Several fundamental questions regarding telomeric function and telomeric DNA-binding proteins have been addressed in the studies reviewed above. The central questions of how the cell protects the telomere end and how the cell is able to measure the length of the telomeric tract remain fundamentally unresolved. In this section, we will describe recent work on end-binding proteins as well as the discovery and potential biological significance of t-loops.
Any current model of telomeric protection must take into account the work of de Lange and colleagues (167), who showed that abrogation of TRF2 function results in a series of changes typical of what would be expected were the telomere end no longer protected. Based on these results, one could argue that TRF2 functions to bind and protect the G-strand overhang. However, unpublished data from this and the de Lange laboratory have shown that TRF2 does not bind single-stranded telomeric DNA. Thus, an alternate explanation is required.
Elegant studies conducted by several groups (24,170) have described the mechanism by which the ends of the short multicopy DNA fragments in the macronuclei of Oxytricha nova are protected. A protein has been isolated that has led to a model of telomere end protection based on a single protein species that would bind and sequester the single-strand overhang. Until recently, few examples of single-stranded telomere overhang binding proteins outside of Oxytricha were known. One such example was found in Candida parapsilosis (171), where a single-stranded binding protein was shown to bind a single-stranded overhang of linear mitochondrial DNA. Recently, a single-stranded telomere-binding protein termed Pot1 (protection of telomeres 1) was isolated and characterized in fission yeast and human cells (28). Pot1 has high affinity for single-stranded DNA with the sequence motif of Schizosaccharomyces pombe telomeric repeat or the G-strand overhang in mammalian telomeres. It has been proposed that Pot1 might serve to protect the telomere end from degradation and regulate the ability of telomerase to elongate the telomere. Whether this protein interacts with the other known telom-ere-binding proteins is presently unclear, and how elimination of TRF2 function may alter Pot1 activity has not been suggested. Another mechanism of end protection, including the formation of G-quartets through a fold-back structure of the G-rich strand, has been proposed (20,21). No evidence for G-quartet formation has been provided, and, as above, no relation to TRF2 function has been suggested.
The two conserved features of nearly all eukaryotic telomeres are a long array of repeating six to eight nucleotide units and a single-stranded overhang on one strand. One facet of this structure that has not been discussed in detail is the highly recombinogenic nature of such a DNA template. Studies of homologous recombination catalyzed by RecA protein have shown that a single-stranded DNA terminating in a 3' hydroxyl is highly recombinogenic when presented with a homologous duplex DNA, and that such a single-stranded tail can find its homologous duplex partner rapidly even in the presence of a 100,000-fold excess of non-homologous DNA (172). Given this, DNA in the form of a telomere would be expected rapidly to undergo self-recombination in the presence of proteins having activities like those of RecA and Rad51. During this event, the 3' single-stranded tail would invade the internal duplex segment, which is fully homologous to the tail throughout its length. Consideration of these facts led us to propose that mammalian telomeres might be arranged as giant loops in which the single-stranded overhang would invade the duplex segment somewhere along its length. The biological consequences of this novel structure would be that it could provide a means of hiding the telomere terminus within the internal repeat segments. Such structural folding also provides additional interesting biological properties. In collaboration between our group and the de Lange laboratory, two means of testing this hypothesis have been developed.
In one approach to test this model, a slipped structure polymerase chain reaction (PCR) method (173) was used to generate a model template containing several kilobases of duplex TTAGGG repeats on the end of a plasmid DNA. The 5' ends were resected to create a DNA with the essential features of a mammalian telomere: long duplex repeats and a G-strand 3' overhang. The effects of the dom inant negative TRF2 allele in cells (discussed above) suggest that TRF2 might be involved in the formation of a looped structure that protects the extreme telomere end. Incubation of the model telomeric DNA with purified TRF2 protein led to the observation of DNA molecules arranged into loops, with TRF2 protein at the loop junction (29) (see Fig. 1B). These looped structures were present at high frequency, but only in the presence of TRF2; neither TRF1 nor tankyrase showed any looping activity. The DNA loops were stabilized by covalent cross linking at A/T steps in the DNA using psoralen and ultraviolet (UV) light, providing evidence that the G-rich overhang had undergone strand invasion into the preceding duplex DNA to form a d-loop structure (Fig. 1C). In these studies, TRF2 localized exclusively to the base of the loop (t-loop junction). It remains unclear how TRF2 may function to form these loops; however, detailed analysis has clarified some of the DNA features required for TRF2-mediated t-loop formation. Specifically, a 3' G-strand overhang and the natural junction between the single-strand and doublestrand repeats are required for loop formation (173).
Direct isolation of telomeric looped molecules from human and mouse cells was also undertaken. Beginning with 3 X108 HeLa cells, nuclei were treated with psoralen and UV to stabilize the t-loop structures. The t-loops were then purified by digesting the genomic DNA with four base restriction enzymes and size fractionating the large telomeric fragments away from the small digested genomic fragments using gel filtration. Examination by electron microscopy revealed DNA molecules greater than 10 kb in length, and in several preparations, 30% of these DNA molecules were present as looped structures (see Fig. 1A). The loops were frequently 15-20 kb in size when the DNA was isolated from mouse cells or from a HeLa strain with long telomeres. Key findings in this study (29) were the following: (a) the t-loop molecules were present in human cultured and primary cells and mouse liver cells in abundance; (b) the looped molecules were formed from telomeric DNA, as shown by their binding TRF1 protein; (c) a single-stranded segment, consistent with a d-loop at the loop junction, could be observed; (d) the size of the t-loop molecules was in good agreement with the known size of the telomeric DNA isolated from the same cells, as determined by Southern blotting; and (e) to observe the t-loops, the telom-eric DNA had to be cross linked in the form of chromatin in situ.
Recently, t-loops have been found to be present not only at human and mouse telomeres, but also at the termini of the micronuclear chromosomes of Oxytricha nova (30) and at the telomeres of Trypanosoma brucei minichromosomes (20). Further, telomeres in Saccharomyces cerevisiae appear to form folded-back structures through protein-protein interactions (174-177). Thus, telomeric looping may be a common theme in telomeric architecture.
The discovery that telomeres are arranged into large duplex loops offers insights into telomeric function. For example, the ability of the cell to distinguish between telomere ends and double-strand breaks has never been explained. Ending chromosomal DNA in loops would provide a simple means of disguising them from the double-strand break recognition and repair factors. Loss of TRF2 had been shown to cause loss of the 3' overhang and apoptosis (167,168). The t-loop model suggests that loss of the overhang may be due to the inability to protect the telomere terminus in a t-loop structure without being TRF2 present.
VIII. REPLICATION INITIATION FROM A CLEAVED T-LOOP JUNCTION: PLACING AN ORIGIN AT THE END
If an origin of replication in the middle of a chromosome fails to fire, or if replication from that origin stalls, bidirectional forks moving in from the left and right should complete the synthesis of both strands. On the other hand, if an origin close to a telomere fails to fire or stalls, then the chromosome may not replicate to its very end and will be lost at metaphase. Thus, mechanisms for ensuring that chromosomes replicate to their full ends may exist. The t-loop formation provides a possible means for initiating replication within the telomere (Fig. 7). Once the G-strand overhang has strand invaded to form a d-loop, this structure could be cleaved by Holliday junction-resolving enzymes. This would result in the loss of the TTAGGG overhang from the end of the telomere and would leave
Figure 7 Initiation of replication from a t-loop. A t-loop structure may serve as a site for initiation of replication following cleavage by Holliday junction-resolving enzymes. These enzymes may cleave the d-loop resulting in the loss of the TTAGGG overhang from the end of the telomere. Replication primers would be left annealed to the C-rich and G-rich strands and may allow replication of the telomeric DNA.
behind a replication primer annealed to the C-rich strand. Recent data from this laboratory suggest that t-loops formed in vitro by TRF2 may involve not only the G-strand overhang, but also a portion of the C-rich strand as well (173). Hence, cleavage would leave primers on both strands. Bidirectional replication also could result from priming on the G-rich strand once synthesis on the other strand had begun. Such a cleavage event also could provide an alternative mechanism for progressive telomeric shortening by increments of ~150 nt per replication cycle. Although this model has not been suggested for telomeres, it has a general parallel in the late phase of bacteriophage lambda and T4 replication (178-180).
Was this article helpful?