Trypanosoma brucei RAP1 maintains telomere and subtelomere integrity by suppressing TERRA and telomeric RNA:DNA hybrids.

Trypanosoma brucei causes human African trypanosomiasis and regularly switches its major surface antigen, VSG, thereby evading the host's immune response. VSGs are monoallelically expressed from subtelomeric expression sites (ESs), and VSG switching exploits subtelomere plasticity. However, subtelomere integrity is essential for T. brucei viability. The telomeric transcript, TERRA, was detected in T. brucei previously. We now show that the active ES-adjacent telomere is transcribed. We find that TbRAP1, a telomere protein essential for VSG silencing, suppresses VSG gene conversion-mediated switching. Importantly, TbRAP1 depletion increases the TERRA level, which appears to result from longer read-through into the telomere downstream of the active ES. Depletion of TbRAP1 also results in more telomeric RNA:DNA hybrids and more double strand breaks (DSBs) at telomeres and subtelomeres. In TbRAP1-depleted cells, expression of excessive TbRNaseH1, which cleaves the RNA strand of the RNA:DNA hybrid, brought telomeric RNA:DNA hybrids, telomeric/subtelomeric DSBs and VSG switching frequency back to WT levels. Therefore, TbRAP1-regulated appropriate levels of TERRA and telomeric RNA:DNA hybrid are fundamental to subtelomere/telomere integrity. Our study revealed for the first time an important role of a long, non-coding RNA in antigenic variation and demonstrated a link between telomeric silencing and subtelomere/telomere integrity through TbRAP1-regulated telomere transcription.

INTRODUCTION

Subtelomeres are regions immediately upstream of telomeres at chromosome ends. The relatively plastic and fragile environment at subtelomeres allows more frequent DNA rearrangements (1) and facilitates gene conversion-mediated antigenic variation in microbial pathogens such as Trypanosoma brucei that causes human African trypanosomiasis, Pneumocystis jerovecii that causes pneumonia, and Borrelia burgdorferi that causes Lyme disease, as the major surface antigens are transcribed from subtelomeric expression sites in these pathogens (2). However, introduction of a single DNA double strand break (DSB) at certain subtelomeric loci results in lethality in >90% of T. brucei cells (3), underlying the importance of subtelomere integrity. Subtelomere integrity is also important for human health. Deletion of subtelomeric D4Z4 repeats derepresses a nearby DUX4 gene in muscle cells and leads to facioscapulohumeral muscular dystrophy (4). 5% of unexplained human mental impairment cases are caused by cryptic unbalanced subtelomeric rearrangements (5). Additionally, olfactory receptor genes and immunoglobulin heavy chain genes are located at human subtelomeric regions (6).

Regular switching of the major surface antigen, VSG, is an important pathogenesis mechanism in T. brucei, which has >2500 complete and truncated VSG genes (7). VSGs are transcribed by RNA Polymerase I (RNAP I) (8) exclusively from one of 15 different subtelomeric expression sites (ESs) (9,10). Genes in T. brucei are arranged in polycistronic transcription units (PTUs), and ESs are typical PTUs with VSG being the last gene in any ES. Monoallelic VSG expression is regulated at multiple levels, including ES promoter activation and silencing, chromatin structure remodeling, and specialized subnuclear localization of the active ES [reviewed in (11)]. Additionally, ES attenuation (12) and the inositol phosphate pathway (13) also affect VSG silencing. Particularly, silent ES promoters are actually mildly active but transcription only elongates for a few kilobases along the PTU, preventing expression of downstream VSGs (14). Therefore, regulation of transcription elongation along ESs is important for VSG silencing. We have shown that TbRAP1, a telomere protein, is essential for repressing subtelomeric VSG expression by telomeric silencing and have proposed that TbRAP1 helps suppress transcription elongation along ESs (15,16).

VSG switching occurs through two major pathways (Supplementary Figure S1). In an in situ switch, the originally active ES is silenced while a different ES becomes expressed simultaneously. DNA recombination mediates the other major class of VSG switching. In crossover (CO)/telomere exchange (TE), the active VSG and a silent VSG (often with its downstream telomeric DNA) exchange places, resulting in the expression of a different VSG from the same active ES without losing any genetic information. In gene conversion (GC), a silent VSG gene is duplicated into the active ES to replace the originally active VSG gene, which is lost. Usually, the term ‘VSG GC’ is used when GC events encompass only the VSG gene and its neighboring sequences, while ‘ES GC’ refers to events that encompass most of the ES, sometimes including the ES promoter. A number of proteins important for DNA recombination, such as RAD51 (17), RAD51-3 (18), BRCA2 (19), RECQ2 (20), TOPO3α (21) and TOPO3α-interacting RMI1 (22) play important roles in VSG switching. In addition, we and others have shown that telomeres and telomere-associated proteins affect VSG switching (23–25).

Telomeres are essential for protecting the natural chromosome ends from being recognized as DNA breaks, and telomere proteins help prevent chromosome ends from being processed illegitimately (26). Recently, we showed that telomere proteins are also important for maintaining subtelomere integrity and stability (24,25,27). In addition, telomeres suppress expression of nearby genes by telomeric silencing (28). In yeasts, it is well accepted that the telomere heterochromatic structure limits the access of the transcription machinery to promoters of subtelomeric genes and silencing is at the transcription activation level rather than at the elongation step (28), while in T. brucei, we have proposed that the TbRAP1-mediated telomeric silencing helps block transcription elongation (15), although the exact mechanism is unknown.

In many organisms including T. brucei, telomeres are transcribed into a long, non-coding telomeric repeat-containing RNA (TERRA), which contains G-rich telomeric sequences (29,30). TERRA is transcribed by RNA Polymerase II in mammalian cells and yeast (31,32), but its transcription is resistant to α-Amanitin in T. brucei, suggesting that it is transcribed by RNAP I (29). In Saccharomyces cerevisiae, TERRA expression and stability are regulated by the telomere-binding factor RAP1 (33). Recent studies showed that TERRA can form R-loops (a three-stranded structure including an RNA:DNA hybrid and a displaced single-stranded DNA) with the telomeric DNA in yeast and human cells, and a high level of telomeric R-loops promotes telomere recombination events (30). R-loops have been hypothesized to interfere with DNA replication, and R-loop-dependent replication impairment has been implicated in transcription-associated recombination (34). The R-loop structure can be resolved by RNaseH, which degrades the RNA moiety in the RNA:DNA hybrid (35). In telomerase-nagative tumor cells, RNaseH1-depleted cells have increased amount of RNA:DNA hybrids, while overexpression of RNaseH1 reduces the occurrence of telomeric R-loops (36). Whether T. brucei TERRA can form telomeric R-loops and influence nearby VSG switching is unknown.

In T. brucei, TbTRF is the duplex telomere DNA binding factor (37), and it interacts with both TbRAP1 (15) and TbTIF2 (25). We have shown that all three telomere proteins are essential for cell viability, TbRAP1 is essential for VSG silencing (15,16), and both TbTRF and TbTIF2 suppress VSG switching (24,25). However, it was not clear why TbRAP1 is an essential gene and whether TbRAP1 affects VSG switching and the expression of TERRA. By performing an in vitro VSG switching assay we found that a transient depletion of TbRAP1 led to a higher VSG switching frequency with most switchers arising from VSG GC. In addition, TERRA and telomeric RNA:DNA hybrid levels were increased in TbRAP1-depleted cells. Depletion of TbRAP1 also led to more DSBs at telomeres and at ES-linked VSG loci. Importantly, expression of an ectopic allele of TbRNaseH1 in TbRAP1-depleted cells lowered the telomeric RNA:DNA hybrids, telomeric/subtelomeric DSBs, and VSG switching frequency back to WT levels. Therefore, depletion of TbRAP1 increases TERRA and telomeric RNA:DNA hybrid levels, the latter of which facilitates DSB-initiated VSG gene conversion and subsequent VSG switching.

MATERIALS AND METHODS

Examination of telomeric RNA:DNA hybrid

One hundred microgram of genomic DNA was isolated from induced (+Dox for 24 h) and uninduced S/RAP1i and S/RAP1i+RNaseH1-2HA cells and sonicated with a BioRuptor (Diagenode) using medium output for eight cycles with 30 s pulse each. Half of the sonicated samples were treated with 20 U of RNaseH (Thermo Fisher Scientific). Both RNaseH treated and untreated samples were equally divided for IP using normal IgG or S9.6 antibodies. After extensive washing, the immunoprecipitated samples were eluted and loaded onto Nylon (+) membrane followed by Southern analysis using a TTAGGG repeat probe.

VSG switching assay, ligation-mediated PCR, cloning of the active VSG, and Pulsed-Field gel electrophoresis

These were performed exactly the same as described in (25). Additional details are described in Supplemental Information. Primers used for LMPCR are the same as those listed in (25).

Quantitative RT-PCR

Quantitative RT-PCR for estimation of VSG expression levels was performed the same way as in (15).

TERRA northern blotting and slot blot hybridization

Total RNA was purified from 100 million T. brucei cells using RNA STAT-60 (Tel. Test Inc.) twice and treated with 10 units of DNase I (Thermo Fisher Scientific) followed by another round of purification with RNA STAT-60. The resulting RNA sample was treated with or without 20 units of RNase One (Promega) and 20 μg of RNase A (Sigma) (as negative controls). For northern blotting, 10 μg of RNA samples were loaded in each lane. For slot blot hybridization, 2 μg of RNA was spotted on the Nylon membrane. RNA samples were denatured at 65°C for 10 min in the presence of formamide and formaldehyde before separated by electrophoresis. To prepare the (CCCTAA)- [or (TTAGGG)-] specific probe, the Klenow primer extension reaction was performed using a duplex TTAGGG repeats as the template in the presence of dA, dT and radioactive dC (or radioactive dG).

γH2A antibody production

A 10-amino acid peptide (CKHAKATPSV) from the C-terminus of T. brucei H2A was synthesized in two versions with the T residue either phosphorylated or not (ProSci), and the phosphorylated peptide was used to immunize two rabbits (ProSci). Both the phosphorylated and unphosphorylated peptides were coupled with SulfoLink Resin (Thermo Fisher Scientific) for affinity purification of the rabbit serum. Rabbit serum was first affinity purified through the column of unphosphorylated peptide to obtain antibodies recognizing the unphosphorylated H2A. The flow through fraction was subsequently affinity purified through the column with the phosphorylated peptide to obtain antibodies that specifically recognize γH2A.

RESULTS

TbRAP1 suppresses VSG switching

We used the HSTB261 strain established by Kim and Cross (21) to examine the effects of TbRAP1 depletion on VSG switching. HSTB261 (hereafter referred to as the S strain) has a blasticidin resistance (BSD) gene immediately downstream of the promoter and a puromycin resistance gene (PUR) fused with the Herpes simplex Thymidine Kinase (TK) gene between VSG2 and 70 bp repeats in the active ES (Supplementary Figure S1) (21). We introduced the inducible TbRAP1 RNAi construct into S cells to establish the S/RAP1i strain. TbRAP1 is an essential protein, and depletion of TbRAP1 leads to cell growth arrest within 24 h (15). Therefore, to recover viable switchers for estimation of switching frequency, we induced TbRAP1 RNAi for only 24 h and removed doxycycline by extensive washing. This decreased TbRAP1 protein levels (Figure 1A and B), slowed cell growth (Figure 1C), and derepressed ES-linked silent VSGs including VSG13 and VSG9 for ∼48 h (Figure 1B). Forty eight hours after removal of doxycycline, the TbRAP1 protein and cell growth recovered to their normal levels (Figure 1A–C), and ES-linked VSG9 and VSG13 were silenced again (Figure 1B). Importantly, inducing TbRAP1 RNAi by doxycycline in the recovered cells resulted in cell growth arrest again (Supplementary Figure S2A and B). Therefore, short-term depletion of TbRAP1 is completely reversible. As controls, we examined TbRAP1 protein levels in uninduced S/RAP1i cells (Supplementary Figure S2C) and in S (Figure 1A) and S/ev cells (S cells carrying an empty RNAi vector; Supplementary Figure S2D) in the presence of doxycycline. The TbRAP1 protein level in these cells did not change.

Figure 1.

A transient depletion of TbRAP1 increased VSG switching frequency. Western analyses of protein levels of TbRAP1 and tubulin in S/RAP1i and S cells (A) and various ES-linked VSGs in Ri-2 (TbRAP1 RNAi) cells (15) (B) at different time points. For transient induction, doxycycline was only added for 24 h. Cells used in (A) and (B) are VSG2-expressors. (C) Growth curves of S, S/ev, and S/RAP1i cells under induced (+Dox), uninduced (−Dox), and transiently induced (+Dox for 24 h) conditions. Average population doublings were calculated from three independent experiments. (D) VSG switching frequencies in S, S/ev, S/RAP1i, S/RAP1i+F2H-TbRAP1, S/RAP1i+RNaseH1-2HA, and S/ev+RNaseH1-2HA cells with (+) or without (−) a transient induction are shown. B1 and C1 are independent clones of S/RAP1i. Average switching frequencies were calculated from at least four independent assays. P values (unpaired t tests) are indicated. The full height of each column represents 100% of analyzed switching events in the corresponding strain (total number of analyzed switchers is shown). Percentages of different VSG switching events are represented by heights of different colored bars. In (C) and (D), standard deviations are shown as error bars.

We next performed the VSG switching assay. All cells were grown for the same number of population doublings so that the switching frequencies in different strains could be compared fairly. We identified potential switchers by their resistance to ganciclovir (GCV) due to loss of TK transcription and confirmed that they no longer expressed the originally active VSG2 by dot blot analysis (Materials and Methods). The final VSG switching frequencies were normalized against the plating efficiencies determined for all strains (Supplementary Figure S2E).

The VSG switching frequency in TbRAP1-depleted cells was increased 6.1–6.7-fold when compared to that in S/ev cells, while S and the uninduced S/RAP1i cells have similar switching frequencies to S/ev cells (Figure 1D). To confirm that the elevated switching frequency was specifically due to the lack of TbRAP1, we introduced an inducible FLAG-HA-HA (F2H)-tagged TbRAP1 expression vector into S/RAP1i cells. Adding doxycycline to S/RAP1i+F2H-TbRAP1 cells induced F2H-TbRAP1 expression (Supplementary Figure S2F) and resulted in a nearly normal growth (Supplementary Figure S2G) and a switching frequency comparable to that in S/ev cells (Figure 1D), indicating that expression of the ectopic F2H-TbRAP1 suppressed the abnormal switching phenotype caused by TbRAP1 depletion. Therefore, TbRAP1 suppresses VSG switching.

VSG gene conversion is the most frequent switching events in TbRAP1-depleted cells

By examining the genotype (detecting the presence of the VSG2 and BSD genes by PCR, Supplementary Figure S3A and B) and marker expression status in recovered VSG switchers, we determined the switching mechanism in each switcher (Supplementary Figure S1; Tables S1–S6) (21). In S, S/ev, and uninduced S/RAP1i cells, most switchers (68-78%) arose from ES GC or ES Loss coupled with an (ESLIS) switch, and 16–32% of the switchers arose from VSG GC (Figure 1D, Supplementary Table S1–S3). However, in TbRAP1-depleted cells, most switchers arose from VSG GC (62–74%), while fewer switchers arose through ES GC or ESLIS (19–26%) (Figure 1D, Supplementary Tables S5 and S6). The portion of cells that switched through an in situ or a CO/TE did not seem to change significantly upon TbRAP1 depletion (Figure 1D). When the ectopic F2H-TbRAP1 was expressed in S/RAP1i, ES GC or ESLIS were again the most frequent events (50%) and fewer cells switched through VSG GC (43%) (Figure 1D; Supplementary Table S4).

To confirm that we classified switchers correctly, we randomly selected switchers derived from induced S/RAP1i cells and analyzed their karyotypes. We first determined which VSGs were expressed in the recovered switchers by RT-PCR and sequencing analysis (Materials and Methods). Subsequently, undigested genomic DNA isolated from these switchers were separated by Pulsed-Field Gel Electrophoresis (PFGE) and analyzed by Southern blotting using the BSD, the originally active VSG2, and the newly active VSG probes. For all randomly selected switchers, their predicted VSG switching mechanisms were confirmed by karyotype analysis (Supplementary Figure S3C–E). Interestingly, karyotype analysis showed that all analyzed switchers that could have arisen from an ES GC or an ESLIS turned out to be the latter type. Although we cannot perform the extremely laborious karyotype studies for all switchers, this observation, based on a random sample, suggests that ESLIS is preferred to ES GC in S/RAP1i cells.

Depletion of TbRAP1 leads to more DSBs at telomeres and subtelomeres

Removal of TbRAP1 led to more VSG GC, which frequently initiates with DSBs. Therefore, we anticipated that depletion of TbRAP1 might increase DSBs at telomeres and subtelomeres. RAD51 is a recombinase that binds to the single stranded 3΄ overhang after 5΄ end resection at DSB sites and mediates strand invasion during homologous recombination (38), and TbRAD51 associates with the chromatin at break sites (17,18). Similarly, the C-terminally phosphorylated histone H2A (γH2A) is deposited at DNA break sites (39). Therefore, we performed Chromatin IP (ChIP) using the TbRAD51 antibody (18) and a γH2A-specific antibody (see below) followed by Southern slot blotting and quantitative PCR (qPCR) in S/RAP1i cells.

We raised rabbit antibodies that specifically recognize γH2A or unphosphorylated H2A (Materials and Methods). After treating WT cells with 1.5 μg/ml phleomycin that induces DSBs (39), we detected γH2A by western blotting (Supplementary Figure S4A), confirming the specificity of the γH2A antibody (Supplementary Figure S4A).

We performed ChIP using TbRAD51 and γH2A antibodies. Normal rabbit IgG was used as a negative control. The ChIP products were first analyzed by Southern slot blotting using a telomeric probe (Figure 2A and Supplementary Figure S4B) and, as a control, a probe specific to the 50 bp repeats that are located upstream of ES promoters (Supplementary Figure S4B and C). Upon depletion of TbRAP1, more TbRAD51 and γH2A were associated with the telomeric chromatin (Figure 2A) but not with the 50 bp repeats (Supplementary Figure S4C). In S/ev cells, adding doxycycline did not change the amount of telomere- or 50 bp repeats-associated TbRAD51 or γH2A (Figure 2A, Supplementary Figures S4B and S4C). These observations suggest more DNA damage at the telomere upon TbRAP1 depletion.

Figure 2.

Depletion of TbRAP1 resulted in more telomeric/subtelomeric DSBs. (A) ChIP experiments were performed using rabbit antibodies recognizing TbRAD51 (18), γH2A, or normal rabbit IgG (as a negative control) in S/ev, S/RAP1i, and S/RAP1i+RNaseH1-2HA cells before (0 h) and after adding doxycycline (24 h). DNA isolated from ChIP products were hybridized with a TTAGGG repeat probe and exposed to a phosphorimager. The hybridization signals were quantified and averages were calculated from at least three independent experiments. (B) DNA isolated from ChIP experiments was analyzed by quantitative PCR (qPCR) using primers specific to VSG2 (active), VSG21 (silent), ES promoters, tubulin, and rDNA. Average enrichment (ChIP/input) was calculated from 3 to 12 independent experiments. In both (A) and (B), standard deviations are shown as error bars. P values of unpaired t-tests are shown. (C and D) LMPCR analyses were performed in S/RAP1i cells. The LMPCR products were hybridized with ES promoter (C) and VSG2 (D) probes: the Ethidium bromide (EtBr)-stained LMPCR products are shown at the top, the Southern blot result is shown in the middle, and the PCR products using primers specific to the TbRAP1 gene (as a loading control) are shown at the bottom. The amounts of input genomic DNA, from uninduced (−Dox) or induced (+Dox) cells, either treated (+) or untreated (−) with T4 DNA polymerase, are marked on top of each lane.

We also analyzed the ChIP products by qPCR using primers specific to VSG2 (active), VSG21 (ES-linked, silent), and ES promoters (one active and multiple silent), tubulin, and rDNA. As shown in Figure 2B and Supplementary Figure S4D, we observed a significant increase in the amount of ES-associated TbRAD51 and γH2A upon depletion of TbRAP1, suggesting that depletion of TbRAP1 resulted in more DNA damage at ESs. No significant change was observed in S/ev cells before and after adding doxycycline.

Although association of TbRAD51 and γH2A with the chromatin strongly suggests that DSBs occurred at the examined loci, ChIP results are still an indirect evidence for DSBs. We therefore performed Ligation-Mediated PCR (LMPCR; Supplementary Figure S5A) and examined ES regions for DSBs before and after depletion of TbRAP1. LMPCR detects DSBs physically (40,41), although it cannot be used to detect DSBs at telomeres quantitatively due to the repetitive sequence. Interestingly, we detected more LMPCR products upon TbRAP1 depletion when ES-linked VSG probes were used, including the active VSG2 (Figure 2D) and derepressed VSG15 (Supplementary Figure S5B) and VSG21 (Supplementary Figure S5C), and this increase was significant (Supplementary Figure S5G). However, no increase in LMPCR products were seen when ES promoter (Figure 2C) or the 70 bp repeat probes (Supplementary Figure S5D) were used. We did not detect any change in DSB levels at VSG loci in S/ev cells (Supplementary Figure S5F). Depletion of TbRAP1 did not increase the number of DSBs at a random chromosome internal SNAP50 locus (Supplementary Figure S5E), indicating that the effect of TbRAP1 depletion is telomere/subtelomere-specific.

It is worth to note that even in WT cells DSBs are detected in ESs, which is the same as demonstrated previously (42). Therefore, TbRAP1 does not suppress subtelomeric DSBs completely. Additionally, in uninduced S/RAP1i cells, a significant amount of LMPCR products of ∼2 kb was detected without T4 DNA polymerase treatment, indicating that these DSBs are blunt-ended and are located downstream of the active VSG2, as the VSG2 gene is 1.43 kb (Figure 2D). Previously, it has been shown that DSBs downstream of the active VSG lead to more ES GC/ESLIS, while DSBs immediately upstream of the active VSG result in mostly VSG GC (3). Therefore, having DSBs downstream of the active VSG2 is likely the reason that ES GC/ESLIS switchers are the most popular in WT cells. In contrast, in TbRAP1-depleted cells, blunt-ended DSBs are rare and nearly all LMPCR products are of ∼700 bp, indicating that these DSBs are located within the VSG2 gene. Since VSG is essential (43) and gene conversion often initiates with DSBs, it is likely that VSG GC is the most efficient repair mechanism to allow cells recover from this detrimental damage, which would explain why VSG GC switchers are predominant in TbRAP1-depleted cells.

The active ES-adjacent telomere is transcribed

To investigate why depletion of TbRAP1 caused more DSBs at telomeres and subtelomeres, we examined other phenotypes and found that TbRAP1 suppresses telomere transcription (see below).

In T. brucei, a long, non-coding RNA containing the 5΄ UUAGGG 3΄ repeat sequence was identified previously (29). This transcript was not affected by 1 mg/ml α-Amanitin, suggesting that it was transcribed by RNAP I (29). However, where this RNA was transcribed from was unknown. We first performed northern analysis to detect the telomeric transcript. Total RNA was extracted from WT bloodstream form cells and treated with excessive amount of RNase-free DNase I. RNA samples were then hybridized with a (CCCTAA)n telomeric probe (Figure 3A, left), a (TTAGGG) telomeric probe (Figure 3A, right), or a TbTR probe (as a loading control, Figure 3A). We only detected the telomeric transcript (hereafter referred to as TERRA) when the (CCCTAA) probe was used (Figure 3A), confirming that TERRA has a G-rich sequence. TERRA sizes range from a few hundred to ∼4k nucleotides in WT cells (Figure 3A). As a control, an equal amount of RNA was treated with RNase One and RNase A, which abolished the hybridization signal (Figure 3A).

Figure 3.

The active VSG-adjacent telomere was transcribed into TERRA. (A) Northern analysis of RNA isolated from WT cells. The northern blot was hybridized with a (CCCTAA) probe (left panel) or a (TTAGGG) probe (right panel). In each panel, the EtBr-stained gel (left), the TTAGGG/CCCTAA hybridization result (middle), and the hybridization result using a TbTR (50) probe (right, as a loading control) are shown. Both RNA samples treated with or without RNase One and RNase A (RNase) were run. (B) Total RNA was purified from VSG2-expressing WT cells and reverse-transcribed using TELC20, TELG20, a random hexamer (as a positive control), or ddH2O (as a negative control) as the RT primer (labeled beneath each gel). The RT products were PCR amplified using primers specific to VSG2 (active), VSG9 (silent), 70 bp repeats, tubulin and rRNA (marked on top of each lane), and the PCR products were separated on agarose gels.

It was hypothesized that TERRA was a product of RNAP I read-through from the active ES into the downstream telomeric repeats (29), although this was not proved. Because ES genes are polycistronically transcribed and trans-spliced, we reason that there must be some nascent RNA molecules containing both telomeric repeats and the upstream VSG sequence if TERRA is a read-through product from the active ES. We performed an RT-PCR experiment to test this possibility. In cells that express VSG2, total RNA was isolated and RT was performed using a TELC20 (5΄-CCCTAACCCTAACCCTAACC-3΄), a TELG20 (5΄-GGGTTAGGGTTAGGGTAAGG-3΄), or a random hexamer as the primer. An RT without any primer was also performed as a negative control. Subsequently, the RT products were amplified by PCR using primers specific to VSG2 (active), VSG9 (silent), 70 bp repeats, rRNA, and tubulin. We found that TELG20-primed RT products did not yield any PCR products using any primer pairs (Figure 3B), while the TELC20-primed RT products yielded a PCR product only when VSG2-specific primers were used (Figure 3B). As a positive control, all actively transcribed genes (including VSG2, rRNA and tubulin) gave a positive PCR product when a random hexamer was used in RT (Figure 3B). As expected, when RT was done without any primer, no PCR products were observed. This result indicates that TERRA is UG-rich, and VSG2 (but not VSG9) and TERRA are co-transcribed, confirming the hypothesis that TERRA is transcribed from the active ES-adjacent telomere.

To further validate our conclusion, we performed the same RT-PCR in a VSG9-expressor (where VSG2 and VSG3 are silent, Supplementary Figure S6A) and a VSG3-expressor (where VSG8 and VSG9 are silent while the VSG2 gene is lost, Supplementary Figure S6B) (15) and obtained similar results: The TELC20-primed RT products were amplified only when the primers specific to the active VSG were used in PCR (Supplementary Figure S6). Therefore, TERRA is transcribed from the active ES-adjacent telomere but not from silent ES-adjacent telomeres.

Depletion of TbRAP1 increases the TERRA level

Because depletion of TbRAP1 derepresses subtelomeric VSGs (Figure 1B) (15,16), we hypothesize that TbRAP1 also regulates TERRA expression.

We first examined TERRA in S/RAP1i cells by northern blotting as described above. When TbRAP1 was depleted, we observed stronger hybridization signals of much larger sizes (up to ∼10 kb) (Figure 4A). However, in S/RAP1i+F2H-TbRAP1 cells, the level and size of TERRA remained unchanged before and after adding doxycycline (Figure 4A).

Figure 4.

Depletion of TbRAP1 increased the TERRA level. (A) Northern analysis of TERRA [hybridized with the (CCCTAA)n probe] in S/RAP1i and S/RAP1i+F2H-TbRAP1 cells before (0 h) and after adding doxycycline (24 and 48 h). The TbTR hybridization was used as a loading control. (B) Dot blot northern analysis of TERRA in Ri-CT (16) and 2/ev cells before and after adding doxycycline using a (CCCTAA) and a tubulin probe. (C) Quantification of dot blot hybridization. Fold changes in TERRA RNA and tubulin mRNA signal intensities between 0 and 24 h after adding doxycycline were calculated from three independent experiments. Standard deviations are shown as error bars. P values of unpaired t-test are shown.

We also performed northern slot blot analysis to better quantify the amount of TERRA in TbRAP1-depleted cells. In a previously established TbRAP1 RNAi strain [Ri-CT, (16)] and the control strain carrying an empty RNAi vector (2/ev), total RNAs were isolated and slot-blotted onto a nylon membrane, then hybridized with the (CCCTAA) probe and a tubulin probe (as a loading control) (Figure 4B). Quantification of the hybridization signals from three independent inductions indicated that the TERRA level in TbRAP1-depleted cells was ∼11-fold of that in uninduced cells (Figure 4C). The TERRA level was not changed in 2/ev cells before and after adding doxycycline (Figure 4C). Therefore, loss of TbRAP1 leads to an increase in the TERRA level.

Depletion of TbRAP1 results in derepression of silent ESs (15,16). Therefore, we hypothesize that TERRA is expressed from the derepressed ES-adjacent telomeres in addition to the fully active telomere upon TbRAP1 depletion. To test this, we performed the same RT-PCR experiment described above. In uninduced S/RAP1i cells, only primers specific to the active VSG2 yielded a PCR product with the cDNA generated from the TELC20-primed RT reaction (Figure 5A), indicating that TERRA was transcribed from the active VSG2-adjacent telomere but not from silent VSG3 or VSG9-adjacent telomeres. As expected, RT using TELG20 or no primer did not yield any PCR products (Figure 5A), and RT using the random hexamer yielded positive PCR products for active genes including VSG2, tubulin and rDNA (Figure 5A).

Figure 5.

The active VSG-adjacent telomere was transcribed in TbRAP1-depleted cells. The same RT-PCR experiment as described in Figure 3B was performed in S/RAP1i cells before (A) and after (B) induction of TbRAP1 RNAi. RT primers were listed below each gel, while primers used in PCR amplification were listed on top of each lane.

After TbRAP1 RNAi was induced for 24 h, we were able to obtain PCR products for all active (VSG2, tubulin and rDNA) and derepressed (VSG3, VSG9 and even 70 bp repeats) genes when the random hexamer was used in RT (Figure 5B), indicating that depletion of TbRAP1 was successful. Surprisingly, when TELC20 was used as the RT primer, a positive PCR product was only obtained with VSG2 primers but not with VSG3 or VSG9 primers (Figure 5B). Therefore, even though depletion of TbRAP1 results in derepression of ES-linked VSGs and increases the TERRA level, TERRA is still mainly expressed from the active VSG-adjacent telomere but little from derepressed VSG-adjacent telomeres, if any.

Depletion of TbRAP1 leads to an increased amount of telomeric RNA:DNA hybrids

In human and yeast cells, TERRA has been shown to form RNA:DNA hybrids with the telomeric DNA, which presumably form a telomeric R-loop structure. R-loops increase DSB formation and thereby promote DNA recombination (30). Because a transient depletion of TbRAP1 leads to increased VSG GC events at subtelomeres and increased TERRA level, we were curious whether TbRAP1 depletion would also increase the number of telomeric RNA:DNA hybrids.

The S9.6 monoclonal antibody specifically recognizes the RNA:DNA hybrid (44) and has been routinely used to examine R-loops (34). We used S9.6 to immunoprecipitate (IP) all RNA:DNA hybrids from S/RAP1i cells before and after induction of TbRAP1 RNAi. The IP products were then analyzed by dot blot hybridization using the (TTAGGG) probe. We observed an increased telomeric RNA:DNA hybrid signal in TbRAP1-depleted cells (Figure 6A). Treating genomic DNA with RNaseH (Thermo Fisher Scientific) before S9.6 IP reduced the precipitated telomeric signals to the background level, confirming that the detected signals are from an RNA:DNA hybrid structure. Quantification of the hybridization signals indicates that the amount of telomeric RNA:DNA hybrids in S/RAP1i cells increased nearly seventeen fold after induction of TbRAP1 RNAi (Figure 6B). Therefore, depletion of TbRAP1 results in many more telomeric RNA:DNA hybrids.

Figure 6.

Depletion of TbRAP1 led to increased amount of telomeric RNA:DNA hybrids, which was suppressed by ectopic expression of TbRNaseH1. (A) Dot blot Southern analysis of input (diluted 20-fold), IgG and S9.6 immunoprecipitated DNA samples in S/RAP1i cells (top) and S/RAP1i+RNaseH1-2HA cells (bottom) before (–Dox) and after (+Dox) adding doxycycline for 24 h. Samples were treated with or without RNaseH (Thermo Fisher Scientific) before IP. A (TTAGGG) probe was used in the hybridization. (B) Quantification of the dot blot hybridization signals. Average values were calculated from at least four independent experiments. Standard deviations are shown as error bars. Unpaired t-test P values are indicated.

To determine whether the highly transcribed active VSG mRNA and the derepressed VSG mRNA in TbRAP1-depleted cells also form an R-loop structure, we performed the same S9.6 IP followed by qPCR using primers specific to different VSG genes. However, we did not detect any VSG sequence-containing RNA:DNA hybrids in S/RAP1i cells before or after induction of RNAi, indicating that the amount of RNA:DNA hybrids formed with the VSG mRNA, if any, is extremely low.

Expressing an ectopic WT allele of TbRNaseH1 in TbRAP1 RNAi cells reduces the telomeric RNA:DNA hybrids back to WT levels

RNA:DNA hybrids can be resolved by RNaseH, which cleaves the RNA strand in the RNA:DNA hybrid (35). In T. brucei, TbRNaseH1 (Tb427.07.4930) was characterized previously (45). We therefore cloned a C-terminally HA-HA (2HA) tagged TbRNaseH1 into an inducible expression vector, pLew100v5 (13), and transfected it into S/RAP1i and S/ev (as a control) cells. In S/ev+RNaseH1-2HA cells, TbRNaseH1-2HA was only expressed upon adding doxycycline (Supplementary Figure S7A). In S/RAP1i+RNaseH1-2HA, however, there was a low level leaky expression of TbRNaseH1 before adding doxycycline, which was induced to a higher level upon addition of doxycycline (Supplementary Figure S7A). Importantly, adding doxycycline in these cells also induced TbRAP1 depletion by RNAi (Supplementary Figure S7A). In addition, we observed derepression of previously silent VSGs, including VSG 3, 8, 13, 18 and mVSG397 (Supplementary Figure S7B), confirming that expression of an ectopic TbRNaseH1 does not suppress the VSG-derepression phenotype in TbRAP1-depleted cells.

Subsequently, we performed S9.6 IP using genomic DNA isolated from S/RAP1i+RNaseH1-2HA cells followed by Southern slot blotting. As expected, the amount of telomeric RNA:DNA hybrid was approximately the same before and after induction, unlike what we observed in S/RAP1i cells (Figure 6). Clearly, expressing an ectopic allele of TbRNaseH1 in S/RAP1i cells suppressed the phenotype of more telomeric RNA:DNA hybrids.

Because silent VSGs were derepressed in S/RAP1i+RNaseH1-2HA cells upon adding doxycycline (Supplementary Figure S7B), we did not expect that expression of TbRNaseH1 would affect TERRA expression. Northern analysis using the C-rich telomeric probe showed that although the TERRA level in induced S/RAP1i+RNaseH1-2HA cells was lower than that in induced S/RAP1i cells, it was still significantly higher than that in WT cells (Supplementary Figure S7C and D). RNaseH1 degrades the RNA strand in the RNA:DNA hybrids (35), which is presumably the reason why less TERRA was detected when an ectopic allele of TbRNaseH1 was expressed.

Expressing an ectopic allele of TbRNaseH1 in TbRAP1 RNAi cells suppresses the phenotype of more DSBs at telomeres/subtelomeres and that of elevated VSG switching frequency

We hypothesize that more telomeric RNA:DNA hybrids is the reason for increased amount of telomeric/subtelomeric DSBs in TbRAP1-depleted cells. To test this, we examined the association of TbRAD51 and γH2A with the telomeric/subtelomeric chromatin by ChIP in S/RAP1i+RNaseH1-2HA cells. In contrast to S/RAP1i cells, adding doxycycline to S/RAP1i+RNaseH1-2HA cells did not increase the association of TbRAD51 or γH2A with the telomeric (Figure 2A and Supplementary Figure S4B) or ES (Figure 2B and Supplementary Figure S4D) chromatin.

We also hypothesize that more VSG switching events in TbRAP1-depleted cells are due to more telomeric/subtelomeric DSBs. To test this, we performed an in vitro VSG switching analysis in S/RAP1i+RNaseH1-2HA cells. Strikingly, the switching frequency in transiently induced S/RAP1i+RNaseH1-2HA cells was very similar to that in S/ev cells (Figure 1D). Furthermore, most switchers arose from ES GC or ESLIS (54%, Supplementary Table S7), which was the same as that in WT cells (Figure 1D). Ectopic expression of TbRNaseH1 in S/ev cells resulted in a VSG switching frequency comparable to that in S cells (Figure 1D, Supplementary Table S8). Therefore, expressing an ectopic allele of TbRNaseH1 in TbRAP1-depleted cells reduced telomeric RNA:DNA hybrids, telomeric/subtelomeric DSBs, and VSG switching frequency to WT levels.

DISCUSSION

TbRAP1 is essential for cell proliferation (15), but the reason was unknown. Now we show that lack of TbRAP1 leads to more DSBs at the telomere and ES-linked VSG loci. Since introducing a DSB near the active VSG results in >90% of cell death (3), increased telomeric/subtelomeric DSBs would be an important factor for growth arrest in TbRAP1-depleted cells. Interestingly, TbRAP1's function in maintaining telomere and subtelomere integrity is linked with its role in telomeric silencing, because TbRAP1-depletion results in an increased TERRA level and more telomeric RNA:DNA hybrids that in turn promote more DSBs at telomeres and subtelomeres, which are a potent trigger for VSG switching (Figure 7) (42). This is different from the protective roles of TbTRF and TbTIF2, two other telomere proteins that are important for maintaining subtelomere stability (24,25,27,37). Neither TbTRF nor TbTIF2 is required for VSG silencing (15,25). Depletion of TbTRF or TbTIF2 led to different subtelomeric DNA damage profiles than depletion of TbRAP1 (25,27). Therefore, although all known T. brucei telomere proteins, when depleted, result in more switchers, the underlying mechanisms are different.

Figure 7.

TbRAP1 suppresses VSG switching by repressing TERRA and telomeric RNA:DNA hybrids. We hypothesize that the presence of TbRAP1 at the telomere (top) helps blocking excessive read-through of RNAP I into the telomeric repeats from the active ES. Removal of TbRAP1 from the telomere (bottom) may allow RNAP I to travel longer along the telomere and yield a higher TERRA level (orange strips), which leads to increased amount of telomeric RNA:DNA hybrids (magnified below the blue shade) and subsequent DNA damage (red bolts) at the subtelomere/telomere vicinity. The telomere complex is extremely simplified in the diagram. Alternative hypotheses are discussed in the text.

A previous study strongly suggests that T. brucei TERRA is transcribed by RNAP I (29). We have confirmed that the active VSG-adjacent telomere is transcribed while telomeres adjacent to silent ESs are not, which is consistent with the hypothesis that TERRA results from read-through of the active ES by RNAP I. However, we do not know whether telomeres that do not harbor any ESs are transcribed. A ribosomal RNA promoter has been found at a minichromosome subtelomere (46), and it may transcribe the downstream telomere.

Depletion of TbRAP1 increases the TERRA level, but TERRA is still mainly transcribed from the active ES-adjacent telomere. Since the derepressed ESs are transcribed at a level that is still 10–100-fold lower than the active ES (16), read-through of the derepressed ESs may be too low to be detected. This is also consistent with our previous observation that loss of TbRAP1 did not affect ES chromatin structure significantly in bloodstream form T. brucei cells (16). The active ES expression is not increased by the loss of TbRAP1 (15). Additionally, no dramatic telomere length change has been detected when TbRAP1 was depleted for two days (15). Therefore, longer TERRA species in TbRAP1-depleted cells suggests that removal of TbRAP1 allows RNAP I to travel longer along the telomere downstream of the active ES (Figure 7). In TbRAP1-depleted cells, the TELC20-primed RT product yielded a low level of PCR product using primers specific to 70 bp repeats (Figure 5B), suggesting that some RNA molecules contain both telomeric and 70 bp repeat sequences. Hence, lacking TbRAP1 may also affect TERRA RNA processing and allow accumulation of more unprocessed TERRA and telomeric RNA:DNA hybrids (see below). Further studies are necessary to elucidate exactly how TbRAP1 regulates TERRA expression.

R-loops are three-stranded structure formed by an RNA:DNA hybrid plus a displaced single-stranded DNA (47), which is sensitive to RNaseH (35) and can be recognized by the S9.6 antibody (44). Excessive R-loops cause genome instability, as R-loops can stall DNA replication and collapsed replication fork often results in DSBs (34,47). In addition, processing of R-loops by transcription-coupled nucleotide excision repair factors can induce DSBs (47). However, exactly how R-loops promote DSB formation is still unclear. TERRA can form the R-loop structure with the telomeric DNA in mammalian and yeast cells, which promotes telomere recombination (30). We detected more telomeric RNA:DNA hybrids and increased association of TbRAD51 and γH2A with the telomere and subtelomere chromatin upon TbRAP1 depletion, suggesting more DNA damage at telomeres/subtelomeres (which is confirmed by LMPCR) due to increased telomeric RNA:DNA hybrids, which is suppressed by ectopic expression of TbRNaseH1. These observations support the model that excessive telomeric R-loops affect telomeric/subtelomeric integrity (Figure 7). So far, it is unclear exactly where R-loop-induced DSBs are located (47). However, we anticipate that telomeric R-loop induced DNA damage is concentrated at the telomere (Figure 7). In our strain, the distance from the end of each VSG gene to the first TTAGGG repeat is 0.7 kb (VSG2), 0.8 kb (VSG15) and 1.5 kb (VSG21), respectively (10). Therefore, it is reasonable that we detect more DSBs by LMPCR within or downstream of these VSG genes, as they are immediately adjacent to the telomere.

Depletion of TbRAP1 increased TERRA level, which presumably leads to more telomeric RNA:DNA hybrids. However, it is possible that TbRAP1 may also suppress the formation of the telomeric RNA:DNA hybrid or facilitates its resolution directly. Alternatively, it is possible that TbRAP1 depletion affects telomeric DNA replication, which results in accumulation of RNA:DNA hybrids as a replication intermediate. Further investigation will be necessary to test these possibilities. TERRA is only transcribed from the active ES-adjacent telomere, but DSBs are detected downstream of both active and silent VSGs, suggesting that telomeric RNA:DNA hybrids are formed at both active and silent telomeres. Whether TERRA can function both in cis and in trans and pair with both active and silent telomeric DNA will need further investigation in the future.

LMPCR did not detect increased DSBs in 70 bp repeats or at ES promoters (which are a few kb or 40–60 kb upstream of the ES-linked VSGs, respectively) in TbRAP1-depleted cells. However, more γH2A and TbRAD51 were associated with the ES promoters and telomeres upon TbRAP1 depletion. This discrepancy may be due to several reasons. First, LMPCR detects physical DSBs, but it is not efficient to detect breaks in repetitive sequences or if DSBs are located far from the place where the locus-specific primer anneals. Second, γH2A can spread more than tens of kilobases around the DSB in yeast (48), and it may spread equally far from the break in T. brucei. Third, RAD51 is important to keep genome integrity when DNA replication is stalled (49). ES replication may be affected by downstream telomeric RNA:DNA hybrids, which leads to accumulation of TbRAD51 along the whole ES. Particularly, the active ES replicates early (20) and may be more sensitive to replication defects.

We did not detect any R-loops at the ES-linked VSG loci in either WT or TbRAP1-depleted cells. In general, R-loops are rare in the genome. R-loops are often associated with high level of transcription, and G-clusters at the nontranscribed strand and negative superhelicity often facilitate R-loop formation (34). VSG sequences are apparently less G-rich than the telomeric repeats. In addition, efficient RNA processing appears to reduce the chance of R-loop formation (34). A large number of VSG proteins (∼107/cell) are synthesized, demanding a very efficient VSG mRNA processing. Therefore, VSG mRNAs are likely efficiently processed so that R-loops at VSG loci are extremely rare.

Many pathogens that undergo antigenic variation express their major surface antigens from subtelomeric regions, as subtelomere plasticity would facilitate antigen switching (2). Indeed, T. brucei subtelomeres are fragile (3), which presumably facilitates DNA recombination-mediated VSG switching (42). However, DSBs at subtelomeres, particularly within the active ES, are deleterious to T. brucei (3). Therefore, it is essential to balance the stability and plasticity at subtelomeres, which is a delicate task. In the TbRAP1-depleted cells, it is detrimental to cell viability to have many DSBs at telomeres/subtelomeres, even though DSBs are a potent trigger for recombination-mediated VSG switching. In summary, our data not only demonstrate that TbRAP1 affects VSG switching by maintaining telomere and subtelomere integrity and stability but also showed that TbRAP1-mediated telomeric silencing is the underlying mechanism of how TbRAP1 suppresses subtelomeric DNA recombination. Importantly, we showed that abnormally high levels of TERRA and telomeric RNA:DNA hybrids cause genome instability that leads to deleterious effects, further underlying the importance of appropriate regulation of the telomeric transcript and RNA:DNA hybrid by TbRAP1. For the first time, our study demonstrated the involvement of a long, non-coding RNA in antigenic variation and the mechanistic link between telomeric silencing and telomere integrity through the regulation of telomere transcription by a conserved telomere protein.

Click here for additional data file.