(Received for publication, February 6, 1997, and in revised form, May 28, 1997)
From the DnaA protein is the initiator of chromosomal DNA
replication in Escherichia coli. We report here our
evidence that thymidine transport across cytoplasmic membranes in
temperature-sensitive dnaA mutants is greatly decreased at
a permissive temperature for growth of the mutants. Complementation
analysis with a plasmid containing the wild type dnaA gene
and P1 phage-mediated transduction confirmed that mutations in the
dnaA gene were responsible for the phenotype. A low level
of nucleoside transport in the dnaA mutant was specific for
thymidine; transport activities for other nucleosides were much the
same as those in wild type cells. Membrane vesicles prepared from the
dnaA mutant showed much the same activity of thymidine
transport as did those from the wild type cells. No significant
difference in the activity of thymidine kinase, which is known to
facilitate thymidine transport, was seen between the mutant and the
wild type cells. An increase in the pool of dTTP, a negative regulator
for thymidine kinase, was observed in the dnaA mutant.
Based on these observations, we suggest that inhibition of thymidine
transport in dnaA mutants is caused by increases in the
dTTP pool.
DnaA protein is the initiation factor of chromosomal DNA
replication in Escherichia coli (1, 2). Initiation of DNA
replication is considered to be regulated by alteration in the activity
of DnaA protein through adenine nucleotide binding, membrane binding, and other factors (3-12). In addition, DnaA protein may regulate cellular processes so that they are coupled with DNA replication (13).
DnaA protein negatively regulates expression of various genes,
including the dnaA gene, through specific binding to DnaA boxes located at the 5 Thymidine transport systems are ubiquitously present in prokaryotic and
eukaryotic cells. Pulse-labeling experimentation with radiolabeled
thymidine has been accepted widely as a method of determining the level
of DNA synthesis in cells. Little attention, however, has been directed
to whether the transport activity of this nucleoside can be altered
experimentally. Genetic and biochemical studies using E. coli identified two different transport systems for nucleosides,
termed the NupC and the NupG systems (16-19), and these can be
distinguished by the lack of potential of the NupC system to transport
guanosine (16-19). Transcriptional regulation of the nupC
and nupG genes has been reported (19). Nucleoside transport
activity plus DNA synthesis seem to be needed for related activity to
occur.
We report here the influence of the dnaA mutation in
thymidine transport in E. coli; thymidine transport across
the cytoplasmic membranes in the dnaA mutants was inhibited
drastically, which in turn suggests a role for DnaA protein in the
regulation of nucleoside transport. A putative mechanism for the
inhibition of thymidine transport in dnaA mutants is
discussed.
All strains of E. coli used in the present experiments are listed in Table
I. E. coli strains ME8923
(nupC3146) and ME8935 (nupG3157) were kindly
provided by Dr. A. Nishimura (National Institute of Genetics, Japan)
(20). P1 phage-mediated transduction was performed as described (21,
22). For complementation analysis, we used the pHB10S plasmid, which
contains the wild type dnaA gene (21, 22). pASTS1 plasmid, a
thyA overexpression plasmid (23), was kindly provided by Dr.
M. Iwakura (National Institute of Bioscience and Human Technology,
Japan).
Table I.
Bacterial strains
Faculty of Pharmaceutical Sciences,
ABSTRACT
INTRODUCTION
EXPERIMENTAL PROCEDURES
RESULTS
DISCUSSION
FOOTNOTES
ACKNOWLEDGEMENTS
REFERENCES
-upstream regulatory region or coding region of
these genes (12). We recently reported an activity of DnaA protein on
DNA topology in vitro through its nonspecific binding to DNA
(14) and alteration in DNA supercoiling in dnaA mutants (15). This alteration in DNA supercoiling may profoundly affect cell
physiology (13, 15). Therefore, studies on functions of DnaA protein
other than that as initiator of chromosomal DNA replication seem
warranted if one is to understand the molecular basis for the
cooperation of DNA replication with other processes.
Bacterial Strains and a Plasmid
Strain
Relevant genotype
Refs.
KS1001
W3110 dnaA+
tnaA::Tn10 lac
21
KS1002
W3110 dnaA5 traA::Tn10
lac
21
KS1003
W3110 dnaA46
tnaA::Tn10 lac
21
KS1004
W3110 dnaA167 tnaA::Tn10
lac
21
KS1005
W3110 dnaA204
tnaA::Tn10 lac
21
KS1006
W3110 dnaA205 tnaA::Tn10
lac
21
KS1007
W3110 dnaA508
tnaA::Tn10 lac
21
KS1008
W3110 dnaA602 tnaA::Tn10
lac
21
KS1009
W3110 dnaA604
tnaA::Tn10 lac
21
ME8970
wild
typenupG3157::Tn10kan
20
ME8923
ME8970
nupC3146::Tn10kan
20
ME8935
ME8970
nupG3157::Tn10kan
20
KY1001
ME8923 dnaA46
tnaA::Tn10
This work
KY1002
ME8935 dnaA46
tnaA::Tn10
This work
Exponentially growing E. coli cells in LB medium were diluted appropriately. The suspension was spread on LB agar plates. The plates were incubated at 30 °C for 12 h. Colonies were counted, and viable cell numbers were determined.
Determination of DNA in CellsExponentially growing
E. coli cells in LB medium were harvested periodically by
centrifugation and were suspended in 300 µl of 0.9% NaCl. Toluene
(10 µl) was added to the suspension, which was then mixed vigorously.
4,6-Diamidino-2-phenylindole dihydrochloride (24) solution was then
added to the suspension to a final concentration of 0.1 µg/ml.
Fluorescence by -excitation (346 nm) and
-emission (452 nm) was
measured.
Exponentially growing E. coli cells in LB medium were treated with nalidixic acid (200 µg/ml, final concentration) at 37 °C for 10 min. Radiolabeled nucleosides were added to the culture, and incubation was run at 37 °C for 20 s. Transport reaction of radiolabeled nucleosides into cells was terminated by adding unlabeled nucleosides at a final concentration of 50 mM. Samples were passed through nitrocellulose membranes (Millipore), radioactivity remaining on the filters was counted in a liquid scintillation counter, and the amount of nucleoside transported into cells was calculated (19).
Preparation of Membrane Vesicles and Transport ExperimentsPreparation of membrane vesicles from the dnaA46 mutant and wild type cells and examination of their activities for nucleosides transport were done as described by Munch-Petersen et al. (19). In brief, exponentially growing cells in LB medium were harvested when the optical density at 660 nm reached 0.8. Vesicles were prepared according to Kaback (25), omitting the final sucrose gradient centrifugation. Protein determination of the membrane vesicles was done as described (26). The standard reaction mixture contained 0.1 M potassium phosphate (pH 6.5), 0.1 M MgSO4, 0.15 mM phenazine methosulfate, 30 mM ascorbate, and 4 µM radiolabeled nucleoside. The transport reaction was initiated by addition of membrane vesicles to the mixtures; the reaction was performed at 20 °C for 4 min and was terminated by adding 50 mM unlabeled nucleoside. Samples were passed through nitrocellulose membranes (Millipore), and the radioactivity remaining on the filters was counted in a liquid scintillation counter. The amount of nucleosides transported into membrane vesicles was calculated (19).
Measurement of Activity of Thymidine Kinase in Cell ExtractsThe activity of thymidine kinase in crude extracts from the dnaA46 mutant and wild type cells was measured, as described elsewhere (27). Exponentially growing cells were collected by centrifugation when the optical density at 660 nm reached 0.8. The precipitates were washed with buffer containing 50 mM Tris-HCl (pH 8.0), 1 mM dithiothreitol, and 1 mM EDTA, then were suspended in the same buffer. After sonication, the suspension was centrifuged at 20,000 × g for 30 min, and the supernatant was used for the assay. Protein in the extracts was determined according to Lowry et al. (28).
The standard reaction mixture contained 62.5 mM Tris-HCl (pH 8.0), 5 mM ATP, 10 mM MgCl2, 2.5 mM NaF, 1 mM dithiothreitol, 250 mg/ml bovine serum albumin, and 10 mM [14C]thymidine (53 mCi/mmol). Crude extracts were incubated in the reaction mixture at 37 °C for 10 min, and the reaction was terminated by boiling the samples. Aliquots (10 µl) were spotted on DEAE-cellulose paper (Whatman DE81), which was washed with 1 mM ammonium formate and 100% ethanol. The radioactivity remaining was measured in a liquid scintillation counter, and the amount of dTMP was calculated, as described (27).
Determination of Nucleotide Pools in CellsThe dTTP pools
in the dnaA46 mutant and wild type strain were determined as
described (29) but with some modifications. Exponentially growing cells
were harvested by centrifugation when the optical density at 660 nm
reached 0.8. The precipitates were subjected to extraction with 0.3 N HClO4, and pH values of the extracts were
adjusted to 5 by KOH solution containing 0.1 M EDTA, followed by centrifugation at 14,000 rpm for 1 min. The amounts of dTTP
in the supernatants were determined, as described (29). The standard
reaction mixture (200 µl) contained 0.5 µM
[-32P]dATP, 1.5 ng of the large fragment of DNA
polymerase I (Takara), 600 ng of poly(dA-dT) (Sigma) pretreated with
DNase I, 50 mM Tris-HCl (pH 8.0), and 5 mM
MgCl2. The reaction of DNA synthesis was carried out at
37 °C for 10 min and terminated by chilling on ice followed by
adding 10% (w/v) trichloroacetic acid. Samples were passed through
glass fiber filters. The amount of radioactivity on the filter was
measured in a liquid scintillation counter. The amount of dTTP was
determined as described (29).
The pools of other nucleotides in the dnaA46 mutant and wild type cells were determined as described (30). Briefly, exponentially growing cells in TG medium (31) containing 640 µM KH2PO4 and [32P]orthophosphate (3-4 µCi/mmol) at 37 °C were passed through nitrocellulose membranes (Millipore), when the optical density at 660 nm reached 0.4. The filters were subjected to extraction with 0.3 N HClO4, and pH values of the extracts were adjusted to 5 by KOH solution containing 0.1 M EDTA, followed by centrifugation at 14,000 rpm for 1 min. Unlabeled dNTP and NTP (50 nM) were added to the supernatant as standard samples and were spotted on polyethyleneimine-cellulose plates followed by soaking in 100% methanol for 5 min. Chromatography was performed in two dimensions; first dimension, 2 M LiCl, 2 M acetic acid (to start line) followed by 2.5 M LiCl, 2 M acetic acid (to 15 cm); second dimension, 0.75 M LiCl, 7.5% boric acid (to 2 cm) followed by 0.75 M LiCl, 10% boric acid (to 15 cm). The plates were autoradiographed, and each spot was identified using authentic nucleotides (dNTP and NTP). Quantitative analysis of the separated compounds was done as described previously (30).
The
incorporation of [3H]thymidine into trichloroacetic
acid-insoluble fractions of a dnaA46 mutant (KS1003) growing
at 37 °C was practically nil (data not shown), and this inhibition
was never evident when the cells were grown at 28 °C (data not
shown). Growth of KS1003 (dnaA46) at 37 °C monitored by
colony formation assay was not significantly different from that of the
wild type strain, KS1001 (Fig.
1A). Generation times of the
mutant and the wild type strain at 37 °C were 31 and 39 min,
respectively. We also examined the rate of DNA synthesis in the
dnaA46 mutant. As shown in Fig. 1B, much the same
rate of synthesis of DNA was observed in the mutant as in the wild type
cells. Thus, inhibition of [3H]thymidine incorporation in
the dnaA mutant cannot be explained by a decrease in the
rate of synthesis of DNA; rather the inhibition of thymidine transport
in the mutant may occur.
Incorporation of [3H]thymidine into trichloroacetic
acid-insoluble fractions in the wild type cells was inhibited (less
than 1% of control) by pretreatment of the cells with 200 µg/ml
nalidixic acid for 10 min (data not shown). Thus, when examining the
activity of thymidine transport in cells, we preincubated the cells
with nalidixic acid so that the effect of the DNA synthesis reaction on
the transport reaction would be eliminated. The activity of thymidine
transport in KS1001 growing at 37 °C was most apparent at a late
stage of log phage of cell growth (Fig.
2). On the contrary, low activity of
thymidine transport was observed in KS1003 growing at 37 °C, at the
corresponding stage (Fig. 2). The Km value for the
thymidine transport in wild type cells was 1 µM (data not
shown), and the transport experiments in Fig. 2 were done in the
presence of 15 µM thymidine. When cells were grown at
28 °C, no significant difference in the activity of thymidine transport was observed between KS1001 and KS1003 (data not shown). These results suggest that the inhibition of incorporation of [3H]thymidine into trichloroacetic acid-insoluble
fractions of the dnaA mutants growing at 37 °C is caused
by a decrease in the activity of thymidine transport in the mutant.
To confirm that the dnaA mutation is responsible for the inhibition of thymidine transport in KS1003, we transformed the strain with pHB10S, a plasmid carrying the wild type dnaA gene (21, 22). The activity of thymidine transport in the transformants recovered to much the same level as that in wild type cells (Fig. 2), whereas the recovery was nil when the cells were transformed with the vector pBR322 (data not shown). Transformation of the wild type strain with pHB10S did not affect the transport activity (data not shown).
We then asked whether the phenotype of the low level of thymidine transport in KS1003 would be cotransducible with the temperature-sensitive growth of this mutant in P1 phage-mediated transduction experiments. In the dnaA46 mutant, Tn10 with a tetracycline-resistant marker is located at a position close to the dnaA gene. P1 phages proliferated in mutant cells, transduced into W3110 cells, and the transductants were selected on tetracycline-containing plates (21, 22). Temperature sensitivities of cell growth and activities of thymidine transport were examined. Of the 53 independent transductants, 7 showed a high temperature-resistant growth and a high activity of thymidine transport, whereas all of the remaining 46 showed temperature-sensitive growth and a low activity of thymidine transport. No transductant showed either a high temperature-resistant growth and a low activity of thymidine transport or a temperature-sensitive growth and a high activity of thymidine transport. Thus, both phenotypes (a temperature-sensitive growth and a low activity of thymidine transport) were cotransducible. From the results of plasmid complementation and P1 phage-mediated transduction, it seems clear that the dnaA46 mutation is responsible for the low activity of thymidine transport in the KS1003 strain.
Temperature-sensitive dnaA mutants have been isolated and
the mutations identified (12). We examined the activity of thymidine transport in several other dnaA mutants. All dnaA
mutants tested showed lower activity of thymidine transport than did
the wild type strain (Fig. 3). The
activity of thymidine transport in the dnaA204 mutant was
more than half that in the wild type cells, whereas the activity in the
dnaA46 mutant was less than one-tenth that in the wild type
cells (Fig. 3). Thus, the extent of inhibition of the transport showed
an allele specificity. Plasmid complementation with pHB10S confirmed
that mutations in the dnaA gene are responsible for the
phenotype of all of these mutants (data not shown).
Effect of nupC or nupG Mutation on the Inhibition of Thymidine Transport in the dnaA46 Mutant
Genetic and biochemical studies
revealed that two transport proteins, NupC and NupG, play a major role
in the nucleoside transport in E. coli (16-19).
Transcription factors that regulate the expression of genes encoding
the proteins, the nupC and nupG genes, have been
identified (19). DnaA protein regulates the expression of a number of
genes through specific binding to DnaA boxes (12) or through other
mechanisms (13-15). Therefore, the possibility that inhibition of
thymidine transport in dnaA mutants may relate to
contribution of DnaA protein to expression of the nupC or
the nupG gene would need to be ruled out. If such is indeed
the case, in the presence of nupC or nupG
mutation, the dnaA mutation should not affect thymidine
transport. The dnaA46 mutation was then transduced into a
nupC or nupG mutant (ME8923 and ME8935,
respectively) by P1 phage-mediated transduction, and the activity of
thymidine transport in the transductants (KY1001 and KY1002,
respectively) was examined. Even in the presence of the nupC
or nupG mutation, the dnaA46 mutation caused
inhibition of thymidine transport (Fig. 4), indicating that the inhibition of
thymidine transport in the dnaA46 mutant cannot be explained
by the effect of the dnaA46 mutation on expression of
either the nupC or the nupG gene.
Nucleoside Specificity for Inhibition of Transport in the dnaA46 Mutant
As thymidine is transported through both NupC and NupG
systems (16-19), one would need to consider the possibility that these two systems are inhibited simultaneously in dnaA mutants.
Because the double mutant of the nupC and nupG
genes showed low activity for thymidine transport (data not shown), we
could not examine the effect of the double mutation of the
nupC and nupG genes on inhibition of thymidine
transport in the dnaA mutant. It was shown previously that
the NupC system can transport adenosine, cytidine, and thymidine but
not guanosine, whereas the NupG system can transport all nucleosides
(16-19). Therefore, if the effect of the dnaA mutation on
thymidine transport is caused by a decreased expression of both the
nupC and nupG genes, transport of other
nucleosides should also be inhibited in the dnaA mutants as
in the case with thymidine. As shown in Fig.
5, transport activities for adenosine,
guanosine, and cytidine in the dnaA46 mutant were much the
same as those in the wild type cells, indicating that inhibition of
nucleoside transport in the dnaA46 mutant is specific for
thymidine. Therefore, the inhibition of thymidine transport in the
dnaA46 mutant cannot be explained by inhibition of both NupC
and NupG systems in the mutant.
Thymidine Transport in Membrane Vesicles Prepared from the dnaA46 Mutant
The effect of the nupC or nupG
mutation and nucleoside specificity on inhibition of thymidine
transport in the dnaA46 mutant suggests that alteration in
the NupC and NupG systems is not involved in the inhibition. Because
unknown transport systems other than the NupC and NupG systems may
contribute to the inhibition of thymidine transport in the
dnaA mutant, we prepared membrane vesicles from the
dnaA46 mutant and wild type cells and examined their transport activities for nucleosides in vitro. As shown in
Fig. 6, the activity for thymidine
transport of membrane vesicles prepared from the dnaA46
mutant grown at 37 °C was slightly higher than from the wild type
cells. Inhibition of thymidine transport in the dnaA46
mutant apparently cannot be reproduced with membrane vesicles in
vitro. We also observed significant transport activities for other
nucleosides in membrane vesicles from the dnaA46 mutant, comparable to that from the wild type cells (data not shown). Thus, a
mechanism other than alteration in transport machinery in membranes is
involved in inhibition of thymidine transport in the dnaA46
mutant. Components in the cytosol, which had been removed during
preparation of membrane vesicles, seem to be responsible for the
inhibition of thymidine transport in the dnaA mutant. We
assumed that these unidentified components would specifically affect
thymidine transport because inhibition of nucleoside transport in the
dnaA mutant was observed only in case of thymidine.
Thymidine Kinase Activity and dTTP Pool in the dnaA46 Mutant
It was reported that thymidine kinase is necessary for
thymidine transport but not so for other nucleosides (32). This enzyme catalyzes the phosphorylation of thymidine and facilitates thymidine transport into cells (32). Therefore, we considered the possibility that the dnaA mutation decreases the activity of thymidine
kinase. We then compared the activity of thymidine kinase in a crude
extract from the dnaA46 mutant with that from the wild type
cells. As shown in Fig. 7, a significant
amount of thymidine kinase activity, comparable to that from the wild
type cells, was observed in the crude extract from the
dnaA46 mutant.
The activity of thymidine kinase is negatively regulated by dTTP (33).
When we examined thymidine kinase activity in the crude extracts, the
samples were diluted to more than 1,000 times before being subjected to
assay (Fig. 7). Therefore, the effect on the enzyme activity of dTTP in
cells would not reflect the result shown in Fig. 7. To rule out the
possibility that the activity of thymidine kinase was inhibited by a
higher concentration of dTTP in the dnaA mutant, the result
being the inhibition of thymidine transport, we compared the dTTP pool
in the dnaA46 mutant with that in wild type cells using
in vitro DNA synthesis (29). As shown in Fig.
8, the dTTP pool in the dnaA46
mutant at 37 °C was four times higher than that in the wild type
cells. At 28 °C no significant difference in the dTTP pool was
observed between the dnaA46 mutant and wild type cells (Fig.
8). We also examined the pools of other nucleotides in the mutant
growing at 37 °C. Exponentially growing cells were continuously
labeled with [32P]orthophosphate in TG medium (31). We
confirmed that transport of [3H]thymidine was indeed
inhibited in the dnaA46 mutant growing at 37 °C (data not
shown). Mixtures of nucleotides were prepared from cells and analyzed
by two-dimensional thin layer chromatography. The dTTP pool in the
dnaA46 mutant was three times higher than that in the wild
type cell in this assay (data not shown). The pools of other
nucleotides in the mutant were much the same as those in the wild type
cells (data not shown). These results suggest that the inhibition of
thymidine transport in the dnaA46 mutant is caused by
increases in the dTTP pool.
To test this notion we examined the effect of an increase in the dTTP
pool created by a factor unrelated to DnaA function. We examined the
effect of overproduction of thymidylate synthase of E. coli
on the dTTP pool and on thymidine transport. Introduction of pASTS1
plasmid, E. coli thymidylate synthase overexpression plasmid
(23), caused a two times higher concentration of dTTP (data not shown)
and one-third lower activity of thymidine transport (Fig.
9) than seen in cells transformed with
the vector pBR322. This result suggests that the inhibition of
thymidine transport in the dnaA46 mutant is at least partly
caused by an increase in the dTTP pool. The possibility that mechanisms
other than an increase in the dTTP pool may be involved in inhibition
of thymidine transport in the dnaA46 mutant would need to be
ruled out.
We obtained evidence that thymidine transport in temperature-sensitive dnaA mutants was drastically inhibited. Experiments on nucleoside specificity for the inhibition of transport, genetic analysis using nupC and nupG mutants, and measurement of transport activity of thymidine in membrane vesicles all support the thesis that the inhibition of thymidine transport in the dnaA mutants is not caused by modulation of transport machinery in membranes. There was a 3- or 4-fold increase of the dTTP pool in the dnaA46 mutant. Based on previous observations that dTTP negatively regulates the activity of thymidine kinase (33) and that thymidine kinase is essential for thymidine transport across the cytoplasmic membrane (32), we propose that the inhibition of thymidine transport in the dnaA mutants is caused by an increase in the dTTP pool. Studies on the effect of overexpression of thymidylate synthase on the dTTP pool and the thymidine transport supported this proposal.
To monitor DNA synthesis in cells, one usually measures the incorporation of radiolabeled thymidine into trichloroacetic acid-insoluble fractions. Some growth conditions have been reported to affect the dTTP pool in cells (30). Therefore, our results show that attention should be directed to alteration in the dTTP pool when pulse labeling cells with radiolabeled thymidine for examination of DNA synthesis.
How the dnaA mutations increase the dTTP pool remains unknown. Because dTTP is one of the substrates of DNA synthesis, the inhibition of DNA synthesis caused by the dnaA mutations seems to increase the dTTP pool. However, our data seem to rule this out. The rates of growth and of DNA synthesis of the dnaA46 mutant growing at 37 °C were indistinguishable from findings in the wild type cells. It was reported that the dTTP pool in cells was not altered when DNA synthesis was inhibited in the presence of nalidixic acid (30). Actually, in our assay for the transport of thymidine, cells were pretreated with nalidixic acid.
Filpula and Fuchs reported that the activity of RDP reductase1 in the temperature-sensitive dnaA mutants was increased when the incubation temperature was shifted to 42 °C, a nonpermissive temperature (34). We recently found that this enzyme was overproduced in the dnaA46 mutant growing at 37 °C.2 Because RDP reductase catalyzes the synthesis of dUDP and dCDP from UDP and CDP, respectively (35), and because dTTP is synthesized de novo from dUDP and dCDP (35), the possibility that the induction of RDP reductase in the dnaA mutant at 37 °C causes increases in the dTTP pool would need attention. The following evidence suggests that this notion can indeed be ruled out. First, the decrease in pool size in the dnaA mutants is specific for dTTP. The overproduction of RDP reductase should increase pool sizes of all four deoxyribonucleotides. Second, the dnaA46 mutant carrying the thyA mutation showed dnaA46 mutation-dependent inhibition of thymidine transport (data not shown). The thyA mutant cannot synthesize dTTP de novo (36, 37). Thus, if the inhibition of thymidine transport in the dnaA mutant is indeed caused by increases in the dTTP pool resulting from a high level of RDP reductase activity, the inhibition of thymidine transport by the dnaA mutation would not be observed in the thyA mutant. Therefore, the induction of RDP reductase may not explain the inhibition of thymidine transport in the dnaA46 mutant.
DnaA protein, the initiator of chromosomal DNA replication in E. coli, plays a key role in regulating DNA replication. Our observations suggest that DnaA protein is also involved in nucleoside transport. DnaA protein may have an important role in the coupling of DNA replication with nucleotide metabolism, which provides substrates of DNA synthesis.
We thank Dr. Y. Komatsu (Shionogi Pharmaceutical Co.) for helpful discussion and M. Ohara for comments on the manuscript. Dr. A. Nishimura (National Institute of Genetics, Japan) and Dr. M. Iwakura (National Institute of Bioscience and Human Technology, Japan) kindly provided the bacterial strains and plasmids, respectively.