The Human Rap1 Protein Complex and Modulation of Telomere Length*

Proper maintenance of telomere length and structure is necessary for normal proliferation of mammalian cells. Mammalian telomere length is regulated by a number of proteins including human repressor activator protein (hRap1), a known association factor of TRF2. To further delineate hRap1 function and its associated proteins, we affinity-purified and identified the hRap1 protein complex through mass spectrometry analysis. In addition to TRF2, we found DNA repair proteins Rad50, Mre11, PARP1 (poly(ADP-ribose) polymerase), and Ku86/Ku70 to be in this telomeric complex. We demonstrated by deletional analysis that Rad-50/Mre-11 and Ku86 were recruited to hRap1 independent of TRF2. PARP1, however, most likely interacted with hRap1 through TRF2. Interestingly, knockdown of endogenous hRap1 expression by small hairpin interference RNA resulted in longer telomeres. In addition, overexpression of full-length and mutant hRap1 that lacked the BRCA1 C-terminal domain functioned as dominant negatives and extended telomeres. Deletion of a novel linker domain of hRap1 (residues 199–223), however, abolished the dominant negative effect of hRap1 overexpression. These results indicate that hRap1 negatively regulates telomere length in vivo and suggest that the linker region of hRap1 may modulate the recruitment of negative regulators of telomere length.

Telomere-binding proteins TRF1 and TRF2 play pivotal roles in telomere protection and maintenance in mammalian cells (1)(2)(3)(4)(5). Several proteins have been shown to associate with TRF1 and TRF2 (3,5,6). Recently, a novel telomere regulator human repressor activator protein (hRap1) 1 was identified as a protein that specifically interacts with TRF2 (7). hRap1 is the human homologue of yeast RAP1. In yeast, RAP1 is a negative regulator of telomere length as well as a regulator of transcription (for review, see Refs. 8 -13). The RAP1 mutants in Saccharomyces cerevisiae (scRAP1) and Schizosaccharomyces pombe (spRAP1) are defective in telomere length control and telomere position effects (14 -16). In human cells overexpression of hRap1 extends telomeres (7,17). It has yet to be fully determined, however, whether endogenous hRap1 is a negative or positive regulator of telomere length. scRAP1 contains two Myb domains and binds telomeric DNA (18,19). scRAP1 has been shown to recruit Rif1, Rif2, and Sir proteins to regulate telomere length, structure, and transcriptional silencing (11,20). Whether similar proteins, as found with yeast RAP1, are complexed with hRap1 remains unknown. In contrast, spRAP1 has only one Myb domain and is recruited to telomeres through Taz1, a yeast homologue of TRF2 (15). Similar to spRAP1, hRap1 also contains a single Myb domain without detectable DNA binding activity (7,21). In addition to the Myb domain, hRap1 has three putative protein-protein interaction domains; they are a BRCT domain, a coiled-coil domain, and a TRF2 interacting RCT domain (7). The function of the hRap1 BRCT and coiled-coil domains has yet to be described.
TRF2 has been shown to associate with the MRN complex (Rad-50, Mre-11, and NBS1) (22). In S. cerevisiae, the MRX complex (equivalent of MRN in mammals) has exonuclease activity, is a positive regulator of telomeres, and is responsible for telomere end processing (23)(24)(25). Although both scRAP1 and the MRX complex localize to the telomere, the MRX complex does so independent of scRap1 (24,25). It is possible that hRap1 may interact with the mammalian MRN components, and this association may be necessary for telomere length control. Alternatively, hRap1 and TRF2 may recruit proteins other than the MRN complex to modulate telomere length.
To address these questions, we undertook biochemical approaches to isolate and study protein complexes that are involved in telomere maintenance. We found several DNA repair proteins in addition to TRF2 to be in the hRap1 complex through affinity purification experiments. Furthermore, a more detailed analysis of hRap1 via RNA interference (26,27) and deletion experiments suggest that hRap1 negatively modulates telomere length in vivo and that multiple proteins may be required for hRap1 function.

Generation of Constructs and Cell Lines of hRap1 and Its Deletion
Mutants-Full-length hRap1 was amplified from a HeLa cDNA library and cloned into the pBabe-puro retroviral vector. hRap1 truncations were generated by PCR using Pfu Turbo DNA polymerase (Stratagene) from full-length hRap1 and cloned into the pBabe retroviral vector. Constructs were named hRap1 (full-length, amino acids 1-399), ⌬Myb (amino acids 1-132 and 192-399), ⌬Link (amino acids 1-198 and 224 -399), ⌬RCT (amino acids 1-290, with an N-terminal nuclear localization signal PRRK), ⌬R⌬C (⌬RCT/⌬Coil-Coil, amino acids 1-227, with an N-terminal nuclear localization signal PRRK), and ⌬BRCT (amino acids 129 -399). Full-length and mutant hRap1 are tagged with FLAG epitopes at both the N and C termini. The retroviral vectors were used to transfect BOSC23 cells by the calcium phosphate method to produce retroviruses for subsequent infection of HeLa S3, HT1080, or HTC75 cells (28). To generate cells stably expressing hRap1 and its mutants, human HeLa S3 and HT1080 cells were infected with the hRap1 retroviruses. These cells were selected with 2 g/ml puromycin after infection for 3 days to obtain cells stably expressing hRap1 and its mutants.
Preparation of Nuclear Extracts-HeLa S3 cells stably expressing FLAG-tagged hRap1 or mutants were grown in suspension up to 1 ϫ 10 6 cells/ml. A total of 1-6 ϫ 10 9 cells were collected for nuclear extract preparation. Briefly, cells were washed in cold phosphate-buffered saline and hypotonic buffer (10 mM Tris, pH 7.3, 10 mM KCl, 1.5 mM MgCl 2 , 0.2 mM phenylmethylsulfonyl fluoride and 10 mM 2-mercaptoethanol), allowed to swell for 15 min in hypotonic buffer, and homogenized until cell membrane lysis was ϳ80% complete. Cells were resuspended in low salt buffer (20 mM KCl, 20 mM Tris, pH 7.3, 25% glycerol, 1.5 mM MgCl 2 , 0.2 mM EDTA) and homogenized briefly to break the nuclear membrane. An equal volume of high salt buffer (1.2 M KCl, 20 mM Tris, pH 7.3, 25% glycerol, 1.5 mM MgCl 2 , 0.2 mM EDTA) was added followed by agitation for 30 min at 4°C, and the samples were spun down at 20,000 ϫ g for 30 min. The supernatant was dialyzed in BC0 buffer (20 mM Tris, pH 7.3, 20% glycerol, 0.2 mM EDTA, 0.2 mM phenylmethylsulfonyl fluoride, and 10 mM 2-mercaptoethanol) for 3 h and centrifuged again. The cleared supernatant was then separated into aliquots, quickly frozen, and stored at Ϫ80°C.
Immunoprecipitation and Mass Spectrometry-For large scale affinity purification, ϳ70 mg of nuclear protein extract was thawed gently and centrifuged at 100,000 ϫ g at 4°C to spin down denatured protein.
The supernatant was immunoprecipitated with 100 l of M2 anti-FLAG-agarose beads (3.3 mg/ml, Sigma) for 3 h at 4°C. The beads were then washed 4 times with NETN (20 mM Tris, pH 8.0, 100 mM NaCl, 0.5% Nonidet P-40, and 1 mM EDTA), and the bound protein was eluted twice with 100 l of 200 g/ml FLAG peptide (DYKDDDDK) (Sigma) in NETN. The eluant and bead fractions were boiled in SDS loading buffer, separated on a precast 8 -12% SDS-PAGE gradient gel (Bio-Rad), and visualized by Coomassie Blue staining. Bands were excised, digested in trypsin and subjected to ion trap mass spectrometry as previously described (29). Peptides were identified using PROWL (prowl.rockefeller.edu).
For small scale immunoprecipitation experiments, 1-3 mg of nuclear extracts were incubated for 2 h at 4°C with 5 l of M2 anti-FLAGagarose beads or 10 g of anti-hRap1 antibody (Bethyl Laboratories) and 15 l of protein A/G agarose beads. Ethidium bromide was added to a final concentration of 100 g/ml. The beads were then washed four times with 0.5 ml NETN, boiled in SDS loading buffer, and separated on 8 or 10% SDS-PAGE gels.
Quantification of Telomere Length-HT1080 and HTC75 cells stably expressing hRap1 or shRNA constructs were passaged and harvested at different population doubling time points for DNA extraction (DNeasy Tissue Kit, Qiagen). Telomere restriction fragment length analysis was performed as described (2, 30) using restriction enzymes HinF1 and RSA1 (New England Biolabs) and the 32 P-labeled oligo probe (TTAGGG) 3 . Quantification of telomere length was performed using ImageQuant software (Molecular Dynamics) and TELORUN (31).
Immunofluorescence-The localization of telomere-associated proteins was visualized through indirect immunofluorescence as previously described (7). Cells were grown overnight on coverslips, permeabilized in the Triton X-100 solution (0.5% Triton X-100 in phosphate-buffered saline), fixed with 3.7% paraformaldehyde in phosphate-buffered saline, and permeabilized again in the Triton X-100 solution containing 300 mM sucrose. The cells were subsequently blocked for 1 h at 37°C in 5% goat serum, stained with various primary and fluorescence-conjugated secondary antibodies for 1 h each at 37°C, and then visualized under a Nikon TE200 fluorescence microscope. Primary antibodies used are polyclonal anti-hRap1 antibody (Bethyl Laboratories), polyclonal anti-FLAG antibody (Sigma), and monoclonal anti-TRF2 antibody (Oncogene Science). Secondary antibodies are Alexfluo 488-conjugated goat anti-mouse antibody (Molecular Probes) and Texas Red goat anti-rabbit antibody (Rockland).

Identification of the hRap1 Protein
Complex-To identify the hRap1 complex, an affinity purification method was utilized (see "Material and Methods"). HeLa S3 cells stably expressing FLAG-tagged hRap1 were first generated. Similar to endogenous hRap1, FLAG-hRap1 remained co-localized with endogenous TRF2 in these cells (see Fig. 5B), suggesting that FLAGtagged hRap1 was targeted to the telomeres in the same fashion as endogenous hRap1 (7). Nuclear proteins were then extracted from these cells and immunoprecipitated with an anti-FLAG antibody that was conjugated to agarose beads. The precipitated proteins were eluted with FLAG peptides and separated by SDS-PAGE (Fig. 1). Distinct bands (compared with uneluted fractions) were subsequently excised, trypsindigested, and sequenced by mass spectrometry.
We found hRap1 to co-purify with TRF2 (band 6) ( Fig. 1 and Table I), in agreement with previous findings that hRap1 associates with TRF2 (7). In addition, Rad50 (band 1) and Mre11 (band 4) were detected. Both Rad50 and Mre11 are subunits of the MRN complex, which has been shown to interact with TRF2 (22). NBS1, the other subunit of the MRN complex, was not detected in our large scale immunoprecipitation experi- ments of hRap1. NBS1 has been shown, however, to associate with TRF2 only in S-phase (22).
Interestingly, a number of proteins involved in DNA damage repair were also found to co-purify with hRap1. The p110 band (band 2) turned out to be poly(ADP-ribose) polymerase (PARP1) (32). Two other PARPs, namely tankyrase 1 and 2, interact with TRF1 and are known regulators of telomere length in mammalian cells (10,(33)(34)(35)(36). Two additional DNA damage repair proteins found in the hRap1 complex are Ku86 and Ku70 (bands 3 and 5). These two proteins heterodimerize and are required for double-strand DNA break repair (37). Previous studies have found that Ku70 and Ku86 associate with telomeres (38 -40). Our data suggest that the Ku70/86 heterodimer may be alternatively recruited to the telomeres through its interaction with the hRap1 complex.
Endogenous hRap1 Associates with Ku86 and Rad50 -We next examined the interaction between endogenous hRap1 and the proteins identified in the hRap1 complex through immunoprecipitation experiments with anti-hRap1 and anti-TRF2 antibodies. Ethidium bromide was added in the reactions to rule out the possibility that some proteins might associate with hRap1 through double-stranded DNA. As shown in Fig. 2, immunoprecipitation of endogenous hRap1 from HeLa cell nuclear extracts was able to specifically pull down TRF2, Ku86, Rad50, and hRap1. Similarly, hRap1, Ku86, and Rad50 were also detected in the immunoprecipitates of endogenous TRF2. This result suggests that hRap1 as well as TRF2 interacts with Ku86 and the MRN complex in vivo.
Structural Organization of the hRap1 Protein Complex-To delineate how different regions of the hRap1 protein may interact with other proteins in the hRap1 complex, FLAG-tagged hRap1 deletion mutants lacking each of the four predicted domains or the linker region between the Myb and coil-coil domains were constructed (Fig. 3A). HeLa S3 cells that stably expressed these mutants were then generated. The expression levels of hRap1 mutants were about 1-5-fold of the endogenous protein (Fig. 3D). The different hRap1 mutant proteins were subsequently immunoprecipitated using the anti-FLAG antibody, and their associations with TRF2 and Rad50 were analyzed. As shown in Fig. 3B, hRap1 mutants lacking the RCT domain (⌬RCT) or both the coiled-coil and RCT domains (⌬R⌬C) failed to immunoprecipitate TRF2. These results suggest that the hRap1 RCT domain mediates hRap1 association with TRF2 in vivo. This is consistent with previous findings that the hRap1 RCT domain could bind TRF2 through yeasttwo-hybrid studies (7).
We showed above that members of the MRN complex can associate with hRap1 in vivo. This interaction may occur through TRF2, as previously hypothesized (6). If this is the case, the hRap1 ⌬RCT mutants that have lost the TRF2 binding abilities should fail to bind the MRN complex. To our surprise, both the ⌬RCT and ⌬R⌬C hRap1 mutants still retained their binding activity with Rad50 and Mre11 (Fig. 3B). In fact, none of the mutations tested ablated or diminished the association of hRap1 with members of the MRN complex. Interestingly, Ku86 association was not abolished among the Rap1 mutants (Fig. 3B), albeit a reduction of Ku86 association was detected for two of the hRap1 mutants (⌬Myb and ⌬BRCT). Because the interactions between hRap1 and Rad50/ Mre11 or Ku86 were maintained even in mutants that failed to bind TRF2, it is therefore likely that hRap1 can recruit Rad50, Mre11, and Ku86 independent of TRF2. It also suggests that multiple domains of hRap1 may be involved in binding of these proteins.
We next examined the interaction of PARP1 with the various hRap1 mutants. In contrast to the proteins examined above, PARP1 no longer co-purified with the hRap1 ⌬RCT mutant (Fig. 3C). This was not the case, however, for the hRap1 ⌬Myb, ⌬Link (Fig. 3C), and ⌬BRCT mutants (data not shown). Therefore, PARP1 likely binds hRap1 either directly to the RCT domain or indirectly through TRF2.
hRap1 Negatively Regulates Telomere Length-Overexpression of full-length hRap1 was shown to extend telomeres in HT1080-derived HTC75 cells (7,17). Whether endogenous hRap1 is a positive or negative regulator of telomere length, however, remains unknown. In addition, if hRap1 does regulate telomere length in vivo, which protein(s) may be necessary for such activity remains to be determined. To address the role of hRap1 in telomere length control, we first generated HTC75 cells in which hRap1 levels were stably reduced using shRNA (26,27,41). Western blot analyses indicated that endogenous hRap1 levels decreased significantly (50.0 Ϯ 10.0%) in hRap1 shRNA cells (Fig. 4A). This inhibition of hRap1 expression was sustained for weeks, allowing for relatively long term studies of telomere length in these cells. We then compared the telomere length of these hRap1 knockdown cells to that of control cells at different cell passages. As shown in Fig. 4B, knockdown of hRap1 resulted in detectable extension of telomeres (ϳ500 bp) as early as population doubling 10. This finding not only im-  2. Endogenous hRap1 co-immunoprecipitated with proteins identified by mass spectrometry. Endogenous hRap1 or TRF2 was immunoprecipitated using HeLa S3 cell nuclear extracts with a polyclonal antibody against hRap1 (Rap1 IP), a monoclonal antibody against TRF2 (TRF2 IP), or an anti-FLAG antibody (Ctrl IP). Immunoprecipitations were carried out in the presence (ϩ) or absence (Ϫ) of 100 g/ml ethidium bromide (EB). The precipitates were separated by SDS-PAGE, blotted, and probed with antibodies against hRap1, TRF2, Ku86, and Rad50.
plicates hRap1 in negatively regulating telomere length but also suggests that this activity is evolutionarily conserved from yeast to human.
Telomeric Localization of hRap1 Mutants and Their Modulation of Telomere Length-We have thus far shown that hRap1 can bind TRF2, PARP1, Ku86, and the MRN complex. To determine which of these proteins may be responsible for hRap1 activity, we tested how the expression of hRap1 and various hRap1 mutant constructs (Fig. 3A) might influence hRap1 localization and telomere length. Both full-length and mutant hRap1 proteins were well expressed in the HT1080 cells. The subcellular localization of the various Rap1 proteins was then determined. Consistent with previous findings (7), endogenous hRap1 (red) exhibited a punctate pattern and costained with endogenous TRF2 (green) (Fig. 5A). Similar results were obtained in cells stably expressing the FLAG-tagged ⌬BRCT mutant (Fig. 5C), indicating that the hRap1 ⌬BRCT mutant can co-localize with endogenous TRF2. However, although the ⌬RCT mutant also exhibited a punctate pattern, it did not co-stain with TRF2 (Fig. 5D). These observations support the model that the RCT domain is critical for hRap1 and TRF2 interaction and telomere association. Genomic DNA was extracted from these stable cells at different passages, and their telomere length was compared using the telomere restriction fragment assay.
Because the mutants differed in their localization and ability to bind TRF2, we expected to see differences in their effects on telomere length. Indeed, results from independent experiments indicated that the mutants extended telomere length to various degrees over generations ( Fig. 6 and data not shown). Notably, the ⌬BRCT mutant cells had consistently longer telomeres at late passages than other mutants. ⌬BRCT telomere length was significantly longer than control (at PD70, p ϭ 0.049). Fulllength hRap1 overexpression also extended telomeres significantly (at PD70, p ϭ 0.012). These results are in agreement with a recent report from Dr. de Lange's group (17). Because hRap1 RNA interference resulted in longer telomeres (Fig. 4), the hRap1 deletion mutants might function as dominant negatives to extend telomere length. Interestingly, the RCT deletion mutant (⌬RCT) had a much weaker telomere extension phenotype compared with full-length hRap1 (at PD70, 5.7 versus 7.2 kilobases, p ϭ 0.004). Therefore, it suggests that the RCT domain may be necessary for the efficient inhibition of endogenous hRap1 function by overexpressed hRap1 mutants. Taken together with our immunoprecipitation results (Figs. 1-3), these data point to a model in which overexpression of hRap1 mutants might block proper telomeric access of signaling components that negatively regulate telomere length. In addition, the RCT-TRF2 interaction may be important for telomere length regulation of hRap1.
The Linker Region May Be Required for the Dominant Negative Effect of hRap1-The hRap1 mutants tested thus far all resulted in dominant negative phenotypes. Upon close examination, we found a small linker region situated between the Myb and coiled-coil domains that had been retained in all the mutants (Fig. 3A). We reasoned that this linker region might participate in recruiting the putative negative regulator(s). To test this possibility, we generated HT1080 cells expressing the hRap1 mutant with the linker region deleted (⌬Link) (Fig. 3A) and found this mutant to co-localize with TRF2 (Fig. 5E). In addition, binding of Rad50, Ku86, and TRF2 was not disrupted by the linker deletion (Fig. 3B). In contrast to full-length hRap1, however, expression of this mutant in HT1080 cells no longer lengthened telomeres (Fig. 6). This intriguing finding suggested that the linker domain of hRap1 might be necessary for hRap1 to negatively regulate telomere length. Because the ⌬Link mutant still has an intact RCT domain and binds TRF2, its failure in telomere extension is likely caused by a mechanism other than RCT-TRF2 interaction. In an effort to understand the mechanism of the ⌬Link affect on telomere length, we counted ⌬Link and TRF2 foci in our immunofluoresence experiments. ⌬Link co-localization with TRF2 was similar to that of full-length hRap1 (42 versus 41%). The slightly low level of total co-localization is likely due to and normally observed in FIG. 3. Deletional analysis of the interaction of the hRap1 complex. A, a schematic representation of various hRap1 deletion mutant constructs. Each construct was FLAG-tagged (see "Materials and Methods" for details). L stands for the linker region. B, nuclear extracts of HeLa S3 cells stably expressing full-length and mutant hRap1 were collected. The nuclear extracts were then immunoprecipitated with an anti-FLAG antibody followed by SDS-PAGE and Western blotting with anti-Rad50, anti-Ku86, anti-TRF2, and anti-FLAG antibodies. C, similarly, FLAG-tagged hRap1 mutants were immunoprecipitated with an anti-FLAG antibody and blotted with anti-PARP1 and anti-FLAG antibodies. D, Western blot of endogenous and hRap1 mutant constructs. The blot was probed with a polyclonal antibody against hRap1, which recognizes both endogenous hRap1 and FLAG-tagged constructs (in C). exogenously expressed protein. We did, however, detect a significant difference in the total number of observable TRF2 foci in ⌬Link cells and full-length hRap1 cells (17.0 versus 22.8 foci per cell focus plane, p Ͻ 0.0001). This may reflect shorter telomere length in the ⌬Link mutant cells. Alternatively, the ⌬Link mutant may alter TRF2 telomere localization. DISCUSSION We have shown that hRap1 interacts with multiple proteins that are involved in DNA damage response pathways, including Rad50, Mre11, Ku70/86, and PARP1. Both the MRN and Ku70/86 complexes are known to regulate double-stranded DNA break repair and homologous recombination (42)(43)(44), whereas PARP1 modulates chromatin structure in response to stress (45). Although the exact role of these DNA-damageresponse proteins at mammalian telomeres remains to be determined, genetic experiments in yeast and Arabidopsis thaliana have linked them to telomere length control (46 -51). Recent studies also indicate that TRF2 is necessary to maintain telomere state and genome stability (4,52). Thus, hRap1 together with TRF2 may function to recruit the DNA damage response proteins for telomere maintenance.
Our data suggest that hRap1 can associate with MRN and Ku70/86 independent of TRF2. It is important to note that none of our mutants abolished MRN or Ku70/86 interaction with hRap1. Therefore, the MRN and Ku complexes may interact with multiple domains of hRap1 (e.g. with both the RCT domain and the BRCT domain or the coiled-coil domain and the BRCT domain).
Our results also support the hypothesis that hRap1 has an evolutionarily conserved role in telomere length regulation; specifically, hRap1 is a negative regulator of telomere length because inhibition of hRap1 by RNA interference resulted in longer telomeres. These findings are in agreement with studies in yeast where null mutations of Rap1 and Taz1 (a TRF2 ortholog) resulted in aberrant telomere elongation (8,15,16).
We were particularly intrigued by the fact that full-length hRap1, and several of the hRap1 mutants we tested lengthened telomeres. It is likely that exogenously expressed hRap1 acts as a dominant negative. Expression of these mutant proteins may FIG. 4. hRap1 negatively regulates telomere length. A, endogenous hRap1 in HTC75 cells was knocked down with a retroviral vector encoding a shRNA against hRap1. A polyclonal anti-hRap1 antibody was used to Western blot for hRap1 expression. Densitometric analysis was performed to determine the difference in hRap1 expression (average knockdown was 50 Ϯ 10% (S.E.)). An anti-GRB2 Western blot was used as loading control and for normalization. B, telomere restriction fragment length analysis of hRap1 knockdown cells. Error bars indicate the range of two different experiments. Mock and hRap1 shRNA cells were collected over generations. Non-telomeric genomic DNA was digested with HinF1 and RSA1, and the remaining telomeric DNA was separated on a 0.6% agarose gel and transferred to Nylon membranes. Southern blotting was performed, and the average size of telomeres was determined using ImageQuant software and TELORUN. kb, kilobases. PD, population doublings.
FIG. 6. hRap1 mutants differentially affect telomere length. Mock HT1080 cells and those expressing fulllength hRap1, ⌬BRCT, and ⌬Link mutants were passaged over 70 generations, and their genomic DNA was collected for telomere restriction fragment length analysis to determine telomere length (see "Materials and Methods"). Error bars indicate S.E. from three different experiments. PD, population doublings. prevent the association of protein(s) that normally controls telomere length with endogenous hRap1. TRF2 may play an important role in this process; for example, heterodimerization of TRF2 and hRap1 may be required for normal hRap1 function. The inability of the ⌬RCT hRap1 mutants to greatly extend telomere length is likely due to the fact that they do not bind TRF2 and, therefore, titrate fewer factors away from the telomere. The hRap1 mutants that contain the RCT domain may be in such excess as to be competing for endogenous TRF2 binding. Our data that the RCT deletion mutant were not as potent as RCT-containing hRap1 mutants in telomere elongation supported this notion. Such a dominant negative titration model of hRap1 overexpression would suggest that the negative regulators of telomere length are segregated away from the telomeres.
Deletion analyses suggested that the linker region of hRap1 protein may modulate recruitment of putative regulators of telomere length, given that the ⌬Link mutant lacking this region was incapable of extending telomeres. There is the possibility that deletion of the linker domain might have affected the folding of hRap1. However, the ⌬Link protein expressed well, maintained its association with components of the hRap1 complex, and was targeted to the telomeres. The fact that the ⌬Link mutant seems to reduce the number of TRF2 foci in cells suggests that, while it is able to interact with TRF2 (Fig. 3B), it may disrupt TRF2 interaction with the telomere, leading to reduction in average telomere length. It is also possible that the linker domain may be involved in or regulate the recruitment of unidentified negative regulators of telomere length.
Interestingly, deletion of the BRCT domain had a greater impact on telomere elongation than other mutants (17). BRCT domains are found in many of the proteins involved in DNA damage response pathways (53). Recently, we demonstrated that BRCT domains mediate phosphorylation-dependent protein-protein interactions (54). It is, therefore, also possible that hRap1 may recruit a regulator of telomere length through its BRCT domain.
What are the negative regulators recruited by hRap1? We showed here that the MRN and Ku70/86 complex associated with hRap1 directly or indirectly, suggesting that the MRN and Ku70/86 complex may be part of the hRap1 binding factor(s) that modulates telomere length in human cells. In further support of this model, the Ku70/86 complex has been shown to be involved in double-stranded DNA break repair and telomere length maintenance (47)(48)(49)(50)(51)55). In yeast and A. thaliana, Ku mutants extended telomeres (48 -50). Furthermore, the MRN complex can regulate telomere length in a Ku-dependent manner in S. cerevisiae (47). Ku70 and TRF2 have been reported to directly interact with each other through yeast two hybrid and in vitro binding experiments (39). Ku has also been proposed to be a direct regulator of human telomerase (56). hRap1/TRF2 may, therefore, indirectly regulate telomerase by either recruitment or regulation of Ku. A reduction of Ku86 association was detected for the ⌬Myb and ⌬BRCT Rap1 mutants, concurrent with the telomere-lengthening phenotype. However, we found that wild type hRap1 but not the ⌬Link mutant extended telomeres yet still bound both Rad50 and Ku86. This implies that the Ku70/86 complex may not be the sole negative factors recruited by hRap1. It is possible that the linker domain may bind an unknown factor, which in turn regulates the activity of telomeric Ku70/86 complexes.