Protection of telomeres 1 proteins POT1a and POT1b can repress ATR signaling by RPA exclusion, but binding to CST limits ATR repression by POT1b

Comprised of telomeric TTAGGG repeats and shelterin, telomeres ensure that the natural ends of chromosomes remain impervious to the DNA damage response. Telomeres carry a long constitutive 3′ overhang that can bind replication protein A (RPA) and activate the ATR Ser/Thr kinase (ATR), which induces cell cycle arrest. A single-stranded (ss) TTAGGG repeat–binding protein in mouse shelterin, POT1a, has been proposed to repress ATR signaling by preventing RPA binding. Repression of ATR at telomeres requires tethering of POT1a to the other shelterin subunits situated on the double-stranded (ds) telomeric DNA. The simplest model of ATR repression, the “tethered exclusion model,” suggests that the only critical features of POT1a are its connection to shelterin and its binding to ss telomeric DNA. In agreement with the model, we show here that a shelterin-tethered variant of RPA70 (lacking the ATR recruitment domain) can repress ATR signaling at telomeres that lack POT1a. However, arguing against the tethered exclusion model, the nearly identical POT1b subunit of shelterin has been shown to be much less proficient than POT1a in repression of ATR. We now show that POT1b has the intrinsic ability to fully repress ATR but is prevented from doing so when bound to Ctc1, Stn1, Ten1 (CST), the complex needed for telomere end processing. These results establish that shelterin represses ATR with a tethered ssDNA-binding domain that excludes RPA from the 3′ overhang and also reveal an unexpected effect of CST on the ability of POT1b to repress ATR.

One of the many tasks of the telomeric shelterin complex is to repress the ATR-dependent DNA damage response at the natural ends of chromosomes (reviewed in Refs. [1][2][3]. The main mode of ATR 3 activation involves the binding of the abundant trimeric RPA (RPA70, RPA32, and RPA14) complex to ssDNA, recruitment of ATR through an interaction between RPA70 and the ATR interacting partner ATRIP, loading of the 9-1-1 (Rad9 -Hus1-Rad1) complex on the end of the duplex, and activation of ATR by 9-1-1-bound TopBP1 (reviewed in Ref. 4). A second pathway involves activation of ATR by RPA-bound ETAA1 and is independent of 9-1-1 and TopBP1 (5,6).
The two DNA structures required for TopBP1-dependent ATR activation are present at telomeres, which carry a 9-1-1loading site and a 3Ј overhang of TTAGGG repeats, sufficiently long to bind one or more RPA trimers. Although telomeres often occur in the t-loop configuration (7,8), this structure is unlikely to guard against ATR signaling because the base of the t-loop retains the critical features required for loading of RPA and 9-1-1 (reviewed in Ref. 3). Rather, ATR repression is achieved by the POT1 subunit of shelterin (9), which binds to the ss TTAGGG repeats (10). Deletion of POT1 leads to activation of ATR at most telomeres in a manner that depends on RPA and TopBP1 (11). Super-resolution STORM imaging showed that this ATR activation takes place despite the presence of t-loops (8).
POT1 proteins bind the sequence 5Ј-TTAGGGTTAG-3Ј either at a DNA end or at an internal position using two oligosaccharide/oligonucleotide binding (OB) folds in their N-terminal half (10,12,13). The interaction of POT1 with ss TTAGGG repeats is neither sufficient nor necessary to target the protein to telomeres (14 -17). Accumulation of POT1 at telomeres requires its association with shelterin, which is mediated by the binding of the shelterin subunit TPP1 to the C-terminal half of POT1 (15,16,18). TPP1 binds to the TIN2 shelterin subunit, which in turn interacts with the duplex telomeric DNA-binding proteins TRF1 and TRF2, thus anchoring POT1 on the duplex telomeric DNA.
As a result of a gene duplication event, rodent telomeres contain two closely related POT1 proteins, POT1a and POT1b, which have diverged in function. Both POT1a and POT1b can repress ATR kinase signaling in G 1 but repression of ATR in S/G 2 requires POT1a (11,19,20). In S phase, POT1b, but not POT1a, governs the postreplicative generation of the telomeric 3Ј overhang (21)(22)(23). In part, POT1b executes this function through the recruitment of CST (Ctc1, Stn1, Ten1) (reviewed in Ref. 24), an RPA-like complex that binds ssDNA and interacts with polymerase ␣/primase (pol␣/primase) to mediate fill-in synthesis that restores the correct overhang length (25).
Telomeres lacking POT1 proteins accumulate RPA, which is not detectable at functional telomeres (11,26). This observation suggested a simple competition model whereby the binding of POT1 to the ssDNA prevents RPA-dependent ATR activation (9, 11, 26 -28). As RPA is ϳ200-fold more abundant than POT1a and POT1b and has the same subnanomolar affinity for ss telomeric DNA (29), it is unlikely that simple competition could explain the repression of ATR. Indeed, POT1 is not able to compete with RPA for ss telomeric DNA in vitro (26). It was therefore proposed that the repression of ATR requires the local tethering of the POT1 proteins to the rest of shelterin, thus increasing the local concentration of POT1. Consistent with this tethering model, repression of ATR requires the association of the POT1 proteins with TPP1 and the TPP1-and TIN2-mediated interaction with TRF1 and/or TRF2. ATR activation throughout the cell cycle is observed when telomeres lack either TPP1 or TIN2 (17,29), or when telomeres contain alleles of TPP1 or TIN2 that do not recruit POT1a/b (30).
In its simplest form, the RPA exclusion model predicts that any shelterin-tethered protein with the ability to bind ss TTAGGG repeats can repress ATR signaling at telomeres. However, this prediction is not met in the context of POT1b, which can repress ATR signaling in G 1 but not in S phase. POT1b and POT1a are equally abundant at telomeres, bind TPP1, and have indistinguishable DNA-binding features (17,19,29,31). Thus, the current knowledge of POT1b argues against the idea that any shelterin-tethered ss TTAGGG repeat-binding protein can exclude RPA from telomeres and thus repress ATR signaling.
There are alternative models for how POT1 represses ATR signaling. The ability of human POT1 to protect telomeric G4 DNA from unfolding by RPA has been proposed to repress ATR signaling (32). It is also possible that the greater affinity of POT1 for the telomeric 3Ј end (when ending in TTAG-3Ј) (13) protects telomeres by providing POT1 with a competitive advantage over RPA. In another proposal, an S/G 2 -specific interplay between hnRNPA1, an RNA-binding protein with high affinity for ss TTAGGG repeats, and the telomeric long noncoding RNA TERRA serves an intermediary function in removing RPA from telomeric DNA and then allowing POT1 to bind (26). Finally, it is possible that the POT1 proteins interfere with 9-1-1 loading. Although POT1a/b do not recognize the telomeric ds-ss junction in vitro (19), it is not excluded that the junction is bound by POT1 in the context of the whole shelterin complex.
Here we report that, consistent with the tethering model, ATR signaling can be partially repressed by the DNA-binding domain of RPA70 when it is tethered to shelterin. Furthermore, we solve the POT1b conundrum by showing that POT1b, like POT1a, can repress ATR signaling in S phase but that its interaction with CST prevents it from doing so. The results provide evidence for the idea that ATR repression involves RPA exclusion by a shelterin-tethered ssDNA-binding protein with no other specialized features.

Engineering a synthetic shelterin-tethered ATR repressor
We sought to test the idea that the exclusion of RPA by POT1 is based on the tethering of the ssDNA-binding domain of POT1 to the duplex telomeric DNA rather than POT1-specific DNA-binding features (e.g. binding to G4 DNA, binding to the 3Ј end, or binding to junctions) (Fig. 1A). We argued that the most stringent test of this tethering model would involve creating a synthetic telomeric ATR repressor based on the DNAbinding domain of RPA70. RPA has the same affinity for ss TTAGGG repeats as POT1 (29) but a different DNA-binding mode: Whereas POT1 binds DNA with two OB-folds, RPA uses three OB-folds (OB-A, -B, and -C) in RPA70 (Fig. 1B) and one OB-fold in RPA32 (33). The C-terminal OB-fold (OB-C) of RPA70 interacts with RPA32, which in turn binds to RPA14.
An obvious complication in creating an RPA-based ATR repressor is that tethering of WT RPA to telomeres would permanently recruit and likely activate the ATR kinase. The interaction of RPA70 with the ATRIP/ATR complex is mediated by its fourth (N-terminal) OB-fold (OB-F), which does not bind to DNA (34). We therefore based our construct on a version of RPA70 that lacks OB-fold OB-F (referred to as ⌬70) (Fig. 1B). This part of RPA70 was fused to the C terminus of POT1a (referred to as POT1aC), which is the part of POT1a that interacts with TPP1 and ensures association with shelterin. The resulting chimera is referred to as ⌬70-POT1aC.
To determine the DNA-binding features of the chimeric proteins, they were co-expressed with StrepII-tagged TPP1 and isolated on Strep-Tactin beads ( Fig. 1, D-G; Fig. S1). TPP1 promotes the binding of POT1 to telomeric DNA but does not bind DNA by itself. The concentration of the partially purified TPP1/chimera heterodimers was adjusted based on immunoblotting ( Fig. 1D) with an antibody to the C terminus of POT1a, and the complexes were analyzed in a gel-shift assay with labeled 34 or 35 nt probes ( Fig. 1, E-G). As expected from previous work (29), the RPA70-containing complexes bound to ss TTAGGG repeat as well as a scrambled sequence probe, whereas the POT1a/TPP1 heterodimer only bound the TTAGGG repeat probe (Fig. 1E). Titration experiments showed that ⌬70-POT1aC/TPP1 had the same affinity (apparent relative K d ϳ0.5 nM) for TTAGGG repeats as POT1a/TPP1 (Fig. 1, F and G). Consistent with the absence of the third RPA70 OB-fold, the ⌬⌬70-POT1aC/TPP1 heterodimer showed a lower affinity (apparent relative K d ϳ2 nM) for ss TTAGGG repeats.

Repression of ATR by ⌬70-POT1aC
To test the ability of the chimeric proteins to prevent the activation of ATR at telomeres ( Fig. 2A), they were expressed in

ATR repression by POT1a and POT1b
SV40-LT immortalized POT1a F/F MEFs from which POT1a can be deleted with the Cre recombinase (22) (Fig. 2B). As controls, the POT1aC fragment and the ⌬70 mutant of RPA70 were expressed individually. Expression vectors were chosen to yield moderate expression levels of the chimeric proteins, because an excess of the mutant forms of RPA70 (⌬70 and ⌬⌬70) might

ATR repression by POT1a and POT1b
interfere with the essential functions of RPA in DNA replication and ATR signaling. Immunoblotting with RPA70 and POT1a antibodies showed that although all exogenous proteins were overexpressed compared with the endogenous POT1a, their abundance was far less than the endogenous RPA70 ( Fig.  2B), making it unlikely that that the RPA variants would have a deleterious effect. Indeed, cells expressing the mutant RPA70 alleles showed no overt proliferation defect and had comparable S phase indices (Fig. S2, A and B). As expected, all exogenous proteins that carried the C terminus of POT1a localized to telomeres (Fig. 2, C and D).
Deletion of POT1a showed the expected induction of a DNA damage response at telomeres. In absence of POT1a, ϳ30% of the cells showed accumulation of ␥-H2AX and 53BP1 at telomeres (Fig. 2, E and F; Fig. S2, C and D). This percentage of cells showing a DNA damage response is consistent with prior reports that showed that POT1a is required for the repression in S/G 2 whereas in G 1 , POT1a is redundant with POT1b (11,22).
As expected, this S/G 2 ATR signaling was fully repressed by complementation of the cells with WT POT1a, but not by POT1aC (Fig. 2, E and F). Remarkably, ⌬70-POT1aC showed a significant reduction in ATR signaling at telomeres, resulting in only 15% of cells with detectable ␥-H2AX foci at their telomeres. As expected, the untethered ⌬70 protein was unable to repress ATR signaling at telomeres (Fig. 2, E and F). Cell cycle analysis indicated that the changes in ATR signaling were not because of changes in S phase index (Fig. S2, A and B) and quantitative analysis of the ss TTAGGG repeats at telomeres (Fig. S2E) showed that the chimeric protein did not reduce the telomeric overhang signal, which could have confounded the interpretation of the reduced ATR signaling. In fact, overexpression of both POT1a and ⌬70-POT1aC induced a slight increase in the telomeric overhang signal, presumably because they compete with POT1b for TPP1 interaction.
In contrast to ⌬70-POT1aC, ⌬⌬70-POT1aC was incapable of blocking ATR signaling at telomeres. We also tested the ability of Escherichia coli SSB to repress ATR signaling when tethered to shelterin (Fig. S3). The SSB-POT1aC chimera, which binds telomeric DNA with an apparent relative K d of ϳ4 nM (Fig. S3, A-D), localized to telomeres but did not repress the DNA damage response (Fig. S3, E-H). We do not know whether the higher relative K d of ⌬⌬70-POT1aC and SSB-POT1aC is the cause of the lack of protection by these chimeras. In principle, in the context of tethered chimeras, changes in the relative K d should not have a great impact. Possibly, the chimeras are defective in other aspects (e.g. exact positioning of the ssDNA-binding domains within the shelterin complex).

CST binding interferes with the repression of ATR by POT1b
It remains to be explained why POT1b is incapable of repressing ATR in S/G 2 although it can do so in G 1 . POT1b interacts with CST, allowing appropriate fill-in synthesis of the telomere terminus during S/G 2 (21). We therefore asked whether the interaction of POT1b with CST interferes with its ability to repress ATR. Prior work had shown that a POT1b allele with mutations in two sites of the protein (Fig. 3A) does not bind to CST and is deficient in regulating the postreplicative processing of the telomere terminus (21). Notably, expression of this mutant version of POT1b fully restored the ATR repression in POT1a-deficient cells. The WT POT1b, even when overexpressed, did not have this effect (Fig. 3, B-D).
To confirm the findings with POT1b⌬CST, we employed shRNA-mediated knockdown of Stn1. Consistent with the results with the POT1b⌬CST mutant, partial knockdown of Stn1 restored the ability of overexpressed POT1b to repress ATR signaling at telomeres. In contrast, knockdown of Stn1 did not restore the ability of the endogenous POT1b to protect telomeres from ATR activation (Fig. 3, E-G). This is most likely because of the residual Stn1 interacting with endogenous POT1b, whereas the residual level of Stn1 is insufficient to bind all overexpressed POT1b. As CST is essential for cell viability, we are unable to test whether complete removal of Stn1 will restore ATR repression by endogenous POT1b.

Discussion
Because mammalian telomeres carry a 3Ј protrusion of ssDNA, ATR signaling has to be repressed throughout the cell cycle. Given the abundance of telomere ends in the nucleus, a system that fails at as few as 1% of the telomeres could lead to induction of cell cycle arrest and senescence or apoptosis. Our data clarify how this stringent ATR repression is achieved by the POT1 subunits of the telomeric shelterin complex.
The shelterin-tethered chimera of RPA70 and the C terminus of POT1a (⌬70-POTaC) is remarkably proficient in repressing ATR signaling at telomeres, although not as effective as WT POT1a. The reduction in ATR signaling at telomeres containing ⌬70-POT1aC suggests that the ability of POT1a to repress ATR signaling is primarily determined by its ability to bind to ssDNA telomeric DNA. Because the DNA-binding domain of POT1a can be replaced with the OB-folds of RPA70, we consider it unlikely that any other feature of POT1-DNA interaction beyond its binding to ssDNA is involved in the control of ATR signaling. The preference of POT1 for a TTAG-3Ј telomere end appears to be largely irrelevant in this context. Indeed, TTAG-3Ј ends only make up 40% of the telomere ends

ATR repression by POT1a and POT1b
in human cells (35), yet all telomeres are shielded from the ATR pathway. Similarly, the ability of POT1 to protect G4 DNA from unfolding by RPA (32) is unlikely to be the primary determinant of ATR repression. Finally, although POT1 does not show a preference for its site next to the ds-ss junction in vitro (19), it could be argued that bound to shelterin, POT1 might have the

ATR repression by POT1a and POT1b
ability to recognize the junction and prevent 9-1-1 loading. This hypothetical attribute is also rendered unlikely to be relevant to ATR repression based on the results with the shelterin-tethered chimeric RPA70 protein.
The inability of POT1b to repress ATR signaling in S/G 2 has presented a conundrum because POT1a and POT1b have identical TPP1-and DNA-binding features. Furthermore, it can be inferred that POT1b resides at telomeres in S/G 2 because it governs telomere end processing during this cell cycle stage. Our data indicate that the interaction of POT1b with the CST and perhaps the polymerase ␣/primase complex prevents POT1b from acting as an ATR repressor. Only when the interaction of POT1b with CST is abrogated, is POT1b proficient in the repression of ATR in S/G 2 . This is an unexpected result because both POT1b and CST bind ss TTAGGG repeats. One possibility is that the recently noted ability of CST to unfold G4 structures (36) leads to a greater ability of RPA to gain access to TTAGGG repeat-binding sites.
The single POT1 subunit in human shelterin has been implicated in the repression of ATR signaling (9,20). Interestingly, the recruitment of CST to human telomeres is mediated by TPP1, not POT1 (37). Human POT1 does play a role in how CST functions at telomeres (38,39) and may therefore have a (transient) interaction with CST, but the recruitment of CST primarily involves binding to TPP1. This arrangement may avoid the complication of CST interfering with the ability of human POT1 protein to repress ATR signaling.  (21). For each retroviral construct, 20 g DNA was transfected into Phoenix packaging cells using CaPO 4 co-precipitation. Medium was changed 12 and 24 h after transfection, and the retroviral supernatant was used for four infections of the MEFs at 12-h intervals. Cells were subjected to selection in 135 g/ml hygromycin for 3 days. To delete endogenous POT1a from POT1a F/F MEFs, cells were subjected to three infections at 12-h intervals with pMMP Hit&Run Cre retrovirus derived from transfected Phoenix cells as described previously (40). Time point 0 was set at 12 h after the first Hit&Run Cre infection. For Stn1 knockdown, the Stn1 shRNA (5Ј-GATCCTGTGTTTCT-AGCCTTT-3Ј) (21) in a pLKO.1 lentiviral vector (Open Biosystems) was produced in Phoenix-ECO cells as described above and introduced with three infections (6-h intervals). Infected cells were selected for 2.5 days in 2 g/ml puromycin.

Electrophoretic mobility shift assay
Labeling reactions and gel-shift assays were performed as described previously (19) with minor modifications. All oligonucleotides were obtained from Sigma. The probe sequences were as follows: Telo34 (5Ј-TTAGGGTTAGGGTTAGGGT-TAGGGTTAGGGTTAG-3Ј) and Scr35 (5Ј-ATGCGACT-CGAGCTAGATGATGTCTTCTGCAATCA-3Ј). Gel-shift reactions were performed in 10 l reaction buffer (50 mM Hepes-KOH, pH 8.0, 50 mM LiCl, 0.1 mM DTT, 50 ng/l ␤-casein) with 0.2 nM polynucleotide kinase end-labeled DNA probes. Proteins were added last and the mixture was incubated for 10 min on ice. Electrophoresis was performed in 0.8% agarose gels run in 0.5ϫ Tris borate/EDTA, pH 8.3. Gels were run for 45 min at 100 V at room temperature, fixed in 20% methanol/10% acetic acid, dried on Whatman DE81 paper at 80°C with vacuum and exposed to phosphorimager screens. The binding fractions were calculated taking all protein-DNA complexes into account (binding fraction (%)) ϭ (proteins-DNA complex)/(free DNA probes) to derive apparent relative K d values.

IF, IF-FISH, and TIF analysis
Cells grown on polylysine-covered coverslips were harvested at time point 96 h after Cre to analyze protein localizations and TIF response. Coverslips were incubated on ice, rinsed once with PBS containing Mg 2ϩ and Ca 2ϩ , and soluble proteins were pre-extracted with ice cold Triton X-100 buffer (20 mM Hepes-KOH, pH 7.9, 50 mM NaCl, 3 mM MgCl 2 , 0.5% Triton X-100, 300 mM sucrose) for 30 s on ice. Coverslips were rinsed twice with cold PBS containing Mg 2ϩ and Ca 2ϩ , transferred to room temperature, and fixed in 3% paraformaldehyde with 2% sucrose for 10 min. Coverslips were washed twice in PBS at room temperature for 5 min. IF was carried out as described previously (40 -42) with the following primary antibodies: ␣-Myc (9B11, Cell Signaling Technology), ␣-␥H2AX (no. 05-636, Millipore), and ␣-53BP1 (ab175933, Abcam). For IF-FISH staining, after the secondary antibody incubation and wash step, cells were fixed again with 2% paraformaldehyde for 5 min; dehydrated in 70, 95, and 100% ethanol for 5 min each and allowed to air dry. Hybridizing solution (10 mM Tris-HCl, pH 7.2, 70% formamide, 1 mg/ml blocking reagent (Roche), containing 100 nM fluorescent PNA probe (TelC-Alexa488 (PNA Bio Inc.)) was added to each coverslip, and the DNA was denatured by heating for 10 min at 80°C on a heat block. After 2 h of incubation at room temperature in the dark, cells were washed twice with washing solution (70% formamide, 10 mM Tris-HCl, pH 7.2) for 15 min each and with PBS three times for 5 min each. DAPI (0.1 g/ml) was added to the second PBS wash. Coverslips were sealed onto glass slides with embedding medium (ProLong Gold Antifade Reagent, Invitrogen). Digital images were captured on a Zeiss Axioplan II microscope with a Hamamatsu C4742-95 camera using Volocity software.

S phase index
Before harvesting, cells seeded on coverslips were incubated for 1.5 h in medium containing 10 M BrdU and subsequently fixed with 2% PFA at room temperature for 10 min. Following an incubation in Nonidet P-40 buffer (0.5% Nonidet P-40 in PBS) at room temperature for 10 min, coverslips were washed three times in PBS for 5 min and then incubated in 4N HCl at room temperature for 10 min. Next, the coverslips were washed three times in PBS for 5 min and incubated with blocking solution (PBS, 0.1% BSA, 3% serum donkey, 0.1% Triton X-100, 1 mM EDTA, pH 8.0) at room temperature for 45 min before addition of ␣-BrdU antibody (B35130, Invitrogen). Cells were incubated with the antibody at 4°C for overnight. Coverslips were washed three times in PBS for 5 min and incubated with secondary antibody at room temperature for 45 min. Coverslips were washed three times in PBS for 5 min, with the second PBS wash containing DAPI (0.15 g/ml) to stain for DNA. Coverslips were sealed and processed as described for IF above.

Telomeric overhang analysis
Mouse telomeric DNA was analyzed on CHEF gels as described previously (21).