Abrogation of the Chk1-Pds1 Checkpoint Leads to Tolerance of Persistent Single-Strand Breaks in Saccharomyces cerevisiae

Abstract

In budding yeast, Apn1, Apn2, Tpp1, and Rad1/Rad10 are important enzymes in the removal of spontaneous DNA lesions. apn1 apn2 rad1 yeast are inviable due to accumulation of abasic sites and strand breaks with 3′ blocking lesions. We found that tpp1 apn1 rad1 yeast exhibited slow growth but frequently gave rise to spontaneous slow growth suppressors that segregated as single-gene mutations. Using a candidate gene approach, we identified several tpp1 apn1 rad1 suppressors. Deleting uracil glycosylase suppressed both tpp1 apn1 rad1 and apn1 apn2 rad1 growth defects by reducing the abasic site burden. Mutants affecting the Chk1-Pds1 metaphase-anaphase checkpoint only suppressed tpp1 apn1 rad1 slow growth. In contrast, most S-phase checkpoint mutants were synthetically lethal in a tpp1 apn1 rad1 background. Epistasis analyses showed an additive effect between chk1 and ung1, indicating different mechanisms of suppression. Loss of Chk1 partially restored cell-growth parameters in tpp1 apn1 rad1 yeast, but at the same time exacerbated chromosome instability. We propose a model in which recombinational repair during S phase coupled with failure of the metaphase-anaphase checkpoint allows for tolerance of persistent single-strand breaks at the expense of genome stability.

THE genetic integrity of organisms is constantly challenged by extrinsic and intrinsic factors. Sources of endogenous damage include reactive oxygen species, the intrinsic lability of DNA chemical bonds, endogenous methylating agents, and DNA replication errors (Hoeijmakers 2001). Common lesions that arise in cellular DNA as result of such damage include abasic (AP) sites, modified or miscoding bases, and single-strand breaks (SSBs) with abnormal termini such as 3′ phosphates, 3′ aldehydes, and 3′ phosphoglycolates (Friedberg et al. 1995). In addition, persistent SSBs can be converted to lethal double-strand breaks (DSBs) during replication (Caldecott 2001). These various DNA lesions block the processes of DNA replication and transcription and lead to an elevated mutation rate, chromosome instability, and ultimately cancer in higher organisms. Multiple highly conserved and redundant processes therefore exist to mitigate the effects of DNA damage. These include DNA repair, surveillance mechanisms known as DNA damage checkpoints, and DNA damage tolerance (Friedberg et al. 2004).

In Saccharomyces cerevisiae, many overlapping repair pathways deal with spontaneous lesions, in particular AP sites and strand breaks with 3′ blocked termini. During base excision repair (BER), the AP endonucleases Apn1 and Apn2 act on AP sites and all 3′ blocking lesions to yield 3′ hydroxyl termini (Guillet and Boiteux 2004). The glycosylase-lyases Ntg1, Ntg2, and Ogg1 also act on AP sites but generate SSBs with 3′ blocking aldehydes through the process of β-elimination (Guillet and Boiteux 2004). The DNA 3′ phosphatase Tpp1, a homolog of human polynucleotide kinase/3′ phosphatase (Vance and Wilson 2001b), specifically removes 3′ phosphates at strand breaks; it is inactive at AP sites and all other 3′ blocking lesions (Vance and Wilson 2001a; Karumbati et al. 2003). The Rad1/Rad10 complex is a 3′ flap endonuclease that acts in nucleotide excision repair (NER) and during recombination (Davies et al. 1995). Genetic and biochemical evidence also suggests the involvement of Rad1/Rad10 in removing 3′ blocking lesions, including covalently bound Top1 protein, 3′ phosphates, 3′ phosphoglycolates, and 3′ aldehydes (Guzder et al. 1992; Guillet and Boiteux 2002; Vance and Wilson 2002; Karumbati et al. 2003). The exact molecular mechanism of these reactions is not clear, however.

Several recent studies have focused on the deleterious effects of spontaneous DNA lesions in haploid yeast when multiple repair pathways, especially BER and NER, have been compromised. Shuttling DNA damage to other repair pathways can often compensate for elimination of one pathway, so that many phenotypes are not observed with single repair mutants. Guillet and Boiteux (2002) showed that apn1 apn2 rad1 mutations were synthetically lethal due to accumulation of AP sites and associated strand breaks with 3′ blocked termini. This synthetic lethality was suppressed by deletion of UNG1, encoding uracil glycosylase, or overexpression of DUT1, encoding dUTP pyrophosphatase (Guillet and Boiteux 2002, 2003). Since both of these processes decrease uracil incorporation in DNA, the repair of deoxyuridine was interpreted to be the major source of AP sites in yeast. Swanson et al. (1999) showed that apn1 ntg1 ntg2 rad1 mutant yeast were slow growing and exhibited a strong mutator phenotype. apn1 ntg1 ntg2 rad1 mutants display oxidative DNA damage, elevated levels of intracellular reactive oxygen species, aberrant cellular morphologies, and transcriptional alterations, analogous to the cellular phenotypes exhibited in human diseases like cancer (Evert et al. 2004). All these studies identify AP sites as the predominant spontaneous lesion leading to toxicity in these strains.

We have found that a haploid strain lacking Apn1, Tpp1, and Rad1 is slow growing (Karumbati et al. 2003). Since Tpp1 acts specifically on 3′ phosphates, a major lesion leading to toxicity in this strain must be strand breaks with 3′ phosphate termini. The source of these lesions remains enigmatic. Reducing the cellular oxidative stress did not suppress the slow growth, nor did the deletion of Tdp1, the only yeast enzyme known to generate 3′ phosphates (Karumbati et al. 2003). Yeast do not have δ-lyases that can generate 3′ phosphates from AP sites (Karumbati et al. 2003).

Checkpoints respond to DNA damage by delaying the cell cycle at several stages. Damage encountered during S phase triggers checkpoints that promote the proper completion of replication, prevent cell division prior to complete replication, and likely also facilitate recombinational repair of the damage (Foiani et al. 1998; Rhind and Russell 2000; Kolodner et al. 2002). Damage encountered during late S/G₂ triggers distinct checkpoints that prevent mitosis in the presence of damage (Weinert et al. 1994, 2000; Sanchez et al. 1999). There is considerable overlap in the proteins required for these different checkpoints, with the kinases Mec1 and Rad53 having major roles in all cell cycle stages (Foiani et al. 2000). Other proteins show greater specificity. In particular, in budding yeast the primary role of the Chk1 kinase is to phosphorylate Pds1/securin and thereby execute a checkpoint at the metaphase-anaphase (M-A) transition (Sanchez et al. 1999; Wang et al. 2001).

In this study we found that tpp1 apn1 rad1 mutant yeast spontaneously gave rise to single-gene slow-growth suppressors at a high frequency. Using a candidate gene approach, we observed that disruption of the M-A checkpoint conferred suppression of tpp1 apn1 rad1 slow growth. In contrast, deletion of S-phase checkpoint genes either had no effect or conferred synthetic lethality in a tpp1 apn1 rad1 mutant background. The implications of these findings are discussed.

MATERIALS AND METHODS

Reagents:

Unless otherwise specified, chemicals were obtained from Sigma Chemical (St. Louis), restriction and DNA modifying enzymes from New England Biolabs (Beverly, MA), and oligonucleotides from Invitrogen (Carlsbad, CA).

Media and growth conditions:

Yeast were grown at 30° with shaking at 280 rpm for liquid cultures. Routine yeast media were either YPAD or synthetic defined media as described by Karumbati et al. (2003). Sporulation medium consists of 1% potassium acetate, 0.1% yeast extract, 0.05% dextrose, and any nutritional requirements of the diploid to 0.5× final concentration. The dissections and analysis of the asci were performed using a Singer MSM 200 system.

Yeast strains:

All yeast strains were isogenic derivatives of YW465; their genotypes are listed in Table 1. Candidate gene disruptions were created by one-step PCR-mediated gene replacement in diploids, in most cases in the tpp1/tpp1 apn1/apn1 RAD1/rad1 strain (YW1041). All disruptions were confirmed by PCR using flanking and KanMX4 primers. In cases where there was no apparent phenotype for the disruption, the gene replacement was also confirmed by the loss of product in PCR using primers in the coding region of the targeted gene. Standard techniques of mating and PCR gene replacement were used to create the base diploid strains used in these studies. Since the tpp1 apn1 rad1 haploid strain is slow growing and capable of generating suppressors, we did not grow it for any extended period of time.

TABLE 1.

Yeast strains used in this study

Strain	Genotypea	Derived from
YW1041	MATa-inc/MATα apn1Δ::HIS3/apn1Δ::HIS3 RAD1/rad1Δ::URA3
tpp1Δ::MET15/tpp1Δ::MET15
YW1135	MATa-inc/MATα RAD1/rad1Δ::URA3
YW1159	MATa-inc/MATα apn1Δ::HIS3/apn1Δ::HIS3 RAD1/rad1Δ::URA3
sml1Δ::ADE2/sml1Δ::ADE2 tpp1Δ::MET15/tpp1Δ::MET15
YW1066	RAD17/rad17Δ::kanMX4	YW1041
YW1116	DDC1/ddc1Δ::kanMX4	YW1041
YW1117	MEC3/mec3Δ::kanMX4	YW1041
YW1176	RAD24/rad24Δ::kanMX4	YW1041
YW1118	CHK1/chk1Δ::kanMX4	YW1041
YW1044	RAD9/rad9Δ::kanMX4	YW1041
YW1162	RAD9/rad9Δ::kanMX4	YW1135
YW1259	RAD9/rad9Δ::kanMX4	YW1159
YW1043	DUN1/dun1Δ::kanMX4	YW1041
YW1163	DUN1/dun1Δ::kanMX4	YW1135
YW1224	DUN1/dun1Δ::kanMX4	YW1159
YW1177	RAD53/rad53Δ::kanMX4	YW1159
YW1237	MEC1/mec1Δ::kanMX4	YW1159
YW1115	TEL1/tel1Δ::kanMX4	YW1041
YW1293	TOF1/tof1Δ::kanMX4	YW1041
YW1348	MRC1/mrc1Δ::ADE2	YW1041
YW1360	MRC1/mrc1Δ::ADE2 TOF1/tof1Δ::kanMX4	YW1348
YW1424	ESC4/esc4Δ::kanMX4	YW1041
YW1359	UNG1/ung1Δ::ADE2	YW1041
YW1377	UNG1/ung1Δ::ADE2 CHK1/chk1Δ::kanMX4	YW1359
YW1494	MATα wild-type strain containing the CEN/ARS plasmid, pRS412	YW465
YW1495	Contains the CEN/ARS plasmid pRS412 with ADE2 marker gene	YW1118

^aFor clarity, common marker alleles have been omitted from all strains. Thus, all strains are additionally ade2Δ0 his3Δ200 leu2 met15Δ0 trp1Δ63 ura3Δ0 (homozygous for diploid strains).

Analysis of tetrads:

The size of the colonies formed after growth of tetrad dissection plates was determined prior to genotyping according to the following scoring system: 5, wild-type colony growth; 4, intermediate between 5 and 3; 3, macrocolony visible after 3 days of growth; 2, microcolony (not macroscopically visible), >10 cells; 1, microcolony, 2–10 cells; and 0, single cell. The category 0 represents ungerminated spores and hence was not included in the interpretation of our data. The genotypes of viable colonies were determined by replica plating onto media for relevant markers. The genotypes of the inviable spores were inferred by the segregation pattern of the viable spores, when possible. Such genotypes are always underrepresented in the tables, however, since not all can be inferred this way.

Growth rate determination:

Five-day tetrad colonies were used to start YPAD cultures with an initial OD of 0.04. To obtain this OD in the case of slow-growing strains, two colonies were pooled to form the initial inoculum. The densities of the samples were then measured at 15- to 30-min intervals by using a GENESYS20 spectrophotometer for at least two doublings. The doubling time was then determined from the linear portion of the growth curve.

Plating efficiency:

Appropriate dilutions of yeast strains from a 5-day tetrad dissection plate were plated to YPAD. The plates were incubated overnight, the number of microcolonies and dead cells was counted microscopically, and the plating efficiency was determined. Any colony with <20 cells of abnormal morphology was regarded as dead. At least 500 colonies from at least six different isolates were analyzed. As with the tetrad scores, this was done prior to genotyping, so that all data were recorded blind to the strain genotype.

Plasmid loss assay:

Plasmid loss rates were determined using the method described by Huang and Koshland (2003). Briefly, YW1495, a diploid strain carrying the pRS412 plasmid with the CEN/ARS and the ADE2 marker gene, was dissected on selective media plates to get the strains with appropriate genotypes and containing the plasmid. Either a 6-day tetrad colony or a wild-type strain (YW465) carrying the plasmid was used to inoculate the initial culture and allowed to grow in YPAD for approximately six to seven doublings. In the case of slow-growing strains, three or more colonies were pooled to form the initial inoculum. The plasmid loss rate was calculated using the formula: loss rate = 1 − 10^m, where m = (log(F_f) − log(F_i))/n, F_i and F_f are the initial and final fraction of cells bearing the plasmid, respectively, and n is the number of doublings. The fraction of plasmid-bearing cells was determined by plating appropriate dilution of cells to both complete and selective synthetic media plates. The number of doublings was calculated from the initial and final colony counts on complete media plates.

Flow cytometric analyses:

Flow cytometry was performed on asynchronous yeast cells stained with propidium iodide. Yeast cells were grown in YPAD to early log phase, washed, and counted. About 1 × 10⁷ cells were resuspended in 1 ml 70% ethanol and fixed overnight at 4°. The cells were then pelleted, resuspended in 50 mm sodium citrate buffer (pH 7.5), and treated with RNase A (final concentration 0.25 mg/ml) for 1 hr at 50°. The cells were next treated with proteinase K (final concentration 1 mg/ml) for 1 hr at 50°. The cells were stained with propidium iodide (final concentration 16 μg/ml) and sonicated at 30 W [five pulses for 1–2 sec with 1- to 2-sec intervals using a Branson (Plainview, NY) Sonifer 250C]. DNA content was analyzed using a Beckman Coulter (Fuller, CA) Cytomics FC500. A total of 80,000 cells were analyzed for each histogram.

Fluorescent microscopy:

Yeast cells were grown in YPAD to early log phase. Approximately 1 × 10⁷ cells were pelleted, resuspended in 50 μl dH₂O, and fixed overnight by addition of 1 ml 70% ethanol. The cells were then washed and resuspended in 1 ml dH₂O and 1 μl of a 1 mg/ml DAPI stock solution was added. The suspension was incubated for 5 min at room temperature in the dark. Cells were subsequently washed twice in dH₂O and resuspended in 50 μl dH₂O. About 10 μl of suspension was then mounted on poly-l-lysine-coated slides for microscopy. Microscopy was performed using a Zeiss Axioskop with a Cool Snap-Pro camera and images were captured using the Image Pro Express software. These experiments again were performed blinded, and 200 cells from at least two different isolates were analyzed.

RESULTS

Spontaneous suppressors of tpp1 apn1 rad1 slow growth:

Yeast bearing combined tpp1 apn1 rad1 mutations grow very slowly due to their substantial deficiency in processing strand breaks with blocked 3′ termini (Karumbati et al. 2003). When tpp1 apn1 rad1 mutant yeast were grown on YPAD plates at 30°, we observed frequent spontaneous appearance of more rapidly growing colonies, although not to wild-type colony sizes (Figure 1A). These slow-growth suppressors were observed essentially every time the strain was streaked or plated and were not petite, and the growth phenotype was stable upon restreaking. We backcrossed eight independent tpp1 apn1 rad1 rapid growers with a tpp1 apn1 RAD1 strain. In the absence of a suppressor, tetrads always showed two large and two small colonies, corresponding to the genotypes tpp1 apn1 RAD1 and tpp1 apn1 rad1, respectively. In contrast, all suppressor dissections showed a mixture of tetrads with zero, one, or two tpp1 apn1 rad1 spores that grew to an intermediate colony size (similar to Figure 1B). This pattern repeated after a second round of backcrossing, indicating heritable suppression of tpp1 apn1 rad1 slow growth by unlinked single-gene mutations.

Figure 1.—

Slow-growth and suppression phenotypes of tpp1 apn1 rad1 yeast. (A) Four-day growth of control (tpp1 apn1) and tpp1 apn1 rad1 strains on YPAD plates. White arrows indicate the spontaneous suppressors. The tpp1 apn1 rad1 colonies that were not suppressed are very slow growing. (B) chk1 mutation is shown as a representative tpp1 apn1 rad1 slow-growth suppressor. The diploid strain YW1118 (tpp1/tpp1 apn1/apn1 RAD1/rad1 CHK1/chk1) was sporulated and dissected. Spore genotypes were determined by replica plating and are indicated below. Colony sizes of wild-type and tpp1 apn1 rad1 spores were designated as 5 and 3, respectively. Suppression of tpp1 apn1 rad1 slow growth by chk1 mutation increased the colony size from 3 to 4, but not to wild-type colony size.

On the basis of the frequency at which suppressors appeared and their varying sizes, we hypothesized that multiple mechanisms of tpp1 apn1 rad1 slow-growth suppression are likely. We therefore used an efficient genetic strategy to screen candidate suppressors. Specifically, known DNA repair, recombination, checkpoint, and various other genes were deleted in diploid strains followed by tetrad dissection to reveal phenotypes of the candidate mutation in combination with tpp1 apn1 and tpp1 apn1 rad1. In addition to revealing suppressors, this analysis allowed us to identify genes whose mutations are synthetically lethal because they impair still further redundant damage response pathways. Below we describe the results obtained with checkpoint and uracil glycosylase mutants. Recombination and other repair mutants will be described elsewhere. To simplify further studies, we developed the colony scoring system shown in Figure 1B (also see materials and methods and Karumbati et al. 2003). In this system, wild-type growth was designated as colony size 5, seen also for yeast with rad1 or tpp1 apn1 deletions. The tpp1 apn1 rad1 strain was a colony size 3. Suppressors were scored as colony size 4, although there was a mixture of colony sizes within this range. Finally, colony sizes 1 and 2 reflect inviable microcolonies. Results using this scoring metric are presented in a tabular format (Tables 2, 3, and 5–7) to facilitate strain comparison.

TABLE 2.

Tetrad dissection growth phenotype of candidate checkpoint genes

	Colony size (% of total)a
Genotype	1	2	3	4	5	Total no.
Wild typeb	1	1	98	327
rad1b	1	0.3	1	98	286
tpp1 apn1b	1	1	5	93	98
tpp1 apn1 rad1b	3	1	96	0.3	358
rad17 tpp1 apn1b	3	97	33
rad17 tpp1 apn1 rad1b	8	8	85	26
rad9 tpp1 apn1	100	31
rad9 tpp1 apn1 rad1	33	67	24
mec3 tpp1 apn1b	100	31
mec3 tpp1 apn1 rad1b	100	29
ddc1 tpp1 apn1	3	97	35
ddc1 tpp1 apn1 rad1	100	36
rad24 tpp1 apn1b	100	30
rad24 tpp1 apn1 rad1b	100	33
chk1 tpp1 apn1b	100	32
chk1 tpp1 apn1 rad1b	100	25
rad53 sml1 tpp1 apn1	100	11
rad53 sml1 tpp1 apn1 rad1	53	47	15
mec1 sml1 tpp1 apn1	13	88	40
mec1 sml1 tpp1 apn1 rad1	17	17	65	23
tel1 tpp1 apn1	100	19
tel1 tpp1 apn1 rad1	9	91	11
dun1 tpp1 apn1	5	95	37
dun1 tpp1 apn1 rad1	10	85	5	20
esc4 tpp1 apn1	14	86	14
esc4 tpp1 apn1 rad1	20	60	20	10
mrc1 tpp1 apn1	2	98	44
mrc1 tpp1 apn1 rad1	27	73	26
tof1 tpp1 apn1	100	34
tof1 tpp1 apn1 rad1	3	3	94	32
mrc1 tof1 tpp1 apn1	91	9	11
mrc1 tof1 tpp1 apn1 rad1	21	79				14

^aThe size of the colonies was determined according the scoring system in materials and methods (colony sizes 3, 4, and 5 are shown in Figure 1). Italic type indicates the predominant colony size observed for each genotype.

^bThese genotypes were repeated in many dissections, and only a subset are presented here. The pattern was the same in all dissections.

TABLE 3.

sml1 suppresses synthetic lethality ofrad9anddun1 withtpp1 apn1 rad1

	Colony size (% of total)a
Genotype	1	2	3	4	5	Total no.
sml1 tpp1 apn1	1	99	110
sml1 tpp1 apn1 rad1	1	3	97	115
rad9 tpp1 apn1 rad1	33	67	24
rad9 sml1 tpp1 apn1 rad1	5	95	20
dun1 tpp1 apn1 rad1	10	85	5	20
dun1 sml1 tpp1 apn1 rad1	7	13	80	15
chk1 tpp1 apn1 rad1	100	25
chk1 sml1 tpp1 apn1 rad1				90	10	10

^aThe size of the colonies was determined according the scoring system in materials and methods (colony sizes 3, 4, and 5 are shown in Figure 1). Italic type indicates the predominant colony size observed for each genotype.

Effect of checkpoint mutations on the tpp1 apn1 rad1 phenotype:

Checkpoint genes transmit DNA damage signals via a cascade, which includes sensors to detect DNA lesions (such as Rad17, a PCNA-like protein), mediator kinases to integrate and transmit signals (such as Mec1 and Tel1, homologs of human ATR and ATM, respectively), and effector kinases that set in motion downstream events (such as Rad53 and Chk1) (Melo and Toczyski 2002). The net effect is to cause delays in the cell cycle to allow time for repair. Because the tpp1 apn1 rad1 strain presumably has a high burden of persistent spontaneous DNA damage, it seemed likely that checkpoint function would be especially important in maintaining the viability of these cells. Initially we tested two candidate DNA damage checkpoint mutants, rad17 and rad9, expecting them to be synthetically lethal in a tpp1 apn1 rad1 background. rad9 was synthetically lethal, but, to our surprise, rad17 was in fact a suppressor (Table 2). To explore this phenomenon, we next tested nearly all known checkpoint genes and found an interesting pattern (Table 2). In addition to rad17, the mutants that acted as suppressors of tpp1 apn1 rad1 slow growth were mec3, ddc1, rad24, and chk1. Both Mec3 and Ddc1 are part of the Rad17 complex (Weinert et al. 2000). Rad24 is an RFC-like protein associated with the Rad17 complex (Majka and Burgers 2003). Chk1 is an effector kinase that acts downstream of these sensors in the M-A DNA damage checkpoint (Sanchez et al. 1999; Weinert et al. 2000). All these mutants conferred a colony size 4, with chk1 being the strongest suppressor. However, as with spontaneous suppressors, no mutant suppressed the tpp1 apn1 rad1 slow growth to wild-type colony size.

Mec1 and Rad53 are protein kinases that play central roles in coordinating the DNA damage response in yeast (Foiani et al. 2000; Rhind and Russell 2000). They also regulate dNTP production during S phase and, hence, DNA replication by inhibiting Sml1, itself a negative regulator of ribonucleotide reductase (RNR) (Zhao et al. 1998). Hence, mec1 and rad53 mutations are lethal, but this lethality is suppressed by further sml1 mutation (Zhao et al. 1998). In our studies, deletion of RAD53 in the tpp1 apn1 rad1 strain was synthetically lethal even in an sml1 background (Table 2). Deletion of SML1 alone had no effect on the growth of tpp1 apn1 rad1 or wild-type strains (Table 3), showing that Rad53 is required for the residual growth of tpp1 apn1 rad1 mutants. Interestingly, deletion of MEC1 had little effect on tpp1 apn1 rad1 slow growth. Most mec1 sml1 tpp1 apn1 rad1 colonies were still size 3, although a significant minority (8 of 23, 34%) were inviable (Table 2).

Two other checkpoint kinases were also mutated. Tel1 is related to Mec1 and homologous to the human gene ATM (Sanchez et al. 1996). Its deletion had no effect on tpp1 apn1 rad1 slow growth (Table 2). Dun1 is a kinase that acts downstream of Rad53 (Zhao and Rothstein 2002). Deletion of DUN1 resulted in lethality in tpp1 apn1 rad1 strains (Table 2).

Esc4 has recently been identified as one of the downstream S-phase-specific targets of Mec1, responsible for resumption of chromosome replication after DNA damage in a Rad53-independent manner (Rouse 2004). Though mec1 mutants had a minor effect, the esc4 mutant was synthetically lethal with tpp1 apn1 rad1 (Table 2). Mrc1 and Tof1 function redundantly as S-phase mediators of stable replication fork pausing (Katou et al. 2003; Zegerman and Diffley 2003). Deletion of TOF1 and MRC1 individually had no effect but their combined mutation was lethal in the tpp1 apn1 rad1 background (Table 2). It was interesting to note that the deletion of DUN1 or RAD9 conferred no growth defect either alone or in a tpp1 apn1 or rad1 strain (Table 2; data not shown). In contrast, deletion of RAD53, MRC1/TOF1, or ESC4 decreased the tpp1 apn1 colony size from 5 to 4 (Table 2).

The dun1 and rad9 effect is due to altered dNTP pools:

Chabes et al. (2003) have shown a correlation between increased survival after DNA damage and dNTP levels. To investigate if the tpp1 apn1 rad1 suppression/lethality observed in checkpoint-deficient strains was due to modulation of dNTP levels, we additionally deleted sml1 in select strains, as already done for rad53 and mec1. As seen in Table 3, the synthetic lethality of dun1 and rad9 with tpp1 apn1 rad1 was suppressed by deleting SML1 such that colonies were now the same size as the tpp1 apn1 rad1 triple mutant alone. Thus, it appears that decreased dNTP levels exacerbate the tpp1 apn1 rad1 phenotype, and also that dun1 and rad9 have little effect on this phenotype once the dNTP levels are corrected. In contrast, no further increase in colony size was observed when SML1 was deleted in a chk1 tpp1 apn1 rad1 background (Table 3), which was expected since the Chk1 checkpoint is not known to affect RNR activity (Zhu and Xiao 2001). Combined with the observation that sml1 itself is not a tpp1 apn1 rad1 suppressor (Table 3), it is clear that nucleotide pools alone cannot account for tpp1 apn1 rad1 growth suppression.

G₂/M arrest of tpp1 apn1 rad1 yeast:

To further characterize the mechanisms of tpp1 apn1 rad1 slow growth and suppression, we next determined the doubling time, cellular and nuclear morphology, DNA content, plating efficiency, and chromosome instability of various strains. For clarity, only experiments with our strongest suppressor, chk1, are presented, but many results were similar with other suppressors (data not shown).

As shown in Figure 2A and also consistent with the colony sizes, the tpp1 apn1 rad1 strain was indeed very slow growing with a doubling time of 175 min as compared to 86 min for the wild-type strain. The tpp1 apn1 strains had doubling times comparable to the wild-type strain (92 min). chk1 mutation decreased the tpp1 apn1 rad1 doubling time but not to wild-type levels (133 min), again consistent with its colony size 4.

Figure 2.—

Effect of M-A checkpoint disruption on tpp1 apn1 rad1 growth, plating efficiency, and chromosome stability. (A) Doubling times of the indicated strains in rich medium (YPAD). Error bars represent standard errors of the mean obtained from three independent experiments. (B) Plating efficiencies of the indicated strains. Error bars represent standard errors of the mean obtained from six to eight independent experiments. (C) CEN plasmid loss rates for the indicated yeast strains. Error bars represent standard errors of the mean obtained from at least five isolates/pools.

tpp1 apn1 rad1 cells were observably larger than typical wild-type cells (Figure 3A; data not shown). Moreover, the triple mutant showed more cells with large or abnormal buds (52 and 21%, respectively, as compared to 42 and 0% in the wild-type strains; Table 4). The designation “abnormal” refers to cells, which were elongated, branched, or bifurcated or had multiple buds from a single mother, or both. Examples of such cells are shown in Figure 3B. These contrast with the monotonous round to oval single-budded wild-type cells (Figure 3A; data not shown). A similar pattern of morphologies was observed by Evert et al. (2004) for apn1 rad1 ntg1 ntg2 mutant yeast. Overall, it suggests a G₂/M arrest. DAPI visualization of the nucleus revealed that nearly 10% of the tpp1 apn1 rad1 cells had a single nucleus stretched between the mother and large daughter cells (Table 4; Figure 3A). This has been called a bow-tie nucleus and again reflects a G₂/M or mitotic arrest (McVey et al. 2001). Mutations in the suppressor checkpoint genes visibly decreased the cell size (Figure 3A; data not shown) and significantly decreased the number of abnormal buds and bow-tie nuclei (Table 4) as compared to the tpp1 apn1 rad1 cells. However, even chk1 did not completely reduce the number of large-budded cells to wild-type levels. DNA content experiments were consistent with the microscopic findings in that tpp1 apn1 rad1 yeast displayed a major peak of G₂/M DNA content (Figure 3A). There was also an increased heterogeneity in the DNA content distribution. chk1 restored a more normal distribution of DNA content with more cells in G₁, although again not to wild-type levels. Thus, suppression by checkpoint dysfunction appears to partially relieve the G₂/M arrest in tpp1 apn1 rad1 cells.

Figure 3.—

Morphology and DNA content of tpp1 apn1 rad1 yeast. (A) Exponentially growing cultures of the different strains were fixed and stained with either DAPI or PI for morphological and DNA content analyses, respectively (see materials and methods). The same magnification was used for all strains. The tpp1 apn1 rad1 cells are G₂/M arrested with large cellular and abnormal nuclear morphology as compared to the wild-type cells. White arrows indicate the bow-tie nuclei. chk1 mutation overrode this arrest, but neither the cell cycle nor the morphology was restored fully to wild type. (B) Examples showing the abnormal cellular morphology exhibited by tpp1 apn1 rad1 cells.

TABLE 4.

Cellular and nuclear morphology oftpp1 apn1 rad1 and suppressor strains

	% cellsa
Genotype	No bud	Small bud	Large bud	Abnormalb	Bow-tie nuclei
Wild type	39 ± 1	21 ± 3	41 ± 2	0 ± 0	3 ± 0
rad1	33 ± 2	24 ± 3	42 ± 5	1 ± 0	3 ± 1
tpp1 apn1	35 ± 2	19 ± 1	46 ± 3	0 ± 0	3 ± 0
tpp1 apn1 rad1	16 ± 3	10 ± 4	52 ± 7	21 ± 1	10 ± 0
chk1 tpp1 apn1	34 ± 1	21 ± 3	44 ± 2	1 ± 0	3 ± 1
chk1 tpp1 apn1 rad1	21 ± 4	15 ± 8	58 ± 3	6 ± 6	5 ± 1

^aMean ±SD of at least 200 cells from two independent isolates.

^bSome examples of cells that were scored abnormal are shown in Figure 3B.

The plating efficiency of the tpp1 apn1 rad1 strain was decreased as compared to the control tpp1 apn1 strain (from 90 to 64%; Figure 2B). The chk1 mutant did not restore the plating efficiency to wild-type levels, although the increase to 82% was significant. Thus, suppression does increase the number of dividing cells in the population as opposed to simply decreasing the cell cycle time. Among the viable microcolonies formed after overnight incubation, wild-type colonies were predominantly larger (> ∼150 cells) while tpp1 apn1 rad1 colonies were predominantly smaller (> ∼40 and < ∼150 cells). Interestingly, the checkpoint suppressors gave rise to a mixture of smaller and larger microcolonies (data not shown).

We also assessed the chromosome stability in these strains by monitoring loss of a CEN plasmid. As shown in Figure 2C, the plasmid loss rate significantly increased in the tpp1 apn1 rad1 triple-mutant strain as compared to the wild type. Adding a chk1 mutation increased the plasmid loss rate still further to a small extent. Thus, unlike all other parameters monitored, chk1 did not suppress but in fact exacerbated the chromosome instability of tpp1 apn1 rad1 yeast.

pds1 suppresses tpp1 apn1 rad1 slow growth:

The G₂/M arrest phenotype of tpp1 apn1 rad1 strains combined with the prominence of bow-tie nuclei led us to hypothesize that impairing the anaphase checkpoint (by pds1 mutation) (Yamamoto et al. 1996) or the mitotic spindle checkpoint (by mad2 or bub2 mutation) (Gardner and Burke 2000) might allow mitotic progression and hence also lead to suppression of slow growth. Indeed, deletion of PDS1 did suppress tpp1 apn1 rad1 slow growth (Table 5), consistent with the fact that Pds1 is phosphorylated by Chk1 in the M-A checkpoint (Wang et al. 2001). However, deletion of BUB2 was surprisingly lethal in the tpp1 apn1 rad1 background, while deletion of MAD2 had no effect (Table 5).

TABLE 5.

Tetrad dissection growth phenotype of mitotic checkpoint genes

	Colony size (% of total)a
Genotype	1	2	3	4	5	Total no.
pds1 tpp1 apn1b	10	90	31
pds1 tpp1 apn1 rad1b	8	88	4	26
bub2 tpp1 apn1	100	8
bub2 tpp1 apn1 rad1	57	43	7
mad2 tpp1 apn1	7	93	30
mad2 tpp1 apn1 rad1		7	93			27

^aThe size of the colonies was determined according the scoring system in materials and methods (colony sizes 3, 4, and 5 are shown in Figure 1). Italic type indicates the predominant colony size observed for each genotype.

^bThese plates were incubated at 25° for 6 days because pds1 strains are inviable at 30°.

Epistasis analysis between checkpoint and ung1 genes:

Guillet and Boiteux (2002) previously reported ung1 to be a suppressor of apn1 apn2 rad1 lethality and argued that uracil glycosylase represents the main source of spontaneous AP sites in yeast (Guillet and Boiteux 2002). Our system overlaps with theirs in that our strains are also apn1 rad1 mutant. It is therefore conceivable that the phenotypes, described above, could result in part from impaired processing of AP sites in addition to 3′ phosphates. We explored the relationship between ung1 and checkpoint suppressors in both the tpp1 apn1 rad1 and the apn1 apn2 rad1 backgrounds to determine the relative contribution of AP sites in the various scenarios. ung1 mutation did suppress tpp1 apn1 rad1 slow growth, but the effect was less pronounced than that of even the weakest checkpoint suppressors, with 16 of 45 (36%) ung1 tpp1 apn1 rad1 spores still scoring as colony size 3 in our blind analysis (Table 6). Epistasis was not observed between chk1 and ung1, such that the chk1 ung1 combination further suppressed tpp1 apn1 rad1 slow growth to the wild-type rate (Table 6). The other checkpoint suppressors showed a similar pattern (data not shown). ung1 itself also had no effect on the abnormal DNA content distribution of tpp1 apn1 rad1 cells, unlike chk1 (data not shown). Like Guillet and Boiteux (2002), we also observed that ung1 suppressed apn1 apn2 rad1 lethality, although this effect was weak in our strain background (Table 7). In contrast, the checkpoint suppressors did not act as suppressors of apn1 apn2 rad1 lethality, as exemplified by chk1 (Table 7). Thus, chk1 and ung1 exemplify distinct mechanisms of tpp1 apn1 rad1 suppression, which suggests that suppression by checkpoint dysfunction is independent of AP sites.

TABLE 6.

Epistasis analyses betweenchk1 andung1

	Colony size (% of total)a
Genotype	1	2	3	4	5	Total no.
ung1 tpp1 apn1b	3	97	61
ung1 tpp1 apn1 rad1b	9	36	56	45
chk1 tpp1 apn1b	100	32
chk1 tpp1 apn1 rad1b	100	25
chk1 ung1 tpp1 apn1b	7	93	14
chk1 ung1 tpp1 apn1 rad1b					100	18

^aThe size of the colonies was determined according the scoring system in materials and methods (colony sizes 3, 4, and 5 are shown in Figure 1). Italic type indicates the predominant colony size observed for each genotype.

^bThese genotypes were repeated in many dissections, and only a subset are presented here. The pattern was the same in all dissections.

TABLE 7.

ung1, but not checkpoint deficiency, suppressesapn1 apn2 rad1 lethality

	Colony size (% of total)a
Genotype	1	2	3	4	5	Total no.
apn1 apn2b	16	84	74
apn1 apn2 rad1b	32	63	5c	116
ung1 apn1 apn2	5	95	22
ung1 apn1 apn2 rad1	22	15	63c	27
chk1 apn1 apn2b	6	94	34
chk1 apn1 apn2 rad1	43	43	14c			21

^aThe size of the colonies was determined according the scoring system in materials and methods (colony sizes 3, 4, and 5 are shown in Figure 1). Italic type indicates the predominant colony size observed for each genotype.

^bThese genotypes were repeated in many dissections, and only a subset are presented here. The pattern was the same in all dissections.

^cThese colonies were much smaller than the tpp1 apn1 rad1 colonies although they were macroscopically visible at 3 days.

DISCUSSION

Differential suppression of AP site and strand break toxicity:

Like other recent studies (Swanson et al. 1999; Guillet and Boiteux 2002, 2003; Evert et al. 2004), this work has used a yeast strain deficient in multiple DNA repair pathways as a means of exploring the disposition of spontaneous DNA damage. Our genotype, tpp1 apn1 rad1, is very similar to the apn1 apn2 rad1 genotype studied by Guillet and Boiteux (2002). In both strains, multiple DNA lesions will be inefficiently repaired because of the diverse activities of Apn1, Apn2, and Rad1/Rad10 (Guillet and Boiteux 2004). For spontaneous damage, a critical consideration is the differential contributions of two lesions: AP sites and strand breaks with blocked 3′ ends. In this context, the important function of Rad1/Rad10 is not NER but its poorly understood ability to remove 3′ blocking lesions, since other NER mutations do not have the same effects (data not shown; Guillet and Boiteux 2002). apn1 apn2 rad1 yeast have a complete AP endonuclease deficiency and are therefore strongly affected by AP sites and strand breaks with 3′ aldehydes created from AP sites by β-lyases (Johnson et al. 1998; Guillet and Boiteux 2002). In contrast, tpp1 apn1 rad1 strains have residual AP endonuclease activity (Apn2), but lack the two dominant 3′ phosphatases, Tpp1 and Apn1 (Vance and Wilson 2001a). Because Tpp1 is rigidly specific for 3′ phosphates (Vance and Wilson 2001a; Karumbati et al. 2003) and because yeast lack δ-lyases to convert AP sites to 3′ phosphates (Karumbati et al. 2003), phenotypes dependent on tpp1 must be due in large measure to deficient processing of primary strand breaks. Beyond this distinct lesion susceptibility, tpp1 apn1 rad1 yeast were useful in that their residual growth allowed for the observation of frequent spontaneous slow-growth suppressors.

Suppression of a DNA repair phenotype might be due to a reduced incidence of a DNA lesion, activation of an alternative repair mechanism, or tolerance to the negative effects of the lesion. The first mechanism is relevant to ung1, which suppresses both apn1 apn2 rad1 lethality and tpp1 apn1 rad1 slow growth by preventing the cleavage of genomic uracil and thereby greatly limiting the occurrence of AP sites (Guillet and Boiteux 2003). In contrast, the M-A checkpoint effect appears to be mediated through primary strand breaks as opposed to AP sites. Specifically, the additive suppression seen in chk1 ung1 double mutants, combined with the persistent G₂/M arrest phenotype of ung1 tpp1 apn1 rad1 yeast, establishes that an important component of the tpp1 apn1 rad1 growth defect is dependent on checkpoint function but not Ung1. This conclusion is supported by the fact that, unlike ung1, checkpoint mutants did not suppress apn1 apn2 rad1 lethality.

Two effects of checkpoint mutations on tpp1 apn1 rad1 slow growth:

We were initially surprised that disruption of DNA damage checkpoints could confer both synthetic lethality and slow-growth suppression in combination with tpp1 apn1 rad1 mutations, since checkpoints are typically associated with resistance to DNA damage. The observed genetic interactions show a clear pattern consistent with the dual nature of checkpoint signaling, however (Figure 4A). Specifically, the Chk1 and Rad53 protein kinases represent two distinct downstream checkpoint pathways that differentially affect the cellular response to DNA damage (Sanchez et al. 1999; Zhu and Xiao 2001). Mec1 and Tel1 are the master kinases that coordinate both of these pathways, with Mec1 being more significant in yeast (Sanchez et al. 1996). The Mec1-Rad53 kinase cascade is most critical for the S-phase checkpoint, although it also has roles in other cell cycle phases such as G₂/M (Allen et al. 1994; Longhese et al. 2003). In contrast, in yeast the Mec1-Chk1 kinase cascade appears to function principally in the M-A checkpoint (Sanchez et al. 1999; Foiani et al. 2000). The fact that rad53 mutation was synthetically lethal in our system while chk1 was a slow-growth suppressor (Figure 4A) suggests distinct actions of the S and M-A checkpoints in the tpp1 apn1 rad1 background.

Figure 4.—

Models of the disposition of persistent strand breaks. (A) In this depiction of known DNA damage checkpoint pathways, ovals are used to indicate checkpoint proteins whose loss leads to suppression of slow-growth phenotype of tpp1 apn1 rad1 yeast. Rectangles are used to indicate checkpoint proteins whose loss leads to synthetic lethality in tpp1 apn1 rad1 yeast. These correspond to the M-A and S-phase checkpoint pathways, respectively. (B) There are various outcomes when the replication machinery encounters a persistent 3′ phosphate lesion (see text for discussion).

It was striking how little effect mec1 mutation had in our assay. However, this can be explained by the fact that it activates checkpoint pathways with opposing effects on tpp1 apn1 rad1 slow growth, i.e., Rad53 and Chk1. These effects appear to offset each other, although early catastrophic events likely lead to the minority of failed mec1 sml1 tpp1 apn1 rad1 colonies. Also, Tel1 might help promote survival in the absence of Mec1. We did not attempt to create a mec1 tel1 double mutation in the tpp1 apn1 rad1 background because mec1 tel1 strains are by themselves very slow growing and genetically unstable (Craven et al. 2002).

Synthetic lethality and the S-phase checkpoint:

Beyond rad53, other checkpoint mutants that were synthetically lethal with tpp1 apn1 rad1 also function in the S-phase checkpoint (Figure 4A). The S-phase checkpoints respond to DNA damage and other influences that impede the progression of replication forks. The Mec1/Rad53-dependent system stabilizes forks so that they can resume replication, delays the firing of late origins, and imposes an anaphase block until replication is completed (Lopes et al. 2001; Agarwal et al. 2003; Clarke et al. 2003). Tof1 and Mrc1 function redundantly in a pausing complex that prevents the uncoupling of the replication machinery when replication is arrested (Katou et al. 2003; Zegerman and Diffley 2003). Esc4 is a target of Mec1 that is not required for the S-phase checkpoint per se, but is essential for resumption of replication after DNA damage (Rouse 2004). The synthetic lethality of esc4 and tof1 mrc1 with tpp1 apn1 rad1 is therefore consistent with a high burden of stalled forks in this background. Importantly, we cannot distinguish whether such fork failure is due to persistent AP sites or strand breaks. In either case, a further role of the S-phase checkpoint is likely to promote the repair of DNA damage and the restart of collapsed replication forks by recombination (Foiani et al. 1998; Rhind and Russell 2000; Kolodner et al. 2002). Indeed, rad52 mutation confers synthetic growth defects with tpp1 apn1 and tpp1 apn1 apn2 mutations (Vance and Wilson 2001a). Moreover, recombination acts independently of Rad1 in the tpp1 apn1 background and is required for chk1-mediated suppression of tpp1 apn1 rad1 slow growth (A. S. Karumbati and T. E. Wilson, unpublished results).

One specific action of the S-phase checkpoint is to increase dNTP pools by inhibiting the RNR inhibitor Sml1, an effect mediated through the kinase Dun1 (Figure 4A; Zhao and Rothstein 2002). sml1 mutation suppressed the synthetic lethality of dun1 tpp1 apn1 rad1 yeast. This demonstrates that depleting dNTP pools itself exacerbates the tpp1 apn1 rad1 slow-growth phenotype and also that RNR-independent roles of Dun1, such as transcriptional activation, are unimportant. dNTP pools cannot explain all S-phase checkpoint phenotypes, however, as rad53 mutation was lethal despite being studied in the sml1 background. Rad9 is the most difficult to interpret. Rad9 acts in coordination with Rad53 in many checkpoints (Toh and Lowndes 2003), and yet rad9 tpp1 apn1 rad1 lethality was suppressed by sml1 mutation. This demonstrates that Rad9 participates in the regulation of dNTP pools. It also suggests that the Rad53 and Rad9 have distinct contributions to different checkpoints, and, indeed, Rad9 contributes to Chk1 activation in a manner separable from Rad53 activation (Blankley and Lydall 2004).

Slow-growth suppression and the M-A checkpoint:

In contrast to the synthetic lethality conferred by rad53 and other S-phase checkpoint mutants, deficiencies in the M-A DNA damage checkpoint acted as tpp1 apn1 rad1 slow-growth suppressors. This checkpoint consists of the Rad17 complex, Rad24, Mec1, Chk1, and Pds1 (Figure 4A; Weinert et al. 2000; Wang et al. 2001). The Rad17 complex (Rad17, Mec3, and Ddc1) and Rad24 are sensors of DNA damage responsible for transducing the damage signal to Mec1 (Weinert et al. 2000). They are less important for recognizing damage in S phase, but are essential in G₂/M (Weinert et al. 1994). Mec1 phosphorylates Chk1, which in turn phosphorylates the anaphase inhibitor Pds1/securin (Wang et al. 2001; Rouse and Jackson 2002). This blocks the targeted degradation of Pds1 by the anaphase promoting complex, thereby preventing mitotic progression (Agarwal et al. 2003). The fact that deleting any gene in this checkpoint pathway (except MEC1, see above) led to slow-growth suppression demonstrates its importance to the tpp1 apn1 rad1 phenotype. Indeed, tpp1 apn1 rad1 yeast displayed a G₂/M arrest phenotype that was significantly, but not completely, overridden by chk1 mutation.

It is important to note that Rad53 also acts in G₂/M, where it inhibits mitotic exit after DNA damage and prevents inappropriate mitosis in the presence of unreplicated DNA. Indeed, at least part of this effect is mediated through stabilization of Pds1 (Sanchez et al. 1999). In our system this function of Rad53 is either secondary to Chk1 or less important than the deleterious effect of Rad53 loss on the S-phase checkpoint. It may partially explain the incomplete rescue of G₂/M arrest by chk1 mutation, however.

Because of the prominent G₂/M arrest of tpp1 apn1 rad1 cells it was also important to consider the mitotic or spindle checkpoints. These checkpoints arrest cells that have failed to achieve bipolar spindle attachment to replicated chromatids, but have also been implicated as a backup mechanism of DNA damage surveillance (Sorger and Gillett 2001). There are essentially two spindle checkpoint pathways: the Bub2 pathway inhibits mitotic exit while the Mad2 pathway is involved in the regulation of Pds1 (Gardner and Burke 2000). We initially considered that disruption of these pathways might suppress tpp1 apn1 rad1 slow growth similar to chk1. bub2 mutation was lethal in combination with tpp1 apn1 rad1, however, while mad2 mutation had no effect. Thus, the spindle checkpoint again seems to act secondarily to Chk1, and, indeed, Chk1 directly blocks the ubiquitination of Pds1 (Agarwal et al. 2003). Together, these results emphasize the specificity of the suppression phenotype caused by loss of the Chk1/securin pathway.

Model for tolerance of persistent strand breaks:

We must finally consider whether disruption of the Chk1-Pds1 axis confers suppression by activation of strand-break repair or by tolerance of persistent DNA damage. Pds1/securin is the inhibitory partner of separase. Degradation of securin in metaphase thus activates separase, which then cleaves cohesin, triggers its degradation, and allows chromatid separation (Nasmyth 2001). Chk1 blocks this sequence by phosphorylating and stabilizing Pds1 (Sanchez et al. 1999; Wang et al. 2001). Recently, Nagao et al. (2004) presented evidence in Schizosaccharomyces pombe that cleavage of cohesin may also be required for recombination-based repair of DNA damage, possibly to allow the repair machinery to access the lesion (Nagao et al. 2004). By analogy, pds1 mutation in S. cerevisiae or Pds1 destabilization by chk1 mutation might facilitate repair of persistent strand breaks by allowing more rapid and/or extensive cohesin loss. However, adding chk1 mutation to a tpp1 apn1 rad1 strain did not suppress but exacerbated loss of a CEN plasmid. Although we cannot exclude it, this result argues against a model in which loss of the M-A checkpoint enhances repair and therefore genome stability.

In contrast, Figure 4B illustrates how recombination combined with an impaired M-A checkpoint could lead to tolerance of persistent SSBs. Replication fork collision with a nick blocked by a persistent 3′ phosphate (structure 1) would result in the aberrant fork collapse (structures 2 or 3). Recombination is inhibited at such structures because either the parent strand bearing the 3′ phosphate must invade its sister (structure 4) or the newly synthesized strand must invade its sister containing the persistent nick (not drawn). In structure 3, replication might continue to copy the entire chromosome (structure 5), which might again require cohesin loss, but the triplicate chromosome would present segregation difficulties during mitosis. Structure 2 is susceptible to formation of a DSB if fork collapse occurs from both sides (structure 6). Now the newly synthesized strand can invade its fully replicated sister and synthesize past the site of the lesion (structure 7). Resolution of this intermediate by synthesis-dependent strand annealing (Paques and Haber 1999) would leave one intact chromosome and one containing a 3′ phosphorylated single-strand gap (structure 8). Precedent for such single-ended invasion and resolution has been provided for DSBs blocked by a single nonhomologous tail (Colaiacovo et al. 1999). Other than the gap, structure 8 is fully replicated. It should elicit a DNA damage signal, however. Abrogation of the M-A checkpoint could make the cell unresponsive to this signal, allowing propagation of the genome despite the persistence of blocked 3′ termini.