Defective Repair of Uracil Causes Telomere Defects in Mouse Hematopoietic Cells*

Background: Telomeres may be susceptible to uracil misincorporation, which is removed by uracil DNA glycosylase (UNG). Results: UNG deficiency increases uracil in telomeres and causes alterations in multiple aspects of telomere maintenance in proliferating mouse hematopoietic cells. Conclusion: Accumulation of uracil interferes with telomere maintenance. Significance: UNG-initiated base excision repair is necessary for the preservation of telomere integrity. Uracil in the genome can result from misincorporation of dUTP instead of dTTP during DNA synthesis, and is primarily removed by uracil DNA glycosylase (UNG) during base excision repair. Telomeres contain long arrays of TTAGGG repeats and may be susceptible to uracil misincorporation. Using model telomeric DNA substrates, we showed that the position and number of uracil substitutions of thymine in telomeric DNA decreased recognition by the telomere single-strand binding protein, POT1. In primary mouse hematopoietic cells, uracil was detectable at telomeres, and UNG deficiency further increased uracil loads and led to abnormal telomere lengthening. In UNG-deficient cells, the frequencies of sister chromatid exchange and fragility in telomeres also significantly increased in the absence of telomerase. Thus, accumulation of uracil and/or UNG deficiency interferes with telomere maintenance, thereby underscoring the necessity of UNG-initiated base excision repair for the preservation of telomere integrity.

Uracil in the genome can result from misincorporation of dUTP instead of dTTP during DNA synthesis, and is primarily removed by uracil DNA glycosylase (UNG) during base excision repair. Telomeres contain long arrays of TTAGGG repeats and may be susceptible to uracil misincorporation. Using model telomeric DNA substrates, we showed that the position and number of uracil substitutions of thymine in telomeric DNA decreased recognition by the telomere single-strand binding protein, POT1. In primary mouse hematopoietic cells, uracil was detectable at telomeres, and UNG deficiency further increased uracil loads and led to abnormal telomere lengthening. In UNG-deficient cells, the frequencies of sister chromatid exchange and fragility in telomeres also significantly increased in the absence of telomerase. Thus, accumulation of uracil and/or UNG deficiency interferes with telomere maintenance, thereby underscoring the necessity of UNG-initiated base excision repair for the preservation of telomere integrity.
Telomeres are nucleoprotein structures at the ends of all linear eukaryotic chromosomes. In mammals, telomeres are composed of the shelterin protein complex and a double-stranded tract of short tandem repeats of TTAGGG that ends in a singlestranded 3Ј overhang on the G-strand (1). Telomeres cap and protect chromosome ends from eliciting a DNA damage response and illegitimate recombination events (2). A critical aspect of telomere maintenance is telomere length homeostasis. Telomere length is regulated by telomerase, the shelterin components POT1, TRF1, and TRF2 that directly bind telomere DNA, and molecular processes such as telomere recom-bination and replication. Telomerase is a ribonucleoprotein complex that adds telomere repeats to the chromosome ends (1), and is regulated by POT1 and its heterodimeric partner TPP1 (3). TRF1 and TRF2 are negative regulators of telomere length (4 -7). In addition, oxidized bases in the telomere repeat sequence disrupt telomere length (8 -10). However, it is unknown if uracil misincorporation impacts the affinity of telomerase and telomere-binding proteins to telomeric DNA, and affects the length, recombination, and replication of telomeres.
Uracil can arise in DNA by misincorporation of dUTP instead of dTTP opposite adenine during DNA synthesis, or by deamination of cytosine to uracil opposite guanine (11). Deamination can occur either spontaneously throughout the genome or by the enzyme activation-induced deaminase in the immunoglobulin loci. Thus, uracil through misincorporation produces U:A base pairs, which are not mutagenic since uracil resembles thymine during replication, whereas uracil through deamination yields U:G base pairs, which are highly mutagenic by producing C:G transitions after replication. Uracil in DNA is primarily recognized and excised by the enzyme UNG 4 during base excision repair. UNG-deficient mice have significantly elevated steady-state levels of genomic uracil but low mutation frequencies (12). These data indicate that the major role for UNG in mice is to remove uracil from misincorporated dUTP, and suggest that misincorporation is the predominant pathway for introducing uracil into mouse genomic DNA. To obtain biochemical data on the impact of this rogue base on telomere function, we dissected the changes in activities of inherent shelterin proteins and telomerase on DNA substrates containing uracil instead of thymine in the telomere TTAGGG repeats. * The work was supported entirely by the Intramural Research Program of the National Institutes of Health, NIA. 1 Both authors contributed equally to this work. 2 To whom correspondence may be addressed. We then examined the effect of uracil misincorporation on telomere length maintenance in proliferating hematopoietic cells by utilizing an Ung Ϫ/Ϫ mouse model that shows enhanced genomic uracil levels (12).

EXPERIMENTAL PROCEDURES
Mice and Primary Hematopoietic Cells-The generation of Ung knock-out (Ung Ϫ/Ϫ ) mice was described elsewhere (12). Wild type (Ung ϩ/ϩ ) and Ung Ϫ/Ϫ mice were derived from heterozygous crosses. Ung and Tert knock-out mice (Ung Ϫ/Ϫ Tert Ϫ/Ϫ ) were generated by crossing Tert (13) and Ung heterozygous mice. All animal procedures were reviewed and approved by the Animal Care and Use Committee of the National Institute on Aging. Bone marrow cells were flushed from femurs and tibias and cultured in RPMI medium (Invitrogen), supplemented with 20% fetal calf serum, interleukin 6 (200 units/ml; Peprotech), and stem cell factor (100 ng/ml; Peprotech).
Electrophoretic Mobility Shift Assay (EMSA)-Oligonucleotide (Midland Certified) preparation and EMSA were performed as described previously (8,14). Recombinant human shelterin proteins POT1/TPP1, TRF1, and TRF2 were purified as described previously (15,16). Mouse POT1a and POT1b were translated in vitro with the TNT coupled reticulocyte lysate kit (Promega) under the manufacturer's recommended conditions. Plasmids encoding mouse POT1a and POT1b were a kind gift from Dr. Sandy Chang (14). To evaluate competition between oligonucleotides with or without uracil, 5 pM 32 P-labeled oligonucleotides were mixed with different concentrations of unlabeled competing oligonucleotides and 50 pM of human POT1 protein in a buffer or 2 l of in vitro translated mouse POT1a or POT1b protein lysate at room temperature for 25 min. Free and protein-bound 32 P-labeled oligonucleotides were separated on a 4% polyacrylamide native gel and quantified by Phosphor-Imager analysis.
Telomerase Direct Primer Extension Assay-Telomerase extension was determined as described previously (17), except glycogen (2 g) was added to each sample instead of linear polyacrylamide as a carrier prior to DNA recovery. Singlestranded substrates (Midland Certified) were assayed using recombinant human telomerase captured on anti-Flag resin and washed with 0.3-0.6 M K-glutamate (17). After telomerase extension, products were resolved on a 10% w/v denaturing acrylamide gel. Quantification of individual radiolabeled species within each lane was carried out using ImageJ64. Individual measurements were assessed for each primer and, where appropriate, normalized to input primer amount and a control primer, d(TTAGGG) 2 , that was 5Ј-endlabeled with [␥-32 P]ATP using polynucleotide kinase (NEB) and added into the reactions prior to product extraction. Statistical analysis was carried out with at least two separate tests: two-way comparisons (e.g. comparing the sum of all repeats added to a dUTP-substituted primer compared with its corresponding unsubstituted primer) were assessed using an unpaired Student's t test with Welch's correction, and multiple comparisons (e.g. individual repeat signal intensities and the sum of all repeats, compared between multiple primers simultaneously) were assessed using 1-way ANOVA with Tukey post-test (Prism 5.0, GraphPad Inc.).
When the null hypothesis could not be rejected using both tests, the results were cited as not significantly different (p Ͼ 0.05).
Detection of Uracil in Mouse Hematopoietic Cells-1 g of genomic DNA from Ung ϩ/ϩ and Ung Ϫ/Ϫ bone marrow cells was incubated with (treated) or without (mock) 10 nM UNG (provided by Dr. J. Stivers, Johns Hopkins University) in a buffer (50 mM HEPES pH 7.5, 20 mM KCl, 4 mM EDTA, and 2 mM dithiothreitol) for 30 min at 37°C. 5 ng APE1 (provided by Dr. D. Wilson III, NIA/NIH) and 10 mM MgCl 2 were then added to the reaction and further incubated at 37°C for 1 h. qPCR analysis was performed on 10 ng of DNA using telomere (18) or Gapdh primers (19) using Power SYBR Green master mix (Invitrogen). Relative amplification was calculated by standard ⌬Ct method. Sensitivity of mock-or UNG-treated samples was calculated against input DNA (2 Ϫ(⌬Ct(mock) Ϫ⌬Ct(input DNA) ).
Chromosome Orientation Fluorescence In Situ Hybridization (CO-FISH) and Q-FISH-CO-FISH and Q-FISH were used to measure telomere sister chromatid exchange (T-SCE) and telomere length, respectively. For CO-FISH, mouse bone marrow cells were cultured in medium with a 3:1 ratio of BrdU/ BrdC at a final concentration of 1 ϫ 10 Ϫ5 M for ϳ12 h. The cells were arrested in mitotic phase by treatment with colcemid (0.1 g/ml) for 2-4 h. Metaphase spread preparation, hybridization and wash conditions, and image quantification were identical to those described previously (9,10,20,21). The telomere specific Alexa 488-labeled (TTAGGG) 3 and/or Cy3-labeled (CCCTAA) 3 peptide nucleic acid probes were used (0.3 g/ml, Panagene). The R statistical package and Graphpad software were used for plotting telomere signal intensities.

Uracil in Telomere DNA Interferes with Binding of POT1-
Rogue bases at telomeric DNA may alter the affinity of telomere-associated proteins. Shelterin proteins TRF1, TRF2, and POT1 directly bind to telomere DNA, to regulate telomeres (2,3). Here, we employed EMSA to investigate how uracil in telomeric DNA may impact recognition and binding of innate telomere-binding shelterin proteins. We focused on testing telomeric DNA substrates in which one or more thymines were replaced by uracil.
POT1 binds the single-stranded (ss) telomeric 3Ј-overhang in diverse eukaryotic chromosomes (22) and regulates telomere length (3). TPP1 facilitates the recruitment of POT1 to telomeres and enhances POT1 binding to ssDNA (15,(23)(24)(25)(26). First, we examined binding of the human POT1/TPP1 protein complex to a 32 P-labeled ss telomere oligonucleotide (GGT 1 T 2 AGGGT 3 T 4 AG; underlined is the minimal sequence required for efficient POT1 binding (27)) in the presence of increasing amounts of unlabeled telomere oligonucleotides that contained no uracil (unmodified) or one or more thymines replaced by uracil (modified). Competition between the 32 Plabeled oligonucleotide and excess of unlabeled oligonucleotide would lead to a decrease in the amount of POT1/TPP1-bound 32 P-labeled oligonucleotide complex in EMSA. This decrease is directly proportional to the binding affinity of POT/TPP1 to the respective unlabeled competing oligonucleotide. Compared with the unmodified competing oligonucleotide (Fig. 1A, lanes 2-6), modified competing oligonucleotides that contained uracil within the POT1 binding site (Fig. 1A, lanes [7][8][9][10][11][12][13][14][15][16][17][18][19][20][21] showed diminished affinity for POT1/TPP1, as evident from increased retention of the protein-oligonucleotide complex. Modified competing oligonucleotides with uracil at T1 and T2 positions displayed lower affinity for POT1/TPP1, than those with uracil at positions T3 and T4 positions (Fig. 1A, lanes 12-16 and lanes 7-11, respectively). A similar observation was also reported when thymine was replaced with adenine (14). This can plausibly be attributed to the fact that although POT1 binds to ss DNA in an extended conformation, where both oligonucleotide/oligosaccharide binding folds contact DNA, it is the N-terminal fold that makes more extensive contacts with the first six nucleotides at the 5Ј-end of the ssDNA (27). Significantly, modified competing oligonucleotides harboring uracil at all four thymines in the POT1 binding sequence (Fig. 1A, lanes 17-21) exhibited weaker POT1/TPP1 binding than those that had uracil at only two positions (Fig. 1A, lanes 7-16). As a control, modified competing oligonucleotides with uracil outside the POT1 binding sequence had no effect on binding (Fig.  1B).
To extend the binding studies of human POT1 in mice, we performed similar experiments with mouse POT1 proteins. Mice have two isoforms of POT1: POT1a and POT1b (14,28), and both share similar DNA binding properties in vitro (14,29). Human POT1 protein combines the functions of two mouse POT1 proteins (29) and also has very similar sequence specificity and DNA affinity as mouse POT1 (30). Similar to human POT1, mouse POT1a and POT1b form heterodimers with TPP1 that can be tethered to double-stranded telomeric DNA FIGURE 1. Affinity of POT1 to uracil-containing telomere substrates. A, representative EMSA showing the affinity of purified human POT1/TPP1 to telomere oligonucleotides with or without uracil. The oligonucleotide GGT 1 T 2 AGGGT 3 T 4 AG (underlined is the POT1 binding sequence; number indicates the position of thymine in the oligonucleotide) was 32 P-labeled, incubated with purified POT1/TPP1 in the absence (lane 1) or presence of increasing concentrations (5-fold molar excess) of various unlabeled competing oligonucleotides (shown at the top; thymine at T1, T2, and/or T3, T4 positions was replaced by uracil (red)). B, the experiment was conducted as in A except for the oligonucleotide GGTTAGGGTTAGGGT 5 T 6 , with thymine at T5 and T6 positions replaced by uracil. C, the experiment was conducted as in A except that the lysates containing in vitro translated mouse POT1a and POT1b were used. Corresponding graphs show the fraction of POT1-bound 32 P-labeled oligonucleotides plotted against fold change of molar excess of competing oligonucleotides. POT1-bound 32 P-labeled oligonucleotide was calculated as a fraction of total radioactivity in each lane. Error bars: S.D. from three independent experiments. by TIN2 (30 -32). We examined the binding of mouse POT1a and POT1b to the telomere oligonucleotides with or without uracil. Modified competing oligonucleotides harboring uracil at all four thymines in the POT1 binding sequence exhibited weaker POT1a and POT1b binding (Fig. 1C). Thus, the presence of uracil in place of thymine within the core POT1 binding site in telomeric DNA significantly diminished the binding affinity of both human and mouse POT1, which depended on the position and number of uracil replacements.
We then examined binding of the two other bona fide telomeric proteins TRF1 and TRF2 to a telomere repeat-containing oligonucleotide duplex without or with uracil replaced by thymine. The binding of human TRF1 and TRF2 to the duplex oligonucleotides with uracil replacing a single T in the TTAGGG sequence was comparable to that without uracil (data not shown). In order to determine if increased uracil load in the telomere repeats affected the binding of TRF1 or TRF2, all the thymines in the telomere repeat region of the duplex were replaced with uracil ( Fig. 2A). Despite this high uracil content, the affinity of both TRF1 and TRF2 to the uracil-modified duplex remained the same as that of the unmodified duplex (Fig. 2, B and C). Collectively, this in vitro analysis demonstrated that uracil in the telomeric substrate had deleterious effects on the binding affinity of POT1, but had no significant impact on binding of TRF1 or TRF2.
Limited Tracts of Uracil in Telomeric DNA Do Not Affect Overall Telomerase Extension-Telomerase recognizes and extends ss DNA primers by virtue of limited primer-template base pairing between telomeric DNA and the telomerase RNA, and by a DNA-binding anchor site within telomerase reverse transcriptase itself (33,34). These and other properties of the enzyme enable it to tolerate substitution of dTTP in the primer without significant impairment of binding or extension (17,(35)(36)(37)(38). We tested whether uracil substitution would affect the ability of recombinant human telomerase to utilize substrates with extensive or limited complementarity to the telomerase RNA template (Fig. 3). We immuno-purified recombinant FLAG-human telomerase reverse transcriptase (hTERT) and human telomerase RNA after reconstitution in rabbit reticulocyte lysate (17) and assayed the ability to extend the primer TT (5Ј-AGGGTTAGGGTTAGGGTT-3Ј) or the same sequence in which the first, second, or third TT registers were replaced by dUTP (UU1, UU2, UU3, respectively) (Fig. 3A). Quantification of the results showed no significant difference in primer utilization, by comparing the intensity of specific repeat sequences, or the sum of all repeats added, between TT and UU1, UU2 or UU3 (Fig. 3, A and C). We also tested the impact of dUTP substitution on a largely non-telomeric primer, GTT (5Ј-AATCCGTCGAGCAGAGTT-3Ј) containing dUTP substitutions at the penultimate TTP (GUT), last TTP (GTU), or both terminal TTP residues (GUU). Again, there was no significant difference in primer utilization by telomerase when comparing the sum of all repeats added, or when comparing the intensity of individual repeats (Fig. 3, B and C). Thus, limited substitution of TTP with uracil did not significantly affect the overall primer extension activity of telomerase in vitro.
Uracil Accumulation Due to Defective Removal Disrupts Telomere Length Homeostasis in Ung Ϫ/Ϫ Hematopoietic Cells-Our in vitro data support the hypothesis that uracil in telomeric DNA substrates diminishes the affinity of POT1. However, it is not known how the accumulation of uracil affects telomeres in vivo. To investigate the potential effects of accumulation of ura-  FEBRUARY 27, 2015 • VOLUME 290 • NUMBER 9 cil at telomeres, we utilized mouse models deficient in the enzyme UNG that removes misincorporated uracil. Though Ung Ϫ/Ϫ -deficient mice show enhanced uracil in the genome (12), accumulation of uracil at telomeres has not been investigated. In order to measure uracil at telomeres, genomic DNA from Ung ϩ/ϩ and Ung Ϫ/Ϫ bone marrow cells were examined by a quantitative PCR method (19). First, genomic DNA was treated in vitro with mock or UNG enzyme. The resulting abasic sites were then nicked by addition of APE1. Thus, samples with higher levels of uracil would have decreased intact DNA. To quantify the levels of intact DNA, quantitative PCR analysis was carried out using telomere specific primers (18). Compared with input DNA, the mock-treated Ung ϩ/ϩ and Ung Ϫ/Ϫ DNA showed no decrease in amplification. However, after UNG treatment, Ung ϩ/ϩ DNA had a 56% decrease in PCR amplification, suggesting the presence of uracil in telomeres (Fig. 4A).

Uracil Misincorporation Disrupts Telomere Maintenance
Furthermore, Ung Ϫ/Ϫ DNA had an additional 23% decrease in amplification efficiency, suggesting that UNG removes uracil at telomeres and in its absence, more uracil accumulates. As a control, uracil content was measured at the Gapdh locus (Fig.  4B). Unlike telomeres, the Gapdh locus is not sensitive to UNG treatment in either Ung ϩ/ϩ or Ung Ϫ/Ϫ DNA. Collectively, these results suggest that telomeres are prone to uracil accumulation and that decreased repair due to lack of UNG exacerbates this phenomenon.
Next, we rationalized that the increased uracil content in telomeres in Ung Ϫ/Ϫ mice may disrupt telomere length homeostasis. Mounting evidence supports the notion that telomere dysfunction primarily affects the hematopoietic lineage in both humans and mice (39,40). We therefore examined telomere length in primary bone marrow cells from Ung Ϫ/Ϫ mice by Q-FISH (Fig. 5, A-C). Ung Ϫ/Ϫ mice exhibited a significant, albeit variable, increase in telomere length compared with Ung ϩ/ϩ mice. Wild-type telomere lengths were occasionally observed in Ung Ϫ/Ϫ mice (data not shown). Because uracil in telomeric DNA had no effect on telomerase activity in vitro (Fig. 3), we further delved into the involvement of telomerase in telomere lengthening in mice deficient for both UNG and telomerase. We generated Ung ϩ/ϩ Tert Ϫ/Ϫ and Ung Ϫ/Ϫ Tert Ϫ/Ϫ mice by crossing Ung Ϫ/Ϫ mice with a strain lacking TERT (13). Consistent with the in vitro data, the majority of Ung Ϫ/Ϫ Tert Ϫ/Ϫ mice continued to show significantly longer telomeres as compared with their Ung ϩ/ϩ Tert Ϫ/Ϫ counterparts (Fig. 5, D and E and data not shown). Thus, UNG deficiency and, in turn, defective uracil removal can lead to lengthening of telomeres, even in the absence of telomerase.
Common fragile sites represent specific chromosomal regions that are sensitive to replication stress, such as treatment with low levels of the DNA polymerase inhibitor aphidicolin and display breaks or gaps in metaphase chromosomes (45). Telomeres are aphidicolin-induced fragile sites, showing aberrant discontinuous multiple telomere signals, namely fragile telomeres (46). Though the underlying cause for the appearance of these structures is still unknown, conditions that cause incomplete DNA replication or stalled replication forks commonly result in telomere fragility (46 -49). UNG is known to localize to replication foci and play an important role in DNA replication (12,50). Furthermore, accumulation of oxidative base lesions results in telomere fragility (10). These reports led us to inquire if uracil misincorporation or UNG deficiency would lead to defective telomere replication, manifesting in telomere fragility. For this purpose, we performed telomere-FISH on primary mouse bone marrow cells from Ung ϩ/ϩ and Ung Ϫ/Ϫ mice (Fig. 6D). Although the frequency of fragile telomeres was comparable in Ung ϩ/ϩ and Ung Ϫ/Ϫ cells (Fig.  6E), UNG ablation caused a significant increase in fragile telomeres in the telomerase null background (Fig. 6F), similar to our observation for T-SCE events. Taken together, these results support the idea that telomerase deficiency exacerbates aberrant telomere homologous recombination and debilitates telomere replication in the absence of uracil removal by UNG.

DISCUSSION
UNG deficiency is associated with B-cell lymphomas and pathological changes in lymphoid organs in mice and humans (51,52). Mounting evidence supports the notion that telomere dysfunction primarily affects the hematopoietic lineage in both human and mice, suggesting the importance of telomere maintenance in highly proliferating cells (39,40). Because of the long arrays of TTAGGG repeats, telomeres might be susceptible to misincorporation of uracil. The finding that uracil is dramatically increased in telomeres from wild type mice, compared with the Gapdh control locus, underscores the susceptibility of DNA ends to instability. Uracil misincorporation could occur more frequently because telomerase has low fidelity (53) and may incorporate dUTP more readily than DNA polymerases delta and epsilon that replicate Gapdh genes. In addition, DNA ends may be protected by shelterin proteins, which would limit access to the UNG enzyme and prevent removal of uracil.
Inappropriately long telomeres may have serious consequences on cells, for instance by resulting in stalled replication forks or the formation of secondary structures that might  FEBRUARY 27, 2015 • VOLUME 290 • NUMBER 9  perturb normal cell cycle progression or by promoting tumorigenesis (54 -59). Overall, telomere integrity depends on the maintenance of length equilibrium, replication, and homologous recombination, all of which could be affected by base alterations such as uracil misincorporation in telomeric DNA. Telomere length is regulated by a complex set of factors including the shelterin complex and telomerase. The shelterin proteins TRF1, TRF2, and POT1, when bound to telomeric DNA substrate, are negative regulators of telomere length (4 -7, 60). Unlike oxidative base lesions of guanine (61), uracil replacement of thymine in telomeric DNA substrates did not affect binding of TRF1 and TRF2 in vitro, although POT1 binding was significantly reduced. This is consistent with the structural reports that the specificity of TRF1 and TRF2 is primarily accredited to their direct contacts with the cluster of G's in the telomeric DNA repeats (62), while the specific conformation imposed on the ssDNA by POT1 precludes its binding to uracil (27,63).

Uracil Misincorporation Disrupts Telomere Maintenance
Recent studies show that mutations in POT1 predispose families to cutaneous melanoma and abnormal telomere lengthening (57,59). Somatic POT1 mutations also lead to telomere lengthening and favor the acquisition of the malignant features of chronic lymphocytic leukemia (56). Further, one of the two mouse POT1 proteins, POT1a, is required for maintaining telomere length homeostasis, with its conditional deletion leading to telomere over-lengthening (64). Our data show a correlation between binding of POT1 proteins to uracil substrates in vitro and abnormal telomere length maintenance in Ung Ϫ/Ϫ mice. The addition of telomere repeats by telomerase requires its access to the terminal residues of the 3Ј singlestranded overhang; however, POT1 is believed to coat the entire 3Ј overhang (27), thereby preventing access of telomerase and acting as a negative regulator of telomere length (3). When uracil replaces thymine at the extreme 3Ј-end of telomeric and non-telomeric substrates in vitro, the overall extension activity by telomerase is unaffected, in agreement with our in vivo observations. However, uracil in telomeric DNA weakens the binding affinity of POT1 in vitro. This might reduce the amount of bound POT1 and thus increase the accessibility of telomerase to the 3Ј overhang, which could potentially contribute to the telomere lengthening phenotype we observed. However, our in vivo data suggests that telomere lengthening could be independent of telomerase. Thus, there may be another route for telomere lengthening in UNG-deficient mouse cells. In 8-oxoguanine DNA glycosylase-null strains of budding yeast, Rad52, a key protein of the homologous recombination pathway, is involved in telomere lengthening in some of the clones (8). Here, we have shown that ablation of UNG in a telomerasedeficient background not only leads to telomere lengthening but also to an increase in T-SCEs in mice. It is possible that uracil in telomeric DNA enhances homologous recombination at telomeres, leading to telomere lengthening in Ung Ϫ/Ϫ Tert Ϫ/Ϫ mice.
UNG plays an important role in DNA replication (12,50); however its deficiency alone did not have a significant impact on telomere fragility or recombination in proliferating hematopoietic cells. The level of uracil accumulated in vivo in UNGdeficient cells might not be substantial enough to result in detectable levels of either fragile telomeres or T-SCEs. Also, the nature and type of base lesions or alterations may differentially influence telomere phenotypes. For instance, it has been shown that TRF1 is required for efficient telomere replication and suppression of telomere fragility (46). Aberrant binding of TRF1 to telomeric DNA containing oxidized bases (61) may contribute to increased telomere fragility, as observed in 8-oxoguanine or endonuclease III-like protein 1 DNA glycosylase-null hematopoietic cells (10). Conversely, uracil does not directly impair the binding of TRF1 to telomeric substrates. In our studies, telomerase deficiency elevated fragile telomeres and T-SCEs in Ung Ϫ/Ϫ mouse cells, suggesting a possible cooperative role of telomerase and UNG-sponsored base excision repair in maintaining telomere integrity in mice. Telomere replication and telomerase-mediated extension are believed to be coupled processes (65,66). Since UNG has been implicated in uracil removal during and after DNA replication (12,50,67), there may be a window of time when both telomerase and UNG collaborate to ensure maintenance of telomere integrity during telomere replication.
In summary, we have demonstrated that enhanced uracil substitutions of thymine in telomeric DNA result in alterations/aberrations in multiple aspects of telomere maintenance in proliferating mouse hematopoietic cells, especially in the absence of telomerase. Since reduced base excision repair capacity has been associated with aging and UNG mutations with hyper IgM syndrome in humans (51,68), our findings in mice could hold true for humans as well.