Repetitive DNA loci and their modulation by the non-canonical nucleic acid structures R-loops and G-quadruplexes

rDNA organization

The rDNA repeats are essential housekeeping genes and are the most abundant genes in the eukaryotic genome. In the model organism Saccharomyces cerevisiae, an rDNA unit comprises 2 open reading frames, 35S and 5S, transcribed by RNA polymerase I (RNA Pol I) and III (RNA Pol III) respectively, and 2 intergenic spacers, IGS1 and IGS2, (Fig. 1a).^{11, 17} A replication fork barrier (RFB) and a bi-directional origin of replication called ribosomal autonomous replication sequence (rARS) are respectively located within the IGS1 and IGS2 regions.¹⁸ Transcription of the rDNA loci is an active and ongoing process due to the cell's exigence for ribosomes, the critical machinery required for protein synthesis. To cope with these demands, cells across species contain multiple copies of rDNA repeats ranging from 7 in Escherichia coli to 12,000 in Zea mays.¹¹ In eukaryotes, these multiple copies are arranged as clusters of tandem repeats and are spatially isolated from the rest of the DNA, making up the bulk of the major nuclear compartment called the nucleolus. Roughly 150–200 rDNA units are all tandemly arranged on chromosome XII of S. cerevisiae, while ∼350 human rDNA units are arranged on chromosomes 13, 14, 15, 21 and 22, with roughly 70 tandem repeats per chromosome (Fig. 1b).^{11, 17} Although such an organization of rDNA repeats is necessary to meet the cells demands, it can also compromise genome stability especially in the absence of the tight controls that typically limit the ability of HR to access or modulate these repeats.

Controlled HR at rDNA repeats

Although organisms contain many rDNA units, only about half are actively transcribed by RNA Pol I.¹⁹ In other words, rDNA repeats typically harbour both active and inactive units. Additionally, within a given rDNA unit, IGS regions can be subject to mechanisms that suppress RNA polymerase II (RNA Pol II)-dependent transcription and promote repeat stability.²⁰ Cells use multiple rDNA-regulatory mechanisms to maintain this repeat stability including: sister chromatid alignment, replication fork progression, IGS transcriptional silencing, and localization of the rDNA repeats. In S. cerevisiae, rDNA units associate with condensin, a 5-subunit protein complex that condenses rDNA and helps align sister-chromatids to warrant accurate HR-dependent repair of damaged rDNA sites and maintain an optimal number of rDNA units.^{19, 21} When rDNA copy numbers are at sub-optimal levels as in cells lacking condensin, cells are hyper-sensitive to DNA damaging agents.¹⁹ Additionally, the shared cellular longevity-sustaining factors are distributed between rDNA and the rest of the genome to ensure genome stability; this distribution is intimately influenced by the size of the rDNA repeat tract.²² Thus, maintenance of an optimal or wild-type number of rDNA units is critical to efficient ribosome biogenesis, cellular longevity, overall genome function, and survival under stress.

To maintain optimal rRNA levels when rDNA copy numbers decrease, cells activate typically inactive rDNA units increasing RNA Pol I localization to the rDNA loci.^19,23 Increased RNA Pol I localization consequently limits the association of condensin with rDNA thereby interfering with condensin/HR-dependent repair of potential DNA damage.^19,23 However, cells are equipped with a compensatory copy number reconfiguration mechanism that can gradually restore optimal rDNA copy numbers. At the heart of this copy number reconfiguration system is a protein called fork-blocking protein 1 (Fob1). Fob1 physically binds to the rDNA intergenic RFB regions, creating a unidirectional block that stalls advancing DNA replication forks (Fig. 1A).^24,25 Stalling of the replication fork leads to the formation of DSBs creating a recombination hotspot. Such DSBs can then be repaired by intra- or inter-chromosomal recombination events that alter rDNA copy number.²⁴ During intra-chromosomal recombination, DSBs are repaired by strand invasion between rDNA units located on the same chromosome, potentially resulting in the loss of repeats located between the break site and the donor site. As such, intra-chromosomal recombination at rDNA loci is responsible for the production of potentially toxic extrachromosomal rDNA circles (ERCs) that can result in subsequent contraction of the rDNA array via the loss of repeats that comprise ERCs.²⁴ Alternatively, repair via inter-chromosomal recombination can occur at DSBs either by equal sister chromatid exchange maintaining rDNA copy number or by unequal sister chromatid exchange (USCE) giving rise to an expanded or contracted rDNA copy number.²⁶ In the context of a short rDNA repeat tract and increased local transcriptional activity, USCE will favor rDNA copy number expansion.²⁶ A key player in this process is a non-coding RNA Pol II-dependent promoter that is located within the IGS1 regions of rDNA and known as the EXP Promoter (E-Pro).²⁶ During the S phase of the cell cycle, the cohesin protein complex associates with intergenic rDNA regions and helps align sister chromatids during replication to ensure equal sister-chromatid exchange (Fig. 1A). However, activation of E-Pro leads to the transcription of long intergenic non-coding RNA molecules effectively limiting local cohesion association and favoring USCE events.²⁶ Cells with an optimal or wild-type rDNA copy number have a selective advantage and optimal copy number will be restored over several cell divisions. E-Pro activity is in turn regulated by the NAD⁺-dependent histone deacetylase Sir2.²⁷ When rDNA copy number is at optimal levels, Sir2 represses E-Pro activity altering the chromatin structure and allowing cohesin to associate at the rDNA array. In contrast, when rDNA copy numbers decrease, Sir2 inhibition of the E-Pro is compromised resulting in a loss of cohesin association at the rDNA loci culminating in USCE with consequential amplification until wild-type optimal copy numbers are reached.²⁷ Importantly, Sir2 is part of a larger molecular network that helps silence intergenic rDNA transcription and maintain rDNA repeat stability.

Although recombination at rDNA is a mechanism used to maintain optimal rDNA units, hyper-recombination events can lead to genomic instability and must therefore be inhibited. To prevent hyper-recombination within the rDNA repeats, yeast cells assemble IGS DNA into silent chromatin structures via a process known as rDNA silencing, which implicates both Sir2-dependent and Sir2-indepednent mechanisms.^20,28 The RENT complex (regulator of nucleolar silencing and telophase exit) comprises the Sir2, Net1 and Cdc14 proteins.²⁹ The RENT complex induces rDNA silencing via the association of both Sir2 and Net1 with coding (Pol I promoter) and non-coding (IGS1/RFB) regions of rDNA (Fig. 1a).²⁸ DNA-bound Fob1 physically associates with the RENT complex trough Net1 and Sir2, favoring their association at the RFB. RENT also localizes around the RNA Pol I promoter via Fob1-independent physical interaction of RNA Pol I with Sir2 and Net1 (Fig. 1a).²⁸ RENT promotes functional interactions between Sir2 and the N-terminal tails of histones H3 and H4. Sir2-dependent deacetylation of lysine residues within the N-terminal tails of H3 and H4 helps establish a more closed or silent chromatin structure that limits access to RNA Pol II within the intergnic rDNA regions.³⁰ In addition to Sir2, the silencing of rDNA is further promoted by other chromosomal factors including the cohibin complex, comprised of Lrs4 and Csm1, and topoisomerase-related factor 2 (Tof2).^{20, 31} Similar to the RENT complex, cohibin and Tof2 are recruited to the IGS1 region in a Fob1-dependent fashion and are required for rDNA silencing (Fig. 1a).³¹ Silencing of the rDNA loci through these mechanisms safeguards against hyper-recombination, thus preserving rDNA integrity.

In addition to rDNA silencing, yeast cells prevent excessive aberrant recombination and maintain the stability of rDNA repeats by regulating their localization within the nucleus. It is well established that rDNA is physically sequestered within the major subnuclear compartment known as the nucleolus, which in yeast is crescent shaped and abuts the nuclear envelope. The subnuclear localization of rDNA plays an important role in rDNA stability, as Sir2-dependent silencing alone is not sufficient to inhibit USCE.²⁰ Within the nucleolus, rDNA is tethered to the inner nuclear membrane (INM) via interaction of the rDNA-associated cohibin complex and the nuclear envelope-embedded chromosome linkage INM proteins (CLIP), namely Heh1 (helix extension helix 1) and Nur1 (nuclear rim 1) (Fig. 1A).²⁰

CLIP and cohibin suppress overall recombination at rDNA while also specifically limiting aberrant recombination events. First, the solid support provided by the nuclear envelope-embedded CLIP allows cohibin to align sister chromatids during rDNA replication, thereby promoting equal and suppressing unequal recombination events.²⁰ Additionally, perinuclear tethering of rDNA by CLIP sequesters the repeats within the recombination-suppressive environment of the nucleolus, shielding the repeats from functional Rad52 recombination proteins enriched in the nucleoplasm.^{20, 32}

In addition, to further prevent uncontrolled recombination between the highly repetitive rDNA repeats, cells temporarily localize rDNA units experiencing DSBs outside of the nucleolus, possibly with at least one other intact donor rDNA unit.³² To warrant a compartment-specific delocalization of HR processes, the Rad52 recombination protein is excluded from the nucleolus by the Smc5-Smc6 and the MRX complexes, and by sumoylation of the Rad52 protein itself.³² Upon induction of a DSB within a given rDNA unit, the MRX complex, comprised of the Mre11, Rad50, and the Xrs2 proteins, recognizes and localizes to the DSB site within the nucleolus. The MRX complex recruits the Smc5-Smc6 complex, whose Mms2 subunit possesses sumoylation activity.³³ Interestingly, although the Smc5-Smc6 complex is required for the exclusion of Rad52 from the nucleolus, it is not required for its sumoylation. Therefore, the role of Smc5-Smc6 in the exclusion of functional Rad52 from the nucleolus could be indirect, possibly involving the sumoylation of other recombination proteins.³² After resection of the MRX-bound DSB, it is relocated from the nucleolus to the nucleoplasm. MRX dissociation is then followed by the recruitment of unsumoylated, and thus functional, Rad52 to the DSB site via RPA proteins, allowing access to the HR machinery downstream of Rad52 and mediating recombinational repair.³²

The various pathways that regulate recombination within the rDNA repeats highlight the importance of regulating repetitive DNA loci. Although HR is critical to the maintenance of genome stability, it can also be a potential source of GCRs and overall genome instability. The action of HR and other specialized mechanisms at telomeres, another type of conserved and repetitive DNA loci located at the ends of linear chromosomes, is also critical to overall genome function and stability.

Introduction to telomeres

Across species, the ends of linear chromosomes are constituted of telomeres, which are repetitive DNA sequences that help counter natural chromosome shortening, prevent chromosome end recognition by the DNA repair machinery, and avoid destabilizing chromosome end-to-end fusions (Fig. 2).^{34, 35} Telomeres are comprised of non-coding double-stranded and single-stranded DNA (ssDNA) sequences consisting of tandem G-rich repeats such as the TG_1–3 of budding yeast and human T₂AG₃.³⁴ The ssDNA regions, known as G-tails, serve as substrates for several DNA-binding proteins whose roles are to preserve telomere length and structure.³⁶ Telomere length varies across species with 300–400 repeats in yeast to tens of thousands in higher eukaryotes.³⁶ Maintaining telomeres at optimal lengths is pivotal to genome stability and health. Disruption of this equilibrium can lead to catastrophic genome instability and telomere-related diseases.

Telomere-linked diseases include cancer and premature aging syndromes, collectively called telomeropathies, which have overlapping symptoms and incompletely penetrant phenotypes, with a highly variable age of onset.^37-39 The DNA replication machinery cannot copy the very ends of linear chromosomes thus the number of protective telomeric repeats should naturally decrease with each round of DNA replication and cell division.^{40, 41} However, an optimal length is maintained through an interplay of shortening and elongation mechanisms. It is imperative that such an interplay be highly regulated to maintain homeostatic telomeric length and subsequently avert the occurrence of telomeropathies.

Telomerase-dependent telomere length maintenance

Telomere length is maintained through the activity of a ribonucleoprotein complex called telomerase, which is recruited to short telomeres to add telomeric DNA repeats to chromosome ends.⁴² The components of telomerase vary across species with the exception of telomerase reverse transcriptase (TERT) and telomerase RNA (TERC). TERC, an RNA subunit, not only serves as a template for reverse transcription but also provides scaffolding to allow telomerase protein components to form the telomerase RNP holoenzyme.⁴³ Although TERC length and sequence varies across species, it contains a conserved catalytic core that is required for substrate binding and telomerase activity.⁴³ If telomerase were left to its own devices, the outcome would be an undesirable increase in telomere length rendering cells immortal which could potentially give rise to cancerous cells.⁴⁴ As such, in addition to regulating telomerase levels, cells also use various factors that highly regulate telomerase recruitment and access to telomeres.

Telomerase recruitment and the frequency/extent of telomere elongation is tightly regulated in cis by the activities of both single- and double-stranded DNA binding proteins.⁴⁵ In yeast, in vivo/in vitro studies reveal that the telomerase accessory factor Est1 directly interacts with the recruitment domain of the ssDNA telomeric-binding protein Cdc13, thereby mediating telomerase access to telomeres.^{46, 47} Additionally, telomerase is recruited to telomeres through a Ku-mediated mechanism; Ku binds both the telomerase RNA (TLC1) and the telomeric silencing protein Sir4, which in turn interacts with the telomeric dsDNA-binding protein Rap1.⁴⁸ Moreover, this interaction subsequently enhances the Cdc13-Est1 interaction at the ssDNA G tail. Recruitment of telomerase by Sir4-Ku interaction is inhibited by Rif1 and Rif2, which compete against Sir4 for Rap1 binding (Fig. 2A(ii)).^48-50 In this setting, Cdc13-Est1-mediated telomerase recruitment is still possible but at a decreased efficiency.⁴⁸ In mammalian cells, telomerase is recruited via a member of the shelterin complex, TPP1.⁵¹ Specifically, TPP1s oligosaccharide binding (OB) fold known as the TEL patch interacts with the TEN-domain of the hTERT component of telomerase (Fig. 2B(ii)).^{52, 53} Thus, various telomere-binding factors help recruit telomerase to chromosome ends.

Multiple telomere-binding proteins can also inhibit telomeric elongation by obstructing telomerase accessibility. Depletion of the Rap1/Rif1/Rif2 complex in budding yeast, or components of the shelterin complex, TRF1, TRF2 and POT1, in mammalian cells, increases telomere length, indicating that these proteins act as negative regulators of telomere length.^50,54-56 In yeast, Rap1 binds telomeres and recruits Rif1/Rif2, which are responsible for inhibiting telomeric elongation (Fig. 2A(ii)).⁵⁰ The N-terminus of Rif2 is able to suppress telomere lengthening in rif1Δ, rif2Δ and rif1Δ rif2Δ mutants in a telomerase-dependent fashion. This suggests that the N-terminal domain of Rif2, termed the BAT domain (Blocks Addition of Telomeres), negatively regulates telomere length by inhibiting their elongation by telomerase.⁵⁰ In mammalian cells, early studies revealed that TRF1 and TRF2, members of the shelterin complex, bind dsDNA regions and inhibit telomere elongation by inducing the formation of a telomeric loop, t-loop, which can physically block telomerase access.^{54, 55} Additionally POT1 binding at the ssDNA region, can physically block telomerase access as well as stabilize the t-loop (Fig. 2B(ii)).⁵⁶

Telomerase-independent elongation of telomeres

In addition to telomerase-mediated elongation, telomeres can also be lengthened by other mechanisms. In mammalian cells, this can occur by a recombination-driven amplification mechanism known as alternative lengthening of telomeres (ALT). This recombination-dependent but telomerase-independent mechanism is a hallmark of many mesenchymal-cancerous cells and occurs in approximately 10–15% of human cancers.^{57, 58} ALT gives rise to abnormally longer telomeres with various lengths.⁵⁹ ALT is able to bypass mechanisms that regulate telomerase potentially rendering cells immortal. Two different mechanisms that rely on the recombination machinery have been proposed for ALT: unequal telomere sister chromatid exchange (T-SCE) and lengthening via break induced replication.^{60, 61}

In telomerase-null yeast cells, a small population of cells is able to maintain telomere elongation via 2 distinct Rad52-dependent pathways, type I and type II, which are collectively known as RTE (RAD52-dependent recombinational telomere elongation). Type II RTE closely resembles mammalian ALT.^62,63 Thus, HR-based mechanisms allowing cells to bypass telomerase-dependent telomere length control are evolutionarily conserved.

To prevent the fusion of telomeres on different chromosomes or chromosome arms to each other, various telomere capping factors ensure the local inhibition of non-homologous end-joining (NHEJ). Telomere-binding proteins cap telomeres thereby shielding them from DNA damage response pathways such as NHEJ. This is crucial as a single NHEJ event at telomeres can give rise to GCRs resulting in genome instability. In budding yeast, Rap1 is required for NHEJ inhibition and several studies have shown that the loss of Rap1 results in a significant increase in NHEJ events, leading to telomere fusions.⁶⁴ As a single NHEJ event at telomeres can be catastrophic to genome stability, cells have used different Rap1-mediated pathways to inhibit such an occurrence. The Rap1 C-terminal domain inhibits NHEJ through distinct Sir4-dependent and Rif2-dependent pathways.⁶⁴ Additionally, the Rap1 central domain inhibits NHEJ in a Sir4 and Rif2 independent manner.⁶⁴ In mammalian cells, Rap1 and TRF2 protect telomeres from being processed by ALT via suppression of telomeric localization of the HR factors: PARP1 and SLX4.⁶⁵ In the absence of Rap1 and TRF2, the accumulation of HR factors at telomeres leads to telomere resection, loss, and ultimately chromosomal fusions.⁶⁵

Taken together, evolutionarily conserved interplays between telomerase, telomeres, their various binding/capping factors, and the basic DNA repair/recombination machineries are critical to genome function and stability.

Introduction to transposable elements

Transposable elements (TEs), arguably the greatest source of repetitive DNA, are widely distributed across the eukaryotic kingdom. They are commonly known as “jumping genes,” due to their ability to move around within the genome.⁶⁶ TEs are comprised of 2 main classes reflecting the mechanism by which they move around in the genome: class I TEs, retrotransposons, move in a copy and paste manner by way of an RNA intermediate, and class II TEs, DNA transposons, move in a cut and paste manner, independent of an RNA intermediate.⁶⁷ Retrotransposons are comprised of 2 sub-classes: long-terminal repeat (LTR) and non-LTR retrotransposons, based on the presence or absence of flanking long-terminal repeat sequences. Retrotransposons replicate with each round of transposition and comprise about 3% and 45% of the yeast and human genomes respectively. Transposon dysregulation has been linked to natural aging, neurologic disorders and various cancers.^13,68-71

In budding yeast, retrotransposons are termed transposons of yeast (Ty). They are flanked with LTRs and are comprised of 5 families, Ty1–5, interspersed across the genome, with Ty1 being the most abundant and active full length Ty element.⁶⁸ In humans, the non-LTR retrotransposons, Long Interspersed Element (LINE-1 or L1), Alu (the most abundant Short Interspersed Element – SINE) and SVA (SINE-VNTR-Alu) are the only active TEs.⁷² Ty and LINE-1 are autonomous elements containing 2 open reading frames, TYA/ORF1 and TYB/ORF2 which encode proteins that are required for retrotransposition (Fig. 3A).^{72, 73} The retrotransposition cycles of both yeast Ty and human LINE-1 elements include reverse transcription of the RNA and integration/insertion of the resulting cDNA product into new genome loci (Fig. 3B-C). As retrotransposons are capable of replicating and inserting themselves into new genomic loci, they can restructure the genome, alter gene expression and trigger deleterious chromosomal rearrangements. Therefore, retrotransposons are considered a major source of potential genome instability. Importantly, to maintain genome integrity, cells have adopted multiple mechanisms to regulate retrotransposition at each stage of the retrotransposon life cycle.

Stage-specific regulation of retrotransposition

The first stage at which cells regulate retrotransposition is at the level of transcription. In yeast, studies using a TYA1:URA3 reporter showed that Ty1 transcription is subject to a negative feedback control known as Ty1 transcriptional co-suppression.⁷⁴ Specifically, in Ty1-less mutants harbouring either a single-copy or multi-copy plasmid of Ty1 elements, only the latter was able repress a Ty1 containing reporter plasmid. This suggests that in the presence of a high or low copy number of Ty1 elements, these jumping genes are either silenced or expressed respectively.⁷⁴ A subsequent report proposed that Ty1 transcriptional co-suppression is independent of Ty1 homology, a hallmark of co-suppression.⁷⁵ This opened the idea that the ability to translate Ty1 mRNA could be driving Ty1 transcriptional silencing in this co-suppression mechanism. Wu and Jiang, 2008, found that Ty1 transcriptional silencing is initiated by non-translatable Ty1 mRNA, mRNA molecules that are poor templates for translation. Over-production of non-translatable Ty1 mRNA results in the inactivation of the nuclear cap-binding complex (CBC) leading to the hyperstimulation of the TOR pathway, which in turn further silences Ty1 transcription through the Tap42-phosphatase complex.^75,76 We note that Ty1 co-suppression was not observed when a TYA1:GFP reporter was used, raising the possibility that it may be affected by some reporters or that it may be an artificial effect of other reporters.⁷⁷ Additionally, Ty1 transcription is also generally inhibited by the chromatin remodelling ATPases Isw1p and Isw2p (Fig. 3B).⁷⁸ In mammalian cells, LINE-1 is silenced through methylation and hypomethylation increases LINE-1 levels and is a key hallmark of many cancers.⁷⁹ In addition, TE silencing is especially critical in germ cells as any TE-related genome instability can be transmitted to the next generation or even potentially decrease fertility. Key to germ line TE silencing are single-stranded noncoding RNAs known as PIWI-interacting RNAs (piRNAs).⁸⁰ piRNA-dependent silencing of LINE-1 requires the PIWI Argonoute protein, MIWI2, which orchestrates DNMT3-dependent DNA methylation at TEs. In a model by Pezic et al 2014, piRNA-dependent methylation of LINE1 is targeted to the promoter-containing 5′ region of LINE-1. In this model, MIWI2/piRNA complexes recognize and bind nascent LINE-1 transcripts allowing MIWI2 to then recruit DNMT3 ensuring the trimethylation of H3K9 and silencing of active LINE-1 elements (Fig. 3C). Thus, cells can repress TE activity via transcriptional processes.

Cells also have an arsenal of post-transcriptional mechanisms allowing for the regulation and repression of retrotransposition. In yeast, Ty1 retrotransposition events are rare despite the relatively high levels of Ty1 RNA, which can be as much as 0.8% of total RNA, indicating that inhibition of Ty1 retromobility exists at one or more post-transcriptional levels.^73,81 Indeed, posttranscriptional control of Ty1 mobility does occur via modulation of each of the stages of the retrotransposition life-cycle including translation (protein synthesis), reverse transcription (cDNA synthesis), and integration into new genomic loci.

In human somatic cells, which lack the piRNA activity found in germ cells, LINE-1 activity is inhibited post-transcriptionally by the micro-RNA miR-128.⁸² miR-128 directly binds a site within the ORF2 region of LINE-1 transcripts resulting in the formation of a LINE-1-miR-128-RISC complex, thereby targeting LINE-1 transcripts for degradation and consequently inhibiting retrotransposition (Fig. 3C).⁸² Although S. cerevisiae lacks RNAi activity, it is able to make use of antisense and non-coding RNAs to inhibit retrotransposition. Ty1 retromobility is post-transcriptionally inhibited by an RNA-dependent gene silencing mechanism involving a non-coding antisense Ty1 cryptic unstable transcript (CUT), termed Ty1-RTL.⁸³ RNA Pol II-dependent and Set1 histone methyltransferase-regulated Ty1-RTL transcription is initiated within Ty1 and inhibits Ty1 expression in trans by acting on newly synthesized Ty1 transcripts via an RNA-dependent silencing mechanism. This in turn leads to a decrease in RNA Pol II occupancy at Ty1 further promoting silencing (Fig. 3B).⁸³ These studies highlight mechanisms by which both human and yeast cells inhibit retrotransposition post-transcriptionally.

Cells have also adopted RNA-mediated mechanisms to modulate retrotransposition at the post-translational level. In S. cerevisiae, this may partially be achieved via the action of antisense Ty1 transcripts. Ty1 mRNA levels generally increase as chromosomal Ty1 copy numbers increase, although this is not always accompanied by increases in retrotransposition.⁸⁴ This is because cells with more chromosomal Ty1 copies have also been shown to have greater levels of antisense Ty1 transcripts, Ty1-AS-RNA, that can inhibit retrotransposition.⁸⁴ There are various size-based classes of Ty1-AS-RNAs. Transcription of Ty1-AS-RNAs is independent of full-length Ty1 transcription, as these antisense transcripts are still present in mutants where Ty1 transcription is inhibited. Ty1-AS-RNA sequences are complementary to the copy number control (CNC) region of Ty1 and are required for post-transcriptional copy number control.⁸⁴ Ty1-AS-RNAs are trans-acting and associate within cytoplasmic virus-like particles (VLPs), thereby suppressing the levels of Ty1 mRNA-encoded proteins (Fig. 3B). Several mechanisms are envisaged through which Ty1-AS-RNAs may be decreasing Ty1 protein levels: 1) base pairing of Ty1-AS-RNA with Ty1 mRNA may prevent Ty1 mRNA dimerization within VLPs; 2) Ty1-AS-RNAs may inhibit processing of the Pol proteins; and 3) Ty1-AS-RNA may block tRNA^met (which acts as a primer for RT/cDNA synthesis) from binding to Ty1 mRNA.⁸⁴ Ty1-AS-RNAs represent one class of Ty1 RNA species that are able to regulate retrotransposition.

In addition to Ty1-AS-RNAs, S. cerevisiae cells can inhibit Ty1 retrotransposition through an internal sense transcript that codes for a trans-dominant protein; the internal Ty1 sense-strand located at the 5′ region, termed Ty1i-RNA, is transcribed from an internal promoter within TYA, about 800 nucleotides downstream of the full-length start site.⁸⁵ It encodes an altered (N-terminally truncated) form of Gag capsid protein (Gag), known as p22 (unprocessed) or p18 (p22 processed by Ty1 Protease), that mediates Ty1 CNC.⁸⁵ p22/p18 complexes interact with Ty1 VLPs changing their spherical morphology and rendering them open or incomplete thereby inhibiting VLP maturation, cDNA synthesis and retrotransposition (Fig. 3B). Inhibition by p22/p18 is partly due to the direct interaction between Gag and p22, as mutations in GAG open reading frames (ORFs) confers resistance to p22, allowing for mature VLP assembly.⁸⁶ Specifically, the N-terminal of p18 interacts with the C-terminal region (CTR) of Gag and inhibits its nucleic acid chaperone (NAC) activity, which is required for Ty1 retrotransposition. Additionally, p18 simultaneously binds Ty1 transcripts with high affinity thus competing with Gag for Ty1 RNA binding. p22/p18 also prevent Ty1 mRNA dimerization and packaging and renders Ty1 mRNA more prone to nuclease degradation, thereby inhibiting retrotransposition.⁸⁷

Even under normal conditions, retrotransposition occurs in “healthy” wild-type cells. To ensure that base-line retrotransposition does not give rise to genome instability, cells target retrotransposon cDNAs to gene-poor regions, resulting in non-deleterious insertions. Ty1 cDNA preferentially integrates upstream of Pol III transcribed genes, mainly the tRNA genes.⁸⁸ Prior to integration, Ty1 cDNA exists as a preintegration complex (PIC) that comprises a cDNA element and an integrase; Ty1 PIC is targeted to regions upstream of RNA Pol III transcribed genes via interaction between the C-terminus of integrase and the AC40 subunit of RNA Pol III (Fig. 3B).⁸⁹ Although RNA Pol I also contains the AC40 subunit, Ty1 does not integrate upstream of RNA Pol I-transcribed genes, suggesting that some other RNA Pol III cofactor is required for the interaction between integrase and the AC40 subunit of RNA Pol III. When this interaction is lost, Ty1 cDNA targets telomeric and subtelomeric heterochromatic regions.⁸⁹ Although the mechanism behind this secondary targeting has not been elucidated, it could plausibly be mediated by interaction between integrase and telomeric proteins. The insertion of retrotransposition into these hotspots ensures non-deleterious integration of Ty1 cDNA.

Although retrotransposon insertions are highly regulated, an increase in the number of repeated interspersed sequences in the genome, a result of retrotransposition, may lead to an increase in non-allelic HR (NAHR) frequency. In other words, an augmentation in the number of retrotransposons, interspersed throughout the genome, might lead to undesirable NAHR events, or ectopic recombination, as a consequence of DSB repair or of the restarting of stalled replication forks. As such, aberrant recombination events at retrotransposons could be detrimental to cells, potentially giving rise to GCRs and genome instability. In budding yeast, DSBs at Ty elements have been shown to be repaired by HR between Ty elements dispersed in the genome.⁹⁰ Moreover, recent studies in human cells, demonstrated that NAHR events between LINE-1 elements occur frequently and that such events can cause structural genomic rearrangements, which can contribute to genome instability and give rise to human genetic disorders.^{91, 92}

Our review thus far illustrates the multitude of mechanisms and factors used by cells to regulate and maintain the integrity of repetitive DNA loci such as rDNA repeats, telomeric repeats, and retrotransposons. In the remainder of this review, we discuss how the regulation of these repetitive DNA loci can intersect in either beneficial or deleterious fashions with different and abundant non-canonical nucleic acid structures.

R-loops

R-loops are 3-stranded nucleic acid structures comprised of an RNA-DNA hybrid and a displaced ssDNA.⁹³ They are often by-products of transcription, arising when nascent transcribed transcripts anneal to the DNA template strand (Fig. 4B). R-loops are potentially DNA damaging structures that have been shown to pause progression of the replication fork, leading to DSBs that may be repaired by HR. Moreover, R-loops are closely connected with GQs, which can be often formed on the displaced ssDNA of R-loops.⁹³ Therefore, the prevention and resolution of R-loops at various loci has to be carefully orchestrated and an arsenal of proteins, including RNase H and Topoisomerase enzymes, play a central role in this process.^94-96

R-loops naturally occur in vivo, and have many important biologic functions such as immunoglobulin class switching and the regulation of gene expression.^{97, 98} Despite their biologic importance, R-loop accumulation poses a threat to genome integrity and is implicated in several cancers and syndromes. For example, mutations in amyotrophic lateral sclerosis (ALS)-linked factors can give rise to R-loop accumulation resulting in aberrant transcript accumulations.⁹⁹

Studies have shown that the long non-coding telomeric repeat-containing RNA (TERRA), a transcript conserved across species, is implicated in R-loop formation/accumulation at telomeres in both yeast and human cells.^94,100-102 Loss of the RNA-DNA hybrid-suppressing RNase H enzymes (RNaseH1 and RNaseH2), which degrade the RNA component of RNA-DNA hybrids, leads to R-loops accumulations at telomeres delaying cellular senescence.^94,100 When the RNase H enzymes are deleted in telomerase deficient yeast cells, recombination at telomeres is increased giving rise to longer telomeres and consequently a delay in cellular senescence.¹⁰⁰ This suggests that R-loop accumulation at telomeres leads to an increase in telomere length in an HR-mediated elongation pathway. In contrast, in HR-deficient yeast cells, R-loop accumulation at telomeres leads to premature cellular senescence.¹⁰⁰ In human ALT cells, RNase H1 is highly enriched at telomeres and knock-down of this enzyme gives rise to an increase in TERRA-associated hybrids, which in turn increase the recombinogenic nature of ALT cells.⁹⁴ These yeast and human studies suggest that telomeric R-loops can both inhibit and accelerate senescence, potentially giving rise to cancerous cells or premature aging, respectively (Fig. 4A).

Therefore, cells tightly regulate R-loop levels at these loci and use various factors, in addition to RNase H, to inhibit their undesirable accumulation at telomeres; such factors include members of the THO complex: Hpr1, Thp2, and Tho2 in yeast cells (Fig. 4A).^{101, 102} The THO complex is required for the assembly and export of mRNA.¹⁰³ In telomerase deficient cells, THO mutants give rise to both premature senescence and telomere elongation through Rad53-dependent telomeric recombination.¹⁰² In these mutants, TERRA-associated R-loop accumulation triggers the Rad53-dependent DNA damage response (DDR).¹⁰² The induction of DDR in this setting is most likely due to the accumulation of DSBs caused by stalled replication forks at TERRA-harbouring R-loops.

As R-loops are transcriptional by-products capable of inducing DSBs, due to their ability to stall replication forks, it is plausible to assume that their accumulation at the highly transcribed rDNA loci may have detrimental effects. Indeed, R-loops have been shown to accumulate at rDNA IGS regions, particularly at E-pro, in the absence of the yeast Ataxin-2 protein Pbp1 or Topoisomerase 1 (Top1).^95,96 This accumulation leads to aberrant recombination events at rDNA, resulting in genomic instability and culminating in a decrease in replicative life span (Fig. 4A). This mechanism seems to be evolutionarily conserved as human ATXN2, the ortholog of yeast Pbp1, was recently shown to prevent R-loop accumulations within nucleoplasmic and nucleolar regions of the nucleus in human cell culture.¹⁰⁴ It remains an intriguing but untested possibility that physiologic R-loops at coding or non-coding rDNA regions may regulate various aspects of rDNA repeats in wild-type cells.

Recent studies have shown that RNA-DNA hybrids accumulate at Ty1 elements in the absence of the yeast RNase H enzymes (Rnh1 and Rnh201).^{105, 106} Whether hybrids accumulate at genomic Ty1 elements or at Ty1 cDNA is unclear: while one study reports that hybrids accumulate at chromosomal Ty1, another reports that hybrids primarily accumulate at Ty1 cDNA resulting in an increase in the frequency of retrotransposition (Fig. 4A).^105,106 Yet another study proposed a model in which RNA-DNA hybrids exist in newly integrated Ty1 cDNA.¹⁰⁷ Based on our current knowledge of R-loops, and the effect that their accumulation has at telomeres and rDNA repeats, one can postulate that the buildup of these structures at chromosomal retrotransposons might lead to DSBs repairable by non-allelic HR using other retrotransposon elements as template. Considering that retrotransposons are interspersed throughout the genome, HR repair at these loci can potentially result in GCRs. Indeed, Stamenova et al., proposed a model in which the Rrm3 helicase, a member of the Pif1 family, prevents the formation of DSBs within RNA-DNA hybrids present in newly integrated cDNA. In the absence of Rrm3, hybrid accumulations lead to an increase in GCRs, retrotransposition and the formation of tandem arrays of Ty1 cDNA.¹⁰⁷ In these mutants, there was an increased recombination of Ty1 cDNA known as Ty1 multimers; high levels of Ty1 multimers increases the numbers of Ty1 cDNA incorporated into the genome per round of retrotransposition; this in turn gives rise to an increase in GCRs.¹⁰⁷ As hybrids have been shown to affect telomere length, rDNA stability and Ty1 retromobility, it is imperative that cells keep these potentially damaging structures in check.

G-quadruplexes

DNA sequences harbouring (GN_xGN_xGN_x)₄ or G4 motifs have the capacity to form the tetrameric structure known as a G-quadruplex or GQ (Fig. 4A). GQs comprise 4 guanine nucleotides forming non-canonical Hoogsteen bonds with 2 bordering guanines resulting in a conformation of stacked square planar structures, or G-tetrads.¹⁰⁸ The enclosed nucleotides form loops that vary in size, with smaller loops giving rise to relatively more stable structures. GQs can be classified based on 2 main features: 1) Intramolecular or intermolecular formation respectively via one or multiple strands; 2) parallel or antiparallel orientation of constituent DNA strands. The presence of a monovalent cation at the center of the G-tetrads stabilizes the GQ structure (for a full review on GQ structures please see¹⁰⁸). G4 rich motifs that have the potential to form GQs are scattered throughout the human genome and are in abundance particularly within repetitive DNA loci.

G4 motifs with GQ-forming potential are enriched at telomeres due to the inherent local concentration of G-rich double-stranded DNA and single-stranded 3′ overhangs. There is mounting evidence for functional roles of GQs at telomeres. For example, telomeric GQs can promote telomere stability by allowing POT1/TPP1, components of the shelterin complex, to outcompete RPA binding to telomeres.¹⁰⁹ ssDNA of telomeric 3′ overhangs interacts with POT1/TPP1 and RPA.¹⁰⁹ RPA initiates the ataxia telangiectasia and Rad3 (ATR) DDR, a key player in the cell cycle checkpoint that induces cell cycle arrest in response to DNA damage.¹¹⁰ If the ATR response is prematurely/incorrectly initiated at telomeres, it can prematurely trigger cell cycle checkpoint and arrest. To avoid such an incorrect activation of this cell cycle checkpoint response, telomeric GQs enhance the binding of POT1 to ssDNA overhangs.¹⁰⁹ Subsequent recruitment of TPP1 blocks RPA binding, protecting telomeres from premature ATR activation. In yeast cells, GQs have also been shown to act as telomere caps when standard capping is compromised.¹¹¹ For instance, in telomere capping mutants, overexpression of the GQ stabilizing protein Stm1 stabilizes telomeric GQs allowing them to act as rudimentary telomere caps.¹¹¹ Studies have also shown that GQs at telomeric 3′ overhangs can block and prevent telomere extension by Oxitricha telomerase in vitro (Fig. 4A).¹¹² Intriguingly, human telomerase is capable of binding and partially unwinding parallel, but not anti-parallel, intermolecular GQs allowing for telomeric extension.¹¹³ These findings suggest that in addition to the mere presence of GQs, the specific conformations adopted by these structures are also important to the ultimate effect that they have on telomere length and stability. Thus, these studies show that GQs can indeed impact telomere stability and function.

G4 motifs are also abundant within rDNA repeats, which are the most actively transcribed loci in the genome and are paramount to overall cellular function. Thus, it is reasonable to postulate that formation of GQs must be highly regulated at these sites. The accumulation of GQ within rDNA repeats can impede progression of replication forks directly via GQ structures and indirectly via stabilization of GQ-containing R-loops (Fig. 4B).^96,114 Indeed overexpression of the yeast GQ-stabilizing protein Stm1 leads to R-loop accumulation within rDNA repeats.¹⁰⁴ In both yeast and humans, a conserved 5′-3′ DNA helicase termed Pif1, maintains the replication fork barrier at rDNA, promoting rDNA integrity.¹¹⁵ A common function of Pif1 is to bind and unwind GQs; in the absence of Pif1, GQ accumulations stall replication forks triggering DSBs that are repaired via aberrant recombination.¹¹⁴ GQs can therefore directly and indirectly impact rDNA repeat stability.

Potential GQ forming sequences have also been identified within the 3′ ends of LINE-1 and plant LTRs^116,117 Although little is known about the role and/or effect that GQs have on retrotransposition, it is an intriguing hypothesis that the formation of GQs at these loci can regulate retrotransposition itself. GQs at the 3′ ends of cDNA may act as a barrier to reverse transcriptase (RT) thus blocking synthesis of second strand cDNA. Additionally, chromosomal formation of GQs on the sense strand can impede access to transcription machinery, transcriptionally silencing retrotransposition (Fig. 4B). Future work in the field should experimentally test these possibilities.