SeqA structures behind Escherichia coli replication forks affect replication elongation and restart mechanisms.

The SeqA protein binds hemi-methylated GATC sites and forms structures that sequester newly replicated origins and trail the replication forks. Cells that lack SeqA display signs of replication fork disintegration. The broken forks could arise because of over-initiation (the launching of too many forks) or lack of dynamic SeqA structures trailing the forks. To confirm one or both of these possible mechanisms, we compared two seqA mutants with the oriCm3 mutant. The oriCm3 mutant over-initiates because of a lack of origin sequestration but has wild-type SeqA protein. Cells with nonfunctional SeqA, but not oriCm3 mutant cells, had problems with replication elongation, were highly dependent on homologous recombination, and exhibited extensive chromosome fragmentation. The results indicate that replication forks frequently break in the absence of SeqA function and that the broken forks are rescued by homologous recombination. We suggest that SeqA may act in two ways to stabilize replication forks: (i) by enabling vital replication fork repair and restarting reactions and (ii) by preventing replication fork rear-end collisions.

INTRODUCTION

The single chromosome of Escherichia coli (E. coli) is replicated bi-directionally by a pair of replisomes moving in opposite directions from the origin (1). During this process, the replication forks can stall, disintegrate, or collapse if they encounter DNA damage, DNA secondary structures or tightly bound proteins (1,2). Restart mechanisms can act on a stalled fork (2–6), but a paused replisome may last a few minutes at most before falling apart (7). If replication forks disintegrate, the resulting double strand breaks (DSBs) or double strand ends (DSEs) can lead to chromosomal fragmentation (8,9). Restart and repair can depend on the recombinase RecA and involve homologous recombination (Figure 1C, ii), or be RecA independent, with digestion of the DSE by RecBCD and a direct restart (Figure 1C, iii).

Figure 1.

Examples of causes of DSEs at replication forks and models for repair and restart of replication forks. (A, i) A nick in the template DNA may cause a DSE when the replisome runs into it and (ii) rear-ending causes two DSEs when the replisomes replicate the final nucleotides of the template DNA that was replicated in the previous round. (B) DSEs are repaired by homologous recombination. RecBCD recognizes the DSE and unwinds and degrades the DNA up to the nearest Chi site (10,11). Next, RecA forms a filament with single-stranded DNA from the processed DSE, invades the intact homologous DNA, and performs recombination. The resulting Holliday junction is resolved by RuvABC, and PriA can restart the replication fork (12,13). (C, i) The replication fork encounters an obstacle/lesion and regresses. The newly synthesized leading and lagging strands anneal in this reaction, which creates a DSE. Under normal circumstances, the reversed fork structure can be resolved and restarted with or without the involvement of RecA (reviewed in (4,5)). (ii) In the RecA-dependent pathway, homologous recombination occurs, leading to two Holliday junctions, which are resolved by RuvABC. (iii) In the RecA-independent reaction, the DSE is digested by RecBCD, and the fork is restarted directly by PriA. However, if the reversed fork structure cannot be processed properly, the result may be cleavage of the fork by RuvABC, causing a DSE (C, i) right panel).

A DSE can arise when the last nucleotide before a nick is replicated and the replisome can no longer find the DNA template (Figure 1A, i). Another cause of DSEs is a replisome stall that lasts long enough for the next replisome to catch up from behind and rear-end it (Figure 1A, ii). In both cases, RecA-dependent homologous recombination performs the repairs to restart the forks and maintain genomic integrity (Figure 1B). RecBCD uses helicase and nuclease activities to unwind and degrade the DNA until it reaches a Chi site, a sequence occurring about every 5 kb on the E. coli chromosome (10,11). RecA recruited to the Chi site forms a filament with single-stranded DNA and invades an intact homologous DNA molecule. Primosomal proteins recognize the resulting DNA structures and can restart replication forks outside of oriC (12,13).

The replication fork reversal (RFR) model (Figure 1C) (14,15) offers a mechanism for how a stalled or disintegrated replisome can be restarted independently of RecA. Reversal of the replication fork occurs when the newly synthesized leading and lagging strands anneal, forming a Holliday junction (HJ) (Figure 1C, i). A replication fork stalled at a leading strand lesion can undergo RFR catalyzed either by RecA or RecG (13). If the reversed fork structure cannot be processed properly, RuvABC may cleave it (Figure 1C, i, right panel). In the RFR model, fork regression, possibly stimulated by RecG (16), can continue to a Chi site where RecA is loaded and recombination occurs (Figure 1C, ii). Alternatively, RecBCD can digest the DSE that is formed in the regression and displace the stabilizing RuvABC (Figure 1C, iii). Both pathways allow PriA-mediated restart of the replication fork (see (5) for a review).

The E. coli SeqA protein was first recognized as an essential factor in origin sequestration (17–20), which inhibits origin re-initiation and helps to ensure only a single initiation per cell cycle (21–23). In E. coli, the newly replicated DNA is transiently hemi-methylated before Dam methylase acts on GATC sites on the new DNA strand (17). GATC sites occur at high density in the origin (oriC), where SeqA protein interacts with them to prevent re-initiation for about one third of the cell cycle (24–27). This sequestration period is reduced to a minimum in SeqA mutants (28).

Evidence suggests other roles for SeqA. It binds to hemi-methylated GATC sites trailing replication forks (24,27,29–34), and most cellular SeqA appears to be associated with newly replicated DNA in complexes following the fork (32,35). The fork-trailing SeqA structures likely also contain oriC DNA throughout the sequestration period (36) and represent a type of hyperstructure of newly replicated DNA from both forks in addition to oriC DNA. Recent work has identified two dynamic SeqA structures held in close proximity (less than 30 nm apart) behind the replication fork, bound to ∼100 kb of DNA on each sister molecule. The distance from these SeqA structures to the replisome has been estimated to be 200–300 nm (37). One possible interpretation is that the DNA between the dual SeqA structures and the replisome might benefit from some additional stability during processes such as recombination and DNA repair. In addition, deletion of seqA results in genomic fragmentation in the absence of functional RecBCD (38–40), and many of the replication forks initiated at oriC in these deletion mutants do not reach the terminus (40).

Together, these findings suggest that SeqA structures associated with replication forks may serve as DNA stabilizers and help prevent DSB formation during replication and/or rescue of stalled replication forks. One possibility is that SeqA supports fork restart (Figure 2A), and another is that it creates a barrier that prevents collisions between new and old forks (rear-ending) (Figure 2B). To assess the possible roles of SeqA complexes behind the replication fork, we used two seqA mutants, seqA2 and seqA4 (18,20,41,42), and compared them to cells expressing functional SeqA protein but unable to exhibit sequestration (43).

Figure 2.

Models of how SeqA may protect the integrity of the replication fork. (A) Schematic illustration showing the SeqA complexes (in red) bound behind the replisome (in green) at a considerable distance (37). The bound SeqA complexes are in close proximity to each other, which means that the sister DNA is held close together behind the fork. This association could potentially provide stabilization during processes such as recombination, repair, and restart of replication forks, as illustrated here by replication fork reversal. (B) Stabilization of the newly replicated DNA by SeqA may create a barrier of bound protein that delays forks replicating a second round and prevents rear-ending, i.e. replication to the end of the new DNA from the first fork.

MATERIALS AND METHODS

Bacterial strains and growth conditions

All strains used are E. coli K-12 and are listed in Table 1. Cells were grown at the specified temperature (37°C or 42°C) in LB medium (LB) (10 g Tryptone, 5 g NaCl, 250 μl 4 M NaOH/l), AB minimal medium (44) supplemented with 10 μg/ml thiamine, and either 0.2% glucose and 0.5% casamino acids (glucose–CAA medium), or 0.2% glucose (glucose medium). The strain IBP05 was made by amplifying the chloramphenicol gene with the primers 5΄GCGCTAAGAACCATCATTGGCTGTTAAAACATTATTAAAAATGTCAATGGCATATGAATATCCTCCTTAG and 5΄CGATTTTTAGCAGACTGATATTTTCACTAATGACTTATTTTCTGCTTACCCTGGAGCTGCTTCGAAGTTCC containing regions homologous to a region in close proximity to oriC and insertion of this fragment into the chromosome, as described in (45). All other strains constructed for this work were made by P1 transduction (46), as indicated in Table 1. The recBC Ts mutations were verified by their sensitivity to UV irradiation, but the seqA2 and seqA4 mutants were verified by sequencing due to a rather low co-transduction of the resistance marker gene and the mutation. The oriCm3 mutants were verified by their asynchronous phenotype as determined by flow cytometry.

Table 1.

Strains

Strain	Relevant features	Source
MG1655	Wild type	(80,81)
AB1157	Wild type	(82)
MG1655oriCm3	oriCm3	(43)
IBP05	oriCm3cam	This work
UF340	seqA2	(20)
UF301	seqA4 dam13	(20)
CAG18433	asnB3057::Tn10	(83)
SF169	UF301 asnB3057::Tn10	UF301 x P1 CAG18433
N1331	Wild type	(84)
N1332	recA Ts	(80,81,84)
NL40	DS941difΔ6::KmR	(85)
DS984	DS941xerC::mini-Mu CmR	(86)
DS9008	DS941xerD::mini-Tn10(-9)	(87)
SS1211	Δrep::cam	(63)
IBP36	MG1655 oriCm3	MG1655 x P1 IBP05 (this work)
IBP37	MG1655 seqA2	MG1655 x P1 UF340 (this work)
IBP38	MG1655 seqA4	MG1655 x P1 SF169 (this work)
SK129	recB270 (Ts) recC271 (Ts)	(53)
IBP04	SK129 oriCm3	SK129 x P1 IBP05 (this work)
IBP01	SK129 seqA2	SK129 x P1 UF340 (this work)
IBP03	SK129 seqA4	SK129 x P1 SF169 (this work)
ER89	SK129 ΔseqA21	(39)
IBP07	N1332 oriCm3	N1332 x P1 IBP05 (this work)
IBP02	N1332 seqA2	N1332 x P1 UF340 (this work)
IBP06	N1332 seqA4	N1332 x P1 SF169 (this work)
IBP24	N1332 Δrep::cam	N1332 x P1 SS1211 (this work)
IBP39	MG1655 Δrep::cam	MG1655 x P1 SS1211 (this work)
IBP40	MG1655 seqA2 Δrep::cam	IBP37 x P1 SS1211 (this work)
IBP41	MG1655 seqA4 Δrep::cam	IBP38 x P1 SS1211 (this work)
IBP98	MG1655 oriCm3 Δrep::cam	MG1655oriCm3 x P1 SS1211 (this work)
IF01	recA938::cam	(88)
IBP87	MG1655 recA938::cam	MG1655 x P1 IF01 (this work)
IBP80	MG1655 seqA2 recA938::cam	IBP37 x P1 IF01 (this work)
IBP81	MG1655 seqA4 recA938::cam	IBP38 x P1 IF01 (this work)
IBP100	MG1655 oriCm3 recA938::cam	MG1655oriCm3 x P1 IF01 (this work)
IBP88	MG1655 difΔ6::KmR	MG1655 x P1 NL40 (this work)
IBP85	MG1655 seqA2 difΔ6::KmR	IBP37 x P1 NL40 (this work)
IBP82	MG1655 seqA4 difΔ6::KmR	IBP38 x P1 NL40 (this work)
IBP101	MG1655 oriCm3 difΔ6::KmR	MG1655oriCm3 x P1 NL40 (this work)
IBP89	MG1655 xerC::mini-Mu CmR	MG1655 x P1 DS984 (this work)
IBP74	MG1655 seqA2 xerC::mini-Mu CmR	IBP37 x P1 DS984 (this work)
IBP83	MG1655 seqA4 xerC::mini-Mu CmR	IBP38 x P1 DS984 (this work)
IBP90	MG1655 xerD::mini-Tn10(-9)	MG1655 x P1 DS9008 (this work)
IBP68	MG1655 seqA2 xerD::mini-Tn10(-9)	IBP37 x P1DS9008 (this work)
IBP84	MG1655 seqA4 xerD::mini-Tn10(-9)	IBP38 x P1 DS9008 (this work)
SMR14323	MG1655 ΔaraBAD567 Δattλ::PBAD zfd2509.2::PN25tetR FRTKanFRT	(70)
SMR13957	MG1655 Δattλ::PN25tetR FRT ΔattλTn7::FRTcatFRT PN25tetO gam-GFP	(70)
EH137	MG1655 PN25tetR PN25tetO gam-GFP	MG1655 x P1 SMR14323 x P1 SMR13957 (this work)
EH138	MG1655 seqA2 PN25tetR PN25tetO gam-gfp	IBP37 x P1 SMR14323 x P1 SMR13957 (this work)
EH139	MG1655 seqA4 PN25tetR PN25tetO gam-gfp	IBP38 x P1 SMR14323 x P1 SMR13957 (this work)

Preparation of DNA in agarose plugs for pulsed field gel electrophoresis

Overnight cultures of recBC (Ts) strains at 22°C (permissive temperature) were diluted to an optical density (OD600) of 0.005 and grown at 42°C (non-permissive temperature) in LB media to OD600 0.15 (exponential phase). A volume of culture corresponding to about 2.5 × 108 cells (Coulter Counter Multisizer, Beckman) was centrifuged and resuspended twice in Tris/NaCl (10 mM Tris–HCl, pH 7.6, 1 M NaCl) (initially 1 ml). Equal amounts of cell suspension (resuspended in Tris/NaCl) and molten clean-cut agarose (Bio-Rad) were brought to 42°C and combined. A total of 90 μl of the agarose/cell suspension was dispensed to wells of disposable molds (Bio-Rad), solidified, expelled into a 50 ml tube containing 2.5 ml EC lysis buffer (6 mM Tris–HCl, pH 7.6, 1 M NaCl, 100 mM EDTA, adjusted to pH 7.6 with NaOH, 1% N-lauryl sarcosine, 1 mg/ml lysozyme and 20 μg/ml RNase A), and incubated overnight at 37°C. After removal of the EC lysis buffer, the plugs were rinsed with TE buffer (10 mM Tris–HCl, pH 8.0, 1 mM EDTA, pH 8.0) and 2.5 ml ESP buffer was added (500 mM EDTA (adjusted to pH 9–9.5 with NaOH), 1% N-lauryl sarcosine, proteinase K (50 μg/ml)), followed by incubation at 37°C overnight (or 48–72 h). The plugs were then washed with TE buffer at 25°C, with three washes of 2 h each using 30 ml TE buffer. The steps above were taken from a standard protocol for preparation of plugs for pulsed field gel electrophoresis (PFGE) of bacterial DNA (47).

CHEF-DR III PFGE and quantification

A CHEF-DR III Pulsed Field Electrophoresis system (Bio-Rad) was used to resolve the DNA. The run time was set to 21 h; temperature was 14°C; the initial and final switch times were 60 and 120 s, respectively; volts/cm was set to 6; the included angle was 120; and 0.5 × TBE was used as the running buffer. The gel was then stained with SYBR Gold Nucleic Acid Gel Stain (Life Technologies) and quantified using Genetool (Syngene) software with the rolling disc method for background subtraction. SYBR Gold Nucleic Acid Gel Stain (Life Technologies) gives a linear relationship between fluorescence intensity and DNA content over at least two orders of magnitude (48), as also applied previously (49). The percentage of chromosomal fragmentation was found by first measuring the DNA present in the well and directly beneath the well (non-fragmented DNA and most likely chromosomes with a single nick, respectively) and then measuring the DNA in the rest of the lane. The fragmented DNA value was then divided by the total DNA value. Quantification of the chromosomal fragmentation of a rep recBC mutant with this method was in agreement with already published results (∼50%) (14).

Flow cytometry and interpretation of DNA histograms

Cells were grown exponentially for several generations to ensure balanced growth. At an OD of 0.15, exponentially growing cells were either harvested directly or treated with rifampicin (300 μg/ml) and cephalexin (10 μg/ml) for the time equivalent of three to four generations before harvesting. Both directly harvested and treated cells were resuspended in TE buffer (10 mM Tris–HCl, pH 8.0, 1 mM EDTA, pH 8.0) and fixed in 70% ethanol. Fluorescein isothiocyanate (FITC, Sigma-Aldrich) (50) was used for protein staining (representing mass), and Hoechst 33258 was used for DNA staining (Sigma-Aldrich) (51). Flow cytometry was performed with a LSR-II flow cytometer equipped with a 488 nm argon ion laser and a 355 nm krypton laser (BD Biosciences), and the results were analyzed using FlowJo software (Tree Star, Inc.).

Rifampicin and cephalexin inhibit initiation of replication and cell division, respectively. Cells cannot initiate a new round of replication after these drugs are added but can complete ongoing rounds of replication (without dividing). DNA histograms of the treated cells thus give an integer number of chromosomes per cell with values 2 or 2, where n = 0, 1, 2, 3…represents the generation in which initiation occurs. However, this pattern holds true only for cells with synchronous initiation of replication (all origins firing simultaneously) and successful completion of replication elongation (all initiated forks reaching the terminus). DNA histograms of cells that exhibit asynchronous initiations will also contain irregular numbers of chromosomes (such as 3, 5, 7, etc.), and problems with replication elongation will be seen as a lack of distinct peaks in the ‘run-out’ DNA histograms (i.e. histograms of cells treated with rifampicin and cephalexin).

Replication run-out in the absence of RecA function

Cultures of cells growing exponentially (to OD ∼0.15) at 30°C were split, and rifampicin (450 μg/ml) and cephalexin (10 μg/ml) both were added to each culture. One portion was kept at the permissive temperature (30°C) whereas the other was shifted to the non-permissive temperature (42°C) for loss of RecA function. Cells were harvested before drug treatment and after 3–4 generation times in the presence of the drugs and fixed for analysis with flow cytometry to compare replication fork run-out at each temperature. The flow histograms were analyzed using FlowJo (Tree Star, Inc.) software. The average per-cell decrease in the number of chromosomes was obtained by first identifying the number of cells in each chromosome peak, multiplying that value by each chromosome number to obtain the total number of chromosomes, and in turn dividing that by the total number of cells counted. A total of 50 000 cells were counted for a proper estimate of the average number of chromosomes/cell. Finally, the total number of chromosomes was compared at the permissive and non-permissive temperatures.

Viability tests

Overnight cultures were serially diluted with growth medium or 1% NaCl to approximately the same OD. A total of 5 μl of dilutions ranging from 10−2 to 10−6 were spotted onto agar plates (agar type indicated in figures).

Gam-GFP induction, fluorescence microscopy imaging and flow cytometry of GFP

The cells were grown to OD ∼0.15 before Gam-GFP was induced by adding 10 ng/ml anhydrotetracycline. Growth was continued for 60 min, at which time the cells were immobilized on a 17 × 28 mm agarose pad (1% containing phosphate-buffered saline (PBS) with 10 ng/ml anhydrotetracycline) and covered with a no. 1.5 coverslip. Images were acquired with a Leica DM6000 microscope equipped with a Leica EL6000 metal halide lamp and a Hamamatsu ImagEM 1k camera. Phase-contrast imaging was performed with an HCX PL APO 100 × /1.40 NA objective. Fluorescence imaging was done using narrow band-pass (BP) filter sets (excitation at BP 470/40 and emission at BP 525/50 for GFP).

In the fluorescence microscopy experiments, care was taken to use the exact same settings for GFP-imaging of wild-type and seqA mutant cells (e.g. the same intensity, exposure time) to avoid misinterpretation. Using the publicly available ImageJ software, in post processing, we adjusted only brightness/contrast, doing so with the exact same cutoff values for images of the seqA mutants and wild-type cells.

Flow cytometry of Gam-GFP was performed with Accuri C6 (BD Biosciences), and the results were analyzed using FlowJo software (Tree Star, Inc.). Cells were grown as described above in glucose-CAA medium with 60-min induction of Gam-GFP (10 ng/ml anhydrotetracycline), harvested and washed in PBS directly prior to analysis with flow cytometry. A total of 50 000 cells were recorded per sample.

Fluorescence microscopy of nucleoids

Nucleoids of fixed cells were stained with Hoechst 33258 and visualized with fluorescence microscopy, as described in (52).

RESULTS

Some replication forks do not reach the terminus in seqA2 and seqA4 mutants

In light of the chromosomal fragmentation (40) and induction of the SOS response (18) reported for seqA deletion strains, we used the seqA2 and seqA4 mutants to elucidate in more detail the importance of SeqA in stabilizing the replication fork. The seqA2 point mutation leads to a change (N152D) in the C-terminus, resulting in a DNA binding deficiency (41), while the seqA4 point mutation leads to a change (A25T) in the N-terminus, inhibiting multimerization (42). Thus, neither strain can form SeqA complexes on newly replicated DNA behind the replication forks.

These strains also cannot sequester oriC from unscheduled, premature initiations and they therefore exhibit an asynchrony phenotype (41,42). An increased number of replication forks per chromosome, as in seqA mutants, could lead to collisions of new and old forks (the so-called replication fork rear-ending scenario; see Figure 1A, ii). To differentiate the two causes of genomic instability—i.e. no sequestration versus having no functional SeqA behind the replication forks—our control was the oriCm3 mutant, which lacks sequestration because of eight mutations that inactivate critical oriC GATC sites but has a fully functional SeqA protein (43). In these cells, extra initiations are allowed, but all cellular proteins are normal.

We use flow cytometry to investigate the DNA content and degree of asynchrony in the three mutants (Figure 3). The mean cellular DNA content was somewhat higher in the two seqA mutants compared to the oriCm3 mutant (Figure 3, left panels; Table 2), but because they also had increased mass, the DNA concentration (DNA/mass) was similar to that of the oriCm3 mutant (Table 2). Compared to the wild type, which has two peaks respectively corresponding to four and eight fully replicated chromosomes, all three mutants had an abundance of additional chromosome peaks, ranging from 3 to about 20 chromosomes. However, the run-out histogram of the oriCm3 differed markedly from that of the two seqA mutants. The peaks of the seqA mutants were much less distinct than for the oriCm3 mutant (Figure 3, middle panels), indicating that some replication forks could not reach the terminus during rifampicin and cephalexin incubation, presumably because of problems during fork elongation.

Figure 3.

Lack of replication fork elongation and induction of the SOS response in the mutant seqA2 and seqA4 cells. Wild-type cells (MG1655), oriCm3 mutant cells (IBP36), seqA2 (IBP37) and seqA4 (IBP38) mutant cells were grown in glucose–CAA at 37°C and analyzed by flow cytometry and microscopy. DNA histograms of exponentially growing cells (left panel) and run-out histograms of cells treated with rifampicin and cephalexin (see Materials and Methods) (middle panel) are shown. The chromosome equivalents (DNA content per cell) are shown on the abscissa and the number of cells per channel on the ordinate. Exponentially growing cells, fixed and stained with Hoechst, were also analyzed with phase contrast and fluorescence microscopy (see Materials and Methods) (right panel).

Table 2.

Mean doubling time, mean relative cellular DNA content, mean relative mass and DNA/mass of the seqA mutants and the oriCm3 mutant

Strain	DT (min) ± STD	Relative DNA content ± STD	Relative mass ± STD	DNA/mass ± STD
Wild type (MG1655)	25 ± 0.0	1.0 ± 0.0	1.0 ± 0.0	1.0 ± 0.0
oriCm3 (IBP36)	25 ± 1.8	1.26 ± 0.1	1.14 ± 0.1	1.11 ± 0.0
seqA2 (IBP37)	30 ± 0.6	1.96 ± 0.4	1.72 ± 0.5	1.19 ± 0.3
seqA4 (IBP38)	29 ± 1	1.78 ± 0.3	1.43 ± 0.2	1.25 ± 0.1

In addition, on microscopy analysis, the seqA mutants showed filamentous cells with unsegregated DNA, but the oriCm3 mutant did not (Figure 3, right panel). The seqA mutants also showed a greater heterogeneity in size/mass distribution in FITC histograms from flow cytometry (Table 2). The doubling time of the two seqA mutants was increased ∼20% compared to wild-type cells, but the oriCm3 mutation did not affect growth rate (Table 2). This increase in doubling time may reflect some cell death or at least a cell division delay among the seqA mutants (based on flow cytometry and microscopy) caused by the SOS response. The SOS response and abortion of replication forks seen here for the two seqA mutants are in agreement with previously reported phenotypes for SeqA-less cells (18,38,40).

The seqA2 mutant cells seemed to show greater problems with DNA segregation and cell division compared to the seqA4 cells, based on nucleoids under microscopy (Figure 3) and cell mass values (Table 2).

The seqA2 and seqA4 mutants exhibit a significantly higher degree of chromosomal fragmentation compared to the oriCm3 mutant

To address whether replication fork disintegration might be why forks in the seqA mutants do not reach the terminus, we investigated the degree of chromosomal fragmentation in these mutants. The seqA2, seqA4 and oriCm3 mutations were transferred to a recBC temperature-sensitive strain, allowing us to inactivate RecBCD (53). These mutants exhibit no nuclease or recombination activity at the non-permissive temperature, so the amount of chromosomal fragmentation can be measured with PFGE. The ΔseqA recBC(Ts) double mutant was included as a positive control because it exhibits chromosomal fragmentation (40).

The seqA mutants of the temperature-sensitive strain had a significantly higher degree of chromosomal fragmentation compared to the oriCm3 mutant (Figure 4; Supplementary Table S1). This result confirms that the seqA mutant cells experienced more problems during replication elongation than cells lacking only sequestration (oriCm3) and that the replication forks in seqA mutant cells disintegrate.

Figure 4.

The seqA2 and seqA4 mutants exhibit significantly more chromosomal fragmentation compared to the oriCm3 mutant. (A) Pulsed field gel electrophoresis of wild-type (AB1157), recBC (Ts) (SK129), oriCm3 recBC (Ts) (IBP04), seqA2 recBC (Ts) (IBP01), seqA4 recBC (Ts) (IBP03), and ΔseqA recBC (Ts) (ER89) after growth at 42°C in LB medium. Standard DNA from S. cerevisiae is shown to the left. The compression zone (containing large chromosomal fragments that are unable to move any further) is indicated with * (see Materials and Methods for more information). (B) Quantification of the chromosomal fragmentation found in (A). The amount of chromosomal fragmentation (%) was calculated by dividing the signal in the whole lane (excluding the well and the area right below the well) by the total signal (see Materials and Methods for more information). Error bars represent standard deviations.

The seqA mutants undergo direct DSBs

The above result showed that the seqA2 and seqA4 mutants exhibit chromosomal fragmentation when recBC is inactivated. Such large genomic fragments may arise from various abnormalities in replication fork elongation. PFGE of the seqA2 and seqA4 single mutants did not reveal any significant fragmentation (i.e. without the recBC Ts mutation) (Supplementary Figure S1); thus, the chromosomal fragmentation in the seqA mutants is mostly repairable when RecBCD is active, as reported for the ΔseqA cells (40). However, chromosomal fragmentation in the absence of RecBCD could result from cleavage of reversed forks, as shown in Figure 1C, i. According to the RFR model, in the absence of RecBCD, the HJ is cleaved by RuvABC, and fragmentation can be observed by PFGE. This restart mechanism thus generates DSEs only when RecBCD is absent and is dependent on RuvABC. We were unable to construct seqA2 or seqA4 strains lacking RuvAB (three rounds of unsuccessful transduction where controls were successful using the same lysate), however, possibly indicating that the seqA mutants depend on homologous recombination to survive.

To check whether this is the case, we also investigated their dependence on RecA. When we combined the two seqA mutations with a recA deletion, the viability of the seqA2 and seqA4 mutants decreased significantly (Figure 5A). This decrease was especially prominent when the seqA recA mutants were grown on LB agar, probably because of greater numbers of replication forks compared to growth on glucose–CAA agar (and the consequently higher demand for SeqA protein). The seqA2 mutant also was somewhat less robust on LB agar than the seqA4 mutant, as previously indicated from microscopy images and flow cytometry mass data (Figure 3; Table 2). Thus, seqA cells depend more than normal on homologous recombination, and DNA fragmentation results from formation of DSEs that require recombinational repair. The ΔrecA seqA2/4 co-inhibition is in accordance with results previously reported for ΔseqA strains (38,40). In contrast, oriCm3 cells did not exhibit decreased viability when recA was deleted (Figure 5A), indicating that direct DSBs in the seqA mutants are attributable to an absence of functional SeqA in the cells rather than to over-replication because of a lack of sequestration.

Figure 5.

Loss of viability in the seqA mutants when combined with ΔrecA, Δdif, ΔxerC, ΔxerD or Δrep. Viability tests: Overnight cultures grown at 37°C were serially diluted and spotted on plates with the indicated medium. The spots are from left to right: undiluted and diluted 10−2, 10−3, 10−4, 10−5 and 10−6. (A) Viability tests of the seqA2 ΔrecA (IBP80), seqA4 ΔrecA (IBP81), and oriCm3 ΔrecA (IBP100) mutants, as well as the single mutants. (B) Viability tests of the seqA2 Δdif (IBP85), seqA4 Δdif (IBP82), oriCm3 Δdif (IBP101), seqA2 ΔxerC (IBP74), seqA4 ΔxerC (IBP83), seqA2 ΔxerD (IBP68) and seqA4 ΔxerD (IBP84) mutants, as well as the single mutants. (C) Viability tests of seqA2 Δrep (IBP40), seqA4 Δrep (IBP41) and oriCm3 Δrep (IBP98), as well as the single mutants.

Dimer resolution is important for seqA2 and seqA4 mutant viability

Recombination and resolution of HJ may result in crossing over, leading to dimeric chromosomes (54–56). Resolution of a dimeric chromosome occurs at the dif site and is coupled to cell division (57). The multifunctional protein FtsK positions the dif sites to the septum and activates the tyrosine recombinases XerC and XerD, which in turn cut and re-join the DNA, resolving the dimeric chromosome (reviewed in (57)). Because the seqA mutants seemed to depend on homologous recombination, we evaluated the effect of removing the dif site, XerC or XerD in these mutants. The cells grew poorly on rich media, so we conducted the viability test using glucose agar (to avoid suppressors). When only dif, XerC or XerD was removed, the mutants exhibited somewhat reduced viability, as demonstrated previously (58), indicating chromosomal dimers arise at a certain (low) frequency in otherwise normal cells (Figure 5B, top panel). However, in the double mutants, the viability was decreased compared to the single dif/XerC/XerD mutants (Figure 5B, middle panels). The effect was more prominent in seqA2/4 dif mutants compared to the seqA2/4 xerC/xerD mutants, presumably because cells may retain some dimer resolution activity when only XerC or XerD is deleted whereas all activity is lost in the dif cells in which XerCD cannot bind to the DNA. In summary, the results indicate a higher frequency of chromosomal dimers in seqA mutant cells compared to wild-type cells and elevated homologous recombinational activity. Removal of dif affected viability of oriCm3 cells to the same extent as wild-type cells (Figure 5B, bottom panel), confirming that lack of sequestration alone does not lead to increased recombinational repair.

The seqA mutants exhibit more ‘reckless’ degradation during replication run-out than the oriCm3 mutant

To gather more information about the nature of replication fork rescue and restart in the seqA mutants, we investigated the so-called ‘reckless’ degradation caused by RecBCD in the absence of the RecA protein. recA (Ts) cells exhibit loss of entire chromosomes during replication fork run-out after treatment with rifampicin and cephalexin at the non-permissive temperature (59). A Δrep mutant that also has elongation problems was included in the experiment for comparison (14,60–63). The Rep helicase helps clear obstacles at the replication fork (64) and interacts with the replicative helicase DnaB (65). Cells lacking Rep stall more frequently than normal and can restart the stalled forks through the RecBCD- and RuvABC-dependent remodeling reaction, which involves RFR and is independent of RecA (14). The recA (Ts) allele was combined with oriCm3, Δrep, and seqA mutations and the resulting strains compared with the recA (Ts) single mutant by flow cytometry (see Table 3 for doubling times, DNA content, mass and DNA/mass values).

Table 3.

Mean doubling times, mean relative cellular DNA content, mean relative mass and DNA/mass of the recA (Ts) strains

Strain	DT (min) ±STD	Relative DNA content ±STD	Relative mass ± STD	DNA/mass ± STD
Wild type (N1331)	66 ± 4.0	1.0 ± 0.0	1.0 ± 0.0	1.0 ± 0.0
recA (Ts) (N1332)	69 ± 0.2	1.0 ± 0.0	1.1 ± 0.1	1.0 ± 0.0
oriCm3 recA (Ts) (IBP07)	70 ± 0.6	1.3 ± 0.1	1.5 ± 0.1	0.9 ± 0.0
seqA2 recA (Ts) (IBP02)	85 ± 3.0	2.4 ± 0.3	2.4 ± 0.4	1.0 ± 0.1
seqA4 recA (Ts) (IBP06)	87 ± 2.7	2.2 ± 0.2	2.2 ± 0.1	1.0 ± 0.1
Δrep recA (Ts) (IBP24)	73 ± 1.0	1.61 ± 0.0	2.07 ± 0.2	0.79 ± 0.1

The loss of DNA during replication run-out in the absence of RecA function in the control cells (recA (Ts) at non-permissive compared to permissive temperatures was ∼15% (Figure 6A; Table 4), as can be seen in the run-out histogram (fewer cells in the four-chromosome peak, more cells in the three-chromosome peak; compare Figure 6B, recA (Ts) right panel and middle panel). It is not clear exactly which DNA is degraded during the replication run-out, but the most straightforward interpretation is that the DSE of a collapsed replication fork is bound by RecBCD and that this enzyme complex then degrades the entire ‘daughter arm’ of a partially replicated chromosome (66–68) (Figure 6C). Because RecA is non-functional, the degradation reaction will keep going past Chi sites and past oriC until the entire daughter arm is degraded and both replication forks eliminated (Figure 6C). The recA (Ts) control cells have a replication pattern in which initiation occurs at two origins and the cells have two replicating chromosomes with two forks each. If one of the forks encounters an obstacle or nick that causes the fork to collapse and a daughter arm to be degraded, after replication run-out, the cell will contain three chromosomes instead of four.

Figure 6.

Degradation of DNA during run-out in the absence of RecA is extensive and does not yield integer numbers of chromosomes in the seqA mutants. (A) DNA histograms obtained by flow cytometry of wt (N1331), recA (Ts) (N1332), oriCm3 recA (Ts) (IBP07), seqA2 recA (Ts) (IBP02), seqA4 recA (Ts) (IBP06), and Δrep recA (Ts) (IBP24) grown in glucose–CAA medium. DNA histograms of cells growing exponentially at the permissive temperature (left panel), and run-out histograms of cells treated with rifampicin and cephalexin at the permissive (middle panel) and non-permissive (right panel) temperature are shown. (B) Bar plot showing the decrease in average number of chromosomes per cell during replication run-out in the absence of RecA function. Error bars represent standard deviations. (C) Model showing how DNA degradation by RecBCD in the absence of RecA may occur. In the model, collapse of the rightmost replication fork is shown. RecBCD binds to the resulting DSE and degrades the entire chromosome arm, eliminating both replication forks. The nicks are ligated, yielding one intact chromosome in a cell that under RecA+ conditions would have ended up with two chromosomes after replication run-out. The figure is adapted from (89).

Table 4.

Mean number of chromosomes/cell after replication run-out and decrease representing reckless degradation after inactivation of RecA

Strain	Chromosomes/cell 30°C ± STD	Chromosomes/cell 42°C ± STD	% Change ± STD
Wild type (N1331)	3.4 ± 0.1	3.5 ± 0.1
recA (Ts) (N1332)	3.5 ± 0.0	3.0 ± 0.1	13 ± 2.4
oriCm3 recA (Ts) (IBP07)	4.3 ± 0.1	2.8 ± 0.1	35 ± 3.1
seqA2 recA (Ts) (IBP02)	8.5 ± 0.1	2.4 ± 0.4	72 ± 4.9
seqA4 recA (Ts) (IBP06)	7.8 ± 0.5	2.3 ± 0.3	70 ± 5.4
Δrep recA (Ts) (IBP24)	5.6 ± 0.2	2.7 ± 0.2	46 ± 3.6

The amount of degradation in the oriCm3 recA(Ts) mutant was ∼35% (Figure 6A; Table 4), and as in the recA(Ts) mutant, most of the cellular DNA was found as integer numbers of chromosomes (Figure 6B). The reason for the increased amount of degradation may be that an increased number of forks and more closely spaced forks (caused by asynchrony and over-initiation) led to an increase number of collapsed forks. Likewise, in the Δrep recA(Ts) cells, integer numbers of chromosomes dominated, indicating that in the absence of RecA, collapsed fork arms were fully degraded but that much of the chromosomal DNA stayed intact (Figure 6B). The increased amount of fork stalling in the absence of Rep protein led to ∼45% degradation during RecA-less replication run-out (Figure 6A; Table 4).

The seqA recA(Ts) mutants, on the other hand, showed a lack of histogram peaks after RecA-less replication run-out (Figure 6B) and also extensive degradation (∼70%) (Figure 6A; Table 4). The greater degradation in the seqA recA(Ts) mutants compared to the oriCm3 recA(Ts) mutant and the Δrep recA(Ts) mutant indicates that DSEs occurred more frequently in the seqA recA(Ts) mutants, possibly because Δrep cells restart stalled or paused replication forks directly (without homologous recombination) by the RecBCD-dependent modulation of the stalled fork (see Figure 1C, iii) (14). Thus, in Δrep cells, collapse of forks would lead to degradation during RecA-less replication run-out. The results indicate that in the absence of SeqA function, more forks collapse directly, more rear-ending of forks occurs, or both events take place. The results are in accordance with the absence of peaks in histograms of seqA2 and seqA4 cultures after growth in run-out conditions in the presence of RecA (Figure 6B, middle panels; Figure 3).

Replication fork remodeling may not be possible in seqA mutants

Stabilization of DNA (200–300 nm) between the SeqA structure and the replisome (37) might be beneficial for replication fork re-start after replication fork stalling or disintegration. Loss of this stabilized area (i.e. by lack of SeqA) thus could affect such replication fork reactions. To investigate a scenario in which replication forks pause and stall more frequently than normal, we combined the seqA and oriCm3 mutations with Δrep. As noted, when Rep helicase is missing, replication forks stall more frequently and can be restarted by a remodeling reaction probably involving RFR (14). If the remodeling reaction requires SeqA, the double mutants would be expected to show a loss of viability.

In accordance with previous findings (61), the Δrep single mutant cells showed the expected run-out histogram with clear peaks (Figure 6A, bottom row, middle column), indicating that the replication forks could complete replication. Thus, although replication forks stalled more frequently in rep-deleted cells, they seem to be restarted rather efficiently through direct restart mechanisms (see Figure 1C, iii) and still reached ter without a significantly affected doubling time (Table 3).

In contrast, cells with either seqA2 or seqA4 in combination with rep deletion displayed a significantly decreased viability compared to single mutants (Figure 5C), as has been shown for ΔseqA rep cells (69). However, the combination of oriCm3 with the rep deletion did not produce this result, possibly indicating that replication forks stalled because of absent Rep helicase are not managed the same way without SeqA as they are with SeqA (i.e. by direct restart mechanisms, at least some of which involve RFR) (see Figure 1C, iii). The unchanged viability of the oriCm3 Δrep cells shows that an increased number of replication forks is not the sole factor leading to greater difficulty for Δrep cells also carrying mutations in seqA.

Microscopy and flow cytometry of Gam-GFP indicates occurrence of DSEs in the seqA mutants

For further evidence of the occurrence of DSEs in seqA mutants, we used a Gam-GFP construct controlled by a doxycycline/tetracycline-inducible promoter (kindly provided by S. Rosenberg (70)). The Gam protein originates from the Mu phage and binds and protects DSEs of linear DNA, thus inhibiting exonuclease activity (71) and subsequently recombinational repair (70). Gam-GFP can therefore be used to visualize DSEs in living cells because it forms fluorescent foci upon DNA DSE binding.

The seqA mutants containing the gam-GFP construct were investigated using fluorescence microscopy and flow cytometry, together with the wild-type equivalent (MG1655 gam-GFP). When visualized with fluorescence microscopy, both the wild-type and the seqA mutants showed a weak background of Gam-GFP fluorescence (pseudo-colored blue, Figure 7A). However, in contrast to the wild-type cells, many of the seqA mutant cells contained structures with a strong GFP signal, and a few cells contained distinct foci (pseudo-colored magenta, Figure 7A).

Figure 7.

Analysis of Gam-GFP signal using fluorescence microscopy and flow cytometry indicates that DSEs arise in the seqA mutants. (A) Fluorescence microscopy of Gam-GFP in wild-type (EH137), seqA2 (EH138) and seqA4 (EH139) mutant cells grown in glucose–CAA at 37°C after 1 h of induction with anhydrotetracycline (10 ng/ml). The GFP-signal is pseudo-colored according to pixel intensity, in which blue represents weak background signal and magenta represents higher intensity values. (B) Bar plot showing the percentage of the cell population containing high-intensity Gam-GFP signal deduced from flow cytometry analysis of GFP in 50 000 cells for wild-type (EH137), seqA2 (EH138) and seqA4 (EH139) mutant strains after 1 h of induction with anhydrotetracycline (10 ng/ml). The gating threshold for definition of high- and low-intensity values was set according to the reported fraction of wild-type cells containing Gam-GFP foci due to spontaneous DSBs (7.5%) (70). Error bars represent standard error of the mean (SEM).

To quantify the number of cells containing high-intensity Gam-GFP signal, we analyzed 50 000 cells per sample using flow cytometry. Shee et al. (70) reported that ∼7.5% of wild-type cells contained Gam-GFP foci as a result of spontaneous DSBs, so we gated the scatter plot of the wild-type cells (GFP signal versus cell size/forward scatter) according to this value. Because background Gam-GFP fluorescence could be observed in all cells, only cells with greater-than-expected GFP intensity based on cell size were included. The gating threshold for defining high- and low-intensity values (respective to cell size) was copied to the scatter plots of the seqA mutant cells. The fraction of the population containing GFP signal above the set intensity threshold yielded, on average, 16.2% and 15.9% for the seqA2 and seqA4 mutant cells, respectively (Figure 7B). The implication is that more than twice as many cells contained DSEs in the seqA mutant populations compared to the wild-type population.

These results support our previous inferences that DSEs are generated during replication in the seqA mutants and that rear-ending of forks, a lack of direct restart of forks, or both may occur in the absence of SeqA function.

DISCUSSION

Lack of SeqA complexes behind the replication forks leads to replication fork disintegration

We show in this work that both seqA2 and seqA4 mutant cells display signs of disintegrated replication forks, resulting in formation of DSEs and rendering these cells dependent on homologous recombination. The observed phenotypes of cells lacking SeqA function are in accordance with previous microarray studies showing that cells without SeqA have an increased oriC/ter ratio and aberrant frequency of gene markers between the origin and the terminus when growing exponentially (40,69). In those studies, the frequency of genes did not follow an exponential pattern but increased in the origin-proximal half of the chromosome, which indicates that not all replication forks launched at oriC reach the terminus within a reasonable amount of time. Both these earlier results and our current findings thus indicate that replication forks disintegrate in cells without proper SeqA function. In contrast, oriCm3 cells performed replication run-out and were not abnormally dependent on recombination enzymes. Together, these findings indicate that (i) binding of fully functional SeqA behind the replication fork can protect cells against formation of DSEs during replication and (ii) the chromosomal fragmentation observed in the seqA mutants is not merely a result of an increased number of replication forks.

Rotman et al. (40) proposed that segregation of sister chromosomes could be the source of DSEs in the absence of SeqA behind the replication forks. In this model, the lack of the SeqA barrier would cause the ‘segregation fork’ to collide with a stalled replication fork, leading to a rupture at a single-stranded region as the segregation machinery pulls the sister strands apart. Although SeqA has been proposed to delay separation of newly synthesized DNA (72), the segregation fork model is not concordant with studies of loci segregation after replication or with differences in SeqA requirements during slow and rapid growth. Loss of SeqA affects the growth rate of rapidly growing but not slowly growing cells (73), indicating that its functions are more important during rapid growth. However, in rapidly growing cells, the replication period spans several generations, so that newly replicated sister molecules reside in the same cell half and co-segregate (36,37) (Figure 8A). Thus, the increased demand for SeqA during rapid growth does not fit with the ‘segregation fork’ hypothesis because the two sister DNA molecules behind a fork do not segregate before cell division during rapid growth. Moreover, in slowly growing cells in which the ‘segregation fork’ in theory could pose a threat to the replication fork (because the sister DNA molecules are segregated to each cell half; Figure 8B), replicated loci stay co-localized for 10–20 min (74). We therefore find it more likely that the observed chromosomal fragmentation in the seqA mutants arises from challenges during chromosome replication rather than from segregation issues.

Figure 8.

Newly replicated sister DNA molecules are segregated to each cell half during slow growth but not during rapid growth. (A) Schematic and simplified figure of the replication pattern and chromosome localization for MG1655 cells grown in glucose-CAA medium at 37°C. The sister DNA from each chromosome co-segregates to each cell half. (B) Schematic and simplified figure of the replication pattern and chromosome localization for MG1655 cells grown in minimal medium at 37°C. The cells contain only one replicating chromosome, and the sister DNA is thus segregated to each cell half.

The dynamic SeqA structures trailing the replication forks might prevent rear-ending

A DSE can theoretically arise if a replication fork catches up with an ‘older’ replication fork in front and replicates the last nucleotide of the newly synthesized strand, the so-called rear-ending (75–78) (Figure 1A, ii). Replication forks stall from time to time (7,61), so replication forks could possibly catch up with each other, especially if sequestration (which causes spacing between the forks of about one third of a generation) is absent. SeqA complexes bound to hemi-methylated DNA trailing the replication fork might delay progression of replication forks approaching from behind, helping to prevent rear-ending collisions (Figure 2B). The rear-ending model thus offers a possible explanation for the formation of DSEs in the seqA mutants, which both over-initiate and lack the proper SeqA complexes behind the fork.

A recent study found that rear-ending does not seem to be the source of DSEs when SeqA is absent (40); however, whether the applied method had sufficient sensitivity to exclude this possibility is uncertain. We show here that DSEs are generated during replication run-out in cells lacking SeqA, a problem that increases with the number of forks following each other on a chromosome. Thus, we argue that it is highly likely that some of the DSEs present in cells lacking SeqA trace to rear-ending of replication forks. This inference seem reasonable because the SeqA protein can prevent rear-ending in two ways: by sequestration, which provides a minimum space between forks (rounds of replication), and by forming a protein barrier that may delay the progress of a new fork moving toward the old one.

Binding of SeqA behind the replication fork might facilitate restart of stalled replication forks

Rear-ending may not be the only problem in cells lacking SeqA function, however. Massive over-initiation because of unregulated levels of ATP-DnaA is reported to lead to cell death as replisomes run into nicks arising from partial oxidative damage repair (79), forming DSEs that apparently exhaust the homologous recombination repair capacity. Cells lacking SeqA function do not over-initiate to this extent, but slow repair of damage relative to the number of ongoing replication forks also could be a problem in the absence of SeqA. A third issue in cells lacking SeqA function could be that ‘easy’ or ‘direct’ restart of stalled replication forks, i.e. restart that does not require homologous recombination, functions poorly. The stretch of DNA between the SeqA complexes trailing the fork and the replisome may be an optimal location for replication fork repair and restart reactions. For instance, replication fork remodeling processes such as RFR may be more easily accommodated, as suggested in the model shown in Figure 2A. The reduced viability of the Δrep seqA mutants could point toward a deficiency in restart. Replication forks in Δrep cells frequently stall and restart via direct restart mechanisms (14). The oriCm3 cells over-initiate in the same manner as the seqA mutants (i.e., because of lack of sequestration) but can exhibit frequent replication restart in the absence of Rep protein (successful elongation of replication by flow cytometry). Thus, an increased number of replication forks is not the reason for the loss of viability in combination with seqA mutation. The cells without Rep are especially dependent on direct restart of replication forks and on the presence of SeqA at the forks, so the result may indicate that the SeqA structures somehow facilitate replication fork restart mechanisms. Binding of SeqA at a distance behind the replisome might provide a stabilized area necessary for such replication fork restart reactions to occur (Figure 2A). It is also possible that SeqA actively participates in restart of stalled forks via a so far unknown reaction.

In summary, we find it likely that replication elongation problems in cells without SeqA function stem from a combination of replication fork rear-ending and deficiencies that occur from aberrant handling of the nascent strands, such as lack of direct replication fork restart.

Click here for additional data file.