Consequences of Telomere Shortening at an Active VSG Expression Site in Telomerase-Deficient Trypanosoma brucei

The cell line used in these experiments, Lister 427 antigenic type MITat 1.2, designated single marker, expresses T7 RNA polymerase, a Tet repressor, and neomycin phosphotransferase and stably expresses VSG 221 (39). Generation of telomerase-deficient clones has been described previously (9). Parental single marker and mutant cells were cultured in HMI-9 containing 2.5 μg/ml G418 (Sigma) at 37°C (7, 11).

Genomic DNA was isolated as previously described (22). Terminal restriction fragments containing specific VSG were detected by genomic blotting, hybridization, and phosphorimaging (14). By use of ImageQuant software, each lane in Fig. 1A was partitioned into 30 equally sized rectangles. The signal intensity in each rectangle was measured as a percentage of the total signal in the entire lane and graphically represented as a function of telomere length.

DNA agarose plugs were prepared as described previously by Navarro and Cross (26). Briefly, 2 × 10⁸ cells were harvested, washed in TDB (5 mM KCl, 80 mM NaCl, 1 mM MgSO₄, 20 mM Na₂HPO₄, 20 mM glucose, pH 7.7) resuspended in 0.5 ml L buffer (0.1 M EDTA, pH 8.0, 10 mM Tris-HCl, pH 7.6, 20 mM NaCl), and incubated for 10 min at 42°C. Five-tenths milliliters of 1.6% low gelling agarose (Sigma) in L buffer was added to 0.5 ml cells, mixed, and poured into plug molds (Bio-Rad Laboratories). Plugs were treated in 3 ml L buffer for 2 days with 1 mg/ml proteinase K at 50°C. After two washes for 15 min with L buffer, proteinase K treatment and washing were repeated. Plugs were stored at 4°C until used. Plugs were embedded in 0.8% agarose in 0.5× Tris-borate-EDTA (TBE) gels. To separate megabase chromosomes, a two-window program was used, consisting of a 100- to 300-s linear ramp pulse time at 120 V for 10 h followed by a 1,000- to 2,500-s linear ramp pulse time at 50 V for 80 h. The rotation angle was 106° at a temperature of 12°C (26). Gels were stained for 45 min in TBE plus 0.02 μg/ml ethidium bromide and destained in TBE for 30 min prior to being photographed under UV light. Southern blotting and hybridization were performed as described above.

Digestion of DNA in genomic plugs was performed as recommended by the enzyme manufacturer (New England Biolabs). Sixty units of ApaI was incubated in New England Biolabs buffer 4 plus bovine serum albumin at 25°C overnight. Plugs were subsequently washed in 0.1 M EDTA, pH 8.0, and 10 mM Tris-HCl, pH 7.6. Restriction fragments were separated on a 0.8% agarose gel in 0.5× TBE. We used a program consisting of a 1- to 12-s linear ramp at a constant 150 V for 15 h at 12°C. The rotation angle was 120°.

T. brucei mRNA was isolated by using RNAstat60 according to the manufacturer's protocol. The dominant VSG expressed by the population was cloned by reverse transcriptase PCR (RT-PCR) using oligonucleotides corresponding to the T. brucei spliced leader sequence (5′-GACTAGTTTCTGTACTATAT-3′) and a VSG C-terminal conserved region (5′-GACTAGTGTTAAAATATATCA-3′). PCR was carried out for 30 cycles of 1 min at 94°C, 1 min at 40°C, and 1 min at 72°C.

Telomere length changes at the active VSG 221 ES can be visualized by digestion with EcoRI and hybridization with a VSG 221 probe (Fig. 1A). The actively transcribed ES telomere is subject to frequent terminal deletions (2, 13, 23, 28, 35). These truncations and their subsequent repair by telomerase account for the telomere length heterogeneity we and others have observed at the active ES in wild-type cells (Fig. 2A, WT lane) (15). In telomerase-deficient cells, breakage and progressive telomere shortening occur. In a newly cloned population, the active ES VSG 221 terminal restriction fragment is a sharp band (Fig. 1A, upper panel, week 0). During subsequent culturing, as a result of telomere breakage in the absence of telomerase, the signal starts to smear towards the bottom of the gel. In contrast, wild-type VSG 221 signal is heterogeneous but remains within a high-molecular-size range throughout the time course (data not shown). To verify equal loading and to confirm the progressive moderate shortening at a silent ES, the blot was rehybridized with a probe against a silent VSG bR2 (Fig. 1A, lower panel). We quantified the extent of telomere loss at the active ES over 8 weeks by measuring signal intensity in each lane as a function of telomere length (Fig. 1B). In several independent experiments, the distribution of telomere signal changed reproducibly over time. We confirmed that loss of VSG 221 and decrease of its transcript levels (Fig. 1C, upper panel) coincided with a gradual increase in VSG 1.8 transcript levels (Fig. 1C, middle panel). By the end of the time course, the population expressed roughly equal amounts of VSG 1.8 and VSG 221, as judged by phosphorimager quantification (see Fig. S1 in the supplemental material), RT-PCR, and sequencing of 30 clones (data not shown), although cells expressing VSG 221 appeared to have a slight growth advantage. In these cells, VSG 1.8 and VSG 221 appeared to be on opposite arms of chromosome VI. The fact that VSG 221 was lost as VSG 1.8 became transcribed suggests that the latter was activated through duplicative gene conversion (see Fig. S2 in the supplemental material). This result led us to investigate the consequences of a short telomere at an active ES in more detail by selecting clones that had lost large amounts of telomeric DNA from the active ES. Six telomerase-deficient clones are shown in Fig. 2A. Clones with very short telomeres, notably, clones 1, 2, 3, and 5, were obtained only after several rounds of continuous propagation and cloning. The sequence between VSG 221 and its telomere is known and indicates that an ∼3-kb terminal restriction fragment contains ∼200 to 400 bp of telomeric repeats (15). As judged by Northern blotting, telomere lengths in this range did not affect VSG 221 mRNA levels at this time (Fig. 2B).

Next, we determined whether further shortening of the active ES telomere would occur and whether it would affect the transcriptional status of VSG 221. We addressed these questions by keeping three independent clones (clones A to C), which already had a very short telomere at the active ES, in continuous culture for 15 weeks, during which we monitored population growth and isolated DNA and RNA every week. We also analyzed the status of VSG 221 at weeks 1, 2, 4, and 6 in an additional eight individual subclones of clone A. The results for weeks 2 and 4 are shown in Fig. S3 in the supplemental material. The short active VSG 221 ES terminal restriction fragment was heterogeneous and ended in ∼200 to 400 bp of telomeric DNA (Fig. 3A, upper panel) but did not shorten further. In contrast to silent ES telomeres harboring VSG bR2 and VSG V02, which continue to shorten at a rate of 3 to 6 bp/PD (Fig. 3A, middle and lower panels) and which also act as equal-loading controls, the VSG 221 signal in the active ES almost disappeared by 5 weeks (Fig. 3A, upper panel). Analysis of individual subclones revealed that some short active ES telomeres became elongated (see Fig S3A and C in the supplemental material), presumably due to repair by break-induced replication (BIR) using another telomere as a template.

The loss of VSG 221 poses two threats to the cell. First, the parasite needs a protective surface coat for survival. Second, loss of VSG 221 and its telomere results in an unprotected chromosome end that requires repair. Since we did not detect any growth retardation, we were interested to learn how the cell addressed these problems. RNA analysis of clone B confirmed that VSG 224 became dominant in the population as VSG 221 transcript levels diminished (Fig. 3B, left four lanes). Two independent clones (clones A and C) also lost VSG 221 over a period of 4 to 5 weeks and also activated VSG 224 (Fig. 3B, right three lanes). In other experiments, VSG 221 was replaced by a different VSG (Fig. 1C).

To clarify the events that led to VSG 224 expression, we generated an ∼9-kb SmaI terminal restriction fragment that hybridized with both VSG 221 and an upstream VSG pseudogene (Fig. 4). After 4 weeks of continuous culture, VSG 221 was again lost from the population. By week 8, the size of the predominant restriction fragments containing the pseudogene increased to ∼23 kb (Fig. 4A, middle panel) and colocalized with a new copy of VSG 224 (Fig. 4A, right panel), suggesting that VSG 224 and part or all of its associated telomere replaced VSG 221 at the previously transcribed ES. In addition to the ∼23-kb bands, a higher molecular band that does not colocalize with the pseudogene appeared (Fig. 4A). We attribute this band to individual breaks in which the SmaI site and the pseudogene marker were deleted. Deletion of the pseudogene occurred after 8 weeks in two analyzed clones, as the duplicated copy of VSG 224 appeared (Fig. 4B and C). To determine whether VSG 224 was activated by gene conversion or reciprocal translocation, we separated whole chromosomal DNA. In all three clones, VSG 221 was lost after ∼4 weeks and VSG 224, initially present as a single-copy gene, was duplicated into the ES that previously carried VSG 221 (Fig. 5A and C).

We independently confirmed that the same ES remained active by using a blasticidin resistance gene that we had integrated immediately downstream of the active promoter (25) while VSG 221 was being expressed. No drug selection was applied during propagation and switching, but the population and individual subclones retained the marker and were highly resistant to blasticidin, indicating that the same ES was in use (Fig. 5B). This result was further confirmed by analyzing individual subclones from each time point. Subclones derived from clone A after 4 or 6 weeks are shown in Fig. 5C. By week 4 subclones 1 and 7 and by week 6 subclones 1, 3, 4, 6, and 8 had all lost VSG 221 (Fig. 5C, upper panel). Most subclones that lost VSG 221 had duplicated VSG 224 (Fig. 5C, lower panel). Subclone 7 from week 4 and subclone 1 from week 6 had lost the VSG 221 gene but did not duplicate VSG 224: these subclones must have undergone independent switch events that activated a different VSG.

Due to the low switching frequency and apparent growth advantage of VSG 221-expressing cells in vitro, our cell lines rarely lose VSG 221 or otherwise result in another VSG becoming dominant in the population, even after extensive propagation. Hence, the appearance of a new VSG at the active ES locus was unexpected and could be explained in two ways. As telomere length at the active ES becomes very short, telomere breakage could lead to terminal truncations and gradual death of the VSG 221-expressing cells in the population. Switched cells, which normally arise at a low frequency, would progressively outgrow the dying cells and lead to the dominance of a new VSG in the population. Alternatively, telomere breaks could be repaired by BIR facilitated by other telomere-linked VSG, with a possible consequent increased rate of antigenic switching.

To address these questions, we attempted to measure switching frequency of strains with a very short active ES telomere by using immune selection in mice. However, due to extensive propagation in laboratory culture or unforeseen consequences of short telomeres in an animal infection that were not evident in culture, these clones did not grow well in mice. Readaptation to mice, during 3 weeks of sequential transfer, appeared to select for recombination-based elongation of the active ES telomere (data not shown).

We also made a mathematical model to predict the growth dynamics of a culture if a short telomere leads to gradual lethality and switchers arise at various frequencies (10⁻³ to 10⁻⁶). Gradual death of cells with a short active ES telomere and concomitant appearance of switchers (at a rate of 10⁻⁶, but with a growth advantage) would affect population growth dynamics only subtly over 4 to 5 weeks. In conclusion, unless the switching frequency is dramatically higher at short telomeres, switchers must have a clear growth advantage in order to outgrow the population within 4 weeks. However, we had to assume parameters, such as relative fitness of short telomere strains and growth advantage of switched parasites, for which we had no experimental data. Thus, we cannot exclude the possibility that switchers, at a frequency of 10⁻⁶, could have outgrown dying parasites without giving rise to a significant decline in population fitness, which would mean that telomere break-induced repair did not necessarily increase the switching rate.