By continuing to browse this site you agree to us using cookies as described in About Cookies
Notice: Wiley Online Library will be unavailable on Saturday 7th Oct from 03.00 EDT / 08:00 BST / 12:30 IST / 15.00 SGT to 08.00 EDT / 13.00 BST / 17:30 IST / 20.00 SGT and Sunday 8th Oct from 03.00 EDT / 08:00 BST / 12:30 IST / 15.00 SGT to 06.00 EDT / 11.00 BST / 15:30 IST / 18.00 SGT for essential maintenance. Apologies for the inconvenience.
Escherichia coli deleted for the tus gene are viable. Thus there must be at least one other mechanism for terminating chromosome synthesis. The tus deletion strain yielded a small fraction of cells that overproduce DNA, as determined by flow cytometry after run-out chromosome replication in the presence of rifampicin and cephalexin. A plasmid, paraBAD tus+, prevented the excess DNA replication only when arabinose was added to the medium to induce the synthesis of the Tus protein. Transduction studies were done to test whether or not additional chromosomal deletions could enhance the excess chromosome replication in the tus deletion strain. A strain containing a second deletion in metE udp overproduced DNA at a high level during run-out replication. Further studies demonstrated that a spontaneous unknown mutation had occurred during the transduction. This mutation was mapped and sequenced. It is polA(G544D). Transduction of polA(G544D) alone into the tus deletion strain produced the high DNA overproduction phenotype. The polA(G544D) and six other polA alleles were then tested in wild-type and in tus deletion backgrounds. The two alleles with low levels of 5′→3′ exonuclease (exo) overproduced DNA while those with either high or normal exo overproduce much less DNA in run-out assays in the wild-type background. In contrast, all seven mutant polA alleles caused the high DNA overproduction phenotype in a tus deletion background. To explain these results we propose a new in vivo function for wild-type DNA polymerase I in chromosome termination at site(s) not yet identified.
The Tus protein/ter sites system for chromosome termination in Escherichia coli appears to be one of the most sophisticated, evolutionarily advanced, systems for terminating a circular chromosome whose terminus region is approached from both directions (see Table 1 for relevant chromosomal genes and sites)
Table 1. Map location of deletions and relevant genes.
Deletion and/or genes
Map location of deletion, gene or site (min)
The first and last gene indicated in each deletion is partially deleted. A more complete map of DE(2337) and DE(2360) is shown in Fig. 1. None of the 10 ter sites, other than terB, are deleted from any of the deletions used (Mulugu et al., 2001).
DE[(helD, 413 bp)::kan-R>]
DE(2360) DE(pntB ydgB)::kan-R
DE(2337) DE(pntB terB tus fumC)::kan-R
DE2211 DE(tar motA)
DE[(gshA, 563 bp)::kanR>]
There are 10 ter sites in two clusters of five that are opposite the origin of replication, oriC. Termination at ter sites is orientation-specific. Thus, when chromosome replication in one orientation reaches any one of a cluster of five, DNA synthesis stops while chromosome replication in the opposite direction is stopped by one of the other five ter sites. The Tus protein is required for these termination reactions (reviewed by Mulugu et al., 2001).
The Tus protein is a contahelicase (Khatri et al., 1989; Lee et al., 1989) that inhibits the DnaB helicase in the DNA polymerase III holoenzyme reaction (Mulugu et al., 2001). Deletion of the tus gene yielded not only a viable phenotype but no detectable phenotype (Morris et al., 1985), until the present study. Thus, one might have suspected that there is a redundant protein that functions by the same mechanism or other termination mechanisms remaining that are evolutionarily conserved. For example, simple collision of replication forks, without other factors, could complete chromosomal DNA sequences. This would have to be followed by homologous or dif/Xer site-specific recombination (Cornet et al., 1996; reviewed in Alexandrov et al., 1999), perhaps in the vicinity of the remaining nine (of 10) ter sites (Sharma and Hill, 1995; Mulugu et al., 2001) to reconstruct a complete circular chromosome. The dif/Xer site-specific recombination system (Table 1) is responsible for resolving normal chromosome dimers before the chromosomes are partitioned during cell division (reviewed in Alexandrov et al., 1999). Either the dif/Xis site-specific recombination system or general homologous recombination might be used with overlapping segments of chromosomes to explain the viability of the tus terB deletions (Morris et al., 1985) and the phenotype we report of excess DNA during run-out replication (see below). The results of this investigation, starting with a terB tus deletion strain (Fig. 1), revealed a new termination system. The chromosomal sites at which this termination system acts are not defined here as they may be dispersed around the chromosome and/or they may be in the E. coli terminus region. This system was identified as a spontaneous missense mutation in the polA gene and not one of the five additional deletions being tested. The polA gene specifies DNA polymerase I, a non-processive DNA polymerase, and a separate 5′→3′ exonuclease (exo) in a single polypeptide chain. DNA polymerase I functions to remove the RNA from Okazaki fragments during lagging strand replication by virtue of its 5′→3′ exonuclease activity (Kornberg and Baker, 1992). Strains defective in the 5′→3′ exonuclease accumulate Okazaki fragments and are deficient in joining those fragments to larger DNA (Konrad and Lehman, 1974). The only hint that DNA polymerase I might be involved in termination is the fact that it is a non-processive polymerase in vitro (Kornberg and Baker, 1992). The most intriguing connection found between DNA polymerase I and DNA polymerase III holoenzyme, by in vitro studies, is the following: all five of the DNA polymerases of E. coli interact with the same site on the β-loading clamp that is necessary for DNA processivity in the DNA polymerase III holoenzyme reaction (Lopez deSaro et al., 2003; Stauffer and Chazin, 2004). On the basis of the in vitro studies cited above and the results, to be presented, we propose the following: Model 1 for the in vivo mechanism of DNA termination activity of DNA polymerase I. Wild-type DNA polymerase I (200 molecules per cell) normally reduces DNA polymerase III activity (10 molecules per cell) during lagging and leading strand synthesis, by competing for the β-clamp; thereby terminating DNA synthesis via the non-processive function of wild-type DNA polymerase I. The mutant alleles specifying DNA polymerase 1 have reduced affinity and compete less well for the β-loading clamp that is required for DNA polymerase III holoenzyme processivity. The excess DNA synthesis would be greatest in the DNA polymerase I mutants deficient in 5′→3′ exonuclease activity. Thus, the mutant forms of DNA polymerase I would ‘antiterminate’ by allowing further DNA polymerase III holoenzyme processivity. Model 2 – excess DNA in the DNA polymerase I mutants is caused by increased template switching during strand displacement synthesis by DNA polymerase III holoenzyme. The excess DNA synthesis would be greatest in the DNA polymerase I mutants deficient in 5′→3′ exonuclease activity. The wild-type DNA polymerase I would inhibit (terminate) this DNA polymerase III holoenzyme reaction by the same mechanism as in Model 1, competition for the site on the β-loading clamp. The testing of these models is beyond the scope of this presentation but some new plausible model is necessary to explain the unusual finding that wild-type DNA polymerase I is, directly or indirectly, involved in termination of DNA synthesis in vivo. The present study also indirectly addresses DNA sites at which termination activity of DNA polymerase I may act. Our data, Models 1 and 2 above, and evidence from the literature do not favour DNA polymerase I action only at the normal terminus region but rather global or random termination sites. However, the answer to which DNA sites, global or the terminus region, can not be definitively answered by the technique we used, flow cytometry.
The technique of flow cytometry was used because it allows one to quantify the DNA content (fluorescence of stained DNA) and mass (light scattering) of each single cell and to record the data from 10 000 cells in a minute with the data collated on a computer screen. These parameters are displayed in dot plots and are most useful for this study. The two dimensional histograms only display the number of cells with a particular quantity of DNA, irrespective of cell size. When exponentially growing wild-type cells are treated with rifampicin (rif) and cephalexin (ceph), they complete the chromosomes that had initiated DNA at oriC but do not initiate new chromosomes at oriC. This treatment is called run-out replication. Wild-type cells yield four and eight completed chromosome equivalents when so treated in complex medium [Luria–Bertani (LB) with glucose]. Two-dimensional histograms conveniently reveal the chromosome equivalents (Boye and Lobner-Olesen, 1990; Lu et al., 1994; reviewed in Skarstad et al., 1996).
Deletions from the chromosome
Five strains with single deletions of the chromosome were available (Table 2). Two of these deletion strains are central to the present study, PK2337 [DE(2337)], and PK2360 [DE(2360)] as a control. The genes deleted and size of the two deletions are presented in Fig. 1. We note that the deletion in strain PK2337 is deleted for terB and the adjacent tus gene but importantly, it is not deleted for any of the other nine ter sites that all lie in the terminus region. (Hill, 1992; Sharma and Hill, 1995; Mulugu et al., 2001). In addition, two strains with previously unavailable deletions were constructed; AMC802 [DE gshA (563 bp)::kan-R>)] and AMC823 [DE helD (413 bp)::kan-R (Table 2 and Experimental procedures). Table 1 presents the map location of the deletions, genes and sites relevant to this study. These deletions were tested because they were deleted for putative sequences that in vitro experiments suggested might be tethered to the cell membrane by the FtsQ protein. That research was not conclusive (A. Markovitz, unpubl. results). However, the flow cytometry studies on those deletion strains, presented here, led to the finding that deletion DE2337 [DE(terB tus)] caused detectable overproduction of chromosomal DNA and all seven mutant polA alleles, in combination with DE(terB tus), caused the highest overproduction of chromosomal DNA (see below).
Table 2. Bacterial strains.
E. coli K-12
Source and construction strain
The *denotes strains made tet-S with fusaric acid.
Rac-0, the absence of the DNA of a defective prophage-like sequence, is most relevant to this study. thr-1 araC14 leuB6 DE(gpt-proA) 62 lacY1 tsx-33 qsr′-0 glnV44(AS) galK2(Oc) LAM- hisG4(Oc) rfbD1 mgl-51 rpoS396(Am) rpsL31(strR) kdgK51 xylA5
Contains plasmid p15-2
S. T. Lovett
Contains plasmid p15-7
S. T. Lovett
DE [gshA (Stu1 to EcoRV, 563 bp)]::kan-R>
DNA→V355, sel kan-R, score amp-S; see text.
DE gshA (563 bp)::kan-R> lac+Iq
P1(AMC802) X DSC8 sel kan-R
srl300::Tn10 (80% linked to gshA) DE gshA (563 bp)::kan-R>
P1(AMC908) X AMC813, sel tet-R, score UV-S and Met+
Mutations as in AMC908
P1(AMC908) X AMC813, sel tet-R, score UV-S and Met–
polA5 DE(metE udp) tet-R
P1(AMC816) X CM4050, sel tet-R, score Met– and UV-S.
polA5 tet-R DE(terB tus)::kan-R
P1(AMC911) X AMC813, sel tet-R, score Met+ and UV-S
polA5 DE(metE udp) tet-R DE(terB tus)::kan-R
P1(AMC911) X AMC813, sel tetR, score Met– and UV-S
DE(metE udp) tet-R DE(terB tus)::kan-R
P1AMC816) X AMC813, sel tet-R, score Met– and UV-R
polA20 DE(metE udp) tet-R
P1(AMC816) X AB3027, sel tet-R, score UV-S and Met–
polA20 tet-R DE(terB tus)::kan-R
P1(AMC915) X AMC813, sel tet-R, score UV-S and Met+
polA591 DE(metE udp) tet-R
P1(AMC816) X S519, sel tet-R, score UV-S Met–
polA591 tet-R DE(terB tus)::kan-R
P1(AMC917) X AMC813, sel tet-R, score UV-S Met+
polA+ tet-R DE(terB tus)::kan-R
P1(AMC905) X AMC813, sel tet-R, score UV-R
polA12 (ts) tet-R DE(terB tus)::kan-R
P1(AMC905) X AMC813, sel tet-R, score UV-S
polA480 (ts exo) tet-R DE(terB tus)::kan-R
P1(AMC906) X AMC813, sel tet-R, score UV-S
polA546 (ts exo) tet-R DE(terB tus)::kan-R
P1(AMC907) X AMC813, sel tet-R, score UV-S
Preparation of strains with multiple chromosomal deletions by P1 transduction
All of the single deletion strains, as well as all multiple deletion strains, were viable on initial transductions (Table 3) with one exception. Strain AM881 [DE(metE udp)] could not be transduced to kan-R using P1 grown on PK2337 [DE(terB tus)::kan-R] in three separate experiments. The transduction frequency from P1 grown on PK2337 was low, even with control recipient strain AB1157 (12–20 kanR transductants per plate versus 300–500 for other markers with the same P1 lysate as a control on the P1). Strain PK2337 has a deletion of 11 098 bp (Fig. 1; Experimental procedures). Thus, the low frequency of kan-R transduction, compared to other markers with the same lysate, is attributed to the non-homology region of 11.1 kb in addition to the kan-R insert. Of course, this result focused our attention on whether or not the double deletion was viable. However, the DE(terB tus) DE(metE udp) strain was easily produced by P1 transduction in the reverse order (Table 3, strains AMC814 and AMC817). Since we later found the spontaneous polA(G544D) mutation in AMC814 (and AMC817), we tested whether both the DE(metE udp) and polA(G544D) might be necessary for viability in the DE(terB tus) background. This hypothesis was disproved when the polA(G544D) and DE(metE udp) were separated in a DE(terB tus) background (Table 2; strains AMC909, AMC910, AMC914). Table 3 outlines the construction of all the multiple deletion strains with up to five different deletions in one strain. All of these strains were tested in run-out replication experiments. None of the deletions, except DE(terB tus), contributed to overreplication during run-out replication. Data to establish this statement are provided in the flow cytometry experiments below.
Table 3. Construction of multiple deletion strains by P1 transduction.
New strain and deletions
Strains with an asterisk (*) were selected for loss of tetracycline resistance using fusaric acid (Bochner et al., 1980) before the transduction indicated. All multiple deletion strains not mentioned in the text were tested by flow cytometry after run-out replication and did not overproduce DNA.
AMC808; contains a deletion but not (terB tus)::kan-R and rac
AMC809; gshA (563 bp)::kan-R and (cheA motB motA)
AMC810; (terB tus)::kan-R and (cheA motB motA)
AMC811; gshA::kan-R and (metE udp)
AMC812; gshA::kan-R and rac
AMC813; (terB tus)::kan-R and rac
AMC814; (metE udp) and (terB tus). Discovered polA(G544D)
AMC815; (metE udp)
AMC816; (metE udp) and rac
AMC817: (metE udp), (terB tus) and rac. Discovered polA(G544D)
AMC829; DE(helD, 413 bp)::kan-R zbc-222::Tn10 and (terB tus) and (metE udp) and (cheA motB motA) and rac. Select Tet-R and score DE(helD, 413 bp)::kan-R by PCR.
Flow cytometry studies on strains with single and multiple chromosomal deletions
Exponentially aerobically growing cells in LB/glucose were treated with rif and ceph (R/C). Aeration was continued at 37°C for 4 h to allow completion of chromosomes that had already initiated at oriC. This procedure is called run-out replication. The cells were prepared and analysed as described (Experimental procedures). Flow cytometry data on derivatives of strain AB1157 (deletion of the defective prophage rac) with 1–4 additional chromosomal deletions are provided in Fig. 2. The data are displayed as DNA histograms and dot plots. The DNA of wild-type strains contains mostly four and eight DNA genome equivalents after run-out replication in complex medium (Lu et al., 1994) as indicated in each figure containing histograms. Each dot in the dot plot measured the DNA content and the mass of that cell. It is important to mention what appears to be a variability in the abscissa of histograms and the ordinate of dot plots of the wild-type between experiments (Figs 2–7). This is caused by the adjustment of the electronics in each experiment. While the scale for DNA (abscissa of histograms and ordinate of dot plots) always reads from 0 to 1000 the two peaks of the histograms can be electronically moved in either direction. Therefore, a wild-type strain must be used for each experiment and comparisons between experiments would be an approximation. The results revealed that deletion of gshA (Fig. 2, 1R/C) produced a DNA histogram and a dot plot identical with an AB1157 control (not shown), but DE(terB tus) produced some cells with detectable quantities of excess DNA(2R/C). Most important, simultaneous deletion of terB tus and metE udp (AMC817) caused a much larger part of the cell population to produce excess DNA and the cell mass of those cells was not larger than the others (3R/C versus 1R/C and 2R/C, dot plots). Thus, cells with a high DNA/mass ratio were produced by AMC817 without significant changes in cell mass. Additional deletion of motA-tar (4R/C) into AMC817 to yield strain AMC820 resulted in somewhat higher DNA/mass ratios than AMC817 but in all later experiments the overproduction of DNA was equivalent between AMC817, AMC820 and AMC819 (AMC819 is equivalent to AMC820 but deletions were added in a different sequence; Table 3) (flow cytometry data not shown). A fifth deletion, gshA into AMC820 (Fig. 1, 5R/C), yielded results similar to those with AMC817 and these were reproduced in other experiments (data not shown). The dot plots made it clear that filaments were not a part of the population producing excess DNA because the masses of AMC817 cells were equivalent to the masses of control cells AMC812 (Fig. 2). The double deletion of terB tus and metE udp, without the rac deletion (strain AMC814), yielded the same results as strain AMC817 with the rac deletion (data not presented). All the run-out replication experiments with multiple deletion strains constructed (Table 3) showed that only strain AMC817 [DE(terB tus) DE(metE udp)], and its derivatives with up to five deletions (all strains in Table 3 were tested) overproduced high levels of excess DNA after run-out replication. The only single deletion strain that overproduced chromosomal DNA after run-out replication contained DE(terB tus).
The cells with a high DNA/mass ratio were produced in strain AMC817 in rif within 2 or 3 h, even in the absence of ceph. Results similar to those with AMC817 were obtained with the additional multiple deletion strains derived from strain AMC817 (data not shown). Thus, a 4 h period is sufficient for all detectable DNA synthesis to be completed.
Flow cytometry experiments in minimal medium M9 plus glucose provide no evidence for non-synchronized oriC replication (Lu et al., 1994) in strain AMC817
The excess DNA apparent in DNA histograms, before or after rif/ceph in multiple deletion strains, was not present in discrete new peaks in experiments in LB + glucose (Fig. 2). Such new DNA peaks in histograms provide evidence for non-synchronized initiations from oriC. However, in other laboratories such discrete peaks were best observed in strains growing in minimal medium (Lu et al., 1994). However, our run-out replication experiments in minimal medium provided no indication that increased unscheduled initiations from oriC occurred (data not presented). Therefore, increased initiations from oriC do not explain the increased DNA/cell.
The effect of a plasmid containing the tus+ gene (pBADtus +) on deletion strains: pBR322 origin plasmids are unstable in double deletion strain AMC817; a new phenotype
It was of interest to determine: (i) Would supplying the Tus protein in trans correct the small overproduction of DNA caused by DE(terB tus) (strain AMC813)? and (ii) Would supplying the Tus protein in trans correct the large overproduction of DNA in the double deletion strain AMC817? Plasmid pBAD18tus+ contained tus+ (Sharma and Hill, 1995) tightly controlled by the BAD promoter that was repressed (glucose)/activated (arabinose) by araC+ gene protein. The araC+ and carbenicillin resistance genes are also on the plasmid (Guzman et al., 1995). The E. coli genes present in pBAD18tus+ are indicated in Fig. 1.
(i) The results demonstrated that plasmid pBAD18tus+ prevented the overproduction of DNA in strain AMC813, but only in the presence of arabinose (Fig. 3, 3R/C versus 2R/C and 1R/C). Thus, the cells with excess DNA resulted from the absence of the Tus protein and not the absence of other genes deleted from that region of the chromosome. This is the first evidence that deletion of the tus gene produces a phenotype that can be reversed by provision of the Tus protein in trans.
(ii) The same experiment was attempted with AMC817 selected to contain pBAD18tus+ using carbenicillin. However, the plasmid could not be established in strain AMC817 although initial transformation plates with carbenicillin yielded colonies. In contrast, the plasmid was completely stable in strain AMC813 under the same conditions (see Experimental procedures for tests of plasmid instability).
Strain AMC817 is UV-sensitive
After it was found that pBR322 origin plasmids were unstable in strain AMC817, we tested and discovered that strains deleted for both (metE udp) and (terB tus) (i.e. AMC817 and AMC814) were UV-sensitive. All of the strains from which they were derived (Tables 2 and 3) were UV-resistant. Three hypotheses were considered: (i) strain AMC814 acquired a spontaneously derived polA mutation during transduction and it was transduced from AMC814 to AMC817 (Table 3); (ii) they acquired an unknown spontaneous mutation; and (iii) the double deletion alone caused these new phenotypes.
Studies with oriC, oriF and oriP1 plasmids
Table 4presents the results of stability studies with oriC and oriF plasmids, both with and without sopABC that provide partition factors that promote plasmid stability in wild-type bacteria (Niki and Hiraga, 1999). The oriC plasmid with the sopABC genes (pXX199) is unstable in strains AMC817* (the asterisk denotes strains made tet-S with fusaric acid, Table 3) and AMC819, but stable in the strains with single deletions. The oriF plasmid with sopABC (pXX704, Niki and Hiraga, 1999) is stable in all strains (Table 4) as is the oriP1 plasmid containing its partitioning factors parABS (pLG44; Erdmann et al., 1999; Surtees and Funnell, 2003; data not shown). The phenotype with the oriC plasmid pXX199 has not been reported for polA mutant strains. The pXX199 plasmid data led to an approach which distinguished between the three hypotheses.
Table 4. Percentage of cells that retain plasmids after 20 cell divisions in the absence of selection.
After transformation with the indicated plasmid procedures were as described in Experimental procedures. After 20 cell divisions without selection bacteria were plated on LB agar. After 24 h at 37°C they were replica plated onto LB agar with and without carbenicillin and counted after 18 h at 37°C. Similar results were obtained by plating directly on LB agar with carbenicillin and comparing the survivors with those obtained by plating directly on LB agar (results not shown). Strains with an asterisk were made tet-S using fusaric acid (Bochner et al., 1980). ND, not done.
The plasmids are (Niki and Hiraga, 1999): oriC+, oriC+ sopABC (pXX199); oriC–, oriC without sopABC (pXX230); oriF+, oriF+ sopABC (pXX704); OriF–, oriF without sopABC (pXX705).
Detection of a spontaneous chromosomal mutation in strain AMC817
It was suspected that strain AMC817 [DE(metE udp) and DE(terB tus)] contained a spontaneous mutation that caused the instability of plasmids with pBR322- type and oriC- type origins and UV sensitivity. A series of Hfr donors encompassing the entire chromosome (CGSC collection) were transformed with pBR322 and each was crossed with strain AMC817. Selection for carb-R (pBR322), streptomycin-R and kan-R (AMC817) yielded some colonies from which pBR322 was isolated. Some of these colonies also required methionine indicating they were DE(metE udp). Since kan-R was inserted in the terB-tus deletion, the results established that there was an unknown mutation that had been removed from AMC817 without removing either of the two deletions. Thus, the two deletions did not cause the pBR322-unstable phenotype by themselves.
Transposon mapping and DNA sequencing of the spontaneous mutation as polA(G544D)
Results of transposon mapping and DNA sequencing established that the spontaneous mutation was in fact a mutation in the polA gene that resulted in replacement of glycine 544 by an apsartate (Experimental procedures). Glycine 544 is highly conserved among the DNA polymerase I proteins in the bacterial data bank. G544 is localized between two α-helices.
What mutation/deletions are responsible for the overproduction of DNA during run-out replication?
Flow cytometry studies demonstrated that polA(G544D), in a wild-type background, caused little increased (just detectable) DNA/cell (Fig. 4). In addition, six polA mutant alleles (Table 5) were studied, first in the wild-type background. The most striking finding was that two polA alleles that were deficient in the 5′→3′ exonuclease overproduced the most DNA during run-out replication [(Figs 4 and 5A, i.e. polA480 (ts exo) and polA546 (ts exo) previously known as polAex1 and polex2 respectively)]. In addition, the polA5 allele, that has a high level of 5′→3′ exonuclease (Table 5), as well as the remaining three polA alleles, produced cells with little increased (just detectable) DNA/cell after run-out replication (Figs 4 and 5A).
Table 5. polA alleles used and their enzymatic defects.
Strain of origin, comments
All the above strains with polA mutant alleles were sensitive to UV and transductions were monitored by measuring sensitivity to UV.
AMC817; this study
Higher than normal
Altered pH opt. and template specificity, lower activity. Processitivity low.
In the DE(terB tus) background all seven polA alleles caused the highest overproduction of DNA in run-out replication (Figs 5B, 6 and 7). DE(met-udp) was not required for the highest run-out replication (Figs 6 and 7). The significance of these results is discussed below.
Flow cytometry results revealed that cells with either of two polA alleles known to have low 5′→3′ exonuclease in vitro contained a cell population that overproduced DNA during run-out replication in rifampicin and cephalexin in a wild-type background. The other five polA alleles overproduced much less DNA during run-out replication in a wild-type background (Figs 4 and 5A). The strain with a tus-terB deletion also detectably overproduced DNA during run-out replication. Provision of the Tus protein in trans eliminated the overproduction of DNA (Fig. 1, map; Fig. 3], indicating the overproduction was due to the absence of Tus protein and not to the deletion of the other contiguous genes from the chromosome. Tus protein prevented overreplication from an oriC plasmid with a properly positioned ter site in an in vitro system (Hiasa and Marians, 1994). Our results are consistent with their findings and extend them to the chromosome in vivo. The results with the tus-terB deletion alone may be interpreted as readthrough the terminus region as the cause of excess DNA during run-out replication although direct evidence for this can not be obtained from flow cytometry studies. Collision of the chromosomes replicating through the remaining nine ter sites, in the absence of Tus protein, and recombination may explain the viability of such strains, as indicated in Introduction. The polA strains (seven mutant alleles) with the tus-terB deletion all overproduced DNA more than either of the single mutant/deletion strains.
The excess DNA as a result of either of two polA exo mutations alone, or all seven polA alleles in combination with the tus-terB deletion, is, in fact, surprising and unexpected. More so, because polA mutants defective in DNA polymerizing activity, as well as those with either increased or decreased 5′→3′ exonuclease activity (Table 5), caused increased DNA synthesis during run-out replication in a terB tus background (Fig. 5B, 6 and 7). Thus, wild-type DNA polymerase I is, directly or indirectly, important in chromosome termination in vivo, although the site(s) of termination caused by polA alleles are more likely to be dispersed rather than near the terminus of normal chromosome replication (see below).
The first question: what mechanisms could explain how DNA polymerase I contributes to chromosome termination, regardless of the site(s) on the chromosome at which it operates? We noted in Introduction that DNA polymerase I is non-processive (Kornberg and Baker, 1992) and that DNA polymerase I and DNA polymerase III holoenzyme compete for the same site on the β-loading clamp that is required for processivity of the DNA polymerase III holoenzyme reaction (Lopez deSaro et al., 2003; Stauffer and Chazin, 2004). Two models were proposed to answer this question in Introduction. We restate the models, both of which can apply to termination at dispersed sites around the chromosome. Model 1: Wild type DNA polymerase I (200 molecules per cell) normally reduces DNA polymerase III activity (10 molecules per cell) during lagging and leading strand synthesis, by competing for the β-clamp that is necessary for processivity, thereby terminating DNA synthesis via the non-processive function of wild-type DNA polymerase I. The mutant alleles specifying DNA polymerase I have reduced affinity and compete less well for the β-loading clamp. The excess DNA synthesis would be greatest in the DNA polymerase I mutants deficient in 5′→3′ exonuclease activity, as observed in Figs 4 and 5A. Thus, the mutant forms of DNA polymerase I would ‘antiterminate’ by allowing further DNA polymerase III holoenzyme processivity. Model 2: Excess DNA in the DNA polymerase I mutants is caused by increased template switching during strand displacement synthesis by DNA polymerase III holoenzyme. The excess DNA synthesis would be greatest in the DNA polymerase I mutants deficient in 5′→3′ exonuclease activity (Figs 4 and 5A). The wild-type DNA polymerase I would inhibit (terminate) this DNA polymerase III holoenzyme reaction by the same mechanism as in Model 1, competition for the same site on the β-loading clamp.
The second question: are the sites where wild-type DNA polymerase I promotes termination dispersed around the chromosome or are they localized, for example, to the terminus region? Flow cytometry experiments done here, as well as a survey of the literature, do not answer this question unequivocally. Proof of where the DNA polymerase I mutants are causing increased DNA synthesis may be obtained, in the future, by comparing the terminus DNA concentration with other chromosomal regions, after DNA run-out replication and cell sorting of the fraction of cells overproducing chromosomal DNA. Recent technological developments (Breier and Cozzarelli, 2004; Viollier et al., 2004), including microarrays (Simmons et al., 2004), make such experiments feasible. However, our study and several in the literature make certain sites and mechanisms very unlikely and worthy of discussion here as we reach the tentative conclusion that dispersed sites are most likely. (i) Increased unscheduled initiations at oriC might have explained excess DNA synthesis but the following argue against oriC. Experiments on run-out replication in minimal medium (Lu et al., 1994) were designed to detect unscheduled initiations at oriC, but no evidence for such initiations were found in our experiments (see Results). Studies on plasmid stability are consistent with this conclusion. An oriC plasmid with the sopABC partitioning elements was unstable in strains with DE(terB tus) and polA(G544D) (strains AMC817*, AMC819), but an oriF plasmid with the same sopABC partitioning elements was stable (Table 4). Thus, the oriC plasmid is not well replicated and partitioned in strains AMC817*/AMC819 but oriF and oriP1, as positive controls, are. (ii) Replication fork collapse occurs randomly in normal cells (reviewed in Simmons et al., 2004). The wild-type DNA polymerase might participate in fork collapse and the mutant forms may participate less well according to Models 1 and 2 above. (iii) An inhibitory role in restart of DNA synthesis for wild-type DNA polymerase I has not been ruled out. However, extensive studies of restart in vitro did not implicate DNA polymerase I as a participant, but did include pol II, pol III and pol V as well as other important factors (Rangarajan et al., 1999; 2002; Sandler, 2000; Sandler and Marians, 2000). Nevertheless, it is logically possible that wild-type DNA polymerase I inhibits restart in vivo. From the perspective of the present study, the effect of wild-type and mutant DNA polymerase I might be tested on an in vitro restart system (Rangarajan et al., 1999; 2002; Sandler, 2000; Sandler and Marians, 2000). (iv) DNA polymerase I is required for cSDR (constitutive stable DNA replication) and iSDR (induced stable DNA replication) and both processes occur in rifampicin. Both of these processes start at origins other than oriC, but since they require DNA polymerase 1 (Kogoma, 1997; Kogoma and Maldonado, 1997), and loss of DNA polymerase 1 activity is required for increased run-out DNA replication, cSDR and iSDR can not explain the present results. (v) The normal termination region. In the absence of the Tus protein, all seven polA mutant alleles caused overproduction of DNA but, in the presence of Tus protein, the two polA alleles deficient in 5′→3′ exoniclease activity overproduced DNA more than did the five other polA alleles (Figs 4–7). These data are consistent with polA alleles defective in 5′→3′ exonuclease being least influenced by the loss of Tus protein. Therefore, their ‘antitermination’ activity may be dispersive, or occur at sites other than the terminus region. The other five polA alleles overproduce DNA dramatically only in a terB tus deletion background. This result can be interpreted in two ways: (i) the overproduction occurs specifically after the terminus region (that is why it is dramatic when the Tus protein is absent); (ii) the system is under great stress and most of the termination due to the five polA alleles is dispersed around the chromosome while a small amount is after the terminus region, similar to the single terB tus deletion. The five mechanisms of termination discussed above lead us to favour a random or dispersed mode of termination for wild-type DNA polymerase I and there appears to be little reason to believe that the normal termination region is preferred. Regardless of by what mechanism and where wild-type DNA polymerase I causes chromosome termination, the present study makes it clear that DNA polymerase I has a previously unrecognized function in vivo. Only direct experiments, as suggested above, will decide whether DNA polymerase I termination is global or localized to the terminus region.
In eukaryotes, DNA polymerase σ is essential for the establishment of sister chromatid cohesion (Wang et al., 2002), but the reason for this requirement has not been obvious. The authors have proposed a switch model in which replicative DNA polymerases are replaced by DNA polymerase σ where the replicating fork approaches a precohesion site (reviewed in Wang et al., 2002). In view of our findings, it is tempting to speculate that such a switch is required to terminate or pause DNA replication at the site where sister chromatid cohesion is initiated or established. For this idea to have merit DNA polymerase σ would have to be non-processive, if the mechanism is similar to Models 1 or 2 above.
Escherichia coli K12 strains and their derivation are listed in Table 2 and strains constructed with multiple deletions are presented in Table 3. Selection for the appropriate antibiotic and testing of the recipients for the linked deletion or mutation was done using appropriate phenotypes [Met– for DE(metE udp), UV sensitivity for polA, kanamycin (50 µg ml−1) for the deletions/insertions in strains such as PK2337 and PK2360 (Morris et al., 1985), loss of motility for DE(motA-tar) DE2211 (Parkinson and Houts, 1982)] and polymerase chain reaction (PCR). Strains that were tet-R(10 µg ml−1) were made tet-S with fusaric acid as described(Bochner et al., 1980).
Strains with deletions and DNA sequencing
Strains PK2337 and PK2360 (Fig. 1, Tables 1 and 2). Strain PK2337 is deleted for at least 4 kb including the terB and tus genes, as determined by Northern hybridization, and kan-R is inserted in the site of the deletion. PK2360 starts at the same left end as PK2337 but ends before the terB tus loci and also has kan-R inserted (P.L. Kuempel, pers. comm.; Morris et al., 1985). PCR studies (see below) and DNA sequencing in this laboratory confirmed that tus and terB were deleted from the chromosome of strain PK2337 and were present in PK2360. The deletions from both PK2337 and PK2360 started from the same site at 10 151 bp of AE000255 (12 965 bp total) and the deletion in PK2360 ended at 3345 bp of AE000256. Thus, the deletion in PK2360 was 6260 bp. The deletion in PK2337 ended 8263 bp into AE000256 making the deletion 11 098 bp. The tus gene is from 6201 to 7130 bp of AE000256. The terB site is included 33 bp before tus (Fig. 1).
Strain AM881 (Table 2) is deleted for the distal part of metE, continues through all of ysgA and finishes in the proximal part of udp (A. Mironov, pers. comm.). DNA sequencing in this laboratory revealed that the deletion occurred between 10 bp direct repeats (CAGGAACGTT) on the E. coli chromosome. When AE000458 and AE000459 are combined (22566 bp total), the 10 bp is at 9664–9673 and 12 172–12 181. The chromosome retains one of the 10 bp repeats and the deletion is 2508 bp.
Strain AB1157 (CGSC1157) is deleted for the defective prophage, rac.
Further verification of deletions from strains AM881, PK2337, PK2360 and RP1571 using PCR
The deletions from strains AM881, PK2337, PK2360 and RP1571 were transduced into strain AB1157 and derivatives so that, for example, strain AMC819 contained the deletions from strains AM881, PK2337 and RP1571 (Table 3). Strain AB1157 [DE(rac)] contained the wild-type sequence of the deletions being tested and was a control in the PCR studies. One 33 mer primer was expected to be internal to the deletion and the other primer was located outside the deletion, as proved by PCR. None of the DNA from the deletion strains yielded a PCR product under these conditions [except the deletion from PK2360 (as expected)], whereas the AB1157 DNA yielded appropriate sized products. The PCR results with a derivative of PK2360 (AMC808) proved that the terB tus loci were present and that those loci were missing from derivatives of PK2337 tested (AMC813, AMC819). DNA sequencing (see above) for the deletions in PK2337, PK2360 and AM881 completed the proof of deletions.
Plasmids used were (i) pBR322, (ii) pBADtus+ (Fig. 1) [contains a 2.7 kb EcoR1-HindIII fragment that, reading from the prmBAD promoter, specifies part of rstB and all of the terB site as well as tus+ and part of fumC (gift of T.M. Hill; Rudd, 1998; Fig. 1; Table 1). Plasmid pBADtus+ also contains araC+ and carbenicillin resistance (Guzman et al., 1995), (iii) pBluescript KS II+ (Stratagene), and (iv) pGEMR-3Z (Promega). Plasmids in (i) to (iv) have a pBR322- type origin. Other plasmids with oriC, oriF and oriP1 are described in Table 4 and in Results.
Purification of plasmids, bacteriophage lambda, lambda DNA, chromosomal DNA, DNA from agarose, DNA from PCR and from restriction reactions used the appropriate Qiagen Kit. DNA concentrations were determined with DNA DipstickTM (Invitrogen). Other procedures were from Ausubel et al. (1987 and supplements).
DNA sequencing procedure
DNA sequences were determined using the Dye-dideoxy method with a Taq kit (Perkin–Elmer), the automated sequencer (Applied Biosystems, Model 373A) and EditView software (ABI) for examination of the data.
Test for plasmid instability with pBR322 origin plasmids and oriC, oriF and oriP1 plasmids in strain AMC817 [DE(terB tus) DE(metE udp) and later discovered polA(G544D)]
Plasmids were transformed into the indicated strains and selected on LB agar containing either carbenicillin (50 µg ml−1; all plasmids except pLG44) or chloramphenicol (25 µg ml−1 for oriP1 plasmid pLG44; Erdmann et al., 1999). An isolated colony was streaked two times for clone purification on the same selective medium. A small sample from a clone was suspended in sterile water and 5 µl was inoculated into 2 ml of LB broth +0.2% glucose (without antibiotic). After overnight growth, 1 µl was used as an inoculum for a second consecutive tube of the broth representing a total of at least 20 cell divisions. After the second overnight growth, dilutions were plated on LB agar plates with and without antibiotic. Colonies on LB agar were always further tested by replic plating on plates with antibiotic. Plasmid instability in strain AMC817 was proved, in some cases (pBR322, pBADtus+) by the complete absence of colonies on both the initial carbenicillin plate and the replica plate from LB agar to LB agar plus carbenicillin. With pBADtus+ in AMC817, cells grown in LB + arabinose as well as LB + glucose yielded a few colonies resistant to carbenicillin in some experiments. Also, among combinations of plasmids and strain AMC817 designated as unstable were some in which 1–100% of colonies were identified as carbenicillin-resistant. However, attempts to isolate the plasmid from selected clones were negative by both gel electrophoresis and the more sensitive transformation procedure. Plasmid instability in strain AMC817 was considered to be proved by the inability to isolate plasmid from selected carbenicillin resistant clones. This category included plasmids pBADtus+ and pBluescriptKSII. Plasmid stability was defined as 99–100% antibiotic-resistant colonies in the same tests used above including the isolation of plasmid DNA. Plasmids pBR322 and pBADtus+ were stable in strains AMC813 [DE(terB tus) and AMC816 [DE(metE udp)] by all the criteria.
Preparation of strains that contain a 563 bp deletion within the gshA gene (Table 2)
Lambda/K446 of the Kohara collection contains the complete gshA gene. (Kohara et al., 1987; Rudd, 1998). One 3.2 kb Pst1-PvuII fragment of lambda/K446, containing gshA+ and adjacent sequences, was ligated to Pst1-Sma1-digested pGEM3Z. Transformants in XL1-blue were obtained containing the appropriate plasmid p33-3. A 563 bp deletion/Kan-R cassette substitution within the gshA gene of p33-3 was obtained using EcoRV and Stu1 and standard techniques. The resulting plasmid, p39, was linearized with Kpn1, and transformed into strain V355 (recD) on media supplemented with 50 µg ml−1 glutathione and 50 µg ml−1 kanamycin to yield AMC802. Finally, kan-R from AMMC802 was transduced into strain DSC8 to produce AMC805 or into AB1157 to yield AMC812 (Table 2). These strains grew normally in liquid or solidified M9-minimal glucose or LB medium with or without glutathione. However, phase microscopy indicated that strain AMC805 grew twice as long in the absence of glutathione. The chromosomal deletion in strain AMC805 was proved by PCR and DNA sequencing of both chromosomal/kan-R joints.
Construction of a plasmid deleted for a 413 bp sequence upstream of helD
Plasmids with a 5 kb BamH1 fragment that contained the complete helD sequence as well as surrounding DNA inserted into the BamH1 site of pBluescript KSII+ were provided by Sue T. Lovett. One, phelD-1, also contained a miniTn10kan-R insert after E. coli bp 1023580 Blattner Number (BN) (Blattner et al., 1997), and the other, phelD-15, the same insert in the same orientation but at approximately 0.5 kb upstream (S.T. Lovett, pers. comm.). Thus, we were able to prepare a plasmid with a 413 bp deletion/kan-R insertion upstream of the helD gene by in vitro recombination in the kan-R gene (strategy of S.T. Lovett). A plasmid, pdelhel, was obtained. DNA sequencing of pdelhel from both directions identified one joint of helD–miniTn10kan (bp 1023580 BN) and the other joint was 413 bp upstream of helD after bp 1023165 BN. Thus, the appropriate deletion/kan-R insertion plasmid was available to delete 413 bp of chromosomal DNA just upstream of helD.
Deletion from the chromosome of a 413 bp helD sequence
Plasmid pdelhel was linearized into two fragments with Pst1 and transformed into strain V355 (recD) with selection for kan-R and scoring for carb-S. Strain AMC823, a presumptive 413 bp deletion in helD, was used to grow P1 with strain AB1157 as a recipient to produce strain AMC824 (Table 2). Primers in helD were selected to amplify the chromosome of the strain directly from chromosomal DNA preparations. DNA sequencing of the PCR products of the two joints established that the chromosomes of AMC823 and AMC824 were deleted for the 413 bp.
Transposition mapping and DNA sequencing of the spontaneous mutation in strain AMC817 as polA(G544D)
Transposition was essentially as described (Way et al., 1984; Kleckner et al., 1991). LambdaNK1316, containing element 103 (miniTn10kan; Kleckner et al., 1991), was used with a wild-type strain DSC8. Twenty thousand kan-R single colonies were selected and pooled. Phage P1 was grown on the kan-R pool and used to transduce AMC851 (an AMC817 derivative, Table 2), selecting for kan-R. A pool of kan-R AMC851 was transformed with pBR322 with selection for carb-R. Colonies that were carb-R and contained pBR322 were presumed to contain kan-R linked to the wild-type of the spontaneous mutation locus. P1 was grown on a single clone of AMC851 kan-R/pBR322, designated AMC852. P1 grown on AMC852 was used to transduce AMC851 and selection for kan-R yielded 80% UV-R colonies; whereas, the recipient AMC851, was UV-S. Thus, kan-R was tightly linked to the spontaneous mutant locus.
Inverse PCR was used to identify the insertion joints of miniTn10kan, essentially as described using the same primers in PCR and initial DNA sequencing reactions (Genevaux et al., 1999). Sequencing of strain AMC852 demonstrated that miniTn10kan had been inserted in the mobB gene that is tightly linked to polA. Further sequencing of the chromosomal polA gene in strain AMC817, in both directions, identified a single base change that leads to a change of glycine to aspartate at amino acid 544. Glycine 544 is highly conserved among the bacteria and is located between two α-helices.
Cell growth, treatment and preparation for flow cytometry
Bacteria were grown in LB broth plus 0.2% glucose or, in one experiment, minimal medium M9 plus 0.2% glucose (Miller, 1972) without CaCl2 and supplemented with the required amino acids (50 µg ml−1) and thiamin (10 µg ml−1). Procedures were generally as described (Skarstad et al., 1996). Cells from aerobic overnight cultures at 37°C were diluted 1–100 or 200 in the same medium. The same conditions of growth were continued. At approximately 108 cells ml−1, samples were removed and rifampicin (rif) and cephalexin (ceph) were added to the remaining cells to final concentrations of 300 µg ml−1 and 12 µg ml−1. Aeration was continued as before for 4 h to allow completion of chromosomes that had already initiated their replication at oriC. All samples were examined by flow cytometry although the data reported exclude the samples from exponentially growing cells because they were not informative. Exceptions to these conditions are noted. Rif blocks initiation of chromosome replication from oriC and ceph blocks further cell division that may result in an underestimation of the number of origins per cell (Skarstad et al., 1996). All culture samples were centrifuged at 16 000 g for 1 min and resuspended in ice-cold 10 mM Tris (pH 8) followed by the addition of ice-cold ethanol to a final concentration of 70%. Samples in 70% ethanol were stored at 4°C until they were prepared for flow cytometry. The cells were pelleted by centrifugation at 16 000 g for 45 min at 4°C, washed in 10 mM Tris (pH 8)/10 mM MgCl (buffer A) by centrifuging as before and resuspended in buffer A. Cells were diluted with buffer A containing propidium iodide at a final concentration of 50 µg ml−1. The samples were kept at 0°C in the dark for 1–2 h before flow cytometer analysis. Cells were examined in a FACScan instrument (Becton–Dickinson) with excitation at 488 nm and emission at 585 nm (fluorescence) for DNA stained with propidium iodide. Data were collected and analysed using Cellquest software for the MacIntosh. The DNA histograms and dot plots were transferred to Photoshop for lettering and boxing of dot plots, as indicated in the figures.
The author thanks the following, without whom this work would not have been completed: Alexander Miranov for the DE(metE udp) strain AM881, and detailed communications; Peter Kuempel for the DE(terB tus::kan-R) strain PK2337 and strain PK2360 and detailed communications; John S. Parkinson for DE(motA-tar) DE2211 and information; Thomas M. Hill for plasmid pBADtus+ Sue T. Lovett for plasmids and a specific idea that permitted specific deletion of the upstream region of helD.
Thanks also to Barbara E. Funnell, Soto Hiraga and Hironoru Niki for plasmids.
The DNA requencing facilities and the Flow cytometetry laboratory of the University of Chicago provided assistance. Deepak Bastia, Nicholas Cozzarelli and Bernard Strauss read the first versiton of this paper and suggested changes and references. In particular, Bernard Strauss suggested that our results implicating DNA polymerase I in termination in E. coli might explain the need for eukaryotic DNA polymerases for sister chromatid cohesion. We thank the unknown referees of our manuscript for their suggested expansion of the manuscript and pertinent questions that helped improve the manuscript.