Introduction

Telomeres, the nucleoprotein complexes located at the ends of linear chromosomes, are essential for genome stability. As conventional DNA polymerases are incapable of replicating the linear DNA ends completely, most eukaryotic cells use telomerase, an RNA-protein complex, to maintain telomeres1,2,3. The telomerase RNA is an integral component of telomerase that dictates the RNA template-dependent telomere synthesis4,5. Lack of telomerase activity leads to telomere shortening in a number of eukaryotic organisms, including human6, mouse7, plant8, worm9, budding10 and fission yeasts11, and protozoan parasites such as Plasmodium falciparum12 and Trypanosoma brucei13.

Telomerase contains a core protein component, Telomerase Reverse Transcriptase (TERT), that provides the catalytic activity and an RNA component, TR, that provides the template for telomere synthesis14. TERT belongs to the reverse transcriptase (RT) family and contains an RT domain with all RT-specific motifs (1, 2, and A-E motifs)15. Unlike classical RTs, telomerase requires an obligatory internal RNA component as a template for DNA synthesis and can copy the template sequence repetitively through efficient translocation. TR also contributes greatly to the processivity of the telomerase16,17. Hence, dysfunctional TR leads to telomere length shortening.

Although the domain structures of TERT are conserved across different species, the TR molecules from different organisms vary greatly in sequence and size (from ∼150 nt in ciliates to ∼1 800 nt in P. falciparum4,18,19,20,21,22). However, a common theme in TR secondary structure has been observed23, which consists of a single-stranded template, a pseudoknot, a template boundary element (TBE), and a telomerase-interacting domain. Mutations in the TR template can introduce mutations into the telomere sequences. This can affect the active-site functions of telomerase due to altered enzyme-substrate interaction24. Mutated telomere sequences may not be recognized by telomere DNA binding factors, which can also lead to telomere dysfunctions.

T. brucei is a protozoan parasite that causes African trypanosomiasis in humans. One major reason for persistent T. brucei infection is that it undergoes antigenic variation and regularly switches its surface antigen (VSG) to evade the host's immune responses25. There are more than 1 000 VSG genes in T. brucei genome26, but only one VSG is expressed exclusively from one of multiple, nearly identical subtelomeric VSG expression sites at any time27. In T. brucei, telomerase activity imparts the predominant mechanism for telomere maintenance28, and telomere elongates at an average rate of 4-6 bp per population doubling (PD) during continuous cell growth29. The TbTERT gene has been identified, and its deletion led to a steady telomere shortening at a rate of 3-4 bp/PD13. Interestingly, when the active VSG-marked telomere is shortened to ∼1 kb, the VSG switching rate is increased by ∼10-fold30. In addition, TbRAP1, an intrinsic T. brucei telomere component, is essential for regulation of subtelomeric VSG gene expression31. Therefore, understanding the telomerase-mediated telomere maintenance in T. brucei can help to elucidate the mechanisms of T. brucei pathogenesis.

Among all the parasitic protozoa for which genome sequences are now available, T. brucei is the most amenable to genetic and biochemical studies due to their rapid growth (as short as 6 h/PD), simple in vitro culturing procedures, efficient knockout/knockin mediated by homologous recombination, and availability of useful molecular tools such as inducible expression32 and RNAi33. Particularly, T. brucei telomere sequence34 and T-loop structure35,36 are identical to those in humans, and the telomere protein complex is largely conserved between T. brucei and vertebrates31,37. Therefore, T. brucei can serve as a useful model system for dissecting the structure and function of telomerase in pathogenic protozoa and for comparative analysis of telomerase function and evolution in general.

In this study, we identify the putative T. brucei TR gene through an in silico approach and further provide in vivo characterization of TbTR's native folding and activity. Additionally, we determine TbTR's in vivo interaction with TbTERT and its critical role in telomere maintenance in T. brucei.

Results

Identification of a putative TbTR gene

We analyzed the telomere sequences in T. brucei brucei and other related subspecies (T. brucei gambiense, T. vivax, T. congolense and T. cruzi). Based on the major telomeric repeat sequence (TTAGGG) identified and considering the previously suggested TR template sequence28,38 (Supplementary information, Figure S1A), we performed a BLAST search for short, nearly exact matching sequences in the T. brucei genome. In this process we identified a number of initial TR candidates (Supplementary information, Figure S1B), many of which overlapped with annotated ORFs, structural RNAs (rRNAs, tRNAs or snoRNAs) and sequences at the telomeric ends of T. brucei genome (Supplementary information, Figure S1B-i). We discarded those initial candidates overlapping annotated protein- or RNA-coding regions or spanning telomeric repeat sequences and narrowed the list down to a few hits that showed a high degree of sequence similarity around the putative TbTR template region (Supplementary information, Figure S1B-ii). We then extended the comparative genome analysis in both directions flanking the TbTR template for all hits to search for syntenic regions among five related Trypanosoma subspecies (Supplementary information, Figure S1B-iii). Eventually we identified a unique sequence of ∼200 nt in T. brucei genome that gave the highest scoring hits when used as BLAST query against each of the other Trypanosoma genomes. These hits are highly conserved in multiple regions including the putative template domain among five Trypanosoma subspecies (Supplementary information, Figure S1B, inset).

Northern blot analysis, using a 35 nt probe specific for this locus, identified a transcript of ∼900 nt (Supplementary information, Figure S2A), which we propose to be the putative TbTR. We found that TbTR is expressed at similar levels in the bloodstream form (BF, when T. brucei is inside a mammalian host) and procyclic form (PF, when T. brucei is inside the midgut of its insect vector) stages. The amount of TbTR in PF cells is estimated to be 92% (Figure 1A, left), 84% (Supplementary information, Figure S3A), and 114% (data not shown) of that in BF cells, respectively, giving an average value of 97%. Likewise, TbTERT appears to be expressed at similar levels in both BF and PF cells (Figure 1A, right).

Figure 1
figure 1

Molecular validation of TbTR. (A) Northern blot analysis of TbTR (left) and TbTERT (right) expression in WT BF and PF cells. Top, blot hybridized with a 900 bp TbTR- or a TbTERT-specific probe. Bottom, ethidium bromide-stained RNA gel showing rRNA species, served as a loading control. (B) SDS-PAGE profile showing a ∼110 kDa and ∼130 kDa band in TbTERT IP sample. (C) IP and western blotting using an anti-TbTERT peptide antibody. The preimmune serum was used as a negative control (Pre). Asterisk represents the TbTERT protein. (D) TbTR is associated with TbTERT. T. brucei cell extract was immunoprecipitated with anti-TbTERT peptide antibody, and the IP product was reverse transcribed after DNase treatment with (+) or without (−) RT using either random primer (RP) or Oligo dT primer (OP), followed by PCR amplification with TbTR-specific primers. Genomic DNA (gDNA) was used as a control. Extra spaces between the marker and lane 2 on the same gel were deleted from the image to save space. (E) Telomerase activity assay using TbTERT IP product pretreated with RNase A or RNasin. The periodicity of repeat was determined by radiolabeled input oligos of 18 and 24 nt (marked on the left). The product with one TTAGGG repeat addition is marked as (+6). In this and other figures, M represents molecular weight marker.

To map both ends of TbTR, we used RNA ligase-mediated RACE (RLM-RACE), which selects full-length mRNAs from total RNA by enzymatic treatments followed by identification of cDNA ends via adaptor-mediated PCR39. One prominent PCR product was identified from the 5′ RACE reaction in addition to two minor amplicons of larger sizes (Supplementary information, Figure S2C). Sequencing analysis of the RACE fragments identified the major 5′ end nucleotides of the TbTR transcript (Supplementary information, Figure S2D), which was also independently verified by primer extension analysis (data not shown). The 3′ end of TbTR also showed ambiguity in end processing, generating three different 3′ RACE products (330, 280 and 150 bp). Cloning and sequencing these RACE products identified alternative 3′ ends of TbTR (Supplementary information, Data S1).

Association of TbTR and TbTERT in vivo

To examine the interaction between TbTR and TbTERT, we raised peptide antibodies against TbTERT using three TbTERT N-terminal peptides. Immunoprecipitation (IP) with one of these antibodies allowed us to obtain a major protein species of 110 kDa and a minor one of 130 kDa (Figure 1B, arrows). In the subsequent western blot analysis of the immunoprecipitated products, ∼40% of the input sample (the 130 kDa band) was pulled down by the anti-TbTERT antibody, but not by the preimmune serum (Figure 1C, asterisk). MALDI-TOF mass spectrometry analysis of this IP product identified several peptides that match the published TbTERT sequence (data not shown), confirming that this is indeed TbTERT. At this point we are unclear about the relationship between the two different-sized bands detected on the SDS-PAGE (Figure 1B). Subsequently, we isolated RNA from the DNase-treated IP product and performed two different reverse transcription reactions, one with random primers (RP) and the other with Oligo dT primers (OP). RT-PCR analysis with TbTR internal primer pairs spanning the RNA template domain identified a band of expected size of about 300 bp (Figure 1D, lanes 2 and 4) from both RP and OP samples, but not from a control reaction lacking RT (Figure 1D, lane 3). Therefore, this data suggests that TbTR is polyadenylated, which associates with TbTERT in vivo. Immunopurified T. brucei telomerase displayed excellent stability and was able to elongate T. brucei telomere substrate in a direct telomerase activity assay (Figure 1E). This activity is sensitive to RNase, but addition of RNase inhibitor, RNasin, can restore the activity during RNase treatment, indicating that this activity depends on an essential RNA component.

Deletion of TbTR leads to progressive telomere shortening

As an independent approach to validate the identity of TbTR, we also established BF and PF TbTR-null cells by replacing the TbTR endogenous alleles with Hygromycin resistance (HYGRO) and Blasticidin S resistance (BSD) genes sequentially. We verified the genotype of TbTR double knockout (DKO) cell lines by PCR analysis (Supplementary information, Figure S3C) and Southern blotting (Supplementary information, Figure S3D). Northern blot analysis also showed that TbTR was not detectable in the TbTR-DKO cells (Supplementary information, Figure S3E).

If TbTR indeed encodes the telomerase RNA component, deletion of TbTR should lead to a telomere shortening phenotype. Telomeres grow at a 6-10 bp/PD rate on average in wild-type (WT) BF T. brucei cells when they are continuously propagated29,40. As a control, we cultured the WT BF cells for more than 130 PDs, and observed the same telomere elongation phenotype (Figure 2A). In contrast, in BF TbTR-DKO clones, telomeres shorten progressively (Figure 2B and 2C). In telomere Southern blot analyses using a TTAGGG repeat probe, all major telomere fragments can be detected (Figure 2B). By calculating the sizes of 5 individual telomere bands in clone C1 (Figure 2B, left panel, asterisks), we estimated a telomere-shortening rate of 3-5 bp/PD during the culturing period. To estimate the rate more accurately, we examined the lengths of telomeres that are specifically marked by a subtelomeric VSG11 (aka VSG BR2) gene by Southern blotting (Figure 2C, left panel). Our T. brucei strain has more than one copy of the VSG11 gene, among which two are telomeric13,41 and are shown as the 20 kb and 7.8 kb bands. We followed the smaller telomeric fragment for more accurate size measurement (Figure 2C, left panel, asterisk) and calculated the telomere-shortening rate to be ∼3 bp/PD. In addition, we examined three VSG8 (aka VSG OD1)-marked bands that are apparently telomeric in the C1 cells (Figure 2C, right panel, asterisks), and the telomere-shortening rate is estimated to be ∼5 bp/PD. Similarly, by calculating the sizes of 6 individual telomere bands in clone C2 (Figure 2B, right, asterisks), we estimated a telomere-shortening rate of ∼5 bp/PD. Therefore, telomere-shortening rate in TbTR-DKO cells is similar to that in TbTERT-null cells (3-4 bp/PD)13,41.

Figure 2
figure 2

Telomere length changes in WT and TbTR-DKO BF cells. Genomic DNA was prepared from WT (A) and TbTR-DKO (B, C) BF T. brucei cells at regular time intervals. A (TTAGGG)4 oligo probe was used in A and B. The VSG8 and VSG11-specific probes reveal both chromosome internal and telomeric DNA fragments (C). C1 and C2 are independent clones. In figures 2-4, PD numbers at each time point are indicated on the top, and telomere bands used for telomere length shortening or elongation rate calculations were marked with asterisks.

Telomere length changes have not been carefully studied in WT or TbTERT null PF cells. We therefore cultured the PF WT cells continuously for over 150 PDs and estimated the telomere elongation rate to be ∼5 bp/PD on average (Figure 3A). In contrast, telomeres shortened at a rate of 5-6 bp/PD in both C6 and C21 TbTR-DKO clones (Figure 3B). Therefore, in both BF and PF cells deleted of TbTR, telomere maintenance is defective, strongly arguing that TbTR encodes the RNA component of telomerase.

Figure 3
figure 3

Telomere length changes in WT and TbTR-DKO PF cells. Genomic DNA was prepared at regular time intervals from WT (A) or TbTR-DKO (B) PF cells. C6 and C21 are independent clones. Southern blot analyses were carried out using the (TTAGGG)4 oligo probe.

Although deletion of TbTR led to telomere shortening, we did not observe any growth defects after long-term (more than 350 PDs) continuous cell culture (Supplementary information, Figure S4), which is the same as TbTERT-deleted cells41.

To further confirm that the telomere-shortening phenotype is caused by the lack of TbTR transcript, we introduced an ectopic WT allele of TbTR into the BF TbTR-DKO cells. A Tet-inducible WT TbTR allele together with its original 5′ and 3′ flanking sequences (∼500 bp) is targeted to the rDNA array. Upon induction, we observed the expression of TbTR at a level higher than that in WT cells (Figure 4A). Examination of telomere lengths in two independent clones by Southern blot analysis showed that telomeres elongate at a steady rate of 8 bp/PD (Figure 4B). Similarly, we targeted a WT allele of TbTR with its original 5′ and 3′ flanking sequences into the tubulin array in PF TbTR DKO cells. This ectopic allele of TbTR is expected to be expressed constitutively, and northern blot analysis showed that it is expressed, although at a level lower than WT (Figure 4C). Telomere Southern blot analysis indicated that telomeres did elongate in these cells, but quantification showed a telomere growth rate of only ∼2 bp/PD, suggesting that the weak expression of TbTR led to a slow telomere growth. Therefore, the telomere-shortening phenotype in TbTR-null cells is complemented by a WT allele of TbTR, indicating that the lack of TbTR function is responsible for the telomere maintenance defect in T. brucei.

Figure 4
figure 4

The telomere-shortening phenotype in TbTR-DKO cells is complemented by an ectopically expressed TbTR WT allele. Northern blots showing expression of the ectopic TbTR allele in BF (A) or PF (C) TbTR DKO cells: Top, blot hybridized with a TbTR-specific DNA probe; Bottom, rRNA species are shown as a loading control. Genomic DNA was isolated at regular intervals and telomere Southern blot was carried out using the (TTAGGG)4 oligo probe on DNA isolated from BF (B) and PF (D) complementation cell lines. B1 and B2 are independent clones of BF cells.

TbTR-DKO cells lack telomerase activity

We adopted a PCR-based TRAP assay42 to determine whether the deletion of TbTR abolished telomerase activity. We first determined the TRAP profile of a telomerase-positive human colon cancer cell line (HT29) using the TS primer as substrate. Human telomerase is expected to add GGTTAG repeats onto TS 3′ end in an RNase-sensitive manner, which is what we observed (Supplementary information, Figure S5, lanes 7 and 8). When T. brucei cell extract was used, a similar RNase-sensitive periodical amplification profile was observed (Figure 5A, lanes 5 and 6). No product was detected when the TS or reverse primer was omitted (Figure 5A, lanes 2 and 3, respectively). When cell extract was omitted from the reaction, we did not observe the ladder of products except one faint band of ∼40 bp (Figure 5A, lane 4), indicating that this band did not result from telomerase activity, but was most likely a primer-dimer product. Careful examination of the sequences of the TS (5′-AATCCGTCGAGCAGAGTT-3′) and the reverse primer (5′-CCCTTACCCTTACCCTTACCCTAA-3′) also revealed that the two primers can anneal with two T:A base pairs, and the primer-dimer product should have a size of 40 bp. WT cell lysate gave a same sized band with stronger intensity as any telomerase-extended product will be amplified with the same TS and reverse primer pair in the TRAP assay.

Figure 5
figure 5

Molecular and biochemical properties of TbTR. (A) TRAP assay with 0.15 μg T. brucei PF cell extract using TS primer as substrate. The TS oligo, reverse primer, or the cell extract was omitted in lanes 2, 3, or 4, respectively. Lane 6 represents an RNase-treated sample. (B) TRAP assay to compare the telomerase activities in WT (lanes 2 and 3), TbTR-DKO (lanes 4 and 5), and TbTR-DKO cells with a complementation TbTR allele (lanes 6 and 7) with or without RNase A treatment. 0.25 μg (lanes 2 and 3), 0.28 μg (lanes 4 and 5) or 0.43 μg (lanes 6 and 7) of cell extract was used in each TRAP reaction, and 20 μl (lanes 2-5) or 40 μl (lanes 6 and 7) of final products were loaded in each lane, respectively. (C) Detecting spliced leader (SL) sequence at the 5′ end of immunopurified TbTR. RT-PCR was done with (+) or without (−) RT using a primer specific to TbTR and a primer specific to the SL sequence. Lanes 2 and 4, RT with random primer (RP) and Oligo dT primer (OP), respectively. Lane 3 represents control reaction without RT. (D) Determination of 5′ cap status of TbTR from total RNA and TbTERT-IP RNA. Total or TbTERT antibody-immunopurified RNA fraction from WT cells were immunoprecipitated with TMG antibody, and the IP products were treated with (+) or without (−) RT followed by amplification with TbTR-specific primers. The supernatant fraction of TbTERT-IP (sup) and the total RNA extract from TbTR-DKO cells were examined the same way. Genomic DNA (gDNA) was amplified with the same TbTR primers as a control. (E) PF T. brucei cells were treated with RNAP III inhibitor Indazolo-sulfonamide for 24 h followed by RNA extraction. Northern blotting was done with TbTR- or U2 snRNA-specific oligonucleotide probes (top). Densitometry-quantified TbTR and U2 snRNA levels were normalized against untreated samples, and the relative changes are shown at the bottom.

Utilizing this TRAP assay, we compared the telomerase activities from WT and TbTR-DKO cells and TbTR-DKO cells containing a TbTR complementation allele using equal amount of cell extracts (Supplementary information, Figure S5A). No telomerase extension products were detected in TbTR-DKO cell lines (Supplementary information, Figure S5A, lane 3), while a weak telomerase activity in the complementation line was observed (Supplementary information, Figure S5A, lane 5, compared with WT in lane 1). To confirm detectable telomerase activity in the complementation line, we repeated the same experiment, but used double amount of cell extract from the complementation line and loaded twice as much of its product on gel (Figure 5B). In this case, we clearly observed telomerase extension products from the complementation line (Figure 5B, lane 6) but not from the DKO cells (Figure 5B, lane 4). Even when double amount of cell extract from TbTR-DKO cells was used, we still did not observe any telomerase extension products (Supplementary information, Figure S5B). Therefore, TbTR-DKO cells lack the telomerase activity, and the complementation line restored the telomerase activity. The apparent low telomerase activity in the complementation cells is presumably due to the fact that the TbTR complementation allele was expressed at a lower level than in WT cells (Figure 4C).

Mutation in the TbTR template region resulted in altered telomere sequences

To further prove that TbTR is used as the template in telomere DNA synthesis, we mutated the TbTR template 5′-CCCTAACCCTA-3′ into 5′-CCCTTACCCTA-3′ (termed TbTRt) and expressed the mutant allele in TbTR null cells. An incorporation of the mutant telomere sequence at the chromosome ends would most definitively confirm the identity of TbTR.

Because our TbTR-null clones have only been cultured for a short period of time, most telomeres are still relatively long. Therefore, it will take a considerable length of time to observe a significant change in telomere length after the ectopic TbTRt is introduced into these cells (so as to verify that TbTRt is functional). It has been shown that the active VSG-marked telomere is more prone to large-sized telomere breaks than other telomeres29,40,43. Therefore, we subcultured TbTR-null cells and picked a clone, in which the active VSG2 (aka VSG221)-marked telomere shortened to ∼1.5 kb. We transfected the TbTRt-expressing construct into these cells, and northern blot analysis showed that TbTRt was expressed at a slightly higher than WT level in all three resulting independent clones (Supplementary information, Figure S6A). Within a week, we were able to detect VSG2-marked telomere growth in TbTRt transfectants (data not shown), which were used for subsequent telomere cloning.

To clone the newly elongated telomeres in TbTRt-expressing cells, we took advantage of the G-rich 3′ overhang structure at the very end of T. brucei telomeres44. We have recently found that at least some of the telomere G-overhangs end in 5′-TAGGGT (Sandhu R and Li B, unpublished data). Therefore, we were able to ligate an adaptor with a 16 bp duplex DNA of unique sequence and a 4 nt 3′ overhang that matches with the terminal 5′-GGGT telomere sequence at chromosome ends (Supplementary information, Figure S6B). Subsequently, the terminal few hundred base pairs of telomere DNA was amplified by one backward primer specific to the unique region of the adaptor and a forward primer containing 4 repeats of TTAGGG. The amplified DNA fragments were then inserted into pGEM-T-easy (Promega) by TA cloning followed by sequencing analysis.

From three independent TbTRt-expressing clones, several individual telomere-containing plasmids were cloned and sequenced. Telomere sequences of six representative clones are shown in Table 1. They all contain a long stretch of TTAGGG WT repeats and a short stretch of TAAGGG mutant repeats at the very end. As a control, telomere fragments cloned from WT cells all contain TTAGGG repeats. Therefore, the mutated TbTRt template sequence was indeed incorporated into telomeres, confirming that TbTR provides the RNA template for T. brucei telomerase.

Table 1 Telomere sequences in several representative telomereclones

Biogenesis of TbTR

Most transcription in T. brucei is polycistronic and few genes have been associated with conventional RNA Pol II promoters. Trans-splicing is a major mechanism of transcript maturation in T. brucei that differs greatly from that of mammals and insects. We therefore examined whether TbTR requires trans-splicing for maturation. RNA sample isolated from immunoprecipitated TbTERT fraction was reverse transcribed using both RP and OP primers. Subsequent PCR analysis using one primer specific to the 5′ end of TbTR and the other specific to the 35 nt spliced leader (SL) RNA common to all T. brucei trans-spliced mRNAs45 resulted in products with the expected size (Figure 5C). A mock reaction without RT failed to yield any product. These data suggest that trans-splicing is involved in the maturation of TbTR, which is different from all other known telomerase RNA species.

T. brucei SL RNA methylations are required for trans-splicing46. However, in contrast to U1, U2 and U4 snRNAs and the U3 snoRNA that contain a 2, 2, 7-trimethyl guanosine cap47, T. brucei SL RNAs do not contain the characteristic U snRNA cap structure. Instead of a TMG cap, SL RNAs provide an unusual cap 4 structure at the 5′ end of all mRNAs48. Because of technical difficulties in determination of the cap 4 structure, we decided to examine whether TbTR contains a TMG cap. We immunoprecipitated total T. brucei RNA with a monoclonal antibody that recognizes the TMG cap. The resulting IP product gave rise to the expected band after RT-PCR amplification using TbTR specific primers when WT cells, but not TbTR-DKO cells were used (Figure 5D). However, when RNA was first immunoprecipitated with TbTERT-specific antibody followed by precipitation with the anti-TMG antibody, no RT-PCR product was detected using the same set of primers (Figure 5D, pellet). Interestingly, TMG signal was detected in the supernatant fraction of the TbTERT IP (Figure 5D, sup), suggesting that the mature TbTR incorporated into telomerase does not possess a canonical cap structure that can be recognized by the anti-TMG antibody. Therefore, we speculate that TbTR might acquire a cap 4 structure49,50 that was not detected by the TMG antibody, which is consistent with our finding that the TbTERT-associated TbTR contains the SL sequence (Figure 5C).

As both human telomerase RNA and SL RNAs in trypanosomes are transcribed by RNA polymerase II (RNAP II), we were curious to know, which RNA polymerase transcribes TbTR. We identified a putative 'TATA' motif, multiple GATA-type transcription factor binding sites (Supplementary information, Figure S1C), and a C/EBP1-type motif upstream of the TbTR transcription start site using Emboss: Tfscan (http://helixweb.nih.gov/emboss/html/tfscan.html), suggesting that TbTR is transcribed by RNAP II. Due to the absence of a good T. brucei cell-permeable inhibitor for RNAP II, we turned to a RNAP III inhibitor that has a broad effect on RNAP III transcription in eukaryotic cells (IC50 = 27 μM and 32 μM for human and Saccharomyces cerevisiae RNAP III, EMD Biosciences). We cultured PF T. brucei cells in the presence (10 μM and 50 μM) or the absence of this inhibitor for 24 h followed by total RNA isolation and northern blot analysis. Because U2 spliceosomal RNA in T. brucei is transcribed by RNAP III51, we used U2 snRNA as a positive control. Northern blot analysis revealed 16% - 23% reduction in the U2 mRNA level when cells were treated with the RNAP III inhibitor, whereas the TbTR level did not change significantly (Figure 5E), further suggesting that TbTR is transcribed by RNAP II.

Secondary structure of TbTR

TbTR folding was first analyzed using parameters that consider sequence conservation information from multiple Trypanosoma subspecies. We recorded nucleotide changes consistent with RNA evolution (covariations) rather than third position variation common to protein coding regions. The 5′ and 3′ RLM-RACE-generated minimal TbTR sequence was used to generate alignments for modeling a consensus secondary structure for Trypanosoma TR molecules. Initially, using Turbofold program52, we were able to compare the common domain architecture of individually folded Trypanosoma TR molecules to derive a consensus structure. This structure was then verified by RNAalifold53, and a potential region to form a pseudoknot was identified by Pknots program54. Finally, covariations were checked manually within conserved domains.

To verify this model, we performed in vivo RNA footprinting. To determine how this RNA folds in vivo, we isolated RNA from T. brucei nuclei that were exposed to DMS and CMCT55. We found that the majority of the nucleotides in the template domain of TbTR is exposed and modified by DMS or CMCT (Figure 6A), suggesting that the TbTR template is generally accessible to single strand-specific hybridization to telomeric repeats and to incoming nucleotides for telomere synthesis. In contrast, very few nucleobases in TBE, which could form a helical structure, are exposed (Figure 6B).

Figure 6
figure 6

Chemical probing of TbTR and secondary structure prediction. Chemical probing of (A) the template region and (B) TBE of TbTR with DMS and CMCT in vivo. Control (CON) represents RNAs not exposed to DMS. Lanes G, U, A and C represent dideoxynucleotide sequencing results. Solid black circles represent full accessibility to chemicals, while open circles represent moderate accessibility. The same symbols are used to mark sites of chemical accessibility along RNA template and helices I-IV on the side in C. (C) Secondary structure model of TbTR RNA. Between two paired RNA strands (marked in the center), solid lines represent Watson-Crick base pairing, open circles represent Wobble pairs, and long dotted lines represent putative long distance/tertiary interactions. Asterisks mark consensus base pairs on both sides. Compensatory mutations/covariations are shaded with grey circles, and single base changes from phylogenetic sequence analysis are circled. Helical domains are numbered I to IV from 5′ to 3′.

Thus, phylogenetic analysis, supported by in vivo chemical footprinting data, showed a strong correlation in developing a TbTR secondary structure model (Figure 6C). In this model, a 5′ proximal stem loop (Helix-I) is present in all Trypanosoma TR analyzed, which is shorter than the yeast telomerase RNA (TLC1) hairpin that interacts with Ku56. We can also detect a conserved template proximal hairpin (Helix-II), which is likely the TBE. Our iterative folding process always detected the template domain to be single-stranded in optimal physiological temperature and salt conditions, which is embedded as a highly conserved domain between TBE and the pseudoknot. The possibility of forming a pseudoknot within the TbTR core 5′ domain was highlighted by the presence of a pair of compensatory mutations identified in this structure. A smaller, alternative, minimum free energy pseudoknot structure within 50 nt of the template domain was identified by iterative folding process (Supplementary information, Figure S1D), which is not supported by conservation or covariation. In addition, a 3′ end multiloop structure with a long hairpin (Helix-IV) was identified mostly in the proximal stem resembling the terminal structure of hTR, which is supported by the presence of covariation. We were not able to identify any typical Sm binding sequence or verify any hTR-like H/ACA motif in TbTR by computational analysis. Our analysis provides a strong but conservative model for native folding of TbTR, which requires further validation with mutational analysis and functional studies. However, the overall consistency among phylogenetic and in vivo structure-probing data provides a convincing paradigm that can help to understand telomerase functions in a genetically tractable human pathogen.

Discussion

Several lines of evidence indicate that the TbTR we identified is indeed the RNA component of T. brucei telomerase. The telomere complex has been shown to play an important role in T. brucei pathogenesis31. Therefore, altering the TbTR template sequence can have a profound effect on the telomere structure, which may have therapeutic implications. In addition, previous studies proposed a putative template domain for Trypanosoma telomerase RNA28,38, which has a unique cytosine-rich motif distinctive from any other known TR molecules, except that from another pathogenic protozoa Plasmodium falciparum (Supplementary information, Figure S1A). Through mutation analysis, we confirmed that this motif indeed is part of TbTR template. This motif may play a novel role in substrate recognition38. Therefore, understanding the functions of TbTR can provide clues to a novel telomere maintenance mechanism in kinetoplastid parasites.

The loss of telomeric repeats after TbTR deletion and restoration of telomere elongation in the presence of an exogenous WT copy indicate that TbTR is encoded by a single-copy gene located on chromosome 11. The major TbTR transcript is a 900 nt RNA, as shown in northern blot analyses. Reverse transcription of TbTERT-associated TbTR (isolated through TbTERT IP) with Oligo dT primer followed by PCR with TbTR-specific and SL-specific primer pairs gave a product of expected size, suggesting that the mature TbTR contains a poly (A) tail and a 5′ SL cap. Capping of mRNAs by SL sequence through trans-splicing is a unique RNA maturation process characteristic of Trypanosoma subspecies. Although different than cis-splicing, trans-splicing in T. brucei still requires canonical spliceosomal U RNA machinery57 and Sm component proteins58,59. Telomerase RNA maturation in fission yeast involves mechanism akin to cis-splicing60, which requires sequential binding of Sm proteins to telomerase RNA61. Although we were unable to detect any typical Sm binding sequence in the TbTR, we cannot rule out the possibility of TbTR interacting with Sm proteins as it could be involved in trans-splicing for TbTR maturation. It is interesting to note that splicing appears to be a common mechanism for maturation of larger telomerase RNAs, such as those encoded by fission yeast60, Plasmodium sp.22 (unpublished data), or Trypanosoma, and could be linked to polyadenylation as seen in budding yeast62. Oligo dT primed RT-PCR analysis was previously used to determine the polyadenylation status of human telomerase RNA63, and to measure the cellular concentration of yeast telomerase RNA, TLC164. Using a similar approach, we were able to identify polyadenylated TbTR transcript not only in total RNA fraction but also from immunopurified telomerase complex (Figure 1D and Supplementary information, Figure S2B). Therefore, TbTR appears to exist in multiple trans-spliced and/or polyadenylated isoforms, as a part of the active telomerase RNP complex or processing intermediates (Supplementary information, Data S1). Their functions in telomere synthesis or developmental regulation in insect and vertebrate hosts, if any, remain to be established.

Like yeast and human telomerase RNA18, TbTR is likely transcribed by RNAP II. However, the 5′ cap modification of RNAP II-transcribed RNAs in Trypanosomes is different from that in other eukaryotes. Usually the SL RNAs possess a complex cap structure, called 'cap 4', which is essential for trans-splicing46,48,65. The RNAP III-encoded spliceosomal RNAs, such as U2 and U4, possess a TMG cap, whereas U5 lacks it66. We did detect TMG-capped TbTR in the total RNA fraction. However, TbTR does not appear to be TMG-capped when associating with TbTERT, thus, based on the cap requirements for trans-spliced products, there is a possibility that trans-spliced TbTR contains the cap 4 structure. However, we are uncertain at this stage whether and why TbTR carries a TMG cap during primary processing, although mRNAs in Caenorhabditis elegans can keep their TMG caps even after trans-splicing67.

The telomere elongation rate in WT cells is very similar in BF and PF cells, which is consistent with our observation that both TbTR and TbTERT are expressed at very similar levels. In TbTR-DKO cells expressing TbTRt mutant, most of the mutant telomeres only had a short stretch of mutant sequence incorporated, suggesting low processivity of T. brucei telomerase. This is consistent with our telomerase primer extension result and previous observations28.

The function of TR serving as a template goes beyond simple substrate recognition and hybridization. Trypanosome telomerase is semiprocessive in nature28,38, although it makes telomeric repeats identical to those synthesized by human telomerase, a highly processive enzyme. Such difference in enzyme processivity can be due to the differences in core template sequences or template adjacent region between host and pathogen. For example, there is a 'UAA' duplication in human TR template that could render enhanced processivity68. In contrast, there is a 'CCC' duplication in TbTR template, which could provide high affinity substrate hybridization38. However, this could also affect the telomere elongation rate due to stronger binding of template to telomeric substrate.

In TbTR complementation cell lines, the telomere elongation rate appears to depend on the level of TbTR expression. This observation strikes a possibility of TbTR being the rate-limiting factor for telomere elongation, although more careful analyses are necessary to reveal the relationship between TbTR/TbTERT expression level, telomerase activity, and in vivo telomere elongation rates. Interestingly, hTERT is usually only expressed in telomerase-positive cells while hTR expression appears to be ubiquitous18, suggesting that hTERT might be the rate-limiting factor in human cells for telomere elongation. However, expression of ectopic hTERT and hTR both appear to be able to increase telomerase activity and lead to telomere elongation, and their effects are additive69. Therefore, it would be necessary to further investigate in details the telomerase regulation in T. brucei to reveal any similar or different underlying mechanisms, which should contribute to antiparasite drug development in the future.

Given major differences in molecular properties and activity of T. brucei and human TR, establishment of a secondary structure model for TbTR is important for future genetic and functional analysis. Our in vivo structure probing data strongly support the phylogenetics-based and computer-developed model. First, iterative folding of predicted TR sequences from five different Trypanosoma subspecies identified the template region to be single-stranded in an unbiased, independent process, which perfectly correlated with in vivo chemical probing data. Thus, the majority of the TbTR template residues are unlikely to be occupied in any hydrogen bonding or tertiary or RNA-protein interactions. Second, the differences in chemical accessibility between the template domain and adjacent 5′ helix-II in in vivo assays suggested a helix-forming potential for TBE. Finally and most importantly, conservation among functional domain sequences or structures provided additional confidence for the structure prediction. For example, the NMR structure of the hTR showed extensive tertiary interactions between the loop and stem nucleotides, rendering an essential triple helix in the pseudoknot structure70,71. A similar structure could possibly form in TbTR pseudoknot (Figure 6C). Additionally, the minimal CR4/CR5 domain sequence of hTR required for reconstitution of active telomerase in vitro is > 70% conserved with TbTR helix IV sequence (Figure 6C), including the P6.1 loop that is shown to be essential for TERT binding and telomerase activity72. Thus, these findings provide a testable model for further analysis.

In TbTERT-null cells, when silent ES-marked telomeres are extremely short (∼40 bp), it was observed that the short telomeres can be stably maintained for several tens of generations without significant changes, indicating the existence of an unusual telomerase-independent telomere maintenance mechanism in T. brucei41. It will be interesting to examine whether the same mechanism can be activated in TbTR-DKO cells that have lost most telomere DNA.

Telomeres and subtelomeres are recombination hotspots for parasitic protozoa such as T. brucei that undergo antigenic variation. In TbTERT-null cells, when the active VSG-marked telomeres are extremely short, an elevated VSG switching frequency was observed30. It has been hypothesized that extremely short telomeres allow more frequent chromosome end breaks that damage the active VSG gene, which forces the cell to undergo antigenic variation73. Whether a similar effect can be seen in TbTR-DKO cells remains to be investigated. As genes encoding virulence factors such as VSGs are expressed from subtelomeric regions, our study lays a foundation for the thorough understanding of TbTR's role in telomere maintenance and virulence gene regulation in a genetically tractable model pathogen.

Materials and Methods

Plasmids

HYGRO together with the actin 5′ and 3′ UTR or BSD together with the Aldolase 5′ and 3′ UTR is flanked by 570 bp of DNA sequences upstream and 500 bp of DNA sequences downstream of the endogenous TbTR gene. The two cassettes are inserted into pBlueScript to generate pSK-TbTR-KO-HYGROcas and pSK-TbTR-KO-BSDcas, respectively. The WT TbTR gene and TbTRt (template changed to 5′-CCCTTACCCTA-3′) together with 670 bp upstream and 370 bp downstream TbTR flanking sequences were inserted into the pLew111 plasmid to generate pLew111-TbTRutr and pLew111-TbTRt-utr, respectively. The same WT TbTR gene with its endogenous flanking sequences was inserted into pHD309-puro to generate a TbTR-expressing construct targeted to the tubulin array.

Cell lines and cell culture

BF T. brucei cells were cultured in HMI-9 medium, while PF T. brucei cells were cultured in SDM-79 medium, both with appropriate antibiotics. BF 427 lister strain SM expressing the T7 polymerase and Tet repressor32 or PF 427 lister strain WT427 was used for generating TbTR-DKO cell lines, C1 and C2 (BF) or C6 and C21 (PF), by sequential transfecting the two TbTR-KO constructs. C1 cell line was cultured for over 250 PD before transfected with pLew111-TbTRutr to generate the TbTR complementation cell lines B1 and B2. C1 cells were subcultured three times to generate a clone that carries a short VSG2-marked telomere. This clone was transfected with pLew111-TbTRt-utr to give rise to the clones that express the mutant TbTRt as the only TbTR allele.

Transfection

T. brucei cell transfection was done using a Nucleofector with Basic Parasites Buffer 1 provided by the manufacturer (Lonza). Program X-001 and W-14 were used to transfect BF and PF cells, respectively.

Telomere Southern blot

Genomic DNA was isolated from T. brucei cells using a PCIA extraction-based protocol and digested with AluI and MboI followed by segregation on 0.7% agarose gel. The agarose gel was then dried and washed with denaturing buffer and neutralization buffer consecutively. The gel was then hybridized with a radiolabeled (TTAGGG)4 oligo probe.

Northern blot

For northern blots shown in Figures 1A, 4A, 4C, Supplementary information, Figure S3A and S3E, total RNA was isolated from T. brucei cells using RNAstat 60 and separated by agarose gel electrophoresis. A 900 bp TbTR fragment was used as the probe. For northern blot shown in Supplementary information, Figure S2A, total RNA was hybridized with an antisense TbTR oligo (5′-TAGTAGGGTTAGGGATCGTATAGCCAAGAAAACAC-3′) probe.

IP and western blotting

A mouse polyclonal anti-sera was raised against TbTERT peptide, aa 526-539 (KRMRSDTLSDPQMR). For IP, cell lysate was prepared from 1-5 × 108 T. brucei cells homogenized in IP buffer (10 mM Tris, pH 7.5, 150 or 300 mM NaCl and 1% Nonidet P-40 with 7 mM PMSF and protease inhibitor cocktail set III, EMD/Millipore) at a 10% wt/vol ratio. 5 μg/ml of anti TbTERT anti-sera or preimmune sera were incubated with 200 μg lysates for 2 h at 4 °C. Samples were processed as described earlier74. The presence of TbTERT in the immunoprecipitated samples was detected by western blotting with anti-TbTERT (1:250). Alternatively, precipitated complex was eluted from protein G-Agarose with epitope peptide (1 mg/ml) for 30 min at room temperature (or overnight at 4 °C) followed by centrifugation for 20 s at 12 000× g.

RNA IP was done following published procedure75 using 5-10 μg total RNA incubated with 15 μl of anti-TMG antibody (K121, Millipore). RNA integrity was checked by a Nanodrop (Thermo) or a Bioanalyzer (Agilent Technologies).

RNA analysis by RT-PCR and RLM-RACE

RT-PCR on total RNA (2-10 μg) or RNA isolated from RNP complex (0.2-0.5 μg) was done after cDNA synthesis with Superscript II RT (Invitrogen) using random primers or TbTR gene-specific primers (Supplementary information, Table S1). To map the 5′ and 3′ end of RNA transcripts, total RNA or IP-RNA was analyzed with FirstChoice RLM-RACE Kit (Life Technologies). For Oligo dT priming, Oligo dT (18)/(20) (Invitrogen) and thermostable reverse transcriptase (Maxima RT, Thermo Scientific) were used.

Chemical probing of RNA structure

To probe the conformation of TbTR in vivo, cellular RNA was isolated from parasite nuclei treated with DMS (dimethyl sulfate) and CMCT (1-cyclohexyl-(2-morpholinoethyl)carbodiimide metho-p-toluene sulfonate), and then analyzed by primer extension according to Ullu et al.65 with following modifications: Nuclei were isolated from 2 × 1010 procyclic cells in 10 mM HEPES–KOH pH 7.9, 1.5 mM MgCl2, 10 mM KCl after homogenization in a Dounce homogenizer by 25 strokes with pestle A and collected by centrifugation at 3 500× g for 5 min in Beckman JS-7.5 rotor. Nuclei suspension was treated with 0.1 mg/ml lysolecithin followed by centrifugation at 2 000 rpm. at 4 °C and resuspended in HEPES-KOH buffer (150 mM sucrose, 10 mM KCl, 10 mM HEPES, pH 7.9, 3 mM MgCl2 and 1 pg/ml leupeptin) before the addition of DMS. Finally, DMS diluted in ethanol (1:4) was added and incubated for 5 min at room temperature. The reaction was stopped by addition of 2-Mercaptoethanol (100 mM), and RNA was extracted by Trizol (Invitrogen). Optimal DMS concentration and time of cell treatment were determined by using a 10-fold DMS concentration range according to suggested procedures76. For the CMCT reaction, CMCT (50 mg/ml in water) was added to nuclei suspension at room temperature for 15 min. Reactions were stopped by adding Trizol followed by RNA extraction. RNAs were DNase treated and precipitated twice after extraction with Trizol. The positions of chemical modifications were mapped by primer extension as described earlier76.

Telomerase activity assay

Direct telomerase activity assays were performed according to published procedure74 using (TTAGGG)3 as the substrate in Figure 1E. 5 μl of the sample was electrophoresed over a 10% urea-polyacrylamide sequencing gel.

For the PCR-based TRAP assay, 10 million T. brucei cells were lyzed in 100 μl of CHAPS buffer from the TRAPeze© Kit (Millipore). In each reaction, 0.1–0.5 μg of protein extract, 40 U of RNasin, 2 pmole of γ-32P end-labeled TS primers, and reverse primers for PCR amplification were mixed with 0.05 mM of dNTP, 20 mM TrisCl pH 8.3, 1.5 mM MgCl2, 63 mM KCl, 0.05% Tween 20 and 1 mM EGTA. 20 ng of RNase A was added together with the cell lysate in reactions treated with RNase. Telomerase-mediated primer extension was carried out at 30 °C for 30 min followed by PCR amplification using the TS and reverse primers. Products from TRAP assays were analyzed on 12.5% nondenaturing PAGE.

Cloning of terminal fragments of telomeres

Unique oligo 5′-CCCTATAGTGAGTCGTATTA-3′ was treated with polynucleotide kinase and ATP and annealed to Guide oligo 5′-ACGACTCACTATAGGGACCC-3′ to generate the adaptor. 10 pmole of adaptor was ligated with 1 μg of genomic DNA followed by PCR using 5′-TTAGGGTTAGGGTTAGGGTTAGGG-3′ and the Guide oligo as primers. The PCR product was then inserted into pGEM-T-easy vector (Promega) and the resulting plasmid was sequenced with T7 and Sp6 primers.