Cell division genes ftsQAZ in Escherichia coli require distant cis-acting signals upstream of ddlB for full expression


  • Klas Flärdh,

    1. Departamento de Biología Celular y del Desarrollo, Centro de Investigaciones Biológicas, Consejo Superior de Investigaciones Científicas, Madrid, Spain.
    Search for more papers by this author
    • Present address: Department of Genetics, John Innes Centre, Colney Lane, Norwich NR4 7UH, UK

  • Pilar Palacios,

    1. Departamento de Biología Celular y del Desarrollo, Centro de Investigaciones Biológicas, Consejo Superior de Investigaciones Científicas, Madrid, Spain.
    Search for more papers by this author
  • Miguel Vicente

    1. Departamento de Biología Celular y del Desarrollo, Centro de Investigaciones Biológicas, Consejo Superior de Investigaciones Científicas, Madrid, Spain.
    Search for more papers by this author

Miguel Vicente E-mail cibmv5j@fresno.csic.es; Tel. (91) 561 18 00, ext. 4375; Fax (91) 562 75 18.


A transcriptional reporter fusion has been introduced into the chromosomal ftsZ locus in such a way that all transcription that normally reaches ftsZ can be monitored. The new Φ(ftsZ–lacZ ) fusion yields four times more β-galactosidase activity than a ddlB–ftsQAZ–lacZ fusion on a lambda prophage vector. A strongly polar ddlB ::Ω insertion prevents contributions from signals upstream of the ftsQAZ promoters and decreases transcription of the chromosomal Φ(ftsZ–lacZ ) fusion by 66%, demonstrating that around two-thirds of total ftsZ transcription require cis-acting elements upstream of ddlB. We suggest that those elements are distant promoters, and thus that the cell division and cell wall synthesis genes in the dcw gene cluster are to a large extent co-transcribed. The ddlB ::Ω insertion is lethal unless additional copies of ftsQA are provided or a compensatory decrease in FtsZ synthesis is made. This shows that ddlB is a dispensable gene, and reinforces the critical role of the FtsA/FtsZ ratio in septation. Using the new reporter fusion, it is demonstrated that ftsZ expression is not autoregulated.


Cell division in bacteria requires the tubulin homologue FtsZ, which is highly conserved in a wide range of prokaryotes, including cell wall-less mycoplasmas (for a recent review, see Bramhill, 1997). As the earliest recognized stage of septation, FtsZ assembles into a contractile ring structure at the future division site (Bi and Lutkenhaus, 1991; Wang and Lutkenhaus, 1993; Ma et al., 1996; Pogliano et al., 1997), and it may there act to recruit additional proteins that are known to localize to this site (Addinall and Lutkenhaus, 1996; Ma et al., 1996; Addinall et al., 1997; Hale and de Boer, 1997; Harry and Wake, 1997; Weiss et al., 1997). The mechanisms controlling assembly of the ftsZ ring and onset of septation are still unknown. FtsZ stoichiometry, however, has a role because ftsZ expression has to be controlled within a very narrow range in relation to both cell mass and the expression of other division genes, notably ftsA, for division to occur at the correct time and place (Ward and Lutkenhaus, 1985; Dai and Lutkenhaus, 1992; Dewar et al., 1992; Palacios et al., 1996; Begg et al., 1998). Interestingly, ftsZ expression is inversely growth rate dependent (Aldea et al., 1990), and the abundance of ftsZ transcripts oscillates in time with the cell division cycle in cultures dividing or replicating synchronously (Garrido et al., 1993; Zhou and Helmstetter, 1994). The natural regulatory signals are required for correct control of the gene, as evidenced by the abnormal behaviour of strain VIP205, in which transcription of ftsZ has been artificially placed under the control of a tac promoter (Garrido et al., 1993; Palacios et al., 1996). Although the significance of these observations is not yet clear, they raise questions about the extent to which ftsZ expression is regulated and what role this may have in controlling the timing of cell division in the Escherichia coli cell cycle.

In E. coli, ftsZ is located at the distal end of a complex gene cluster comprising 16 genes involved in division and cell wall synthesis (dcw gene cluster) (Ayala et al., 1994). Six promoters within the immediately upstream ddlB, ftsQ and ftsA genes contribute to ftsZ expression (Rothfield and Garcia-Lara, 1996; Flärdh et al., 1997). In addition, transcriptional readthrough from distant upstream promoters may reach the gene, as the dcw cluster contains no recognizable terminators (Ayala et al., 1994). However, as no transcript mapping experiments have directly addressed the role of such upstream promoters (Aldea et al., 1990; Zhou and Helmstetter, 1994; Cam et al., 1996), and as ftsZ mRNAs are processed by RNase E at two sites just 5′ to the structural gene (Cam et al., 1996), it has not been possible to conclude which are the major ftsZ promoters and what controls they may be subjected to.

We describe here the introduction of an Φ(ftsZ–lacZ ) fusion at the chromosomal ftsZ locus. In contrast to previously used constructions, this allows monitoring of all transcription normally reaching ftsZ. Using this system, we show that the six ftsQAZ promoters within or downstream of ddlB only give rise to around one-third of the normal level of ftsZ transcription, and that upstream cis-acting signals, probably promoters, are required for full expression. As a consequence of this work, attention should be directed towards characterizing the putative upstream promoters that control the co-transcription of several cell division genes in the dcw gene cluster.


A lacZ reporter fusion in the chromosomal ftsZ locus

To monitor all transcription that normally enters ftsZ, the gene was replaced by a transcriptional Φ(ftsZ–lacZ ) fusion at its chromosomal locus. This was achieved by integrating pKFV116 into the chromosome of strain VIP374 (Fig. 1A). The resulting plasmid co-integrate strain VIP406 is ampicillin resistant, Lac+ and, because ftsZ is expressed from the tac promoter, IPTG dependent. The integrated plasmid was transduced into MC1061 to create strain VIP407. The structures of the ftsZ regions in VIP406 and VIP407 were confirmed by Southern blot analysis of EcoRV-digested chromosomal DNA using ftsAZ and lacZ probes. As shown in 1Fig. 1B, a 1350 bp ftsAZ fragment present in MC1061 was replaced by two characteristic fragments of 2216 and 2011 bp in the plasmid co-integrates, the latter of which also hybridized to the lacZ probe. In VIP406 and VIP407, the absence of the 2877 bp fragment, characteristic for the intact pKFV116, was of particular importance as it showed that only a single copy of pKFV116 had been integrated (Fig. 1B).

Figure 1.

. Introduction of a transcriptional Φ(ftsZ–lacZ ) reporter fusion into the chromosomal dcw gene cluster. A. Plasmid pKFV116 was integrated into the chromosome of strain VIP374 by homologous recombination at the ftsA–ftsZ junction. The ftsZ gene is represented by a grey box, whereas the other genes of the dcw gene cluster, the reporter gene lacZ and the bla and lacI Q genes are shown as white boxes. Chromosomal genes that do not belong to the dcw cluster are shown as black boxes. Not all parts of pKFV116 are drawn to scale, and only the approximate location of the bla gene is shown. Direction of transcription of bla and lacZ is indicated by arrowheads inside them. Relevant EcoRV (E) restriction sites, probes used for Southern blots, and the approximate locations of the oligonucleotide primers (small arrows) used to verify the resulting strains are indicated. B. Southern blot analysis of plasmid pKFV116 co-integrate strains. EcoRV-digested total DNA of strains MC1061 (lanes 1 and 5), VIP406 (lanes 2 and 6), VIP407 (lanes 3 and 7) and CC118(λpir ) with pKFV116 as an independent replicon (lanes 4 and 8) were probed with non-radioactively labelled ′ftsA–ftsZ′ (lanes 1–4) and lacZ (lanes 5–8) fragments.

The ftsZ gene is not autoregulated

When VIP407 was grown in LB medium containing different concentrations of IPTG, the average cell volume was close to that of the parent MC1061 at 20 μM IPTG, whereas cell volume increased at both higher and lower concentrations (Table 1). At IPTG concentrations below 2 μM and at 200 μM or higher, cell division was inhibited and cultures failed to grow exponentially (data not shown). These results show that the IPTG dependence of VIP407 is similar to that previously reported for strain VIP205, in which ftsZ also is controlled from a tac promoter (Garrido et al., 1993; Palacios et al., 1996). At the same time as the cellular FtsZ levels can be varied by altering the IPTG concentration, transcription of the ftsZ locus can be quantified using the Φ(ftsZ–lacZ ) fusion in VIP407. Taking advantage of these features, we addressed the question of whether ftsZ is autoregulated. Although the FtsZ content varied with IPTG concentration, expression of the Φ(ftsZ–lacZ ) fusion was not significantly affected within the range that allows balanced growth to be established (Table 1). Similar experiments with VIP406 showed no significant alterations in β-galactosidase activity within 2–50 μM IPTG (data not shown). Thus, within a close to 20-fold range of FtsZ levels in which exponential growth was sustained, FtsZ did not affect transcription of its own gene.

Table 1. . Effect of altered level of ftsZ expression on cell volume and Φ(ftsZ–lacZ ) expression in strain VIP407. a. An exponentially growing culture of VIP407 in LB thi Ap 20 μM IPTG was washed and divided into media with different IPTG concentrations. Cultures were kept in exponential growth by dilutions and were sampled after 3 h. Similar results were obtained after 5 h.b. Values are related to MC1061 growing exponentially in the same medium lacking antibiotics. The median cell volumes, determined using a Coulter Counter, provide an underestimation of the real change in cell volume for the abnormally long cells present in 6 and 100 μM IPTG (Palacios et al., 1996).c. Very long filaments were formed which prevented determination of cell volumes using the Coulter Counter.Thumbnail image of

A polar insertion in ddlB decreases expression of ftsA and ftsZ

The β-galactosidase activity produced by VIP407 (1320 Miller units) was 4.4 times higher than that produced by an MC1061 λKFV2 lysogen (300 Miller units) in the same growth conditions (Table 2). λKFV2 contains a ‘ddlB–ftsQAZ ’ fragment fused to lacZ (Flärdh et al., 1997). As the Φ(ftsZ–lacZ ) fusion joints and the lacZ genes are identical in VIP407 and λKFV2, the discrepancy in β-galactosidase activities could be due either to cis-acting signals that are missing from λKFV2, or to their different locations on the chromosome, with λKFV2 being located at the attλ site and the fusion in VIP407 being integrated at the ftsZ locus.

Table 2. . β-Galactosidase activities produced from single-copy Φ(ftsZ–lacZ ) fusions. a. Cultures were grown for more than 10 generations at 37°C in LB thi supplemented with 10 μM IPTG and, when appropriate, Ap and Sp.b. Average values for cultures grown at 20–200 μM IPTG. Growth rates and β-galactosidase activities did not vary significantly between the different IPTG concentrations.Thumbnail image of

To distinguish between these two possibilities, the Ω interposon containing strong transcriptional terminators was introduced into the ddlB gene of strain VIP406, close to the position corresponding to the starting point of the insert in λKFV2 (Fig. 2A). As it had not been established whether ddlB is an essential gene, a ddlB merodiploid was constructed using λKFV120, which carries the ddlB ::Ω disruption. Southern blot analysis of the VIP406 λKFV120 lysogens demonstrated that λKFV120 had integrated by a recombinational cross-over downstream of the chromosomal ddlB locus, leaving the ddlB ::Ω insertion located immediately upstream of the Φ(ftsZ–lacZ ) fusion (Fig. 2A and data not shown). Strain VIP433 produced only 34% of the β-galactosidase activity of its parent VIP406 (Table 2). This suggested that most of the difference in activity between VIP407 and MC1061 λKFV2 can be accounted for by upstream cis-acting signals that are missing from λKFV2. The β-galactosidase values might, however, have been affected by division problems in VIP433, which grows as short filaments and displays morphological aberrations at division sites and cell poles (data not shown). A likely explanation for this defect is the alteration in ftsA expression imposed by the genetic structure of VIP433. The end point of the fragment cloned in the parent of λKFV120 is a HindIII site within ftsA (Begg et al., 1980). Thus, VIP433 is expected to synthesize a truncated form of FtsA from the allele present to the left of the inserted lambda, and a full-length version from the allele present downstream of ddlB ::Ω (Fig. 2A). Indeed, a new, smaller protein was recognized by an FtsA antiserum in Western blots of VIP433. In addition, the normally sized FtsA was present at a significantly diminished level, which indicates that ddlB ::Ω has a polar effect on expression of the full-length ftsA (data not shown).

Figure 2.

. Integration of the polar Ω interposon upstream of the Φ(ftsZ–lacZ ) reporter fusion in strain VIP406. A. Lysogenization of strain VIP406 (Δattλ ) with the ddlB ::Ω transducing phage λKFV120. The single cross-over that leads to integration of λKFV120 can occur on either side of the Ω element, but only the integration event leading to strain VIP433 is shown. The ddlB gene is represented by a cross-hatched box and the Ω element by a doubly cross-hatched box. The hairpin symbol denotes one of the transcriptional terminators contained in the Ω element. Arrows show positions of known promoters in the ddlB-ftsQAZ region. The rest of the dcw cluster is drawn as in Fig. 1A. B. Curing of the λKFV121 prophages. The recombination events that lead to excision of λKFV121 from strains VIP434/439 are schematically represented by crosses. Recombination at I yielded spectinomycin-resistant ddlB ::Ω progeny (VIP491/490) whereas recombination at II yielded sensitive ddlB+. The probe and the NaeI (N) and XbaI (X) restriction sites that were used in Southern blot analysis, and the oligonucleotide primers (small arrows) used in PCR reactions are indicated. With the exception of the NaeI site in Ω, there are no sites for these enzymes in λKFV120.

The ddlB gene is dispensable and ddlB::Ωis polar on ftsQA

The anomalies of strain VIP433 would be eliminated if it could be cured from its prophage to make a ddlB ::Ω ftsA+ haploid strain. However, a failure to obtain such SpR strains would indicate that ddlB ::Ω is a lethal mutation, either because ddlB is essential or because of polarity on the essential genes ftsQ and ftsA. To facilitate curing experiments, phage λKFV121 was used to create strain VIP434, which is identical to VIP433 except for the prophage being immλ cI857. Cured derivatives of VIP434 were isolated after counterselection of the thermosensitive lysogens (Fig. 2B). It was found that cured SpR strains could not be recovered at 20 μM or more IPTG, indicating that ddlB ::Ω is lethal during these conditions (Table 3). However, lethality could be suppressed by lowering the IPTG concentration and, although they grew poorly and required 48 h of incubation to acquire normal colony size, SpR strains were recovered at 5 and 10 μM IPTG with a significant frequency (Table 3). In contrast, SpR strains showing no signs of growth deficiency were obtained at high frequency at 10–60 μM IPTG from strain VIP439, which contains additional copies of ftsQA on a plasmid of moderate copy number (Table 3). We conclude that ddlB ::Ω is polar on ftsQA and that this may be lethal unless ftsZ expression is decreased or additional copies of ftsQA are provided. It is not known why the frequency of cured SpR strains at some IPTG concentrations was similar to or even higher than the number of the cured strains obtained without Sp selection. If the recombination event that leads to loss of the prophage can occur with equal probability on either side of the prophage, only around 50% of the cured strains would have been expected to be SpR.

Table 3. . IPTG-dependent frequency of ddlB ::Ω (SpR) and ddlB+ (SpS) haploid derivatives produced by curing of λKFV121 from VIP434 and VIP439. a. Cultures of VIP434 and VIP439 were heat shocked, incubated 1 h at 30°C and plated on LB thi Ap with or without Sp and with different IPTG concentrations and incubated for 24 h at 42°C.b. Number of colonies on Sp-containing plates divided by the number of colonies on plates without Sp.c. Colonies grew very slowly and were counted after 48 h of incubation at 42°C.ND, not determined.Thumbnail image of

All cured SpR derivatives of VIP434 and VIP439 that were tested (six isolates of each) lacked the ddlB+ and contained only the ddlB ::Ω allele as deduced from PCR amplifications (data not shown). For two such strains, designated VIP491 and VIP490, respectively, Southern blots confirmed the structure of their dcw gene clusters (data not shown). VIP491 grew poorly, presumably because of deficient ftsQA expression. On the other hand, strain VIP490 grew well and was similar to the ddlB+ counterpart VIP489 in terms of growth rate and morphology (data not shown). These results prove that ddlB is a non-essential gene, and, although an exhaustive analysis of the mutants falls outside the scope of this paper, no obvious phenotype could be associated with inactivation of ddlB.

cis-Acting elements upstream of the ddlB gene are required for two-thirds of total ftsZ transcription

To avoid unwanted side-effects of growth deficiencies and decreased ftsQA expression, analyses of the effect of ddlB ::Ω on the Φ(ftsZ–lacZ ) fusion were made in strains VIP490 and VIP489. The presence of pKFV122 does not influence expression of Φ(ftsZ–lacZ ) as strain VIP489 produces β-galactosidase activities similar to those of VIP406 (Table 2). In agreement with the results from VIP407 (Table 1), there were no significant effects of IPTG-concentration on β-galactosidase activity in either VIP490 or VIP489 between 20 and 200 μM (data not shown). Averaged over this interval, VIP490 produced 34% of the β-galactosidase activity of VIP489 (Table 2). Thus, the ddlB ::Ω mutation reduces transcription of the ftsZ locus by two-thirds in a strain that grows normally and lacks the truncated ftsA allele of VIP433.


Are promoters upstream of ddlB required for a major fraction of ftsZ transcription?

We have shown that distant sequence elements upstream of the ftsQ2p1p promoters in ddlB are required for approximately two-thirds of the total transcription that normally reaches the ftsZ gene. First, the promoter activity of a ddlB–ftsQAZ fragment fused to lacZ on a lambda phage is only about 30% of the activity of the same Φ(ftsZ–lacZ ) fusion placed in the ftsZ locus with all the upstream sequences of the dcw gene cluster intact and affecting the fusion in cis. Second, a strongly polar ddlB ::Ω insertion reduced the activity of the Φ(ftsZ–lacZ ) fusion at the ftsZ locus by two-thirds. Third, the ddlB ::Ω mutation is polar on ftsA expression, showing that the effect is not specific for expression of the transcriptional fusion. Furthermore, we can exclude the possibility that ddlB in itself has an effect on ftsZ expression because the ddlB+/ddlB ::Ω merodiploid VIP433 has a β-galactosidase activity close to that of the ddlB ::Ω haploid VIP490 (Table 2).

The simplest explanation of the polarity of ddlB ::Ω on Φ(ftsZ–lacZ ) is that distant upstream promoters drive two-thirds of ftsZ expression. Alternatively, the insertion may disturb the interaction between a hypothetical upstream activating element and any of the promoters within or downstream of ddlB. Although the latter possibility cannot be excluded by the results presented here, other observations support the importance of upstream promoters. First, there are no recognizable transcriptional terminators within the dcw gene cluster, making it possible that RNA polymerases starting at the head of the cluster can transcribe all the way through to the last gene, envA (Ayala et al., 1994). Second, a report by Dai and Lutkenhaus (1991) suggests that significant contributions to ftsZ expression are made from the upper half of the gene cluster. They found that prophage λ16–2, which carries a sequence extending from within ftsW to the end of the dcw cluster, failed to complement an ftsZ null mutation in trans and produced only 60–70% of the normal cellular FtsZ concentration. Third, Hara et al. (1997) have identified recently a promoter at the 5′ end of the dcw cluster and showed that this is required to attain the minimal levels of expression of some of the genes from mraZ to, and including, ftsW that are necessary for growth and viability. It is likely that this and other promoters that have been suggested to exist upstream of ddlB have an effect on ftsZ (see Discussion and references in Hara et al., 1997). When combined with previously reported estimations of the activities of the ftsQAZ promoters (Flärdh et al., 1997), our results imply that during rapid growth around 66% of total ftsZ transcription may derive from promoters that are upstream of ddlB or are sensitive to signals present in this region; ≈15% from the ftsQ2p1p promoters within ddlB ; 12% from ftsZ4p3p inside ftsA ; whereas only around 4% and 2% appear to originate from the ftsAp and ftsZ2p, respectively (Fig. 3).

Figure 3.

. Estimated roles of different chromosomal regions in the total transcription of ftsZ. The estimations of contributions from different individual promoters within or downstream of ddlB have been published previously (Flärdh et al., 1997). It is demonstrated in this paper that around two-thirds of total ftsZ transcription requires cis-acting signals upstream of ddlB. These signals may be distant promoters or enhancer-like elements that act on any of the ftsQAZ promoters. Scissors indicate the position of two RNase E processing sites.

Implications for ftsZ regulation

Several studies have addressed the regulation of the ftsQAZ promoters within or downstream of ddlB. It has been suggested that the Gearbox promoter ftsQ1p contributes to producing constant amounts of FtsQ, A and Z per cell cycle (Vicente et al., 1991). It is now known that the alternative sigma factor σS recognizes ftsQ1p, and that the SdiA activator, possibly responding to a cell density-dependent signal, affects ftsQ2p (Wang et al., 1991; García-Lara et al., 1996; Sitnikov et al., 1996; Ballesteros et al., 1998). As demonstrated here, however, these promoters are responsible for a small fraction of ftsZ transcription and their significance is not clear as they may only have modulating effects on the total amount of gene product produced or be important under specific growth conditions.

One aim of this work was to understand better the regulation of ftsZ expression and in particular the growth rate and cell cycle-dependent controls acting on the gene. Garrido et al. (1993) have detected cell cycle-dependent alterations in the abundance of ftsZ mRNA with 5′ ends close to the start of the gene. This mRNA population was likely to consist mainly of processed messages (Cam et al., 1996; Flärdh et al., 1997), but Zhou and Helmstetter (1994) extended the observations by demonstrating oscillations also in longer transcripts with starting points within or upstream of the ftsA coding region. Our demonstration of the role of upstream signals means that such oscillations could arise at distant cell cycle-dependent promoters and thus affect many genes within the cluster. Reporter strains like VIP407 will enable studies of the impact of such promoters or mechanisms on the overall transcription of ftsZ. Here, we show that variation of the cellular FtsZ level within a close to 20-fold range has no autoregulatory effects on transcription of the gene itself. As the FtsZ protein appears to switch between a soluble cytoplasmic location and a polymeric membrane-associated state during the cell cycle (Bi and Lutkenhaus, 1991), it was possible that an autoregulatory function of either of the forms caused oscillations in ftsZ transcription. Although it cannot be completely excluded that the average level of a hypothetical form of FtsZ active in regulation was unaffected despite the changes in total FtsZ, our results suggest that autoregulation does not contribute to ftsZ mRNA oscillations. In addition, we have found that the dam methyltransferase has no effect on Φ(ftsZ–lacZ ) expression in VIP407 (unpublished). Thus, the oscillations in ftsZ transcripts appear not to be controlled in the same manner as dnaA, the transcription of which is sensitive to dam-dependent DNA methylation in such a way that expression of dnaA is inhibited during an interval of the cell cycle when the newly replicated gene is hemimethylated (Campbell and Kleckner, 1990).

Importance of the FtsA/FtsZ ratio

We have found that ddlB ::Ω causes severe division problems and a decrease in the cellular content of FtsA. The division problems are due to polarity on ftsQA as they are suppressed by a plasmid carrying ftsQA. Although we have not experimentally distinguished between the importance of FtsQ and FtsA, it is likely that the effects primarily are caused by alterations in the FtsA level because E. coli is comparatively insensitive to changes in ftsQ expression in rich media (Storts et al., 1989; Carson et al., 1991; Dai and Lutkenhaus, 1992; Begg et al., 1998). In the absence of the ftsQA plasmid, ddlB ::Ω is lethal at normal FtsZ concentrations, but lethality is suppressed by a compensating decrease in ftsZ expression. Previous studies have shown that moderate overexpression of one of FtsA or FtsZ may block cell division, and that this is suppressed by overexpressing the other gene as well (Dai and Lutkenhaus, 1992; Dewar et al., 1992). Our results indicate that also the detrimental effects of decreased ftsA expression can be overcome by lowering the FtsZ level, and, hence, they reinforce the importance of the FtsA/FtsZ ratio in cell division. FtsA and FtsZ are known to interact (Wang et al., 1997), and FtsA localizes to the septation site in an FtsZ-dependent manner (Addinall and Lutkenhaus, 1996; Ma et al., 1996). Apparently, the interaction is disturbed by the altered ratio between the two proteins.

The ddlB gene is dispensable

It is shown here that ddlB is a non-essential gene. As its gene product catalyses a process expected to be essential, namely the formation of D-alanyl-D-alanine required for cross-linking of peptidoglycan strands, it is implied that the isoenzyme-encoding ddlA gene (Zawadzke et al., 1991) provides sufficient D-alanine:D-alanine ligase activity to sustain peptidoglycan assembly and cell viability. A previously isolated conditional lethal ddlB(Ts) mutation was complemented by ddlA on a lambda vector, and it was therefore not clear why the original ddlB(Ts) mutant ST640 would be thermosensitive (Zawadzke et al., 1991). One possibility was that ddlB would have a function that could not be complemented by the chromosomal ddlA. This is, however, ruled out by our finding that ddlB can be disrupted without any obvious phenotype. S. typhimurium contains a homologue of ddlA, which is non-essential, suggesting the existence of two ddl genes also in that organism (Daub et al., 1988). On the other hand, only one ddl gene is present in H. influenzae (Fleischmann et al., 1995) and no other bacteria are known to posses more than one ddl. As suggested by our results, expression of ddlB may be intimately coupled to the surrounding genes in the dcw cluster. A possible advantage of having two ddl genes is that it would enable an independent regulation of this step in peptidoglycan synthesis.

Experimental procedures

Bacterial strains and phages and their cultivation

The strains of Escherichia coli K-12 and bacteriophage lambda that were used in this study are listed in Table 4. Lambda and P1vir phages were propagated and handled according to the protocols of Silhavy et al. (1984). Bacterial strains were isolated and cultivated on solid LB medium supplemented, when appropriate, with thiamine (thi) at 5 μg ml−1; ampicillin (Ap) at 50 μg ml−1; spectinomycin (Sp) at 25 μg ml−1 for single-copy markers and 50 μg ml−1 for plasmids; chloramphenicol (Cm) at 20 μg ml−1; IPTG at 20 μM if not stated otherwise; and X-gal at 40 μg ml−1.

Table 4. . Bacterial strains and phages used in this study (Herrero et al. (1990); Casadaban and Cohen (1980); Simons et al. (1987)).Thumbnail image of

DNA manipulations

General DNA manipulations were according to standard protocols (Sambrook et al., 1989). Oligonucleotide primers used in this study are listed in Table 5. For ligations, DNA fragments were isolated in low melting point agarose gels (gift from S. A. Hispanagar) and ligated directly in the agarose. Transformations were performed using the PEG/DMSO one-step procedure (Miller, 1992). Chromosomal DNA was extracted using cetyltrimethylammonium bromide as described previously (Ausubel et al., 1991). Probes for Southern blots were non-radioactively labelled and hybridization and colorimetric detection were carried out using Boehringer's Nucleic Acid Labelling and Detection kit according to the manufacturers instructions.

Table 5. . Oligonucleotide primers used in this study. a.+ signifies that the oligonucleotide sequence is present in the sense strand and − signifies that it is present in the antisense strand of the specified gene.Thumbnail image of

Plasmid construction

Plasmid pKFV116 (Fig. 1) was constructed as follows. A 1550 bp NruI–XbaI fragment containing the lacI Q gene and the tac promoter from plasmid pJF119HE (Fürste et al., 1986) was cloned between the EcoRV and XbaI sites of pGP704 (Miller and Mekalanos, 1988) to create pKFV115. This and other derivatives of pGP704 were propagated in CC118(λpir ). A 3717 bp AscI–NruI fragment from pKFV2 (Flärdh et al., 1997) carrying the C-terminal part of ftsA and an Φ(ftsZ–lacZ ) transcriptional fusion was made blunt-ended using Klenow polymerase and cloned in the Klenow-treated XbaI site of pKFV115, so that the tac promoter of pKFV115 would direct transcription into the truncated ftsA sequence.

To clone the ddlB gene, a 2565 bp DNA fragment from phage λddl + was amplified using PCR and oligonucleotides KF7 and LY19 (Table 5). A 2484 bp HindIII–Bgl II fragment of the PCR product was cloned between the HindIII and BamHI sites of pSU18 (Bartoloméet al., 1991) to create plasmid pKFV117. The ddlB gene on pKFV117 was disrupted by digesting at the Sal I sites within the gene, filling in the ends with Klenow polymerase, and inserting the Ω interposon (Prentki and Krisch, 1984) as a 2.0 kb SmaI fragment, thus creating plasmid pKFV119, which has basepairs 123–133 of the ddlB structural gene replaced by Ω.

Plasmid pKFV122 was made by inserting a 2966 bp XbaI–Pst I fragment from pKFV2 (Flärdh et al., 1997) between the NdeI and Pst I sites of pSU19 (Bartoloméet al., 1991). Therewith, the lac promoter of pSU19 was deleted, as it otherwise might have interfered with the regulation of the tac promoter in strain VIP406.

Construction of Φ( ftsZ–lacZ) reporter strains

Plasmid pKFV116, which only replicates in hosts expressing the π protein (Miller and Mekalanos, 1988), was electroporated into the non-permissive host VIP374, and ApR Lac+ colonies were selected on plates containing 10–30 μM IPTG. The IPTG concentrations were selected on the basis of previous experiences with VIP205, which also has its single ftsZ gene controlled from a tac promoter (Garrido et al., 1993). IPTG dependence was tested by streaking on plates where IPTG had been omitted or substituted for 0.1% glucose. Genetic linkage of integrated pKFV116 to leu was tested by growing phage P1 on the resulting co-integrate strain VIP406, transducing strain VIP183 to ampicillin resistance, and scoring for tetracycline sensitivity. The location of the integrated plasmid at the ftsAZ locus was confirmed using PCR with primer pairs TG17-MA10, TG20-MF19 and TG20-TG17 (Table 5). The physical structure of the ftsAZ region of the chromosomes of VIP406 and VIP407 was confirmed in Southern blots probed with a 1885 bp Bgl II–HpaI fragment (‘ftsAZ’ probe) from plasmid pZAQ (Ward and Lutkenhaus, 1985) and a 3348 bp BamHI–NruI fragment (lacZ probe) from pTL61T (Linn and St. Pierre, 1990).

Construction of ddlB:: Ωtransducing phages

The disrupted ddlB ::Ω allele on pKFV119 was crossed onto phage λddl + by growing λddl + on MC1061 containing pKFV119. Recombinant phages were isolated by lysogenization of MC1061, selection of SpR CmS lysogens, UV induction of such a lysogen, and another round of lysogenization and induction. The resulting lysate contained a purified imm21 SpR phage, which was designated λKFV120.

λKFV121, a thermoinducible derivative of λKFV120, was obtained by growing λcI857 on an MC1061 λKFV120 lysogen, using the lysate to infect a RYC1000 λRS74 lysogen, and selecting SpR Lac lysogens. After testing thermosensitivity and lambda immunity phenotype, one lysogen was purified and heat induced to obtain λKFV121. The ddlB ::Ω alleles on λKFV120 and λKFV121 were confirmed by PCR and subsequently in Southern blots of lysogens as described below.

Construction and curing of ddlB:: Ωlysogens

Strain VIP406 was lysogenized with λKFV120 and λKFV121, yielding strains VIP433 and VIP434 respectively. VIP434 was transformed with pKFV122 to produce VIP439. Curing of λKFV121 prophages from strain VIP434 and VIP439 was carried out essentially as described previously (Xiao et al., 1991), except that plating was performed on LB thi Ap, with and without Sp and supplemented with IPTG at 2, 5, 10, 20, 40 and 60 μM. Cm was included in plates used for strain VIP439. All ddlB ::Ω strains were confirmed in Southern blots of NaeI–XbaI-digested chromosomal DNA, probed with a Bst XI–Pst I fragment form pKFV119 containing parts of murC, ddlB and the Ω interposon.

β-Galactosidase assays

Liquid cultures for determination of β-galactosidase activities were grown at 37°C with vigorous shaking in LB medium supplemented with IPTG, thi and when appropriate Ap and Sp. The cultures were maintained in exponential growth at an optical density at 600 nm (OD600) lower than 0.25 by dilutions into prewarmed medium for at least 10 doubling times before samples were taken for β-galactosidase assays at OD600 between 0.05 and 0.2. β-Galactosidase assays were performed as described previously (Flärdh et al., 1997).

Measurement of FtsZ and FtsA concentrations and cell volumes

Cellular FtsZ and FtsA contents were determined by Western blotting as described previously (Pla et al., 1990; Palacios et al., 1996), except that blots were revealed using a luminescence detection kit (Boehringer Mannheim). Cell volumes were determined using a Coulter Counter as described previously (Palacios et al., 1996). Phase-contrast microscopy was performed using cells fixed in 0.75% formaldehyde using a Zeiss photomicroscope III.


  1. Present address: Department of Genetics, John Innes Centre, Colney Lane, Norwich NR4 7UH, UK


We thank Víctor de Lorenzo and Ken Begg for gifts of strains and Keith Chater for comments on the manuscript. This research has been funded by projects from the European Commission (BIOTECH) and Ministerio de Educación y Ciencia to M.V.K.F. was supported by fellowships from the Swedish Natural Research Council and Comisión Interministerial de Ciencia y Tecnología.