|Home | About | Journals | Submit | Contact Us | Français|
The Werner and Bloom syndromes are caused by loss-of-function mutations in WRN and BLM, respectively, which encode the RecQ family DNA helicases WRN and BLM, respectively. Persons with Werner syndrome displays premature aging of the skin, vasculature, reproductive system, and bone, and those with Bloom syndrome display more limited features of aging, including premature menopause; both syndromes involve genome instability and increased cancer. The proteins participate in recombinational repair of stalled replication forks or DNA breaks, but the precise functions of the proteins that prevent rapid aging are unknown. Accumulating evidence points to telomeres as targets of WRN and BLM, but the importance in vivo of the proteins in telomere biology has not been tested. We show that Wrn and Blm mutations each accentuate pathology in later-generation mice lacking the telomerase RNA template Terc, including acceleration of phenotypes characteristic of latest-generation Terc mutants. Furthermore, pathology not observed in Terc mutants but similar to that observed in Werner syndrome and Bloom syndrome, such as bone loss, was observed. The pathology was accompanied by enhanced telomere dysfunction, including end-to-end chromosome fusions and greater loss of telomere repeat DNA compared with Terc mutants. These findings indicate that telomere dysfunction may contribute to the pathogenesis of Werner syndrome and Bloom syndrome.
Clues to aging mechanisms have come from segmental progeroid disorders, genetic diseases that hasten certain features of aging. Chief among these is Werner syndrome, characterized by premature aging of the skin, vasculature, and bone and elevated rates of certain cancers, particularly sarcomas (12, 37). A related disease is Bloom syndrome, characterized by limited features of aging, including premature menopause, and by short stature and elevated rates of most cancers (10). Werner syndrome and Bloom syndrome are caused by loss-of-function mutations in WRN and BLM, respectively, which encode 3′-5′ DNA helicases of the RecQ family (11, 60). The WRN protein also includes a 3′-5′ exonuclease domain not shared with BLM (20, 47). Based on the genome instability in Werner and Bloom syndromes, cell biological studies, biochemical studies of the purified proteins, and genetic studies of RecQ homologues in yeasts and Escherichia coli, it is widely accepted that the proteins function in the recombinational repair of stalled replication forks or double-strand breaks (25, 35, 36). Furthermore, there is evidence that WRN and BLM function in S-phase checkpoint and apoptotic responses to DNA damage (25, 35). Several observations indicate that the proteins also play important roles in telomere maintenance (36). Although the maintenance of DNA stability throughout the genome by WRN and BLM presumably contributes substantially to their anticancer activities, it is not yet known what particular functions of the proteins are most important for preventing rapid aging.
Evidence pointing to telomere functions for the WRN and BLM proteins includes the observation that serially passaged Werner syndrome fibroblasts shorten telomeres more rapidly than controls and senesce prematurely (45). Telomere dysfunction likely contributes to the premature senescence because the senescence can be bypassed via artificial expression of telomerase (58). Expression of telomerase also rescues the elevated sensitivity of Werner syndrome cells to certain genotoxic compounds (18), consistent with a possible role for telomere dysfunction in mediating the sensitivity to the compounds. Furthermore, WRN and BLM interact with several proteins critical for telomere function (37), including the TRF2 protein (38), which plays a central role in telomere capping. WRN and BLM occasionally localize to nuclear foci containing telomere DNA (38, 59), particularly in so-called ALT cells that most likely maintain telomeres via recombination (23, 42), and overexpression of BLM lengthens telomeres in ALT cells (49). Additionally, in cells lacking telomerase, the Saccharomyces cerevisiae RecQ protein Sgs1p prevents rapid senescence and maintains telomeres via recombination (8, 19, 23).
These observations indicate that WRN and BLM might help maintain telomere integrity, possibly through the repair of damaged telomeres. Consistent with this idea, mouse Wrn null mutants are essentially normal and without signs of premature aging (29). The long telomeres and relatively high levels of telomerase activity in murine cells might mask WRN functions otherwise required to respond to telomere dysfunction. To test this, we asked if shortening of telomeres, via inactivation of the gene encoding the telomerase RNA template component (Terc), might expose functions for WRN. Although they possess some unique functions, WRN and BLM appear to have some overlapping activities (22, 25, 36), and so in our experiments we also combined the Wrn null mutation with the BlmM3 mutation to uncover such functional redundancy. While mice homozygous for Blm null alleles die as embryos (7, 14), mice homozygous for the BlmM3 hypomorphic mutation develop normally but have elevated rates of cancer after 1 year of age (30, 32). Here we present an analysis of phenotypes that arise in mice with combinations of mutations in the Wrn, Blm, and Terc loci.
Wrn+/− Blm+/− Terc+/− mice were generated by crossing Wrn−/− Blm−/− mice with Terc+/− mice (7, 50). These were in turn crossed to generate Wrn−/− Blm−/− Terc+/− mice, which were crossed with Wrn+/− Blm+/− Terc+/− mice to generate the different G1 mutant lineages. The frequencies of offspring of the different genotypes were at expected Mendelian ratios. Each G1 lineage was crossed to itself to generate G2 lineages, and so forth, using cousin by cousin matings to avoid generation of substrains. All mice were in a mixed genetic background of 50% C57BL/6, 37.5% 129Svj, and 12.5% BALB/c. For the second set of experiments (“new” crosses), mice carrying the three mutations in heterozygous form were crossed for six generations to homogenize the genetic background prior to the generation of G1 lineages. All mice were weighed weekly from weaning at 3 weeks of age until at least 100 days of age. For reasons of space limitation, studies were restricted to mice under 14 months of age. Mice were examined daily, and those judged to be suffering from ill health were sacrificed for analysis. Mouse experimental protocols were approved by the University of Pennsylvania Institutional Animal Care and Use Committee.
Tissues were fixed in 10% formalin overnight at 4°C, followed by storage in 70% ethanol at 4°C; 5-μm sections were stained with hematoxylin and eosin. Quantitation of apoptotic cells and mitoses was performed with the observer blind to genotype, and at least 80 crypts were counted per mouse. For terminal deoxynucleotidyl transferase-mediated dUTP nick end labeling (TUNEL) analyses, 5-μm tissue sections were stained with the ApopTag kit (Intergen) according to the manufacturer's instructions, and at least 50 crypts were counted per mouse.
Dual energy X-ray absorptiometry (DEXA) scans were performed of the vertebrae in the abdominal region of anaesthetized mice with a Lunar PIXImus 2. Scans were calibrated with a phantom model with 0.058 g/cm2 bone mineral density.
After hair removal with electric clippers, full-thickness 3-mm (original crosses) or 4-mm (new crosses) punch biopsies were made through to the panniculus carnosus over each scapula. Two wounds were made per mouse, and five mice per genotype were tested. The long and short axes of each wound were measured daily, and area was calculated with the formula for an ellipse (πab). Hair regrowth assays were done as described (53).
Ears were washed for 10 s in 70% ethanol and then in phosphate-buffered saline containing 250 μg of G418 per ml and then minced and digested four times for 30 min in 0.25% trypsin-EDTA at 37°C. Released cells and partially digested tissue were cultured in a 24-well plate in Dulbecco's modified Eagle's medium with 10% heat-inactivated fetal bovine serum, penicillin-streptomycin, 3% O2, and 5% CO2 until confluent, and then split into six-well plates. 3T3 proliferation assays were performed in ambient O2 and 5% CO2 as described (6). The usual 3-day passage interval was extended in cases where culture growth slowed substantially; in such cases the cells were refed every 3 days and allowed to become nearly confluent before they were passaged.
Splenocytes were cultured for 48 h in RPMI with 10% heat-inactivated fetal bovine serum, 10 μg of lipopolysaccharide per ml, and 5 μg of concanavalin A per ml and then arrested for 30 min with 0.75 μg of colcemid per ml. Metaphases were fixed, stained with 4′,6′-diamidino-2-phenylindole (DAPI), and analyzed as described (44). For samples from the new crosses, metaphases were scored by an observer blind to genotype. For telomere signal-free end analyses, metaphase spreads were stained with DAPI and a indocarbocyanine-conjugated peptide nucleic acid (PNA) telomere repeat probe, indocarbocyanine-OO-(CCCTAA)3 (Boston Probes) as described (Unit 18.4, Current Protocols in Cell Biology Online, 2003 [www.interscience.wiley.com]). Telomere flow fluorescence in situ hybridization was performed on bone marrow cells with a fluorescein isothiocyanate-labeled PNA telomere repeat probe as described (2). Telomere fluorescence per cell was calculated for cells gated in the G1 phase of the cell cycle based on DNA content, and telomere fluorescence for each sample was corrected for background fluorescence by control hybridizations of the same sample without the telomere PNA probe. All samples to be compared directly were processed and measured by flow cytometry on the same days to avoid interrun variability. Repeat measurements of the same sample yielded a coefficient of variation of 6%.
Graphpad Prism 3 software was used for statistical analyses. Unpaired two-tailed t tests were used to compare each genotypic group for the data in Fig. Fig.11 to to44 and and6.6. Kaplan-Meier analyses were performed for the data in Fig. Fig.5,5, and significance was determined with a log-rank test. Error bars in all figures indicate the standard error of the mean.
Earlier work had demonstrated that mice doubly homozygous for the BlmM3 hypomorphic allele and a Wrn null allele were viable and fertile (28, 50). Such Wrn−/− BlmM3/M3 mice were crossed with Terc+/− mice to generate Wrn+/− Blm+/M3 Terc+/− triple heterozygotes, which were in turn crossed to generate four lineages: wild type, Wrn−/− BlmM3/M3 (wb), Terc−/− (t), and Wrn−/− BlmM3/M3 Terc−/− (wbt) (see Materials and Methods). Terc−/− mice must be bred for several generations before telomeres become sufficiently short to elicit pathology (44), and so we carried each lineage through multiple generations.
The first and second generations (G1 and G2) of each lineage were fertile, and no abnormalities were observed for the first year of life in general appearance, body weight, or histologic surveys of the following organs: thymus, lymph nodes, spleen, heart, lung, gastrointestinal tract, liver, pancreas, submandibular gland, kidney, adrenal gland, skeletal muscle, bone marrow, femur, vertebrae, skin, ovaries, and testes (not shown). However, a decline in fertility was evident in G3 wbt mutants, and G4 wbt mutants were sterile; in contrast, the wb and t lineages each generated viable offspring through G7 (Fig. (Fig.1A)1A) . Gross and histological examination of gonads from G4 triple mutant mice revealed marked atrophy. In males, this was characterized by a decline in testes mass (Fig. (Fig.1B)1B) and by the loss in seminiferous tubules of spermatogenic cells and mature sperm (Fig. (Fig.1C).1C). These findings are similar to those observed in later-generation Terc−/− mutants (17, 26, 44) (Fig. (Fig.1B,1B, G6 t mice) but occur approximately two to three generations earlier in wbt mutants.
Although the differences in fertility between the t and wbt mice were large, it was possible that the mixed genetic background of the mice or differences in telomere length among the founders of each lineage might have contributed in a random fashion to the phenotypic differences. To control for these possibilities, we repeated the experiment; hereafter we refer to the first set of crosses as original crosses and the repeat set of crosses as new crosses. Furthermore, prior to beginning the new crosses, we first homogenized the genetic background of the mice through six generations of parent-offspring and intersibling crosses of mice carrying the three mutations in heterozygous form. In addition, in the new crosses, Wrn−/− (w), BlmM3/M3 (b), Wrn−/− Terc−/− (wt), and BlmM3/M3 Terc−/− (bt) lineages were also generated to gauge the individual contributions of Wrn and Blm mutations to the phenotypes. Consistent with results from the original crosses, wbt mice from the new crosses became infertile rapidly, at G3 (one generation earlier than they had in the original crosses), while the other genotypes were fertile through at least G4 (data not shown). Moreover, for the original four genotypes (wild type, wb, t, and wbt), all of the phenotypes tested in both the original and new crosses (fertility, body weight, intestinal pathology, wound healing, and cytogenetics; see below) were affected consistently, indicating that the differences observed between lineages are a consequence of their different Wrn, Blm, and Terc genotypes.
wbt triple mutants, like the other genotypes, had normal body weight in G1 (data not shown), but G3 and G4 wbt mutants weighed less and were smaller than controls. Weekly measurements revealed differences in the body mass apparent at 3 weeks of age and which increased as the mice grew older (Fig. (Fig.2A2A and and2B).2B). G3 bt mutants also weighed less than controls, but not as little as G3 wbt mutants, indicating that the Blm and Wrn mutations each contribute to low body weight in wbt mutants (Fig. (Fig.2C2C).
G3 and G4 wbt mutants frequently developed a kyphotic (hunched) and scruffy appearance (Fig. (Fig.2A).2A). Consistent with the observed kyphosis, DEXA (dual energy X-ray absorptiometry) scans of G3 mice revealed reduced bone density in the vertebral skeleton of wbt mutants compared with controls (Fig. (Fig.2D).2D). Although increased hair loss was sometimes observed in aging G3 and G4 wbt mutants, hair regrowth assays did not reveal significant differences between the different mutant lineages (data not shown).
We noticed that G3 and G4 wbt and bt mutant mice suffered from chronic diarrhea and frequently developed prolapse of the rectum (Fig. (Fig.2A).2A). At the histological level, the appearance of the small intestines of all G1 genotypes was normal, but compared with controls, the small intestines of G3 and G4 wbt mice were characterized by distortion of crypt architecture and signs of acute inflammation, including blood vessel dilatation and neutrophil infiltration (Fig. (Fig.3A).3A). Under normal conditions, intestinal epithelial cells undergo apoptosis at the end of their developmental cycle as they are sloughed off the tips of villi, but in contrast, epithelial cells in the crypts do not normally undergo apoptosis. However, crypt cells do apoptose in late-generation telomerase mutants (17, 44).
To test if Wrn and Blm mutation influenced the level of crypt apoptosis associated with Terc mutation, we quantified apoptosis by direct histologic examination of hematoxylin- and eosin-stained tissue sections and confirmed these data by quantifying TUNEL-positive nuclei in the tissue sections (Fig. (Fig.3B).3B). Levels of apoptosis in intestinal crypts were slightly elevated in G3 t and wt mutants but markedly elevated in G3 and G4 wbt mice compared with controls and reached the levels observed in G7 t mutants (Fig. (Fig.3B).3B). G3 bt mutants had levels of apoptosis as high as that of G3 wbt mutants, while G3 wt mutants had levels similar to that of G3 t mutants, indicating that, in this particular tissue, Blm may have a greater role than Wrn in the context of telomerase inactivation. Furthermore, we confirmed through TUNEL analysis that levels of apoptosis in G1 wbt mutants were not elevated (Fig. (Fig.3B),3B), indicating that this phenotype depends on telomere loss and not loss of telomerase per se.
Individuals with Werner syndrome suffer from chronic skin ulcers, particularly around the ankles, and late-generation Terc mutants display slow wound healing (12, 44). We therefore tested our mice for rates of wound healing and observed slow wound healing in G3 wbt mutants compared with controls (new crosses; Fig. Fig.4A).4A). Although the Wrn and Blm mutations each contributed to the slow healing of wbt mutants, the Wrn mutation appeared to contribute more than the Blm mutation because G3 wt but not G3 bt mutants healed wounds more slowly than the controls. Wound healing involves the interplay of several cell types, and we do not yet understand the basis for this defect. However, one cell type important for wound healing is the dermal fibroblast, and we observed significant growth defects in cultured adult skin fibroblasts from wbt mutants compared with controls (Fig. (Fig.4B4B).
Skin fibroblasts were obtained from the ears of 10-month-old G1 siblings (new crosses) and cultured serially in 3T3 assays. No significant differences in the rates of culture doubling were observed between wild-type and t cells, consistent with earlier studies (6), while wb and wbt cultures doubled more slowly but similarly to each other in early passages. However, after seven population doublings, the wbt cells slowed and failed to proliferate further for over 1 month. Based on the sloughing of cells that failed to exclude trypan blue from the plate into the medium and based on the enlarged and flattened appearance of the remaining adherent cells, a mixture of cell death and senescence might be responsible for the poor growth of the triple mutant cultures. It is possible that the impaired growth of wbt cells from G1 mice after serial culture under the damaging conditions of standard cell culture (41) reflects defects that also occur at later generations in vivo in wbt mice. Regardless, these observations provide another example of how mutations in Wrn, Blm, and Terc synergize to affect cell function.
Consistent with their frail appearance, we observed a dramatically reduced life span in G3 wbt mutants versus controls. As shown in Fig. Fig.5,5, the median age of death for G3 wbt mutants was 7 months, compared with greater than 10 months for all other genotypes (P < 0.0001 compared with all other genotypes by log-rank test). In addition, G3 bt mice also have a significantly reduced life span, with a median age of death of 11 months (P ≤ 0.002 compared with all other genotypes by log-rank test). Because the G3 wbt life span is significantly shorter than the G3 bt life span, the Wrn mutation contributes to the mortality of the wbt mutants, even though it has no effect on life span in the presence of active telomerase and also does not appear to shorten life span dramatically in G3 wt mice (Fig. (Fig.5)5) (28). There was no increased mortality for any of the genotypes in G1 animals during the first year of life (data not shown), indicating that the rapid mortality of G3 wbt and bt mutants apparently depends on telomere shortening.
The pathology observed in wbt mutants could arise from direct exacerbation by Wrn and Blm mutation of telomere dysfunction initiated by loss of telomerase, or it might instead reflect the summation of genomewide instability caused by mutation of Wrn and Blm combined with genome instability caused by loss of telomerase. To begin to address these possibilities, we tested aspects of telomere function. A cardinal indicator of telomere dysfunction is end-to-end chromosomal fusion, which becomes more frequent in later-generation Terc−/− mice (15). Fusion levels in metaphase spreads from splenocytes from G1 mice were not elevated for any genotype, but were significantly elevated in G3 and G4 wbt mutants (Fig. (Fig.6A6A and and6C).6C). Neither G3 bt nor wt mutants had significant elevations in fusions, indicating that the Wrn and Blm mutations each contribute to this wbt phenotype.
Telomere dysfunction in wbt triple mutants could reflect a propensity of telomeres, at a given degree of shortening, to malfunction in the absence of WRN and BLM or could reflect an elevated loss of telomere repeat DNA in wbt compared with Terc−/− mutants or both. Flow fluorescence in situ hybridization analysis of bone marrow cells indicated no differences in telomere length among G1 genotypes but revealed a significantly shorter mean cellular telomere length in G3 and G4 wbt samples compared with t samples (Fig. (Fig.6B).6B). Consistent with these measurements, the frequency of chromatid termini without detectable telomere repeat DNA in metaphase chromosome spreads (signal-free ends) was elevated in samples from G3 wbt compared with G3 t samples (Fig. (Fig.6C6C and and6D).6D). Therefore, at least one contributor to the severe phenotypes of late-generation wbt mutants appears to be accentuated loss of telomere repeat DNA in this lineage compared with the t lineage. Furthermore, the occurrence of signal-free ends was not increased in wb mutants, indicating that, at least in the presence of telomerase, the Wrn and Blm mutations do not themselves cause significant telomere repeat loss. This contrasts with a recent demonstration of stochastic loss of complete telomere repeat arrays at chromosome ends in a telomerase-positive human tumor cell line overexpressing a dominant-negative form of WRN (3) and suggests that, in murine splenocytes, the Wrn and Blm mutations either do not cause such complete loss events or that telomerase is sufficient to repair such events.
To address the possibility that the WRN and BLM proteins have important functions in the setting of telomere dysfunction, we tested the phenotypic consequences of Wrn and Blm mutations in mice with telomeres that had been shortened artificially via mutation of the Terc RNA component of telomerase. We found that the Wrn and Blm mutations each contributed to the acceleration, by two to three generations, of several phenotypes characteristic of later-generation Terc mutants, including infertility, apoptosis of intestinal crypt epithelial cells, impaired wound healing, and chromosome end-to-end fusions. These findings are consistent with the accentuated telomere repeat loss observed in the wbt triple mutants compared with Terc−/− single mutants. However, other phenotypes, including reduced body mass at early ages and reduced bone density, have not been observed in Terc−/− mice and so may reflect the recreation in the mouse of pathology seen in Werner or Bloom syndrome: low bone density is found in Werner syndrome, and reduced body mass is found in Bloom syndrome and to a lesser degree in Werner syndrome (10, 12). Consistent with this view, the Blm mutation contributed more than the Wrn mutation to the low body weight of wbt mice. Furthermore, some wbt phenotypes are characteristic of both the human syndromes and late-generation Terc−/− mice, including fertility defects in Werner syndrome and Bloom syndrome, and chronic wounds in Werner syndrome, and so it is unclear if they reflect a simple acceleration of Terc−/− phenotypes or additional interactions between shortened telomeres and the Wrn and Blm mutations.
An important aspect of wbt phenotypes is that several reflect synergistic genetic interactions among the Wrn, Blm, and telomerase mutations: the defects, including those in fertility and testes mass, intestinal apoptosis, chromosome fusion, rates of wound healing, and growth of fibroblast cultures, are more severe in the triple mutants than the sum of these defects in each single mutant. Thus, the mutations are not acting independently, but the three genes function together, possibly through direct action of the WRN and BLM proteins at telomeres. Because G1 mutants were unaffected, the effects of Wrn and Blm mutation in wbt mutants are not caused by loss of telomerase activity per se, but apparently require telomere shortening. Faster loss of telomere repeat DNA and higher levels of chromosome end-to-end fusions in wbt mutants suggest that WRN and BLM might have direct effects at telomeres. Alternatively, telomere dysfunction may synergize with nontelomeric defects caused by Wrn and Blm mutation: DNA damage signals from shortened telomeres might add to damage signals throughout the genome to exceed a threshold that signals cell death or growth arrest, or a reduction in DNA repair capacity caused by telomere dysfunction (55a) might synergize with DNA repair defects caused by Wrn and Blm mutation. Both models may be correct, and the interaction of WRN and BLM with TRF2 and other telomere maintenance proteins, including Ku and DNA-PKcs (37), supports the first idea, while the phenotypes observed in wbt but not Terc−/− mice support the second.
The first idea is also supported by studies from our laboratory investigating how mutation of the SGS1 RecQ family helicase of S. cerevisiae accelerates the rate of senescence of yeast telomerase (tlc1) mutants. We have found that null mutations in a variety of loci required for normal DNA replication, repair, and recombination, including TOP1, MUS81, SLX4, RRM3, and SRS2, have no effect on the rate of senescence of tlc1 mutants (M. Azam, J. Y. Lee, and F. B. Johnson, unpublished results), consistent with published findings that mutations in RAD1, RAD10, and POL3 also do not affect the rate of senescence of telomerase mutants (43). Thus, accelerated senescence in sgs1 tlc1 mutants is unlikely to be a consequence of generic genome instability but rather reflect a specific function of the Sgs1 protein, perhaps at telomeres. Regardless of the precise mechanisms by which the Wrn, Blm, and Terc mutations interact, our findings indicate that the Wrn and Blm mutations can accentuate the pathology associated with telomere shortening, a process associated with natural human aging, and thus indicate that the pathology of Werner and Bloom syndromes may involve telomere dysfunction.
How might telomere defects contribute to premature aging in Werner syndrome? Werner syndrome aging pathology occurs principally in mesodermal tissues, such as bone, dermis, and vascular endothelium, which normally have slow rates of cell turnover and low levels of telomerase activity (13). In contrast, highly proliferative tissues, such as the hematopoietic system and gastrointestinal tract, which have relatively high levels of telomerase, and the central nervous system, which is effectively postmitotic and does not shorten telomeres with age (2), are relatively spared premature aging. Given the data presented here and elsewhere indicating that telomerase compensates for some defects caused by loss of RecQ family proteins (8, 19, 23), the distribution of defects in Werner syndrome could indicate tissues where cell turnover rates are high enough to cause telomere shortening with age but telomerase activity is too low to prevent telomere dysfunction. In contrast, Terc−/− mice have no telomerase activity in any tissue, and so tissues with the highest rate of cell turnover, and thus presumably the fastest telomere shortening, are most affected by combined mutations in Terc, Wrn and Blm. This explains why wbt mice but not humans with Werner and Bloom syndrome develop degenerative pathology in the small intestine: humans with these diseases are presumably replete with telomerase in this organ, protecting them from telomere defects.
This may also explain why premature aging is less pronounced in Bloom syndrome than Werner syndrome, because unlike WRN, which is expressed broadly in many tissues, BLM is preferentially expressed in highly proliferative tissues, including the male germ line and thymic lymphocytes (24, 60). These tissues express substantial levels of telomerase and so may be relatively protected from telomere dysfunction (27, 57). This model cannot fully explain the fertility defects in Werner syndrome and Bloom syndrome, because germ cells express high levels of telomerase, but it is likely that meiotic functions of WRN and BLM contribute to normal fertility (33, 55). Furthermore, cancer can arise in both Werner syndrome and Bloom syndrome in tissues with high levels of telomerase, for example, in the hematopoietic system, but this may reflect principally the widespread genome instability characteristic of these diseases rather than a telomere-specific defect.
If WRN and BLM indeed play direct roles at telomeres, by what mechanisms might they maintain telomere function? First, WRN and BLM might play a role in the recombinational repair of telomeres. This is supported by the high levels of WRN and BLM protein in foci containing telomeric DNA in ALT cells (23, 59), which most likely use recombinational mechanisms to maintain their telomeres (42), by the ability of overexpressed BLM to lengthen telomeres in ALT cells (49), and by the requirement for yeast Sgs1p for the generation of type II survivors of telomerase deletion, which use recombination to maintain telomeres (8, 19, 23). The recent finding that downregulation of RAD51D expression causes telomere shortening and dysfunction in primary murine cells is consistent with a role for recombination in the normal maintenance of telomeres (52).
Second, WRN and BLM could regulate T-loop assembly or disassembly. WRN and BLM interact functionally with TRF2, which catalyzes T-loop formation in vitro (37, 48), and thus might modulate the ability of TRF2 to regulate T-loop assembly. T-loops include a D-loop at their base, and WRN helicase and exonuclease activities catalyze the disassembly of D-loops in vitro (38), and further, the degradation of synthetic T-loops by the WRN exonuclease was recently shown to be stimulated by TRF2 in vitro (31). Third, WRN and BLM may participate in the replication or processing of telomere ends. FEN-1 (flap endonuclease 1), which is required for proper Okazaki fragment processing during lagging-strand DNA synthesis, is strongly stimulated in vitro by purified WRN and BLM (46) and may thus cooperate with WRN and BLM in telomere replication. Together, the proteins might be particularly important for telomere lagging-strand synthesis because of the absence of a “backup” 3′ end at the telomere terminus from which repair synthesis to replace defective Okazaki fragments might otherwise be primed (39). Consistent with this possibility, S. cerevisiae cells with a mutation of the FEN-1 ortholog RAD27 incur high levels of single-stranded DNA principally at telomere ends (39), and shorten their telomeres more rapidly in the setting of genetic inactivation of telomerase (40). Fourth, G quadruplexes, which form at telomeric DNA in vitro and are favored substrates of RecQ helicases (21), might accumulate at shortening telomeres and require WRN or BLM to unwind them. One or more of these mechanisms may be operative, and our mouse models provide useful materials to further determine how WRN and BLM may participate in telomere maintenance.
Finally, we note that although it is clear that human aging is accompanied by telomere shortening in several tissues, it remains uncertain if telomere dysfunction plays a causal role in aging (4, 54). While telomere shortening is one contributor to telomere dysfunction and telomere dysfunction can in turn cause apoptosis or senescence in cultured cells (5), whether telomeres ever become critically short in aged humans remains undetermined. The remarkable similarities between the Werner syndrome phenotype and certain features of natural aging, in combination with our findings and those of others that indicate a role for WRN in the setting of telomere dysfunction, provide circumstantial support for the hypothesis that telomeres may play an important role in the aging process, and the tissues affected by premature aging in Werner syndrome might point to those that are most susceptible to telomere dysfunction during natural aging. Furthermore, because telomere dysfunction contributes to overall genome instability and is itself interpreted by the cell as a form of DNA damage (9, 51), our findings are consistent with the many observations indicating a role for genome maintenance in the promotion of longevity (16, 56).
Findings consistent with ours, that Wrn and Terc mutations cause synergistic phenotypes in late-generation mice, have been reported recently by S. Chang et al. (Nat. Genet. 36:877-882, 2004).
We thank the Center for Neurological and Degenerative Disease Research and the Center for Molecular Studies in Digestive and Liver Disease for histology assistance, the Mouse Phenotyping, Physiology and Metabolism Core for DEXA scans, J. Martin for database assistance, S. Chang for communication of unpublished findings, and R. A. Marciniak and members of the Johnson lab for helpful discussions.
This work was supported by R03 and R55 awards from the NIA and a Paul Beeson Award from AFAR (F.B.J.) and by grants from the NIH, Ellison Medical Foundation, and the Howard and Linda Stern Fund (L.G.).