Telomerase-Independent Stabilization of Short Telomeres in Trypanosoma brucei

In cancer cells and germ cells, shortening of chromosome endsis prevented by telomerase. Telomerase-deficient cells havea replicative life span, after which they enter senescence.Senescent cells can give rise to survivors that maintain chromosomeends through recombination-based amplification of telomericor subtelomeric repeats. We found that in Trypanosoma brucei,critically short telomeres are stable in the absence of telomerase.Telomere stabilization ensured genomic integrity and could haveimplications for telomere maintenance in human telomerase-deficientcells. Cloning and sequencing revealed 7 to 27 TTAGGG repeatson stabilized telomeres and no changes in the subtelomeric region.Clones with short telomeres were used to study telomere elongationdynamics, which differed dramatically at transcriptionally activeand silent telomeres, after restoration of telomerase. We proposethat transcription makes the termini of short telomeres accessiblefor rapid elongation by telomerase and that telomere elongationin T. brucei is not regulated by a protein-counting mechanism.Many minichromosomes were lost after long-term culture in theabsence of telomerase, which may reflect their different mitoticsegregation properties.

INTRODUCTION

Top

Introduction

Linear chromosomes are capped by telomeres, which are nucleoproteincomplexes consisting of tandem DNA repeats and proteins thatbind to them. Telomeres serve two major purposes: they protectchromosome ends from the DNA repair machinery that would otherwiserecognize them as double-stranded breaks, and, with the helpof telomerase, they compensate for the gradual sequence lossthat would otherwise arise from the inability of conventionalDNA polymerases to replicate chromosome ends (21, 31, 51). Telomeraseis present in germ cells, and its activation is a crucial stepin tumorigenesis (13, 30). In human and yeast cells, telomeraseactivity is tightly regulated by proteins that relay informationabout telomere length (32, 52). The ability of telomerase toextend a telomere is inversely proportional to the initial lengthof the telomere DNA repeats (36, 37, 56). This protein-countingmodel explains how short telomeres can be rapidly elongatedwhereas the length of long telomeres is restricted (37). Inhuman somatic cells, which lack telomerase, telomeres shortenthrough each replicative cycle (25). This molecular clock limitsthe proliferation of somatic cells to $~$ 50 to 60 population doublings(PD), after which they enter a nonreversible growth arrest calledsenescence. Telomerase-deficient cells occasionally escape senescenceor apoptosis. Yeast and human survivor cells acquire the abilityto maintain chromosome ends through an alternative lengtheningmechanism (ALT) (35, 48). In the absence of telomerase, Saccharomycescerevisiae can survive by amplification of subtelomeric repeats,recombination among telomeres, or palindrome-based amplificationof subtelomeric regions (35, 38). In addition to telomere-telomererecombination-based lengthening, Schizosaccharomyces pombe canovercome the loss of telomeric DNA by chromosome circularization(43).

From 5 to 20% of human cancers lack telomerase activity anduse ALT to dramatically increase their telomere length (5).Although telomere-telomere recombination is the major mechanismunderlying ALT, fluorescent in situ hybridization of metaphasespreads of human ALT cells reveals that telomeric repeats areundetectable at many chromosome ends (5, 17). It remains unclearwhether signal-free ends represent short telomeres, how muchtelomeric DNA they carry, and whether these ends might be temporarilystabilized until they are able to undergo recombination.

Here we describe the consequences of telomere shortening inTrypanosoma brucei, the causative agent of African sleepingsickness. This parasite persists in the bloodstream of its mammalianhost by stochastically changing its surface coat, which consistsof a homogeneous layer of variant surface glycoproteins (VSG)(2, 4). VSG genes are dispersed throughout the genome, but theactively transcribed VSG gene is always located in one of $~$ 20subtelomeric transcription units called expression sites (ES)(15, 45, 55). Antigenic switching can be achieved by duplicativegene conversion events, in which a transcriptionally silenttelomeric or a nontelomeric VSG gene is the donor (28, 49),or by coupled ES silencing and activation, called an in situswitch (28). Despite intensive investigations, the mechanismsthat regulate antigenic variation remain elusive, but telomeresare likely to play an important role.

T. brucei has 11 diploid megabase chromosomes (MBC), which containall the essential genes and most ES; a variable number of intermediatechromosomes (IC), which appear to contain ES and little else;and $~$ 100 minichromosomes (MC), which consist mainly of 177-bptandem repeats, silent VSG genes, and telomeres (40, 58). MCare thought to provide a reservoir of VSG genes that can beactivated by duplicative transposition. Despite their similarityand number, MC are mitotically stable and faithfully segregateduring several years of continuous propagation (1, 60). We previouslyidentified and deleted the gene encoding telomerase reversetranscriptase (TERT) in order to investigate the role that telomeresmight play during antigenic variation (16). Here we report theconsequences of telomere shortening for different chromosometypes. Most interestingly, our results show that telomerase-deficientT. brucei can maintain the length of critically short telomeresby telomere stabilization. MC are lost as telomeres become criticallyshort. By restoration of the TERT gene, we were able to explorethe dynamics of telomere elongation at transcriptionally activeand silent ES. Our results suggest that T. brucei has evolveda novel way of regulating telomerase-dependent telomere elongation.We propose a model in which transcription modulates telomericchromatin structure and renders it accessible to the telomerasecomplex.

MATERIALS AND METHODS

Top

Materials and Methods

Trypanosome cell lines and complementation constructs. T. brucei Lister 427, expressing MITat 1.2 (VSG 221), was previouslymanipulated to constitutively express T7 polymerase and theTet repressor protein and cultured in HMI-9 medium at 37°C(27, 59). In this cell line, designated "single marker," bothalleles of the TERT open reading frame (ORF) were replaced bygenes conferring resistance to hygromycin and puromycin, respectively(16). To complement the TERT deletion phenotype, the T. bruceiTERT open reading frame was amplified using primers that introduceNdeI restriction sequences at the 5' end and HpaI site at the3' end of the gene. The PCR product was cloned into pGEM-T Easy(Promega), and clones from independent PCRs were sequenced toexclude the possibility of PCR-derived mutations. To put theORF under the control of an inducible T7 promoter, the PCR productwas released by NdeI/HpaI digestion and cloned into similarlydigested pCO57 (this step removes a green fluorescent protein[GFP] cassette from pCO57, a derivative of pLew82 [59]). Properintegration was verified by restriction analysis and sequencing.To introduce a GFP tag at the telomerase C terminus, NdeI siteswere introduced by PCR at both ends of the TERT ORF, clonedinto pGEM-T Easy, released, and ligated into NdeI-cut and dephosphorylatedpCO57. The constructs were linearized by digestion with NotIand transfected into telomerase-deficient cell lines, and integrantswere selected with 2.5 µg/ml phleomycin. Upon reintroduction,TERT-expressing clones were verified by Southern blotting (datanot shown). Expression of the fusion product was induced byaddition of 100 ng/ml doxycycline. DNA was extracted on a weeklybasis, and telomere length changes at transcriptionally silentand active ES were monitored by Southern hybridization. To assesswhether TERT-GFP is fully functional, its ability to elongatea short telomere was compared with those of a reintroduced wild-typeTERT allele and a hemagglutinin-tagged version. Wild-type andTERT-GFP constructs were fully functional, whereas hemagglutinin-TERTdid not elongate telomeres (data not shown).

Bal 31 digestion. Bal 31 digestion was performed as described previously (41).Briefly, 100 µg of DNA in 500 µl volume were incubatedat 30°C with 5 units of Bal 31. One-hundred-microliter sampleswere removed periodically, added to 2 µl 0.5 M EDTA, incubatedat 65°C for 10 min, and stored at –20°C untilDNA was extracted by phenol-chloroform-isoamyl-alcohol, ethanolprecipitated, and dissolved in 10 mM Tris (pH 7.4), 0.1 mM EDTA.DNA was digested with EcoRI and loaded on an 0.8% agarose gel.Southern blotting and hybridization procedures were performedas described previously (16).

Telomere tailing procedure and PCR amplification. Telomere tailing and PCR amplification were carried out as describedelsewhere (19), with minor modifications. To increase the efficiencyof telomere tailing, terminal restriction fragments (TRF) correspondingto the size of stabilized VSG 121 telomere ( $~$ 1.6 kb) were isolatedfrom an 0.8% agarose gel. Approximately 100 ng of TRF in One-Phor-Allbuffer (Pharmacia), 1mM dCTP, and 5 U of terminal deoxynucleotidyltransferase (Amersham Biosciences) was incubated for 30 minat 37°C. The efficiency of the tailing reaction was verifiedby adding radiolabeled dCTP to the reaction mix and runningthe products on agarose gels (data not shown). The tailing reactionswere PCR amplified using 0.75 µM primer A (5'-ACACAGGCGAACCAGAATGACGCTGCAGCCAAAGCA-3'),1.0 µM primer B (CCCCC)₃, 320 µM deoxynucleosidetriphosphates, 10x Taq polymerase buffer, and 2.5 U of Taq polymeraseby incubating for 2 min at 94°C, followed by 45 cycles of45 seconds at 94°C, 45 seconds at 65°C, and 70 secondsat 72°C. PCR-amplified products were separated on a 1.2%agarose gel, and nontailed TRF were amplified in parallel asa negative control. Specific products were gel purified andsubcloned into pGEM-T (Promega). Transformed bacteria (Escherichiacoli DH 10B) were grown at 30°C to minimize recombinationevents. Restriction digests were performed as described by themanufacturer (New England Biolabs). Sequencing reactions werecarried out by GeneWiz Inc. (North Brunswick, NJ).

RAGE. DNA agarose plugs were prepared as described by Navarro andCross (45). Briefly, 2 x 10⁸ cells were purified from the bloodof infected mice, resuspended in 0.5 ml L-buffer (LB) (0.1 MEDTA [pH 8.0], 10 mM Tris-HCl [pH 7.6], 20 mM NaCl), and incubatedfor 10 min at 42°C. A 0.5-ml quantity of 1.6% low-temperature-gellingagarose (Sigma) in LB was added to 0.5 ml cells, mixed, andpoured into plug molds (Bio-Rad Laboratories). Plugs were treatedin 3 ml LB with 1 mg/ml proteinase K at 50°C for 2 days.After two washes for 15 min with LB, proteinase K treatmentand washing were repeated. Plugs were embedded in 0.8% agarosein 0.5x Tris-borate-EDTA gel. Chromosome-sized DNA was separatedby rotating agarose gel electrophoresis (RAGE) (Stratagene,Inc.) under conditions described previously (44): a linear rampof 100 to 300 s for 10 h at 120 V and then 1,000 to 2,500 sfor 80 h at 50 V. Gels were dried at room temperature and hybridizedas described above. Separation of MC and IC took place at arotation angle of 120° in 0.5x Tris-borate-EDTA at 12°C.The program consisted of one ramped pulse time of 60 to 20 sat 110 V for 60 h (Luisa Figueiredo, unpublished data). Forin-gel hybridization with radiolabeled telomeric oligonucleotides,gels were dried at room temperature, denatured, and neutralized.In-gel hybridization was preformed as described previously (16).Alternatively, DNA was blotted onto an N+ Hybond membrane, UVcross-linked, and hybridized. Signals were quantified usinga phosphorimager screen and ImageQuant software (Molecular Dynamics).

To quantify the extent of MC loss, gels were run in duplicate:gel 1 was in-gel hybridized using a radiolabeled (TTAGGG)₄ probe,whereas gel 2 was blotted onto Hybond N+ membrane and sequentiallyhybridized with probes for the 50-bp repeat, 177-bp repeat,and individual VSG genes; 177- and 50-bp probes were obtainedby PCR amplification from genomic DNA as previously described(18). Hybridized membranes were exposed to phosphorimager screens,and signal intensities of individual bands were quantified usingImageQuant software. To analyze the reduction in MC 177-bp signal,50-bp repeat and VSG probes were used to normalize loading.

Slot blot hybridization. A 0.4- to 0.8-µg quantity of DNA from wild-type and long-term-culturedTERT-deficient strains was added to 6x SSC (900 mM NaCl, 90mM sodium citrate, pH 7.0), boiled for 10 min, and placed onice. Duplicate samples (200 and 400 µl) of denatured DNAwere applied to a vacuum manifold device containing a 6x SSC-equilibratedHybond N+ membrane, which was subsequently UV cross-linked andhybridized with 177-bp and 50-bp repeat probes. Signal intensitieswere quantified using phosphorimager screens.

RESULTS

Top

Results

Long-term consequences of telomerase deletion. Deletion of the TERT gene leads to progressive shortening ofthe T. brucei telomere tract, at a rate of 3 to 6 bp/populationdoubling (16). The average telomere of the Lister 427 laboratorystrain used in these studies is $~$ 15 kb (41). To monitor long-termtelomere shortening and its consequences, several independentTERT^–/– clones were continuously cultured withoutany selection regimen for $~$ 2.5 years. Telomeres of wild-typeT. brucei grow at a constant rate of 7 to 9 bp/PD with no apparentlimit during continuous propagation in the laboratory (3, 46).After $~$ 1 year in culture, telomeres of a TERT-deficient cloneshortened significantly, and a dramatic reduction of telomeresignal was evident after $~$ 2.5 years of continuous propagation(Fig. 1A). The six strong bands in Fig. 1A represent classesof telomeres of similar length and not six individual telomeres.A slowly migrating band also appeared in extensively propagatedTERT-deficient strains (Fig. 1A). As judged by two-dimensionalgel electrophoresis with DNA from human ALT cells as a positivecontrol, this band does not appear to represent circular double-strandedDNA (data not shown). In-gel hybridization of native agarosegels, using strand-specific probes, excluded the possibilitythat this DNA is single stranded, and preliminary experimentsshowed that the band is susceptible to Bal 31 digestion, indicatingthat it is either nicked or terminal (data not shown). We havebeen thwarted in all our attempts to determine the origin andsignificance, if any, of this band.

View larger version (77K):
[in this window]
[in a new window]
FIG. 1. Telomere loss in telomerase-deficient T. brucei. (A) Lanes 1 to 3 show wild-type telomeres. Due to their abundance and length, they are not well resolved on this gel. Telomerase-deficient clones exhibit gradual telomere shortening at a rate of 3 to 6 bp/PD during 6-week time courses after 1 or 2.5 years in continuous culture. DNA was digested with MboI and AluI and hybridized using a (TTAGGG)₄ probe. Equal loading was ensured by fluorimetric quantification of DNA prior to loading. The asterisk indicates the appearance of a new heterogeneous band in TERT-deficient cells. (B) Telomere loss on different chromosome types. MBC were separated from IC and MC. Left panel: telomeric signal decreased in clones cultured for 0.5, 1.0, and 2.5 years. Right panel: a duplicate gel hybridized with a 50-bp probe as a loading control. The arrowhead marks rearrangements among IC. WT, wild type.

To measure the decline in telomere signal in different chromosomeclasses, gels were dried and probed with a radiolabeled (TTAGGG)₄probe (Fig. 1B). To ensure equal loading, the telomeric signalwas normalized to a 50-bp repeat signal, which is uniquely presentin long arrays upstream of ES promoters (Fig. 1B, right panel).Comparable results were obtained when a single-copy gene wasused as a loading control (data not shown). Quantitative analysisof these blots revealed that the telomere signal on MBC declinedto $~$ 12% of its initial intensity after $~$ 2.5 years in culture.Surprisingly, the telomere signal on MC reproducibly decreasedto 6.5% (±1.2%) (Fig. 1B, middle panel). We observedone example of a rearrangement among MCB, which is consistentwith the very infrequent rearrangements observed many yearsago and with the fact that the diploid chromosomes of T. bruceican differ greatly in size, due to variations in long chromosome-internalrepetitive regions, transposon-like sequences, and other pseudogenesthat accumulate in large subtelomeric regions and possibly facilitaterecombination (39, 42). We also noticed DNA rearrangements withinthe IC region of the gel (Fig. 1B, left panel). By separationof individual IC and hybridization with IC-specific genes, weconfirmed the origin of these rearrangements to IC (data notshown). These changes could be the result of chromosome fusionsor rearrangements. However, since we have very little understandingof IC function or relevance, we did not investigate these eventsfurther. To investigate what could account for this differenceof telomere loss between MBC and MC and to study the effectsof telomere loss on MC stability, we analyzed the karyotypeof TERT-deficient cell lines in more detail by hybridizationto a 177-bp repeat motif that is unique to MC. The 177-bp repeatsignal decreased to approximately half of its initial intensity,relative to 50-bp and MBC-internal (data not shown) loadingcontrols, over the course of $~$ 2.5 years (Fig. 2A). These resultswere reproducible for independently propagated clones.

View larger version (55K):
[in this window]
[in a new window]
FIG. 2. Loss of minichromosomes in telomerase-deficient T. brucei. (A) Left panel: whole chromosomes, separated by RAGE and analyzed by Southern blotting. Upper panel: 50-bp repeat probe serves as loading control (the arrowhead indicates IC rearrangements). Lower panel: the same gel upon rehybridization with MC-specific 177-bp repeat probe (only the MC part of the gel is shown). Right panel: phosphorimager quantification of MC signal. The 177-bp signal was normalized to the 50-bp signal; bars represent the average 177-bp signal from three independent experiments (standard deviation is indicated) with up to three clones. (B) Left panel: slot blots using the 50-bp repeat probe as loading control and the 177-bp probe to confirm MC loss in long-term-cultured TERT^–/– strains (clones 1 and 2). Right panel: quantified results of four experiments.

If MC had fused to each other, the 177-bp signal could havebeen redistributed over the gel and thereby reduced in the designatedMC region. Although we did not observe any obvious chromosomefusions, we also quantified MC in unfractionated DNA by "slotblotting" (Fig. 2B). Quantitative phosphorimager analysis forfour independent experiments confirmed that the 177-bp MC signalwas reduced to $~$ 50% of its initial intensity (Fig. 2B). We attributethe decline of the 177-bp signal to loss of MC, which can alsobe detected by ethidium bromide staining (data not shown). Thisinterpretation is consistent with the 50% reduction in TTAGGGsignal on MC compared to MBC.

No significant rearrangements were detected in MBC using 50-bprepeat, VSG gene, and chromosome-internal hybridization probes(Fig. 1 and 2 and data not shown). Despite dramatic telomereshortening, MC loss, and rearrangements among IC in long-term-culturedtelomerase-deficient strains, we could not detect any growthor viability problems, as judged by growth curves and platingassays (data not shown).

Telomere length stabilization in the absence of telomerase. We monitored several MBC telomeres for $~$ 2.5 years. Once theybecame exceptionally short, the fate of individual telomereswas followed by hybridization to unique telomeric VSG genesduring further propagation. VSG 121 and bR2 telomeres were chosenfor analysis because they had become particularly short. Thereare three copies of VSG 121, one of which is telomeric. Thetelomeric restriction fragment, released by digestion with EcoRI,gradually shortened during the first 2 weeks of observation(Fig. 3A). Between weeks 3 and 5, the rate of shortening decreased,and the length of this restriction fragment ultimately stabilized(weeks 6 to 11). The length of the VSG 121 chromosome end remainedstable during several additional months of continuous culture(data not shown). We confirmed this result by analyzing theVSG 121 telomere in a second independent clone, using a differentrestriction enzyme to liberate the terminal fragment (Fig. 3B).

View larger version (40K):
[in this window]
[in a new window]
FIG. 3. Telomere length stabilization at dramatically shortened VSG 121 and VSG bR2 telomeres during continuous culture of telomerase-deficient clones (1 week corresponds to $~$ 24 population doublings). Bands corresponding to chromosome-internal copies of VSG 221 and bR2 do not change in size. (A and B) Stabilization of the VSG 121 terminal restriction fragment (asterisk) liberated by EcoRI (A) or BglI (B) digestion in two clones after 6 or 10 weeks of culture (chromosome-internal BglI fragments migrate higher and are not shown). (C) EcoRI liberates the two telomere-linked copies of VSG bR2. One copy is located in an ES with a long telomere (arrowhead), and the second copy is located in an ES with a shorter telomere, whose length stabilized after 12 weeks (asterisk). (D) The population from panel C was cloned at week 14, and one clone was propagated for an additional 13 weeks. The short telomere (asterisk) remained stable throughout the remaining 13 weeks, whereas the longer terminal restriction fragment continuously shortened (arrowhead).

To exclude the possibility that stabilization was occurringexclusively at the VSG 121 telomere, we analyzed a silent EStelomere harboring VSG bR2 (Fig. 3C and D). There are four copiesof VSG bR2, two of which are ES-linked. Nontelomeric VSG bR2restriction fragments did not change in size but the ES-linkedfragments gradually shortened. One ES-linked copy of VSG bR2had a short telomere that allowed good resolution of size changeson an agarose gel. This telomere progressively shortened upto week 12 and then stabilized. At week 14, a clone was pickedout of the population and kept in culture for an additional13 weeks (Fig. 3D). The VSG bR2 band was initially sharp (week16). Over time, the average length of the terminal fragmentdid not change significantly (compare Fig. 3C and D), but theband became more heterogeneous. Upon further culturing, telomerestabilization also occurred on chromosome arms harboring VSG224 and VSG 1.8 (data not shown).

To confirm that the stabilized short chromosome ends remainedterminal and did not fuse with other chromosome arms, we treatedthe DNA with the exonuclease Bal 31, after which the DNA wascut to liberate terminal restriction fragments that could bevisualized with VSG gene probes. Telomere-linked copies of VSG121 and VSG bR2, regardless of whether they were stabilizedor not, were highly susceptible to Bal 31 treatment (Fig. 4),indicating that they remained terminal. Nontelomeric copieswere unaffected by Bal 31. Telomerase-deficient S. pombe cansurvive by circularizing its chromosome (43). Circularized chromosomesremain trapped in the slot during RAGE separation. Using VSG121 and bR2 as hybridization probes, we verified that the chromosomesharboring these VSG genes remained linear and that VSG 121 andbR2 did not recombine onto other chromosomes (data not shown).

View larger version (61K):
[in this window]
[in a new window]
FIG. 4. Short stabilized restriction fragments are terminal. Genomic DNA was digested with exonuclease Bal 31. Stabilized terminal restriction fragments containing VSG 121 (left panel) or VSG bR2 (right panel) were liberated by EcoRI digestion and were susceptible to Bal 31 digestion (asterisk), whereas chromosome-internal copies (arrowheads) were not.

Taken together, telomere length analyses of different chromosomearms over several months indicated that, upon reaching a criticallength, terminal restriction fragments were stabilized by atelomerase-independent mechanism. This mechanism seems to actexclusively at critically short telomeres, since longer telomerescontinued to progressively shorten. Such a distinction has notbeen observed in other species in the absence of telomerase.

Cloning and sequencing of short stabilized telomeres. Up to this point, we did not know whether short stabilized chromosomearms still contained telomeric repeats or whether they weremaintained by amplification of subtelomeric sequences. To addressthis question, the short stabilized VSG 121 chromosome end wascloned and sequenced using a telomere tailing method (19). Poly(C)tails were added to purified terminal restriction fragmentsby using terminal transferase. VSG 121 and poly(G) primers wereused to PCR amplify the region encompassing the 3' end of theVSG 121 ORF, the subtelomeric region, and the end of the chromosome.PCR products were ligated into a cloning vector and analyzedby restriction digestion and sequencing. The size of the clonedPCR products (Fig. 5A) matched the size range observed on Southernblots and in the uncloned PCR products (data not shown). Sequencing(Fig. 5B) revealed terminal TTAGGG repeat tracts that rangedfrom 41 to 200 bp, matching the size heterogeneity observedby Southern blotting and PCR amplification (Fig. 3 and datanot shown). The end of the sequence varied among individualclones. By adapter cloning of telomere ends, it was previouslyshown that T. brucei telomeres terminate in TTAGGG (11). Thisand other work suggested that the templating region of telomeraseRNA ends in AATCCC-5' (6). To verify that no changes occurredin the subtelomeric region, we PCR amplified this region byusing VSG 121 and (CCCTAA)₅ primers. Comparison of wild-typeand mutant cells showed no amplifications or rearrangementsin the $~$ 200-bp subtelomeric region between the VSG gene and theTTAGGG repeat tract. In conclusion, our results indicate thata potentially novel mechanism maintains 40 to 200 bp of terminalTTAGGG repeats in the absence of telomerase.

View larger version (49K):
[in this window]
[in a new window]
FIG. 5. Terminal sequence of short stabilized telomeres. (A) Terminal restriction fragments were C tailed, PCR amplified, and cloned. Size variation among the cloned products reflected the heterogeneity observed on Southern blots and in the uncloned PCR product. (B) Six representative sequences showed 7 to 27 perfect hexameric repeats; the terminal nucleotide varied.

Rates of telomere elongation at transcriptionally silent and active ES. Previous studies showed that an artificially seeded short telomereat the transcriptionally active ES was rapidly elongated bytelomerase (29). To determine the rate of elongation at naturallyshortened stabilized telomeres and to verify that these shorttelomeres could be extended by telomerase, we restored TERTto cells that had been in culture for $~$ 2.5 years and studiedthe dynamics of telomere elongation. As shown in Fig. 6A, ashort 3.5-kb terminal restriction fragment containing the activelytranscribed VSG 221 and less than 1 kb of telomeric repeatswas rapidly elongated. The rate of elongation, measured fromthe initial day of complementation to the first possible timepoint at which sufficient DNA could be obtained from the cells(PD 28), was $~$ 160 bp/PD. This experiment was repeated with sixindependently complemented clones with similar results, so weare confident the rapid elongation is a consequence of telomeraserestoration rather than random recombination in individual clones.When the elongated telomere reached 8 to 10 kb, the rate ofelongation decreased to $~$ 6 to 8 bp/PD (Fig. 6A, right panel).Although the resolving power of an 0.8% agarose gel is limitedas telomere restriction fragments became large, the elongationrate clearly remained at $~$ 6.5 bp/PD for the last 7 weeks of thetime course. When telomerase was restored to clones having differenttelomere lengths at the active ES, the rate of telomere elongationwas inversely related to initial telomere length.

View larger version (44K):
[in this window]
[in a new window]
FIG. 6. Telomere elongation dynamics at transcriptionally active and silent ES telomeres of various lengths after restoration of telomerase. The initial telomere length and restriction fragments used are schematically represented under each panel. (A) Left panel: a short transcriptionally active VSG 221 ES telomere was rapidly elongated. As the telomere reached a length of $~$ 8.5 kb, the rate of elongation decreased. Right panel: elongation of a 6.5-kb active VSG 224 ES telomere. During the initial $~$ 90 population doublings, the rate of telomere elongation was $~$ 25 bp/PD, but it subsequently decreased to $~$ 6.5 bp/PD. (B to D) Elongation of three short stabilized silent ES telomeres (VSG 224, VSG 121, and VSG bR2) after telomerase restoration. (E) Two silent ES-linked copies of VSG 1.8, with very different initial telomere lengths, are elongated at the same rate. (F and G) Long VSG bR2 and VSG 121 telomeres elongate at a constant rate in wild-type T. brucei.

In contrast, the rate of elongation at transcriptionally silenttelomeres was only 6 to 8 bp/PD when telomerase was restored,regardless of their initial length. Telomeres harboring VSG224, VSG 121, and one copy of VSG bR2 were short (40 to 200bp) and stable, prior to reintroduction of telomerase (Fig.6B, C, and D). The same extension rate was observed at the largerof the two VSG bR2 telomeres (Fig. 6D) and at two longer telomeresof $~$ 4.2 kb and 1.3 kb at the VSG 1.8 ES (Fig. 6E).

Over 20 years ago, it was observed that T. brucei telomeresof 5 to 9 kb grow at a constant rate (3, 46). We wondered whether,in accordance with the conventional protein-counting model oftelomere length homeostasis, elongation would cease when a telomerereached a specific length. To address this question, we turnedto extensively propagated wild-type T. brucei and found thatsilent ES telomeres longer than 10 or 15 kb continue to growat a rate of 6 to 8 bp/PD (Fig. 6F and G).

In conclusion, very short transcriptionally silent and activeES telomeres have very different elongation dynamics. Shortactive ES telomeres are very rapidly elongated by telomerase.As telomere length increases, the rate of elongation decreasesto 6 to 8 bp/PD, as at silent ES.

DISCUSSION

Top

Discussion

We monitored telomere loss, in the absence of TERT, at severalMBC and IC during $~$ 2.5 years. Shortening of nontranscribed telomeresoccurred at a constant rate of 4 to 6 bp/PD, and we did notobserve any telomere-telomere recombination. Telomere shorteningled to the loss of many apparently nonessential MC and to rearrangementsamong IC. MBC remained stable. Several possibilities could accountfor these differences. Loss of MBC, which contain essentialgenes, would presumably be lethal. IC appear only to containES, which can be active or silent, so rearrangements or lossmay be tolerated to some degree. It is possible that MC withvery short telomeres could be more susceptible to nucleolyticdegradation. MBC and MC exhibit different positional dynamicsduring mitosis (12, 18, 24). Although segregation of MC normallyoccurs with high fidelity, as shown by their stable inheritanceover 5 years of propagation (1), it remains unclear how thelarge number of MC are faithfully replicated and segregated.We speculate that MC might be attached to the mitotic spindlethrough their telomeres. As a consequence of dramatic telomereshortening, MC could be missegregated and give rise to cellslacking a particular MC or containing extra copies. Cells harboringextra MC might require slightly more time to replicate. During2.5 years, cells that lost a minichromosome could have a minutereplication advantage and become dominant within a population.A comparable situation has been observed between different typesof survivors in telomerase-deficient S. cerevisiae. Type I survivors,which maintain their chromosome ends by extensive amplificationof subtelomeric Y' elements, were outgrown by type II (telomere-telomererecombination) survivors, presumably because the extensive amplifications(10% increase in genome size) increased the time needed forreplication (53).

Short MBC telomeres stabilized as they reached a length of $~$ 40to 200 bp. Bal 31 digestion confirmed that the remaining telomericrepeats were indeed terminal and that stabilized chromosomesdid not fuse, suggesting that as little as 40 bp of telomericrepeats is sufficient to stabilize a chromosome end. The terminalnucleotide varied between different clones, which differs fromthe situation in wild-type cells (11). Telomere stabilizationappears to be sufficient to ensure genomic integrity of MBCin a large majority of cells, as there was no discernible impacton long-term population growth. Nevertheless, TERT^–/–T. brucei might have an enhanced propensity to undergo homologousrecombination, similarly to sir/tlc-deficient yeast cells thatbecome survivors without going through crisis (34). In Drosophila,chromosome end protection depends upon heterochromatin protein1, which binds chromosome ends in a sequence-independent manner.It has been suggested that the epigenetic state of the chromatinrather than a particular sequence mediates end protection (33).It is possible that protection of short silent ES telomeresis in part mediated by the epigenetic state of its chromatin,but we have no information on this.

We did not detect extrachromosomal telomeric circles, whichcould explain why short telomeres remained stable during severalmonths of continuous culture and did not undergo the dramatictelomere elongation observed in human ALT cells and type IIS. cerevisiae survivors (10, 48, 53).

Immortal human ALT cells are characterized by increased telomerelength and telomere heterogeneity and by the presence of extrachromosomaltelomeric circles and ALT-associated PML bodies (5, 8, 26, 57).There is evidence that telomere-telomere recombination accountsfor the increase in telomere length in ALT cells (17). It hasalso been suggested that extrachromosomal telomeric circlescould serve as a template for rapid rolling-circle-based elongationof telomeres (14, 17, 54). However, telomere fluorescent insitu hybridization on metaphase nuclei of ALT cell lines showsmany chromosome arms with little or no telomeric signal (47).The lack of subtelomeric marker genes in ALT cells has hamperedefforts to follow the fate of individual telomeres. A recentreport identified a human ALT cell line that maintains shortyet stable telomeres in the absence of characteristic ALT markers(7). In this cell line, the average telomere length was $~$ 5 kb,as assessed by telomere blotting of RsaI/HinfI-digested genomicDNA. The precise subtelomeric locations of the RsaI and HinfIsites are unknown, but the size of individual telomeres couldbe shorter than 5 kb. Could the mechanism that stabilized telomerelength in this cell line be similar to the mechanism that stabilizesT. brucei telomeres as short as 40 bp?

We propose a speculative model that would restrict the extentof telomere elongation and lead to the telomere stabilizationthat we observe. Telomeres of human, mouse, plants, and T. bruceican invade upstream tandem repeats, giving rise to a loop-likestructure called the t-loop (9, 22, 23, 41). During telomerestabilization, a conventional DNA polymerase could extend theinvading 3' telomere terminus (intrastrand break-induced replication)(Fig. 7A). The displaced strand would give rise to a bubble.Subsequent unwinding of the downstream double-stranded region,realignment, and lagging-strand synthesis would complete replication.The extent of elongation would be limited by the size of theloop, which in turn is limited by the length of the telomeretract. Elongation could thus be restricted and lead to telomerestabilization. Alternatively, interstrand break-induced replication,in conjunction with machinery that tightly controls elongationto 40 to 200 bp, could facilitate telomere stabilization inthe absence of telomerase (Fig. 7B).

View larger version (28K):
[in this window]
[in a new window]
FIG. 7. (A) Speculative model for telomerase-independent telomere stabilization by intrastrand break-induced replication. Conventional DNA polymerases could extend the invaded 3' end, using the C-rich strand as a template. Extension might be limited by the 5' end of the telomere terminus. Subsequent unwinding and strand displacement, followed by lagging-strand synthesis (in red), could result in a limited extension of short telomeres. The RNA primer is shown in green. (B) An alternative model for stabilization by interstrand break-induced replication in conjunction with some mechanism that tightly limits elongation of the 3' terminus to 40 to 200 bp. (C) Model for telomere length regulation in T. brucei. Left panel: silent ES telomeres. Telomere termini are tucked away, forming a t-loop structure (41). During some particular stage of the cell cycle, t-loops are resolved, and telomerase accesses the short 3' overhangs and adds 6 bp ending in TTAGGG (11). Elongation occurs at the same rate regardless of the initial telomere length. Right panel: short active ES telomere. VSG gene transcription extends into the telomere, disrupts telomere structure (such as a t-loop), and allows telomerase unrestricted access to extend the 3' end. As the telomere length increases, transcription does not reach the telomere end, allowing the assembly of a telomerase-inaccessible structure. As a consequence, the extension rate decreases to that of transcriptionally silent telomeres.

Short stabilized telomeres in telomerase-deficient T. bruceienabled us to study telomere length regulation upon reintroductionof telomerase. Telomere length regulation has been studied innumerous organisms. In human cells, telomere length homeostasisis regulated by a negative feedback loop, executed through numeroustelomere-binding proteins that relay information about telomerelength to the telomere terminus (51). How these proteins accomplishthis task remains unclear, but recent work suggested that chromatinremodeling (or t-loop resolution) could regulate telomeraseaccess to telomere termini (52). However, this protein-countingmodel does not appear to apply to T. brucei, where transcriptionallysilent telomeres were elongated at the same rate, regardlessof their initial telomere length. Furthermore, during in vitroculture, telomere homeostasis is not achieved even at long telomeres,which appear to grow indefinitely. This result implies thatthe negative feedback loop that regulates telomere length inother organisms is absent from transcriptionally silent telomeresin T. brucei.

In contrast to silent ES, a short active ES telomere was elongatedat $~$ 160 bp/PD. As the length of an active ES telomere increased,the rate of elongation decreased until it matched the rate atsilent ES. The only known difference between active and silentES is transcriptional activity, and it has been shown that transcriptionextends into telomeric repeats (50). Very recently, a directcorrelation between the strength of an inserted subtelomericpromoter and the rate of telomerase-dependent extension at theadjacent short seeded telomere was demonstrated (20). Basedon all these observations, we propose a model in which transcriptiondisrupts telomeric chromatin and enhances telomerase accessand elongation (Fig. 7C). After rapid elongation, transcriptionno longer reaches the telomere terminus, which can now reassembleinto a t-loop, rendering the 3' end inaccessible during mostof the cell cycle. The elongation dynamics now match those oftranscriptionally silent ES.

In conclusion, we speculate that T. brucei telomeres play animportant role in a variety of cellular processes, includingchromosome segregation, genome integrity, and the regulationof VSG gene switching. Furthermore, due to numerous unique subtelomericVSG genes, T. brucei telomerase-deficient mutants offer a uniqueopportunity to study the molecular mechanism underlying telomerestabilization, which may enhance our understanding of telomeremaintenance in some cancerous human cells.

ACKNOWLEDGMENTS

We are grateful to Titia de Lange and Joachim Lingner for helpfuldiscussions and protocols, to Jenny Li for technical assistancewith growing trypanosomes in animals, to Sean Rooney for humanALT DNA, to Luisa Figueiredo and Bibo Li for inspiring discussionsand probes, and to Veena Mandava and Christian Janzen for discussionand critical reading of the manuscript.

This work was supported by NIH grant AI 50614.

FOOTNOTES

* Corresponding author. Mailing address: Laboratory of Molecular Parasitology, The Rockefeller University, 1230 York Avenue, New York, NY 10021. Phone: (212) 327-7571. Fax: (212) 327-7845. E-mail: george.cross{at}rockefeller.edu.

REFERENCES

Top