Two Essential DNA Polymerases at the Bacterial Replication Fork

See allHide authors and affiliations

Science  23 Nov 2001:
Vol. 294, Issue 5547, pp. 1716-1719
DOI: 10.1126/science.1066351


DNA replication in bacteria is carried out by a multiprotein complex, which is thought to contain only one essential DNA polymerase, specified by the dnaE gene inEscherichia coli and the polC gene inBacillus subtilis. Bacillus subtilis genome analysis has revealed another DNA polymerase gene,dnaEBS , which is homologous to dnaE. We show that, in B. subtilis, dnaEBS is essential for cell viability and for the elongation step of DNA replication, as is polC, and we conclude that there are two different essential DNA polymerases at the replication fork of B. subtilis, as was previously observed in eukaryotes.dnaEBS appears to be involved in the synthesis of the lagging DNA strand and to be associated with the replication factory, which suggests that two different polymerases carry out synthesis of the two DNA strands in B. subtilis and in many other bacteria that contain both polC and dnaEgenes.

The paradigm of the bacterial replication fork is that of Escherichia coli (1). The fork includes the DNA polymerase III holoenzyme, which contains 10 different subunits. One of the subunits, α, is the catalytic DNA polymerase. There are two α molecules in the holoenzyme, each of which copies a different DNA strand. The holoenzyme from B. subtilis is not as well characterized, but was reported to contain a similar number of polypeptides, including a catalytic DNA polymerase subunit (2). The E. coli and B. subtilis catalytic subunits of the holoenzyme, encoded bydnaE and polC genes, respectively, are prototypes of two different DNA polymerase classes within the so-called family C, which groups replicate DNA polymerases from eubacteria (3). The B. subtilis genome sequence (4) indicated the existence of two other genes encoding family C DNA polymerases, in addition to polC, both of the E. coli class. One of these genes, yorl, is carried on a prophage that can be eliminated from B. subtilis (5) and thus is not essential for cell growth. The other gene, referred to here asdnaEBS , is encoded chromosomally. Genes homologous to polC and dnaEBS are present in all fully sequenced genomes of bacteria belonging to theBacillus/Clostridium group and in the genome of a thermophilic microorganism species called Thermotoga maritima. dnaEBS encodes a protein with DNA polymerase activity (6), as does its homolog fromStreptococcus pyogenes (7).

The polC gene is essential for B. subtilis cell growth (8). We show here that dnaEBS is also essential. First, inactivation of the gene by recombination with an insertional plasmid vector, pMUTIN (9), carrying an internal segment of the gene, was not successful. The failure was not due to a polar effect of insertion, which we know because a downstream gene (ytsJ; Fig. 1, top) could be inactivated. Similar results were obtained with the polCgene and the downstream ylxS gene (Fig. 1, top). Second, it was possible to place the dnaEBS gene under the control of the Pspac promoter, which is regulated by the LacI repressor and is therefore induced by isopropyl-β-d-thiogalactopyranoside (IPTG), by insertion of pMUTIN carrying a DNA segment overlapping the 5′-end of the gene (Fig. 1, top). However, the resulting strain required IPTG for growth (Fig. 1, middle) (10). Analogous observations were obtained with the polC gene (Fig. 1, middle). Third, we isolated 11 different thermosensitive dnaEBS mutants (10).

Figure 1

dnaEBS andpolC genes are essential for B. subtilis growth and DNA replication. (Top) dnaEBS andpolC regions in B. subtilis. DNA segments overlapping the 5′ end of each gene (thick lines) were used for single cross-over integration of pMUTIN2 (depicted as a thin line; boxes, lollipops, and bent arrows indicate genes, transcription terminators, and promoters, respectively; the emR gene derives from plasmid pE194) to obtain the strains in which the dnaE orpolC genes are placed under the control of thePspac promoter and were designated HVS614 and HVS609, respectively. (Middle) Growth of HVS614 (left) and HVS609 (right). Cells were grown in Luria broth (LB) medium supplemented or not supplemented with IPTG (open and solid symbols, respectively) and were diluted periodically to maintain exponential growth. Optical density (squares) and colony-forming units (on IPTG-supplemented plates, triangles) were determined. (Bottom) Macromolecular synthesis in HVS609 and HVS614 grown with or without IPTG, measured as described previously (11). Average values and dispersion from four experiments for DNA (circles) and two experiments for RNA (triangles) or protein (squares) are shown.

To test whether dnaEBS is required for DNA synthesis, we cultivated cells having the gene underPspac control, with or without IPTG, and monitored incorporation of [3H]-thymidine into acid-insoluble material (11). DNA synthesis was inhibited by depletion of DnaEBS whereas RNA and protein syntheses were not (Fig. 1, bottom). Analogous results were obtained for thepolC gene (Fig. 1, bottom). We conclude that because both genes are required for DNA synthesis, their products must not have the same role in this process.

Because PolC is required for elongation (8), it seemed possible that DnaEBS might be required for initiation. Two lines of evidence show that DnaEBS is essential for elongation. First, incorporation of labeled thymidine was arrested immediately upon a shift of a thermosensitivednaEBS mutant from 30° to 47°C (Fig. 2, left). Second, the relative abundance of chromosomal markers close to the origin and terminus of replication did not change upon depletion of DnaE or PolC (Fig. 2, right). The origin-to-terminus ratio is between 2:1 and 4:1 in B. subtilis cells during exponential growth (12). We would expect this ratio to remain constant if the elongation of replication were arrested, whereas it would change to 1:1 if the elongation continued but the initiation ceased. We derived two conclusions from these observations. First, depletion of the unaltered catalytic subunit of the B. subtilisholoenzyme PolC interferes with elongation of DNA replication. This suggests that the polymerase is frequently lost from and then reloaded onto the progressing replication fork, which might explain the presence of high PolC levels in B. subtilis, revealed by the analysis of cells carrying the protein fused to green fluorescent protein (GFP) (13). Second, depletion of DnaEBS also arrests elongation, indicating that DnaEBS is required for this replication step.

Figure 2

DnaEBS is required for elongation. (Left) The dnaEBS thermosensitive strain EDJ51 and the parental strain 168 were grown at 30°C in the presence of [3H]-thymidine. At the 1-hour point, half of each culture was transferred to 47°C and the [3H]-thymidine incorporation was determined at 30° and 47°C (open and solid symbols). Squares and circles refer to the mutant (dnaE) and parental (Wt) strains, respectively. (Right) Exponentially growing cells havingdnaEBS or polC genes under the control of Pspac (HVS614 or HVS609, respectively) were inoculated into the rich LB medium with or without IPTG and grown for 3 hours. Their DNA was extracted, cleaved with Eco/RV, and analyzed by Southern blotting with probes from seven different chromosomal regions (positions are indicated in degrees) (10). Gray and white bars refer to samples with and without IPTG.

Because both PolC and DnaEBS are required for the elongation step of DNA replication, they must have different roles in this process. We considered that each might be involved in the synthesis of a different DNA strand, and so we used a pAMβ1-derived plasmid to test this possibility. Replication of pAMβ1 is initiated by DNA polymerase I, which progresses for about 200 base pairs, thus generating a D loop intermediate (14). A PriA-dependent primosome then assembles at a specific site present within the single-stranded part of the D loop (15–17) and promotes assembly of the holoenzyme, which completes the replication of the plasmid. This process mimics the restart of the arrested replication fork. pAMβ1 replication is unidirectional, and the leading and lagging strands are thus known (18). A derivative plasmid was introduced intodnaEBS or polC thermosensitive mutants, and the resulting strains were propagated at 37°C, the temperature at which we speculated the polymerase activity might already be affected. Total cell DNA was prepared and hybridized with probes specific for one or the other of the plasmid strands (Fig. 3). As expected, only the double-stranded plasmid DNA was detected in the control wild-type strain. In contrast, in thednaEBS mutant, the probe complementary to the leading strand revealed, in addition to double-stranded DNA, a fast-migrating DNA form. This form did not hybridize with the probe specific for the other strand, which shows that it was single stranded and corresponded to only one DNA strand. The amount of single-stranded DNA was relatively low (about 3%), possibly for two reasons. One is that inactivation of DnaEBS at 37°C was incomplete, because sufficient polymerase activity had to be maintained to allow cell viability, and thus the uncoupling of the synthesis of the two strands may have taken place in only a fraction of molecules. The other is that the conversion of plasmid single-stranded DNA into a double-stranded form is efficient in B. subtilis. For instance, rolling circle plasmids, in which leading and lagging strand synthesis are fully uncoupled, often accumulate no more than 10% of single-stranded DNA (19). We conclude that synthesis of the lagging strand of the plasmid is affected in thednaEBS mutant. A different plasmid phenotype was observed in the polC mutant, in which the plasmid copy number was drastically reduced (Fig. 3). A low level of single-stranded plasmid DNA could have therefore escaped detection. However, inefficient synthesis of the leading DNA strand could reduce the plasmid copy number, without any accumulation of single-stranded DNA. A simple explanation of these observations is that the plasmid replication fork normally contains two polymerases, PolC and DnaEBS, each of which synthesizes one DNA strand. We suggest that an incomplete fork can assemble without DnaEBSand progress far enough to allow formation of a full plasmid single-stranded DNA (about 5 kb). Formation of a fork containing only one catalytic polymerase subunit has been reported for E. coli (20).

Figure 3

Synthesis of plasmid pAMβ1 lagging strand is deficient in dnaEBS mutant cells. Bacillus subtilis cells harboring plasmid pIL252p1 (30), were grown exponentially at 30°C and shifted to 37°C for 90 min. Their total DNA was prepared and hybridized with probes homologous to the plasmid. +, E, and C refer to the wild-type strain,dnaEBS thermosensitive strain EDJ51, andpolC thermosensitive mutant dnaF33, respectively. Probes were double-stranded (ds) plasmid DNA or oligonucleotides complementary to the leading or lagging plasmid DNA strand. Positions of ds and single-stranded (ss) DNA are indicated.

Is the fork that assembles on plasmid pAMβ1, in a process that mimics the restart of replication, identical to that which normally replicates the B. subtilis chromosome? An argument in favor of this hypothesis is that, in B. subtilis, the process that leads to holoenzyme assembly during restart of the replication fork and initiation of chromosomal replication involves the same four proteins (DnaB, DnaC, DnaD, and DnaI) and only one specific protein [PriA, involved in restart, or DnaA, in initiation at the origin, (16, 17, 21)]. It has been predicted, then, that PolC and DnaE should both be present within the replication factory (13). We therefore examined the localization of the two proteins in the cell, using functional polymerase-GFP protein fusions (10). As shown in Fig. 4, A and B, cells containing a PolC-GFP fusion protein usually exhibited a single discrete focus localized at about mid-cell (lines in Fig. 4A), as reported previously (13). The foci were frequently offset from the central long axis of the cell, which is suggestive of a submembrane localization. In some cells (33%), generally the longer ones, there were two foci, one on either side of mid-cell (arrowheads in Fig. 4A). Under the conditions used, most cells would be expected to have a single ongoing round of DNA replication, with initiation of new rounds occurring just before cell division (22). The localization of a DnaE-GFP fusion was similar to that of a PolC-GFP fusion (Fig. 4, C and D). Most cells showed a single mid-cell focus (lines), with a minor proportion (30%) showing two foci (arrowheads). Again, the foci were often offset from the central long axis of the cell, and the dual foci were spaced out on either side of mid-cell. The low levels of fluorescence associated with the fusions prevented us from testing for colocalization directly. However, the striking similarity of the distributions of the two fusion proteins indicate that the PolC and DnaE proteins colocalize in the replication factory.

Figure 4

Subcellular localization of PolC-GFP and DnaE-GFP in B. subtilis. Strains carrying gfpfusions to the polC (A and B) ordnaE (C and D) genes were grown in TS medium (22) to exponential phase, and samples were examined by microscopy (10). (A and C) GFP fluorescence. White lines show examples of cells with a single central focus; arrowheads point to the foci in cells with two foci. Scale bar, 2 μm. (B and D) Quantitative analysis of the positions of PolC-GFP (B) and DnaE-GFP (D) foci relative to cell length. The position of each focus was measured relative to the nearest cell pole, and the distance was expressed as a percentage of the length of the cell. Results for cells with a single focus are shown in black and results for cells with two foci are in white. Results are plotted as the percentage of total cells counted, based on measurements of (B) 65 and (D) 80 cells.

The B. subtilis replication fork thus appears to contain two different polymerases. This is similar to the situation regarding the eukaryotic replication fork, which also contains two polymerases—delta and epsilon—which have been proposed to be involved in the synthesis of different DNA strands (23, 24). However, the catalytic domain of the latter polymerase is not essential, indicating that, if missing, it can be replaced by another polymerase, whereas a noncatalytic domain has an essential structural role (25). Further work should clarify whether an analogous situation holds true for B. subtilis. It was recently reported that the E. coli holoenzyme is asymmetric, with distinguishable leading and lagging polymerases (26). The need for the difference between the leading and lagging polymerase may thus be universal, the latter being more amenable to recycling than the former (26); and only the means to achieve the difference may vary, relying either on holoenzyme properties, as in E. coli, or on two different enzymes, as appears to be the case inB. subtilis, phylogenetically related bacteria, and eukaryotes.

Our results do not rule out additional nonessential roles of DnaEBS in the cell, such as mismatch correction or DNA repair [B. subtilis mutants strongly affected in these processes are viable (27)]. In particular, theB. subtilis genome sequence indicates that DNA polymerase II is absent, and DnaEBS could possibly fulfill the role this enzyme plays in E. coli (28).

  • * To whom correspondence should be addressed. E-mail: ehrlich{at}


View Abstract

Stay Connected to Science

Navigate This Article