Small RNA-directed DNA elimination: the molecular mechanism and its potential for genome editing.

Transposable elements have both detrimental and beneficial effects on their host genome. Tetrahymena is a unicellular eukaryote that deals with transposable elements in a unique way. It has a separate somatic and germline genome in two nuclei in a single cell. During sexual reproduction, a small RNA directed system compares the germline and somatic genome to identify transposable elements and related sequences. These are subsequently marked by heterochromatin and excised. In this Review, current knowledge of this system and the gaps therein are discussed. Additionally, the possibility to exploit the Tetrahymena machinery for genome editing and its advantages over the widely used CRISPR-Cas9 system will be explored. While the bacterial derived CRISPR-Cas9 has difficulty to access eukaryotic chromatin, Tetrahymena proteins are adept at acting in a chromatin context. Furthermore, Tetrahymena based gene therapy in humans might be a safer alternative to Cas9 because the latter can trigger an immune response.

Introduction

Transposable elements (TEs) pose a threat to their host as they are able to move from one genomic location to another. However, TEs are also a source of genetic diversity that can drive evolution of the host genome. Therefore, hosts have developed defence mechanisms to detect TEs in their genome and render them harmless while at the same time maintaining them as genetic reservoirs. Mammals for example tolerate TEs in their genome but silence them at the transcriptional level. The unicellular Tetrahymena on the other hand does not merely suppress the activity of TEs, but it employs the more rigorous measure of removing them from its somatic genome.

Tetrahymena is able to clear its somatic genome from TEs because it has two nuclei in a single cell: a micronucleus (MIC) containing the germline genome and a macronucleus (MAC) containing the somatic genome. During sexual reproduction, Tetrahymena compares the MIC and MAC genome to identify TEs and subsequently eliminate them from the new MAC. In total, approximately 35% of the somatic genome is reproducibly removed and the remaining sequences are rejoined [1]. These radical DNA rearrangements are tightly coordinated by an intricate RNA framework [2,3].

This Review covers the later stages of programmed DNA elimination in Tetrahymena, starting from the search for a target by the Tetrahymena Piwi protein Twi1p (see Table 1 for an overview of the involved proteins). For a detailed overview of all stages, the reader is referred to a previous Review by Mochizuki and Noto [4]. The focus of this Review is on the open questions that remain regarding the Tetrahymena DNA elimination mechanism and how components of this system can potentially be repurposed as a gene-editing technique. Although CRISPR-Cas9 has revolutionized gene-editing and has become indispensable for life science, it has several drawbacks, especially when applying it in humans. We discuss the potential of a gene-editing system based on Tetrahymena DNA elimination proteins or a combination of the two systems to overcome these drawbacks.

Table 1.

Proteins involved in DNA elimination in Tetrahymena

Protein	Description	Reference
Coi6p	Boundary-protecting factor	Suhren et al. [21]
Coi7p	Boundary-protecting factor	Suhren et al. [21]
Dcl1p	Dicer-like 1 protein, processes scnRNAs	Malone et al. [7]
Ema1p	RNA helicase, unwinds nascent RNAs transcribed from IESs	Aronica et al. [19]
Ezl1p	Histone methyltransferase, catalyses H3K27 methylation at IESs	Liu et al. [16]
Giw1p	Regulates nuclear localization of Twi1p	Noto et al. [18]
Jmj1p	Boundary-protecting factor	Suhren et al. [21]
Lia5p	Boundary-protecting factor	Suhren et al. [21]
Pdd1p	Chromodomain protein, binds H3K27me3 and H3K9me3	Liu et al. [16]
TKu80p	Ku80 homologue, protects the ends of double-strand breaks created by Tpb2p	Lin et al. [23]
Tpb1p	PiggyBac-like transposase, excises a small subset of IESs	Cheng et al. [9]
Tpb2p	PiggyBac-like transposase, excises the majority of IESs	Cheng et al. [13]
Tpb6p	PiggyBac-like transposase, excises a small subset of IESs	Cheng et al. [9]
Twi1	Piwi protein, uses a scnRNA to locate IESs	Mochizuki et al. [2]
Twi11	Piwi protein, uses a scnRNA to locate IESs	Mochizuki et al. [2]

Tetrahymena life cycle

Tetrahymena is a unicellular eukaryote that lives in freshwater. When nutrients are plentiful, the MIC and MAC divide mitotically and amitotically, respectively, and binary fission follows to produce two daughter cells (Fig. 1A). This changes under starvation conditions, when pairs of cells with different mating types reproduce sexually through conjugation (Fig. 1B-J). During conjugation, both the new MAC and MIC initially contain the same genome. However, two types of programmed genome rearrangements occur in the developing MAC: chromosome breakage and DNA elimination. Additionally, the MAC genome undergoes multiple rounds of endoreplication.

Figure 1.

. Each cell contains a macronucleus (MAC) and a micronucleus (MIC). In the presence of sufficient nutrients, Tetrahymena reproduces asexually by binary fission (A). However, when there is a lack of nutrients, it reproduces sexually by conjugation (B-J). To start conjugation, two cells of complementary mating types fuse (B) and their MICs undergo meiosis (C). Three of the meiotic products are degraded, while the surviving nucleus undergoes mitosis (D). The fused cells then exchange a pronucleus (E) and the pronuclei fuse to create a zygotic nucleus (F), which undergoes two rounds of mitosis (G). Two products will develop into new MACs, one will form a new MIC and the fourth product is degraded. The parental MAC is also degraded and the fused cells separate (H). Finally, the MIC divides mitotically (I), which is followed by binary fission (J). The approximate time-scale of events is indicated in hours post-mixing (hpm)

The first type of programmed genome rearrangement, chromosome breakage, leads to the fragmentation of chromosomes at conserved chromosome breakage sequences [5]. The second, DNA elimination, results in the removal of approximately 12,000 internal eliminated sequences (IESs), from the new MAC genome [1]. The eliminated sequences mainly consist of TE-related sequences and elimination of these sequences is essential for cell viability, as indicated by the inability of mutants with defective DNA elimination to produce viable progeny [6-8]. A small subset of 12 IESs have conserved inverted terminal repeats. These IESs are excised in a transposon-like fashion by the piggyBac transposases Tpb1p and Tpb6p. However, the majority of IESs do not share one common motif. Yet, they are excised from the genome with near base-pair accuracy by another domesticated piggyBac transposase: Tpb2p [1,9,10]. Unlike the piggyBac transposases in other organisms such as Paramecium or Trichoplusia ni, Tpb2p does not appear to have strict specificity [11-13].

Excision of IESs by Tpb2p is orchestrated by small RNAs, termed scan RNAs (scnRNAs) [2]. In the current model, early-scnRNAs are produced from certain IESs (referred to as type-A IESs) and their surrounding regions in the MIC genome during the early-conjugation stage (Fig. 2A) [3]. These double-stranded ~28 nucleotide RNAs are processed by the Dicer-like protein Dcl1p [7], after which they are transported to the cytoplasm. There they form a complex with the Piwi protein Twi1p and the complex is transported to the parental MAC (Fig. 2B) [14]. Next, scnRNA selection occurs by degradation of scnRNAs that are complementary to the MAC genome (Fig. 2C) [15]. The remaining scnRNA-Twi1p complexes translocate to the new MAC where they induce heterochromatin formation at matching sequences (Fig. 2D) [16]. This includes IES from which no early-scnRNAs were produced (type-B IESs), which have repetitive sequences in common with the type-A IESs [3]. Next, another round of scnRNA production occurs (Fig. 2E). These late-scnRNAs further trigger heterochromatin formation [3]. Finally, the heterochromatinized regions are excised and the DNA ends are ligated back together (Fig. 2F).

Figure 2.

Programmed DNA elimination. Early scan-RNAs (scnRNAs) are produced from type-A internal eliminated sequences (IESs) and surrounding regions (A). Next, they are transported to the MAC (B) and scnRNAs complementary to the MAC genome are degraded. Meanwhile, a new MIC and MAC have been formed (C). The remaining scnRNAs are transported to the new MAC, where they recognize both type-A and – B IESs (D). The scnRNAs induce heterochromatin formation and the production of late-scnRNAs, which further spread the heterochromatin (E). Finally, the parts of the genome marked by heterochromatin are excised and the ends are ligated (F)

Early-scnRNA induced heterochromatin formation

One of the crucial components of the programmed DNA elimination machinery is the Piwi protein Twi1p [2]. In total Tetrahymena has 12 Twi proteins with varying functions, of which Twi1 and Twi11 are the only ones known to be essential for small RNA-directed DNA elimination [17]. Shortly after scnRNA biogenesis and the subsequent transportation to the cytoplasm, Twi1p stabilizes the double-stranded scnRNAs and removes one of their strands. Once mature, scnRNA-Twi1p complexes are bound by Giw1p, enabling translocation to the new MAC [18]. There the search for a sequence homologous to the scnRNA starts (Fig. 3A). Finding a matching target requires another component to join the complex: Ema1p. Aronica and colleagues showed that this putative RNA helicase is necessary for the interaction between the scnRNA-Twi1p complex and chromatin [19]. Their study suggests that Ema1p unwinds nascent RNAs that are transcribed from IESs, thereby facilitating base-pairing between these non-coding RNAs and the scnRNAs. In this way, the scnRNA-Twi1p complex thus gains indirect access to the genomic sequence.

Figure 3.

From Twi1p-scnRNA target search to IES elimination. Together with Ema1p, the Twi1p-scnRNA complex searches for targets in nascent transcripts (A). At matching sequences, histone H3 is methylated at lysine 9 (H3K9me3) and 27 (H3K27me3) by Ezl1p (B). Pdd1p is recruited to the heterochromatin (C). This is a requirement for the production of late scnRNAs that further spread the heterochromatin. At the same time, boundary-protecting factors confine the heterochromatin to IESs (D). Finally, Tpb2p excises IESs marked by heterochromatin (E)

In the search for matching targets, the early-scnRNA-Twi1p complex identifies type-A IESs from which the early-scnRNAs were originally derived. Additionally, type-B IESs are found because they share repetitive sequences with type-A IESs [3]. Once a sequence is identified that matches the scnRNA, the histone methyltransferase Ezl1p is recruited by the scnRNA-Twi1p complex through a mechanism that remains elusive (Fig. 3B). A study by Liu and colleagues showed that Ezl1p catalyses methylation of histone H3 at lysine 27 (H3K27me3), which is a histone mark characteristic of heterochromatin [16]. Additionally, they showed that Ezl1p-catalysed H3K27me3 regulates the methylation of histone H3 at lysine 9 (H3K9me3). In Tetrahymena, this second hallmark of heterochromatin is exclusively found at IESs in the developing MAC and the methylated histones attract additional proteins necessary for DNA elimination [20].

Late-scnRNAs and IES boundary determination

The role for small RNAs in DNA elimination does not stop here. Noto and colleagues found that a second group of scnRNAs is produced during late conjugation [3]. They showed that spreading of the heterochromatin induced by early-scnRNAs triggers the production of late-scnRNAs. This occurs through an unknown mechanism that requires Pdd1p, a protein that recognizes both H3K27me3 and H3K9me3 (Fig. 3C) [3]. Late-scnRNAs are loaded into zygotic Twi1p and Twi11p, but whether the two zygotic Piwi proteins have distinct functions is unknown. Guided by the late-scnRNAs, they further promote heterochromatin formation at IESs in an Ezl1p-dependent manner. This can trigger the production of more late-scnRNAs from other IESs in trans. Thus, a positive feedback loop is formed (Fig. 3D), which increases the robustness of the DNA elimination system. According to simulations by Noto and colleagues, this mechanism ensures that almost no IESs fail to be eliminated, even in the extreme case where only 1% of the type-A IESs produce early-scnRNAs [3].

To prevent the excision of DNA outside IESs, heterochromatin spreading must be stopped exactly at IES boundaries (Fig. 3D). In a study by Suhren and colleagues, several proteins were identified that are crucial for defining IES boundaries [21]. One of these boundary-protecting factors is the heterochromatin-binding protein Coi6p, which localizes to heterochromatinized IESs. In its absence, heterochromatin is not sharply confined to IESs and late-scnRNAs are also produced from regions outside IESs. Two proteins that interact with Coi6p, Coi7p and Lia5p were also found to negatively regulate heterochromatin formation. A second level of boundary control comes in the form of the histone demethylase Jmj1p. It reverses heterochromatin marks and is crucial for proper DNA elimination. However, the exact mechanism through which Jmj1p contributes to defining IES boundaries has not yet been revealed.

DNA elimination

IESs marked by heterochromatin are excised from the genome (Fig. 3E). The endonuclease responsible for this is the domesticated piggyBac transposase Tpb2p [14,15]. At IES boundaries it creates double-strand breaks (DSBs) with 4-nt long 5ʹ overhangs. Although piggyBac transposases are known to recognize and cleave 5ʹ-TTAA-3ʹ sequences, Tpb2p does not seem to have a strict sequence preference [13]. It does however have the ability to interact with histone marks specific for heterochromatin [13]. Hitherto, the exact mechanism ensuring that Tpb2p cleaves specifically at IES boundaries has not been elucidated.

Upon excision of the first IESs, aggregates of heterochromatinized IESs form [22]. These structures, referred to as heterochromatin bodies, also appear in response to DNA damage induced by UV [22]. This suggests that the function of heterochromatin bodies may be to facilitate DNA repair. Supporting this hypothesis is the finding by Lin and colleagues that a Ku80 homologue, TKu80p, is essential for the assembly of heterochromatin bodies [23]. They show that TKu80p binds to and protects the ends of the DSBs created by Tpb2p. This is followed by repair of the break through the non-homologous end joining (NHEJ) pathway. Another possible role for heterochromatin bodies, not excluding the former, could be to sequester freed IESs. This would prevent them from interacting with other genomic sequences before they are degraded.

Discussion

The general system through which Tetrahymena clears its somatic genome of TEs and TE-related sequences has been unravelled and important players have been identified (Fig. 3). However, several key questions concerning the molecular mechanism remain unanswered. A first question is how early-scnRNA induced heterochromatin leads to late-scnRNA production. It seems counter-intuitive that heterochromatin, which generally corresponds to a transcriptionally inactive state, triggers expression. However, transcription might be promoted by one of the proteins recruited to the heterochromatin. The role of many of these proteins remains elusive, further studies into their function might reveal the link between heterochromatin formation and late-scnRNA production.

Arguably the most intriguing open question is how Tetrahymena is able to excise thousands of sequences from its genome with high precision without any apparent sequence motif at the IES boundaries. The molecular machinery led by scnRNAs marks IESs with heterochromatin, which at best demarcates IESs with nucleosome resolution. What further narrows down the IES boundary? A recent study by Lin et al. suggests the existence of cis-acting boundary elements [24]. They identified multiple inverted repeats each of which potentially regulate a subset of IESs. These sequences are not located exactly at the IES boundaries but in a flanking region of about a hundred base pairs. Remarkably the same element is found at near equal distance from the two opposite boundaries of an IES. Altogether, they found six major groups of inverted repeats that are present in the flanking regions of approximately 60% of all IESs. Further studies focussing on the remaining 40% will point out whether cis-acting boundary elements are the global regulatory mechanism of determining IES boundaries or whether the boundaries of these IESs are set by a different mechanism. Additionally, identifying and characterizing proteins that bind the cis-acting elements could provide further insights into how these elements govern IES boundary determination.

Another open question is whether Tpb2p is recruited specifically to IES boundaries or whether it localizes to all heterochromatin. Although Tpb2p does not cleave DNA within the IES body, it could still bind to all heterochromatin if it is only activated at IES boundaries. In that case, an accessory protein is needed that localizes to euchromatin-heterochromatin transition regions and provides the trigger for Tpb2p activation. Studying the localization of Tpb2p, for example by ChIP analysis or high-resolution microscopy, could provide a first clue as to which of the two scenarios is correct. Additionally, finding interaction partners of Tpb2p will help to further unravel the mechanism regulating IES boundary determination.

Besides accurately removing such a large number of sequences with high accuracy, Tetrahymena also adequately degrades the excised sequences and ligates the remaining parts of the new MAC genome. Repair of the DSBs occurs through the NHEJ pathway, but these DSBs are different from typical DNA damage. When DNA damage occurs, the ends of the break are held close together to facilitate repair. However, in the case of DNA elimination, the ends that are to be ligated are different from the ends of the original DSB. It is crucial that the IES is removed and that the ends adjacent to the IES boundaries are the once that are ligated. Otherwise, IESs could be reintegrated in the genome or the genome could become scrambled. Little is known about this final stage of programmed DNA elimination. The aforementioned cis-acting boundary elements could play a role in keeping neighbouring MAC-destined regions together.

Perspective: potential for genome editing

Although multiple open questions remain, one can envision the Tetrahymena DNA elimination machinery being repurposed as a genome editing tool. Programmed DNA elimination in Tetrahymena is one of a wide variety of systems that organisms have developed to eliminate or silence foreign nucleic acid sequences. Another defence system against foreign DNA, the bacterial CRISPR-Cas9 system, has already been successfully employed for genome editing. Although it has quickly turned into an essential tool for life science research, the technique has several drawbacks including off-target effects [25]. Because it acts through a different mechanism than CRISPR-Cas9, the Tetrahymena DNA elimination machinery might provide valuable tools to complement the current genome editing toolbox.

One of the drawbacks of the CRISPR-Cas9 system is that it is not optimized to act in a chromatin context. The system is originally derived from prokaryotes, where it acts on naked DNA. In a recent study, Horlbeck and colleagues observed that nucleosomes hinder Cas9 in binding to the genome and cleaving it [26]. They noted that efficient targeting was almost exclusively achieved in nucleosome poor regions of the genome. Tetrahymena on the other hand is a eukaryote. Therefore, its DNA elimination machinery is tailored for accessing the eukaryotic genome. Thus, proteins from the Tetrahymena DNA elimination system might provide a gateway to chromatinized DNA, thereby increasing genome editing efficiency in eukaryotes.

A second drawback of the CRISPR-Cas9 system is that it might trigger an immune response in humans [27]. The most frequently used Cas9 variants are derived from two bacterial species: Staphylococcus aureus and Streptococcus pyogenes. Infection with these bacteria is common among humans and as a result the majority of the human population has an adaptive immune response to Cas9 [27]. This raises concerns regarding the safety and effectiveness of Cas9-based therapies as it might lead to the elimination of Cas9-expressing cells. Because Tetrahymena are not known to infect humans, there is expected to be no pre-existing adaptive immunity to Tetrahymena derived proteins. They might therefore provide a safer alternative to CRISPR-Cas9 for gene-editing in humans.

While CRISPR-Cas9 only requires a guide RNA and the Cas9 protein, a genome editing system based on the Tetrahymena DNA elimination machinery would consist of more components. First, a scnRNA-Twi1p complex is required to locate a target. This also requires Ema1p or another RNA helicase to provide access to nascent transcripts. The next crucial component is an endonuclease, which has to be recruited to the target site. Because many questions about Tpb2p are still unanswered, further research is required before it becomes clear whether Tpb2p can be used for genome editing. Tpb2p has for example been shown to have some sequence specificity, albeit not very strict [13]. Can it be directed to cleave at a specific site of choice? If so, it provides an advantage over Cas9 which is limited by the requirement of a PAM sequence adjacent to the target. Finally, Tpb2p interacts with the histone H3 tail, especially in the presence of H3K9me3 or H3K27me3 [28]. In Tetrahymena these heterochromatin marks are restricted to IESs, however in other organisms they are widespread through the genome. This could lead to off-target effects or inefficient targeting of euchromatin regions.

As an alternative, Cas9 could be used as the endonuclease. It would be interesting to explore ways to implement a form of ‘two-factor authentication’ by using a combination of Tetrahymena machinery and CRISPR-Cas9. Here a scnRNA with Twi1p would perform the first step of locating the target. Next, a conditionally active form of Cas9 would be used which is activated by a moiety carried by Twi1p. Cas9 would thus be activated at the target site, where its guide RNA ensures that it induces a DSB exactly at the right site. By using such a ‘two-factor authentication’ system, off-target effects are expected to be greatly reduced.

Conclusion

In conclusion, programmed DNA elimination in Tetrahymena is an intricate process of which the overall mechanism has been revealed. Further studies focussing on amongst others the open questions presented above are required to complement the current knowledge and obtain a detailed picture of the molecular mechanisms involved in programmed DNA elimination. This will not only increase our understanding of Tetrahymena biology, but it will also open new doors for exploiting the Tetrahymena DNA elimination system as a genome editing technique.