Assembly of Functional ALT-associated Promyelocytic Leukemia Bodies Requires Nijmegen Breakage Syndrome 1¹

Telomeres, specialized structures at the ends of linear chromosomes, are comprised of GC-rich tandem repeats. The protection of telomere ends and maintenance of their length prevent chromosome degradation, end-to-end fusions, and rearrangements, events detrimental to chromosome stability and genome integrity. Yeast, germ-line cells, and the majority of cancer cells typically maintain telomere length by de novo addition of DNA by telomerase (1, 2). In addition to the standard mechanism, subsets of cells maintain telomere length through a telomerase-independent mechanism, termed the ALT³ pathway (3).

Genetic studies in yeast have implicated a recombination-dependent mechanism in the ALT pathway. In this model, members of RAD52 epistasis group, including the DNA recombination and DSB repair factors Rad52, Rad51, and the Rad50/Mre11/Xrs2 complex participate, along with a helicase, Sgs1, and mismatch repair factors (4, 5, 6, 7, 8, 9, 10, 11). Two distinct ALT pathways have been proposed based on participation of either Rad50 or Rad51. Rad52, on the other hand, is required for both pathways (5, 7).

In humans, ∼10–20% of tumors are telomerase negative and have been proposed to maintain telomere length by ALT pathways (3). Selection of the exact mechanism of ALT in the cells of these tumors, although not well understood, may be related to the germ layer of their origin (12). Evidence hints that homologous recombination plays a primary role in most of the mammalian ALT pathways, similar to the nonreciprocal recombination used by yeast that are telomerase defective (5, 7). Marker cassettes integrated into specific telomeres of mammalian cells can be copied to other short telomeres, presumably through a gene conversion-like recombination process (13). In mammalian ALT cells, recombination factors, including Rad51, Rad52, and the Rad50/Mre11/NBS1 complex (the human counterpart of yeast Rad50/Mre11/Xrs2), are specifically associated with a fraction of nuclear PML bodies. Also present are telomere binding factors TRF1, TRF2, replication factor RPA and telomere repetitive DNA (14, 15, 16, 17). BrdUrd incorporation associated with the APBs in late S-G₂ phase indicates that the DNA synthesis associated with telomere lengthening may be an important function in these subnuclear compartments (16).

DNA recombination and DNA DSB repair involve a number of factors, defects in which are implicated in mammalian carcinogenesis. The product of the BRCA1 has essential roles in cellular responses to DNA damage, in checkpoint control, and in DSB repair (18, 19, 20, 21, 22, 23, 24, 25, 26, 27). Consistent with these functions, BRCA1 associates with both the Rad50/Mre11/NBS1 complex and Rad51 in vivo (18, 20). In yeast, the Rad50/Mre11/Xrs2 complex participates in several processes, including homologous recombination and nonhomologous end-joining, S-phase replication, DNA damage response, and cell cycle checkpoint control (review in Ref. 28). Moreover, the Rad50/Mre11/Xrs2 complex is proposed to have a role in telomerase-dependent maintenance (5). Whether the Rad50/Mre11/NBS1 complex has a similar role in mammalian cells is unknown. Intriguingly, however, NBS1 physically interacts with TRF1, a telomere-specific binding protein, supporting a potential role for NBS1 complex in mammalian telomere homeostasis (16). Although the association of BRCA1 with Rad50/Mre11/NBS1 complex is critical for DNA damage response, it remains unclear whether BRCA1 plays a role in telomere length maintenance in mammalian cells. Furthermore, how the multiple repair/recombination factors with distinct function assembled in APBs during telomere elongation is also unknown.

Here, we present evidence that BRCA1 and hRap1 are new components of APBs. The association of repair proteins with IRIF is distinct from their association with APBs. Using overexpressed dominant-negative proteins, we show that NBS1 is required for recruitment and/or assembly of a subgroup of proteins into APBs, including Mre11, Rad50, and BRCA1 but not Rad51 or TRF1. The BRCT domain of NBS1 is required for this function. Although TRF1 interacts with NBS1, it is not critical for localization of either NBS1 or Rad51 within APBs. Finally, colocalization of Mre11, Rad50, and BRCA1 at APBs is important for BrdUrd incorporation at these subnuclear sites. These results suggest that NBS1 plays a critical role in mediating the assembly of functional APBs that is important for the propagation of human ALT cells.

ALT cell lines, including SV40 large T-antigen-immortalized human fibroblasts (GM847 and VA13) and two osteosarcoma lines (Saos-2 and U-2 OS), have previously been described (16). Cells were cultured in DMEM supplemented with 10% fetal bovine serum in 10% CO₂. Telomerase-positive cells T24 and MCF7 were cultured similarly. Cell synchronization at late S-G₂ was achieved by a double thymidine block-and-release protocol as described previously (16). For BrdUrd pulse labeling, cells were incubated 30 min with reagents in a cell proliferation kit (Amersham Pharmacia Biotech, Inc., Piscataway, NJ), fixed, and stained as described below.

Indirect immunofluorescence staining was performed essentially as described previously (16). Briefly, cells plated on cover slips in 6-cm culture dishes were washed with PBS and fixed for 20 min in 3.7% formaldehyde in PBS with 0.1% Triton X-100. Cells were then permeated with 0.05% Saponin at room temperature for 30 min, followed by washing five times in PBS. After blocking with 10% goat serum in PBS/0.5% NP40 at room temperature for 30 min, cells were incubated with primary antibodies for 2 h at room temperature. FITC or Texas Red-conjugated secondary antibodies (Southern Biotechnology Associates, Inc., Birmingham, AL) were used to detect primary antibodies. Cells were counter stained with 4′, 6′-diamino-2-phenylindole and mounted in Permafluor solution (Shandon Lipshaw, Pittsburgh, PA). Fluorescence microscopy was performed with a Nikon Eclipse TE300, and the captured images were processed with Metamorph software and Adobe Photoshop. Immunoprecipitation was performed in lysis 250 buffer [50 mm Tris-Cl (pH 7.4), 250 mm NaCl, 5 mm EDTA, 50 mm NaF, 0.1% NP40, and 0.5 mm phenylmethylsulfonyl fluoride] at 4°C as described previously (18). Coprecipitates were resolved by 7.5% SDS-PAGE and immunoblotted with Rad50, Mre11, and/or NBS1 antibodies (29).

For TRF1 and hRap1, mouse polyclonal sera were raised against GST-TRF1 and GST-hRap1, respectively, which had been expressed and purified from bacteria. Mouse anti-Mre11(12D7), mouse anti-Rad50 (13B3), and mouse anti-Rad51 (14B4) monoclonal antibodies have been described previously (16, 18). Rabbit anti-Rad51 and mouse anti-BRCA1 monoclonal (Ab-1) antibodies were purchased from Oncogene Research Products (Boston, MA). Rabbit anti-NBS1 antibody was obtained from Novus Biologicals (Littleton, CO). Mouse anti-PML antibody was from Santa Cruz Biotechnology (Santa Cruz, CA). Rabbit anti-Sp100 antibody was purchased from MBL International Corporate (Watertown, MA).

To generate cell clones that express a GFP-NBS1C fusion protein, we used a tetracycline-inducible expression system (30). NBS1C (amino acids 543–754 of NBS1) was translationally fused to GFP in the pUHD10 vector. Constructs pUHD10-NBS1C and pCHTV (bearing tet regulator and hygromycin-resistant gene, respectively) were cotransfected (10:1 molar ratio) into human osteosarcoma U-2 OS cells, which were subsequently selected with hygromycin (150 μg/ml) in the presence of tetracycline (1 μg/ml). Hygromycin-resistant colonies were isolated and assayed for conditional expression of GFP-NBS1C. One of the positive clones, named UNBS1C, with tight tet-dependent regulation, was used in this study.

Constructs directing the expression of NBS1C, GFP, or GFP-TRF1Δmyb were prepared according to a simplified adenovirus amplification system with minor modifications (31). To facilitate virus amplification and evaluation of the titer, TRF1Δmyb and NBS1C were both tagged with GFP. Adenoviruses were produced and amplified in E1-immortalized human embryonic kidney cells 293. ALT cells were infected and harvested at 36 h after infection. To perform double staining (such as Mre11 with PML), cells were treated with 1% H₂O₂ after fixation until the GFP signal was fully quenched and then stained as described above.

BRCA1 plays an essential role in the maintenance of genome stability, which is partially attributed to its interaction with the DNA damage response and repair complex NBS1/Rad50/Mre11 (18). In ALT cells, the NBS1/Rad50/Mre11 complex is localized with APBs during late S-G₂ phase, suggesting its involvement in telomere length maintenance (16, 17). We first asked if BRCA1 is involved in the ALT pathway. Indirect immunofluorescence studies showed that BRCA1 colocalized with APB-associated proteins, TRF1, Sp100, and NBS1 in several ALT cells, including GM847 (Fig. 1,A). Although BRCA1 forms speckles in S-phase cells, they are not APB associated (32). However, when ALT cells are enriched at late S-G₂ phase, after release from a double-thymidine block, BRCA1 associates with APB foci in ∼24% of cells, compared with 5–7% in unsynchronized culture (Fig. 1,B). These data suggest that the association of BRCA1 with APBs is specific for late S-G₂ stage, similar to that of NBS1 and Rad51 (Fig. 1,B and Ref. 16). In contrast, colocalization was not observed in telomerase-positive cells such as human bladder carcinoma cells, T24, and human breast cancer cells, MCF7 (Fig. 1 C), suggesting that the association of BRCA1 with APBs is specific to late S-G₂ in ALT cells. These data implicating BRCA1 in the ALT pathway are consistent with its roles in homology-directed DNA repair and microhomology-mediated DNA end-joining (27, 33).

BRCA1 is known to colocalize with NBS1/Mre11/Rad50 complex and Rad51 in response to treatment of cells with DNA damaging agents (18, 20). The NBS1/Mre11/Rad50 complex and Rad51 participate in distinct DNA repair events (34, 35). However, both colocalize with APBs in ALT cells (14, 15, 16). To demonstrate these functionally distinct recombination factors and BRCA1 are localized in APBs, we performed pair-wise coimmunostainings in either randomly growing or synchronized ALT cells at late S-G₂ (Fig. 1,D). We found that Rad51 colocalized with BRCA1 (Fig. 1 D, a–d), NBS1 (e–h), and Mre11 (i–l) reciprocally, suggesting that these factors may work in concert during a recombination-mediated telomere lengthening that potentially occurs in APBs.

In addition to recombination proteins, it is known that many factors are found at APBs, including telomere repeat binding proteins TRF1 and TRF2 (14, 16). The human homologue of yeast Rap1, hRap1, was recently identified as a telomere-associated protein, which contains a BRCT and is critical for telomere length regulation (36). Using an antibody specific for hRap1, we detected immunoreactivity that colocalized with APBs (Fig. 1 E). Similar to BRCA1 and the NBS1/Mre11/Rad50 complex, the association of hRap1 with APBs is also most evident when cells are arrested at the late S-G₂ phase (data not shown). Combined, the results of these localization experiments suggest that a wide spectrum of both telomere and DSB repair-associated factors may be operational in human ALT pathway.

NBS1/Mre11/Rad50 complex and Rad51 are known to play an important role in the repair of ionizing radiation-induced DSBs. Although NBS1/Mre11/Rad50 and Rad51 form nuclear foci upon ionizing radiation, Rad51 cannot be found to be colocalized with NBS1 or Mre11 in a given cell (Ref. 37 and our unpublished observations). This observation is in sharp contrast to what was observed in Fig. 1 D that both groups of proteins are colocalized at APBs in ALT cells. Therefore, it is likely that recombination-dependent telomere length maintenance is mechanistically related but not identical to DSB repair-associated homologous recombination.

To distinguish the relationship between these two types of nuclear foci, we test whether IRIF and the localization to APBs are dependent on each other. After treatment with γ-irradiation, increased percentage of cells exhibited granular nuclear speckles of BRCA1, Rad51, and NBS1, characteristic of DNA repair foci. These DNA-damaged induced foci did not colocalize with PML or TRF1 in ALT cells such as human osteosarcoma U-2 OS (Fig. 2) and Saos-2 cells (data not shown). However, it was noted that a small fraction of cells demonstrated APB-associated foci positive for the BRCA1, NBS1, and Rad51 antigens in irradiated and mock-irradiated cells. These same patterns of localization were also observed at earlier times (1.5 and 4 h) after exposure to ionizing radiation (data not shown). Combined, these findings suggest that DNA damage-induced foci and APB-associated foci form independently and are morphologically different.

The association of the above factors with telomeres or APBs in ALT or telomerase-positive cells, their colocalization with BrdUrd under various conditions, and their inclusion in IRIFs are summarized in Table 1. The results were obtained from experiments using either GM847 (ALT line) or T24 (telomerase positive) cells in this study or from other studies using various cell lines (14, 15, 16, 19, 37, 38, 39, 40). Cumulatively, these data indicate that the distribution of both DSB repair and telomere-associated factors in the nucleus appears to be dynamic and multifaceted. In both telomerase-positive and ALT cells, the NBS1 complex is associated with telomeres and APBs, within which DNA synthesis takes place during late S-G₂ of the cell cycle. The NBS1 complex also associates with replication forks throughout S-phase progression and is associated with DNA damage-induced foci when cells are irradiated. On the other hand, BRCA1 and Rad51 are associated with APBs and are also found in replication foci in the presence of DNA damage, regardless of the cell type.

Elucidating the assembly of functional APBs is central to understanding the ALT pathway. Given the complex spectrum of APB-associated factors, we sought to identify a few key molecules that are critical for their assembly. Because TRF1 is known to interact with both telomeric DNA (41) and NBS1 (16), we hypothesized that both proteins are critical for recruitment and/or assembly of at least BRCA1, Mre11, and/or Rad50 into APBs. To test this model biochemically and cell biologically, we established a subclone of U-2 OS cells, named as UNBS1C, that harbors a stably integrated tet-off system that controls the expression of a GFP-tagged truncated form of NBS1, specifically, a COOH-terminal region encompassed by amino acids 543–754. The rationale for this design is that, although the COOH-terminal region of NBS1 is sufficient for nuclear localization and binding Mre11, the mutated complex does not associate with IRIF (42, 43), an activity that apparently resides within the NH₂-terminal part of the NBS1. If NBS1 is required for APB association, we expected that conditional expression of GFP-NBS1C would result in formation of Rad50/Mre11/NBS1C complexes incapable of APB association. In Fig. 3,A, immunoblot data demonstrate that removal of tetracycline from the media resulted in a significant induction of GFP-NBS1C expression (compare Lanes 7 and 8). Coimmunoprecipitation with α-Mre11 and α-myc and immunoblot detection showed that Rad50 and Mre11 are present in a complex with GFP-NBS1C (Fig. 3 A, Lanes 3–6).

The effects of the truncated NBS1C/Mre11/Rad50 complex on the association of Mre11 and BRCA1 with APBs and DNA synthesis are summarized in Fig. 3,B. There was a significant reduction in the localization of BRCA1 at APBs upon removal of tetracycline (P = 0.0153). The most dramatic effect was on Mre11-associated APBs (P < 0.0001), which are normally detected in ∼30% of the cells arrested at late S-G₂ stage (16). In contrast, Mre11-associated APBs were completely abolished in late S-G₂ cells upon induction of GFP-NBS1C expression. Consistent with these localization data were the fraction of APBs that displayed BrdUrd incorporation, which was also effectively reduced upon the expression of GFP-NBS1C (P = 0.0157). These data suggest that DNA synthesis was reduced in the APBs. Although not investigated here, the data imply an impairment of telomere length maintenance because telomere DNA is the only type known to be present at APBs (14). In contrast to the elimination of Mre11 and reduction of BRCA1 in APBs, the association of TRF1 and Rad51 proteins was only slightly reduced (Fig. 3 B), although GFP-NBS1C also binds to TRF1 (16). Thus, the mechanism involved in the recruitment of Rad51 to APBs remains unknown.

To eliminate the possibility that the above results were attributable to clonal variations, we designed a system for the expression of GFP-NBS1C using adenovirus, with which highly efficient infection and expression could be achieved. Several ALT lines, including GM847 and VA13, were infected and assessed for the cellular effects of GFP-NBS1C overexpression. Results similar to those with the inducible U-2 OS subline UNBS1C were obtained (data not shown).

These findings suggest that NBS1 is essential for recruitment and/or association of the Mre11/Rad50 complex and BRCA1 with APBs and for APB-specific DNA replication during late S-G₂ phase. It was noted that short-term expression of GFP-NBS1C (within 4 days) did not correlate with significant cell cycle arrest, as quantified by fluorescence-activated cell sorting analysis and immunostaining of the cell cycle marker phosphohistone H3 (data not shown). On the other hand, persistent expression of GFP-NBS1C for >2 weeks led to cell death (data not shown), suggesting that the association of intact NBS1 complex with APBs may be required for growth and proliferation of human ALT cells.

We next tested whether TRF1 is required for the association of NBS1, Mre11, and Rad51 with APBs. We used a recombinant adenovirus capable of directing expression of GFP-TRF1Δmyb that forms a nonauthentic TRF1Δmyb/TRF1 dimer, which acts as a dominant negative by depleting endogenous TRF1 from telomere DNA (41, 44). We reasoned that its overexpression would reduce or inhibit the association of NBS1 with telomeres and/or APBs. Unexpectedly however, NBS1 antigen is detected with APBs in a fraction (Fig. 4,B) of GM847 cells in the presence of overexpressed GFP or GFP-TRF1Δmyb (Fig. 4,A). Similar results were obtained in two other ALT lines, U-2 OS and VA13 (data not shown). Interestingly, instead of a reduction, a slight increase of the association was observed in cells overexpressing GFP-TRF1Δmyb, compared with the control cells expressing GFP alone (Fig. 4 B). Taken together, these results suggest that the association of NBS1/Mre11/Rad50 with APBs is independent of TRF1. It is possible that the association could be mediated by another APB component, by other telomeric factors, or by telomere DNA because all of the members of NBS1/Mre11/Rad50 complex have intrinsic DNA binding activity (34, 35).

On the basis of the results described above, we concluded that NBS1 plays a central role in the assembly of the functional APBs, which depend on the recruitment of Mre11, Rad50, and BRCA1. Fig. 5 summarizes several sets of experiments designed to elucidate the structural requirement of NBS1 for association with the Mre11/Rad50 complex and localization with APBs. The FHA and BRCT domains are two known conserved modules residing at the NH₂-terminal part of NBS1 (Fig. 5,A and Ref. 45). In addition, two regions that are also relatively conserved from human to chicken were found and designated as CR1 (amino acids 197–280) and CR2 (amino acids 643–754), respectively (Fig. 5 A). CR1 is juxtaposed to several SQ-containing phosphorylation motifs that are important for DNA damage response (46, 47, 48, 49). CR2 is the interaction domain for several proteins including Mre11 and TRF1 (16, 42, 43).

Using this structural knowledge, we generated a panel of NBS1 mutants and determined their ability to associate with Mre11 complex and APBs by coimmunoprecipitation and/or immunostaining (Fig. 5,B). Mutation of a conserved residue (His45Ile, Fig. 5 B) in the FHA domain reduced the association of NBS1 with APBs. Similar reductions were also observed with several other mutants such as NBS1ΔCR1A (CR1 completely deleted) and NBS1ΔCR2. Strikingly, deletion of the BRCT domain (or both FHA and BRCT domains) completely eliminated the association of NBS1 with APBs. In contrast, deletion of Mre11 interaction domain (NBS1ΔCR2N) did not affect this association, suggesting that the association of NBS1 with APBs is Mre11 independent. Furthermore, expression of NBS1ΔBRCT in U-2 OS cells reduced the colocalization of Mre11/Rad50 and BRCA1, consistent with our observations derived from UNBS1C (data not shown). Serine 343 of NBS1 has been identified as a critical phosphorylation site during DNA damage responses (46, 47, 48, 49). However, changing serine 343 to alanine (unphosphorylatable) or glutamate (mimicking phosphorylation) did not affect the association of NBS1 with APBs, suggesting that DNA damage-induced modifications did not play a role in relocalization of NBS1. This result is consistent with our observation that association with DNA damage-induced foci by DSB factors is distinct from their ability to localize with APBs. Taken together, these results strongly suggest that the BRCT domain of NBS1 may mediate an interaction with a PML body component, and this interaction is sufficient to bring in NBS1 and its associated factors critical for telomere elongation to APBs in ALT cells.

In summary, our results implicated the BRCA1 and NBS1 in the human ALT pathway. We established that NBS1 is essential for the assembly of functional APBs, which are probably critical for telomere elongation in human ALT cells. Association of Mre11/Rad50 and BRCA1 with APBs depends on the integrity of NBS1, especially the BRCT module. The recruitment of Rad51 at APBs is NBS1 independent, consistent with the notion that Rad50 complex and Rad51 are required for the generation of different types of ALT cells in yeast (5). Although TRF1 is an integral part of the telomeres, it is dispensable for the recruitment of NBS1 and Rad51 to APBs. Therefore, TRF1 may not be essential for telomere length maintenance in the ALT pathway, but its role at telomeres could be more structurally related. Recombination-associated events responsible for telomere elongation in ALT cells would provide a unique stage for functional cross-talk between BRCA1, Rad51, and the NBS1/Mre11/Rad50 complex, which, in parallel to their roles in DNA recombination and/or DSB repair, contributes to the maintenance of genome stability. Determining whether the BRCA1/NBS1/Mre11/Rad50 complex functions mechanistically in the actual telomere lengthening process or serves as a signal transducer to recruit other recombination and/or DNA synthesis factors awaits additional studies.

We thank Chi-Fen Chen, Yumay Chen, Chang-Ching Liu, Paula Garza, and Diane Jones for their assistance, Qing Zhong for helpful discussion, and Nicolas Ting and Dave Sharp for critically reading the manuscript. We also thank Bert Vogelstein for providing Adeasy system for generating recombinant adenovirus.