The proper maintenance of telomeres is essential for global genome stability. Mammalian telomeres are composed TTAGGG repeats bound to shelterin, a complex of six core proteins, including the double-stranded DNA binding proteins TRF1 and TRF2 and Protection of Telomeres 1 (POT1) that interacts with its binding partner TPP1 to protect single-stranded G-rich overhangs1. Telomeres rendered dysfunctional by the removal of TRF2 are recognized as DSBs, activate ATM and are ligated by the NHEJ pathway to generate fused chromosomes3. The MRN complex recruits ATM to sites of DSBs, where it initiates a signaling cascade leading to checkpoint responses9, 10. While MRN localizes to deprotected telomeres11, 12, its role in NHEJ-mediated repair is not well understood. We sought to uncover the roles of the MRN complex in NHEJ by utilizing two mouse alleles of Mre11: the Mre11^Δ/Δ null allele that also abolishes the formation of the MRN complex, and the Mre11^H129N/Δ nuclease-deficient allele that eliminates both endo and exonuclease functions of Mre118. To engage the NHEJ pathway, we removed TRF2 from telomeres with retrovirus-mediated shTRF2 in SV40LT immortalized mouse embryonic fibroblasts (MEFs) (Supplemental Fig. 1a). Metaphase spreads harvested 72 to 120h after shTRF2 treatment revealed a progressive increase in the number of fused telomeres until nearly all chromosomes are joined end-to-end (Supplemental Fig. 1b). This telomere fusion phenotype was suppressed when MEFs were first complemented with shRNA-resistant murine TRF2 cDNA prior to shTRF2 treatment (Supplemental Fig. 1c, d, Supplemental table S1), indicating that the telomere fusions observed are not due to off target effects of shTRF2 treatment.

To determine whether Mre11 is required for the DDR initiated by dysfunctional telomeres, mouse embryo fibroblasts (MEFs) containing the conditional Mre11^F/Δ allele were infected with adenovirus expressing Cre recombinase. The floxed Mre11 allele was efficiently deleted, with concomitant loss of Mre11 protein as confirmed by Western blotting (Supplemental Fig. 2b, c). Since TRF2 represses ATM activation at telomeres4, 5, removal of TRF2 from Mre11^F/Δ cells resulted in robust ATM activation, phosphorylation of Chk2 and induction of γ-H2AX and 53BP1 telomere-induced DNA damage foci (TIF) that colocalized with telomeric DNA in 90% of cells examined (Figure 1a, c, Supplemental Fig. 3a). However, both ATM phosphorylation and TIF formation were reduced to 12% in Mre11^Δ/Δ MEFs in response to TRF2 depletion (Figure 1a, c; Supplemental Fig. 3a). To determine whether the Mre11 nuclease activity is required for ATM activation after induction of telomere dysfunction, Mre11^H129N/Δ MEFs were generated (Supplemental Fig. 2a–c). In contrast to Mre11^Δ/Δ MEFs, the phosphorylation of ATM and induction of TIFs upon TRF2 depletion in Mre11^H129N/Δ MEFs remained robust (Fig. 1b, d; Supplemental Fig. 3b). These results suggest that ATM phosphorylation in response to TRF2 depletion requires MRN, while the Mre11 nuclease activity is dispensable for this function. At murine telomeres, ATR activation is repressed by Pot1a4, 5. To test whether ATR activation at telomeres requires MRN or Mre11 nuclease activity, shPot1a was utilized to remove Pot1a in Mre11^Δ/Δ and Mre11^H129N/Δ MEFs. Pot1a depletion resulted in robust ATR phosphorylation and TIF formation in both cell types (Figure 1e, f, Supplemental Fig. 3c, d), suggesting that Mre11 is not required to activate an ATR-dependent DDR at telomeres due to Pot1a loss.

Our data demonstrate that removal of TRF2 results in telomeres being recognized as DSBs that elicit MRN dependent DDR. To determine whether repair of dysfunctional telomeres is impacted in the setting of Mre11 nuclease deficiency or upon removal of the MRN complex, TRF2 was depleted in Mre11^Δ/+, Mre11^F/Δ, Mre11^Δ/Δ and Mre11^H129N/Δ MEFs. While Mre11^F/Δ cells displayed extensive end-to-end chromosome fusions in a Lig4-dependent manner, Mre11^Δ/Δ cells showed a 15-fold reduction in the number of chromosome-chromosome fusions (Fig 2a, Supplemental Fig, 4a–c and Supplemental Table S2). Unexpectedly, chromosome-chromosome fusions were similarly reduced in nuclease-deficient Mre11^H129N/Δ cells, even though the DDR is activated upon TRF2 loss (Fig 1d, Fig 2a, and Supplemental Table S2). These results suggest a critical role of the Mre11 nuclease activity in mediating NHEJ of dysfunctional telomeres.

The dramatic reduction in the number of chromosome-chromosome fusions observed in Mre11^H129N/Δ cells treated with shTRF2 suggest the possibility that nucleolytic processing of telomeric ends might be required for NHEJ of dysfunctional telomeres. To test this hypothesis, we examined the status of the 3’ single-strand overhang using an in gel hybridization assay. In contrast to shTRF2 treated Mre11^F/Δ cells, which showed rapid reduction of the 3’ overhang due to NHEJ-mediated processing13, the 3’ overhang persists in Mre11^Δ/Δ and Mre11^H129N/Δ cells with uncapped telomeres (Fig. 2b). Treatment with the 3’ end specific exonuclease ExoI revealed that these overhangs are indeed single-stranded telomeric repeats (Supplemental Fig, 5a, b). `The lack of overhang degradation in Mre11^H129N/Δ cells treated with shTRF2 suggests that nucleolytic processing of 3’ telomeric overhang by Mre11 may be required for efficient NHEJ of telomeres rendered dysfunctional through the removal of TRF2.

Interestingly, telomere fusions were not completely abolished in shTRF2 treated Mre11^Δ/Δ cells. Instead, chromosome-orientation FISH (CO-FISH) revealed that ~90% of these rare fusions involved the leading strands of sister chromatids (Fig. 2c, 2d). After telomeric replication, telomeres formed by leading strand synthesis are initially blunt, while lagging strand DNA synthesis result in the formation of 3’ overhang after the removal of the last synthesis primer. Leading strand telomeric ends have to be enzymatically processed after replication to produce a 3’ overhang that is protective against NHEJ-mediated fusion. We postulate that the MRN complex prevents NHEJ of newly synthesized leading strand telomeres by mediating 5’ end resection of the leading strand to generate a 3’ overhang. In support of this notion, leading-leading chromatid fusions comprised ~60% of all chromatid fusions in shTRF2 treated Mre11^H129N/Δ cells, suggesting that the nuclease activity of Mre11 is at least in part responsible for the formation of the 3’ overhang at the leading strand (Fig. 2c, 2d).

While the Pot1a-TPP1 shelterin sub-complex14 binds to the 3’ single-stranded telomeric overhang and represses ATR-mediated DDR4, 5, little is known how the overhang is protected from engaging DNA repair pathway(s). The presence of intact single-stranded telomeric overhang in shTRF2 treated Mre11^Δ/H129N cells, and the ability of this mutant to activate ATM/ATR signaling pathways represents an unique opportunity to address this question. Using immunostaining and telomere FISH to detect the localization of endogenous TPP1 at single-stranded overhangs and the double stranded telomeric binding protein TRF1 to mark telomeres, we observed a 3.5-fold reduction of TPP1, but not TRF1, at telomeres in Mre11 proficient cells following TRF2 depletion (Supplemental Figs. 6a–c), reinforcing the notion that telomeric accumulation of the Pot1a-TPP1 complex is reduced when TRF2 is depleted15. In contrast, TPP1 immunostaining remained robust in shTRF2 treated Mre11^H129N/Δ cells (Supplemental Figs. 6a–c), suggesting that Pot1a-TPP1 remain complexed to the single-stranded telomeric overhangs in these cells.

We next asked whether removal of Pot1a-TPP1 rendered the overhang amenable to DSB repair. We first depleted TRF2 in Mre11^H129N/Δ MEFs, then removed Pot1a, Pot1b, or TPP1 from the single-stranded overhang using shRNAs (Supplemental Figure 7a). Depletion of Pot1a but not Pot1b resulted in a 3-fold increase in telomere end-to-end fusions (Figure 3a, b), supporting the hypothesis that in mouse cells, Pot1a helps protect telomeres from engaging inappropriate repair, while Pot1b is involved in the protection of the C-strand16–19. Removal of TPP14 from the 3’ overhang of shTRF2 treated Mre11^H129N/Δ cells resulted in a 5-fold increase in chromosome end-to-end fusions, with 30% of all chromosomes fused (Fig. 3a, b, Supplemental Figs. 7a, 8a, b). Taken together, these results suggest that at telomeres lacking TRF2 and Mre11 nuclease activities, the TPP1-Pot1a complex is able to fully protect the telomeric overhang from engaging inappropriate repair pathways.

ATM, ATR and 53BP1 have been implicated in mediating NHEJ of dysfunctional telomeres following removal of TRF25, 20. We found that depletion of ATR, but not ATM, resulted in a 3-fold reduction in the number of end-to-end chromosomal fusions induced by retroviral expression of full-length TPP1^ΔRD in shTRF2 treated Mre11^H129N/Δ cells (Figure 3c, d, Supplemental Figs. 7b, 9a–c). Similarly, depletion of 53BP1 led to a 4-fold reduction of chromosomes fused in Mre11^H129N/Δ MEFs expressing TPP1^ΔRD (Figure 3c, d, Supplemental Fig. 7c). Collectively, these data argue that in the absence of the Pot1a-TPP1 complex, processing of the single-stranded overhang does not require Mre11 nuclease activity to mediate ATR and 53BP1-dependent chromosomal fusions. We postulate that other nucleases in the cell, for example the nucleotide excision repair nuclease ERCC1/XPF that has been shown to remove the 3’ overhang when coupled to the NHEJ machinery13, might substitute for Mre11 nuclease function at this stage.

Finally, we asked which pathway (NHEJ vs. HR) is required for repair of the single-stranded overhang. Removal of Pot1a/TPP1 from Mre11^H129N/F MEFs resulted in elevated HR at telomeres, detected as a 5-fold increase in telomere sister chromatid exchanges (T-SCEs)21 (Supplemental Fig. 10a, b). Interestingly, T-SCEs were not increased significantly in Pot1a/TPP1 depleted Mre11^H129N/Δ cells, suggesting that the Mre11 nuclease activity plays an important role in telomeric HR (Supplemental Fig. 10a–c). When TRF2 is removed from Pot1a/TPP1 depleted Mre11^H129N/Δ cells, robust Lig4 dependent chromosomal fusions were observed (Fig. 4a–c). However, T-SCE levels in these cells remained similar to those observed in control cells (Fig. 4d, Supplemental Fig. 11). These results supports the hypothesis that the main function of Pot1a-TPP1 is to repress HR at telomeres, while TRF2 is primarily responsible to repress NHEJ at telomeres.

The data presented here demonstrate multiple roles for the MRN complex at telomeres. Mre11 interacts with TRF2 and localizes to telomeres throughout the cell cycle11. Together with TRF2, it plays a protective role in preventing NHEJ-mediated fusion of leading strand telomeric DNA after replication, likely by promoting 5’ leading strand resection to generate Pot1a-TPP1 bound 3’ overhangs that prevents NHEJ (Supplemental Fig. 12). While the Mre11 nuclease activity plays a role in this process, additional factors are likely to be involved22. One such candidate is CtIP23, 24, since recent data suggests that the yeast homolog Sae2 cooperates with Mre11 to remove nucleotides at DSBs in a 5’-3’ endonucleolytic manner before extensive resection is performed by other nucleases including ExoI and Dna225–26. However, generation of 3’ single-stranded overhangs has the undesired consequence of initiating inappropriate HR at telomeres. Indeed, the Mre11 nuclease activity is required for telomere HR after Pot1a-TPP1 is removed from the 3’ overhang. Since initiation of HR requires RPA binding to single-stranded DNA, we speculate that the Pot1a-TPP1 complex represses HR by preventing RPA access to the 3’ overhang.

Removal of TRF2 from telomeres stimulates the MRN complex to activate ATM and 53BP1 to promote NHEJ. Surprisingly, removal of TRF2 also liberates the Mre11 3’-5’ nuclease activity to processes 3’ telomeric overhangs prior to NHEJ of chromosome ends. Since single stranded telomeric overhangs are incompatible with DNA ligation3, we speculate that degradation of the 3’ overhang by Mre1127, likely in conjunction with other components of the NHEJ pathway13, is required to generate telomeric substrates amenable for joining by Lig4. This model is supported by the observation that Mre11 nuclease activities are required for efficient NHEJ during class switch recombination in developing lymphocytes28.

Our studies also uncover an unexpected role for the Pot1a-TPP1 complex in protecting the 3’ overhang from NHEJ in the absence of TRF2 and Mre11 nuclease activity. Since TRF2 is thought to be essential for the formation of the t-loop structure proposed to protect chromosome ends from engaging the NHEJ pathway, our result support the notion that under certain conditions the t-loop is dispensable for telomere end protection29. We suggest that the Pot1a-TPP1 complex cooperates with TRF2-Rap1 during telomere replication, when the t-loop is transiently lost, to protect the integrity of linear chromosomal ends30.