INTRODUCTION

Top Abstract Introduction Materials and methods Results Discussion References

Telomeres are the nucleoprotein complexes that make up the ends of eukaryotic chromosomes. In the majority of organisms examined, telomeres consist of short repeated DNA sequences and their associated sequence-specific binding proteins (reviewed in reference 11). In the yeast Saccharomyces cerevisiae, telomeres consist of the heterogeneous repeat (TG)_1-6TG_2-3, abbreviated as TG_1-3, which contains irregularly spaced binding sites for the major yeast telomere binding protein Rap1p (reviewed in reference 54). The length of the TG_1-3 repeats is regulated, being maintained within a small range of 325 ± 75 bp (40). Telomeres can be formed on linear yeast plasmids by transformation followed by the addition of TG_1-3 repeats. Sequencing of these linear plasmid repeats has shown that the precise TG_1-3 sequences of independently formed telomeres are different (48). When individual telomeres derived from the same formation event are isolated from different cells, they have different TG_1-3 sequences for the half of the TG_1-3 tract nearest the chromosome end. This sequence heterogeneity is thought to arise from telomeres going through cycles of telomere shortening followed by the addition of TG_1-3 repeats with a different sequence (48). This shortening and lengthening by the addition of different TG_1-3 sequences probably also occurs with chromosomal telomeres because the lengths of individual telomeres vary between different cell lineages (43). A direct consequence of this sequence heterogeneity is that the spacing between individual Rap1p molecules bound to these repeats varies between different telomeres, and so the heterogeneity of the telomere DNA sequences causes heterogeneity in telomere chromatin structure.

The fact that telomere length varies as cells grow but still remains within a specific length range has led to the hypothesis that telomere length is regulated by balancing lengthening and shortening mechanisms (11). Presumably, telomere length is somehow measured, and depending on the result, telomere lengthening or telomere shortening occurs. Telomere lengthening occurs primarily via telomerase, a cellular reverse transcriptase with its own RNA template (11, 24), and can also occur through recombination-dependent mechanisms (36). Telomere shortening occurs by incomplete replication of the 5' chromosome end, owing to removal of the RNA primer for DNA synthesis (reviewed in reference 54), and by active degradation (6, 49) and deletion mechanisms (23). Similar processes occur in other organisms, including humans (11, 28, 32). How telomere length is measured and how these processes are controlled are not yet understood.

Telomere binding proteins play an important role in regulating telomere length. The major yeast telomere binding protein Rap1p can be divided into at least two domains, a large DNA binding domain (amino acids 361 to 596 [16]) and a large C-terminal domain (amino acids 600 to 827) which contains multiple subdomains involved in telomere length control, transcriptional silencing, and transcriptional activation (5, 12, 20, 26, 45). Rap1p is known to play a negative regulatory role by inhibiting telomere elongation. Missense or deletion mutations in the Rap1p C terminus cause yeast telomere length to increase (20, 21, 25, 45). These rap1 mutations eliminate interactions between Rap1p and other proteins important for telomere length control (13, 52). Overproduction of the Rap1p C-terminal domain in vivo causes chromosomal telomeres to lengthen (5, 12, 41, 51), presumably because Rap1p-interacting factors are titrated away from telomeres. Experiments with the Rap1p homolog of Kluyveromyces lactis indicate that the C termini of Rap1p molecules bound to the very ends of the chromosome interact with proteins that inhibit telomere elongation, presumably by inhibiting the access of telomerase to the chromosome terminus (19). These data have led to the hypothesis that the Rap1p C terminus and proteins bound to it inhibit telomere lengthening (11).

The Rap1p C terminus can also play a positive role in telomere elongation. The Rap1p C terminus can stimulate the formation of chromosomal telomeres in a transformation assay, a model system for the elongation of short telomeres (38). As with the inhibition of telomere lengthening, the ability of the Rap1p C terminus to stimulate telomere elongation most likely involves other proteins that bind to Rap1p.

Several factors that interact with the Rap1p C terminus and regulate telomere length and transcriptional silencing have been identified. These factors include Rif1p, Rif2p, and Sir3p (13, 33, 52). Sir3p forms a complex with Sir2p, Sir4p, and chromatin to silence transcription of nearby genes, a phenomenon known as telomere position effect (14, 15, 27, 39, 44). Overproduction of Rif1p and Rif2p causes telomeres to shorten, deletion of either gene causes telomeres to lengthen by ~200 bp, and simultaneous deletion of both genes causes telomere length to increase from ~325 bp to 2 to 4 kb of TG_1-3 (52). The rif1 rif2 double-mutant phenotype is identical to that of rap1^t mutants which lack the Rap1p C terminus (20). The 2- to 4-kb TG_1-3 repeats in rap1^t and rif1 rif2 cells are thought to arise by telomere elongation occurring in an unregulated fashion (20, 21). Thus, Rif1p and Rif2p are important components for telomere length regulation. Recent work has shown that yeast normally measure telomere length by counting Rap1p molecules (29), and Rif1p and Rif2p probably play essential roles in this process. However, the mechanism by which Rap1p molecules are counted is unknown.

In measuring telomere length, the cell must determine where the telomere starts, where it ends, and how many Rap1p molecules are present in between. How the beginning and the end of a telomeric TG_1-3 tract are demarcated is unknown. While in vitro studies show that Rap1p binds to sites present at an average of about 1 per 18 bp, site spacing is irregular and some sites overlap (7). This variable placement of Rap1p binding sites means that these telomeric Rap1p molecules are not presented to the cell in a uniform manner. Thus, it is unknown whether all of these Rap1p molecules are used to measure telomere length in vivo.

To investigate the mechanism of telomere length regulation, we constructed a series of synthetic telomeres with additional Rap1p molecules internal to the terminal TG_1-3 repeats and determined their effects on telomere length. The goal of this approach was to mimic a portion of the internal telomere nucleoprotein complex and determine how changes in this structure alter length regulation. If the internal sequences are detected by the cell as part of the telomere, the terminal TG_1-3 tract will be elongated to a length less than that of a natural telomere. We constructed synthetic telomeres that separated adjacent TG_1-3 tracts with different lengths of nontelomeric DNA or that contained regularly spaced arrays of Rap1p sites internal to the terminal TG_1-3 tract. As the separation between adjacent TG_1-3 tracts increased in size from 38 to 50 bp, cells counted smaller portions of the internal TG_1-3 tract as part of the TG_1-3 tract. Cells could not accommodate a 138-bp disruption; the internal sequences were not counted, and a new telomere-nontelomere junction was established. Placing tandem arrays of regularly spaced Rap1p binding sites just internal to the TG_1-3 repeats showed that yeast cells can accommodate different spacings of 13 to 35 bp between consecutive Rap1p sites while measuring telomere length and that cells equated one Rap1p molecule with about 19 bp of TG_1-3 DNA in vivo. These results indicate that all of the Rap1p molecules at the telomere-nontelomere junction participate in telomere length regulation. The internal tandem Rap1p sites had the unexpected effect of eliminating the variation in average telomere lengths normally seen between independently formed telomeres. These data suggest that the boundaries of the telomere, i.e., the chromosome terminus and the junction between the Rap1p-containing and non-Rap1p-containing chromatin (the telomere-nontelomere junction), communicate with one another when telomere length is measured. We therefore propose that telomere length measurement involves forming a folded chromatin structure that is dependent on the presence of a threshold number of telomeric Rap1p molecules. This folded structure links the chromosome terminus and the telomere-nontelomere junction and inhibits telomere lengthening.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and methods Results Discussion References

Strains. Recombinant DNA manipulations were in Escherichia coli MC1066 (r m⁺ pyrF::Tn5 trpC leuB). S. cerevisiae strains used were KR36-6L (MATa ade2-1 or 101 ade8-18 ura3-52 trp11 leu2-RC his3) (40) and the gal4 strain YM708 (MAT ade2-101 ura3-52 trp1-901 his3-200 lys2-801 LEU2 can^R gal4-542; from Mark Johnston).

Plasmids. Plasmid YIpADH35 contains ADH4-URA3-TG_1-3 sequences where the 29-bp segments of TG_1-3 sequences consist of the oligonucleotides GATCCGGGTGTGTGGGTGTGTGGGTGTGGGTGTGC and GGCCGCACACCCACACCCACACACCCACACACCCG in pBR322 as described elsewhere (38) (boldfaced and underlined bases denote Rap1p binding sites). The HincII-NotI fragment bearing the 256 bp of TG_1-3 was isolated from pCT300 (256 bp of TG_1-3 in pVZ-1; from K. Runge, R. Wellinger, J. Wright, and V. Zakian), filled in, and then cloned into the filled-in BamHI site of YIpADH35 to give YIpADH256-50 with the orientation of the TG_1-3 repeats the same as the terminal 29-bp TG_1-3 repeats (see Fig. 1B and C). This plasmid was previously called YIpADH-275 (38). This 256-bp tract was derived from a ~350-bp tract of TG_1-3 sequences (as determined by Southern blotting) cloned by circularization of a yeast linear plasmid that was directly transformed into yeast (40). Upon rescuing the plasmid into E. coli, some of the TG_1-3 sequences were deleted. Sequencing of multiple YIpADH256-XX constructions gave the same sequence for the 256-bp insert (Fig. 1C), showing it was stable in bacteria. YIpADH256-50 has a 50-bp polylinker spacer between the terminal 29 bp of TG_1-3 and 256 bp of TG_1-3. YIpADH652-42 has the 256 bp of TG_1-3 repeats in the opposite orientation to telomere with a 42-bp polylinker spacer between the terminal 29-bp TG_1-3 and 256-bp TG_1-3.

Telomere formation and integration. Digestion of all of the YIpADH plasmids with SalI and NotI releases a SalI-ADH4-URA3-insert-TG_1-3-NotI fragment which can replace the left telomere of chromosome VII. Transformations using 5 µg of SalI- and NotI-digested plasmid into S. cerevisiae KR36-6L or YM708 were performed as described previously (38). The relative efficiencies of telomere formation previously noted (38) were confirmed in these experiments. In most cases, 100 to 200 transformants of each type were picked to a master plate lacking uracil, grown overnight at 30°C, replica plated to a plate containing uracil, grown overnight, and assayed for telomere position effect by testing for growth on YC-Ura and 5-fluoro-orotic acid plates. Subsequently, several transformants of each type predicted to have a URA3 telomere were then used for Southern analysis using a PCR fragment equivalent to the StuI-NsiI 3' fragment of URA3 as the probe. Integration of YIpADH256-50 at internal loci (URA3) does not give detectable levels of telomere position effect (data not shown).

DNA isolation and Southern blotting. Genomic yeast DNA was isolated according to the protocol of V. Schulz (42a). Briefly, stationary-phase cells (5 ml) were washed with water and resuspended in 250 µl of lysis buffer (100 mM Tris-HCl [pH 8.0], 50 mM EDTA, 1% sodium dodecyl sulfate [SDS]), and acid-washed glass beads (Sigma type V; diameter of 400 to 500 µm) were added to the 0.5-ml mark. The tube was shaken for 10 min in a Vortex-Genie (VWR Scientific) with the shaker attachment at top speed, and 150 µl of 7.5 M ammonium acetate was then added. The tube was heated at 65°C for 15 min and placed on ice for 15 min, and then 500 µl of CHCl₃ was added. After mixing, the tube was spun for 8 min at top speed in a microcentrifuge. The aqueous phase was added to a 1.5-ml tube with 0.5 ml of room temperature isopropanol and spun at room temperature for 5 min. The DNA pellet was washed with 70% ethanol and resuspended in 50 µl of TE (10 mM Tris-HCl [pH 8.0], 1 mM EDTA). To this preparation, 1.5 µl of RNase A (5 mg/ml) was added, and the mixture was incubated at 37°C for 15 min; this was followed by the addition of 1.5 µl of proteinase K (10 mg/ml) and incubation at 37°C for another 15 min. The solution was brought to 300 µl with water, extracted once with an equal volume of phenol-CHCl₃ (4:1), and extracted once with CHCl₃. The DNA was precipitated by adding 150 µl of 7.5 M ammonium acetate (pH 7) and 1 ml of 95% ethanol at 20°C, spun, and resuspended in 50 µl of TE. Five microliters was digested for each gel lane. Southern analysis was done as described in reference 42. Prehybridization solution contained 6× SSC (1× SSC is 0.15 M NaCl plus 0.015 M sodium citrate), 5× Denhardt's solution, 0.5% SDS, and 100 µg of denatured, fragmented salmon sperm DNA per ml, and prehybridization was at 68°C for 2 to 3 h. Denatured probe was added to the filter in hybridization solution (6× SSC, 0.5% SDS, 100 µg of denatured salmon sperm DNA per ml), and the filter was hybridized overnight in a hybridization oven at 68°C. The filters were washed as described elsewhere (42).

Telomere length measurement. In the Southern blots used for these analyses, the molecular size standards at 1.636 and 0.517 kb were used to construct the standard curve for migration versus molecular weight and then checked against the migration of the 1.018-kb standard. In all cases, the derived molecular size of the 1.018-kb standard was within 8 bp of its actual molecular size. In addition, the molecular size of the chromosomal URA3 StuI band was within ±15 bp in these blots. The amount of TG_1-3 equivalent to the TEF35-6, TEF18-6, or TEF13-6 tandem arrays was determined by subtracting the size of the YIpADHTEFXX-6 StuI fragment from the YIpADH control with the same insert size (e.g., YIpADH210 size  YIpADHTEF35-6 size). For example, since YIpADH210 contains an insert which is not counted as TG_1-3 whereas YIpADHTEF35-6 contains an insert that is equivalent to TG_1-3, this difference gives the "equivalent TG_1-3 length" of the TEF35-6 tandem array. Two advantages of this approach were that (i) the minor differences in migration for the YIpADHTEF insert also occurred for the YIpADH insert on the same blot, and so deviations in gel migration for these terminal restriction fragments on an individual blot are similar, and (ii) only one comparison is performed, and so errors inherent in each measurement are added together only once. In addition, each telomere was also cleaved with StuI plus BamHI to verify that the inserts were still present in vivo and of the correct size. For YIpADHTEF13-6, the YIpADH88 telomere was used as a control. Corrections for the 4- to 6-bp differences between the insert sizes of the YIpADH and YIpADHTEF telomeres were included in these calculations. For Fig. 3C, length measurements were from the blots in Fig. 3A and B and two other blots similar to that in Fig. 2B that ran more uniformly (not shown).

PCR. The terminal portions of the 26- and 38-bp spacers between the 256- and 29-bp TG_1-3 sequences were detected by a hot-start PCR method with the 5' URA3 StuI and 3' polylinker primers (Table 1). For this purpose, the primer mix (50 pmol of both primers in 50 µl of H₂O) was heated at 94°C for 10 min, and the remaining components (~2 µg of genomic DNA plus deoxynucleoside triphosphates, buffer components and Taq DNA polymerase in 50 µl) were separately heated at 94°C for 2 min. After these two reaction mixtures were combined, the mixture was topped with 50 µl of mineral oil and PCR was performed by 5' denaturation at 94°C for 1 cycle, then 5 cycles of 94°C-1 min, 60°C-1 min, and 72°C-1 min, followed by 30 cycles of 94°C-1 min, 55°C-1 min, and 72°C-1 min, and ending with 72°C-10 min.

	RESULTS

Top Abstract Introduction Materials and methods Results Discussion References

Model systems for altering telomere length regulation. Two classes of synthetic telomeres were constructed to investigate the mechanism of telomere length regulation. The first class involved separating an internal 256-bp tract of TG_1-3 repeats from a terminal 29-bp TG_1-3 tract with a nontelomeric spacer of various lengths (Fig. 1, the YIpADH256-XX and YIpADH652-42 telomeres). These nontelomeric spacers contained no Rap1p binding sites. After transformation into yeast and conversion of the synthetic telomere to a chromosomal telomere (Fig. 1A), the 29-bp TG_1-3 tract will be elongated to 325 ± 75 bp of TG_1-3 (40) if the internal tract is not seen as part of the telomere (i.e., not counted as telomeric TG_1-3). However, if the internal tract is seen as part of the telomere, the 29-bp TG_1-3 tract will not be elongated as much. These spacer telomeres allowed us to determine how a disruption in the continuous TG_1-3 tract altered length measurement. These telomeres also tested the minimum size of the disruption that functionally separated the internal and terminal TG_1-3 sequences to form a new internal boundary for the telomere. The orientation dependence of the internal TG_1-3 tract on telomere length measurement was also examined (Fig. 1, YIpADH652-42).

View larger version (40K): [in this window] [in a new window]   FIG. 1.   Introduction of synthetic telomeres into yeast. (A) Replacement of the left telomere of chromosome VII by homologous integration at the ADH4 locus, using a synthetic telomere adjacent to URA3. Different inserts, shown in panel B, were cloned between URA3 and the TG1-3 sequences to attempt to alter telomere length regulation. (B) The synthetic telomeres used in this work. The nontelomeric spacers between the TG1-3 tracts are described in Materials and Methods. The orientation of the TG1-3 repeats is indicated by arrowheads. YIpADH652-42 has the internal 256-bp TG1-3 tract in the reverse orientation relative to the terminal 29-bp tract. The specific YIpADHTEF and YIpADH telomeres are presented in Fig. 5. (C) Sequences of the TG1-3 repeats and spacer in the YIpADH256-50 construction. The sites used to construct the plasmids in panel B are indicated. The TG1-3 repeats are from positions 54 to 309 in the 5' KpnI site, and the single A in the TG1-3 repeats is at position 98.

View larger version (54K): [in this window] [in a new window]   FIG. 2.   Some of the internal TG1-3 sequences are considered part of the telomere across a 50-bp nontelomeric spacer. (A) Predicted sizes of the telomere restriction fragments on Southern blots. Telomere length, i.e., the length of the elongated terminal TG1-3 tract, was determined by subtracting the StuI-BamHI fragment length from the StuI fragment length measured on the same blot. Analysis on the same blot was important for giving reproducible and accurate length determination for different control telomeres. With lanes 14 and 15 of B as an example, the length of the elongated terminal TG1-3 sequences (the ? in the diagram) is 150 bp. (B) Representative genomic DNAs from cells bearing the YIpADH256-26, -38, and -50 telomeres and the YIpADH652-42 telomere were cleaved with StuI or StuI plus BamHI and analyzed by Southern blotting using the URA3 fragment in panel A as probe. All telomeres were formed in yeast strain YM708. The 0 spacer is YIpADH35. Each pair of lanes (indicated by a bar under the number) represents an individual transformant. Lane 1, YM708 with no synthetic telomere; lanes 10 and 11, a YIpADH256-50 transformant with short telomeres; lanes 14 and 15, a YIpADH256-50 transformant with long telomeres; lanes 12 and 13, a YIpADH256-50 transformant with short and long telomeres (mixed telomeres). Note that the YIpADH652-42 construction places the BamHI site closer to URA3 (Fig. 1B). Telomere fragment sizes were determined as described in Materials and Methods.

An internal 256-bp TG_1-3 tract can be partially counted as part of the telomere in both orientations and across a 50-bp spacer. As described above, the first class of synthetic telomeres contained an internal 256-bp tract of TG_1-3 sequences (Fig. 1B and C). The synthetic telomeres containing this sequence were named YIpADH256-XX, where XX is the 26-, 38-, or 50-bp nontelomeric spacer between the 256-bp tract and the terminal 29-bp tract (Fig. 1B). In these constructs, the terminal and internal TG_1-3 tracts were in the same orientation. In addition, a telomere with the 256-bp tract in the opposite orientation followed by a 42-bp spacer and the terminal 29 bp TG_1-3 tract, YIpADH652-42 (Fig. 1B), was also constructed.

The internal 256-bp TG_1-3 tract is not counted as part of the telomere across a 138-bp spacer. Because the 50-bp spacer allowed only ~130 bp of the internal 256 bp TG_1-3 tract to be counted as part of the telomere, larger spacers were constructed to determine how large a disruption was required to abrogate counting of the internal TG_1-3 sequence as part of the TG_1-3 tract and establish a new internal boundary for the telomere (i.e., a telomere-nontelomere junction). Nontelomeric DNA spacers of 138 and 266 bp, which did not contain Rap1p sites, were constructed (Fig. 1B and C) and then introduced into yeast.

View larger version (44K): [in this window] [in a new window]   FIG. 3.   A 138-bp nontelomeric spacer establishes a new telomere-nontelomere junction. (A) Individual YM708 transformants bearing the YIpADH256-138 telomere were analyzed as for Fig. 2. Lanes 6 and 7, a YIpADH256-138 transformant with long telomeres; lanes 10 to 13, transformants with short telomeres; lanes 8 and 9 and 14 to 19, transformants with short and long telomeres. The transformant in lanes 8 and 9 had lost the BamHI site in the long telomeres and was not used for the analysis in panel C. The chevron-headed arrow indicates the long telomeres, the filled arrowhead indicates the StuI-BamHI long telomere fragment, and the hollow arrowhead indicates the short telomeres. Lane 1 contains DNA from the untransformed strain. (B) Individual transformants bearing the YIpADH256-266 telomere were analyzed as for Fig. 2. Lanes 10, 11, 14, 15, and 18 show transformants with short telomeres, while lanes 6 to 9, 12, 13, 16, and 17 show transformants with short and long telomeres. Arrows are used as in panel A. Lane 1 contains DNA from the untransformed strain. (C) The amount of the internal 256-bp tract counted as part of the elongated TG1-3 tract plotted against the length of the nontelomeric spacer. The gray box represents the length heterogeneity due to normal variation and was calculated by using seven independent YIpADH35 telomeres. The amount of internal TG1-3 counted was determined by subtracting the average length of the terminal TG1-3 tract distal to the BamHI site for different nontelomeric spacer telomeres (the ? in Fig. 2A) from the average modal length of the TG1-3 tract for the control YIpADH35 telomere (Fig. 2A). The standard error of the telomere length measurements for each spacer telomere was ±10% except for YIpADH652-42 (±53%).

Telomere processing may not occur immediately after integration of the synthetic telomere. Because some single transformants contained mixed telomeres (Fig. 2B, 3A, and 3B), it appeared that a single transformation event had given rise to a colony containing individual cells with either short or long telomeres. One hypothesis to explain these data is that integration of the synthetic telomere construct and its conversion to a chromosomal telomere by the cellular machinery did not occur in the same cell cycle (Fig. 4A). Yeast telomeres as short as ~30 bp are sufficient for chromosomal maintenance since tel1 hdf1 cells have telomeres in this size range and are viable (37). After the synthetic telomere was integrated, the cell may have divided to produce two progeny before altering the length of the terminal 29-bp TG_1-3 tract. In these two progeny cells, one cell may have elongated the 29-bp tract to give rise to the long telomeres, while the other cell may have degraded the terminal TG_1-3 tract (and thus the BamHI site in the spacer) and formed a telomere by elongating the internal 256-bp TG_1-3 tract to give rise to the short telomeres. In this model, the short and long telomeres exist in separate cells within the colony and are stable once formed after the first few cell divisions. A second, alternative model is that the telomere is immediately formed after integration and long telomeres undergo deletion events during colony growth to form the short telomeres. In the first model, single-cell subclones derived from the mixed telomere colonies should have either short telomeres or long telomeres but not both. Long telomeres should be stable to subsequent growth, and so transformants with long telomeres should yield subclones with only long telomeres (Fig. 4A). In the second model, where the long telomeres undergo deletion events, cells with long telomeres should frequently give rise to subclones that have mixed telomeres.

View larger version (50K): [in this window] [in a new window]   FIG. 4.   The transformants with mixed telomeres are colonies of cells with short or long telomeres. (A) Hypothesis for the formation of single transformant colonies bearing mixed telomeres and the predicted outcomes. After integration of the construction, the cell replicates the telomere and divides prior to either elongating the terminal 29 bp TG1-3 tract or deleting the terminal 29-bp tract and elongating the internal 256-bp tract. After these events, the telomeres are stably maintained in separate cells within the same colony. Subclones derived from single cells from these individual transformants bearing two types of telomeres should have either short or long telomeres but never both. (B) Genomic DNAs from single colonies derived from a single YIpADH256-50 transformant bearing either mixed telomeres (lanes 2 to 11) or only long telomeres (lanes 12 to 16). DNAs were cleaved with PstI, which cleaves in URA3 (see panel A) and analyzed as for Fig. 2. Lane 1 is DNA from a cell transformed with YIpADH35. (C) Genomic DNAs from a single colony derived from a single YIpADH256-138 transformant bearing long and short telomeres (Fig. 3A, lane 14) cleaved with StuI and analyzed as for Fig. 2 (lanes 2 to 15). Lane 1 is DNA from a cell transformed with YIpADH35; lanes 16 to 18 are single-cell subclones of the short telomere transformant in Fig. 3A, lane 12. In panels B and C, the chevron-headed arrows indicate long telomeres that retain the nontelomeric spacer and hollow arrows indicate short telomeres that have lost the spacer. All transformants were in strain YM708.

Cells measure one Rap1p molecule as 19 bp of TG_1-3 sequences in vivo. The irregular spacing of Rap1p sites within yeast telomere sequences makes it unclear whether all Rap1p sites are counted to monitor telomere length, and no experiments have yet determined whether all Rap1p molecules are counted in vivo. This information is particularly important because a model of yeast telomere length regulation predicts that several Rap1p molecules at the telomere-nontelomere junction do not participate in length regulation (3, 8). To determine how many Rap1p molecules are counted to measure telomere length, we varied the spacing between six tandem Rap1p sites at the telomere-nontelomere junction and determined the effect on telomere length. The fact that a 26- or 38-bp nontelomeric spacer does not disrupt counting (Fig. 2 and 3) indicates that the cell can accommodate variable spacing between Rap1p sites. This consideration indicates that if six Rap1p molecules are arranged in regular arrays with different spacing, then they should be counted as a constant length of TG_1-3 sequences even though the total DNA length of the Rap1p binding site array varies.

View larger version (59K): [in this window] [in a new window]   FIG. 5.   Cells count one Rap1p molecule as ~19 bp of TG1-3. (A) Telomere constructions contained either phased arrays of Rap1p sites internal to the 29 bp TG1-3 tract or the same-size fragment of  DNA. The orientation of the Rap1p site array is indicated by the arrowheads as in Fig. 1. The locations of the BamHI site and the URA3 fragment used to probe the Southern blots are shown. (B) Genomic Southern analysis of telomeres bearing six Rap1p sites spaced one site every 18 bp in strain KR36-6L. DNAs were digested with StuI or StuI plus BamHI and analyzed by Southern blotting using the URA3 probe shown in panel A. Two YIpADH108 telomeres were included as negative controls for telomeres bearing inserts of equal size with no Rap1p sites. The leftmost lane of B and C is DNA from the untransformed KR36-6L strain; 35 denotes YIpADH35 (Fig. 1B). (C) Genomic Southern analysis of telomeres bearing six Rap1p sites spaced one site every 35 bp in strain KR36-6L analyzed as for panel B (reprinted from reference 38 with permission). A YIpADH210 telomere was included as a negative control for an insert of equal size with no Rap1p sites. (D) Genomic Southern analysis of telomeres in strain KR36-6L bearing six Rap1p sites spaced one site every 35 bp. All DNAs were digested with StuI. All lanes are from the same gel. (E) Genomic Southern analysis of telomeres in strain KR36-6L bearing six Rap1p sites spaced one site every 13 bp. A YIpADH88 telomere was included as a negative control for an insert of nearly equal size with no Rap1p sites.

Regularly spaced arrays of internal Rap1p sites eliminate telomere length heterogeneity between individual transformants. The terminal chromosome restriction fragment gives a dispersed band on Southern blots because the length of a specific telomere fragment in individual cells differs. An internal restriction fragment gives a sharp band because it is the same length in all cells (e.g., the telomeric StuI restriction fragments versus the StuI-BamHI fragments in Fig. 5). This terminal restriction fragment length heterogeneity is due to sequence differences between the terminal TG_1-3 tracts of different telomeres (48). An additional heterogeneity is observed between telomeres formed in independent transformants, because when new synthetic telomeres are formed, the sequence of all but the first 11 bp of TG_1-3 added is random (18, 48) (Fig. 6A).

View larger version (53K): [in this window] [in a new window]   FIG. 6.   Phased arrays of Rap1p sites eliminate the length variation between transformants. (A) Structures of steady-state telomeres in independent transformants derived from the YIpADH35 (35) and YIpADHTEF (TEF) constructions. When telomeres are formed with YIpADH35, 90% of the telomere consists of newly synthesized TG1-3 repeats. Since the TG1-3 sequences added after the first 11 bp are random (18), each of the independent YIpADH35 transformants will have a different telomere sequence near the telomere-nontelomere junction. In contrast, all of the YIpADHTEF telomeres retain the same nontelomeric Rap1p sites at the junction. The terminal 120 to 150 bp of TG1-3 sequences for both the YIpADH35 and YIpADHTEF telomeres are randomized during cell growth (48), and so the only differences between the YIpADH35 and YIpADHTEF telomeres are the sequences near their telomere-nontelomere junctions. (B) Genomic DNAs digested with StuI from seven independent YIpADH35 transformants, five independent YIpADHTEF35-6TG transformants, and four independent YIpADHTEF35-6CA transformants (all in strain KR36-6L) were analyzed by Southern blotting as for Fig. 5. All lanes are from the same gel. The length of each telomere was measured at the point indicated by the small white bar in each lane, which is the point of most intense hybridization (the modal telomere length). The lengths of the YIpADH35 telomeres varied over a ~60-bp range, while the lengths of most of the YIpADHTEF35-6 telomeres were within a ~10-bp range (see text). (C) The same DNAs as in panel B were digested with StuI and BamHI to show that the length heterogeneity between the independently formed telomeres was due to different lengths of the terminal TG1-3 sequences.

	DISCUSSION

Top Abstract Introduction Materials and methods Results Discussion References

The placement of TG_1-3 sequences or nontelomeric Rap1p sites adjacent to the telomeric TG_1-3 tract caused yeast cells to maintain the TG_1-3 tract at a shorter equilibrium telomere length. These results indicated that yeast measure telomere length by counting Rap1p molecules instead of TG_1-3 sequences. Yeast detected internal TG_1-3 sequences as part of the telomere, or counted the internal sequences, across a 50-bp spacer. However, only part of the internal TG_1-3 tract was counted across a spacer larger than 38 bp, and a 138-bp nontelomeric spacer was sufficient to establish a new internal boundary and separate the telomere from the subtelomeric sequences. Cells counted each of the six telomeric Rap1p molecules in the TEF telomeres as ~19 bp of TG_1-3 in vivo. Interestingly, the conserved spacing of Rap1p molecules near the telomerenontelomere junction in different transformants eliminated the modal length heterogeneity commonly observed between independent telomere formation events.

Cells transformed with the YIpADH256-XX telomere formed colonies containing individual cells bearing either short or long telomeres (Fig. 2 to 4), and these telomere lengths were stably maintained over many divisions (Fig. 4; data not shown; reference 38). The mixed transformants occurred at high frequency and with all nontelomeric spacers, suggesting that the events that gave rise to them are normal cellular processes. A hypothesis consistent with these findings is that after integration, the original telomere construct was transiently stable for one or more cell divisions before either the terminal 29-bp TG_1-3 tract was elongated or the terminal 29-bp TG_1-3 tract was deleted and the internal 256-bp tract was lengthened (see Results and Fig. 4A). The high frequency of mixed transformants in the YIpADH256-138 and -266 telomeres (Fig. 3A and B) suggests that most integration events can give rise to a mixed transformant. Therefore, these results strongly suggest that telomere formation was not completed in the first cell cycle after integration.

The phased array of Rap1p sites at the telomere-nontelomere junction in the YIpADHTEF telomeres showed that the spacing between the six internal Rap1p molecules could be varied widely and still be counted as nearly the same length of TG_1-3 sequences, such that 78 to 210 bp of nontelomeric Rap1p sites was counted as 90 to 128 bp of TG_1-3. The results presented here indicate that yeast equated each Rap1p molecule with ~19 bp of TG_1-3 (Table 2). In vitro experiments by others have shown that Rap1p bound to cloned telomeric sequences at a frequency of slightly greater than one molecule per 18 bp of TG_1-3 (7). The fact that the frequency of Rap1p binding sites in vitro and the equivalence of Rap1p for a given amount of TG_1-3 sequences in vivo is so close indicates all of the Rap1p molecules near the telomere DNA-nontelomere DNA junction participate in telomere length measurement.

All three arrays eliminated the modal length heterogeneity between individual transformants (Fig. 5 and 6). Thus, all three arrays had similar overall effects on the cellular telomere length measurement apparatus in that the most internal Rap1p molecules can affect the lengthening and shortening reactions that occur at the chromosome end. These data indicate that the chromosome end and telomere-nontelomere junction must somehow communicate when telomere length is measured.

While this work was in progress, Marcand et al. showed that internal TG_1-3 sequences or fusion proteins containing the Gal4p DNA binding domain and the Rap1p C terminus tethered internal to the terminal TG_1-3 tract could cause telomeres to be maintained at a shorter equilibrium length (29). They concluded that telomere length is regulated by a negative feedback mechanism that counts Rap1p molecules. Their data and ours are consistent with a wide variety of previous genetic experiments that showed that titration of telomere components from chromosome ends altered telomere length regulation (40) and that the Rap1p C terminus tethers these components to telomeres (5, 13, 19, 20, 25, 51, 52). The use of the Gal4p-Rap1p fusion by Marcand et al. required that only low levels of the protein be produced because overexpression of the Rap1p C terminus causes telomere lengthening (5, 12), most likely by titrating away negative telomere length regulatory components (13, 52). Otherwise, telomere lengthening caused by overproduction of the Rap1p C terminus would mask the telomere shortening caused by tethering Rap1p C termini to telomeric sites. This low level of fusion protein raised the possibility that the Gal4p binding sites were not completely occupied (footnote 18 in reference 29), preventing quantitative analysis of the number of bp of TG_1-3 that the cell equates with one Rap1p. In contrast, our approach was quantitative because the nontelomeric Rap1p sites in this work did not perturb the cellular levels of telomere length regulatory components. Thus, the length change of these telomeres (Fig. 5) was due only to the arrangement of Rap1p molecules at the telomere-nontelomere junction. In addition, comparing the modal lengths of the YIpADHTEF telomeres with the control YIpADH telomeres allowed us to determine that the cell equates one Rap1p molecule with ~19 bp of TG_1-3 in vivo (Table 2). Since the spacing of Rap1p sites is one per 18 bp of TG_1-3 in vitro (7), our results indicate that all six of the arrayed Rap1p molecules at the telomere-nontelomere junction participated in telomere length measurement.

Models for telomere length measurement in yeast need to account for the partial counting of the internal 256-bp TG_1-3 tract across the 42- and 50-bp spacers (Fig. 2B), for a 138-bp nontelomeric spacer establishing a new internal telomere-nontelomere junction (Fig. 3A and C), for all Rap1p molecules near the telomere-nontelomere junction participating in telomere length measurement, and for the effect of these Rap1p molecules on telomere length (Fig. 5 and 6). In addition, a wide variety of studies in S. cerevisiae and K. lactis indicate that the Rap1p C terminus at the ends of established telomeres plays an important role in limiting telomere elongation (5, 12, 19, 20, 25). Finally, the Rap1p C terminus at the telomere-nontelomere junction plays a role in the transcriptional silencing of genes near the telomere by nucleating a complex of proteins that then spreads internally by coating chromatin with Sir3p (15, 27, 39, 44). Thus, a model for yeast telomere length regulation must accommodate these different functions.

One current model for telomere structure suggests that the telomere is divided into counted and uncounted regions where the uncounted region consists of Rap1p molecules near the telomere-nontelomere junction bound by Sir3p and Sir4p (3, 8). This model is inconsistent with our data showing that all Rap1p molecules at the telomere-nontelomere junction are counted by the cell. The recent data of Marcand et al. (29) are also inconsistent with this model. Their studies showed that tethering just two Rap1p C termini at the telomere-nontelomere junction can cause the elongated TG_1-3 tract to be maintained at a length ~30 bp shorter than that of control telomeres (29), indicating that at least one or possibly both of these Rap1p C termini participate in telomere length measurement. Given our results that each Rap1p molecule at the telomere-nontelomere junction is counted as ~19 bp of TG_1-3 (Table 2), both of these Gal4-Rap1p fusions must have participated in telomere length measurement. However, data from several labs indicate that the Sir3p-Rap1p interaction is required to establish silencing of genes near telomeres (15, 27, 33, 44). A simple explanation to accommodate these two distinct Rap1p functions is to propose that they are temporally separated, occurring at different times in the cell cycle.

Our working model for telomere length measurement is that after telomeres have been replicated, Rap1p molecules are counted by the formation of a folded telomere structure that is dependent on many weak interactions between Rap1p and negative regulators of telomere length (Fig. 7). This model of the yeast telomere as a highly folded structure is consistent with previous studies showing that the chromatin that includes yeast TG_1-3 repeats is a single unit as defined by micrococcal nuclease digestion (53). In addition, Rap1p binding causes a 90° to 100° bend in DNA in vitro (7), and ~18 Rap1p molecules should be present in a 325-bp telomere, giving rise to a highly folded structure. We hypothesize that many interactions between Rap1p and other proteins constrain the folding of the TG_1-3 sequences so that the chromosome end is brought close to the telomere-nontelomere junction (Fig. 7). Formation of this structure then blocks telomere elongation because the structure itself blocks telomerase access to the chromosome end, or the structure recruits an inhibitory complex that blocks telomerase access. If the structure is not formed, telomere elongation by telomerase may occur. We propose that subsequent to structure formation and elongation of short telomeres (Fig. 7), the yeast Ku proteins may rebind to the chromosome terminus (10) and recruit Sir proteins (46) to form the heterochromatin complex of Sir proteins and Rap1p involved in telomeric silencing. Once silencing is established, genes would then become refractory to transcriptional activation by transactivators.

View larger version (37K): [in this window] [in a new window]   FIG. 7.   A working model for telomere length regulation. Telomere length is monitored by counting Rap1p molecules to keep telomeres within a set range of lengths. We propose that Rap1p molecules are counted by the formation of a transient, highly folded three-dimensional structure stabilized by many weak interactions between Rap1p molecules and negative regulators of telomere length (drawn here as a hairpin for simplicity). When the newly replicated TG1-3 repeats are 325 ± 75 bp, a highly folded structure that links the telomere-nontelomere junction to the chromosome terminus forms. This structure blocks telomerase access to the chromosome terminus, and thus telomere elongation, either by sequestering the end or by recruiting an inhibitory complex (e.g., a complex formed between the Rap1p C terminus, Cdc13p, and Stn1p [9, 34]). When telomeres are short, the structure is not formed, no inhibition occurs, and telomeres are elongated. Lengthening of short telomeres would be enhanced by the Rap1p molecules which are no longer sequestered by the folded structure (38). Subsequent to these events, telomeric heterochromatin is formed and maintained until the next S phase. A simpler model that linked the telomere-nontelomere junction to the chromosome terminus without counting the intervening sequences had been previously proposed for K. lactis telomeres (31).

A number of observations regarding yeast telomere function support our model in limiting telomere length measurement and possible telomere elongation to a period late in S phase that is temporally separated from the establishment of telomere position effect or silencing. Telomere position effect can switch between transcriptionally repressed and active states, and the repressed state can be overcome by transcriptional activators only in S phase (1), indicating that telomeric gene silencing is not stably established at this point in the cell cycle. Late in S phase is when telomeric DNA is replicated (30) and a 50- to 150-bp 3' extension of the TG_1-3 strand is formed (50); therefore, the telomere repeats are undergoing length changes at this time. Slow passage through S phase increases telomeric silencing (22), which is consistent with this hypothesis (Fig. 7) because an increased amount of time after telomere length measurement would allow assembly of telomeric heterochromatin before the chromosome is refolded and the cell cycle proceeds. So the hypothesis that length regulation occurs only in late S phase just after telomeres are replicated, and that afterwards a stable, transcriptionally silenced heterochromatin structure is established and maintained until the next S phase, is consistent with the current data on telomere replication and telomeric silencing.

Our model for telomere length regulation (Fig. 7) can explain all of the results presented here. First, the slight counting differences observed between the different TEF arrays (Table 2) would be due to one set of arrays (e.g., TEF35-6) allowing more efficient structure formation than another (e.g., TEF13-6) because the presentation of Rap1p molecules is slightly different for each array. The more compact Rap1p spacing in the TEF13-6 telomeres may make fewer protein-protein interactions required to form the structure and block telomerase access. As a result, a slightly longer terminal TG_1-3 tract is required to introduce an additional Rap1p molecule to stabilize the structure. Second, all TEF telomeres have nearly the same modal telomere length because the positioning of the six Rap1p molecules at the telomere-nontelomere junction is the same in independent transformants. Therefore, the chromatin complex at the terminus (Fig. 7) will see the same chromatin arrangement at the telomere-nontelomere junction in each transformant. The remaining heterogeneity in the TEF telomeres, i.e., the dispersed band formed by the telomeric restriction fragment, results from individual cells having different terminal TG_1-3 sequences (48). In contrast, telomeres composed of all TG_1-3 sequences have different arrangements of Rap1p molecules at both the telomere-nontelomere junction and the chromosomal terminus in each transformant. Therefore, the chromatin structures of both the telomere-nontelomere junction and the terminus are different in each transformant; thus, structure formation differs in independent transformants, and the modal lengths show large variation (Fig. 6B). Finally, the partial or incomplete counting of the internal 256-bp TG_1-3 tract across the 42- and 50-bp spacers (Fig. 3C) would occur because the nontelomeric sequences partially disrupt the protein-protein interactions between the chromosome terminus and the telomere-nontelomere junction and so the structure is not formed. Slight elongation of the terminal TG_1-3 tract adds Rap1p-Rap1p interactions that overcome the disruption caused by the 42- or 50-bp spacer. In this model, the 138-bp spacer requires so much elongation of the terminal tract that lengthened TG_1-3 repeats can interact more easily with themselves than with the internal tract. This explanation for partial counting can also explain the ~50 bp of telomere shortening seen in sir3 and sir4 mutants (35). The binding of preexisting Rap1p-Sir3p-Sir4p complexes to random sites in the telomere just after DNA replication would interfere with structure formation in wild-type cells. In sir3 and sir4 mutants, structure formation would be more efficient because more Rap1p molecules can interact with negative regulators of telomere length, and so a shorter tract of TG_1-3 and fewer Rap1p molecules would be required to form the structure that blocks telomere lengthening. The result would be slightly shorter telomeres in sir3 and sir4 cells.

While the heterogeneous telomeric sequences in yeast are distinct from the homogeneous repeats found in humans, both of these organisms could share folded, heterogeneous telomeric chromatin structures. Both Rap1p and the human telomeric binding protein TRF1 bend DNA and bind to DNA in a noncooperative fashion (2). Because the human telomeric sequences are homogeneous repeats, the exact positions where telomeric proteins bind within the sequence are not defined. Therefore, the precise arrangement of telomeric binding proteins in a homogeneous sequence can be different at individual telomeres. Thus, the chromatin structures for different human telomeres will be heterogeneous at the protein level, whereas yeast telomeres are also heterogeneous at the DNA sequence level. As both yeast and humans appear to use feedback mechanisms that count telomere binding proteins to measure telomere length (references 29 and 47 and this work), the results discussed here for yeast may apply to humans.