Normal human chromosomes have long G-rich telomeric overhangs at one end

Abstract

Telomeres protect the ends of linear chromosomes from degradation and abnormal recombination events, and in vertebrates may be important in cellular senescence and cancer. However, very little is known about the structure of human telomeres. In this report we purify telomeres and analyze their termini. We show that following replication the daughter telomeres have different terminal overhangs in normal diploid telomerase-negative human fibroblasts. Electron microscopy of those telomeres that have long overhangs yields 200 +/- 75 nucleotides of single-stranded DNA. This overhang is four times greater than the amount of telomere shortening per division found in these cells. These results are consistent with models of telomere replication in which leading-strand synthesis generates a blunt end while lagging-strand synthesis produces a long G-rich 3' overhang, and suggest that variations in lagging-strand synthesis may regulate the rate of telomere shortening in normal diploid human cells. Our results do not exclude the possibility that nuclease processing events following leading strand synthesis result in short overhangs on one end.

Figure 1

Purification of telomeres. (A) Sequence specificity of the purification of telomeres. HinfI-digested human placental DNA was annealed to various biotinylated oligonucleotides, and telomere/oligonucleotide complexes were captured with streptavidin-coated magnetic beads. The DNA remaining in the supernatant vs. that bound to the beads was then analyzed on agarose gels and probed with a ³²P-labeled (TTAGGG)₄ oligonucleotide. CTR₄ and CTR₆ contain four and six copies of the C-rich terminal repeat (CCCTAA), GTR₆ contains six copies of the G-rich terminal repeat (TTAGGG), and ClaHin is a nontelomeric biotinylated oligonucleotide. Only the C-rich oligonucleotides complementary to the G-rich telomeric overhang were able to retrieve the double-stranded placental telomeres. (B) Purification requires single-stranded overhangs. Treatment of the DNA with the single-stranded exonuclease Exo 1 (1 U/μg) largely abolished the ability to retrieve telomeres. Noncanonical G structures (Henderson 1995; Kipling 1995) may make a small fraction of the overhangs resistant to complete digestion. (C) Purification requires ⩾12 bases of overhang. A 5-kbp artificial telomere containing single-stranded G-rich overhangs of variable lengths was annealed to a biotinylated C-rich oligonucleotide and purified using streptavidin-coated magnetic beads. The material bound to the magnetic beads or remaining in the supernatant is large and does not enter a denaturing polyacrylamide gel (uncut lanes). The radioactive telomeric repeats were released by digestion with an enzyme that cuts the plasmid just before the start of the telomeric repeats (ClaI). Sequences containing as few as 12 nucleotides of G-rich overhangs can be recovered even if they are part of a 5-kbp-long artificial telomere.

Figure 2

Telomeres with long overhangs contain newly synthesized C-rich daughter strands. (A) Schematic model for telomere replication. This model postulates that lagging-strand synthesis leaves a 3′ overhang of the parental G-rich strand. Following one round of replication in BrdU, labeled G-rich strands are present in blunt-ended telomeres, whereas labeled C-rich strands are present in telomeres that have overhangs. (B) Retrieval of BrdU-labeled daughter strands. Telomeres from BJ fibroblasts, in which both strands contained thymidine (Thy:Thy), only one strand contained BrdU after a single round of replication (Thy:BrdU), or both strands contained BrdU after four rounds of replication (BrdU:BrdU), were purified using biotinylated C-rich oligonucleotides, melted, and then precipitated with anti-BrdU antibodies. The antibody-bound DNA was then released by boiling in SDS, analyzed on agarose gels, and probed with oligonucleotides specific for each strand. The amount of purified telomeres used in each sample was not identical, as the efficiency of magnetic bead purification varied between experiments. The exposure of each lane has been adjusted to represent equivalent amounts of input telomeres (antibody bound + unbound for each strand). The newly synthesized (BrdU-containing) strand on those telomeres that contained long overhangs was the C-rich and not the G-rich strand.

Figure 3

T4 gp32 decoration of single-stranded DNA and telomeric overhangs. (A) The single-stranded region of linearized plasmid DNA containing a 450-nucleotide gap of single-stranded telomeric repeats was decorated with the single-strand binding T4 gp32. Initial magnification, 100,000×. (B) Linear relationship between size standards and measured lengths. Different lengths of single-stranded gaps (•) or overhangs (○) coated with gp32 were examined. The 48- and 450-nucleotide single-stranded regions contained G-rich telomeric repeats; the 200- and 1000-nucleotide gaps contained plasmid sequences. Except for the 48-nucleotide overhang, 40–70 molecules of each type were examined. It was difficult to distinguish the very short decorated region from background for the 48-nucleotide overhang sample, and we consider 50 nucleotides of overhang to be the limit of detection for this technique. Error bars indicate 1 standard deviation (S.D.). (C) Purified BJ fibroblast telomere decorated with gp32. Initial magnification, 25,000×. (D) Histogram of the length of the gp32-decorated regions of purified telomeres from BJ fibroblast DNA. A total of 108 and 69 molecules from population doubling level (PDL) 20 (solid bar) and 87 (shaded bar) were examined. Between 70% and 80% of the molecules purified on the basis of having G-rich overhangs had one decorated end. The remaining undecorated molecules may represent fragments of broken telomeres. None of the telomeres was decorated at both ends. Average overhang lengths (±S.D.) were 157 ± 69 nucleotides for PDL20 and 226 ± 88 nucleotides for PDL87 fibroblasts. A higher background of free T4 gp32 in the PDL87 preparation may have compromised our ability to detect the shortest overhangs in that sample.

Figure 4

Telomere size changes after chromosome replication in the absence of telomerase. A replication fork is shown proceeding into a telomere of length L with a 3′ G-rich overhang 1 unit long. The size of the telomere before replication is the average of the two strands, which is [(L + 1) + L] ÷ 2 = L + 0.5 units. Lagging strand synthesis is illustrated as a series of discrete Okazaki fragments that would be joined together to form a continuous strand. Following replication, lagging strand synthesis would leave a long 3′ overhang; leading strand synthesis would generate a blunt end. After replication is complete (step 1) the average size of the four strands would be [(L + 1) + L + L + L] ÷ 4 = L + 0.25 units. The net shortening after replication would be (L + 0.5) − (L + 0.25) = 0.25, implying that the rate of telomere shortening should be one-quarter of the length of the G-rich overhang. A recent model (Makarov et al. 1997) has been postulated, in which extensive nuclease processing produces symmetrical long overhangs (step 2). If this were to occur, the average size of the four strands would be [(L + 1) + L + L + (L − 1)] ÷ 4 = L. The net shortening after replication and processing would thus be (L + 0.5) − L = 0.5, suggesting that the rate of telomere shortening in the absence of telomerase should be one-half the length of the G-rich overhang. Our data indicate that the rate of shortening (50 bp/division) is one quarter the length of the overhang (200 ± 75 nucleotides) in BJ fibroblasts.

Figure 5

Telomere shortening in BJ fibroblasts. (A) Terminal restriction fragment (TRF) gel of DNA from BJ foreskin fibroblasts at different mean population doubling levels. (B) Rate of telomere shortening. Data from three different TRF gels using DNA prepared from two different lifespan studies are shown. The average rate of shortening was 49 bp/population doubling.

Figure 5

Telomere shortening in BJ fibroblasts. (A) Terminal restriction fragment (TRF) gel of DNA from BJ foreskin fibroblasts at different mean population doubling levels. (B) Rate of telomere shortening. Data from three different TRF gels using DNA prepared from two different lifespan studies are shown. The average rate of shortening was 49 bp/population doubling.