|Home | About | Journals | Submit | Contact Us | Français|
The ataxia telangiectasia-mutated and Rad3-related (ATR) kinase is a master checkpoint regulator safeguarding the genome. Upon DNA damage, the ATR-ATRIP complex is recruited to sites of DNA damage by RPA-coated single-stranded DNA and activated by an elusive process. Here, we show that ATR is transformed into a hyperphosphorylated state after DNA damage, and that a single autophosphorylation event at Thr 1989 is crucial for ATR activation. Phosphorylation of Thr 1989 relies on RPA, ATRIP, and ATR kinase activity, but unexpectedly not on the ATR stimulator TopBP1. Recruitment of ATR-ATRIP to RPA-ssDNA leads to congregation of ATR-ATRIP complexes and promotes Thr 1989 phosphorylation in trans. Phosphorylated Thr 1989 is directly recognized by TopBP1 via the BRCT domains 7 and 8, enabling TopBP1 to engage ATR-ATRIP, to stimulate the ATR kinase, and to facilitate ATR substrate recognition. Thus, ATR autophosphorylation on RPA-ssDNA is a molecular switch to launch robust checkpoint response.
ATR, ATM (ataxia telangiectasia-mutated), and DNA-PKcs (DNA dependent protein kinase) are three members of the phosphoinositide-3-kinase-like protein kinase (PIKK) family and key regulators of DNA damage signaling and DNA repair. Although all these PIKKs are activated by DNA damage, their DNA damage specificities are distinct, and their functions are not identical. ATM and DNA-PKcs are activated by double-stranded DNA breaks (DSBs), whereas ATR responds to a broad spectrum of DNA damage that induces single-stranded DNA (ssDNA)(Ciccia and Elledge, 2010; Cimprich and Cortez, 2008; Flynn and Zou, 2010). This unusual versatility of ATR enables it to play a particularly important role in the cellular responses to intrinsic genomic stresses during cell proliferation. Unlike ATM and DNA-PKcs, ATR is essential for cell survival even in the absence of extrinsic genomic insults (Brown and Baltimore, 2000; Cortez et al., 2001). Elucidation of the mechanism by which ATR is activated is central to understanding how genomic integrity is maintained in humans.
In response to DNA damage, ATR, ATM, and DNA-PKcs are regulated by distinct DNA damage sensors. ATM is recruited and activated by the Mre11-Rad50-Nbs1 (MRN) complex (Berkovich et al., 2007; Lee and Paull, 2005; Uziel et al., 2003), whereas DNA-PKcs is recruited and activated by the Ku70-Ku80 heterodimer (Smith and Jackson, 1999). ATR, through ATRIP, recognizes RPA-ssDNA at sites of DNA damage or stressed replication forks (Ball et al., 2005; Costanzo et al., 2003; Namiki and Zou, 2006; Zou and Elledge, 2003). In contrast to ATM and DNA-PKcs, ATR-ATRIP is not fully activated by the sensor-DNA complex, RPA-ssDNA (MacDougall et al., 2007). The full activation of ATR-ATRIP requires additional regulators including Rad17, the Rad9-Rad1-Hus1 (9-1-1) ‘checkpoint clamp’, and TopBP1 (Kumagai et al., 2006; Liu et al., 2006; Zou et al., 2002). Thus, ATR is a unique PIKK that is recruited and activated by different factors through a multi-step process.
A major gap in our understanding of ATR activation is how the ATR-ATRIP kinase is activated on RPA-ssDNA. Even in the absence of DNA and other proteins, TopBP1 directly stimulates the ATR-ATRIP kinase in vitro (Kumagai et al., 2006). The ATR-activating domain (AAD) of TopBP1 interacts with ATR-ATRIP in vitro, but this interaction is weak and not regulated by DNA damage (Mordes et al., 2008). How the stimulation of ATR-ATRIP by TopBP1 is regulated by DNA damage in vivo is still poorly understood. TopBP1 interacts with Rad9, a 9-1-1 component, through two constitutively phosphorylated residues at its C terminus (Delacroix et al., 2007; Lee et al., 2007; Takeishi et al., 2010). In response to DNA damage, 9-1-1 is loaded onto dsDNA by a Rad17-containing, RFC-like ‘clamp loader’ that recognizes junctions of RPA-ssDNA and double-stranded DNA (dsDNA) (Ellison and Stillman, 2003; Zou et al., 2003). The interaction between Rad9 and TopBP1 may help to recruit TopBP1 to sites of DNA damage and/or facilitates ATR-ATRIP activation (Delacroix et al., 2007; Lee and Dunphy, 2010). However, it remains elusive how the 9-1-1 and TopBP1 recruited to dsDNA engage the ATR-ATRIP on RPA-ssDNA.
A second major question on ATR activation is how ATR recognizes its substrates and transmits DNA damage signals. Several proteins implicated in ATR signaling, such as Rad17 and Chk1, are phosphorylated by ATR on chromatin (Smits et al., 2006; Zou et al., 2002), suggesting that ATR functions on damaged DNA. Furthermore, Rad17, Claspin, and Chk1 are known to associate with each other via a series of ATR-orchestrated events after phosphorylation (Kumagai and Dunphy, 2003; Wang et al., 2006). The role of these ATR substrates in signal transduction and the phosphorylation-mediated interactions among them suggest that ATR directs assembly of a dynamic signaling complex on DNA. Nonetheless, how ATR engages this signaling complex remains unknown.
To delineate the process of ATR activation, we sought to capture ATR in its active state, to molecularly define the state, and to dissect the biochemical events leading to this state. We found that during its activation, ATR, like ATM and DNA-PKcs, is transformed into a hyperphosphorylated state with multiple sites phosphorylated. Surprisingly, however, among the phosphorylation sites of ATR that we identified, only Thr 1989 is critical for robust ATR activation. The phosphorylation of Thr 1989 occurs in trans among the ATR-ATRIP complexes that congregate on RPA-ssDNA. Phosphorylated Thr 1989 is directly recognized by TopBP1, enabling TopBP1 to stably engage the ATR-ATRIP complex, to efficiently stimulate the kinase, and to act as a scaffold for ATR-substrate interactions. These findings reveal unexpected links among the recruitment, stimulation, and substrate recognition of ATR-ATRIP, presenting a clearer picture of how ATR is fully activated at sites of DNA damage.
To determine whether ATR is phosphorylated during activation, we used mass spectrometry to analyze Flag-tagged ATR purified from hydroxyurea (HU)-treated 293E cells. Our data showed that ATR was phosphorylated at Ser 428, Ser 435, Thr 1989, and possibly at Ser 436 and Ser437 (Figs. 1A and S1A). The phosphorylation of Ser 428 was previously shown by others using an antibody from Cell Signaling (www.cellsignal.com/products/2853.html). The phosphorylation of Ser 435 and Thr 1989 was documented by large-scale studies on protein phosphorylation (Daub et al., 2008; Dephoure et al., 2008). To date, none of these phosphorylation sites have been functionally characterized. The location of Thr 1989 in the FAT (FRAP, ATM, TRRAP) domain, a potential regulatory element conserved among PIKKs, prompted us to focus our initial analysis on this phosphorylation site.
We first asked whether the phosphorylation of T1989 is induced by DNA damage. To monitor T1989 phosphorylation in vivo, we generated phospho-specific antibodies to this site. In cells irradiated with ultraviolet (UV) light, the phospho-T1989 antibody specifically recognized Flag-tagged wild-type ATR (ATRWT), but not the T1989A mutant (ATRT1989A; Fig. 1B). Endogenous ATR was also recognized by the phospho-T1989 antibody after UV irradiation in several cell lines (Fig. S1B). Treatment of cell extracts with phosphatase reduced the recognition of ATR by the phospho-T1989 antibody (Fig. 1C). In addition to UV, ionizing radiation (IR) and HU also induced T1989 phosphorylation (Fig. 1D). Together, these results demonstrate that ATR is phosphorylated at T1989 in a DNA damage-induced manner.
The phosphorylation of ATR at T1989 occurs rapidly after DNA damage. Like the ATR-mediated Chk1 phosphorylation, T1989 phosphorylation was detected within 0.5 h after UV treatment (Fig. S1C). Unlike Chk1 phosphorylation, which declined after 2 h, T1989 phosphorylation persisted until 12 h post UV treatment. The UV-induced T1989 phosphorylation is dose-dependent. T1989 phosphorylation was readily detected in cells treated with 5 J/m2 of UV, and was maximally induced by 50 J/m2 of UV (Fig. S1D). Phosphorylated ATR was coimmunoprecipitated by ATRIP (Fig. 1E), showing that T1989 is phosphorylated in the ATR-ATRIP complex. Furthermore, T1989 phosphorylation was detected only in the chromatin fractions but not in the soluble fractions (Fig. 1F), suggesting that ATR is phosphorylated on chromatin. These features of T1989 phosphorylation are consistent with a potential role in ATR activation.
We next used the ATRT1989A mutant to investigate whether T1989 phosphorylation is implicated in ATR activation. Like ATRWT, ATRT1989A was able to phosphorylate a Rad17-derivative substrate (GST-Rad17) in vitro (Fig. 2A), showing that the T1989A mutation does not significantly alter the kinase domain. When inducibly expressed in stable cell lines, ATRT1989A, but not ATRWT, attenuated the ATR-mediated Chk1 phosphorylation after UV treatment (Fig. 2B; cell-cycle distributions shown in S2A). Moreover, even in the absence of UV, induction of ATRT1989A elicited H2AX phosphorylation in a large fraction of cells (Figs. S2B–C), indicating an increase in genomic instability. These results suggest that although ATRT1989A possesses an intact kinase domain, it interferes with the function of endogenous ATR.
To directly determine whether ATRT1989A is functional, we established ATRflox/−-derivative cell lines (Cortez et al., 2001) allowing inducible expression of Flag-HA-tagged ATRWT or ATRT1989A in cells devoid of endogenous ATR. Both ATRWT and ATRT1989A were expressed at levels similar to that of endogenous ATR in these cell lines (Fig. 2C). As expected, the ATRT1989A mutant expressed in the cell line was detected by ATR antibodies but not the phospho-T1989 antibody after UV irradiation, and it retained the ability to phosphorylate GST-Rad17 in vitro (Figs. S2D–E). In cells lacking endogenous ATR, Chk1 and Rad17 were efficiently phosphorylated by ATRWT but not ATRT1989A after UV treatment (Fig. 2C; lanes 4 and 8). To rule out the possibility that the compromised checkpoint response in ATRT1989A expressing cells is due to unexpected events during cell line generation, we tested additional independently generated cell lines that express ATRWT or ATRT1989A. Consistent with the experiment above, all ATRT1989A expressing cell lines displayed defective Chk1 activation (Fig. S2F). Furthermore, similar results were obtained using ATRflox/− cells infected with Cre-expressing adenovirus (Ad-Cre) and transfected with plasmids encoding Flag-ATRWT or Flag-ATRT1989A (Fig. 2D). Together, these results demonstrate that the ATRT1989A mutant is compromised in its ability to initiate robust checkpoint signaling.
In marked contrast to ATRT1989A, neither ATRS428A nor ATRS435/436/437A failed to activate Chk1 (Figs. 2D and S2G), showing that among the phosphorylation sites of ATR that we identified, T1989 is the only one critical for ATR activation. Since ATR is critical for genomic stability in cycling cells, deletion of ATR from ATRflox/− cells resulted in loss of cell viability (Cortez et al., 2001). Expression of ATRWT, but not ATRT1989A, suppressed the growth defects of Ad-Cre-infected ATRflox/− cells in both cell proliferation and colony formation assays (Figs. 2E and S2H). These results suggest that T1989 is not only critical for the activation of ATR by extrinsic DNA damage, but also for its essential function in cycling cells.
While the ATRT1989A mutant is defective for checkpoint response, the phospho-mimetic ATRT1989D mutant is fully functional in Chk1 activation after DNA damage (Fig. S2I). Furthermore, we note that ATRT1989D did not induce Chk1 phosphorylation in the absence of DNA damage, suggesting that T1989 phosphorylation is necessary but not sufficient for initiating robust checkpoint signaling. As described below, phosphorylated ATR functions in concert with other DNA damage sensors and TopBP1 to activate checkpoint response.
We next investigated which kinase is responsible for T1989 phosphorylation. T1989 is followed by a Pro residue, raising the possibility that it is a substrate of CDKs. However, inhibitors of various CDKs did not affect T1989 phosphorylation after UV (Fig. S3A). Treatment of cells with 50 μM of roscovitine for 14 h completely abolished the CDK-dependent Mcm2 phosphorylation (Montagnoli et al., 2006), but did not alter the UV-induced T1989 phosphorylation (Fig. 3A). In marked contrast, T1989 phosphorylation was clearly diminished by caffeine, a pan-inhibitor of ATR and ATM (Fig. 3B). To pinpoint the PIKK responsible for T1989 phosphorylation, we tested the effects of specific ATM and DNA-PKcs inhibitors. Even when used in combination at high concentrations, ATM and DNA-PKcs inhibitors did not eliminate T1989 phosphorylation (Figs. 3B and S3B). These results suggest that ATR, rather than ATM and DNA-PKcs, is likely responsible for the UV-induced T1989 phosphorylation.
If T1989 is autophosphorylated by ATR, one would expect that the kinase-deficient ATR mutant (ATRKD) is not phosphorylated at T1989. To test this possibility, we transiently expressed Flag-ATRWT and Flag-ATRKD in cells lacking endogenous ATR (Fig. 3C). In the absence of endogenous ATR, Flag-ATRWT but not Flag-ATRKD was phosphorylated at T1989 after UV treatment, showing that T1989 phosphorylation is dependent upon ATR activity. Inhibition of Chk1 did not alter T1989 phosphorylation, suggesting a direct role of ATR in this phosphorylation event (Fig. 3B).
To test whether T1989 is directly phosphorylated by ATR, we generated a GST-fusion protein that contains a peptide encompassing T1989 and its surrounding residues (GST-T1989). Like GST-Rad17, GST-T1989 was significantly phosphorylated by ATR (Fig. 3D). This phosphorylation of T1989 by ATR was specific because the kinase did not phosphorylate GST-T1989A, and the phosphorylation of T1989 was compromised when ATRKD was used. Together, these results suggest that T1989 is a direct substrate of ATR in vitro.
To pinpoint the role of T1989 phosphorylation in ATR activation, we asked if this event is dependent upon TopBP1. Knockdown of TopBP1 with siRNA dramatically reduced UV-induced Chk1 phosphorylation, but did not affect T1989 phosphorylation (Fig. 4A). On the other hand, in the absence of endogenous ATR and the presence of ATRT1989A, UV-induced TopBP1 phosphorylation at T1062 was compromised (Fig. 4B). Together these results show that T1989 phosphorylation is independent of TopBP1, but the phosphorylation of TopBP1 requires T1989 phosphorylation. Thus, T1989 phosphorylation likely functions upstream of TopBP1 during ATR activation.
In response to UV damage, ATR was coimmunoprecipitated by TopBP1 from the chromatin fractions (Fig. 4C). This damage-induced interaction of ATR and TopBP1 was abolished by phosphatase treatment of extracts (see Fig. 5A), showing its dependence on phosphorylation. In cells expressing ATRWT or ATRT1989A, only ATRWT but not ATRT1989A was efficiently coprecipitated by TopBP1 (Fig. 4C). Furthermore, TopBP1 was efficiently coprecipitated by ATRWT, but not ATRT1989A (Fig. 4D). These results suggest that T1989 phosphorylation is important for the interaction between ATR and TopBP1 after DNA damage.
To understand how TopBP1 interacts with phosphorylated T1989 (Figs. 4C–D, ,5A),5A), we generated two biotinylated peptides that contain phosphorylated or unphosphorylated T1989 and its surrounding residues. Only the phospho-T1989 peptide, but not the unphospho-peptide, captured endogenous TopBP1 from extracts (Fig. 5B). Using this binding assay and Flag-tagged TopBP1 fragments, we mapped the phospho-T1989-binding motif of TopBP1 to its BRCT domains 7 and 8 (Figs. 5C–D and S4A), which are required for activation of the ATR pathway (Gong et al., 2010). When overexpressed in human cells, a TopBP1 fragment containing only BRCT 7–8 associated with endogenous ATR (Fig. S4B). Furthermore, TopBP1 fragments lacking BRCT 1–6, expressed and purified from E. coli., directly bound to the phospho-T1989 peptide (Figs. 5E and S4C). These results demonstrate that TopBP1 directly engages phosphorylated ATR via BRCT 7–8.
To confirm the specificity of the interaction between BRCT 7–8 and phospho-T1989, we characterized the interaction using point mutants of both binding partners. When the phosphate-binding pocket of BRCT 7–8 was disrupted by the S1273A, R1280Q, or K1317M mutations (Leung et al., 2011), the interaction between purified BRCT 7–8 and phospho-T1989 was compromised (Fig. S4C). We recently showed that the binding of phospho-T1133 of BACH1 to BRCT 7–8 depends on its neighboring residues at the +3/+4 positions (Leung et al., 2011). Similarly, Ala substitutions of the +3 Glu and +5 Lys residues of T1989, which are highly conserved among the ATR orthologs in mammals (see Fig. S6D), significantly reduced the binding of phospho-T1989 to BRCT 7–8 (Fig. S4D). These results suggest that both phospho-T1989 and the +3/+5 residues contribute to the specific binding to BRCT 7–8.
T1989 phosphorylation is induced by DNA damage and it occurs on chromatin. ATRT1989A colocalized with RPA at DNA damage-induced foci (Fig. S5), suggesting that T1989 is not required for the localization of ATR to sites of DNA damage. In cells treated with ATRIP or RPA1 siRNA, UV-induced T1989 phosphorylation was diminished (Fig. 6A–B). In contrast, knockdown of Rad17, a regulator of ATR that does not affect the recruitment of ATR-ATRIP to RPA-ssDNA (Zou et al., 2002), did not alter T1989 phosphorylation (Fig. 6C). These results suggest that T1989 phosphorylation may be directly regulated by the recruitment of ATR-ATRIP to RPA-ssDNA.
The recruitment of ATR-ATRIP by RPA-ssDNA may bring multiple ATR-ATRIP complexes together and promote ATR autophosphorylation in trans. Consistent with this possibility, purified Flag-tagged ATR-ATRIP pulled down increased amounts of GFP-ATR from extracts in the presence of RPA-ssDNA (Fig. 6D), suggesting that multiple ATR-ATRIP complexes congregate on RPA-ssDNA. To test whether ATR can phosphorylate T1989 in trans, we generated an ATR mutant lacking the kinase domain at the C terminus (ATRΔC). In the presence of endogenous ATR, ATRΔC was efficiently phosphorylated at T1989 after UV treatment (Fig. 6E). This result, although does not exclude the possibility of ATR cis autophosphorylation, demonstrates that ATR can indeed autophosphorylate T1989 in trans after DNA damage.
The crosstalk among ATR molecules on RPA-ssDNA raised the possibility that the defect of ATRT1989A might be complemented in trans. To assess this possibility, we generated an ATRKD,T1989D double mutant. Since ATRKD,T1989D is inactive as a kinase, it does not directly contribute to ATR substrate phosphorylation. However, the phospho-mimetic mutation of ATRKD,T1989D may allow it to bring in TopBP1 and facilitate activation of neighboring ATRT1989A molecules on RPA-ssDNA. Indeed, coexpression of ATRKD,T1989D and ATRT1989A in cells lacking endogenous ATR partially rescued UV-induced Chk1 activation (Fig. 6F). This result strongly suggests that the defect of ATRT1989A stems from its compromised ability to interact with TopBP1, and this defect can be partially complemented in trans by the phospho-mimetic ATRKD,T1989D mutant.
The results above suggest that T1989 phosphorylation is a crucial event linking ATR-ATRIP recruitment to TopBP1-mediated ATR-ATRIP activation. To directly test whether T1989 is important for stimulation of the specific kinase activity of ATR-ATRIP, we purified ATRWT-ATRIP and ATRT1989A-ATRIP complexes from 293E cells and GST-tagged TopBP1 from E. coli. To ensure that the in vitro kinase assay measures the effect of TopBP1 on ATR-ATRIP activity rather than substrate binding, we used GST-Rad17, which only contains a short peptide from Rad17, as substrate. Compared to ATRWT-ATRIP, ATRT1989A-ATRIP was stimulated by full-length TopBP1 (TopBP1WT) less BRCT7-efficiently (Fig. 7A). Furthermore, a TopBP1 fragment lacking BRCT 7–8 (TopBP1ΔBRCT7–8) stimulated ATRWT-ATRIP less efficiently than TopBP1WT (Fig. S6A). Thus, the efficient stimulation of ATR-ATRIP by TopBP1 relies on the ATR-TopBP1 interaction mediated by the phospho-T1989 of ATR and the BRCT 7–8 of TopBP1.
A number of proteins participating in ATR signaling are phosphorylated after DNA damage. One example of these proteins is Rad9 (Roos-Mattjus et al., 2003), which binds to TopBP1 via its BRCT 1–2. The ability of TopBP1 to bind ATR-ATRIP and Rad9 through distinct BRCT domains raised the possibility that TopBP1 might function as a scaffold to facilitate substrate phosphorylation. To test this possibility, we performed in vitro ATR-ATRIP kinase assays in the absence of DNA and the presence of TopBP1WT or TopBP1ΔBRCT1–2 (a TopBP1 fragment lacking BRCT 1–2). Purified 9-1-1 complex and GST-Rad17 were used as TopBP1-bound and free substrates, respectively (Fig. S6B) (Yang and Zou, 2006; Zou et al., 2003). Both TopBP1WT and TopBP1ΔBRCT1–2 stimulated the phosphorylation of GST-Rad17 by ATR-ATRIP (Fig. 7B), suggesting that the function of BRCT 1–2 in recruiting TopBP1 to DNA is bypassed in vitro. In contrast, only TopBP1WT but not TopBP1ΔBRCT1–2 enhanced the phosphorylation of Rad9 by ATR-ATRIP (Fig. 7C), showing that the BRCT 1–2 of TopBP1 promote Rad9 phosphorylation independently of ATR-ATRIP stimulation. Together, these results suggest that in addition to the ability to stimulate ATR-ATRIP, a distinct scaffolding function of TopBP1 is also needed for efficient Rad9 phosphorylation.
In this study, we identified ATR autophosphorylation at T1989 as a hallmark of the active state of ATR. Furthermore, we elucidated how this phosphorylation event functions as a molecular switch for checkpoint activation.
The recruitment of ATR-ATRIP by RPA-ssDNA plays a crucial role in the spatial regulation of ATR. It has been proposed that the colocalization of ATR-ATRIP and its substrates at sites of DNA damage facilitates phosphorylation of ATR substrates (Zou and Elledge, 2003). Whether RPA-ssDNA has a direct role in activating the kinase is not known. In contrast to RPA-ssDNA, TopBP1 directly stimulates the activity of ATR-ATRIP (Kumagai et al., 2006). While TopBP1 is clearly important for ATR signaling, how TopBP1 is regulated by DNA damage and whether it is needed for the phosphorylation of all ATR substrates in humans is not addressed. This study suggests that the recruitment and the stimulation of ATR-ATRIP are, in fact, two mechanistically coupled events. RPA-ssDNA not only spatially regulates ATR-ATRIP, but also directly contributes to its activation by promoting ATR autophosphorylation at T1989, which directs TopBP1 to stimulate ATR-ATRIP (Fig. 7D). Since T1989 phosphorylation is independent of TopBP1, ATR, through its basal activity, plays a surprising role in promoting its own activation. Together, our results suggest that the full activation of ATR by DNA damage is sequentially driven by RPA-ssDNA, ATR autophosphorylation, and TopBP1, thus unifying the previous studies that demonstrated the dependence of ATR signaling on RPA-ssDNA and TopBP1 (Ciccia and Elledge, 2010; Cimprich and Cortez, 2008; Flynn and Zou, 2010).
The length of ssDNA induced by DNA damage is an important determinant for ATR activation (Byun et al., 2005; MacDougall et al., 2007; Sartori et al., 2007; Shiotani and Zou, 2009). In vitro, ATR-ATRIP binds to RPA-ssDNA in a length-dependent manner (Zou and Elledge, 2003). We suggest that the binding of ATR-ATRIP to RPA-ssDNA leads to congregation of ATR-ATRIP complexes and T1989 phosphorylation in trans (Fig. 7D). In this model, RPA-ssDNA is not only a quantitative signal for ATR-ATRIP recruitment, but also a length-dependent platform that promotes ATR autophosphorylation. The coupling of ATR-ATRIP recruitment and one of the initial events of ATR activation may be the key mechanism ensuring the quantitative regulation of ATR signaling by ssDNA.
In addition to the length of ssDNA, the number of ssDNA-dsDNA junctions is another important determinant for the strength of ATR signaling (MacDougall et al., 2007; Van et al., 2010), suggesting that the DNA damage sensors on dsDNA also quantitatively regulate ATR activation. Nonetheless, how exactly the ATR-ATRIP on ssDNA is regulated by the DNA damage sensors on dsDNA remains elusive. Independently of ATR, the Rad17-RFC complex recognizes ssDNA-dsDNA junctions and loads 9-1-1 onto dsDNA (Ellison and Stillman, 2003; Zou et al., 2003). Rad9 in the 9-1-1 complex interacts with TopBP1 (Delacroix et al., 2007; Lee et al., 2007; Makiniemi et al., 2001), whereby recruiting TopBP1 to sites of DNA damage. Although 9-1-1 brings TopBP1 to the vicinity of ATR-ATRIP, the TopBP1 and 9-1-1 on dsDNA are physically separate from the ATR-ATRIP on ssDNA. Our data that TopBP1 stably engages phosphorylated ATR via the BRCT domains 7 and 8 reveals a previously unknown step during ATR activation (Fig. 7D). The TopBP1 recruited by 9-1-1 needs to be redirected to ATR-ATRIP by this phosphorylation-dependent interaction, allowing the DNA damage sensors on ssDNA and dsDNA to act jointly to fully activate ATR-ATRIP. The interaction between BRCT7–8 and phospho-T1989 may poise the AAD of TopBP1 to stimulate ATR-ATRIP (Mordes et al., 2008). Additionally, as indicated by the trans complementation of ATRT1989A, phospho-ATR may tether TopBP1 to stimulate multiple ATR-ATRIP complexes on RPA-ssDNA (Choi et al., 2010).
In addition to stimulating the ATR-ATRIP kinase, the interaction between TopBP1 and ATR may facilitate ATR to recognize its substrates. In fission yeast, Rad4/Cut5, the homologue of TopBP1, interacts with Rad3/ATR substrates Rad9, Crb2 and Chk1 (Furuya et al., 2004; Mochida et al., 2004; Saka et al., 1997). In vertebrates, TopBP1 interacts with Rad9 through its N terminal BRCT domains 1 and 2 (Delacroix et al., 2007; Lee et al., 2007). Xenopus TopBP1 remains associated with 9-1-1 and Rad17 after DNA damage (Lee and Dunphy, 2010). When phosphorylated by ATR in human cells, Rad17 interacts with the mediator protein Claspin (Wang et al., 2006). ATR-dependent phosphorylation of Claspin enables it to associate with Chk1, promoting Chk1 phosphorylation by ATR and stimulating Chk1 activity (Kumagai et al., 2004; Lindsey-Boltz et al., 2009). Together, these findings suggest that a checkpoint signaling complex is physically linked to the N terminus of TopBP1. Our result that the BRCT 1–2 of TopBP1 promote Rad9 phosphorylation independently of their function in TopBP1 recruitment provides an example of how TopBP1 functions as a scaffold to bring ATR to its substrates (Fig. S6C). Interestingly, Dpb11, the budding yeast homologue of TopBP1, is known to associate with phosphorylated Sld3 and Sld2 through its BRCT repeats at the N and C termini (Tanaka et al., 2007; Zegerman and Diffley, 2007). During the initiation of DNA replication, these interactions allow Dpb11 to function as a scaffold and recruit the GINS complex to the MCM-Cdc45 complex at replication origins (Muramatsu et al., 2010). It is tempting to speculate that human TopBP1 functions analogously as a scaffold of phosphorylation-mediated protein complexes during the initiation of DNA replication and activation of the ATR checkpoint.
Reminiscent to the assembly of the replication complex at origins, ATR, TopBP1 and other proteins are assembled into a signaling complex at sites of DNA damage. The phosphorylation-mediated ATR-TopBP1 interaction may be one of the key events for the assembly of this signaling complex. Consistent with the model in which ATR functions in concert with Rad17, 9-1-1, and TopBP1 on damaged DNA, phospho-ATR is unable to phosphorylate Chk1 after DNA damage in the absence of Rad17 or TopBP1 (Figs. 4A and and6C),6C), and the phospho-mimetic ATRT1989D mutant cannot induce Chk1 activation in the absence of DNA damage (Fig. S2I). While the interaction between phospho-ATR and BRCT 7–8 is specific, it is weak and only captured under mild binding conditions (data not shown), which may explain why phospho-ATR is not sufficient to efficiently recruit TopBP1 in the absence of Rad17. These findings emphasize the requirement of both ssDNA and dsDNA for the assembly of the ATR signaling complex. It is interesting to note the phosphorylation of ATR at T1989 persists much longer than Chk1 phosphorylation after DNA damage (Fig. S1D), suggesting that the ATR signaling complex is dynamically regulated during the course of DNA damage response.
Like ATR, ATM and DNA-PKcs undergo autophosphorylation in or near the FAT domain during activation, indicating that this is a common feature of PIKKs (Bakkenist and Kastan, 2003; Chan et al., 2002; Kozlov et al., 2006). However, the autophosphory-lation of each of the PIKKs appears to be functionally distinct (Bakkenist and Kastan, 2003; Berkovich et al., 2007; Daniel et al., 2008; Ding et al., 2003; Lee and Paull, 2005; Pellegrini et al., 2006; So et al., 2009; Uematsu et al., 2007), suggesting that these modifications may have adapted to different regulatory roles during evolution. Interestingly, the number of BRCT domains of TopBP1 has risen during evolution, suggesting that it may also have acquired new regulatory roles (Garcia et al., 2005). Among the three PIKKs activated by DNA damage, ATR is the only one that is recruited and activated by different factors, and the autophosphorylation of ATR plays a key role in integrating the functions of these factors. We note that the sequence conservation of T1989 is only apparent in mammals (Fig. S6D). It will be interesting to elucidate whether the homologues of human ATR in other organisms are autophosphorylated during activation, and whether the role of ATR autophosphorylation is conserved.
Plasmids expressing Flag-ATRWT and Flag-ATRKD were previously described (Tibbetts et al., 2000). Plasmids encoding Flag-ATRT1989A, Flag-ATRS428A, Flag-ATRS435/436/437A, and Flag-ATRΔC (amino acids 2350–2644 deleted) were derived from Flag-ATRWT using site-directed mutagenesis. The pNZ2-ATRWT plasmid encoding HA-ATRWT was provided by Dr. David Cortez (Mordes and Cortez, 2008). The pNZ2-ATRT1989A plasmid was derived from pNZ2-ATRWT using site-directed mutagenesis. The entire coding sequence of ATR in the plasmids encoding ATRT1989A was confirmed by DNA sequencing. For purification of the ATR-ATRIP complex from 293E cells, Flag-ATR and His-ATRIP were cloned into the pTT3 vector. The Flag-CMV2-TopBP1 plasmid is a gift from Dr. Jiri Lukas. Plasmids expressing Flag-TopBP1Δ1–2, Flag-TopBP1Δ1–5, Flag-TopBP1Δ1–6, Flag-TopBP1Δ7–8, Flag-TopBP1-BRCT7+8, and GST-TopBP1Δ1–6 were generated either by direct cloning or with the pUNI system (Liu et al., 1998).
The phospho-specific antibodies to ATR pT1989 were generated by Cell Signaling and Bethyl against the peptide NH2-Cys-FPENE(pT)PPEGK-COOH. ATR, TopBP1, phospho-Rad17, phospho-TopBP1, and phospho-Mcm2 antibodies are from Bethyl. ATRIP antibodies were previously described (Cortez et al., 2001). Rad17 and Chk1 antibodies are from Santa Cruz. Phospho-Chk1 antibody is from Cell Signaling. Phospho-H2AX antibody is from Millipore.
To precipitate ATR and TopBP1 from the chromatin fractions, cell extracts were first fractionated as previously described (Zou et al., 2002). To release chromatin-bound proteins, chromatin-enriched pellet (P3) were sonicated and digested with benzonase in binding buffer (100 mM Na2HPO4, 2 mM K2HPO4, 137 mM NaCl, 2.7 mM KCl). The resulting lysates were centrifuged 14,000 rpm for 10 m, and the supernatants were collected and precleared with protein G/A sepharose. The solubilized chromatin fractions were subsequently incubated with primary antibodies and 20μl Protein G/A beads, and the beads were collected and washed four times with binding buffer.
The pT1989 (CFPENEpTPPEGKNML) and the corresponding unphosphorylated peptides were synthesized by the Tufts University Core Facility. In pull-down assays, biotinylated peptides were attached to streptavidin-coated Dynalbeads, and incubated with nuclear extracts or purified proteins. The beads were subsequently retrieved, extensively washed with PBS containing 0.12–0.15% Triton X-100, and bound proteins were subjected to SDS-PAGE and immunoblotting.
The ATR kinase assays were performed essentially as previously described (Canman et al., 1998) with the following modifications. HEK 293E cells were transfected with Flag-ATR and His-ATRIP expressing plasmids, and Flag-ATR was immunoprecipitated with anti-Flag M2 antibody in the TGN buffer [50 mM Tris-HCl (pH 7.5), 150 mM NaCl, 50 mM phosphoglycerol, 10 % glycerol, 1% Tween 20, 1 mM PMSF, 1 mM NaF, 1 mM Na3VO4, and 1 mM DTT, and protease inhibitors]. The precipitates were washed twice with the TGN buffer, once with the TGN buffer supplemented with 0.5 M LiCl, and twice with the reaction buffer [10 mM Hepes (pH 7.5), 50 mM NaCl, 10 mM MgCl2, 50 mM glycerophosphate, 1 mM DTT, and protease inhibitor] without ATP. The in vitro kinase reactions were conducted in the presence of 5 μM ATP and purified GST-T1989 (GST-PENETPPETPPEGK) or GST-T1989A as substrates. The phosphorylation of T1989 was monitored using the phospho-T1989 antibody.
To analyze the stimulation of ATR-ATRIP by TopBP1, the ATR-ATRIP complex was purified using a two-step protocol. The ATR-ATRIP complex, which contains Flag-ATR and His-ATRIP, was first purified using Ni-beads and eluted with imidazole. The proteins eluted from Ni-beads were subsequently incubated with anti-Flag M2 beads, and the ATR-ATRIP complex was eluted with 200 μg/ml 3x Flag peptide. The kinase reactions were conducted with purified ATR-ATRIP complex, GST-Rad17, and 5 μCi [γ-32P] ATP in the presence or absence of GST-TopBP1.
We thank S. Elledge, A. Ciccia, J. Bartek, J. Lukas, D. Cortez, D. Chen, B. Chen, Q. Ge, R. Polakiewicz for reagents, and members of the Zou lab for discussion. This work was supported by grants from NIH (GM076388), ACS (RSG-08-297), and the Federal Share of MGH Proton Program to L. Z.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.