Preferential binding of ATR protein to UV-damaged DNA

Cell Culture and Antibodies.

The simian virus 40-transformed human embryonic kidney (HEK)293T cells were maintained in DMEM (GIBCO/BRL) supplemented with 10% FBS and 100 units of penicillin and streptomycin/ml. Mouse monoclonal anti-Flag antibody was obtained from Sigma.

Plasmid Constructs.

Two Flag-ATR constructs containing the full-length cDNA clone of human ATR and its kinase-dead version (K2327R) in pcDNA3 (Invitrogen) expression vector were kindly provided by R. T. Abraham [Duke University (17)]. By using PCR, various regions of the cDNA encoding ATR were amplified (see delineation of the regions in Fig. 1). In each case, the 5′ primer contained an ATG codon followed by a Flag-epitope in frame with the coding region that was amplified. These PCR products were digested with BamHI and XhoI restriction enzymes and ligated into pcDNA4.1 (Invitrogen) expression vector through the same enzyme sites to generate the N-terminal Flag epitope-tagged ATR fragments.

Transfections and Cell Lysis.

Transfections were carried out by using the calcium phosphate method. For transfections, 8 × 10⁶ 293T cells were plated per 225 cm² flask and allowed to grow for 24 h. Plasmid DNA (25 μg) was mixed with CaCl₂, and this mixture was added to a 2 × HEBS (140 mM NaCl/1.5 mM Na₂HPO₄/50 mM Hepes, pH 7.05) solution and incubated for 8 min at room temperature. The CaCl₂/DNA/HEBS precipitate was then added dropwise to the cell medium. After 16-h incubation at 37°C in 5% CO₂, cells were washed twice in serum-free DMEM and fresh media added to the cells for a further 48 h of incubation. Cells were lysed in ice-cold lysis buffer [50 mM Tris, pH 7.5/150 mM NaCl/10 mM β-glycerophosphate/10% glycerol/1% Tween-20/0.1% Nonidet P-40/1 mM Na₃VO₄/1 mM NaF and protease inhibitors (Roche Molecular Biochemicals)] for 30 min. The cell lysates were centrifuged for 30 min at 32,000 × g or frozen in dry ice and stored at −80°C until use.

Protein Purification.

The supernatants from the spun lysates were incubated at 4°C for at least 4 h with anti-FLAG M2 affinity gel (Sigma) that had been preequilibrated in TBS buffer (50 mM Tris⋅HCl, pH 7.5/150 mM NaCl). The beads were then washed once with 1 M NaCl containing TBS buffer and three times with 150 mM NaCl containing TBS. The ATR bound to beads was either used directly for pull-down experiments or eluted with TBS buffer containing 200 μg/ml FLAG peptide (Sigma).

Human p53 protein overexpressed in SF9 cells was generated in the laboratory of J.D.G., as described (18).

DNA Substrates.

Duplex DNAs of 46 bp, containing a single T[6–4]T photoproduct and its unmodified version, were prepared from four complementary oligomers. The sequences of the oligonucleotides are:

 5′-ctagcgggatccggtgcaggtaT[6-4]Tatggaattcgtagatctgcgtc-3′

 3′-gatcgccctaggccacgtccataataccttaagcatctagacgcag-5′

The T[6–4]T photoproduct containing substrate was constructed by ligating the damage-containing oligomer (8-mer), which was 5′-terminally labeled, with other partially overlapping oligonucleotides to obtain the internally labeled duplex as described (19, 20). The resulting 46-bp duplex contains the (6–4) photoproduct at positions 23 and 24. The full-length duplexes were separated from the ligation mixture by gel purification through 5% nondenaturing polyacrylamide gel. These substrates were used for pull-down assays. For protein–DNA crosslinking experiments, a 50-mer substrate with an azidophenacyl photosensitizer of the following sequence was used:

5′-gcgccagctggccaccctgagaagctacgagcgagcgccaagcttgggct-3′.

These 50-mer duplex DNAs were purchased from Operon Technologies (Alameda, CA). The azidophenacyl moiety was coupled to the oligonucleotide as described (21).

For electron microscopy and kinase assays, pGEMEX1 dsDNA cooled on ice in a microtiter plate was irradiated with 254-nm light at a rate of 1 mW/cm² by using a germicidal UV-C (254-nm) lamp (General Electric).

Electron Microscopy.

Fifty nanograms of ATR protein was incubated with 200 ng of linearized or supercoiled pGEMEX1 dsDNA in kinase buffer containing 25 mM Hepes, pH 7.9, 50 mM KCl, 10 mM MgCl₂, 2 mM MnCl₂, 20% glycerol, 0.1% Nonidet P-40, 1 mM DTT, and 0.5 mg/ml of BSA for 15 min at 30°C. After incubation, DNA–protein complexes were fixed with 0.6% glutaraldehyde for 10 min at 20°C. The reaction mixture was chromatographed on 2 ml of BioGel A5m (BioRad), equilibrated with TE buffer (10 mM Tris⋅HCl/1 mM EDTA, pH 8.0) to remove fixatives and unbound protein. The fractions containing DNA–protein complexes were collected and prepared for electron microscopy as described (22) and analyzed by using a Philips (Eindhoven, The Netherlands) CM12 electron microscope. The images were scanned by using a Nikon-4500 film scanner and contrast adjusted by using Adobe photoshop software (Adobe Systems, Mountain View, CA). The size of protein particles bound to DNA was estimated as described elsewhere (23).

Pull-Down Assays.

The Flag-ATR protein bound to the beads was washed first with the kinase buffer and then incubated with internally radiolabeled DNA substrates at 30°C for 30 min in a reaction buffer of 25 μl. The beads were washed three times with 0.1% Nonidet P-40 containing TBS buffer at 4°C. The DNA-bound ATR on the beads was either loaded onto 10% SDS/PAGE or eluted with TBS buffer containing 200 μg/ml of FLAG peptide (Sigma) and then visualized by silver staining. The amount of DNA bound to ATR was measured with a PhosphorImager and the imagequant system (Molecular Dynamics) by comparing the radioactivity bound to ATR-containing resin to the radioactivity in control lanes containing known amounts of DNA.

Protein–DNA Crosslinking.

Azidophenacetyl-derivatized DNA (2 fmol) was incubated with approximately 100 ng of ATR in 10 μl of kinase buffer at 30°C for 30 min. Reaction mixtures were transferred to a microtiter plate and irradiated with a 366-nm light from BLB lamps (General Electric) at a rate of 1 mW/cm² at a distance of 5 cm. The samples were covered with the polystyrene lid of the microtiter plate to shield them from wavelengths below 300 nm and cooled on ice during illumination. Crosslinked proteins were resolved on 7.5% SDS/PAGE, and gels were silver stained, dried, and exposed to film.

Kinase Assays.

Kinase assays were performed in 20-μl reactions by using 50 ng of purified ATR, 50 ng of purified p53, and linearized pGEMEX1 DNA with and without UV treatment at the indicated concentrations in a kinase buffer containing 25 mM Hepes, pH 7.9, 50 mM KCl, 10 mM MgCl₂, 2 mM MnCl₂, 20% glycerol, 0.1% Nonidet P-40, 1 mM DTT, and 0.5 mg/ml of BSA. ATR and DNA mixtures were incubated on ice for 5 min before substrate and ATP addition. ATP was added to a final concentration of 50 μM along with 5 μCi[γ-³²P]-ATP. Reaction mixtures were incubated at 30°C for 30 min. Phosphorylated proteins were subjected to 10% SDS/PAGE, visualized by silver staining, and exposed on a Molecular Dynamics PhosphorImager screen for quantitation by imagequant software.

Purification of Full-Length ATR and ATR Fragments for DNA–Protein Interactions.

We wished to determine whether ATR interacts directly with DNA and if it does, whether this interaction is confined to a specific domain within the protein. Fig. 1A schematically shows the organization of the ATR protein (24). This large protein contains a limited number of sequence motifs as indicated: these include an N-terminal leucine-rich domain, an ATM homology domain of about 100 amino acids in its middle region, and a C-terminal domain with homology to PIKK family proteins, which contains the kinase active site residues. To investigate ATR-DNA interaction and the effect of this interaction on kinase activity, we used FLAG-tagged wild-type and kinase-dead ATR constructs in a mammalian expression vector as described (17, 25) as well as FLAG-tagged constructs expressing the N-terminal half (N), the middle (M), and the C-terminal half (C) of the protein as indicated in Fig. 1A. First we purified the full-length wild-type and kinase-dead ATR by affinity chromatography and tested them for purity by SDS/PAGE and silver staining and for activity by an autophosphorylation assay. As seen in Fig. 1B, the proteins are of high purity, and the kinase-dead mutant does not exhibit detectable kinase activity in our assay. Fig. 1C shows the expression of the ATR fragments on transfection of HEK293T cells with the appropriate constructs. In general, we found that the ATR full-length, ATR-N, and ATR-M proteins were expressed at comparable levels, but ATR-C was expressed at a significantly lower amount.

Binding of ATR to DNA.

Analysis of ATR-DNA binding by conventional electrophoretic mobility-shift assay in conjunction with “supershift” by anti-ATR antibodies indicated that ATR directly binds DNA (data not shown). However, because of the large size of the protein, data obtained by this method were qualitative and variable. Therefore we used a “pull-down” assay to analyze ATR–DNA interactions. HEK293T cells transfected with vectors expressing full-length ATR or ATR fragments were lysed, and the tagged ATR proteins were purified by binding to FLAG-agarose beads. Radiolabeled DNA was added to the beads and, after incubation and exhaustive washes, the proteins on the beads were released with SDS, visualized by silver staining, and the DNA was detected by autoradiography. Fig. 2 Top shows the relative levels of ATR and its derivatives on the beads. The full-length protein and the N-terminal and middle fragments are expressed at comparable levels and the C-terminal fragment is at a lower amount. Fig. 2 Middle shows the autoradiographic analysis of bound DNA and Bottom shows the quantitative analysis of the data from Middle. It appears that only full-length ATR binds DNA as the amount of DNA associated with ATR fragments is not above background. Although the use of those fragments in the pull-down experiment shows that ATR fragments alone cannot bind to DNA, the presence of ATR fragments on the beads confirms the specificity of full-length ATR interaction with DNA, indicating that the full-length ATR interaction with DNA is not a result of nonspecific binding of DNA onto resins.

Our ATR preparations are quite pure (see Fig. 1), and therefore the most likely explanation of the data in Fig. 2 is that ATR directly binds DNA. However, it was conceivable that binding was mediated by small quantities of a contaminant not detectable by our analyses. Therefore, we wished to confirm direct ATR-DNA binding by photocrosslinking. DNA–protein crosslinking by UV occurs only when the two are in direct contact, and therefore this method is referred to as “zero-distance” crosslinking (26). To perform UV crosslinking, full-length ATR and ATR fragments were purified, mixed with azidophenacetyl derivatized and radiolabeled DNA, and exposed to 366-nm light. The reaction products were separated on an SDS/PAGE, which was then analyzed by silver staining (Fig. 3A), and autoradiography (Fig. 3B). We find that full-length ATR is crosslinked to DNA, whereas the ATR fragments are not (Fig. 3B). Thus, we conclude that ATR is a DNA-binding protein, and that either the DNA binding requires amino acids from the entire protein or the fragments we have expressed are not in a conformation conducive to bind DNA.

Preferential Binding of ATR to UV (6–4) Photoproduct.

An important question in the field of DNA damage checkpoint is the identity of the signal initiating the checkpoint cascade. Although it is generally believed that lesions processed by the replication or repair systems generate a common structure that is recognized by the DNA damage checkpoint pathway, the possibility exists that under certain circumstances the damage per se is bound by the sensor. UV induces two major photoproducts in DNA: cyclobutane pyrimidine dimers, which make up 80–90% of the lesions, and (6–4) photoproducts, which constitute 10–20% of the damage. Cyclobutane pyrimidine dimers, in contrast to (6–4) photoproducts, are poorly recognized by damage sensors involved in DNA repair (27), and we reasoned that the same may be true for the damage sensors in checkpoint response as well. Therefore, we compared the binding of ATR to an oligomer of 46 bp with and without a (6–4) photoproduct by using the pull-down assay. As shown in Fig. 4, ATR has a significantly higher affinity for the oligomer with the (6–4) photoproduct. Considering that a single photoproduct increases the affinity of ATR to the 46-mer 1.6-fold, it is calculated that ATR has about 46 × 1.6≈80-fold higher affinity to a (6–4) photoproduct relative to an undamaged dinucleotide.

Analysis of ATR-DNA Complexes by Electron Microscopy.

Solution studies on ATR-DNA interaction were complemented by electron microscopy. Fig. 5A shows representative pictures of ATR-DNA complexes. The protein binds with equal affinity to linear and supercoiled DNA; under our assay conditions, 10% of supercoiled or linear DNA molecules were bound by ATR (Fig. 5A). It appears that ATR is a globular monomeric protein with no special affinity for DNA ends. However, in about half of the micrographs, the protein was seen bound to two DNA segments, indicating that ATR may have two DNA-binding sites or that alternatively two monomers bound to different regions of the DNA molecule bringing these regions into proximity by protein–protein interaction. Obviously, this point deserves further investigation by other methods.

In addition to providing information regarding the size and the shape of ATR, electron microscopy also enabled us to measure the affinity of ATR to damaged DNA by an independent method. Unirradiated or UV-irradiated plasmid DNA was mixed with ATR, and the fractions of DNA molecules with bound ATR were scored. As seen in Fig. 5B, gradually more ATR is bound to DNA with increasing UV dose. Thus, this experiment, combined with the pull-down assay shown in Fig. 4, strongly suggests that this binding may enable ATR to act as a sensor in the DNA damage checkpoint cascade.

Stimulation of ATR Kinase Activity by DNA Damage.

If specific binding of ATR to DNA damage is an early step in the UV-induced DNA damage checkpoint response, DNA damage might be expected to modulate the kinase activity of ATR. To test this prediction, ATR was incubated with one of its better known substrates, p53, in the presence of undamaged and UV-damaged DNA (Fig. 6A). There is a high basal level of p53 phosphorylation in the absence of DNA, which is stimulated by DNA (28) and to a significantly higher level by UV-damaged DNA (Fig. 6B). Thus, our data are consistent with the notion that ATR is a damage sensor and is directly activated by the primary DNA lesion.