TIBS 1 6 - OCTOBER1991
IF THE SINGLE linear double-stranded DNA molecule that comprises the genetic material of a eukaryotic chromosome were completely sequenced from end to end, the first and last sequences encountered would be the telomeric DNA sequences. These terminal sequences stand out from internal DNA sequences because of both their primary sequence and their functio~al properties. They are required for the complete replication of these linear chromosomes, since a conventional DNA polymerase would leave gaps at the 5' end of newly replicated strands, losing essential genetic information. Also, broken chromosome ends that are unprotected by telomeres may fuse with other broken ends and they are also susceptible to degradation by nucleases. Thus without these essential sequences, the chromosome is unstable [for review see Refs 1-3 (although see Refs 4, 5 for apparent exceptions)]. Telomeres are maintained by the ribonucleoprotein enzyme telomerase e-9 (for review see Ref. I0), which synthesizes one strand of the telomeric DNA in an unusual fashion: an RNA sequence within the telomerase RNA is used as the template for the synthesis of this strand. This review describes this specialized reverse transcriptase, and considers how telomere synthesis by telomerase relates to other processes acting on telomeres in vivo.
The structure of telomeres Telomeres are DNA-protein complexes. The form, although not the prec';se sequence, of telomeric DNA is highly conserved among eukaryotes. It consists of very simple, tandemly repeated sequences, with one strand containing clusters of G residues. Representative exa,'npl~ of telomeric repeated sequences are shown in Table I. A terminal stretch of this simple-sequence DNA ranges from
A
B
A
i
A First replication round completed
/
+
+
+
ProduCtSrOf~ie:~i;d round
Rpre 2 Model for telomere replication. (a) One telomeric region of a chromosome. The G-rich strand is shown as a thick liqe with its 3' extension bound by the telomere terminal protein (hatched). (b) A DNA replication fork initiates from an origin within the chromosome and moves toward the chromosomal terminus. The helicase associated with the replication complex separates the parental strands (marked A and B). The telomere terminal protein is displaced, allowing extension of the transiently free single-stranded G-rich DNA 3' terminus of strand B by telomerase. Discontinuous lagging-strand synthesis by the joint action of a primase and polymerase copies strand B (the zigzag represents RNA primer; the straight arrow is DNA).(c) Leading-strand synthesis toward the chromosomal terminus, copying parental strand A, produces one blunt-ended daughter chromatid. After the newly extended terminus of G-rich strand B is copied, removal of the most distal RNA pdmer (dotted zigzag) leaves a 5' terminal gap (i.e. the G-rich 3' overhang). The telomere terminal protein may now rebind. (d) In the next replication cycle, when the replication complex moves to the end of the blunt-ended chromatid marked A in (c), helicase separates the ends of the strands, producing a transiently single-stranded G-rich strand terminus which can be elongated by telomerase. Replication of the chromatid marked B in (c) occurs as in (a)-(c).
to chromosomal ends lacking telomeres. Before considering this question, the action of telomerase in vitro needs to be clarified. In vitro, telomerase exhibits a preference for elongating Grich single-stranded DNA oligonucleotlde primers, including oligonucleotldes with the same length and sequence as the short, G-rich 12-16 nucleotide telomeric overhang. While AT-rich, C-rich or random sequence oligonucleotides are generally not used efficiently as primers, in some specific cases such oligonucleotides can be elongated if provided at relatively high concentrations in vitro (M. Lee and E. H. Blackburn, unpublished). These results indicate that, normally, the preferred primer substrate of telomerase is a preexisting telomeric end. Thus, the in vitro preference of telomerase for primers with telomeric DNA sequences, while not absolute, does not directly support a role for telomerase in de novo addition to non-telomeric ends. However, in a number of systems telomeric repeats have been found joined to a broken chromosome end, suggesting that telomerase could be responsible for such chromosome 'healing' (for review see Ref. I).
380
appears to either lengthen or shorten, in relatively short increments, in each cell generation (for reviews see Refs 1, 3). This was particularly apparent in a study in haploid yeast, in which the behavior of the telomeres of Individual chromosomal molecules could be inferred. The telomeres of the replicative descendents of each individual chromosomal molecule became progressivefy more heterogeneous in length with each round of cell division or DNA replication, and a model for telomere replication was proposed based on these results zz. in vitro, telomerase cannot use a blunt-ended duplex telcmeric DNA as a pritnc.:, even under conditions when it can elongate non-telomeric singlestranded primers n9 (M. Lee and E. H. Blackburn, unpublished). This is consistent with the mechanism of telomerase shown in Fig. Ib, in which the 3' end of A m ~ l for telomre mpllcatlon In vlvo the primer to be elongated is baseA number of questions about telo- paired to the RNA template sequence. mere structure and replication are still Here, a model for telomere rep:ication unanswered. In most eukaryotes, within is considered which elaborates the prea population of telomeres the number vious one z2, but now accounts for: (I) of duplex telomeric repeat sequences is this inability to elongate a blunt-ended variable, while the overhang length duplex; (2) the ability of DNA polymappears constant ~T,2~. Each telomere erases, at least in vitro, to copy the last
Direct evidence that telomeras~ does indeed add repeats to broken nontelomeric ends has been obtained in recent experiments in which a mutated telomerase RNA gene was introduced into Tetrahymena cells. The mutation was in the RNA template sequence, and had been shown previously to specify synthesis in vivo of telomeres containing the corresponding mutated sequence m, thus distinguishing newly added telomeric repeats from pre-existing ones. Cells expressing the mutated telomerase RNA gene were induced to undergo site-specific fragmentation of their chromosomes, which results in de novo addition of telomeric DNA to nontelomeric sequences. In de novo-add~d telomeres, mutated telomerase telomeric repeats were found directly adjacent to the non-telomeric end sequences (G-L. Yu and E. H. Blackburn, submitted).
TIBS 16 - OCTOBER1991
(5') nucleotide of their DNA template; (3) the available knowledge of the structure and behavior of telomeres in viuo. The model is illustrated on Fig. 2, which shows one telomeric end of a chromosome undergoing two successive rounds of replication. The telomere terminal protein that is originally bound to the 3' protrusion of the G-rich strand is displaced, perhaps by the replication complex as it moves towards the chromosome end (Fig. 2b). This allows ~xtension of the singlestranded 3' terminus of strand B by telomerase. This strand is replicated by discontinuous lagging-strand synthesis and removal of the last RNA primer leaves the G-rich 3' overhang (Fig. 2c). Rebinding of the telomere terminal protein may curtail further elongation of the telomere by competing with telomerase for the single stranded telomeric end 27. By contrast, strand A is replicated by leading-strand synthesis towards the chromosomal terminus, producing a blunt-ended daughter chromatid (Fig. 2c). The telomere of this newly synthesized strand is not extended until the next replication cycle, when strand separation of chromatid A transiently exposes a single-stranded G-rich terminus which can then be elongated by telomerase (Fig. 2d). Subsequent stages in the second replication cycle are as described for the first. This model makes a number of predictions. As proposed previously, such repeated cycles generate the progressive increase in telomere length and length heterogeneity that are observed in various systems in vivo (for review see Ref. I). It predicts that telomerase acts only during S phase, using the opportunity afforded by the displacement of either the complementary strand and/or the telomere terminal protein by the semi-conservative replication machinery. The model also predicts that half of the telomeric termini on chromosomes may be blunt-ended. Do the available data support or refute this possibility? The G-rich strand protrusion at chromosomal termini was first demonstrated by DNA sequence analysis of both strands of hypotrichous ciliate telomeres (whose overall duplex length is regulated more strictly than in most eukaryotes) 2s. This analysis did not suggest the presence of blunt ends, although it should be noted that it was not specifically designed for their detection. In Tetrahymena and slime mold telomeres the G-rich overhang
was measured by analysis of the length National Institutes of Health grants of the overhang susceptible to single- GM26259 and GM32565 to E.H.B. strand-specific chemical modification |7. Blunt ends would not have been seen in References 1 Blackbum, E. H. (1991) Nature 350, 569-573 this analysis. Hence, it is not estab2 Blackburn, E. H. and Szostak, J. W. (1984) lished whether some telomeric termini Annu. Rev. Biochem. 53, 163-194 are, in fact, blunt-ended in oivo. If blunt3 Zakian, V. A. (1989) Annu. Rev. Genet. 23, 579-604 ended DNA telomeres are found definitively not to exist in uivo, this will sug- 4 Biessmann, H., Carter S. B. and Mason, J. M. (1990) Proc. Natl Acad. Sci. USA 87,1758-1761 gest either: (I) in contrast to the situ5 Levis, R. W. (1989) Cell 58, 791-801 ation in vitro, telomerase in uivo can 6 Greider, C. W. and Blackburn, E. H. (1989) Nature 337, 331-337 elongate the newly synthesized blunt7 Shippen-Lentz, D. and Blackburn, E. H. (1989) ended (or even 3' recessed) leading Mol. Cell. Biol. 9, 2761-2764 strand (possibly facilitated by inter8 Modn, G. B. (1989) Cell 59, 521-529 actions with telomere structural pro9 Zahler, A. M. and Prescott, D. M. (1988) Nucleic Acids Res. 16, 6953-6972 teins or the replication machinery); Blackburn, E. H. (1990) Science 249, 489-490 or (2) other as yet uncharacterized 10 11 Lustig, A. J., Kurtz, S. and Shore, D. (1990) activities process the DNA ends to creScience 250, 549-552 12 Conrad, M. N., Wright, J. H., Wolf, A..~. and ate the observed 3' overhang. V. A. (1990) Cell 63, 739-750 Additional regulatory mechanisms 13 Zakian, Gottschling, D. E. and Zakia n, V. A. (1986) Cell operating on overall telomere lengtil, 47, 195-205 which become apparent over the long 14 Price, C. M. and Cech, T. R. (1989) Biochemistry 28, 769-774 term, have been inferred through sevPrice, C. R. (1990) MoL Ceil. Biol. 10, eral studies of telomere length variation 15 3421-3431 in different systems (for reviews see 16 Raghuraman, M. K., Dunn, C. J., Hicke, B. J. and Cech, T. R. (1989) Nucleic Acids Res. 17, Refs I, 3). An intriguing possibility is 4235-4253 that feedback mechanisms act on indi17 Henderson, E. R. and Blackburn, E. H. (1989) vidual telomeres, perhaps mediated by MoL Cell. BioL 9, 345-348 proteins that are structurally associ- 18 Yu, G-L., Bradley, J. D., Attardi, L. D. and Blackburn, E. H. (1990) Nature 344, 126-132 ated with telomeres. The telomere-bindC. W. and Blackburn, E. H. (1987) Cell ing protein RAP1 of yeast is an attrac- 19 Greider, 51, 887-898 tive candidate for such a regulator of 20 Shippen-Lentz, D. and Blackburn, E. H. (1990) telomere length",~2,, since it also plays various regulatory roles in gene expression. Indeed, the potential exists for interactions between all the components known to act on telomeric DNA: double- and single-stranded telomeric DNA-binding proteins as well as telomerase. Whether and how such interactions occur are still to be unravelled.
Acknowledgements I thank my colleagues in the laboratory for helpful comments on the manuscript. Research support was from
Science 247, 546-552 21 Shampay, J., Szostak, J. W. and Blackburn, E. H. (1984) Natur.~=.310, 154-157 22 Shampay, J. ano Blackburn, E. H. (1988) Proc. Natl Acad. ScL USA 85, 534-538 23 Lundblad, V. and Szostak, J. W. (1989) Cell 57, 633-643 24 Lundblad, V. and Blackburn E. H. (199C} Cell 60, 529-530 25 Kipling, D. and Cooke, H. J. (1991) Nature 347, 400-402 26 Klobutcher, L. A., Swanton, M. T., Donini, P. and Prescott, D. M. (1981) Proc. Natl Acad. Sci. USA 78, 3015-3019 27 Raghuraman, M. K. and Cech, 1. R. (1990) Nucleic Acids. Res. 18, 4543-4552 28 Blackburn, E. H. and Gall, J. G. (1978) 1 Me/. BioL 120, 33-53
TIBS reference lists Authors of 7"IBS articles are asked to limit the number of references citea to provide non-specialist readers with a concise list for further reading. It is hoped that the citation of other, more extensive review articles rather than a comprehensive list of original articles enables interested readers to delve more immediately into the topic. 381
IF THE SINGLE linear double-stranded DNA molecule that comprises the genetic material of a eukaryotic chromosome were completely sequenced from end to end, the first and last sequences encountered would be the telomeric DNA sequences. These terminal sequences stand out from internal DNA sequences because of both their primary sequence and their functio~al properties. They are required for the complete replication of these linear chromosomes, since a conventional DNA polymerase would leave gaps at the 5' end of newly replicated strands, losing essential genetic information. Also, broken chromosome ends that are unprotected by telomeres may fuse with other broken ends and they are also susceptible to degradation by nucleases. Thus without these essential sequences, the chromosome is unstable [for review see Refs 1-3 (although see Refs 4, 5 for apparent exceptions)]. Telomeres are maintained by the ribonucleoprotein enzyme telomerase e-9 (for review see Ref. I0), which synthesizes one strand of the telomeric DNA in an unusual fashion: an RNA sequence within the telomerase RNA is used as the template for the synthesis of this strand. This review describes this specialized reverse transcriptase, and considers how telomere synthesis by telomerase relates to other processes acting on telomeres in vivo.
The structure of telomeres Telomeres are DNA-protein complexes. The form, although not the prec';se sequence, of telomeric DNA is highly conserved among eukaryotes. It consists of very simple, tandemly repeated sequences, with one strand containing clusters of G residues. Representative exa,'npl~ of telomeric repeated sequences are shown in Table I. A terminal stretch of this simple-sequence DNA ranges from
A
B
A
i
A First replication round completed
/
+
+
+
ProduCtSrOf~ie:~i;d round
Rpre 2 Model for telomere replication. (a) One telomeric region of a chromosome. The G-rich strand is shown as a thick liqe with its 3' extension bound by the telomere terminal protein (hatched). (b) A DNA replication fork initiates from an origin within the chromosome and moves toward the chromosomal terminus. The helicase associated with the replication complex separates the parental strands (marked A and B). The telomere terminal protein is displaced, allowing extension of the transiently free single-stranded G-rich DNA 3' terminus of strand B by telomerase. Discontinuous lagging-strand synthesis by the joint action of a primase and polymerase copies strand B (the zigzag represents RNA primer; the straight arrow is DNA).(c) Leading-strand synthesis toward the chromosomal terminus, copying parental strand A, produces one blunt-ended daughter chromatid. After the newly extended terminus of G-rich strand B is copied, removal of the most distal RNA pdmer (dotted zigzag) leaves a 5' terminal gap (i.e. the G-rich 3' overhang). The telomere terminal protein may now rebind. (d) In the next replication cycle, when the replication complex moves to the end of the blunt-ended chromatid marked A in (c), helicase separates the ends of the strands, producing a transiently single-stranded G-rich strand terminus which can be elongated by telomerase. Replication of the chromatid marked B in (c) occurs as in (a)-(c).
to chromosomal ends lacking telomeres. Before considering this question, the action of telomerase in vitro needs to be clarified. In vitro, telomerase exhibits a preference for elongating Grich single-stranded DNA oligonucleotlde primers, including oligonucleotldes with the same length and sequence as the short, G-rich 12-16 nucleotide telomeric overhang. While AT-rich, C-rich or random sequence oligonucleotides are generally not used efficiently as primers, in some specific cases such oligonucleotides can be elongated if provided at relatively high concentrations in vitro (M. Lee and E. H. Blackburn, unpublished). These results indicate that, normally, the preferred primer substrate of telomerase is a preexisting telomeric end. Thus, the in vitro preference of telomerase for primers with telomeric DNA sequences, while not absolute, does not directly support a role for telomerase in de novo addition to non-telomeric ends. However, in a number of systems telomeric repeats have been found joined to a broken chromosome end, suggesting that telomerase could be responsible for such chromosome 'healing' (for review see Ref. I).
380
appears to either lengthen or shorten, in relatively short increments, in each cell generation (for reviews see Refs 1, 3). This was particularly apparent in a study in haploid yeast, in which the behavior of the telomeres of Individual chromosomal molecules could be inferred. The telomeres of the replicative descendents of each individual chromosomal molecule became progressivefy more heterogeneous in length with each round of cell division or DNA replication, and a model for telomere replication was proposed based on these results zz. in vitro, telomerase cannot use a blunt-ended duplex telcmeric DNA as a pritnc.:, even under conditions when it can elongate non-telomeric singlestranded primers n9 (M. Lee and E. H. Blackburn, unpublished). This is consistent with the mechanism of telomerase shown in Fig. Ib, in which the 3' end of A m ~ l for telomre mpllcatlon In vlvo the primer to be elongated is baseA number of questions about telo- paired to the RNA template sequence. mere structure and replication are still Here, a model for telomere rep:ication unanswered. In most eukaryotes, within is considered which elaborates the prea population of telomeres the number vious one z2, but now accounts for: (I) of duplex telomeric repeat sequences is this inability to elongate a blunt-ended variable, while the overhang length duplex; (2) the ability of DNA polymappears constant ~T,2~. Each telomere erases, at least in vitro, to copy the last
Direct evidence that telomeras~ does indeed add repeats to broken nontelomeric ends has been obtained in recent experiments in which a mutated telomerase RNA gene was introduced into Tetrahymena cells. The mutation was in the RNA template sequence, and had been shown previously to specify synthesis in vivo of telomeres containing the corresponding mutated sequence m, thus distinguishing newly added telomeric repeats from pre-existing ones. Cells expressing the mutated telomerase RNA gene were induced to undergo site-specific fragmentation of their chromosomes, which results in de novo addition of telomeric DNA to nontelomeric sequences. In de novo-add~d telomeres, mutated telomerase telomeric repeats were found directly adjacent to the non-telomeric end sequences (G-L. Yu and E. H. Blackburn, submitted).
TIBS 16 - OCTOBER1991
(5') nucleotide of their DNA template; (3) the available knowledge of the structure and behavior of telomeres in viuo. The model is illustrated on Fig. 2, which shows one telomeric end of a chromosome undergoing two successive rounds of replication. The telomere terminal protein that is originally bound to the 3' protrusion of the G-rich strand is displaced, perhaps by the replication complex as it moves towards the chromosome end (Fig. 2b). This allows ~xtension of the singlestranded 3' terminus of strand B by telomerase. This strand is replicated by discontinuous lagging-strand synthesis and removal of the last RNA primer leaves the G-rich 3' overhang (Fig. 2c). Rebinding of the telomere terminal protein may curtail further elongation of the telomere by competing with telomerase for the single stranded telomeric end 27. By contrast, strand A is replicated by leading-strand synthesis towards the chromosomal terminus, producing a blunt-ended daughter chromatid (Fig. 2c). The telomere of this newly synthesized strand is not extended until the next replication cycle, when strand separation of chromatid A transiently exposes a single-stranded G-rich terminus which can then be elongated by telomerase (Fig. 2d). Subsequent stages in the second replication cycle are as described for the first. This model makes a number of predictions. As proposed previously, such repeated cycles generate the progressive increase in telomere length and length heterogeneity that are observed in various systems in vivo (for review see Ref. I). It predicts that telomerase acts only during S phase, using the opportunity afforded by the displacement of either the complementary strand and/or the telomere terminal protein by the semi-conservative replication machinery. The model also predicts that half of the telomeric termini on chromosomes may be blunt-ended. Do the available data support or refute this possibility? The G-rich strand protrusion at chromosomal termini was first demonstrated by DNA sequence analysis of both strands of hypotrichous ciliate telomeres (whose overall duplex length is regulated more strictly than in most eukaryotes) 2s. This analysis did not suggest the presence of blunt ends, although it should be noted that it was not specifically designed for their detection. In Tetrahymena and slime mold telomeres the G-rich overhang
was measured by analysis of the length National Institutes of Health grants of the overhang susceptible to single- GM26259 and GM32565 to E.H.B. strand-specific chemical modification |7. Blunt ends would not have been seen in References 1 Blackbum, E. H. (1991) Nature 350, 569-573 this analysis. Hence, it is not estab2 Blackburn, E. H. and Szostak, J. W. (1984) lished whether some telomeric termini Annu. Rev. Biochem. 53, 163-194 are, in fact, blunt-ended in oivo. If blunt3 Zakian, V. A. (1989) Annu. Rev. Genet. 23, 579-604 ended DNA telomeres are found definitively not to exist in uivo, this will sug- 4 Biessmann, H., Carter S. B. and Mason, J. M. (1990) Proc. Natl Acad. Sci. USA 87,1758-1761 gest either: (I) in contrast to the situ5 Levis, R. W. (1989) Cell 58, 791-801 ation in vitro, telomerase in uivo can 6 Greider, C. W. and Blackburn, E. H. (1989) Nature 337, 331-337 elongate the newly synthesized blunt7 Shippen-Lentz, D. and Blackburn, E. H. (1989) ended (or even 3' recessed) leading Mol. Cell. Biol. 9, 2761-2764 strand (possibly facilitated by inter8 Modn, G. B. (1989) Cell 59, 521-529 actions with telomere structural pro9 Zahler, A. M. and Prescott, D. M. (1988) Nucleic Acids Res. 16, 6953-6972 teins or the replication machinery); Blackburn, E. H. (1990) Science 249, 489-490 or (2) other as yet uncharacterized 10 11 Lustig, A. J., Kurtz, S. and Shore, D. (1990) activities process the DNA ends to creScience 250, 549-552 12 Conrad, M. N., Wright, J. H., Wolf, A..~. and ate the observed 3' overhang. V. A. (1990) Cell 63, 739-750 Additional regulatory mechanisms 13 Zakian, Gottschling, D. E. and Zakia n, V. A. (1986) Cell operating on overall telomere lengtil, 47, 195-205 which become apparent over the long 14 Price, C. M. and Cech, T. R. (1989) Biochemistry 28, 769-774 term, have been inferred through sevPrice, C. R. (1990) MoL Ceil. Biol. 10, eral studies of telomere length variation 15 3421-3431 in different systems (for reviews see 16 Raghuraman, M. K., Dunn, C. J., Hicke, B. J. and Cech, T. R. (1989) Nucleic Acids Res. 17, Refs I, 3). An intriguing possibility is 4235-4253 that feedback mechanisms act on indi17 Henderson, E. R. and Blackburn, E. H. (1989) vidual telomeres, perhaps mediated by MoL Cell. BioL 9, 345-348 proteins that are structurally associ- 18 Yu, G-L., Bradley, J. D., Attardi, L. D. and Blackburn, E. H. (1990) Nature 344, 126-132 ated with telomeres. The telomere-bindC. W. and Blackburn, E. H. (1987) Cell ing protein RAP1 of yeast is an attrac- 19 Greider, 51, 887-898 tive candidate for such a regulator of 20 Shippen-Lentz, D. and Blackburn, E. H. (1990) telomere length",~2,, since it also plays various regulatory roles in gene expression. Indeed, the potential exists for interactions between all the components known to act on telomeric DNA: double- and single-stranded telomeric DNA-binding proteins as well as telomerase. Whether and how such interactions occur are still to be unravelled.
Acknowledgements I thank my colleagues in the laboratory for helpful comments on the manuscript. Research support was from
Science 247, 546-552 21 Shampay, J., Szostak, J. W. and Blackburn, E. H. (1984) Natur.~=.310, 154-157 22 Shampay, J. ano Blackburn, E. H. (1988) Proc. Natl Acad. ScL USA 85, 534-538 23 Lundblad, V. and Szostak, J. W. (1989) Cell 57, 633-643 24 Lundblad, V. and Blackburn E. H. (199C} Cell 60, 529-530 25 Kipling, D. and Cooke, H. J. (1991) Nature 347, 400-402 26 Klobutcher, L. A., Swanton, M. T., Donini, P. and Prescott, D. M. (1981) Proc. Natl Acad. Sci. USA 78, 3015-3019 27 Raghuraman, M. K. and Cech, 1. R. (1990) Nucleic Acids. Res. 18, 4543-4552 28 Blackburn, E. H. and Gall, J. G. (1978) 1 Me/. BioL 120, 33-53
TIBS reference lists Authors of 7"IBS articles are asked to limit the number of references citea to provide non-specialist readers with a concise list for further reading. It is hoped that the citation of other, more extensive review articles rather than a comprehensive list of original articles enables interested readers to delve more immediately into the topic. 381
