Multiple links connect central carbon metabolism to DNA replication initiation and elongation in Bacillus subtilis.

DNA replication is coupled to growth by an unknown mechanism. Here, we investigated this coupling by analyzing growth and replication in 15 mutants of central carbon metabolism (CCM) cultivated in three rich media. In about one-fourth of the condition tested, defects in replication resulting from changes in initiation or elongation were detected. This uncovered 11 CCM genes important for replication and showed that some of these genes have an effect in one, two or three media. Additional results presented here and elsewhere (Jannière, L., Canceill, D., Suski, C., et al. (2007), PLoS One, 2, e447.) showed that, in the LB medium, the CCM genes important for DNA elongation (gapA and ackA) are genetically linked to the lagging strand polymerase DnaE while those important for initiation (pgk and pykA) are genetically linked to the replication enzymes DnaC (helicase), DnaG (primase) and DnaE. Our work thus shows that the coupling between growth and replication involves multiple, medium-dependent links between CCM and replication. They also suggest that changes in CCM may affect initiation by altering the functional recruitment of DnaC, DnaG and DnaE at the chromosomal origin, and may affect elongation by altering the activity of DnaE at the replication fork. The underlying mechanism is discussed.

1. Introduction

DNA replication must be coordinated with the cell cycle and cell division to ensure that chromosomes are faithfully replicated once and only once per cell cycle. This is carried out by overlapping mechanisms that control the activity of replication origins and replication enzymes for initiating DNA strand opening and recruitment of the replication machinery at origins every cell cycle.

It has been known for several decades that replication is also subject to a metabolic control. In steady-state bacteria, this control couples the rate of replication to the rate of growth afforded by nutrients by modulating the initiation frequency and/or the speed of replication forks. The net result is a precise and reproducible timing of DNA synthesis in the cell cycle across a wide range of nutritional conditions. Intriguingly, this timing primarily depends on the energy and precursors extracted from nutrients rather than on the actual nature of the carbon and nitrogen sources, as Escherichia coli cells growing at the same rate in different media exhibit similar temporal replication patterns. The temporal control of replication in the cell cycle defines two major periods: the C period, which corresponds to the time taken to replicate the chromosome, and the D period, which corresponds to the time between termination of DNA replication and cell septation. At the growth rates analysed in this study (1–3 doublings/h (db/h)), the C and D periods are rather constant in E. coli, while they substantially vary in Bacillus subtilis (see below). In chemostat cultures of Saccharomyces cerevisiae and possibly higher eukaryotes, the metabolic control confines DNA synthesis to the reduction phase of a redox metabolic cycle that is repeated several times per cell cycle.

Despite a long history of investigation, the exact nature of the determinants involved in the metabolic control of replication remains elusive. Equally elusive is the mechanism at play and the way it acts in concert with classical control functions of replication initiation. The long-standing hypothesis is that the metabolic control of replication depends on the concentration of the active form of the replication initiator (DnaA–ATP) or on restricting DNA polymerases activity by limiting precursor concentrations. However, these concepts have been recently challenged. Moreover, several groups argue that this control is a multifactorial process, which varies with nutrient richness and may involve sensing the cell’s metabolism and communicating it to the replication machinery. One example of such signalling involves the guanosine tetra- and penta-phosphate [(p)ppGpp]. This nucleotide analogue signals the metabolic status of bacteria and accumulates under nutritional stresses to inhibit the initiation or elongation phase of replication and to impair the activity of the DnaG primase, an enzyme that synthesizes the short RNA primers used by DNA polymerases to replicate genomes., However, although the replication inhibitory activity of (p)ppGpp at high concentration is well established, its role in DNA synthesis at low concentration (that is in the absence of nutritional stress) is still in debate.,

Central carbon metabolism (CCM) extracts the precursors and energy needed for macromolecular synthesis and biomass production from nutrients. This breakdown process involves about 30 key reactions that are highly conserved across the phyla. CCM is fed at various positions by different metabolites and the metabolite entry point determines the polarity of the carbon flux travelling through it, either glycolytic or gluconeogenic. The CCM catabolic reactions are grouped in pathways of which glycolysis, gluconeogenesis, the pentose phosphate pathway and the tricarboxylic acid (TCA) cycle form the main routes for breaking down nutrients (see the schematic representation in Supplementary Fig. S1). By directly sensing the supply and the demand in biosynthetic reactions, CCM is in a strategic position for producing signals that cells may use for adapting the main cellular activities to nutrient richness as recently demonstrated for cell division in E. coli and B. subtilis.

A role for CCM in replication was first suggested in the model organisms B. subtilis, E. coli and S. cerevisiae by a series of related genetic studies that analysed the viability of thermosensitive replication mutants at high temperature in cells compromised in CCM genes. In B. subtilis, this analysis uncovered strong genetic links between the 3-carbon part of the glycolysis/gluconeogenesis pathway (i.e. gapA, pgk, pgm, eno and pykA see Supplementary Fig. S1) and three replication enzymes, namely the DnaC helicase, DnaG primase and DnaE polymerase. Significantly, these three replication enzymes are loaded early at oriC during the initiation of replication and form a ternary complex in the replisome to carry out DNA melting and lagging strand synthesis., This would make these enzymes good target candidates for modulating replication initiation and elongation in response to changes in CCM activity. In E. coli, studies have uncovered strong genetic links between pta, ackA and genes encoding pyruvate dehydrogenase subunits (lpdA and aceE) (Supplementary Fig. S1) and the replication initiator protein DnaA, linking CCM to initiation. In the budding yeast, three glycolytic enzymes (HXK2, ENO2 and PGM1) were found genetically linked to MCM1 a multifunctional protein required for the stable maintenance of minichromosomes that binds DNA sequences to stimulate initiation of DNA synthesis and to regulate transcription of genes involved in diverse cellular functions including replication and the cell cycle. Other, independent, studies further support the hypothesis of CCM-replication links. For instance, an alteration of replication initiation was detected in cells defective in the glycolytic enzymes GapA, PykA and PdhB in B. subtilis and several CCM genes were found to be important for replication in eukaryotic cells. The CCM-replication link may even be more complicated as a striking inter- and intra-cellular diversity in E. coli and B. subtilis metabolism was found and we hypothesized that DNA replication may influence cell’s metabolism. To get insights into how CCM is linked to replication in B. subtilis, we analysed here the coupling between growth and replication in a library of 25 CCM mutants cultivated in three different rich media.

2. Materials and methods

2.1. Strains and plasmids

Strains and plasmids are listed in Supplementary Table S1. B. subtilis cells were grown at 37°C in LB supplement or not with malate 0.2% or in minimal medium (K2HPO4: 80 mM; KH2PO4: 44 mM; (NH4)2SO4: 15 mM; C6H5Na3O7 2H20: 3, 4 mM; CaCL2: 50 mM; MgSO4: 2 mM; FeIII citrate: 11 µg/mL; MnCl2: 10 µM; FeSO4: 1 µM; FeCl3: 4 µg/mL; Trp 50 µg/mL) supplemented with glucose 0.4%, casein hydrolysate 0.2%, malate 0.4%, glutamine 0.4%, proline 0.4% and/or succinate 0.4%, as listed in Supplementary Table S2. Unless stated otherwise, antibiotics were used at the following concentrations: spectinomycin (Sp, 60 µg/mL); kanamycin (Km, 5 µg/mL); erythromycin (Em, 0.6 µg/mL); chloramphenicol (Cm, 5 µg/mL); phleomycin (Pm, 10 µg/mL). Liquid cultures were performed in a water bath under strong shaking (200–230 rpm). The LB medium is a complex rich medium that supports steady-state growth at OD600nm <0.3 and a gradual decrease of the growth rate up to saturation (OD600nm > 3). The regimen is glycolytic in the first part of the growth curve as shown by direct measurement of the activity of promoters sensitive to the glycolytic (gapA promoter) and neoglucogenic (pckA and gapB promoters) carbon flux.

The E. coli plasmid pMUTIN2-Pm is a derivative of pMUTIN2 (EmR) in which the erm gene was replaced by a Pm (pmr) marker. The plasmid pLJH88 and pLJH89 were constructed by replacing the small HindIII-NheI fragment of the E. coli plasmid pDR111 (SpcR) by PCR HindIII-NheI fragments produced from pALS10 and containing the relAt or relAi ORFs preceded by a ribosome-binding site, respectively. The relA constructions were then inserted at the amyE locus of the B. subtilis chromosome by double crossover event to permit RelAt and RelAi expression from the IPTG-dependent Pspank promoter. The plasmid pLJH90 is a CmR derivative of pLJH88. It was constructed by replacing the spc-containing HindIII-SacII fragment of pLJH88 by the cat-containing HindIII-SacII fragment of pDG1661. The plasmids pLJH92 and pLJH93 are derivatives of pMTLBs72 which is a low (6) copy number, CmRB. subtilis plasmid. They contain at the SmaI site a PCR fragment generated from pLJH88 or pLJH89 and containing the relAt or relAi gene under the control of Pspank. The pLJH160 plasmid is a pMUTIN2 (EmR) derivative in which the small EcoRI-BamHI fragment was replaced by an EcoRI-BamHI PCR fragment containing the C-terminal region of pykA fused to a 12 bp long linker and the FLAG tag (5'AGCGGTAGCGGTGACTACAAAGACGATGACGACAAA3'). Upon insertion in the chromosome by single crossover, B. subtilis encodes a PykA protein tagged with FLAG. The plasmid pLJH162 is a pMUTIN2 (EmR) derivative in which the small EcoRI-BamHI fragment was replaced by an EcoRI-BamHI PCR fragment containing the C-terminal region of gapB fused to the 12 bp long linker and the FLAG tag. The gapB region contains in addition a point mutation that inactivates the catalytic site of GapB (151SCT>AAA; TCATGCACC>GCGGCTGCA). The insertion of the plasmid in the B. subtilis chromosome by single crossover results in the production in an inactive, FLAG-tagged GapB protein. Plasmids pLJH198 and pLJH199 are derivatives of the E. coli pJL87 (CmR) that harbours between the HindIII-NheI restriction sites, a PCR fragment containing the 5' end of gapB or pckA, respectively. Once inserted in the B. subtilis chromosome by single-crossover, the gapB or pckA gene are expressed from Phyper-spank, an IPTG-dependent promoter. The pLJH200 plasmid is a pDR111 (SpcR) derivative in which a HindIII-NheI PCR fragment containing the B. subtilis pgk ORF and its ribosome binding site was inserted between the corresponding restriction sites of the recipient plasmid. This plasmid allows insertion of the pgk gene at the amyE locus by double crossover and the inserted gene is expressed from the IPTG-dependent Pspank promoter. The plasmid pLH206 is a pLJH200 derivative in which the spc marker was replaced by a cat gene. This was done by inserting between the LguI-EcoRI restriction sites of pLJH200 a LguI-EcoRI PCR fragment containing the cat marker of pMTLBs72. Plasmid constructions were carried out using the In-Fusion® HD Cloning kit (Clontech, France) and Stellar E. coli competent cells (Clontech, Saint-Germain-en-Laye, France). Constructions were checked by DNA sequencing. B. subtilis strains were constructed by transforming competent cells with genomic or plasmid DNA as indicated in Supplementary Table S1. The genotype of the transformants was controlled by PCR, phenotypes analysis and/or DNA sequencing.

2.2. Quantitative PCR

2.2.1. Measurement of the ori/ter ratio

Wild-type cells cured of prophages (TF8A) and containing or not CCM mutations were grown overnight in different liquid media in triplicates (with a very few exceptions, the three independent cultures were inoculated with different constructs). Saturated cultures were then diluted 500-fold in the same media and growth was carefully monitored using spectrophotometry. Samples for qPCR analysis were collected at low cell concentration (OD600nm 0.1–0.15) to ensure that cell cycle parameters are determined in steady-state cells and are not affected by the approach to the stationary phase or by changes in medium composition. The genomic DNA was extracted using a standard procedure and every qPCR reaction was carried out using two technical repeats of 4–5 serial dilutions. A non-replicating control DNA (stage II sporlets) was analysed simultaneously to the samples in about 1/3 of the qPCR plates. Primers for the marker frequency analysis (see Supplementary Table S3 for details) were used at 200 nM in 12 µl of the 1× SYBR qPCR Premix Ex Taq (Tli RNaseH Plus) (Ozyme, St Quentin en Yvelines, France) and DNA amplification was carried out as follows: denaturation: 95°C, 30s; amplification: (33 cycles) 95°C, 5 s and 60°C, 1 min. For every qPCR reaction, the primer efficiency (97 to 103%) was controlled. The ori/ter ratio of the non-replicating sporlets was 0.585 +/– 0.019, a value consistent with the size of the PCR products (Supplementary Table S3). The amplifications were carried out on a Mastercycler® ep realplex (Eppendorf, Le Pecq, Fr) and the ΔCq were determined automatically from the cycle threshold.

The pgmIP and enoLP metabolic mutants (DGRM353 and DGRM354, respectively) are thermosensitive (Supplementary Fig. S2). To determine the effect of these mutations on cell cycle parameters, the following procedure was followed. The strains were precultured over-night at 30°C in triplicates, diluted 1,000-fold for growth at restrictive temperature (47°C for pgmIP and 45°C for enoLP) and cell cycle parameters (growth, ori/ter ratio and the C period) were determined as above. Results (Supplementary Table S4) were then used to calculate the corresponding parameters at 37°C. For this, we first analysed the effect of temperature on growth, the ori/ter ratio and C period of the wild-type (TF8A), ΔytsJ (DGRM350) and ΔytoI (DGRM360) strains (these mutations are present in the pgmIP and enoLP strains, Supplementary Table S1). Results showed that temperature has nearly no effect on the ori/ter ratio while it changes the growth rate and the C period in a similar way in the three strains (Supplementary Fig. S3), as expected from previous studies., They in addition showed that the ΔytsJ and ΔytoI mutations have no effect on the growth-ori/ter ratio coupling (compare data Supplementary Fig. S3 37°C with Figs 1 and 2). The equation of the slopes Supplementary Fig. S3 were then used to derive the cell cycle parameters of the thermosensitive CCM mutants at 37°C (Supplementary Table S5).

Figure 1.

Growth-replication coupling in wild-type cells. Cell cycle parameters of wild-type cells (TF8A) growing exponentially for more than 10 generations in different media were determined: cell growth was monitored by spectrophotometry, the ori/ter ratio and C period by qPCR and the number of origins/cell as well as the D period by flow cytometry (see the text and Materials and Methods section for details). Left panel: ori/ter ratios and C periods plotted against growth rates. Right panel: numbers of origins/cell and D periods plotted against growth rates. The equations of the ori/ter ratio and C period curves (dotted lines) are: y = 1.0942e0.4469x and y = –4.0435x + 52.835, respectively. The ori/ter ratio and number of origins/cell are means of at least three independent replicates (SD/means <10%). The C periods were determined from representative experiments using 18 loci scattered on both arms of the chromosome and the D period was calculated from the number of origins/cell and the C period as indicated in the text.

2.2.2. Measurement of the C period

Cell growth, genomic DNAs extraction and amplification reactions were as above. Primer pairs, their positions on the chromosome map and the size of the PCR products are given Supplementary Table S3. Results were reported to the terC signal and corrected for the size of the PCR products. The C period was extracted from the slope of the straight line of the plot (log2Nm)τ against m where Nm is the relative copy number of a sequence N at position m (with 0 < m < 1 along the ori-ter axis) and τ the generation time in minutes. Control experiments showed that the marker frequency of the different loci in the non-replicating DNA was close to 1 (1.011 +/– 0.045) as expected. Similar data were obtained with primers on the right and left arm of the chromosome.

2.3. Flow cytometry analysis

2.3.1. Measurement of the number of origins/cell

Strains were grown as indicated in the previous section and at OD600nm 0.1–0.15, chloramphenicol (200 µg/mL final) was added to the cultures to impede replication initiation, cell division and allow completion of ongoing rounds of replication. After 4–6 hours of drug treatment, 108 cells were fixed in filtered ethanol 70% and stored at 4°C. Stored cells were then washed in filtered 0.1 M of phosphate buffer (PB) pH 9, resuspended in 1 mL of Tris-buffered saline buffer (TBS 150, filtered) (20 mM Tris-HCl pH 7.5, 150 mM NaCl) and stained with Hoechst 33258 (1.5 µg/mL) for at least 30 min as described elsewhere. Flow cytometry analysis was performed using a MoFlow Astrios cell sorter (Beckman Coulter, Life Sciences) equipped with a 355 nm krypton laser and a 448/59 nm bandpass filter used to collect Hoechst 33258 fluorescence. Data were analysed with the Kaluza software (Beckman Coulter, Life Sciences). We counted 50,000 to 100,000 events. In most of the tested samples, DNA histograms showed 1–2 main peaks with a 2n distribution. The number of origins/cell was obtained from the proportion of cells with 1, 2, 4, 8 and/or 16 chromosomes.

2.3.2. Measurement of the D period

The D period was calculated from the equation D = [(ln(ori/cell)τ/ln(2)] – C where τ is the generation time and C the C period.

2.4. Suppression assay

Three independently constructed mutants containing a dna(Ts) and the ackA deletion were grown overnight in LB at 30°C. Appropriate dilutions were then spread on solid LB medium and incubated at permissive (30°C) or restrictive temperature as indicated previously.

2.5. Microscopy analysis

For analysing the production of filament in dna(ts) mutants, cells were grown overnight at 30°C in LB and diluted 500-fold in the same medium for further growth at permissive or restrictive temperature. After 2 hours of incubation, cells were collected and the nucleoid was stained with 4',6'-diamidino-2-phenylindole (DAPI, 10 µg/mL) (Sigma). For analysing the effect of (p)ppGpp production on cell size, relevant strains were grown at 44°C in LB in the presence of different concentrations of IPTG. After about 10 generations of growth and at OD650nm = 0.4, cells were collected and the membrane was stained by FM5-95 (Biotium Inc, Fremont, CA). Stained cells were then mounted on glass slides (VWR) covered with 1.2% agarose (low melting point, Sigma) in minimal medium supplemented with malate 0.2% and casa 0.2% and a 0.17 mm glass coverslip (VWR) was placed on top. The microscopy was performed on an epifluorescence microscope (Zeiss, Axio Observer.Z1) with a ×100 magnification oil-immersion objective (Plan-APOCHROMAT Ph3) and a CMOS camera (Orca Flash 4.0 LT Hamamatsu). Digital images were acquired and analysed using Zen 2012 software. All experiments were independently performed at least three times and representative pictures are shown.

2.6. (p)ppGpp analysis

The (p)ppGpp production in WT cells and (p)ppGpp mutants was analysed before and after induction of the stringent response by serine hydroxamate for E. coli and arginine hydroxamate (250 µM) for B. subtilis) as previously described.

2.7. β-Galactosidase activity

The β-galactosidase activity encoded by the bgaB gene of B. stearothermophilus was measured as described previously.

3. Results

3.1. Growth-replication coupling in wild-type cells

A hallmark of growing populations of E. coli and B. subtilis cells is the concurrent increase in the rate of growth and DNA synthesis. This coupling implies that (i) bacteria growing slowly in poor carbon sources host a chromosome that either is not being replicated or is undergoing a single round of replication while (ii) fast-growing bacteria in rich media contain multi-forked chromosomes undergoing up to three overlapping replication cycles. To characterize in more detail the growth-replication relationship in B. subtilis, we quantified cell cycle parameters of a wild-type strain growing at rates ranging from 0.5 to 3 doublings per hour (db/h) in different media. This strain (TF8A) and the derivatives used below, are free of prophages to avoid any interference of phage induction on chromosomal replication. Growth was assessed by spectrophotometry, and DNA replication by marker frequency using qPCR. In the latter analysis, the amount of replication terminus (ter) sequences was compared with the amount of 18 other loci regularly scattered along both arms of the chromosome and including the replication origin (ori). This gave us two major replication parameters: the relative copy number of ori versus ter sequences (the ori/ter ratio) and the C period. This latter parameter was extracted from the slope of the straight line of the plot (log2Nm)τ against m where Nm is the relative copy number of a sequence N at position m (with 0 < m < 1 along the ori-ter axis) and τ the generation time in minutes. The results show a strong positive correlation between growth and the ori/ter ratio, as expected,,, (Fig. 1, left panel, Supplementary Table S2). They also show a significant negative correlation between growth and the C period: the time taken to replicate the chromosome increases from about 40 to 50 min when the growth rate decreases from 3 to 0.5 db/h. Calculation of the C period (C) from the ori/ter ratio and τ using the formula ori/ter = 2C/τ confirms this negative correlation (Supplementary Table S2).

To get further insights into the growth-replication relationship, the number of origins/cell and the D period of the wild-type cells grown in various media were measured by flow cytometry. To that goal, we first analyzed the size of events detected by the flow cytometer in order to discriminate single cells from the chains of cells that often accumulate in fast growing B. subtilis cultures. This entailed using density plots of area versus width of fluorescence signals generated by cultures stained with a DNA specific dye (Hoechst 33258). Representative results are presented Supplementary Fig. S4 (left panels). The width axis corresponds to the flight time of particles through the laser beam spot. It reveals small (S) and large (L) events and that the proportion of large events strongly increases (from 1 to up to 60%) with the medium richness and growth rate. The area axis gives the integral of fluorescent signals. It shows that the intensity of fluorescence signals of small events varies while that of large events is rather constant and strong. Taken together, these results show that small events correspond to single cells while large events correspond to cell-chains and/or to clump of cells.

The samples used to measure the number of origins/cell and the D period corresponds to exponentially growing cultures treated for 4–5 hours with Cm. This treatment allows completion of ongoing replication cycles but prevents re-initiation of new rounds of replication and cell division. Hence, at the end of the treatment, cells contain an integral number of fully replicated chromosomes that corresponds to the number of origins at the time of drug addition (replication run-out assay). It is of interest to note here that the proportion of cell-chains in Cm-treated cultures is greatly reduced (to 1–5%) as shown by density plots in Supplementary Fig. S4, right panels. DNA histograms of Cm-treated, single cells (Supplementary Fig. S5) show a smooth increase (from 2 to 6) of the number of origins/cell at growth rates ranging from 0.5 to 2 db/h followed by a stronger increase and a plateau to 9 origins/cell for cells growing at rates >2 db/h (Fig. 1, right panel and Supplementary Table S2). The D period, calculated from the equation D = [(ln(ori/cell)τ/ln(2)] – C, varies with the growth rate in B. subtilis as expected (Fig. 1 and Supplementary Table S2).

Overall, the cell cycle parameters determined here show that the coupling between growth and replication in B. subtilis cells growing at rates >0.5 db/h depends on a concerted modulation of the initiation and elongation phases of replication. In agreement with E. coli studies, they also support the hypothesis that this coupling depends on the precursors and energy extracted from nutrients rather than on the nature of available carbon sources.

3.2. Identification of CCM genes important for the growth-replication coupling

To look for a possible involvement of CCM in the growth-replication coupling, we analysed cell cycle parameters of CCM mutants grown in rich media and compared the results with data obtained for the wild-type strain grown in different carbon sources. We assumed that this approach is relevant as coupling depends on the energy and precursors extracted from nutrients and not on the amount and nature of nutrients available in the environment (see above). This assumption is further supported here by a coupling analysis in B. subtilis cells extracting energy and precursors from a rich medium (LB) at different rates because of inhibition of ribosome production. Such an assay was developed in E. coli using cells expressing a truncated form of the E. coli (p)ppGpp synthetase RelA from an inducible promoter. This truncated protein (termed RelAt) synthesizes constitutively the inhibitor of ribosome synthesis (p)ppGpp without the need for metabolic stress and binding to idling ribosomes., This genetic system was proved to modulate E. coli growth rate and cell size in LB in an inducer-dependent manner. To develop a similar approach in B. subtilis, the E. coli relAt gene was inserted in the B. subtilis chromosome and put under the control of an IPTG inducible promoter. A related strain expressing an inactive form of E. coli RelA (RelAi) was constructed as a control. As in E. coli, an inverse correlation between (p)ppGpp production, growth rate and cell size was found in B. subtilis (see Supplementary data and Supplementary Fig. S6). Measurement of the ori/ter ratio shows a wild-type coupling over the large range of growth rates generated by degrees of (p)ppGpp production in LB (Supplementary Fig. S7). Hence, B. subtilis sets the replication rate according to the growth rate determined by its metabolism and not by the richness of the growth medium. It is thus reasonable to compare data from wild-type cells grown in various media to data from CCM mutants grown in rich media. This underpins our hypothesis that CCM mutants growing at various rates in rich media will have a wild-type growth-replication relationship unless the metabolic mutation alters cell cycle parameters.

We used a library of 25 CCM mutants and three rich media to analyse the growth-replication coupling in 53 metabolic conditions. The mutations correspond to deletions (16), point mutations (5) or inducible transcriptional fusions (4). They decrease or abolish the activity of 19 genes (i.e. two-thirds of the key CCM genes; see Supplementary Table S1 for details on the strain library) that map in glycolysis (ptsG, pgi, gapA, pgk, pgm, eno and pykA), gluconeogenesis (pckA and gapB), the pentose phosphate pathway (zwf), the TCA cycle (citC and mdh), the overflow pathway (pta and ackA) and in the pyruvate metabolism (pdhB, pdhC, ytsJ, pycA and pps) (see the CCM map Supplementary Fig. S1 for details). The rich media (LB; LB + malate and the minimal medium + malate + casa – casein hydrolysate) feed both the upper and lower parts of CCM and support growth of mutants at rates ranging from about 0.5 to 3.5 db/h (see Materials and Methods and Supplementary Table S5 for details) (note that the ΔpdhB and ΔpdhC strains did not grow in malate + casa and that the Δpgi mutant was not analysed in LB and LB + malate as cells had abnormal shapes in these media – not shown).

The growth rate, ori/ter ratio and C period in the 53 metabolic conditions were measured and compared with curves typifying the growth-replication coupling in wild-type cells (Fig. 2, top panels and see Supplementary Table S5 for details). Based on the SD/mean of experiments (<10%), we assigned the metabolic conditions into one of three classes. The first class (class I) contains the majority of conditions (34/53) and the growth-replication coupling appeared to be wild-type as the ori/ter ratio and the C period deviate by less than 20% from the typical wild-type curves. The second class (14/53) (class II) contains the conditions associated with a significant alteration of the growth-replication coupling with both the ori/ter ratio and the C period deviating by 20–120% from the typical wild-type curves. The third class (5/53) (class III) contains conditions in which a mixed phenotype is observed: the metabolic mutants exhibit a defect in only one replication parameter (the ori/ter ratio or the C period). Flow cytometry analysis shows that metabolic conditions conferring a class II phenotype at 37°C are also often associated with an increase in the number of origins/cell but not with a major change in the D period (this is of importance as changes in the D period may affect the growth-replication coupling by altering the timing of replication initiation) (Supplementary Fig. S8) [note that (i) mutations—pgmIP and enoLP—conferring a class II phenotype at high temperature were not analysed and that (ii) this analysis was limited to conditions conferring a growth rate ≤2 db/h as the growth-origins/cell reference curve is inaccurate at higher growth rates (Fig. 1, right panel)].

Figure 2.

Identification of CCM mutations important for the growth-replication coupling in rich media. Top panels: Growth-ori/ter ratio and growth-C period coupling in CCM mutants grown in rich media (LB, LB + malate and malate + casa) (see Supplementary Table S5 for details). Closed dots and dotted line: results from wild-type cells (Fig. 1); Green and red symbols: CCM mutants with parameters that deviate from the typifying curve by less or more than 20%, respectively. Bars in the growth-ori/ter ratio relationship stand for mean and standard deviation obtained from at least three independent biological repeats (obtained generally from independent constructs) and two technical repeats (SD/mean < 10%). Bottom panels: Green symbols: CCM genes dispensable for both the growth-ori/ter ratio and growth-C period relationships (class I mutants). Red: CCM genes important for both relationships (class II mutants). Blue: CCM genes important for at least one relationship (class III mutants). zwf: glucose 6-phosphate dehydrogenase; ptsG: phosphoenolpyruvate: carbohydrate phosphotransferase enzymeIICBA; pgi: glucose 6-phosphate isomerase; gapA: glyceraldehyde-3-phosphate dehydrogenase, NAD-dependent; gapB: glyceraldehyde-3-phosphate dehydrogenase, NADP-dependent; pgk: phosphogycerate kinase; pgm: phosphoglycerate mutase; eno: enolase; pykA: pyruvate kinase; pdhB: pyruvate dehydrogenase (E1 beta subunit); pdhC: pyruvate dehydrogenase (dihydrolipoamide acetyltransferase E2 subunit); pta: phosphotransacetylase; ackA: acetate kinase; citC: isocitrate dehydrogenase; mdh: malate dehydrogenase; ytsJ: malic enzyme (NAPDH-producing); pckA: phosphoenolpyruvate carboxykinase; pycA: pyruvate carboxylase; pps: phosphoenolpyruvate synthase.

Taken together, the results show that the alteration of the growth-replication coupling in CCM mutants cultivated in rich media mainly results from an early firing of replication initiation and/or from a decrease in the mean velocity of fork progression. Hence, CCM is important for DNA replication. As initiation and elongation defects perturb replication cycles, our results suggest that a balanced CCM activity is needed for a normal timing of DNA replication in the cell cycle. Significantly, we did not observe a consistent correlation between reduced growth and replication defect indicating that a mere alteration of CCM activity and cellular metabolism is not sufficient to alter replication (Fig. 2).

Eleven CCM genes appear to be important for replication in rich media (Fig. 2, bottom panels). These genes are located in the main routes of carbon source degradation: the 6- and 3-carbon part of glycolysis/gluconeogenesis, the pentose phosphate pathway and the overflow pathway. Interestingly, the sets of genes important for replication differ from one medium to another with some genes being involved in just one medium (zwf, pgi, gapA, gapB, pta and mdh), and others being involved in two (pgk, pykA and ackA) or in three (eno and pgm) media (Fig. 2, bottom panels). It can thus be inferred that CCM is connected to DNA replication through multiple links that are committed in replication in a medium manner.

3.3. CCM mutations affect the initiation or elongation phase of replication

To determine which replication step is defective in CCM mutants, we constructed a series of strains replicating their genome from a plasmid replicon rather than from the chromosomal origin oriC. The replicon used (termed oriN thereafter) originates from the low copy-number plasmid pLS32. It encodes an initiator protein RepN that binds the plasmid origin embedded in the repN gene to initiate replication., In addition to RepN, initiation at oriN depends on the host-encoded factors DnaD, DnaB (two helicase loaders), DnaC (the helicase) and DnaN (the DNA polymerase clamp). It is however independent of the initiator DnaA and the oriC sequence. Once assembled at oriN, the initiation proteins (RepN, DnaB, DnaC and DnaD) recruit the rest of the chromosomal replication machinery to carry out bidirectional replication. In the series of strains constructed here, the plasmid replicon was inserted 2 kb to the left of the chromosomal origin and normal initiation of chromosomal replication was inactivated by introducing three ochre mutations into the DnaA-encoding gene or by deleting oriC sequences.,

The plasmid replicon used responds poorly to differences in carbon sources (see below). This means that if uncoupling between growth and replication in CCM mutants were to result from changes in initiation (i.e. the productive interaction of DnaA with oriC), cells replicating their genome from the plasmid replicon would not suffer from the metabolic mutation and would thus exhibit the coupling of oriN-dependent cells, wild-type for metabolic functions. In contrast, if uncoupling were to result from changes in DNA elongation (or in an initiation event downstream of the formation of an active DnaA/oriC complex), plasmid replicon-dependent cells would still suffer from the metabolic mutation and would thus not exhibit the coupling typical of the CCM+, oriN-dependent cells (i.e. growth and replication would be uncoupled).

We first analysed the growth-ori/ter coupling in dnaA or ΔoriC cells (which replicate the chromosome from oriN) growing in different media. As a control, an isogenic strain containing oriN and a functional chromosomal origin (DnaA+ OriC+) was analysed. The results show that this control strain (which contains oriN but does not depend on it) had the strong, positive relationship between growth and the ori/ter ratio characteristic of B. subtilis (Supplementary Fig. S9, top Table and panel) whereas oriN-dependent cells had a much weaker positive relationship (Supplementary Fig. S9, bottom Table and panel). This weaker response is probably because the plasmid replicon (i) is relatively insensitive to differences in the richness of media and (ii) cannot support high growth rates since it is unable to provide the high frequency of DNA initiation events required in rich media.

To determine which replication step is defective in CCM mutants, the growth-ori/ter ratio coupling in oriN-dependent derivatives was analysed as above. This analysis was focused on class II metabolic conditions where both the ori/ter ratio and the C period are affected at 37°C (i.e. the class II metabolic conditions affecting replication at high temperature and involving the pgmIP and enoLP mutations were not tested here, as the replication activity of the oriN replicon at high temperature is not known). Out of the nine tested metabolic conditions, four give a coupling typical of oriN-dependent cells while five exhibit a perturbed oriN-type coupling (Fig. 3 and inserted table). As reasoned above, the loss of the uncoupling phenotype in oriN-dependent CCM mutants shows that the corresponding CCM mutations affect initiation (i.e. the formation of an active DnaA/oriC complex) at the chromosomal origin in oriC-dependent strains, while the persistence of this phenotype in oriN-dependent cells shows that the CCM mutations affect DNA elongation (or in a step downstream of the formation of an active DnaA/oriC complex) in the parental oriC-dependent strains. This interpretation is strengthened by control experiments showing that the assay is not sensitive to the way initiation at the chromosomal origin is abrogated (by the dnaA or ΔoriC mutation, Supplementary Fig. S10). Collectively, our results reveal that CCM genes are connected to the initiation (i.e. the formation of an active DnaA/oriC complex) or elongation (or an initiation step downstream of the formation of an active DnaA/oriC complex) phase of replication in a medium-dependent manner (Fig. 3). Although we investigated a limited number of class II conditions, links between CCM genes and different key replication phases were found in two out of the three tested media. This suggested that the mechanism of the growth-replication coupling in B. subtilis, which depends on a concerted modulation of the initiation frequency and elongation velocities (Fig. 1), may involve the combined effect of groups of CCM genes on the initiation and elongation phase of replication. This hypothesis is supported by the results presented below.

Figure 3.

Identification of the replication step responding to CCM mutations. The growth and ori/ter ratio of CCM–oriN-dependent strains were determined in class II metabolic conditions and compared with the corresponding coupling curve of the CCM+oriN-dependent strain (Supplementary Fig. S9). Inserted table: Cell cycle parameters of the oriN-dependent CCM mutants grown in different media and replication target of CCM mutations (Initiation refers to the formation of an active DnaA/oriC complex and Elongation refers to a downstream initiation step or to the elongation phase of replication). Panels: Close diamonds and dotted lines: results from oriN-dependent, CCM+ cells (Supplementary Fig. S9); Open triangles: CCM mutants with a relationship typical for oriN-dependent, CCM+ cells; Close triangles: CCM mutants with a relationship different from the parental oriN-dependent, CCM+ strain.

3.4. Genetic links between CCM and replication genes in LB

The cell cycle studies of oriN-dependent cells presented above show that the genes encoding catalysts of the terminal reactions of glycolysis (gapA, pgk, pgm, eno and pykA) and of downstream reactions (pta and ackA) are important for replication initiation (i.e. the formation of an active DnaA/oriC complex) or elongation (or a step downstream of the formation of an active DnaA/oriC complex) in the LB medium as summarized Fig. 4A, right part of the panel. Intriguingly, a previous study carried out in the same medium showed that pgk, pgm, eno and pykA are also genetically linked to the replication genes dnaC, dnaG and dnaE, while gapA is genetically linked to dnaE (Fig. 4A, left part of the panel). Interestingly, the enzymes encoded by these replication genes (the helicase DnaC, the primase DnaG and the lagging strand polymerase DnaE) were recently shown to be required for the initiation and elongation phase of replication.,, Indeed, during initiation of DNA replication, the conserved replication initiator DnaA protein, aided by the initiation factors DnaB, DnaD and DnaI (a helicase loader), melts the origin, oriC, to mediate anti-parallel loading of two hexameric rings of the DnaC helicase, one on each melted strand. The primase, DnaG, and the lagging strand polymerase, DnaE, are then recruited to synthesize RNA primers and start DNA polymerization., The remaining replisomal components, including the clamp, DnaN, the clamp loader and the polymerase PolC, are then assembled to form two fully fledged replisomes with two different DNA polymerases (DnaE and PolC) that move away from oriC in opposite directions and replicate in a concerted manner the leading and lagging strands up to the terminus where the two replication forks merge.,

Figure 4.

Suppression of dnaE(Ts) mutations in the ΔackA context. (A): Functional (dark lines) and genetic (dotted lines) links between (i) CCM genes, (ii) the initiation and elongation phase of replication and (iii) the DnaC, DnaG and DnaE replication enzymes. (B): Viability and microscopy analysis of a ΔackA dnaE (Ts) double mutant. Wild-type cells and strains carrying the thermosensitive dnaE2.6 mutations combined or not with the deletion of the ackA gene were grown in LB at permissive temperature. They were then streaked on LB plates or inoculated in liquid LB broth and then incubated at permissive or restrictive temperature. Left panels: colony formation at permissive (30°C) and restrictive (42°C) temperatures. Right panels: Microscopic analysis of cells grown at permissive or after 2 hours of incubation at restrictive temperature. Merge images of phase contrast and DAPI stained nucleoid (colored in blue). White arrows: anucleated minicells.

To extend to ackA our search for genetic links between CCM and replication genes in LB, we used the same method as previously (note that this method did not detect genetic links between pta and replication genes). For this, we first constructed a series of ΔackA strains containing a thermosensitive (Ts) mutation in replication genes. The double mutants were then tested for their viability (by plating) and ability to prevent filament formation (by microscopy) at restrictive temperature. Our results show that the ackA deletion restores viability and prevents filament formation at high temperature in every dnaE(ts) mutant tested (4/4) but failed to suppress thermosensitive mutations affecting other replication enzymes (namely: DnaB, DnaD, DnaC, DnaG, PolC, and DnaN) (Fig. 4B and Supplementary Table S6). Taken together, our results suggest that pgk, pykA (and possibly pgm and eno) are important for initiation (from the formation of the DnaA/oriC complex up to the recruitment of DnaC, DnaG and DnaE and the loading of two fully fledged replisomes) via an effect on DnaC, DnaG and DnaE while gapA and ackA are important for elongation via an effect on the lagging strand polymerase DnaE (Fig. 4A). This latter hypothesis is enforced by data showing that the deletion of gapA or ackA suppresses dnaE(ts) mutants mainly affected in DNA elongation.

4. Discussion

We report here on a rather large study aimed at better understanding the important and likely universal relationship between central carbon metabolism (CCM) and DNA replication. To achieve this, we characterized the cell cycle parameters in three rich media of a collection of 25 CCM B. subtilis mutants. The results reveal defects in the initiation and/or elongation phase of replication in about one-third of the metabolic conditions tested. These defects are rather strong, causing a significant increase in the concentration of the origin sequences per cell and an up to 2-fold decrease in the mean velocity of replication forks. We infer from these results that CCM is important for replication and, in particular, for the coupling between growth and replication that positions the replication cycle within the cell cycle. These roles for CCM in replication are even more surprising as the replication phenotypes reported here were found in cells that are fully proficient in replication enzymes and replication control functions.

Overall, 11 genes mapping in different CCM pathways (the pentose–phosphate, glycolysis/neoglucogenesis, TCA and overflow pathways) appear to be important for replication in rich media (Fig. 2). Interestingly, their commitment in replication is medium dependent, forming partly overlapping subgroups with some genes only being needed for a single medium and others being needed for two or three media (Fig. 2). Moreover, some CCM genes are important for initiation while others are important for elongation or for both replication steps (pgk) (Fig. 3). Taken together, our results suggest that cells evolved a toolbox of intermingled CCM-replication links that it partially activates to cope with requirements imposed by the environment.

The results reported here show that the coupling between growth and replication in wild-type cells depends both on the modulation of the frequency of initiation and the mean velocity of replication forks (Fig. 1). Studies on CCM mutants confirm this conclusion (Figs 3 and 4) and suggest that this coupling may depend on both CCM-initiation and CCM-elongation links. It is thus tempting to speculate that these links are part of the metabolic control of DNA replication in B. subtilis.

Additional results presented here and in a previously publication uncover strong genetic links between six CCM genes (gapA, pgk, pgm, eno, pykA and ackA) and three replication genes (dnaC, dnaG and dnaE) in cells growing in the rich LB medium (Fig. 4A). These links allow CCM mutations to restore the viability of thermosensitive replication mutants at restrictive temperatures. Interestingly, the links were suggested to depend on metabolic-induced conformational changes that restore the activity of the thermosensitive replication enzymes at restrictive temperature. They however may not dependent on a (i) deep change in the replication machinery that make thermosensitive replication enzymes dispensable for replication, (ii) the overexpression of dna(Ts) genes, (iii) a decay in the degradation of the thermosensitive replication protein, (iv) a reduction in ATP concentration or (v) an induction of a stress response. Moreover, the replication enzymes responding to CCM alteration encode the helicase DnaC, the primase DnaG and the lagging strand polymerase DnaE that are loaded early at oriC during the initiation of replication and form a ternary complex in the replisome for carrying out DNA melting, RNA priming and essential polymerase functions in lagging strand synthesis., Although the suppression assay and the CCM-replication studies are based on distinct approaches and concepts, their combination gives us for the first time a coherent picture on how cell’s metabolism and DNA replication may be coupled in LB. Indeed, results showed that at least two metabolic genes important for initiation (pgk and pykA) are genetically linked to dnaC, dnaG and dnaE while CCM genes important for elongation (gapA and ackA) are genetically linked to dnaE (Fig. 4). This suggests that, in the LB medium, CCM may modulate initiation by altering the functional recruitment of DnaC, DnaG and DnaE to oriC and may modulate DNA elongation by altering the activity of the polymerase DnaE in replication forks. This latter hypothesis is consistent with data showing that fork progression depends on lagging strand operations and that the rate of DNA melting by helicases depends on the speed of DNA polymerases. Along this line, the physical interaction between the helicase DnaC and the polymerase DnaE may provide a direct means for adapting the rate of duplex unwinding and the concurrent synthesis of the leading and lagging strand to the activity of the DnaE polymerase.

How might CCM impact DNA replication? Long-standing hypotheses that the overall process of regulating replication depends on the concentration of the active form of the replication initiator (DnaA–ATP), on the sub-saturation of the replication machinery in precursors and/or on the concentration of (p)ppGpp have been challenged. The results presented here also argue against the sub-saturation hypothesis, as there is no obvious correlation between the impact (positive or negative) of CCM mutations on replication precursor pathways (the main source of (d)NTPs and ATP are the pentose phosphate pathways and the TCA cycle, respectively) and their coupling or uncoupling phenotype (Fig. 2). Post-translational modification of replication enzymes is an alternative mechanism for the metabolism-replication coupling system as such modifications are important parts of the eukaryotic arsenal for regulating DNA replication., Moreover, several lines of experimental evidence indicate that this level of regulation may also exist in bacteria. First, recent advances in proteomic technologies have uncovered post-translational modifications of numerous bacterial proteins including replication enzymes. Second, independent studies show that the E. coli replication initiator DnaA is acetylated and its level of acetylation may play a critical role in regulating DNA replication initiation, and that phosphorylation of the single-stranded DNA-binding (SSB) protein is important for replication in B. subtilis., Third, indirect data suggest that the CCM-replication links detected in B. subtilis and S. cerevisiae depend on a metabolism-dependent allosteric regulation of replication enzymes.,

Post-translational modifications of replication enzymes are carried out by protein kinases/phosphatases and protein acetyltransferases/deacetylases in eukaryote cells. Such enzymes are thus potential effectors of the coupling between CCM and replication, although little is known on how their activity is regulated by metabolism (see however Ref. 87). An alternative way to achieve metabolism-dependent, allosteric-regulation of replication enzymes is via CCM itself. This group of catabolite reactions is central to the supply of precursors and energy from nutrients and to the demand in biosynthetic reactions, moreover, its activity responds almost immediately to changes in carbon sources. CCM is thus at a strategic position to dynamically integrate metabolism with replication. This dynamic integration may involve underexplored activities/functions of CCM metabolites and proteins. Indeed, several CCM metabolites have emerged as important modulators of various cellular activities by modulating dynamically protein activities via post-translational modifications or allosteric binding (see as reviews)., Key metabolites among others are acetyl-coenzyme A and ATP, which are the major donors of acetyl and phosphoryl groups, respectively, and acetyl phosphate which lies at the crossroads between protein phosphorylation and protein acetylation. In addition, CCM enzymes from bacteria to eukarya are often multi-functional proteins carrying out both catalysis of a canonical metabolic reaction and non-metabolic functions involved in processes as diverse as DNA replication, transcription, cell division, signalling, apoptosis and pathogenicity. These moonlighting activities, found in the cytoplasm and the nucleus, involve novel enzymatic activities (like protein kinase activity) and/or interactions with non-metabolic enzymes (see for instance Refs. 94–97). We are currently exploring the fascinating possibility that the CCM-replication links in B. subtilis involve moonlighting activities of CCM enzymes and/or signalling metabolites.

DNA double-stranded breaks are chromosomal lesions with a high mutagenic potential., Their frequency increases in response to replication control defects and the importance of this threat is highlighted by the presence of multiple mechanisms ensuring faithful genome duplication once per cell cycle, and preventing re-initiation of replication., In cells with aberrant timing of initiation, re-initiation of replication may increase the risk of double-stranded breaks by collision between forks or by forks encountering gaps in the DNA template. Such a risk may also increase in cells with a lengthened C period in consequence of a higher probability of fork breakage over time. Through their effect on replication and the timing of DNA synthesis in the cell cycle, the CCM-replication links described in this study may thus form a new determinant of genetic stability. In support of this hypothesis, it was found in E. coli that CCM mutations influence the fidelity of DNA replication and that mutants of the ackA-pta overflow pathway, which is genetically connected to the replication initiator DnaA,, accumulate double strand breaks. Moreover, mutants of the budding yeast impeded in the temporal control of DNA replication may exhibit substantial increases in the spontaneous mutation rate. Further investigations are required to better evaluate the mechanism of CCM-replication links and its role in genetic stability and potentially in cancer as early events engaging normal human cells into the path in tumorigenesis involve major changes in CCM activity (the Warburg effect) and genetic instability.

Click here for additional data file.