Cell cycle checkpoints function by sensing DNA damage and transmitting a signal that results in arrest at defined stages of the cycle. The S phase “checkpoint” halts cells during the DNA synthetic period in response to DNA damage ( DNA damage responses) and is somewhat different from other checkpoints in two ways. First, in the classic paradigm, cell cycle progression is halted at various discrete points in the cycle (e.g., in G₁, at the G₂/M transition, and during mitosis), and cells are usually arrested for many hours prior to reentry into the cycle. In contrast, the S phase damage-sensing pathway arrests cell cycle progression for only an hour or so and seemingly at any position within the S phase; therefore, it is difficult to think of it as a checkpoint per se (although we will refer to it as such in this review). As more is learned about the molecular basis of all of the checkpoints, it is becoming clear that each may have multiple targets. Thus, the underlying strategies for mobilizing these vital damage-sensing pathways may turn out to have many similarities.

A second possible difference is that the classic G₁, G₂/M, and mitotic checkpoints are thought to serve an anticipatory or surveillance function by arresting cells prior to critical processes such as DNA replication or mitosis, both of which are capable of converting potentially repairable damage into catastrophic lesions, such as double-strand breaks or chromosome nondisjunction. However, there is little evidence to suggest that the classic S phase damage-sensing pathway has an anticipatory function, and it is formally possible that the arrest of DNA synthesis in response to damage results from a competition for trans-acting factors between multiple DNA-templated responses (repair, transcription, and replication).

At what point(s) in S phase does the S phase checkpoint operate? It has been known for decades that when exponentially growing cells are subjected to ionizing radiation, DNA synthesis is rapidly and significantly inhibited. Compelling evidence has been obtained that damage-induced inhibition of DNA synthesis represents a bona fide damage-sensing signal transduction pathway, as opposed to a simple arrest of replication forks by damage in the template (i.e., a cis-acting mechanism). For example, a dose of 5 Gy of gamma radiation results in a single-strand break every 25 replicons on average, yet inhibits overall DNA synthesis by ∼50%. Therefore, it has been presumed for years that the S phase checkpoint is a global cellular response that functions in trans and inhibits initiation of DNA replication as opposed to chain elongation. Indeed, there is direct experimental evidence from studies on mammalian viruses that doses of ionizing radiation too small to damage individual viral templates nevertheless downregulate viral replication, suggesting a trans-regulatory mechanism. Results from three different experimental systems also provide direct evidence that the S phase checkpoint downregulates DNA synthesis by inhibiting initiation  radiation sensitivity. Alkaline sucrose gradient analysis of pulse-labeled DNA showed that ionizing radiation inhibits the appearance and maturation of the smallest nascent fragments to a much greater extent than the maturation of larger fragments. Secondly, DNA fiber autoradiographic analysis of the changes in the pattern of replication after irradiation suggested that initiation is preferentially inhibited. Thirdly, a two-dimensional gel replicon mapping approach that can distinguish between initiation and elongation of DNA synthesis showed clearly that ionizing radiation preferentially downregulates initiation in a defined chromosomal replicon (the amplified, early-firing dihydrofolate reductase domain). To determine whether damage-induced inhibition of initiation also occurs at origins that fire later in the S period, advantage was taken of the naturally amplified ribosomal gene cluster, in which a restriction site polymorphism distinguishes between early- and late-firing rDNA replicons. The two-dimensional gel replicon mapping approach showed that both the early- and late-firing origins were inhibited by ionizing radiation. By extension, it is likely that all origins in mammalian chromosomes will be inhibited in response to DNA damage, regardless of when they are activated.

An interesting observation suggests that there may be more than one checkpoint or pathway responsible for the overall inhibition of the rate of DNA synthesis that occurs when ionizing radiation is delivered to an asynchronous culture of growing cells. When cells are synchronized and replicate cultures are irradiated at hourly intervals during S phase, the subsequent inhibition of DNA synthesis never exceeds more than about 25% at any time in the S period, whereas DNA damage delivered to an asynchronous population results in 50% or greater inhibition of the overall rate of DNA replication. By careful fluorescence-activated flow sorter analysis of gamma-irradiated log cultures, it was possible to demonstrate the existence of a population of late G₁ cells whose entry into S phase was prevented for several hours. Since this potentially new checkpoint was uncovered in CHO cells, which are deficient in p53 activity  p53 Family, it is distinct from the well-known p53-mediated G₁ checkpoint. This pathway is therefore analogous to a G₁/S checkpoint that has been described in Saccharomyces cerevisiae, which acts between the cell cycle steps affected by the DBF4 and CDC7 gene products.

It is possible that downregulation of entry into the S phase in mammalian cells (i.e., at the G₁/S transition) functions by inhibiting the earliest S phase origins and is therefore mechanistically related to damage-induced inhibition of late-firing origins (i.e., it does not differ from the S phase checkpoint per se). Since irradiation delays the entire S period for several hours (i.e., it effectively repositions the S phase along the cell cycle time axis), it would have to be argued that late-firing origins cannot fire until early-firing origins have done so. Interestingly, however, cells arrested near the G₁/S boundary in either mimosine or aphidicolin (both effective inhibitors of chain elongation) are resistant to radiation-induced inhibition of DNA synthesis, lending weight to the argument that there is a unique damage-sensing pathway that operates at the G/₁S transition prior to origin firing. Since the events preceding initiation of DNA synthesis in mammalian cells are only now being characterized, it will be some time before the molecular nature of this potential new checkpoint is uncovered.

What proteins are involved in the S phase damage-sensing checkpoint? The ataxia telangiectasia mutated (ATM) ( ATM protein) and the ATM Rad3-related (ATR) protein kinases are critical checkpoint factors in human cells. Both are members of the phosphatidylinositol 3-kinase-related family of protein kinases (PI3K) that play a major role in sensing damage and triggering repair of DNA lesions in mammalian cells. ATM and ATR are preferentially activated by different forms of DNA damage. ATR is activated following induction of DNA adducts by UV or by chemical cross-linking agents, whereas ATM is activated in response to double-strand DNA breaks induced by irradiation, etc.

The ATM gene was cloned in 1995. Cells derived from patients with ataxia telangiectasia (AT) fail to inhibit DNA synthesis in response to irradiation. This phenomenon was termed radiation-resistant DNA synthesis (RDS), which reflects a failure of intra-S phase checkpoint control. ATM is thought to play a central role in the cellular responses (including the S phase checkpoint) to DNA double-strand breaks (DSBs) and other genotoxic stresses. ATM exists as a catalytically inactive dimer in the absence of DNA damage. In response to DNA damage, especially DSBs, ATM undergoes rapid autophosphorylation on serine 1981, resulting in dissociation of the inactive dimers to yield active monomers. However, when the 1987 site was mutated (the mouse equivalent of serine 1981) and expressed in mice as the sole ATM species, the mutated ATM kinase was activated by radiation, suggesting that an alternative mode for stimulation of the ATM kinase must exist. The fact that several different serine/threonine phosphatases type 2 (PPP2) alpha isoform and type 5 (PPP5) regulate ATM activity raises the question whether additional phosphorylation sites exist in the ATM protein that regulate its activity. In addition to DNA damage, chromatin remodeling molecules are also able to induce rapid activation of ATM/ATR in the absence of detectable DNA strand breakage. The activation of ATM/ATR by DNA strand breakage might thus be mediated, at least in part, by a change in chromatin structure.

In 2000, the Seckel syndrome (an autosomal recessive disorder characterized by developmental abnormalities including growth retardation and microcephaly) gene was localized to human chromosome 3q22.1-q24 (SCKL1) by homozygosity mapping. This gene has subsequently been identified as ATR. ATR is thought to be activated by both stalled replication forks and bulky adducts produced by agents such as mitomycin C. ATR function in cell cycle checkpoint signaling pathways is dependent on an ATR-interacting protein known as ATRIP. ATRIP contains domains similar to the ATM-binding domain of Nbs1, which is required for the interaction of ATRIP with ATR. ATR and ATRIP both localize to intranuclear foci after DNA damage or inhibition of replication. ATRIP is phosphorylated by and interacts with the single-stranded DNA (ssDNA)-binding protein RPA. RPA coats ssDNA at replication forks to form an RPA–ssDNA complex in response to damage. Deletion of ATR in mice mediated by the Cre recombinase causes the loss of both ATR and ATRIP expression, the loss of DNA damage checkpoint responses, and cell death. Depletion of RPA inhibits ATR association with chromatin and abrogates the aphidicolin-induced DNA replication checkpoint. Moreover, RPA is also required for activation of ATR-mediated phosphorylation of Chk1 and Rad17.

 Nijmegen breakage syndrome (NBS) is a recessive genetic disorder characterized by elevated sensitivity to ionizing radiation, chromosome instability, and a high frequency of malignancies. Since cellular features partially overlap with those of ataxia telangiectasia (A-T), NBS was long considered an A-T clinical variant. Nbs1, the product of the gene underlying the disease, contains three functional regions: a forkhead-associated (FHA) domain, a BRCA1 C-terminal (BRCT) domain at the N-terminus, an Mre11-binding region at the C-terminus, and several SQ motifs (consensus phosphorylation sites by ATM and ATR kinases) in the central region. Nbs1 forms a multimeric complex with hMre11/hRad50, which is recruited to the vicinity of DSB sites by direct binding to histone H2AX via the FHA and BRCT domains of Nbs1. Studies suggest that the Mre11–Rad50–Nbs1 (MRN) complex, and possibly other proteins, play a role in the recruitment of ATM to the region of DNA strand breaks. In addition, BRCA1 ( BRCA1/BRCA2 germline mutations and breast cancer risk) appears to be necessary for recruitment of ATM to damage foci. In response to IR, ATM fails to localize to damaged sites in the cells lacking full-length BRAC1 or full-length Nbs1, despite the fact that ATM is activated. Once ATM is recruited to damage sites, it can phosphorylate substrates at the break, including Nbs1, 53BP1, BRAC1, H2AX, and Smc1. Smc1 is a member of the structural maintenance of chromosomes protein (SMC) (family of proteins) and is thought to enhance the efficiency of DSB repair. Nbs1 therefore acts both as an upstream modulator of ATM as well as a downstream target of ATM.

In response to different genotoxic stresses, ATM and ATR transduce the damage signal to checkpoint control proteins to activate checkpoints. There are multiple parallel pathways underlying the intra-S phase checkpoint. The first pathway involves the ATM/ATR-Chk1/Chk2-Cdc25A-Cyclin E(A)/Cdk2-Cdc45 cascade, which links the upstream checkpoint kinases with the core cell cycle machinery in S phase cells. DNA-damage-activated ATM and/or ATR phosphorylate Chk1 (serines 317 and 345) and Chk2 (threonine 68). Chk1 and Chk2 inhibit Cdc25A activity by phosphorylation of its serine 123 residue followed by ubiquitin-mediated degradation. Cdc25A is a dual-specificity phosphatase, removing inhibitory phosphates from threonine 14 and tyrosine 15 from Cdk2. Thus, Cdk2 is inhibited, resulting in decreased phosphorylation of Cdc45. This prevents its association with chromatin and thereby decreases initiation of DNA replication at origins.

Another S phase pathway involves the ATM–Nbs1–Smc1 cascade. Following IR, the MRN/Smc1 complex is recruited to sites of DSBs. Once ATM is recruited to the complex, it phosphorylates Nbs1 on serine 278 and serine 343. Mutation of these residues results in an S phase checkpoint defect following ionizing radiation. In addition, phosphorylation of the protein Smc1 on serines 957/966 by ATM is also necessary for activation of the S phase checkpoint. The ATM–Nbs1–Smc1 axis is clearly distinct from the ATM–Chk2–Cdc25A pathway.

 Fanconi anemia (FA) is a recessive genetic disease characterized by cellular hypersensitivity to DNA interstrand cross-linking agents, mild sensitivity to other genotoxic agents, and clinical features that overlap with NBS, A-T, and ATR–Seckel syndrome. Phosphorylation of serine 222 of Fanconi anemia protein D2 (FANCD2) by ATM is dependent on Nbs1 and is necessary for the IR-activated S phase checkpoint. Since the ATM–FANCD2 pathway apparently acts independently of SMC1 phosphorylation, its downstream effect is presently unknown.

What are the targets of the S phase checkpoint? Precise regulation of the initiation of DNA synthesis is critical, since it ensures that the genome is replicated once and only once per cell cycle. Much of what we know about the regulation of initiation at origins comes from yeast. Autonomously replicating sequence (ARS) elements were first identified in a functional assay by virtue of their ability to support plasmid replication. In Saccharomyces cerevisiae, it is now known that a stepwise assembly of proteins onto origins precedes origin firing. In yeast, a multi-protein complex (the origin replication complex (ORC)) has been isolated and shown to interact with ARS elements. ORC is comprised of six proteins and remains bound to yeast origins throughout the cell cycle. During M phase, the Cdc6 protein is recruited to ORC replication licensing system, which, in turn, recruits minichromosome maintenance (MCM) proteins to form the pre-replicative complex (pre-RC) on origins. An S phase cyclin is then thought to be necessary for the association of the Cdc45 protein with the pre-RC just prior to the onset of DNA synthesis. Several other proteins such as Cdt1, GINS, and MCM10 play critical but poorly understood roles in effecting initiation at origins. Origin firing is likely downregulated through MCM2 phosphorylation, which itself is regulated by checkpoint pathways signaling through Cdc45.

What is the biological significance of the S phase checkpoint? Cell cycle checkpoints by definition provide an adaptive cellular advantage following genotoxic stress. In the case of the S phase checkpoint, radiation sensitivity has been uncoupled from checkpoint function, arguing that the S phase checkpoint may not always function to enhance survival following DNA damage. However, these experiments were performed on AT or AT-like cells in which it is not possible to exclude other mutations that are epistatic to the checkpoint defect. Furthermore, ATM likely directly influences DNA repair possibly via SMC1. Another possibility is that the S phase checkpoint functions to maintain genomic stability. Irradiation produces single-strand DNA breaks, and replication through single-strand breaks has the potential to produce double-strand lesions that, if not correctly repaired, can lead to chromosomal rearrangements and carcinogenesis. An interesting further possibility is that the S phase checkpoint functions in normal cell division to cope with endogenous oxidative or other genotoxic stresses. Future work will clarify these questions.