Chromosome segregation control by Escherichia coli ObgE GTPase

Authors


*E-mail lovett@brandeis.edu; Tel. (+1) 781 736 2497; Fax (+1) 781 736 2405.

Summary

Escherichia coli cells depleted of the conserved GTPase, ObgE, show early chromosome-partitioning defects and accumulate replicated chromosomes in which the terminus regions are colocalized. Cells lacking ObgE continue to initiate replication, with a normal ratio of the origin to terminus. Localization of the SeqA DNA binding protein, normally seen as punctate foci, however, was disturbed. Depletion of ObgE also results in cell filamentation, with polyploid DNA content. Depletion of ObgE did not cause lethality, and cells recovered fully after expression of ObgE was restored. We propose a model in which ObgE is required to license chromosome segregation and subsequent cell cycle events.

Introduction

Studies of chromosome organization in bacterial cells show that the chromosome is an exquisitely organized and dynamic structure (reviewed recently in Thanbichler et al., 2005). Chromosome segregation in bacteria does not occur all at once but in sequential phases (Lau et al., 2003; Viollier et al., 2004; Bates and Kleckner, 2005; Nielsen et al., 2006). After replication at mid-cell, the origin region (oriC) is rapidly segregated outward. The speed at which this occurs (reviewed in Gordon and Wright, 2000) rules out passive models for bacterial chromosome segregation, which proposed that outward cellular growth could drive the movement of a fixed chromosome. As the loci of the chromosome are replicated, they are moved outward to the poles in a sequential fashion (Lau et al., 2003; Viollier et al., 2004; Bates and Kleckner, 2005; Nielsen et al., 2006). In Escherichia coli, there may be a period of sister chromosome cohesion between duplication and subsequent segregation, although its length is disputed (Sunako et al., 2001; Bates and Kleckner, 2005; Nielsen et al., 2006). It has been proposed that SeqA, a protein that binds newly replicated (hemimethylated) DNA, may mediate this cohesion (Nielsen et al., 2006). After the bulk of the chromosome has segregated, cohesion at the terminus region (ter) appears to persist (Bates and Kleckner, 2005; Nielsen et al., 2006). Once ter has separated, septation ensues. Therefore, segregation of oriC, ter and the rest of the chromosome could have unique determinants.

The mechanism by which specific regions of the chromosome are organized and segregated prior to division remains a mystery. The speed at which chromosomes segregate appears to require an active process, but the machinery remains elusive. It has been proposed that replication itself can provide the force for segregation. In the ‘extrusion-capture’ model, replication through an anchored replisome provides the force to move daughter chromosomes to the poles (Lemon and Grossman, 2001). In Bacillus subtilis and Caulobacter crescentus, the Soj/Spo0J ParA/B system may serve to capture or constrain a dynamically mobile oriC (Ireton et al., 1994; Mohl and Gober, 1997). The MigS system of E. coli may play a similar role (Yamaichi and Niki, 2004; Fekete and Chattoraj, 2005). The bacterial condensin-like proteins (Hirano, 2005), Smc of Bacillus and MukB of E. coli have chromosome segregation defects (Hiraga et al., 1991; Britton et al., 1998). Progressive condensation of the chromosome as it emerges from the replisome may provide force or organization of the chromosome to facilitate segregation. The MreB filament may also play a role in segregation. The bacterial actin homologue (van den Ent et al., 2001), MreB, forms a helix that transverses the length of the cell (Jones et al., 2001; Shih et al., 2003; Defeu Soufo and Graumann, 2004), and is a determinant of rod shape in many bacteria (Wachi and Matsuhashi, 1989; Jones et al., 2001; Figge et al., 2004). Soon after depletion of MreB or after depolymerization with a compound A22, chromosome segregation defects are apparent (Soufo and Graumann, 2003; Gitai et al., 2005; Kruse et al., 2006). In Caulobacter, an intact MreB filament is required for segregation of oriC, but not of bulk chromosomal DNA (Gitai et al., 2005). Segregation of both oriC and bulk chromosomal DNA is affected in E. coli (Kruse et al., 2006). MreB, therefore, has been proposed to provide a track on which newly replicated chromosomescould translocate or the force for segregation itself: polymerization could drive the outward movement of an attached chromosome. RNA polymerase has also been proposed to provide force for segregation (Dworkin and Losick, 2002) and offers a convenient means to link biosynthetic status to chromosome segregation. Other mechanisms proposed for chromosome segregation involve transertion of membrane proteins (Norris et al., 2004) or release of sister chromosome cohesion (Bates and Kleckner, 2005).

Even more mysterious is the regulation of cell cycle events, such as chromosome segregation, in response to cell nutritional status. Treatment of E. coli filaments with protein synthesis inhibitors causes arrest of ongoing chromosome segregation and nucleoids appear to coalesce (van Helvoort et al., 1996; Van Helvoort et al., 1998). Upon relief of the inhibition, nucleoids rapidly segregate. These findings confirm that bulk chromosome organization and segregation are active processes and possibly subject to checkpoint control, correlating with transcriptional/translational status.

The Obg/CgtA protein is an essential GTPase conserved throughout the eubacteria and implicated, in E. coli, for control of chromosome segregation (Kobayashi et al., 2001). We discovered a mutation in obgE that conferred extreme sensitivity to replication inhibitors and has characteristic phenotypes of a cell cycle checkpoint control defect (Foti et al., 2005). Despite the fact that it is a well-conserved and essential protein, mutants in Obg/CgtA exhibit a wide variety of phenotypes (reviewed in Czyz and Wegrzyn, 2005 and Brown, 2005). In B. subtilis, roles in sporulation initiation, DNA replication, stress response and ribosome function have been suggested (Trach and Hoch, 1989; Vidwans et al., 1995; Scott and Haldenwang, 1999). Other roles for Obg homologues in cellular morphogenesis have been reported. Obg from Streptomyces ceolicolor controls mycelium formation in response to intracellular pools of guanine nucleotides (Okamoto and Ochi, 1998). A mutational analysis of the C. crescentus orthologue suggests accumulation of 50S ribosomal subunits at the expense of 70S assembled ribosomes (Lin et al., 2004), a phenotype also seen in E. coli (Sato et al., 2005). However, the essential function of Caulobacter Obg (CgtA) appears to be progression through the cell cycle (Datta et al., 2004) rather than defects in ribosome assembly. The temperature-sensitive mutant arrests prior to replication but not with additional ribosome defects at its non-permissive temperature. Given the connection between ribosome function and chromosome organization in E. coli, it is still unclear whether the effect of CgtA on the cell cycle is likely to be direct or indirect.

To study the essential function of ObgE, we depleted cellular levels of ObgE using the regulatable araBAD promoter. Previously, the essential role of E. coli ObgE and Caulobacter CgtA had been examined using temperature-sensitive alleles (Kobayashi et al., 2001; Datta et al., 2004), which could have complex effects on the function of the protein. Our results are consistent with a role for ObgE in chromosome partitioning, with defects in chromosome segregation observed early after shut-off of the ara promoter even before depletion of ObgE protein. No detriment in protein synthesis was observed at times when cell cycle defects were obvious. The delay in nucleoid segregation was accompanied by cell filamentation, caused by inefficient FtsZ-ring formation. The SOS response to DNA damage was not evident, indicating that ssDNA and damaged chromosomes were not accumulating along with this cell cycle block. The effects of ObgE on cell cycle progression were fully reversible, with cells recovering growth capacity after many hours of depletion of ObgE. Our results suggest that ObgE, at least in E. coli, is required for cell cycle progression, independent of effects on translation.

Results

Depletion of ObgE delays chromosome segregation and cell division in a non-lethal manner

We created a strain to determine the null phenotype of ObgE, STL7961 (Foti et al., 2005). The chromosomal locus of ObgE is deleted and complemented by pSTL346, a plasmid expressing wild-type ObgE (pBAD33::ObgE). Expression from the parental plasmid, pBAD33, is controlled by the araBAD promoter that can be turned on or off by the addition of l-arabinose and d-glucose respectively (Guzman et al., 1995). Levels of ObgE protein were followed by Western blotting, using affinity-purified polyclonal antibody to ObgE. Nearly wild-type levels of ObgE were observed in STL7961 grown in Luria–Bertani (LB) medium, prior to depletion, as compared with wild-type strain MG1655 (Fig. 1B).

Figure 1.

Depletion of ObgE in strain and chromosome segregation.
A. Micrographs of phase-contrast images overlayed with the DAPI-stained image before and after 3 h of ObgE depletion. Scale bar = 5 μm.
B. Western blot analysis comparing levels of ObgE (top panel) and FtsZ (bottom panel) from cell lysates of MG1655 and STL7961. Each lane was loaded with 25 μg of total protein from cell lysates as measured by a Bradford assay. Wild-type (WT) at time zero and a time-course of STL7961 in LB 0.2% d-glucose medium are shown from time zero (0) to 3 h (3). Polyclonal ObgE and FtsZ antibodies were used to probe the same cell lysate samples.
C. Plating efficiency of STL7961 on LB 0.2% arabinose media following growth in LB 0.2% arbinose media (Ara to Ara) or LB 0.2% glucose media (Glue to Ara) for the indicated time.
D. Scatter plot of cell length versus nucleoid number for wild-type MG1655 cells (open circles) and ObgE-depleted cells (closed squares) following growth in LB 0.2% glucose media.

Chromosome segregation defects were observed early after ObgE depletion. Confirming the results of Kobayashi et al. (2001) with the temperature-sensitive allele of obgE, large DAPI-staining nucleoids were seen after only 1 h of ObgE depletion, when levels of ObgE were still detectable at about 15% of that prior to the downshift (Fig. 1B). In a few cells segregated nucleoids were apparent. After 3 h in LB glucose medium, when ObgE was barely detectable by Western blots (Fig. 1B), many cells displayed segregated chains of nucleoids whereas others contained segregated nucleoids in addition to large nucleoid masses (Fig. 1A and D), similar to those seen at 1 h.

Despite this problem in chromosome partitioning, DNA replication continued after ObgE depletion. DNA content in these ObgE-depleted cells was measured by PicoGreen staining, followed by flow cytometry (Fig. 2). The median DNA content for ObgE-depleted cells between 3 and 5 h was 15–20 chromosome equivalents. (A 1N DNA content reference value was obtained using a dnaA46 control strain, grown at 42°C to block replication initiation.) Therefore, DNA replication continued throughout the depletion to generate extensively polyploid cells. The DNA histograms are very broad, indicating considerable heterogeneity in chromosomal content. Despite the heterogeneity in the population, the DNA content of the cell is still correlated with cell size as suggested by the scatter plots, Fig. 2.

Figure 2.

Cells depleted of ObgE filament and accumulate DNA as measured by flow cytometry. Cell size and DNA content histograms and scatter plots of STL7961 after 0 (median size = 11, median DNA = 30) and 3 (median size = 41, median DNA = 150) h of depletion. The median cell size and DNA content of STL7961 was compared with a dnaA46TS at 42°C (median size = 14, median DNA = 10) for a 1N chromosome content control. Measured in A.U.

Cell growth continued after ObgE depletion, although cell division was partially blocked. We measured the size of ObgE-depleted cells by microscopic examination and flow cytometry. ObgE-depleted populations accumulated much larger cells (median length 7.6 μm versus 2.5 μm for mock-depleted cells, Fig. S1), with high DNA content (Figs 2 and 3). These large cells (Figs 1 and 3) were filamentous and contained multiple nucleoid masses. However, some cells did seem to be able to divide: even after 3 h of ObgE depletion, a non-filamentous subpopulation was evident as a sharp peak in the flow cytometry forward scatter histogram (Fig. 2).

Figure 3.

Localization of oriC or ter visualized by GFP–ParBΔ30 binding to a nearby parS site: oriC in wild-type mock-depleted (A) or ObgE-depleted (B) cells after 3 h; ter in wild-type mock-depleted (C) or ObgE-depleted cells. Scale bars = 5 μm. ObgE-depleted cells with GFP–ParB oriC::parS (E) or ter:parS (F) costained with DAPI (pink). Enlargements of individual cells are shown without the DAPI staining (bottom).

Even after long periods of ObgE depletion, cellular survival was unaffected when expression of ObgE was restored. Depletion of ObgE was not lethal: flow cytometric analysis of LIVE/DEAD-stained cell populations revealed that only 1.4% of the population was dead after ObgE has been depleted for 3 h. Furthermore, cells depleted for ObgE in glucose medium for as long as 6 h formed no fewer colonies when returned to growth on LB–arabinose plates as the culture prior to depletion (Fig. 1C).

Localization of the origin and terminus chromosomal regions in ObgE-depleted cells

To gain more information into the nature of the chromosome segregation defect, we examined the location of the origin (oriC) and terminus (ter) regions of the chromosome in ObgE-depleted cells, using a ParBΔ30::GFP system developed by Austin and collaborators (2002). A parS site, to which ParB binds, was introduced near oriC or ter in our ObgE-depletion strain with an additional plasmid expressing ParBΔ30::GFP. This system allows us to determine the number and organization of oriC and ter in ObgE-depleted cells.

In normally growing E. coli cells, the origin region of wild-type chromosomes duplicates at mid-cell and is quickly relocalized to about the 1/4 and 3/4 positions, where secondary initiation occurs (Li et al., 2002; Lau et al., 2003; Bates and Kleckner, 2005). In ObgE-depleted cells (Figs 3 and 4) we observed multiple oriC foci within cells, indicating that initiation had occurred multiple times. Compared with mock-depleted wild-type cells, ObgE-depleted cells had a significantly increased number of oriC foci (Fig. 4), with a median value of 5 oriC foci versus 2 for wild-type. Odd numbers of oriC foci were observed, consistent with our previous observation of replication initiation asynchrony in obgE mutants (Foti et al., 2005).

Figure 4.

Histograms of the number of OriC (A) and Ter (B) GFP-foci per cell determined using the ParB/parS system, and location of these OriC (C) and Ter (D) GFP-foci as fraction of cell length. Black bars represent 100 wild-type cells subjected to mock deplete conditions (3 h in 0.2% glucose); grey bars represent 100 ObgE-depleted cells.

Localization of terminus by the ParSΔ30::GFP system appeared quite different. Usually only one ter focus was apparent per nucleoid (Fig. 3F) and the number of ter foci in ObgE-depleted cells was no different from wild-type mock-depleted cells (median value of 1, for both). Brightly fluorescent ter foci were located generally in the centre of unpartitioned nucleoids, as stained by DAPI (Fig. 3F). Occasional fainter satellite foci were seen in close proximity to a brighter ter focus (cells 1 and 3 in Fig. 3F). The intensity of the ter focus was roughly proportional to the intensity of the DAPI-stained nucleoid. Cells with larger DNA masses with one ter focus tended to be brighter, for example cells 1, 3, and 4, than cells with smaller DNA masses and multiple ter foci (e.g. cell 2).

Measurement of the location of the ter foci in relation to cell length showed that, like wild-type cells, ObgE-depleted cells had preferential localization of ter to 1/4 or 1/2 positions (Fig. 4D). However, whereas 20% of wild-type mock-depleted cells possessed ter in a polar location, this class was absent in the ObgE-depleted cells. Polarly located oriC was also absent in ObgE-depleted cells (Fig. 4C).

Our observation of reduced number of ter foci relative to oriC foci in ObgE-depleted cells may indicate that replication is incomplete. Alternatively, it may represent ter that have been duplicated but remain colocalized. This could be caused by some kind of aggregation or cohesion that persists between multiple ter regions. Supporting the latter hypothesis, marker frequency analysis by Southern blot (Fig. S2) showed that the oriC/ter ratio is not elevated after ObgE depletion, 1.7 (3 h post depletion) versus 2.2 (prior to depletion), nor is it different relative to wild-type mock-depleted cells (oriC/ter = 1.8). This is inconsistent with the notion that initiation continues to occur but progress of the replication fork from oriC to ter is blocked.

Moreover, if replication were incomplete in ObgE-depleted cells, we would expect induction of the SOS response as a consequence of accumulation of replication intermediates. Difficulties in replication or processing of broken chromosomes elicit the SOS response, involving the co-ordinated transcriptional induction of a set of genes that promote cellular survival (Sutton et al., 2000). Using luciferase gene fusions to two promoters that report induction of the SOS response (Van Dyk et al., 2001), we measured their expression during the period of ObgE depletion. We observed no significant fold induction (3 h growth in LB 0.2% glucose versus 3 h in LB) of either the dinB (1.98 ± 0.84) or the recA (2.15 ± 1.12) promoters normalized to the promoterless control. Because the SOS induction signal is the binding of RecA to ssDNA gaps or ends, this argues against the accumulation of stalled replication forks or other damaged DNA in ObgE-depleted cells.

SeqA foci are perturbed in ObgE-depleted cells

The SeqA protein of E. coli binds to newly replicated DNA via interactions with hemimethylated GATC sites that are produced after DNA replication and can be seen as foci colocalized with the replisome (Onogi et al., 1999; Brendler et al., 2000). Aggregation of SeqA protein has been proposed to organize the chromosome into a filamentous structure after passage of the replication fork (Han et al., 2003; Han et al., 2004). Overproduction of SeqA causes a delay in chromosome segregation and accumulation of anucleate cells (Bach et al., 2003). Using a GFP–SeqA fusion, we examined the nature of SeqA foci with and without ObgE depletion for 3 h. SeqA normally exists as punctate, single or paired foci, colocalized to the replisome, at 1/4 and 1/2 celllular positions. In ObgE-depleted cells, SeqA failed to form foci and formed a more diffuse pattern (Fig. 5), sometimes throughout the nucleoid. This may indicate a more persistent association of SeqA with the nucleoid in the absence of ObgE.

Figure 5.

Depletion of ObgE perturbs SeqA localization. GFP-SeqA foci in wild-type MG1655 mock-depleted cells (left panel) versus cells depleted for ObgE for 3 h (right panel).

Inefficient FtsZ-ring formation in ObgE-depleted cells

The tubulin-like protein, FtsZ, forms a ring at mid-cell which marks the future site of cell division (reviewed recently in Goehring and Beckwith, 2005). This Z ring recruits a cascade of other factors that stabilize the ring and constrict the cytoplasm during cytokinesis. In the presence of MinE, the spatially oscillating proteins MinC and MinD regulate the location of ring formation in cooperation with the nucleoid occlusion protein SlmA (Yu and Margolin, 1999; Shih et al., 2003; Bernhardt and de Boer, 2005). The DNA damage-inducible SulA protein inhibits FtsZ polymerization and thereby delays cell division in the presence of DNA damage (Bi and Lutkenhaus, 1993; Dai et al., 1994). Because cells lacking ObgE became filamentous, we investigated whether they were defective in FtsZ-ring formation by immunofluorescence of ObgE-depleted cells.

After 3 h in LB 0.2% glucose medium, the majority of cells did not display bright FtsZ staining when compared with wild-type (Fig. 6 top panel and data not shown). FtsZ rings that did appear were often malformed, as dots or aberrant bands. We noted that localization of FtsZ in ObgE-depleted cells did not overlap with the DAPI-stained nucleoids, suggesting that nucleoid occlusion, the restriction of Z-ring formation over the nucleoid, is still functional in ObgE-depleted cells. Inefficient Z-ring formation could be rescued by return to growth in arabinose medium after depletion of ObgE (Fig. 6 bottom panel). Sharp Z-ring bands formed between well-condensed nucleoids after transfer to arabinose medium for 2 h, even for very long cells containing as many of 45 distinct nucleoids, as shown in Fig. 6. Rescue of FtsZ-ring formation by arabinose suggests that its failure to form in ObgE-depleted cells is not due to insufficient concentrations of FtsZ. Indeed, expression of FtsZ, as detected in Western blots of ObgE-depleted cells, appeared normal (Fig. 1), confirming that the filamentous nature of ObgE-depleted cells is not due to a downregulation of FtsZ gene expression.

Figure 6.

Immunofluorescent microscopy using polyclonal FtsZ antibody. FtsZ (blue) and DNA (pink) in STL7961 following 3 h of depletion (top panel). STL7961 cells grown in LB 0.2% arabinose for 2 h following depletion show the reversibility of inefficient ring formation (bottom panel). Scale bar = 5 μm.

The block to FtsZ-ring formation is unlikely to be due to the induction of the SulA inhibitor or other SOS factors. As discussed above, SOS induction, as measured in luceriferase gene fusions (fold induction of SulA promoter 1.76 ± 0.69), did not occur concomitant with ObgE depletion. Both SulA-dependent and SulA-independent filamentation have been observed to depend on RecA gene function (Hill et al., 1997). We found that cell filamentation in ObgE-depleted cells was not dependent on functional RecA. Cell size, as determined by flow cytometric analysis, of ObgE-depleted recA cells [median = 115 arbitrary fluorescence units (A.U.)] was equally large as that of ObgE-depleted cells in a recA+ background (median = 126 A.U.) at 3 h. It remains possible that RecA-independent filamentation is produced as response to the chromosome organization or segregation difficulties; however, the nature of such a mechanism is not known.

ObgE depletion does not reduce translational capacity

It has been previously reported that ObgE interacts with ribosomal proteins, and depletion of ObgE effects pre-16S-rRNA processing and ribosomal protein levels (Kobayashi et al., 2001). Both E. coli and C. crescentus have been reported to copurify with the ribosome, although this interaction has been reported to be weak (for Caulobacter) (Lin et al., 2004) or variable (for E. coli) (Kobayashi et al., 2001; Sato et al., 2005). It has been suggested that ObgE is required for ribosome biogenesis and that observed cell cycle effects may be an indirect consequence of loss of translation capacity (Sato et al., 2005).

We investigated whether translational capacity was diminished during ObgE depletion. To determine the translation rate of ObgE-depleted cells, total protein was labelled with 35S-methionine, precipitated with trichloroacetic acid (TCA), and translation rates were determined from total incorporated radioactivity. The rate of methionine incorporation (V = picomoles of methionine incorporated into TCA precipitate/microgram of cell dry weight per minute) for STL7961 grown in LB glucose medium to deplete ObgE for 3 h was not lower than that before the depletion (V = 1.24 × 10−3 ± 3.19 × 10−4, depleted, versus 7.13 × 10−4 ± 4.67 × 10−4, non-depleted). However, both values for STL7961 were somewhat lower than that for wild-type strains, MG1655 (7.22 × 10−3 ± 1.32 × 10−3). The profile of SDS-PAGE-separated proteins labelled before or after ObgE depletion did not appear different, so specificity of proteins that were translated after ObgE depletion was not grossly affected (data not shown).

Discussion

ObgE is necessary for efficient chromosome partitioning

Previous studies of Obg orthologues have examined different temperature-sensitive alleles and point mutations to deplete functional Obg in Gram-negative bacteria (Kobayashi et al., 2001; Datta et al., 2004; Sato et al., 2005; Sikora et al., 2006). The results of these studies have suggested various and diverse cellular roles, including ribosome biogenesis, stress response and chromosome segregation. Which of these roles is essential for bacterial growth has remained elusive. The effects of temperature-sensitive mutations can be difficult to interpret, as they may reflect protein misfolding or protein stability/turnover causing loss of function, gain of function, or the loss of one of many functions. In this study, we wished to elucidate the essential function of ObgE by simply depleting the protein, using a regulatable promoter.

An early phenotypic consequence of cells depleted for ObgE levels is a chromosome-partitioning defect. This occurred even after a modest reduction in ObgE levels, 15% of that prior to depletion. ObgE is an abundant protein, reported to be 34 000 molecules per exponentially growing cell, and our experiments suggest that this high level is required to sustain chromosome segregation. ObgE-depleted cells became elongated with large DNA masses or irregular, segregated nucleoids, indicating that some chromosome segregation may have been occurring but was inefficient. Replication initiation and cell growth continued to occur, and cells became filamentous and highly polyploid. Surprisingly, this block was reversible and cells did not sustain loss of viability during ObgE depletion as determined by cell plating measurements and vital staining. No induction of the SOS response was observed, indicating a lack of chromosomal damage. This, and the lack of any detectable inviability associated with ObgE depletion, argues that chromosomes are intact but some signal for cell cycle progression is lacking.

The defects in chromosome partitioning were accompanied by changes in chromosome organization at oriC and ter, visualized by GFP–ParB fusions. The mechanism by which the spatial organization of chromosomes appears disrupted is unclear. We observed multiple dispersed oriC foci in unsegregated nucleoids but only one ter focus. The ter foci appeared brighter as the area of DAPI staining increased, suggesting cohesion or aggregation of chromosomes near ter. In some cells containing large DNA masses we observed satellite ter foci, suggesting that these cells may be in the process of separating chromosomes at ter. Dominant mutants in the bacterial actin, MreB, display a strikingly similar phenotype: multiple oriC foci but only one ter focus in an unsegregated nucleoid (Kruse et al., 2003). Because MreB has also been implicated in bacterial chromosome segregation (Soufo and Graumann, 2003; Gitai et al., 2005; Kruse et al., 2006), this raises the possibility that both ObgE and MreB may operate in or regulate a common chromosome segregation pathway or apparatus.

The position of ter in cells is not random: it preferentially resides at mid-cell or at the 1/4 positions. In ObgE-depleted cells, this pattern was retained. Therefore, segregation defects in ObgE-depleted cells do not appear to reflect gross mislocalization of ter. However, we did observe the loss of a subpopulation of cells with polarly localized ter. Because ter is seen at the poles in newly divided cells (Bates and Kleckner, 2005; Nielsen et al., 2006), this class may be lacking in ObgE-depleted cells because of a block to cell division.

SeqA mediates chromosome cohesion near the terminus that is trapped in ObgE-depleted cells?

Two recent studies have shown that the terminus region is slow to segregate after duplication (Bates and Kleckner, 2005; Nielsen et al., 2006). Our results and those of Gerdes and coworkers (2003) suggest that transient cohesion or colocalization of the terminus regions is trapped in cells lacking ObgE or expressing mutant MreB. The ability of DNA-bound SeqA to aggregate in vitro (Han et al., 2003; 2004) suggests that SeqA could be the factor that mediates cohesion: SeqA bound to GATC sites in the two newly replicated sister chromosomes could coalesce the chromosomes at ter. The observation that MreB and ObgE are required to relieve this cohesion suggests that this could be a regulated, orchestrated step in the bacterial cell cycle. In support of this idea, overexpression of SeqA does perturb chromosome segregation and delay cell division (Bach et al., 2003). Likewise, overexpression of ObgE perturbs the cellular localization of SeqA (Foti et al., 2005), and we observed more extensive and diffuse GFP–SeqA through the nucleoid in the absence of ObgE.

The mechanism of bacterial chromosome segregation is still mysterious, although the demonstration of segregation defects associated with mutants in mreB suggests that the bacterial actin may organize the chromosome to facilitate this process (Soufo and Graumann, 2003; Gitai et al., 2005; Kruse et al., 2006). Outward polymerization of MreB could also provide force for movement of chromosomal regions. It has been proposed that DNA replication through a fixed replisome at mid-cell could provide force for chromosomal segregation, the ‘extrusion-capture’ model (Lemon and Grossman, 2001). Our results and those of others (Kobayashi et al., 2001; Kruse et al., 2003) show that DNA replication and chromosome segregation can be uncoupled; that is, both obgE and mreB mutants accumulate unsegregated replicated chromosomes. At least for ObgE, the segregation block appears to be reversible (Kobayashi et al., 2001). Force generated by a fixed RNA polymerase remains a viable mechanism for chromosome partitioning (Dworkin and Losick, 2002). The interaction of RNA polymerase and MreB supports a model whereby RNA polymerase tethered to the MreB filament pumps DNA outward to the cell poles (Kruse et al., 2006). Mutants in RNA polymerase also exhibit chromosome segregation defects, with cohesion persistent at ter (Kruse et al., 2006). ObgE also has been shown to interact with RNA polymerase (Butland et al., 2005), but the consequences of this interaction are not known. A reasonable hypothesis is that ObgE is required to license chromosome segregation and, in its absence, chromosomes are tethered at ter and not engaged with the segregation apparatus, whatever its nature.

Mysterious delay of cell division

Concomitant with chromosome-partitioning defects, ObgE-depleted cells appeared block in cell division and continue to grow as highly polyploid filaments. This block may not be absolute: a subpopulation of non-filamentous cells was observed in the cell size histogram by flow cytometry. The mechanism by which division is blocked remains unclear. Immunofluorescent experiments suggest that FtsZ rings fail to form between segregated nucleoids in depleted cells. Re-expression of ObgE after a period of depletion revealed sharp FtsZ rings between segregated nucleoids, confirming that FtsZ was of sufficient cellular concentrations to initiate ring formation. Western blotting of FtsZ after ObgE depletion also indicated that levels of FtsZ were normal. The SOS response was not induced in ObgE-depleted cells, as assayed by three promoter reporter fusions, nor did recA mutations block cell filamentation, ruling out a SulA, RecA-dependent inhibition of FtsZ-ring formation. These results suggest another unidentified mechanism for cell division arrest in ObgE-depleted cells. It remains possible that the division block is an indirect consequence of other chromosome organization or segregation problems in ObgE-depleted cells.

ObgE and its role in translation

Previous experiments using a number of methods had suggested that Obg function was essential for growth because of a role in ribosome biogenesis or function (Brown, 2005; Sato et al., 2005). Our experiments suggest that ObgE is essential for cell division, but not for cell growth or viability. Cells continue to elongate and replicate DNA, but fail to divide. Translational capacity did not decline to any detectable degree during ObgE depletion; therefore, ObgE cannot be required for bulk ongoing protein synthesis. This conclusion is similar to recent studies of the Obg orthologue, CgtAC, where it was deduced that the essential role of the protein was for cell cycle progression and not for ribosome assembly (Datta et al., 2004). It remains possible that ObgE is required to assemble new ribosomal subunits, without affecting the function of previously assembled subunits, or affects in a more subtle way ribosome stability or turnover. Nevertheless, cells depleted for ObgE rapidly exhibit cell cycle defects without loss of translational capacity, suggesting that the cell cycle defects are the direct phenotypic consequences of loss of ObgE. In addition, the phenotypes of obgE mutants with respect to replication inhibitors and fork instability (Foti et al., 2005) are difficult to reconcile with a singular role for Obg in ribosome biogenesis. We favour the unifying hypothesis that Obg may link cell cycle events with translational capacity, explaining both the genetic properties of obgE mutants and its association with the ribosome. Further experimental work will hopefully clarify the connection between Obg, ribosome assembly, stability and cell cycle.

Obg and the stringent response

The stringent response to amino-acid starvation results in reduced transcription of stable RNAs, concomitant with increased expression of certain biosynthetic genes, mediated by the signalling molecule ppGpp (Magnusson et al., 2005). Recent results in Vibrio cholerae (Raskin et al., 2007) suggest that the Obg orthologue, CgtA, is required to downregulate the stringent response, perhaps by modulation of the ppGpp hydrolase activity of SpoT. Cells depleted for CgtA show gene expression patterns characteristic of constitutive induction of the stringent response. Furthermore, in the absence of the ppGpp synthase, RelA, V. cholerae cgtA mutants are viable. Our results with the E. coli ObgE are inconsistent with this finding: relA does not suppress the inviability of obgE (D.J. Ferullo and S.T. Lovett, unpublished results), nor is the ObgE-depletion phenotype equivalent to that produced by a sustained stringent response. Induction of the stringent response blocks DNA replication in E. coli (Levine et al., 1991; Schreiber et al., 1995), whereas ObgE deletion strains continue to replicate, producing polyploid filamentous cells. Although this does not rule out a role for E. coli Obg in regulation of the stringent response, it does suggest an additional and essential function for the E. coli protein.

Experimental procedures

Bacterial strains, plasmid constructions and growth conditions

Strains (Table 1) were grown at the indicated temperature as previously described in LB (Miller, 1992). MG1655 isogenic strains were constructed by P1 virA transduction (Table 1) (Lovett and Sutera, 1995). For transductions and P1 phage lysates, cultures were grown in LCG or LCA as described previously (Foti et al., 2005). Antibiotics were used at the following concentrations: ampicillin 100 μg ml−1, kanamycin 60 μg ml−1, and tetracycline and chloramphenicol 15 μg ml−1.

Table 1. Escherichia coli K-12 strains and plasmids.
StrainRelevant genotypeSource or derivation
MG1655K-12 wild-type F-rph-1Blattner et al. (1997)
CC4711lacΔU169 gal490λcI857Δ(cro-bioA) pstA::P1parSLi et al. (2002)
CC4713lacΔU169 gal490λcI857Δ(cro-bioA) gadB::P1parSLi et al. (2002)
CAG12072sfsB203::Tn10Singer et al. (1989)
GN5002obgEΔ::kan[pGK14]Kobayashi et al. (2001)
GN5003ObgEΔ::kan[pGK15]Kobayashi et al. (2001)
STL7222seqAΔSutera and Lovett (2006)
STL7961obgEΔ::kan[pSTL346]Foti et al. (2005)
STL8297dnaA46 tna::Tn10Foti et al. (2005)
STL8690obgEΔ::kan[pSTL346]Foti et al. (2005)
STL9178sfsB::Tn10 obgEΔ::kan[pSTL346]Tcr P1 transductant CAG12072 × STL7961
STL9357pstA::P1parS sfsB::Tn10 obgEΔ::kan[pSTL346]Tcr P1 transductant of STL9178 × CC4711
STL9220gadB::P1parS sfsB::Tn10 obgEΔ::kan[pSTL346]Tcr P1 transductant of STL9178 × CC4713
STL10424gadB::P1parS sfsB::Tn10 seqAΔCmr P1 transductant of STL7222 × CC4713
STL10488gadB::P1parS sfsB::Tn10 seqAΔobgEΔ::kan[pSTL346]Tcr P1 transductant of STL9178 × STL10424
PlasmidRelevant genotypeSource or derivation
pSTL346cat araC obgEWTFoti et al. (2005)
pET11a::obgEobgEWTTan et al. (2002)
pALA2705GFP-Δ30::ParBLi et al. (2002)
pDEW201luxCDABEVan Dyk et al. (2001)
pDEW238PrecA:::luxCDABEVan Dyk et al. (2001)
pDEW236PdinB::luxCDABEVan Dyk et al. (2001)
pDEW201::SulApPsulA::luxCDABEThis work

The plasmid pDONR201::sulAP was constructed by polymerase chain reaction (PCR) amplifying the sulA promoter with the following primers and Pfu polymerase (Stratagene) (primers: sulAP1: GGG GAC AAG TTT GTA CAA AAA AGC AGG CTT CGG GGC AAG ATT AAT TTA TG and sulAP2: GGG GAC AAG TTT GTA CAA AAA AGC AGG CTT CGG GGC AAG ATT AAT TTA TG). After purification (QiaQuick PCR Purification Kit, Qiagen), the fragment was inserted into pDONR201 by Gateway site-specific recombination (Invitrogen). To construct pDEW201::sulAP, pDEW201 (Van Dyk et al., 2001) was digested with SmaI (New England Biolabs) and ligated to Gateway Conversion Cassette B (Invitrogen). The resulting plasmid was reacted with pDONR201::sulAP following the manufacturer's instructions to create pDEW201::sulAP.

Depletion of ObgE from cells

For promoter depletion, cells were grown in LB media containing the appropriate antibiotics overnight at 37°C. To deplete cells of ObgE, the culture was subsequently diluted 1:100 in LB media containing 0.2% d-glucose and incubated at the indicated temperature with aeration. For recovery, the culture was resuspended in fresh LB medium containing 0.2% l-arabinose. Aliquots were collected and fixed as indicated. For methanol fixation, 80% ice-cold methanol was added to cells concentrated in 50 μl of LB media. For ethanol fixation, 1 ml of concentrated cells was added to 9 ml of 95% ice-cold ethanol. Fixed cultures were stored at 4°C.

Luciferase assays

STL7961 cells containing plasmids expressing LuxCDAB under the control of the RecA, DinB or SulA promoters were depleted of ObgE as described above. Luciferase expression was measured by scintillation counting as described previously (Goldfless et al., 2006). The number of cells per ml in each culture was determined using the microscope and a Petroff Hauser counting chamber. Arbitrary luminescence units per cell (A.U.) was then calculated from scintillation cpm/cells in each culture.

Flow cytometry analysis for DNA content, cell size and vital staining

For DNA content and cell size, ethanol fixed cells suspended in 0.5 ml of phosphate-buffered saline (PBS) (10 mM potassium phosphate buffer pH 7.4, 137 mM sodium chloride, 2.7 mM potassium chloride). PicoGreen (Invitrogen) was diluted 1:100 in 25% DMSO. The samples were incubated with 100 μl of diluted dye for 3 h at room temperature and then diluted with 1 ml of PBS. For vital staining, 1 ml of culture was resuspended in 1 ml of PBS without fixing. The cells were stained with LIVE/DEAD®BacLightTM Bacterial Viability Kit according to the manufacturer's instructions (Invitrogen). Exponentially growing MG1655 cells were stained with and without fixing for dead and live controls. All cultures were analysed using a Becton Dickinson FACSCalibur flow cytometer and CellQuest software.

Microscopy

For nucleoid and FtsZ staining, 20 μl of a methanol-fixed culture was dried on poly-l-lysine hydrobromide-coated (Sigma Aldrich, 1 mg ml−1) slide. Dried cells were incubated with 100 μl of lysozyme solution (2 mg ml−1 in 25 mM Tris-HCl pH 8.0, 50 mM glucose, 10 mM EDTA) for 10 min. After lysis, cells were washed three times with TBST (25 mM Tris-HCl pH 7.4, 3 mM KCl, 140 mM NaCl, 0.25% Tween-20). The cells were dried with methanol and acetone washes for 1 min each and then air dried. Non-specific antibody interactions were blocked by TBST with 2% BSA for 15 min. The slides were incubated for 1 h with a 1:3000 dilution of polyclonal anti-FtsZ (Addinall et al., 1996). Cells were washed three times with TBST, and non-specific antibody interactions were then blocked by TBST 2% BSA. Slides were incubated with 5 μg ml−1 Alexa Fluor® 488 goat anti-rabbit IgG (Invitrogen) for 1 h and then washed three times with TBST. Nucleoids were stained with 10 μl of DAPI (10 μg ml−1 4′,6′ diamido-2-phenylindole) for 10 min. The slides were washed three times with TBS and mounted with 5 μl VectaShield (Vector Laboratories) mounting medium. FtsZ and DAPI slides were analysed with an Olympus BX51 microscope equipped with a RGB liquid crystal colour filter and a Qimaging Retiga Exi camera. Images were acquired and analysed with OpenLab DarkroomTM (Improvision) and manipulated with Adobe Photoshop Elements.

For nucleoid staining only, cells were adhered to poly-l-lysine slides as above. Cells were washed three times with PBS, stained with DAPI for 10 min, and were washed five times with TBS, mounted with 5 μl VectaShield mounting medium and analysed as above. For visualization of GFP–ParB bound near oriC or ter (strains described in Table 1), we used published procedures (Li et al., 2002), with the exception that levels of GFP–ParB were reduced by omitting induction of expression with IPTG.

Marker frequency analysis using Southern hybridization

Cells were grown exponentially for 3 h in LB medium with and without 0.2% glucose. Samples were taken and chromosomal DNA was extracted using the MasterPure™ DNA Purification Kit (Epicentre®). Restriction digestion and preparation of the probe were performed as described (Nyborg et al., 2000). Briefly, the chromosomal DNA was triple digested with EcoRI, HindIII and EcoRV, and the fragments were separated on a 1.0% agarose gel. The DNA was vacuum-transferred to a nylon membrane (Amersham). The membrane was prehybridized with BSA for more than 1 h at 65°C and hybridized overnight at 65°C with 32P α-dATP. The probe consisted of three DNA fragments that anneal to the chromosomal regions gidA (84.3 min), relB (34.8 min) and cya (85.9 min). The DNA fragments were labelled using Random Primer Labeling (Molecular Cloning). After hybridization, the membrane was washed with 0.5 M Na2HPO4 (pH 7.3)/20% SDS/0.5 M EDTA 3× for 10 min at 25°C, followed by 2× for 5 min at 65°C. The membrane was exposed on a Phosphoimaging screen (Molecular Dynamics) and scanned on a Bio-Rad Molecular Imager® FX (Bio-Rad). Analysis of bands was carried out using Quantity One® imaging software (Bio-Rad). Normalization of the bands was accomplished using genomic DNA from dnaA46 that was grown at the non-permissive temperature for 2 h.

Western blot analysis

ObgE-depleted cells were resuspended in 100 μl of Tris-sucrose (50 mM Tris-HCl pH 7.5, 10% sucrose) and quick frozen on a dry-ice ethanol and stored at −80°C. Crude lysates were then prepared by adding 16 μl of 25 mM EDTA, 1 μl of 0.1 M DTT, and 4 μl of 10 mg ml−1 lysozyme in Tris-sucrose to thawed cells. Cells were incubated on ice for 5 min, followed by the addition of 4 μl of 5 M NaCl, and incubation on ice for 25 min. Samples were then heat shocked four times at 37°C and then centrifuged at 4°C for 15 min. Total protein in the supernatant was determined by a Bradford assay (Bradford, 1976) (using the reagent purchased from Bio-Rad), and 25 μg of total protein was separated on a 15% SDS-PAGE acrylamide gel prior to transfer to a PVDF membrane. ObgE or FtsZ was visualized with SuperSignal West Pico Chemiluminescent Substrate (Pierce) using polyclonal serum of ObgE or FtsZ with horseradish peroxidase-linked anti-rabbit Ig, from donkey (GE Healthcare/Amersham Biosciences) as a secondary antibody.

Polyclonal ObgE antibody production and affinity purification

Escherichia coli BL21 cells harbouring pET11a::ObgE was purified as previously described with the following modifications (Tan et al., 2002). Cells were grown to OD600 = 0.8 and induced with 1 mM IPTG for 2 h. The culture was 100-fold concentrated in 10% sucrose 50 mM Tris-HCl pH 7.5 and stored at −80°C. Cells were lysed by heat shock at 37°C and incubation with 0.2 mg ml−1 lysozyme in 100 mM NaCl, 1 mM DTT, 10% sucrose, and 50 mM Tris-HCl pH 7.5. Cleared cell lysate was loaded onto a pre-equilibrated 5 ml Q-column (Bio-Rad), and eluted in 20 mM Tris-HCl pH 7.5, 1 mM DTT, 0.1 mM EDTA, 10% ethylene glycol and 100–1000 mM NaCl. ObgE-containing fractions (480–560 mM NaCl) were loaded onto a 5 ml HA column (Bio-Rad), washed and eluted (20–400 mM potassium phosphate buffer pH 6.8 and 10% glycerol). ObgE-containing fractions (130–170 mM phosphate) were pooled and stored in 90 mM potassium phosphate pH 6.8 and 50% glycerol at −20°C. Protein was concentrated to 600 μg ml−1 as determined by the Bradford method (1976). Polyclonal rabbit anti-ObgE serum was raised from these concentrated fractions by ProSci, Poway, CA.

Crude lysates of MG1655 cells were separated on a 15% SDS-PAGE gel and transferred to a PVDF membrane as previously described for the Western blot. The membrane was cut in half and Western blotted as above with anti-ObgE. The two pieces of the membrane were aligned, and a strip corresponding to ObgE (approximately 50 kDa) was excised and incubated for 2 h in antisera. Bound protein was eluted for 10 min with 100 mM glycine pH 2.5 and the buffer neutralized with 1/10 volume 1 M Tris pH 8.0.

Labelling total protein with 35S-methionine

Cells were depleted of ObgE or grown as described above in LB media and subsequently washed twice in M9 minimal media with rhamnose or glucose. The concentration of the culture was adjusted to OD600 = 0.1. 35S-methionine (0.01 mCi) was added to 100 μl of each sample and incubated at 37°C for 30 min. Total protein was precipitated by adding 100 μl of 10% TCA. A total of 10 μl aliquot was resuspended in 1.5 M Tris pH 8.8 and separated on a 15% SDS-PAGE gel. The remaining portion of the sample was filtered through GF/C filters, washed four times with 5 ml of 5% TCA, once with 95% ethanol, and air dried. Total radiation was measured by scintillation counting, and the rate of incorporation (V) was determined with V = picomoles methionine incorporated/microgram of dry cell weight per minute (Svitil et al., 1993).

Acknowledgements

We thank S. Austin, J. Bardwell, N. Kleckner and T. VanDyk for strains, J. Lutkenhaus for FtsZ antibody, and R. Ren and the Flow Cytometry Core Facility for flow cytometry. This work was supported by NIH RO1 G51753 to S.T.L., T32 G07122 to J.J.F., and T32 G07596 to N.S.P.

Ancillary