All cells have mechanisms to maintain the integrity of their genomes. Cell cycle checkpoint pathways have evolved to regulate cell cycle progression, DNA replication, and repair of damaged DNA [1,2]. The ataxia telengiectasia and Rad3-related (ATR) protein kinase is the apical kinase of one such checkpoint pathway. The best known role of ATR is its activation of checkpoint kinase 1 (Chk1) after replication fork stalling, leading to cell cycle arrest. A growing body of data, however, now shows roles for ATR far beyond the activation of Chk1. For example, ATR also phosphorylates numerous substrates in other DNA repair pathways.  In addition, ATR plays a role in normal replication of undamaged DNA.

Many current cancer treatments, including chemotherapeutic agents and ionizing radiation, induce DNA damage and replication fork stalling, thereby activating cell cycle checkpoint pathways. A variety of studies have shown that this response is an important mechanism that helps cancer cells survive these treatments. These findings have prompted the development of agents targeting DNA-damage response signaling pathways. Although several inhibitors of Chk1 have been developed and have proceeded to clinical trials, no inhibitors of ATR have yet been developed.

In this review, we summarize current understanding of ATR signaling pathways. We also examine the effects of ATR downregulation in combination with chemotherapeutic agents and discuss the possibility of developing ATR inhibitors for clinical use.

ATR is a member of the phosphoinositide 3-kinase related kinase (PIKK) family [3,4]. PIKKs are large serine-threonine protein kinases that exhibit a high degree of sequence homology particularly in their kinase, FAT (named after the kinases FRAP, ATM, and TRRAP) and FATC domains. The PIKKs show a similar overall architecture (Figure 1). The kinase region is bordered by the FAT domain on the N-terminal side and the FATC domain on the C-terminal side. The FAT domain is part of a long N-terminal region predicted to fold into alpha-helical HEAT repeats, each 37-47 amino acids long. Although the functions of the FAT and FATC domains are not firmly established, it has been speculated they may be important for proper functioning of the kinase domain or for interactions with other proteins [5]. The PIK regulatory domain (PRD), located between the kinase and FATC domains, is a regulatory domain in at least ATR, ATM, and the mammalian target of rapamycin (mTOR), and is poorly conserved. The N-terminus, also poorly conserved, allows for interaction with accessory proteins. Electron microscopy structures, which have been reported for DNA-PKcs and ATM only, show that PIKKs assume a large globular conformation in which distant regions of the primary sequence can be brought together through folding [6,7].

A more detailed understanding of the structure of ATR and how it interacts with other proteins has emerged in recent years. ATR is a large kinase consisting of 2644 aa, with a molecular weight of 300 kD. ATR-interacting protein (ATRIP), the 85 kDa binding partner of ATR, binds to the N-terminus of ATR [8]. Because there is no difference in phenotypes that result from the loss of ATR or ATRIP (reviewed in ref. [9]), ATRIP should probably be considered a subunit of the ATR holoenzyme.

Some ATR-interacting proteins actually bind to ATRIP. For example, ATRIP binds to replication protein A (RPA) on single-stranded DNA [10]. This interaction occurs through binding of an acidic alpha helix in ATRIP with the basic cleft of an N-terminal OB-fold domain in the large RPA subunit [11]. In addition, the primary binding site for the topoisomerase binding protein I (TopBP1) activation domain is also within ATRIP [12]. Activation of ATR also requires the PRD domain, as mutations in this region prevent activation of ATR by TopBP1, causing severe checkpoint defects [12]. Five phosphorylation sites have been mapped on ATR/ATRIP, yet none is a clear early marker for ATR activation [13,14,15].

A comparison of ATR and ATM shows structural as well as broad functional similarities. ATR shares considerable sequence homology with ATM. The FAT domain of ATR shares 23 of 43 amino acids with ATM; the kinase domain shares 35 of 55 amino acids; and a functionally undefined N-terminal region has 21 of 39 amino acids in common with ATM [16]. In addition to the sequence homology and similar structural architecture, ATR and ATM both prefer to phosphorylate serine or threonine residues followed by a glutamine [17,18]. These S/T-Q sites are clustered in some substrates but can be found singly in others. Although ATR and ATM each have specific substrates, there are some substrates common to both, including RPA [19] and BRCA1 [20].

ATR is essential for cell survival [21]. ATR^-/- cells develop widespread chromosome breaks and then undergo apoptosis. As a consequence, knockout of Atr in mice is embryonic lethal [21]. In contrast, even though ATM^-/- cells are hypersensitive to ionizing radiation and other agents that cause double-strand breaks (see below), ATM is not essential for cell survival.

ATR and ATM are the apical kinases for the two checkpoint pathways triggered by DNA damage. Chk1 and Chk2 are critical kinases downstream of ATR and ATM, respectively, playing roles in cell cycle arrest, DNA repair, or apoptosis (Figure 2).

ATR and ATM are generally thought to respond to different types of DNA damage [22]. ATR is activated in response to ultraviolet light, certain chemotherapeutic drugs, hydroxyurea, and replication stress. When these agents cause polymerases to stall during replication at damage sites on DNA, helicases will continue to unwind the DNA, leading to long stretches of ssDNA [10]. Replication protein A (RPA) coats ssDNA. The ATRIP portion of the ATR-ATRIP complex then directly binds RPA through conserved binding regions. This binding of ATR-ATRIP to RPA-coated ssDNA does not in itself, however, activate ATR. Activation also requires the Rad9-Hus1-Rad1 (9-1-1) complex and topoisomerase binding protein I (TopBP1) (Figure 3). The 9-1-1 complex, a heterotrimeric ring similar in structure to the replicative sliding clamp PCNA, is loaded by Rad17 and the four small replication factor C (RFC) subunits, in an ATP dependent manner, onto RPA-coated ssDNA [23,24]. More specifically, 9-1-1 is preferentially loaded onto DNA with 5’-recessed ends, including stalled replication forks at sites of damage, recombination sites, and nucleotide-excision repair sites [25]. Next, the BRCT I and II domains of TopBP1 interact with phosphorylated serine³¹⁷ on chromatin-bound Rad9 [26]. This brings the activation domain of TopBP1 in proximity to RPA-bound ATR, allowing the activation of ATR [26]. Although this interaction with TopBP1 increases the kinase activity of ATR toward its substrates [27], it is not known how this occurs. One possibility is the binding of ATR to the activation domain of TopBP1 changes the conformation of the kinase domain of ATR [28].

Among the many substrates phosphorylated by activated ATR, the most widely studied is Chk1 [29]. Another protein, claspin, serves as an adaptor to bring ATR and Chk1 together [30,31,32,33]. The Tim/Tipin complex may also serve to bring ATR and Chk1 together [34,35]. After ATR phosphorylates Chk1 on Ser³¹⁷ and Ser³⁴⁵, activated Chk1 then phosphorylates many serine residues on the phosphatase Cdc25A (Figure 4), leading to its ubiquitylation and degradation [36]. As a result, Cdc25A is not available to remove inhibitory phosphorylations on Cdk1/cyclin B, causing cells to arrest in G2 phase of the cell cycle. Activated Chk1 can also phosphorylate Cdc25C, causing it to bind 14-3-3 proteins and be exported from the nucleus [37,38]. This also leads to G2 arrest, as the cytoplasmic Cdc25C cannot remove the inhibitory phosphorylations on nuclear Cdk1/cyclin B.

The phosphorylation of Chk1 by ATR can also lead to an intra-S phase checkpoint [39]. When DNA is damaged, ongoing processes that involve the DNA, such as replication, can lead to further damage. ATR signaling is critical for inhibiting the firing of new origins of replication [40,41,42], thereby diminishing this replication-associated damage. The regulation of replication origins by ATR signaling involves, at least in part, Chk1-mediated phosphorylation of Cdc25A, which then cannot de-phosphorylate and activate the Cdk2/cyclin A complex required for S-phase progression [36]. In both yeast and mammalian cells, ATR acting through checkpoint kinases slows elongation of replication forks [43]. The ATR mediated S-phase checkpoint also directly inhibits the rereplication of DNA, i.e., replication more than once in a cell cycle, as measured by the percentage of cells with > 4N DNA [44].

In contrast to the ATR pathway, the ATM pathway (Figure 2) is activated after DNA double-strand breaks (DSBs), which may result from ionizing radiation or certain chemotherapeutic drugs. ATM, activated by its interaction with the Mre11-Rad50-NBS1 complex (MRN), phosphorylates a variety of substrates [45], most notably Chk2. Once activated by this phosphorylation, Chk2 not only phosphorylates Cdc25A and Cdc25C as described above, but also the transcription factor p53, which activates the gene encoding p21, an inhibitor of the Cdk2/cyclin A and Cdk2/cyclin E complexes. This sequence of events constitutes the G1 checkpoint, which prevents cells with damaged DNA from entering S phase. The activation of p53 may also lead to apoptosis of cells if damage is beyond repair [46,47].

Results from a large-scale proteomic analysis of proteins phosphorylated in response to DNA damage on consensus sites recognized by ATR and ATM identified over 700 substrate proteins with more than 900 phosphorylation sites [62]. Although many of these proteins are substrates of ATM, there are also a large number of potential ATR substrates in this set. This set of proteins is highly interconnected and describes a much more complicated network for the DNA damage response than was previously understood.

The most thoroughly studied ATR substrate is the kinase Chk1. As discussed above, Chk1 is activated when ATR phosphorylates it on Ser³¹⁷ and Ser³⁴⁵. Once phosphorylated, Chk1 is released from chromatin to phosphorylate its own substrates, leading to diminished replication and slowed cell cycle progression, thereby allowing time for repair to occur. Additionally, ATR and Chk1 help regulate DNA replication in undamaged cells. Specifically, ATR and Chk1 regulate the timing of the firing of origins during normal, unperturbed replication [63]. In the absence of damage, loss of Chk1 activity results in the frequent stalling and, possibly, collapse of active forks as well as activation of adjacent, previously suppressed origins [64].

Additional ATR substrates involved in replication and stabilization of stalled forks include the replication factor C complex, RPA1 and RPA2, and the MCM2-7 complex [65,66,67,68]. Phosphorylation of these substrates can lead to inhibition of replication, which goes along with the checkpoint function of ATR, or can lead to the restarting of replication, i.e., recovery from replication inhibition. For example, MCM2 phosphorylation by ATR recruits Polo-like kinase 1 (Plk1), which then promotes recovery of DNA replication [69].

Additional ATR substrates that play roles in various checkpoint and repair pathways have been identified. BRCA1 (breast cancer suppressor protein 1) plays a role in both non-homologous end joining (NHEJ) and homologous recombination (HR). Although the exact role of BRCA1 in these processes is not clear, it is phosphorylated by ATR following damage by IR or hydroxyurea [70]. Notably, BRCA1 is also a substrate for ATM, which phosphorylates it following IR.

WRN and BLM are two additional ATR substrates that are thought to play critical roles during replication. WRN, the protein missing in Werner's syndrome, an autosomal recessive disorder characterized by genomic instability, is phosphorylated through an ATR/ATM dependent pathway in response to replication blockage [71]. Bloom helicase (BLM), the protein mutated in Bloom syndrome, which is characterized by predisposition to almost all forms of cancer, is phosphorylated by ATR at Thr⁹⁹ after replication stress. This phosphorylation is critical for BLM to interact with and colocalize with 53BP1 [72]. BLM and WRN are thought to act at stalled replication forks to prevent recombination and nuclease action at the site [73,74,75].

ATR also participates in regulation of the Fanconi anemia repair pathway, a pathway specifically implicated in repair of DNA cross-links. ATR phosphorylates FANCD2 on Thr⁶⁹¹ and Ser⁷¹⁷, thereby promoting FANCD2 monoubiquitination and enhancing cellular resistance to DNA cross-linking agents such as cisplatin [76].

Because the ATR-Chk1 checkpoint pathway serves to insure cell survival after replication stress, a normal and robust checkpoint is thought to be a mechanism of resistance to chemotherapy. As a result, ATR-Chk1 pathway components are considered promising therapeutic targets. In particular, inhibition of ATR-Chk1 pathway components could potentially enhance the effectiveness of replication inhibitors. A potential advantage of sensitizing cells in this way would be the use of lower doses of the replication inhibitor, thus reducing toxicity to hematologic, gastrointestinal, and other organ systems, if the normal cells are not sensitized to the same extent.

A variety of observations suggest that checkpoint inhibition might indeed selectively sensitize cancer cells. It has been known for two decades that most tumor cells are deficient in the G1 checkpoint. For example, many cancers have mutations in p53 or other components of the p53 pathway, leading to reliance on the S and G2 checkpoints to arrest the cell cycle and provide for repair and survival [77,78,79]. Inhibition of the S and G2 checkpoints may then preferentially kill these p53 deficient tumor cells. This is illustrated by the effects of caffeine, a PIKK inhibitor, which preferentially sensitizes p53 mutant cells to ionizing radiation [80,81]. Alternatively, one or more of the other components of checkpoint responses may be mutated or impaired in cancer cells. Here, too, inhibition of the remaining checkpoints is likely to have greater deleterious effect on cancer cells than on normal cells because of intact function of all of the checkpoints in normal cells.

Specificity for cancer cells may also be insured by the fact that untransformed cells have more robust S and G2 checkpoints than tumor cells, as illustrated in recent work on the role of checkpoints at replication forks stalled by hydroxyurea treatment [82]. In normal cells, both the Chk1 pathway and several Chk1-independent pathways are activated after replication fork stalling. The Chk1-independent mechanisms in untransformed cells include activation of p38, which collaborates with Chk1 to prevent mitotic entry. Down-regulation of cyclin B1 promoter activity also occurs, independent of Chk1 or p38, after replication fork stalling in untransformed cells. Tumor cells, in contrast, rely entirely on the Chk1 pathway for the needed response, perhaps because they lack one or more additional parallel pathways that, in untransformed cells, ensure a replication checkpoint response in the absence of Chk1. Thus, inhibition of Chk1 along with hydroxyurea enhances the killing of tumor cells but not untransformed cells [82].

Collectively, these studies provide strong support for the use of inhibitors of the ATR-Chk1 pathway in combination with DNA damaging chemotherapies.

Inhibition of checkpoint pathways is currently being investigated as a way to broaden the therapeutic window of anticancer agents such as antimetabolites, alkylating agents, platinating agents, and topoisomerase inhibitors. The use of ATR siRNA and other molecular techniques described in the preceding section, as well as the current clinical use of Chk1 inhibitors, give a glimpse of the potential efficacy of ATR inhibitors.

Despite the promising results in cell lines, the clinical use of ATR inhibitors presents several potential problems. The most obvious concern is that ATR is an essential gene, and knockout of Atr in mice is embryonic lethal [21,101]. This raises the concern that inhibition of ATR may be lethal to normal cells. However, knockout of Chk1 is also embryonic lethal in mice [29,101], yet small molecule Chk1 inhibitors are now being used in clinical trials in combination with chemotherapeutic drugs.

A second concern is that ATR regulates the firing of origins of replication in S phase of every cell. As a consequence, replication may not proceed properly without ATR. Also, errors in replication occur regularly in normal cells. It is not known whether cells can fully repair replication errors if ATR is not functional or is only active at low levels.

A third concern is that ATR inhibition may also increase breaks at fragile sites in the DNA of normal cells. These fragile sites may be especially prone to breakage after treatment with a drug that causes replication stress together with an inhibitor of ATR

A fourth issue involves the relationship between ATR and ATM signaling. As mentioned above, current evidence suggests significant crosstalk between these two kinases. Inhibition of ATR may affect the ability of the ATM pathway to function optimally. This could be harmful to normal cells if they are also exposed to DNA damaging agents. Additionally, lack of optimal ATM signaling through p53 could allow cancer cells to avoid apoptosis after treatment with DNA damaging drugs. However, the fact that many cancers are defective in p53 or another protein in the ATM-p53 pathway argues that ATR inhibition may not be a problem with regard to its effect on ATM and p53.

Although specific ATR inhibitors have not yet been described, several inhibitors of checkpoint kinases have been developed to date. These include UCN-01, AZD7762, XL844, PF-00477736, CBP501, and SCH 900776 (Table 1). All of these drugs inhibit both Chk1 and Chk2, except SCH 900776, which is reportedly specific for Chk1. Of the remaining five inhibitors, PF-00477736 is the most selective for Chk1 (approximately 100-fold). XL844 is also 10-fold more selective for Chk1, while UCN-01, AZD7762, and CBP501 inhibit Chk1 and Chk2 to a similar degree.

Results of clinical trials of UCN-01 in combination with cisplatin [102], carboplatin [103], topotecan [104], irinotecan [105], and fluorouracil [106] have been published. Partial responses were seen in 10% of patients in the topotecan + UCN-01 trial and in one of ten patients in the cisplatin + UCN-01 trial. No objective responses were seen in the trials of UCN-01 with carboplatin, irinotecan, or fluorouracil. Clinical trials of all other Chk1 inhibitors are currently ongoing, and publication of results from these trials is awaited with interest.

If clinical trials of these checkpoint inhibitors have limited success, targeting ATR might be a logical next step. Because ATR has functions beyond the phosphorylation of Chk1, inhibition of ATR may be more effective in enhancing the killing of cancer cells than the Chk1 inhibitors now in clinical trials. In the near future, we should learn whether inhibiting Chk1 increases the clinical antitumor efficacy of certain agents. In addition, questions regarding toxicity of combination therapies with checkpoint inhibitors and the risk of secondary cancers should be answered to some degree. Thus, these results could help guide future trials with ATR inhibitors.

Our understanding of the ATR-Chk1 signaling pathway has greatly increased over the past several years. A much broader role for ATR has also emerged, including functions in different DNA repair pathways. However, with hundreds of potential substrates of ATR identified by proteomics analysis, much work remains to thoroughly understand the consequences of ATR signaling.

The near future will provide results of the first clinical trials of checkpoint inhibitors in combination with a variety of chemotherapeutic agents. These trials use inhibitors of Chk1 (or dual inhibitors of Chk1 and Chk2). However, the effectiveness of ATR inhibition in combination with several different chemotherapeutic agents has also been demonstrated in the laboratory by several groups. The sensitization of cancer cells to chemotherapy by ATR inhibition is striking, and in some instances, much greater than the sensitization observed with Chk1 inhibition. In light of this finding, the pharmaceutical development of ATR inhibitors is a potentially useful and exciting strategy for cancer therapy.