For speculation about the catalytic acid of MazF and other
related toxins, the structures of these proteins in their active
conformation are needed. At the position corresponding to
H17, the third active site residue of Kid, a histidine or func-
tionally similar serine is found. A significant overlap exists
between the RNA and antitoxin binding sites of the two tox-
ins Kid and MazF. This explains the mode of inhibition by
Kis and MazE, respectively, on a molecular level. The
CcdB-CcdA and RelE-RelB complexes might be too large to
enter their specific site of action. The catalytic residues of
YoeB have been proposed to be E46 and H83, being the
catalytic base and acid, respectively. R65 is thought to bind
the reactive phosphate. The observed absence of catalytic
residues in RelE and their presence in YoeB is consistent
with the intrinsic ribonuclease activity of YoeB and the fact
that RelE mediates cleavage of mRNA in a ribosome-
dependent manner.

MazF-MazE and CcdB-CcdA, the obtained information
seems to point out a general ‘architecture’ of the systems.
Likely, those and related TA systems all function according
to a same scheme, following here. In a bacterial cell experi-
encing a normal growth situation, the antitoxin:toxin ratio is
approximately 4:1. The complexes formed at these condi-
tions are multimers with antitoxin:toxin ratios close to 1:1;
the toxicity is neutralised and transcription of the TA operon
is autoregulated properly. Possibly, the 1:1 stoichiometry of
the TA multimers is favourable for protein-DNA interac-
tions. Upon loss of the plasmid or other stressful conditions,
antitoxin synthesis is abolished and the antitoxin:toxin ratio
drops significantly; T2A2T2 heterohexamers are formed to
neutralise the toxicity as much as possible. For plasmid-
encoded systems, a progressive activation of the toxin will
occur in time. Gradually, toxin dimers will co-exist with the
heterohexamers and affect cell growth increasingly. The
same might happen for chromosomal TA systems. However,
since de-repression of the operon follows upon diminishing
of the antitoxin, a continued synthesis of antitoxin might take
place. This could result in survival of the cell.
For the structurally related TA systems, i.e. Kid-Kis,
ACKNOWLEDGEMENTS

sion. This project was financially supported by the European
Union (QLK2-CT-2000-00634, HPRI-CT-2001-00172 and
NMR Large Scale Facility scheme). M.B. Kamphuis was
further supported by the Center for Biomedical Genetics,
M.C. Monti by a Short-Term FEBS Fellowship and the
We would like to thank G.E. Folkers for valuable discus-

Page 11

Structure and Function of Bacterial Kid-Kis Protein & Peptide Letters, 2007, Vol. 14, No. 2 123
lab of R. Díaz-Orejas by the Spanish MCyT (SAF2002-
04649) and MEC (BFU2005-03911/BMC).
REFERENCES
[1]
[2]
Hayes, F. (2003) Science, 301, 1496-1499.
Holcik, M. and Iyer, V. N. (1997) Microbiology, 143 ( Pt 11),
3403-3416.
Engelberg-Kulka, H. and Glaser, G. (1999) Annu. Rev. Microbiol.,
53, 43-70.
Gerdes, K., Rasmussen, P. B. and Molin, S. (1986) Proc. Natl.
Acad. Sci. USA, 83, 3116-3120.
Jensen, R. B. and Gerdes, K. (1995) Mol. Microbiol., 17, 205-210.
Gerdes, K., Christensen, S. K. and Lobner-Olesen, A. (2005) Nat.
Rev. Microbiol., 3, 371-382.
Greenfield, T. J., Ehli, E., Kirshenmann, T., Franch, T., Gerdes, K.
and Weaver, K. E. (2000) Mol. Microbiol., 37, 652-660.
Condon, C. (2006) Mol. Microbiol., 61, 573-583.
Anantharaman, V. and Aravind, L. (2003) Genome Biol., 4, R81.
Engelberg-Kulka, H., Sat, B., Reches, M., Amitai, S. and Hazan, R.
(2004) Trends Microbiol., 12, 66-71.
Hayes, C. S. and Sauer, R. T. (2003) Cell, 112, 2-4.
Nystrom, T. (2003) Bioessays, 25, 204-211.
Nystrom, T. (2004) Annu. Rev. Microbiol., 58, 161-181.
Pedersen, K., Christensen, S. K. and Gerdes, K. (2002) Mol. Mi-
crobiol., 45, 501-510.
Pandey, D. P. and Gerdes, K. (2005) Nucleic Acids Res., 33, 966-
976.
Gerdes, K. (2000) J. Bacteriol., 182, 561-572.
Clissold, P. M. and Ponting, C. P. (2000) Curr. Biol., 10, R888-
890.
de la Cueva-Mendez, G., Mills, A. D., Clay-Farrace, L., Diaz-
Orejas, R. and Laskey, R. A. (2003) EMBO J., 22, 246-251.
Picardeau, M., Le Dantec, C., Richard, G. F. and Saint Girons, I.
(2003) FEMS Microbiol. Lett., 229, 277-281.
Yamamoto, T. A., Gerdes, K. and Tunnacliffe, A. (2002) FEBS
Lett., 519, 191-194.
Gabant, P., Van Reeth, T., Dreze, P. L., Faelen, M., Szpirer, C. and
Szpirer, J. (2000) Biotechniques, 28, 784-788.
Xu, W., Zhang, Y., Yeh, L. Y., Ruprecht, C. R., Wong-Staal, F.,
McFadden, B. A., Reddy, T. R. and Ruprecht, R. M. (2002) Bio-
techniques, 32, 1266-8, 1270.
Bernard, P., Gabant, P., Bahassi, E. M. and Couturier, M. (1994)
Gene, 148, 71-4.
Ogura, T. and Hiraga, S. (1983) Proc. Natl. Acad. Sci. USA, 80,
4784-4788.
Bravo, A., de Torrontegui, G. and Diaz, R. (1987) Mol. Gen.
Genet., 210, 101-110.
Pullinger, G. D. and Lax, A. J. (1992) Mol. Microbiol., 6, 1631-
1643.
Lehnherr, H., Maguin, E., Jafri, S. and Yarmolinsky, M. B. (1993)
J. Mol. Biol., 233, 414-428.
Roberts, R. C. and Helinski, D. R. (1992) J. Bacteriol., 174, 8119-
8132.
Tian, Q. B., Ohnishi, M., Tabuchi, A. and Terawaki, Y. (1996)
Biochem. Biophys. Res. Commun., 220, 280-284.
Gronlund, H. and Gerdes, K. (1999) J. Mol. Biol., 285, 1401-1415.
Metzger, S., Dror, I. B., Aizenman, E., Schreiber, G., Toone, M.,
Friesen, J. D., Cashel, M. and Glaser, G. (1988) J. Biol. Chem.,
263, 15699-15704.
Masuda, Y., Miyakawa, K., Nishimura, Y. and Ohtsubo, E. (1993)
J. Bacteriol., 175, 6850-6856.
Gotfredsen, M. and Gerdes, K. (1998) Mol. Microbiol., 29, 1065-
1076.
Hargreaves, D., Santos-Sierra, S., Giraldo, R., Sabariegos-Jareno,
R., de la Cueva-Mendez, G., Boelens, R., Diaz-Orejas, R. and Raf-
ferty, J. B. (2002) Structure (Camb), 10, 1425-1433.
Ceglowski, P., Boitsov, A., Chai, S. and Alonso, J. C. (1993) Gene,
136, 1-12.
Brantl, S., Behnke, D. and Alonso, J. C. (1990) Nucleic Acids Res.,
18, 4783-4790.
[3]
[4]
[5]
[6]
[7]
[8]
[9]
[10]
[11]
[12]
[13]
[14]
[15]
[16]
[17]
[18]
[19]
[20]
[21]
[22]
[23]
[24]
[25]
[26]
[27]
[28]
[29]
[30]
[31]
[32]
[33]
[34]
[35]
[36]
[37] Loris, R., Dao-Thi, M. H., Bahassi, E. M., Van Melderen, L.,
Poortmans, F., Liddington, R., Couturier, M. and Wyns, L. (1999)
J. Mol. Biol., 285, 1667-1677.
Kamada, K., Hanaoka, F. and Burley, S. K. (2003) Mol. Cell, 11,
875-884.
Loris, R., Marianovsky, I., Lah, J., Laeremans, T., Engelberg-
Kulka, H., Glaser, G., Muyldermans, S. and Wyns, L. (2003) J.
Biol. Chem., 278, 28252-28257.
Takagi, H., Kakuta, Y., Okada, T., Yao, M., Tanaka, I. and Kimura,
M. (2005) Nat. Struct. Mol. Biol., 12, 327-331.
Madl, T., Van Melderen, L., Mine, N., Respondek, M., Oberer, M.,
Keller, W., Kathai, L. and Zangger, K. (2006) J. Mol. Biol., on-
line.
Kamada, K. and Hanaoka, F. (2005) Mol. Cell, 19, 497-509.
Kamphuis, M. B. et al. (2007) Proteins, on-line, doi: 10,1002/
prot.21254.
Tsuchimoto, S., Ohtsubo, H. and Ohtsubo, E. (1988) J. Bacteriol.,
170, 1461-1466.
Bravo, A., Ortega, S., de Torrontegui, G. and Diaz, R. (1988) Mol.
Gen. Genet., 215, 146-151.
Lemonnier, M., Santos-Sierra, S., Pardo-Abarrio, C. and Diaz-
Orejas, R. (2004) J. Bacteriol., 186, 240-243.
Ruiz-Echevarria, M. J., Berzal-Herranz, A., Gerdes, K. and Diaz-
Orejas, R. (1991) Mol. Microbiol., 5, 2685-2693.
Santos-Sierra, S., Pardo-Abarrio, C., Giraldo, R. and Diaz-Orejas,
R. (2002) FEMS Microbiol. Lett., 206, 115-119.
Tsuchimoto, S. and Ohtsubo, E. (1993) Mol. Gen. Genet., 237, 81-
88.
Marianovsky, I., Aizenman, E., Engelberg-Kulka, H. and Glaser,
G. (2001) J. Biol. Chem., 276, 5975-5984.
Tam, J. E. and Kline, B. C. (1989) Mol. Gen. Genet., 219, 26-32.
Salmon, M. A., Van Melderen, L., Bernard, P. and Couturier, M.
(1994) Mol. Gen. Genet., 244, 530-538.
Zhang, J., Zhang, Y. and Inouye, M. (2003) J. Biol. Chem., 278,
32300-32306.
Afif, H., Allali, N., Couturier, M. and Van Melderen, L. (2001)
Mol. Microbiol., 41, 73-82.
Van Melderen, L., Thi, M. H., Lecchi, P., Gottesman, S., Couturier,
M. and Maurizi, M. R. (1996) J. Biol. Chem., 271, 27730-27738.
Christensen, S. K., Mikkelsen, M., Pedersen, K. and Gerdes, K.
(2001) Proc. Natl. Acad. Sci. USA, 98, 14328-14333.
Tsuchimoto, S., Nishimura, Y. and Ohtsubo, E. (1992) J. Bacteri-
ol., 174, 4205-4211.
Hargreaves, D., Giraldo, R., Santos-Sierra, S., Boelens, R., Rice, D.
W., Diaz Orejas, R. and Rafferty, J. B. (2002) Acta Crystallogr. D
Biol. Crystallogr., 58, 355-358.
Santos-Sierra, S., Lemonnier, M., Nunez, B., Hargreaves, D., Raf-
ferty, J., Giraldo, R., Andreu, J. M. and Diaz-Orejas, R. (2003)
Plasmid, 50, 120-130.
Kamphuis, M. B., Bonvin, A. M. J. J., Monti, M. C., Lemonnier,
M., Muñoz-Gómez, A., van den Heuvel, R. H. H., Díaz-Orejas, R.
and Boelens, R. (2006) J. Mol. Biol., 357, 115-126.
Munoz-Gomez, A. J., Lemonnier, M., Santos-Sierra, S., Berzal-
Herranz, A. and Diaz-Orejas, R. (2005) J. Bacteriol., 187, 3151-
3157.
Zhang, J., Zhang, Y., Zhu, L., Suzuki, M. and Inouye, M. (2004) J.
Biol. Chem., 279, 20678-20684.
Pimentel, B., Madine, M. A. and de la Cueva-Mendez, G. (2005)
Embo J., 24, 3459-3469.
Loverix, S. and Steyaert, J. (2001) Methods Enzymol., 341, 305-
323.
Loverix, S. and Steyaert, J. (2003) Curr. Med. Chem., 10, 779-785.
Raines, R. T. (1998) Chem. Rev., 98, 1045-1066.
Steyaert, J. (1997) Eur. J. Biochem., 247, 1-11.
Ruiz-Echevarria, M. J., de la Torre, M. A. and Diaz-Orejas, R.
(1995) FEMS Microbiol. Lett., 130, 129-135.
Dominguez, C., Boelens, R. and Bonvin, A. M. (2003) J. Am.
Chem. Soc., 125, 1731-1737.
Kamphuis, M. B. (2005) Utrecht University, Utrecht.
Gogos, A., Mu, H., Bahna, F., Gomez, C. A. and Shapiro, L. (2003)
Proteins, 53, 320-322.
Pellegrini, O., Mathy, N., Gogos, A., Shapiro, L. and Condon, C.
(2005) Mol. Microbiol., 56, 1139-1148.
[38]
[39]
[40]
[41]
[42]
[43]
[44]
[45]
[46]
[47]
[48]
[49]
[50]
[51]
[52]
[53]
[54]
[55]
[56]
[57]
[58]
[59]
[60]
[61]
[62]
[63]
[64]
[65]
[66]
[67]
[68]
[69]
[70]
[71]
[72]
Not For Distribution

Page 12

124 Protein & Peptide Letters, 2007, Vol. 14, No. 2 Kamphuis et al.
[73] Lemos, J. A., Brown, T. A., Jr., Abranches, J. and Burne, R. A.
(2005) FEMS Microbiol. Lett., 253, 251-257.
Zhu, L., Zhang, Y., Teh, J. S., Zhang, J., Connell, N., Rubin, H.
and Inouye, M. (2006) J. Biol. Chem., 281, 18638-18643.
Aizenman, E., Engelberg-Kulka, H. and Glaser, G. (1996) Proc.
Natl. Acad. Sci. USA, 93, 6059-6063.
Christensen, S. K., Pedersen, K., Hansen, F. G. and Gerdes, K.
(2003) J. Mol. Biol., 332, 809-819.
Zhang, Y., Zhang, J., Hoeflich, K. P., Ikura, M., Qing, G. and
Inouye, M. (2003) Mol. Cell, 12, 913-923.
Zhang, Y., Zhang, J., Hara, H., Kato, I. and Inouye, M. (2005) J.
Biol. Chem., 280, 3143-3150.
Munoz-Gomez, A. J., Santos-Sierra, S., Berzal-Herranz, A.,
Lemonnier, M. and Diaz-Orejas, R. (2004) FEBS Lett., 567, 316-
320.
Li, G. Y., Zhang, Y., Chan, M. C., Mal, T. K., Hoeflich, K. P.,
Inouye, M. and Ikura, M. (2006) J. Mol. Biol., 357, 139-150.
Hazan, R., Sat, B. and Engelberg-Kulka, H. (2004) J. Bacteriol.,
186, 3663-3669.
Sat, B., Reches, M. and Engelberg-Kulka, H. (2003) J. Bacteriol.,
185, 1803-1807.
Sat, B., Hazan, R., Fisher, T., Khaner, H., Glaser, G. and Engel-
berg-Kulka, H. (2001) J. Bacteriol., 183, 2041-2045.
Amitai, S., Yassin, Y. and Engelberg-Kulka, H. (2004) J. Bacteri-
ol., 186, 8295-8300.
Kolodkin-Gal, I. and Engelberg-Kulka, H. (2006) J. Bacteriol.,
188, 3420-3423.
Hiraga, S., Jaffe, A., Ogura, T., Mori, H. and Takahashi, H. (1986)
J. Bacteriol., 166, 100-104.
Jaffe, A., Ogura, T. and Hiraga, S. (1985) J. Bacteriol., 163, 841-
849.
Miki, T., Yoshioka, K. and Horiuchi, T. (1984) J. Mol. Biol., 174,
605-625.
Aguirre-Ramírez, M., Ramírez-Santos, J., Van Melderen, L. and
Gómez-Eichelmann, M. C. (2006) Canad. J. Microbiol., 52, 24-30.
Bernard, P. and Couturier, M. (1991) Mol. Gen. Genet., 226, 297-
304.
[74]
[75]
[76]
[77]
[78]
[79]
[80]
[81]
[82]
[83]
[84]
[85]
[86]
[87]
[88]
[89]
[90]
[91] Dao-Thi, M. H., Charlier, D., Loris, R., Maes, D., Messens, J.,
Wyns, L. and Backmann, J. (2002) J. Biol. Chem., 277, 3733-3742.
Bahassi, E. M., Salmon, M. A., Van Melderen, L., Bernard, P. and
Couturier, M. (1995) Mol. Microbiol., 15, 1031-1037.
de Feyter, R., Wallace, C. and Lane, D. (1989) Mol. Gen. Genet.,
218, 481-486.
Tam, J. E. and Kline, B. C. (1989) J. Bacteriol., 171, 2353-2360.
Maki, S., Takiguchi, S., Miki, T. and Horiuchi, T. (1992) J. Biol.
Chem., 267, 12244-12251.
Bernard, P. and Couturier, M. (1992) J. Mol. Biol., 226, 735-745.
Dao-Thi, M. H., Van Melderen, L., De Genst, E., Afif, H., Buts, L.,
Wyns, L. and Loris, R. (2005) J. Mol. Biol., 348, 1091-1102.
Maki, S., Takiguchi, S., Horiuchi, T., Sekimizu, K. and Miki, T.
(1996) J. Mol. Biol., 256, 473-482.
Critchlow, S. E., O'Dea, M. H., Howells, A. J., Couturier, M.,
Gellert, M. and Maxwell, A. (1997) J. Mol. Biol., 273, 826-839.
Kampranis, S. C., Howells, A. J. and Maxwell, A. (1999) J. Mol.
Biol., 293, 733-744.
Bahassi, E. M., O'Dea, M. H., Allali, N., Messens, J., Gellert, M.
and Couturier, M. (1999) J. Biol. Chem., 274, 10936-10944.
Pedersen, K., Zavialov, A. V., Pavlov, M. Y., Elf, J., Gerdes, K.
and Ehrenberg, M. (2003) Cell, 112, 131-140.
Christensen, S. K. and Gerdes, K. (2003) Mol. Microbiol., 48,
1389-1400.
Nieto, C., Pellicer, T., Balsa, D., Christensen, S. K., Gerdes, K. and
Espinosa, M. (2006) Mol. Microbiol., 59, 1280-1296.
Cherny, I., Rockah, L. and Gazit, E. (2005) J. Biol. Chem., 280,
30063-30072.
Grady, R. and Hayes, F. (2003) Mol. Microbiol., 47, 1419-1432.
Cherny, I. and Gazit, E. (2004) J. Biol. Chem., 279, 8252-8261.
Christensen, S. K., Maenhaut-Michel, G., Mine, N., Gottesman, S.,
Gerdes, K. and Van Melderen, L. (2004) Mol. Microbiol., 51,
1705-1717.
Koradi, R., Billeter, M. and Wuthrich, K. (1996) J. Mol. Graph.,
14, 51-5, 29-32.
Kraulis, P. J. (1991) J. Appl. Crystallogr., 24, 946-950.
Merritt, E. A. and Murphy, M. E. (1994) Acta Crystallogr. D Biol.
Crystallogr., 50, 869-873.
[92]
[93]
[94]
[95]
[96]
[97]
[98]
[99]
[100]
[101]
[102]
[103]
[104]
[105]
[106]
[107]
[108]
[109]
[110]
[111]


Received: September 11, 2006 Revised: November 03, 2006 Accepted: November 11, 2006

Not For Distribution