Telomere and Subtelomere R-loops and Antigenic Variation in Trypanosomes

The R-loop structure

R-loops are three-stranded RNA-DNA structures with an RNA:DNA hybrid and a displaced single-stranded DNA ¹¹. The replication-associated RNA:DNA hybrids are usually 11 bp long (in Okazaki fragments) and transcription-associated RNA:DNA hybrids are generally 8 bp long (within the RNA polymerase active site) ¹². However, longer R-loops (> 1 kb) can also form ¹³. Currently the thread back hypothesis is the best working model for R-loop formation: DNA behind a transcription bubble is negatively supercoiled, which has a tendency to unwind, allowing the nascent RNA to anneal with the template strand easily ^{1, 14}. In support of this, defective transcription elongation and termination, RNA splicing, and relaxation of supercoiled DNA all lead to elevated R-loop levels ^15–18. In addition, DNA nicks on the nontemplate DNA strand downstream of the promoter and G clusters facilitate initial R-loop formation, while subsequent expansion and stabilization of the RNA:DNA hybrid are enhanced by high G density and negative supercoiling ¹⁹.

R-loops play important roles and are necessary for several cellular processes ¹. First, in mammals, R-loops drive the programmed genomic rearrangement during immunoglobulin class switch in activated B-cells ²⁰. Second, R-loops formed at the promoter regions can modulate gene expression ^21–23. Third, R-loops are associated with H3 S10 phosphorylation, a heterochromatic marker, suggesting that it has an important role in chromatin structure modulation ²⁴. Fourth, in human mitochondria, origin-specific DNA replication is initiated with a two step process: (1) transcription from an upstream promoter leads to accumulation of R- loops, and (2) processing the R-loops by RNase H1 generates 3’ ends for DNA replication by DNA polymerase γ ²⁵. However, R-loops are also well-known to be a genome instability factor ²⁶. The displaced single-stranded DNA can be easily mutated and contributes to transcription-associated mutagenesis ²⁷, recombination, and DSBs ^{28, 29}. A stable R-loop structure can also block the progression of the replication fork ^19,30, although the underlying mechanism is still not fully understood ³. It is also possible that R-loops can trigger chromatin compaction ^31–33, which in turn can block replication fork progression ³. Additionally, nucleotide excision repair endonucleases can process R-loops into DSBs ³⁴.

R-loops are quite stable, as RNA:DNA interactions are thermodynamically more stable than DNA:DNA pairing ³⁵, and many enzymes are involved in dissolution of the R-loop. Several RNA helicases including Rho ³⁶, DHX9 ³⁷, and Senataxin ³⁸ can unwind the RNA:DNA duplex. Ribonuclease H1 (RNase H1) can degrade the RNA strand of the RNA:DNA hybrid and requires a tract of at least four ribonucleotides in the substrate ³⁹. Eukaryotic Ribonuclease H2 (RNase H2), usually a trimer ^{40, 41}, can resolve the R-loop ^{42, 43} and remove single ribonucleotides from genomic DNA ^44–46. These enzymes play important roles in maintaining an appropriate level of R-loops in cells ⁴⁷.

Recent studies in a kinetoplastid parasite, Trypanosoma brucei, have shown that telomere and subtelomere R-loops can induce DNA damages, not only contributing to genome instability but also emerging as an important factor that influences antigenic variation ^6–8. These results further indicate that R-loops can have both beneficial and detrimental effects, and its level needs a tight regulation.

Antigenic variation in T. brucei

Trypanosomsa brucei causes human and animal African trypanosomiasis, which are usually fatal without treatment. T. brucei is transmitted by tsetse (Glossina spp.) and threatens millions of people in sub-Saharan Africa ⁴⁸. The bloodstream form (BF) T. brucei proliferates in the extracellular space of the mammalian host. Its major surface antigen, Variant Surface Glycoprotein (VSG), forms a dense protein layer ⁴⁹, masks other invariant surface antigens, and can elicit strong immune responses ⁵⁰. However, T. brucei regularly switches its VSG coat, effectively evading its elimination by the host ⁵. This antigenic variation is a critical pathogenesis mechanism and has two major aspects: VSG monoallelic expression and VSG switching.

There are more than 2,500 VSG genes and pseudogenes (all are located at subtelomeric regions) in the T. brucei genome ⁵¹. Many of these are in large VSG gene arrays on megabase chromosomes (Fig. 1A, top) ^52–54. Individual VSG genes are also found at two-thirds of minichromosome subtelomeres (Fig. 1A, bottom) ^{51, 55, 56}. VSG genes located at these loci are normally not expressed. At the BF stage, VSGs are expressed exclusively from BF VSG Expression Sites (B-ESs, Fig. 1B, top), which are large polycistronic transcription units (PTUs) transcribed by RNA Pol I and are located immediately upstream of the telomere on megabase or intermediate chromosomes (Fig. 1A, top and middle) ^{57, 58}. VSG is the last gene in any B-ES and is located within 2 kb from the telomeric TTAGGG repeats in all completely sequenced B-ESs, while the B-ES promoter is located 40 – 60 kb upstream (Fig. 1B, top) ^{52, 59, 60}. 70 bp repeats are found immediately upstream of the VSG gene in long arrays (3 – 20 kb) in B-ESs ^{52, 59}. T. brucei has multiple B-ESs (~ 15 in the Lister 427 strain), all with similar gene organization and ~ 90% sequence identity ⁵⁹, but often having different VSG genes ⁵⁹. However, at any moment, only one B-ES is fully active, presenting a single type of VSG on the cell surface ⁴⁹. Monoallelic VSG expression ensures effectiveness of VSG switching, as the previously active VSG needs to be silenced for the parasite to avoid elimination by the host. Although detailed mechanisms are not fully understood, many factors have been shown to regulate VSG expression ⁶¹, such as chromatin structure ^62–72, transcription elongation ^73–75, inositol phosphate pathway ^{76, 77}, nuclear lamina ^{78, 79}, recruitment of sumoylated protein(s) to the active ES promoter ⁸⁰, DNA replication initiation factors ^81–84, a subtelomere and VSG-associated VEX complex ^{85, 86}, and telomeric silencing ^{64, 87}. The VSG coat is lost when T. brucei is ingested by its insect vector, tsetse ⁸⁸. At the same time, T. brucei differentiates into the procyclic form (PF). After migrating to the salivary gland of tsetse, the infectious metacyclic form T. brucei expresses VSG again and is ready to be injected into a mammalian host ⁸⁹. At this stage, VSGs are expressed from metacyclic ES (M-ESs, Fig. 1B, bottom) that are also located at subtelomeric regions and are transcribed by RNA Pol I ^{90, 91}, except that M-ESs are monocistronic transcription units with the promoter located < 5 kb from the telomere ^92–96.

VSG switching only occurs in proliferative BF T. brucei cells ⁹⁷. It has two major pathways ^{98, 99}. In-situ switching occurs at the transcriptional level, where the originally active ES is silenced while a previously silent ES becomes fully transcribed, and no gene rearrangement is involved (Fig. 2A, bottom left) ^98,99. The second and more frequent switching event is DNA recombination-mediated and includes two major types of process ¹⁰⁰. In crossover (CO, or telomere exchange, Fig. 2A, bottom right), a silent VSG (often ES-linked or at a minichromosome subtelomere) exchange places with the active VSG reciprocally, resulting in a new VSG being expressed from the originally active ES with no loss of genetic information ^{101, 102}. In gene conversion-mediated VSG switching (GC, Fig. 2A, top right), a previously silent VSG gene is copied into the active ES to replace the originally active VSG, which is subsequently lost ^{103, 104}. Theoretically, any functional VSG gene in the genome can act as a donor in GC-mediated VSG switching. However, sequences of different B-ESs are 90% identical ⁵⁹, GC can often occur between B-ESs as they have long homologous sequences ^{103, 105}. Additionally, all ES-linked and minichromsome VSGs are expected to be good GC donors as they have long telomere repeats downstream and 70 bp repeats upstream, the same as the active VSG (Fig. 1). Indeed, telomere conversion-mediated VSG switching events have been observed ¹⁰⁶. When telomere-adjacent VSGs are used as donors, it is possible that VSG switching may occur through break-induced-replication (BIR) ¹⁰⁷ instead of a classical GC (Fig. 2B). However, whether BIR occurs in VSG switching has not been investigated. Another type of GC has also been observed in VSG switching: several VSG genes can act as donors simultaneously, each donating a piece of the gene segment, resulting in a novel mosaic VSG gene ^108–111. In many published switching assays, in situ switching events are less frequent than recombination-mediated ones, while gene conversion is the most prevalent ^6,112–115.

Many proteins, especially those involved in DNA recombination, have been shown to play important roles in VSG switching. For example, RAD51 that mediates strand invasion in homologous recombination ¹¹⁶, RAD51–3 (a RAD51-related protein) ¹¹⁷, and BRCA2 ¹¹⁸, all facilitate VSG switching, while Topoisomerase 3 alpha ¹¹⁴, the RMI1 homologue ¹¹⁵, replication origin binding factor TbORC1 ⁸¹, and a RecQ helicase, RECQ2 ¹¹⁹, suppress VSG switching. Additionally, telomere proteins have been shown to suppress VSG switching ^{6, 112, 120, 121}, while cells harboring an extremely short active VSG-adjacent telomere have a ~ ten-fold higher VSG switching rate than WT cells ¹²². In addition, cells with defective RNase H enzymes appear to have a higher VSG switching frequency than WT cells ^{7, 8}. However, how VSG switching is initiated and regulated is less well-understood ^{123, 124}.

T. brucei telomere proteins and antigenic variation

VSG is expressed exclusively from subtelomeric regions ⁶⁰, and the telomere complex has been shown to play important roles in antigenic variation in T. brucei ^{6, 87, 112, 120, 121, 125}. Telomeres are nucleoprotein complexes that are essential for genome integrity and chromosome stability in eukaryotes ^{126, 127}. Proteins that directly bind the telomere DNA or associate with the telomere chromatin through protein-protein interaction play critical roles in all aspects of telomere functions, including proper maintenance of the telomere length ¹²⁸, suppression of illegitimate DNA degradation, recombination, and repair at the chromosome ends ¹²⁹, assembly of a telomere heterochromatin that represses subtelomeric gene expression ¹³⁰, and regulation of the telomeric transcript level ¹³¹.

In T. brucei, besides the telomerase components that are required for telomere maintenance ^132–134, several other telomere proteins have been identified ^{87, 121, 125, 135}. TbTRF directly binds the duplex TTAGGG repeats through its C-terminal Myb domain ¹³⁵, and the telomere DNA binding activity is essential for cell viability and telomere/subtelomere stability: cells transiently depleted of TbTRF and the TbTRF mutants with weakened telomere-binding activities have more VSG switching events that mostly involve the loss of the active ES ¹¹². A TbTRF-interacting factor, TbTIF2, is essential for normal cell growth and has a critical role in maintaining telomere/subtelomere integrity: depletion of TbTIF2 results in increased amount of DSBs in ESs, and a transient depletion of TbTIF2 leads to more frequent VSG switching, with most events involving the loss of the active ES ¹²¹. TelAP1 is identified in the TbTRF protein complex and in the complex that can interact with an oligonucleotide of telomeric sequence ¹²⁵. It is the only non-essential telomere protein identified so far, and TelAP1 null cells exhibit a faster VSG silencing kinetics when cells are differentiated from BF to PF in vitro ¹²⁵.

TbRAP1 is a TbTRF-interaction factor that associates with the telomere chromatin and is essential for cell viability ⁸⁷. Depletion of TbRAP1 leads to derepression of all ES-linked silent VSGs (upto several thousand-fold), while the originally active VSG is expressed at ~ 50% of its WT level ^{64, 87}. Because transcription from silent B-ES promoters is detected but transcription elongation attenuates within a few kb ^{74, 75}, it is hypothesized that depletion of TbRAP1 removes a blockage of transcription elongation, resulting in basal level VSG expression from all silent ESs ⁸⁷. This is consistent with the observation that promoter-less VSGs at minichromosome subelomeres are not derepressed after TbRAP1 depletion ^64,87. Additionally, the TbRAP1-mediated telomeric silencing is position-dependent, exerting strongest effects on telomere-adjacent VSG genes, weaker effects on VSG pseudogenes in the middle of B-ESs, and weakest effect on reporter genes inserted immediately downstream of the B-ES promoter ⁸⁷. The PF T. brucei cells proliferate in the midgut of tsetse and express procyclins as its major surface protein ¹³⁶. VSGs are normally silenced at this stage ⁸⁸ but derepressed upon TbRAP1 depletion ⁶⁴. Additionally, telomeric and subtelomeric chromatin structure is less compact when TbRAP1 is removed in PF cells ⁶⁴. TbRAP1 also plays an important role in maintaining telomere/subtelomere integrity and stability, which helps suppress VSG switching ⁶. Interestingly, the underlying mechanism involves the telomeric transcript (TERRA) and telomere R-loops.

TERRA in T. brucei

The UUAGGG repeat-containing telomere transcript (TERRA) was first identified in T. brucei cells and several closely related kinetoplastid parasites three decades ago ¹³⁷. In recent years, TERRA has been detected in all eukaryotes tested ^138–142 and has been shown to be the product of telomere repeat transcription in humans, mouse embryonic stem cells, both budding and fission yeasts, and plants ^{140, 141, 143–149}. Often, all telomeres are not transcribed ^{150, 151}, and transcription from intrachromosomal telomeric sequences can be abundant ¹⁴⁹. TERRA has been shown to play important roles in telomere protection ^{138, 151,153}, length regulation ^{148, 154, 157}, and recombination ¹⁵⁸ in mammalian cells and yeasts. TERRA may also play a role in gene expression regulation in mouse embryonic stem cells ¹⁵². In addition, an excessive amount of TERRA led to telomere and subtelomere instability in yeast ¹⁵⁸, presumably due to its propensity to form telomere R-loops ^{159, 160}, which is known to induce DSBs and cause genome instability ^4,161_.

The telomere transcript was detected in both BF and PF T. brucei cells and only a fraction of this RNA is polyadenylated ¹³⁷. Most strikingly, the TERRA level is resistant to 1 mg/ml alpha-amanitin, suggesting that TERRA is transcribed by RNA polymerase I ¹³⁷. Knowing that the active VSG is transcribed at a very high level from a subtelomeric ES by RNA Pol I that is resistant to alpha-amanitin, it was hypothesized that TERRA is transcribed by RNA Pol I as a product of read-through into the telomere repeats downstream of the active VSG gene ¹³⁷. This hypothesis was confirmed recently by Nanavaty et al. ⁶. All B-ESs are PTUs ^{52, 59, 162}, and nascent polycistronic transcripts are processed through trans-splicing, where a common spliced leader (SL) sequence is added to the 5’ end of individual mRNAs ^{163, 164}. After reverse transcription of total RNA using a CCCUAA primer, Nanavaty et al. were able to detect a PCR product using primers specific to the active VSG gene, indicating that the un-processed nascent RNA containing both the telomeric UUAGGG and the active VSG sequences exists, which confirms that TERRA indeed is transcribed from the telomere downstream of the active ES ⁶. No TERRA product was detected from silent ES-adjacent telomeres ⁶. However, T. brucei has more than 200 telomeres that do not host any ESs ⁵². Whether TERRA can be transcribed from ES-free telomeres is still unknown. Our lab has now performed a TERRA FISH analysis in BF T. brucei cells using the TELC-PNA probe (PNA Bio). We frequently observe only one or two TERRA foci in each T. brucei nucleus (Fig. 3A), suggesting that TERRA is transcribed from few telomeres if not only from the active ES-adjacent telomere. Additionally, we performed TERRA FISH and TbTRF IF analysis simultaneously and found that TERRA is colocalized with TbTRF at the telomere (Fig. 3A).

Interestingly, TERRA is detected in PF T. brucei cells where VSG is not transcribed ^{137, 165}, suggesting that not all TERRA is transcribed as a read-through product. Our lab has also detected both UUAGGG and CCCUAA repeat-containing TERRA species in PF T. brucei cells (Fig. 3B). In addition to the TERRA species with various sizes (from 0.5 kb to 10 kb, shown in northern blotting as a smear), PF T. brucei also transcribes both G-rich and C-rich TERRAs with a more discrete size (~ 1 kb) (Fig. 3B). However, the origins of both UUAGGG and CCCUAA repeat containing TERRA species in PF T. brucei cells are currently unclear.

Although the function of T. brucei TERRA is not clear, TERRA has been shown to form telomere R-loops ⁶, and higher than WT levels of telomere R-loops result in an increased amount of subtelomeric and telomeric DSBs and an elevated VSG switching rate ^6–8.

Telomere and subtelomere R-loops and antigenic variation

Telomere R-loops are detected in WT T. brucei cells ⁶ by the monoclonal antibody S9.6 that specifically recognizes the RNA:DNA hybrid ¹⁶⁶. Depletion of TbRAP1 leads to not only a higher level of TERRA but also more telomere R-loops and an increased amount of telomeric and subtelomeric DSBs, which in turn cause more frequent VSG switching ⁶. The increased amount of DSBs at telomeres and subtelomeres is mainly mediated by the increased amount of telomere R-loops, as expression of an ectopic allele of TbRNase H1 in TbRAP1-depleted cells brings telomere R-loops and telomeric/subtelomeric DSBs back to WT levels, which further reduces the VSG switching rate back to its WT level ⁶.

Although a higher level of TERRA is frequently associated with an elevated amount of telomere R-loops and vice versa ^{158, 160, 167, 168}, this may not always be the case in T. brucei. It has been shown that the R-loop level is influenced by RNA processing ^{169, 170}. High transcription level and poor RNA processing can both lead to an increased level of R-loops. However, it is unknown whether TbRAP1 plays any direct roles in dissolution of the telomere R-loop or whether TbRAP1 affects premature RNA (containing both UUAGGG repeats and the upstream VSG sequences) processing. When an ectopic allele of TbRNase H1 is expressed in the TbRAP1-depleted cells, the TERRA level is still much higher than that in WT cells, even though the amount of telomere R-loop is reduced to the WT level ⁶, suggesting that TERRA and the telomere R-loop may be regulated independently. It would be interesting to further examine the relationship between telomere R-loop and TERRA in T. brucei.

Two recent studies on T. brucei ribonuclease H enzymes also showed that R-loops at the telomere and the subtelomere influence VSG switching frequencies ^{7, 8}. T. brucei has two RNase H enzymes, a non-essential TbRNase H1 ¹⁷¹ and an essential TbRNase H2 ⁸. DRIP-seq experiments (R-loop immunoprecipitation followed by high-throughput sequencing analysis) were done to map which genomic loci have R-loops in WT and RNase H defective cells ^{7, 8, 172}. In WT cells, R-loops are detected at the region immediately downstream of the active VSG gene, and a much lower level of R-loops is observed immediately downstream of a silent VSG gene ^{7, 8}. More R-loops are clearly detected at both of these regions in TbRNase H1 null cells ⁷. Depletion of TbRNase H2A (the catalytic subunit of TbRNase H2) has a similar phenotype ⁸, indicating that both RNase H enzymes influence R-loop levels at the telomere and subtelomere junction.

Interestingly, in both TbRNase H1 null and TbRNase H2A-depleted cells, increased amounts of R-loops are detected across the active and silent ESs, with the most prominent increase at the 70 bp repeats ^{7, 8}. The heterogeneous ~ 70 bp repeats are found upstream of most VSG genes (Fig. 1 & 2) ⁵¹. The repeats often contain varying numbers of tandem TAA triplets and their sizes range from 66 to 81 bp ^{173, 174}. Two highly conserved motifs (AGTGTTGTGAGTGTG and TATAATAAGAGCAGTAAT) have been identified and 83% of the 70 bp repeats being studied contain one or both of these motifs ¹⁷⁵. In VSG gene arrays, usually few copies of 70 bp repeats are upstream of a VSG gene ^{52, 162}. While in B-ESs, 3 – 20 kb of 70 bp repeats are upstream of the VSG gene ⁵⁹. In WT BF cells, no stable transcripts from the 70 bp repeats have been detected, even though the active ES is highly transcribed by RNA Pol I, suggesting that RNA processing is very efficient. However, in TbRNase H1 null and TbRNase H2A-depleted cells, a significant amount of R-loops are detected at the 70 bp repeat region in both active and silent ESs ^{7, 8}. Consistent with the notion that R-loops often induce DNA damages ³, more γH2A (DNA damage associated histone H2A with phosphorylated T130 ¹⁷⁶) associates with the active ES chromatin, particularly at the telomere-proximal region, in TbRNase H1 null cells ⁷, suggesting telomere-proximal R-loops are more stable. When TbRNase H2A is depleted, much more γH2A associates with both active and silent ES chromatin throughout the whole ES ⁸. Furthermore, more switchers are observed to have shed the originally active VSG on the cell surface in the RNase H defective cells, suggesting that these cells have an increased VSG switching rate than WT cells ^{7, 8}. Therefore, increased amounts of telomere/subtelomere R-loops in RNase H defective cells are linked with elevated DNA damage levels at ESs and more frequent VSG switching ^{7, 8}

In TbRNase H defective cells, an increased amount of R-loops is detected at the telomere/subtelomere junction ^{7, 8}. It is likely that telomere R-loops are also stabilized in these cells, although this has not been tested directly. It is also unknown whether TERRA levels are increased when the RNase H enzymes are deleted or depleted. In TbRNase H1 null and TbRNase H2A-depleted cells, the mRNA levels of a number of silent VSGs are increased (upto several ten-fold) ^{7, 8}, suggesting that TbRNase H1 and H2 may be important for VSG silencing. R-loops have been shown to affect gene expression in other organisms ^21–23. Therefore, it is possible that VSG expression is affected by nearby R-loops and that TERRA level is also increased in TbRNase H defective cells. On the other hand, frequent VSG switching in the RNase H defective cells ^{7, 8} could also lead to mildly increased VSG mRNA levels when a cell population is examined. Cells expressing both the originally active VSG and an originally silent VSG simultaneously are observed by IF in the TbRNase H defective cells ^{7, 8}. However, these cells might be in the middle of a VSG switching process. Whether the TERRA level is affected by the RNase H enzymes needs further investigation.

It is important to note that functions of TbRNase H1 and H2 are not limited at ES regions ^{7, 8, 172}. A recent DRIP-seq analysis detected R-loops in multiple T. brucei genome loci in WT cells ¹⁷². In TbRNase H1 null cells, increased levels of R-loops are also observed at transcription start sites of RNA Pol II transcribed PTUs, while no significant increase of DNA damage at these sites is seen ¹⁷². It is also unknown whether TbRNase H1 affects mRNA levels of genes located outside of ESs. On the other hand, depletion of TbRNase H2A leads to a dramatic increase in the amount of genomic DNA damage, particularly at the transcription initiation sites ⁸. Additionally, several tens of genes other than VSG, ESAG, or procyclin genes exhibit increased mRNA levels when TbRNase H2A is depleted ⁸.

Subtelomeric regions are often composed of various repeats and gene families ^177–179. High polymorphism in the subtelomere is frequently observed among different chromosome ends and individuals in humans ^{180, 181}, yeast ^{177, 182}, fly ¹⁸³, plant ¹⁸⁴, and fungal pathogens ^{185, 186}. T. brucei subtelomeres also exhibit dynamic variations: the T. brucei homologous megabase chromosome pairs often differ greatly in size (Fig. 1A) ¹⁸⁷. Several factors contribute to this size polymorphism: subtelomeric ESs and VSG gene arrays have different sizes, telomere lengths vary at different chromosome ends, and repetitive chromosomal regions vary in size ¹⁸⁸. Importantly, two-thirds of the size polymorphisms are due to variations in subtelomeric regions, while chromosomal core regions, containing all essential genes, are relatively stable ⁵³.

Telomere dysfunctions are well-known to induce genome instabilities. At the telomere vicinity, unprotected telomeres lead to chromosome end-to-end fusions ^{189, 190}, anaphase bridges ^{191, 192}, and telomere recombination ^{193, 194}. Telomere fusions in human cells can further induce a persistent mitotic arrest that leads to greatly increased cell lethality ¹⁹⁵. At a global level, telomere crisis results in dicentric chromosome formation and subsequent chromothripsis and kataegis ¹⁹¹. Dysfunctional telomeres can also lead to subtelomere instability in yeast ^{196, 197}. In telomerase null cells, Type I and Type II survivors use DNA recombination-dependent mechanisms to maintain their telomere length ¹⁹⁶, where Type I survivors amplify their subtelomeric Y’ elements in a Rad51-dependent pathway ^{197, 198}. Studies in T. brucei have shown that depletion of telomere proteins, TbTRF, TbRAP1, and TbTIF2, all result in unstable subtelomeres ^{6, 112, 120, 121}.

Most telomere dysfunctions result in severe genome instability and cause cell growth defects ¹²⁹. In T. brucei, VSG is essential ¹⁹⁹, and damages to the active VSG gene are generally poorly tolerated: Introducing an artificial DSB (an I-SceI cut) within or near the active VSG gene leads to cell death in more than 80% of the cell population ²⁰⁰, which also leads to a 250-fold higher VSG switching rate ²⁰¹. It is possible that the I-SceI cut is not repaired efficiently due to continued I-SceI expression. However, the location of the damage site appears to be a critical factor, as inducing the same I-SceI cut in a silent ES is much better tolerated ²⁰⁰. TbTIF2 and TbRAP1 are essential proteins that associate with the telomere chromatin ^{87, 121}, and depletion of these proteins induces DNA damages mainly in the active and silent ESs ^{6, 121}. These observations are consistent with the idea that DNA damages in the active ES are poorly tolerated. Depletion of TbRNase H2A results in global increased amounts of DNA damages at transcription initiation sites ⁸, suggesting that TbRNase H2A may be important for genome integrity at non-subtelomeric regions. Therefore, it is hard to interpret whether the increased amounts of DNA damages in ESs contribute significantly to the growth defect in TbRNase H2A-depleted cells. On the other hand, some telomeric dysfunction and subtelomeric damages are better tolerated, contributing to subtelomeric plasticity. Single damages within silent ESs do not severely affect cell growth ²⁰⁰. Telomeres downstream of the silent ESs can be as short as < 100 bp without inducing any cell growth defect ²⁰². Additionally, cells do not experience cell cycle arrest when a single telomere (without any adjacent ES) is deleted ²⁰³. Therefore, damages that do not directly disrupt the active VSG gene seem less detrimental to cell growth than the ones that do. However, TbRNase H1 null cells represent an exception, where an increased amount of DNA damages is detected in the active ES, but the cells are viable ⁷. Although it is not strictly comparable among different studies, loss of TbRNase H1 does not seem to cause an as high level of DNA damage in ESs as depletion of TbTIF2 or TbRAPl ^{6, 7, 121}. In the latter two cases, DNA damages are observed in both active and silent ESs, while loss of TbRNase H1 only causes a moderate increase in the amount of DNA damages in the active ES ^{6, 7, 121}. Therefore, at T. brucei subtelomeres, the amount of damages probably also contributes to its effect on cell fitness. In consistence with this, some DNA breaks are detected at 70 bp repeat region in the active ES in WT cells ²⁰¹, suggesting that cells can tolerate a low level of subtelomeric damages, although the exact amount of telomere and subtelomere damages in WT cells is unknown.

These observations suggest that the balance between plasticity and integrity at the T. brucei subtelomere is a key factor of keeping a balance between antigenic variation and cell fitness, both important for parasite survival. A similar balancing act is necessary for proper telomere maintenance in human ALT cells ¹⁶⁸, where a telomerase-independent and DNA recombination-dependent telomere maintenance mechanism is critical for cell survival ²⁰⁴. In ALT cells, overexpression of TERRA leads to an increased level of telomere R-loops and many more telomere recombination events ¹⁶⁸. Importantly, depletion of RNase H1 in these cells leads to an accumulation of telomere R-loops and C-circle excision-mediated rapid telomere shortening ¹⁶⁸.

Overexpression of RNase H1 in ALT cells reduces the amount of telomere R-loops, which hinders telomere recombination potential and results in gradual telomere shortening ¹⁶⁸. Therefore, perturbing telomere R-loop levels in ALT cells disrupts the delicate balance between recombination-mediated telomere attrition and maintenance.

Achieving a good balance between telomere/subtelomere stability and subtelomere variation may be feasible through an introduction of the right amount of telomere/subtelomere damages ²⁰⁵. In Kluyveromyces lactis, variation in a subtelomeric gene family encoding β-galactosidase allows yeast to better cope with different nutrition ²⁰⁶, a scenario not too different from VSG switching in T. brucei. Mild telomere dysfunction that does not induce global genome instability leads to increased variation of the subtelomere β-galactosidase-coding genes, while severe telomere dysfunction causes complete deletion of these genes ²⁰⁶. Therefore, mild telomere dysfunction can serve as a beneficial drive for subtelomere variations that allow cells to better adapt to environmental stresses ²⁰⁵. The TbRNase H1 null cells ⁷ may be a good example of having a mild telomere/subtelomere dysfunction in T. brucei, as the amount of R-loop-induced subtelomeric DNA damages appears not severe enough to trigger cell growth defect but sufficient to stimulate VSG switching, which is presumably beneficial for a long-term parasite survival inside the mammalian host. Since TbRNase H1 null cells are viable, it would be interesting to examine whether these cells have increased virulence when infecting an animal host.

Telomere/subtelomere R-loops and initiation of VSG switching

Homologous recombination (HR) is a major pathway for VSG switching ¹⁰⁰. For ES-linked and minichromosome VSGs, the telomere sequence is found downstream of VSG^{51, 52, 59}. Additionally, most VSG genes have a common 14 nt sequence in their 3’UTR region ⁵¹. It has been proposed that the common 14 nt VSG 3’UTR sequence and sometimes the telomere sequence can serve as the downstream homologous arm, while the 70 bp repeats can serve as the upstream homologous arm for efficient homologous recombination that mediates a VSG switch ²⁰⁷. However, an earlier study showed that the 70 bp repeats upstream of the VSG gene in an ES is not required for VSG switching ¹⁰³. In a recent study that introduces an I-SceI cut in the active ES to initiate VSG switching, it is found that 70 bp repeats in the active ES promote selection of VSG donors from the genomic archive rather than only from silent ESs ²⁰⁷.

Although a DSB is not absolutely required for HR, it is a good inducer for HR and frequently repaired by HR ²⁰⁸. Therefore, it has been hypothesized that DSBs are an important trigger of VSG switching ^{123, 201}. In support of this, introducing an artificial DSB immediately upstream of the active VSG leads to a more than 250-fold higher VSG switching frequency ²⁰¹, and DSBs introduced near the active VSG (both upstream and downstream) result in efficient VSG switching ²⁰⁰. DSBs can also be detected in 70 bp repeats by Ligation mediated PCR analysis in WT T. brucei cells ²⁰¹, although the amount of telomeric and subtelomeric DNA damage in WT cells has not been carefully quantified. Interestingly, the active ES-adjacent telomere experiences frequent large fragment deletions ²⁰⁹, which may be a consequence of TERRA transcription by RNA Pol I ^{6, 137} that depletes most nucleosomes in the active ES ^{67, 68} and may also interfere with the binding of TbTRF at the telomere. Therefore, DNA breaks in both the telomere and 70 bp repeats are possible to serve as an inducer for VSG switching. Several mechanisms are possible for DSB formation in these repetitive sequences.

First, as repetitive sequences, both telomeres and 70 bp repeats can be difficult to be replicated due to strand slippage during DNA replication ²¹⁰ and secondary structure formation ²¹¹ (such as the G-quadruplex structure formed by the telomere G-rich strand DNA ²¹² and the non-H bonded structure formed by TTA/TAA in the 70 bp repeats ²¹³). In mammalian and yeast cells, several telomere proteins have been shown to play important roles in ensuring proper telomere replication ^214–216. Whether T. brucei telomere proteins have similar functions still requires further investigation. Nevertheless, obstacles in DNA replication often lead to stalled replication fork and eventual DSBs ^{217, 218}. Second, it is shown that the active B-ES is replicated in early S phase ¹¹⁹. Hence, the transcription machinery may collide with the DNA replication machinery in the active B-ES and its adjacent telomere, which often causes fork stalling and collapsing followed by DSB formation ²¹⁹. Third, the active ES and its adjacent telomere are transcribed at high levels by RNA Pol I ^{6, 220}, which can induce more DSB formation with or without the R-loop formation. RNA Pol I transcription depletes nucleosomes from the active ES ^{67, 68} and can, presumably, also remove some TbTRF from the telomere. The exposed DNA in the active ES and the adjacent telomere is likely more vulnerable to nuclease attack, which can result in more DSBs. Additionally, the traverse of RNA polymerase along repetitive DNA allows formation of DNA secondary structure, which helps DSB formation ²²¹. Furthermore, the high levels of transcription can induce transcription-associated recombination (TAR) ²²², and RNA Pol I transcription in BF T. brucei cells stimulates homologous recombination > 3 fold ²²³. Finally, the telomere and subtelomere R-loops may further increase the chance of DSB formation in this area. We now know that TERRA forms R-loops with the telomeric DNA ⁶, and an increased level of telomere R-loops leads to an increased amount of DSBs in ESs and more frequent VSG switching ⁶. Additionally, R-loops are stabilized at the 70 bp repeats in TbRNase H1 null and TbRNase H2A-depleted cells, which is also linked with increased amounts of DNA damages in ESs and more frequent VSG switching ^{7, 8}. Transcribing the active VSG with the flanking repetitive sequences provides a couple of intrinsic mechanisms to induce DSBs in the local region, either with or without the formation of local R-loops, which ensures a good chance of VSG switching through homologous recombination.

R-loops at the telomere and 70 bp repeats clearly contribute to more efficient antigenic variation ^6–8. However, the underlying mechanisms of the formation and dissolution of these R-loops are not well-understood. Functions of T. brucei TERRA are not clear, and whether high levels of TERRA always lead to high levels of telomere R-loops is not known. Similarly, it is unclear whether the RNA transcribed from the 70 bp repeats can only exist as part of the R-loop in TbRNase H defective cells. Finally, TbRNase H1 null cells appear to have achieved a good balance between improved subtelomeric plasticity and sufficient genome stability ⁷. A better understanding about the detailed functions of TbRNase H1, particularly at the subtelomere, would be revealing for the positive and negative effects of telomere/subtelomere R-loops on antigenic variation and genome stability.

Telomere biology in other microbial pathogens and the potential link with pathogen virulence

Two kinetoplastid parasites that are closely related to T. brucei also cause debilitating human diseases: Trypanosoma cruzi causes Chagas disease, which can lead to serious heart and digestive problems ²²⁴. Leishmaniasis is caused by over 20 species of Leishmania. The disease severely decreases the life quality and causes heavy economic burdens ²²⁵. Telomeres in all three kinetoplastid parasites have the TTAGGG repetitive sequence ^{162, 226, 227}. Homologues of all known T. brucei telomere proteins can be easily identified in T. cruzi and Leishmania genomes ^{228, 229}. Additionally, TERRA has been detected in Leishmania ¹⁶⁵. However, it is unknown whether T. cruzi transcribes its telomeric sequence. It is also unclear whether R-loops are formed in T. cruzi and Leishmania. Still, the essential telomere function in maintaining genome stability is expected to be conserved in both T. cruzi and Leishmania. In addition, subtelomere stability may be an important factor for T. cruzi virulence. T. cruzi is able to infect any host cell but mainly macrophages, fibroblasts and epithelial cells ²³⁰. Infective forms of T. cruzi express surface transsialidase proteins ^{231, 232}, and their conserved peptide motifs are important for interacting with cytokeratin and mediate host-parasite interaction ²³³. Decreasing trans-sialidase expression also contributes to the loss of T. cruzi virulence ²³⁴. Interestingly, the gp85 gene family that encodes trans-sialidase is located at subtelomeric regions ²³⁵, and maintaining telomere and subtelomere stability is expected to help maintain stable gp85 gene copy numbers. Therefore, it will be interesting to investigate whether gp85 expression is affected by telomeric silencing, whether any R-loop is formed at telomere/subtelomere, and whether these R-loops affect gp85 gene family stability or their expression.

Several other microbial pathogens that undergo antigenic variation also host their variable surface antigens at subtelomeres ¹⁷⁹, including Pneumocystis jirovecii that causes pneumonia in immunodeficient patients, in which DNA recombination appears to be the major pathway of antigenic variation ¹⁸⁵. A better understanding of telomere functions and how telomere R-loops influence telomere and subtelomere stability will help us better understand the mechanisms of antigenic variations in these human pathogens.