If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
* This work was supported by National Science Foundation (NSF) Grant MCB9728782 and National Institutes of Health (NIH) Grant CA82633 (to G. F. M.) and by NSF Grants MCB9916576 and MCB0137060 (to C. K. M.). This publication was made possible, in part, by the Cell Tissue and Analysis Core of the Environmental Health Sciences Center, Oregon State University, NIEHS, NIH, Grant P30 ES00210. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. § Supported by a predoctoral fellowship from the government of Turkey.
The relationship between dNTP levels and DNA synthesis was investigated using α factor-synchronized yeast treated with the ribonucleotide reductase inhibitor hydroxyurea (HU). Although HU blocked DNA synthesis and prevented the dNTP pool expansion that normally occurs at G1/S, it did not exhaust the levels of any of the four dNTPs, which dropped to about 80% of G1 levels. When dbf4 yeast that are ts for replication initiation were allowed to preaccumulate dNTPs at 37 °C before being released to 25 °C in the presence of HU, they synthesized 0.3 genome equivalents of DNA and then arrested as dNTPs approached sub-G1 levels. Accumulation of dNTPs at G1/S was not a prerequisite for replication initiation, since dbf4 cells incubated in HU at 25 °C were able to replicate when subsequently switched to 37 °C in the absence of HU. The replication arrest mechanism was not dependent on the Mec1/Rad53 pathway, since checkpoint-deficient rad53 cells also failed to exhaust basal dNTPs when incubated in HU. The persistence of basal dNTP levels in HU-arrested cells and partial bypass of the arrest in cells that had preaccumulated dNTPs suggest that cells have a mechanism for arresting DNA chain elongation when dNTP levels are not maintained above a critical threshold.
). The simplest explanation for HU inhibition of DNA synthesis is that it starves the DNA polymerase at the replication forks for dNTPs. HU treatment has been shown to reduce the purine dNTP pools in a variety of mammalian cells (
); however, conflicting data exist concerning its modulation of pyrimidine dNTP pool levels. Furthermore, even for purine dNTPs, HU has only rarely been shown to cause a complete depletion of the dGTP or dATP pools (
). The complicated, often reciprocal, changes in individual dNTP pools that occur in HU-treated mammalian cells may be due to the compensatory activities of deoxyribonucleotide salvage pathways in higher eukaryotes. Budding yeast offers a simpler system in which to study the mechanism by which HU affects replication. Yeast possess no deoxyribonucleoside kinase activities, and thus deoxyribonucleotide synthesis is entirely dependent on ribonucleotide reductase. Also, yeast can easily be synchronized in G1 using mating pheromone, and the availability of several temperature-sensitive cdc mutations allows cell cycle progression to be reversibly halted at specific points throughout the cell cycle (
) showed that RNR1 gene mutations that free RNR enzyme from allosteric feedback inhibition result in elevated dNTP levels and increased resistance to DNA-damaging agents. There is no published information on the effects of HU on dNTP levels in yeast.
In mammalian cells, the mechanism by which HU inhibits replication has been investigated by attempting to reverse HU inhibition by administration of exogenous deoxyribonucleosides (
). In yeast, the absence of deoxyribonucleotide salvage enzymes makes it difficult to assay the ability of exogenous deoxyribonucleosides to reverse the effects of HU. However, the availability of cdc mutants that cannot initiate S phase at the nonpermissive temperature represents a potential means of genetically elevating dNTP levels prior to application of HU. The Dbf4 polypeptide, together with Cdc7, forms a heterodimeric protein kinase that is essential for replication initiation (
), experiments have shown that dNTPs accumulate as synchronized cells enter S phase. The concomitant accumulation of dNTPs and synthesis of DNA suggests that there might be a link between the initiation of these two events. In the current study, we tested whether a threshold level of dNTPs was needed for replication initiation by releasing α factor-arrested dbf4 yeast into HU medium at the permissive temperature and then monitoring their ability to replicate when switched to the nonpermissive temperature in the absence of HU. A similar reciprocal switch experiment in which DNA synthesis and cell division were monitored has already established that Cdc7, the protein with which Dbf4 interacts, acts prior to the HU-sensitive step during the cell cycle (
). This suggests that dNTP accumulation is not required for replication initiation. However, in this and all other previous reciprocal switch experiments involving HU and cdc mutations, dNTP levels were never measured. Furthermore, although Cdc7 has been analyzed, that Dbf4 acts prior to HU in preventing replication has not formally been shown.
The checkpoint kinases Mec1 and Rad53 act sequentially in a pathway that, among other things, appears to boost or preserve dNTP levels in cells treated with DNA-damaging agents or HU. In response to DNA damage, the Mec1/Rad53 pathway positively regulates RNR activity in at least three ways. It stimulates transcription of the large and small subunit RNR genes (
). The occurrence of DNA synthesis at late origins in rad53 cells suggested that HU-treated cells must have a residual pool of dNTPs. To determine the extent to which dNTP pools fall in HU-treated cells, the levels of all four dNTPs were determined in synchronized wild-type and rad53 cells incubated in the absence and presence of HU.
Our analyses showed that, although HU prevented dNTP accumulation at G1/S and blocked DNA synthesis, it did not exhaust the basal dNTP pools. HU inhibition of replication was bypassed if dbf4 cells were allowed to preaccumulate dNTPs at the nonpermissive temperature prior to exposure to the drug at the permissive temperature. Accumulation of dNTPs was not a prerequisite for replication initiation, since dbf4 cells released from α factor into HU at the permissive temperature were able to replicate when switched to the nonpermissive temperature in the absence of HU. Finally, the arrest mechanism was not dependent on the Mec1/Rad53 pathway, since HU treatment did not exhaust the dNTP pools in checkpoint-deficient rad53 mutants. Our results suggest that cells have evolved a mechanism for arresting DNA chain elongation when dNTP falls below a minimum threshold. Elongation arrest may represent a means of conserving basal dNTPs for processes such as DNA repair.
Media, Strains, and General Methods—Cells were grown in YEPD medium. Yeast strains used are listed in Table I. MY273 and SY2626 are derivatives of W303-1 that carry a bar1 mutation to facilitate α factor synchronization. To derive MY273, W303-1α was mated to SY2626 (G. Sprague, University of Oregon, Eugene, OR), and a MATα segregant carrying the bar1 allele was isolated by tetrad dissection. To derive the dbf4-1 bar1 strain MY317, MY273 was mated to L128-20 (L. Johnston, National Institute for Medical Research, Mill Hill, London, UK), and random spores from the resulting diploid were screened for α factor and temperature sensitivity. To derive the rad53-11 bar1 strain MY376, MY273 was mated to BS883 (R. Sclafani, University of Colorado Medical Center, Denver, CO), and tetrads from the resulting diploid were dissected and tested for α factor sensitivity and uracil prototrophy. (The rad53-11 mutation was linked to pep4:URA3.) The wild-type RAD53 strain (MY377) was derived from the same tetrad used to derive MY376.
G1-arrested cells were obtained by incubating asynchronously growing cells with 100 ng/ml α factor for 3 h at 25 °C. To achieve synchrony, arrested cells were collected on Whatman #1 filters, washed once with fresh YEPD and once with 100% conditioned medium, resuspended in 25% conditioned medium, and incubated at either 25 or 37 °C. Conditioned medium was prepared by growing W303-1a cells to saturation in YEPD and removing the cells by filtration.
Flow Cytometry—About 5 × 106 cells were collected by centrifugation and fixed in 70% ethanol overnight. After microcentrifugation (10,000 rpm, 5 s), the cell pellet was washed with 100 μlof50mm sodium citrate buffer (pH 7), resuspended in 100 μl of sodium citrate buffer, and sonicated for 6 s. After a 1-h incubation at 55 °C with RNase A (0.25 mg/ml) and a 1-h incubation with proteinase K (1 mg/ml), 100 μl of sodium citrate buffer containing 16 μg/ml propidium iodide was added, and samples were incubated at 4 °C overnight in the dark. A Beckman Coulter Epics XL flow cytometer was used to analyze the samples. For each histogram, 25,000 cells were assayed.
dNTP Pool Measurements—Approximately 3 × 108 cells were harvested and extracted as described by Muller (
), which is based on the incorporation of a limiting dNTP into an alternating co-polymer template poly(dA-dT) or poly(dI-dC) by Klenow DNA polymerase in the presence of an excess of 3H-labeled complementary dNTP.
HU Prevents dNTP Accumulation at G1/S in α Factor-synchronized Yeast—Wild-type cells (MY377) were synchronized by α factor arrest and release. As shown by the flow cytometry data in Fig. 1A, asynchronously growing cells arrested with a G1 DNA content when incubated with α factor, and arrested cells entered S phase about 40 min after α factor was removed. Parallel dNTP pool measurements (Fig. 1B) showed that the levels of all four dNTPs were lower in G1-arrested cells than in asynchronous cells and that the levels of all four dNTPs increased 2.5–4-fold as cells entered S phase. The relative size of each dNTP pool differed reproducibly, with dTTP always being the largest and dGTP always being the smallest. For example, 108 asynchronous cells contained 293 pmol of dTTP, 183 pmol of dATP, 74 pmol of dCTP, and 65 pmol of dGTP. Assuming a cell volume of 42 fl (
). In general, the purine dNTP pools are the most strongly affected by HU; however, in experiments using different mammalian cells, the effects of HU on individual dNTP pools varied quantitatively and qualitatively. Yeast represented a system where cell synchrony could be easily achieved and where the effects of HU on individual dNTP pools as cells entered S phase could be determined. In particular, we were interested in determining whether any dNTP pool was exhausted as cells began replicating.
As shown by the flow cytometry data in Fig. 1C, when α factor-arrested MY377 cells were released to fresh medium containing HU, DNA content did not increase, confirming that HU blocked replication. As shown in Fig. 1D, HU prevented the accumulation of dNTPs that normally began about 40 min after α factor release in the absence of HU (see Fig. 1B). The result suggested that HU blocked the synthesis of all four dNTPs in yeast. HU affected the dNTP pools evenly and did not significantly change the ratios of the pools relative to each other. For example, the dTTP/dATP/dCTP/dGTP ratio was 39:19:11:10 before release into HU medium and was 45:20:13:10 after 1 h in HU. After 80 min in HU, the pools had decreased 21% for dTTP, 18% for dATP, 29% for dCTP, and 15% for dGTP. On average, dNTP pools contracted to about 79% of G1-arrested levels. Thus, although HU completely blocked DNA replication, it had only a modest effect on dNTP levels.
Preaccumulation of dNTPs Permits Replication in the Presence of HU—The persistence of significant dNTP pools in HU-treated yeast belied the simple idea that replication arrest in HU was due to exhaustion of the substrates needed for chain elongation by DNA polymerase. Careful examination of the results of earlier studies in mammalian cells showed similar persistence of dNTPs and suggested that HU can inhibit replication without exhausting a bulk dNTP pool. Several alternative explanations for HU-induced replication arrest are conceivable. First, cells may have a mechanism for sensing the size of the dNTP pools and halting replication fork progression when a dNTP pool falls to a low, but non-zero, level. Second, inhibition of RNR may completely exhaust a privileged dNTP pool at replication forks and yet only mildly affect the bulk dNTP pools measured in whole cells. Third, inhibition of flux through RNR may inhibit replication by a dNTP-independent mechanism. For example, inhibition of RNR activity may affect the redox state of the co-substrate thioredoxin, and thereby affect the activity of regulatory proteins that are sensitive to the redox state of thioredoxin. Fourth, although RNR is the only characterized target of HU, the drug may affect the activity of an as yet uncharacterized target that is required for replication fork progression. The latter three of these explanations predict that HU should continue to block replication if a means of artificially raising the bulk dNTP pools were achieved. The first of these explanations predicts that cells should be able to bypass the HU-induced replication block if a means of raising the bulk dNTP pools were achieved.
As a means of raising the bulk dNTP pools in yeast, we investigated the use of a temperature-sensitive dbf4 mutation. Dbf4, together with Cdc7, forms a holoenzyme with protein kinase activity. The Dbf4/Cdc7 protein kinase is located at replication origins and catalyzes the last known step in the cascade of events linking START to replication origin firing (
). At the nonpermissive temperature (37 °C), dbf4 mutants arrest at G1/S with unfired origins of replication.
Fig. 2 shows the results obtained when α factor-arrested dbf4 mutants were released from pheromone at the nonpermissive temperature in the absence of HU and were subsequently switched to the permissive temperature in the presence of HU. Previous continuous labeling studies, in which [14C]uracil incorporation into DNA was monitored, indicated that HU inhibits DNA synthesis in yeast within 10 min of its administration (
). To ensure that HU had adequate time to enter the cell and inactivate RNR, the drug was added 20 min before the temperature shift down. Fig. 2B shows that all four dNTP pools decreased when asynchronous dbf4 cells (black bars) were arrested in G1 with α factor and that all four dNTP pools increased 2.4–4.4-fold when α factor was removed and cells were incubated at the nonpermissive temperature (37 °C) for 80 min (shaded bars, labeled 20, 40, 60, and 60 + 20). Expansion of the dNTP pools at the nonpermissive temperature implied that the mechanism responsible for increased dNTP synthesis in late G1 was not dependent on Dbf4. Having preaccumulated dNTPs at 37 °C, would dbf4 cells switched to the permissive temperature be able to synthesize DNA in the presence of HU? The decrease in dNTP levels that was observed when cells were given HU at 60 min and switched to 25 °C at 80 min (Fig. 2B, open bars labeled 60 + 20, 60 + 40, 60 + 60, 60 + 80, and 60 + 100) was consistent with the idea that DNA synthesis was occurring and that dNTPs were being consumed through their incorporation into DNA.
Flow cytometry was used to directly monitor DNA synthesis in cells released from a dbf4 block in the presence of HU. As shown in Fig. 2A, dbf4 cells arrested with a G1 DNA content when incubated in α factor, and they did not enter S phase for at least 60 min when α factor was removed and cells were incubated at the nonpermissive temperature (37 °C). However, when cells were given HU at 60 min and switched to the permissive temperature (25 °C) at 80 min (60 + 20), they began to synthesize DNA. By 20 min after the temperature shift-down (60 + 40), DNA content had increased 27% above G1 levels. By this time, dNTP pools had fallen to nearly G1 levels (see Fig. 2B). By 40 min after the temperature shift down (60 + 60), DNA content had increased by 30%, and dNTP pools had fallen to G1 levels. With longer incubations, out to 80 min after the temperature shift down (60 + 100), DNA content had not increased further, and dNTP pools in general had fallen to a constant, non-zero level. The data in Fig. 2, A and B, indicate that when cells were allowed to preaccumulate dNTPs before HU was added, they were able to replicate about 30% of their genome in the presence of the drug. Then, once dNTPs had dropped to about G1 levels, DNA synthesis once again became arrested.
In the experiment shown in Fig. 2, it was important that HU be added several minutes prior to the temperature shift-down. If HU was added simultaneously with the temperature shift-down, many cells seemed to replicate almost their entire genome. We speculate that when HU is added at the same time as the temperature shift-down, significant amounts of DNA were synthesized before HU had entered the cell and inhibited RNR activity.
From the results in Fig. 2, we concluded that preaccumulation of dNTPs allowed cells to replicate 30% of their genome in the presence of HU. The result was inconsistent with models in which the HU-induced replication arrest was due to an effect of HU on a privileged dNTP pool, on the redox state of RNR co-substrates such as thioredoxin, or on an as yet uncharacterized target of the drug. The results support a model in which DNA replication in yeast requires a critical dNTP threshold, below which replication forks are arrested. According to this model, HU blocks replication because, in the absence of RNR activity, cells cannot maintain their dNTP pools above the threshold required to prevent replication arrest.
Deoxyribonucleotide Accumulation Is Not Required for Dbf4 Activity—Although the data in Fig. 2 showed that Dbf4 activity was not required for dNTP accumulation at G1/S, the reciprocal question of whether dNTP accumulation was required for Dbf4 activity at G1/S has not formally been addressed. To investigate this question, dNTP levels and DNA content were measured in α factor-synchronized dbf4 cells released from pheromone at the permissive temperature in the presence of HU and subsequently switched to the nonpermissive temperature in the absence of HU. To ensure that Dbf4 protein was inactivated prior to restoration of RNR activity, cells were switched to the nonpermissive temperature 10 min before HU was removed. As shown in Fig. 3B, dbf4 cells released from pheromone at the permissive temperature in the presence of HU did not accumulate dNTPs (open bars, labeled 20, 40, and 60). When subsequently switched to the nonpermissive temperature in the absence of HU, the cells gradually accumulated dNTPs over the next 60 min (shaded bars, labeled 60 + 20, 60 + 40, and 60 + 60). Parallel flow cytometric determination of DNA content (Fig. 3A) showed that cells remained with a G1 DNA content during the initial 60-min incubation at the permissive temperature in the presence of HU and that most cells acquired a G2 DNA content during a subsequent 60-min incubation at the nonpermissive temperature in the absence of HU. The increase in DNA content at the nonpermissive temperature shows that the majority of cells were able to traverse the Dbf4 execution point during the initial incubation at the permissive temperature, although HU was present and prevented dNTP accumulation. Since the execution point for the Dbf4/Cdc7 kinase lies very close to the act of replication origin firing, it is likely that origins were able to initiate replication in the absence of dNTP accumulation.
Persistence of dNTP Pools in HU-arrested rad53 Cells—The persistence of dNTP pools in HU-arrested wild-type cells (see Fig. 1) and bypass of the HU arrest in dbf4 cells that had preaccumulated dNTPs prior to HU exposure (see Fig. 2) suggested that cells have a mechanism for arresting DNA chain elongation when dNTP levels fall below a minimum threshold. One protein known to play a role in arresting cell cycle progression in HU-treated cells is the protein kinase Rad53 (
). Checkpoint-deficient rad53 cells incubated in HU enter mitosis with unreplicated DNA and die. It was recently shown that checkpoint-deficient rad53 cells are able to synthesize DNA from late firing origins in the presence of HU (
). Since replication from late firing origins would require dNTPs, we speculated that the mechanism responsible for replication arrest at low dNTP levels was defective in rad53 cells. In other words, DNA replication in HU-treated rad53 may continue until one of the dNTP pools is totally exhausted. To investigate this possibility, DNA content and dNTP levels were determined in α factor-synchronized rad53 cells incubated in the presence or absence of HU.
As shown by the flow cytometry data in Fig. 4A, rad53 cells arrested with a G1 DNA content when incubated in α factor and began to synthesize DNA within 40 min of α factor removal. Like wild-type cells, when rad53 cells were released from pheromone into medium containing HU (Fig. 4C), most cells retained a G1 DNA content throughout the course of the experiment. Fig. 4B shows dNTP levels in α factor-synchronized rad53 cells. Asynchronous cell dTTP and dATP levels in rad53 cells were similar to wild-type levels, and dCTP and dGTP levels were about 1.7-fold higher than wild-type levels (compare solid bars in Fig. 4B with those in Fig. 1B). Higher asynchronous cell dCTP and dGTP levels in rad53 mutants were somewhat surprising, since the Mec1/Rad53 pathway is known to positively influence RNR activity, at least in DNA-damaged cells (
). As was previously observed for wild-type cells, the levels of all four dNTPs fell when asynchronous rad53 cells were arrested in G1 by incubation in α factor (Fig. 4B, shaded bar, time 0). Also as previously observed for wild-type cells, dNTP levels increased when rad53 cells were released from α factor. However, the size of the increase was generally smaller in rad53 cells than in wild-type cells. For example, by 60 min after pheromone release, dTTP, dATP, dCTP, and dGTP increased 4.0-, 2.1-, 4.0-, and 2.1-fold, respectively, in wild-type cells (see Fig. 1B) but increased only 1.6-, 3.1-, 1.2-, and 1.3-fold, respectively, in rad53 cells. On average, dNTP levels increased 310% when wild-type cells entered S phase and increased only 80% when rad53 cells entered S phase. Reduced accumulation of dNTPs in rad53 cells was also evident when maximal absolute levels were compared. Maximal levels of dTTP, dATP, dCTP, and dGTP in rad53 cells were 46, 118, 49, and 66% that of wild-type cells, respectively. The reduced ability of rad53 cells to accumulate dNTPs as cells entered S phase supports the hypothesis that Rad53 is part of a pathway that positively regulates RNR activity (
Fig. 4D shows dNTP levels in α factor-synchronized rad53 cells released to medium containing HU. As was previously observed for wild-type cells, HU blocked dNTP accumulation but did not result in exhaustion of any one of the four dNTP pools. Thus, at least qualitatively, the rad53 mutation did not eliminate the mechanism by which cells arrest replication prior to dNTP exhaustion. After 80 min in HU, the pools had decreased by 19% for dTTP, 25% for dATP, 64% for dCTP, and 18% for dGTP. On average, the dNTP pools contracted to about 69% of G1 levels in HU-treated rad53 cells. In comparison, the dNTP pools contracted to about 79% of G1 levels in HU-treated wild-type cells (see Fig. 1D). The 64% decrease in dCTP observed in HU-arrested rad53 cells was noticeably greater than the 29% decrease observed in similarly treated wild-type cells. This inequality was primarily a function of the relatively high levels of dCTP in G1-arrested rad53 cells. If the absolute levels to which dCTP fell were compared, the dCTP pool in HU-arrested rad53 cells was only 18% smaller than in HU-arrested wild-type cells. The rad53 mutation had a similarly small effect on the absolute level to which the other dNTP pools fell in the presence of HU. The dATP pool in HU-arrested rad53 cells was 27% smaller than in HU-arrested wild-type cells, and the dTTP and dGTP pools in HU-arrested rad53 cells were 19 and 27% larger, respectively, than in HU-arrested wild-type cells. Thus, based on either the percentage decrease from G1 levels or the absolute size of the residual dNTP pool, the rad53 mutation had little effect on the extent to which dNTPs were consumed following HU inhibition of RNR activity. The persistence of dNTP pools in HU-treated rad53 cells indicated that the mechanism responsible for elongation arrest when dNTP levels are low was not dependent on the Mec1/Rad53 pathway.
Most studies on HU-induced replication arrest in S. cerevisiae have focused on the triggering of checkpoint mechanisms that prohibit inappropriate cell cycle transitions and thereby preserve genome stability or prevent catastrophic mitoses (reviewed by Elledge (
)). In contrast, our work has focused on the most proximal effect of HU, the inhibition of RNR, and the effect of RNR inhibition on dNTP levels and DNA synthesis. Although HU is widely believed to block replication by exhausting the supply of dNTPs required by DNA polymerase, the effect of HU on dNTP levels in yeast had not previously been determined, and published work on the effects of HU on dNTP levels in mammalian cells has been difficult to interpret, perhaps due to the compensatory activities of deoxyribonucleotide salvage pathways. In α factor-synchronized yeast, we found that HU prevented the accumulation of dNTPs that normally occurs as cells enter S phase. Flow cytometry confirmed that HU blocked DNA replication. Although HU blocked dNTP accumulation, it did not completely deplete any one of the four dNTP pools. Our results thus indicated that DNA chain elongation in HU-treated yeast is arrested by a mechanism that preserves basal dNTP pools. Similar preservation of dNTP pools may also occur in HU-treated mammalian cells. For example, when HU is added to exponentially growing human diploid fibroblasts, dATP and dGTP levels decrease only by about 70%, and dCTP and dTTP levels actually increase about 30% (
). Similarly, in mitogen-induced human lymphocytes, although HU prevents the dATP pool expansion that normally accompanies activation, it only lowers dATP levels to about 50% of that present in uninduced cells, and it has even less effect on the three other dNTP pools (
). In general, HU has stronger effects on purine deoxyribonucleotides, but even for purines, only rarely has HU been shown to cause complete exhaustion of a dNTP pool. In serum-synchronized primary mouse embryo cells (
). In the above studies, strong effects of HU on purine dNTP levels were observed only in synchronized cells. In the same studies, HU had relatively mild effects on dNTP levels in asynchronous cells. For example, in asynchronous L929 cells, HU had little effect on the size of any dNTP pool, yet the drug strongly inhibited DNA synthesis (
). Furthermore, in isoleucine-synchronized Chinese hamster ovary cells, although HU blocked the increase in dGTP levels that normally occurs as cells enter S phase, it had a relatively mild effect on basal dGTP levels, which dropped to about 70% of G1 levels. In this respect, the Chinese hamster ovary cell data were similar to our yeast results, which showed that HU prevented dNTP accumulation as cells entered S phase but did not have much of an effect on basal dNTP levels. The existence of compensatory salvage pathways in mammalian cells may have contributed to the complicated effects of HU on individual dNTP levels. Nevertheless, the persistence of dNTPs in most cases where mammalian cells were treated with HU is consistent with our conclusion that replication arrest in HU-treated cells is not due to exhaustion of the activated substrates required by replicative DNA polymerases.
Although our data indicate that HU arrests replication without completely depleting any dNTP pool, the replication arrest mechanism is not independent of dNTP levels. If dNTP levels were artificially raised by incubating yeast dbf4 mutants at the nonpermissive temperature for S phase entry, HU did not block replication when the cells were subsequently switched to the permissive temperature (Fig. 2). These results are consistent with earlier mammalian cell studies, which showed that exogenous deoxyribonucleosides partially or totally reversed the effect of HU on replication (
). The reversal data indicate that RNR activity per se is not needed for DNA chain elongation. Thus, models in which HU-induced replication arrest is due to inhibition of flux through RNR are not tenable. As long as dNTP levels are maintained above a critical level, RNR activity is dispensable for DNA replication. The results of our dNTP preaccumulation experiment in yeast (Fig. 2) as well as those of others in mammalian cells (
) are also incompatible with the idea that HU-induced replication arrest is due to complete exhaustion of a privileged dNTP pool that is generated near replication forks and does not equilibrate rapidly with a much larger bulk dNTP pool. If such a privileged dNTP pool existed, it would be quickly depleted when RNR activity was inhibited by HU, and therefore preaccumulation of dNTPs would not be expected to bypass the HU-induced replication block. Furthermore, rapid equilibration of nuclear and cytosolic dNTPs is suggested by rodent cell experiments that showed that nuclear dNTPs are depleted about as rapidly as cytosolic dNTPs when Chinese hamster ovary cells are incubated in HU (
The smallest dNTP pool in yeast is dGTP. We determined that α factor-arrested G1 cells contained 312,000 dGTP molecules/cell, or about 10% of the minimal amount theoretically required to replicate the 1.2 × 107-bp haploid yeast genome. When cells were released from α factor and allowed to enter S phase in the presence of HU, the dGTP pool fell to a minimum of about 240,000 molecules/cell. Consistent with the pool size reduction being due to DNA replication, the 72,000-molecule decrease in dGTP was roughly matched by a 66,000-molecule decrease in dCTP, and the 114,000-molecule decrease in dATP was roughly matched by a 126,000-molecule decrease in dTTP. If one assumes that the decrease in dNTP levels was due entirely to deoxyribonucleotide incorporation into DNA, HU-treated cells incorporated about 380,000 deoxyribonucleotides before DNA chain elongation became arrested. Santocanale et al. (
) showed that replication bubbles about 4500 base pairs in length accumulate at two early replication origins when α factor-arrested yeast are released to HU-containing medium. Whole-genome analyses indicated that yeast utilize about 330 efficient origins of replication per S phase (
). If all of these origins fired in the presence of HU, and the resulting elongation complexes incorporated 4500 deoxyribonucleotides per replication fork before arresting, consumption of 3 × 106 dNTPs/cell would be expected. Since we witnessed consumption of only 380,000 dNTP molecules/cell, this suggests that only 13% of yeast origins were of the early type capable of firing in the presence of HU.
Although early replication origins fire in the presence of HU, late replication origins do not (
). This prompted us to investigate whether dNTP pools became completely exhausted when rad53 cells were released from α factor into medium containing HU. Our results showed that checkpoint-deficient rad53 cells also ceased replication before any dNTP pool became completely exhausted. The degrees to which the dNTP pools were affected by HU were roughly the same in wild-type and rad53 yeast. The failure of any dNTP pool to be completely depleted in rad53 cells suggests that the elongation complexes originating from late origins also become arrested shortly after initiation in the presence of HU. Our results indicate that the mechanism sensing dNTP pool levels in HU-treated cells does not require an intact Rad53/Mec1 pathway.
Why would cells evolve a mechanism for arresting replicative DNA elongation before dNTP pools are exhausted? One possibility is DNA damage. It has been shown that excision repair is sensitive to dNTP pool depletion in cultured human fibroblasts (
). Similarly, in yeast expressing an engineered ribonucleotide reductase that is refractive to allosteric dATP feedback inhibition, the resulting increase in dNTP levels is correlated with increased cell survival following DNA damage (
), suggests that it is important for cells to boost dNTP levels in response to DNA damage. If it takes time for the transcriptional and posttranscriptional processes triggered by the Mec1/Rad53 pathway to boost dNTP levels, it would be advantageous for cells to arrest DNA synthesis at replication forks to conserve the dNTP pools for DNA synthesis at repair sites. According to this view, replication arrest would not only allow time for DNA repair to take place, it would also preserve the activated precursor pools needed for DNA repair. Consistent with this idea, unscheduled DNA synthesis, as measured by [3H]thymidine incorporation in UV-irradiated quiescent fibroblasts, is refractive to HU inhibition (
The mechanism for replication arrest when RNR is inhibited remains unknown. If RNR inhibition resulted in significant expansion of the rNTP pools, it is possible that an increased rNTP/dNTP ratio might result in misincorporation of ribonucleotides into DNA and thereby trigger a replication arrest. However, the rNTP pools in yeast, calculated from published data to be 0.34 mm rUTP, 2 mm rATP, 0.26 mm rCTP, and 0.34 mm rGTP (
) and are therefore unlikely to be affected by the presence or absence of a relatively small flux of rNDPs through RNR. Indeed, in α factor-synchronized yeast, rNTP levels are constant throughout the cell cycle (
). Similarly, we observed that the aggregate nucleobase, nucleoside, and nucleotide pool, as determined by acid-soluble A260 measurements, did not change during the cell cycle or after treatment with HU (data not shown). Given the large and unchanging size of the rNTP pool, it is unlikely that HU inhibits replication via an effect on rNTP metabolism. Given the observation that dNTP preaccumulation delays the replication arrest (Fig. 2), we feel that the simplest explanation fitting the data is that cells have a mechanism for shutting down replication before dNTP pools become exhausted. The arrest mechanism does not require an intact Rad53/Mec1 pathway, since checkpoint-deficient rad53 cells also do not exhaust their dNTP reserves when incubated in HU. It is not clear whether all or a subset of the four dNTP pools is sensed. If the smallest pool, dGTP, is the one sensed, DNA elongation is arrested when dGTP levels fall below 240,000 molecules/cell. Although we have no specific model for the dNTP-sensing mechanism, we speculate that in order for elongation complexes to progress more than a few thousand base pairs from their point of origin, dNTP levels must be maintained above a critical threshold.