Modes of action
RNAi is based on the action of RISC complex. Once the small RNAs of different sizes are processed, one of the duplex strands is loaded onto one or more AGO proteins. The Arabidopsis genome encodes 10 AGO proteins (Mi et al., 2008; Montgomery et al., 2008; Takeda et al., 2008; Zhu et al., 2011). Base-pairing between the small RNA guide strand and the target RNA determines locus specificity, whereas the distinct activity of 24-nt vs. 21-22-nt small RNAs determines the type of silencing. Indeed, 24-nt siRNAs mediate transcriptional gene silencing (TGS) through DNA methylation and chromatin modifications while 21- and 22-nt siRNAs and miRNAs mediate post-transcriptional gene silencing (PTGS) through RNA cleavage and translational inhibition (Ding and Voinnet, 2007; Mallory and Vaucheret, 2009; Voinnet, 2009). A common set of enzymatic activities (RDRs, DCLs and AGOs) and mechanisms (synthesis of dsRNA, dicing into small RNA, and small RNA-directed RNA cleavage) are shared between TGS and PTGS pathways. However, the two processes have different outcomes as well as specific actors.
Transcriptional gene silencing
TGS is a function of nuclear RNAi, which in plants and fungi can repress target genes at the transcriptional level by epigenetic modification of chromatin. Different studies have shown the transcriptional silencing in yeast, cytosine methylation in plants and genome rearrangement in ciliates. Plants use DNA methylation to modulate the expression of transposons and repeats. In plants, DNA methylation in the promoter region can lead to transcriptional inhibition while methylation in the coding parts do not generally affect the transcription (Mette et al., 2000; Sijen et al., 2001). The first example of the RNA-directed DNA methylation (RdDM) in plants was reported during viroid infection in tabacco. DNA methylation of a transgene sequence homologous to that of the viroid genome was observed (Stam et al., 1998; Wassenegger et al., 1994).
As mentioned before, p4/p5-siRNAs are the main actors in directing RdDM. The transcripts from Pol IV serve as substrate for siRNA generation. RNA Pol IV interacts with RDR2 to produce dsRNA, which then are diced into 24 nt-siRNAs by the action of DCL3 (Kasschau et al., 2007; Law et al., 2011). These 24-nt siRNAs are exported into the cytoplasm where they can be loaded to one or more AGO proteins. AGO4 is the first AGO protein that has been shown to play a role during RdDM but there are at least 3 out of 10 Arabidopsis Argonautes, which are involved in this pathway (Zilberman et al., 2003). In fact, AGO6 was also found to act in certain cases, additionally to AGO4 to guide DNA methylation and TGS (Zheng et al., 2007). Also it has been shown that AGO9 can interact with 24 nt siRNAs and is required in silencing repetitive genomic regions involved in heterochoromatin formation (Duran-Figueroa and Vielle-Calzada, 2010).
Once the siRNA/AGO complex is formed, it will be imported to the nucleus and be guided to complementary pol V intergenic non-coding transcripts. The final result of this co-transcriptional silencing is the deposition of repressive cytosine methylation at the loci, which are transcribed by Pol V. In plants, DNA methylation occurs at cytosine bases in all CG, CNG (where N is any nucleotide) and CHH (any asymmetric site, where H is A, C or T) sequences. The genetic requirements for each kind of methylation are different and specific (Chan et al., 2005; Law and Jacobsen, 2010).
In general, plant DOMAINS REARRANGED METHYLTRANSFERASE 2 (DRM2) catalyses the de novo methylation at asymmetric CHH sites, whereas the maintenance of CG methylation is carried the maintenance of CHG methylation by CHROMOMETHYLASE 3 (CMT3)(Cao and Jacobsen, 2002; Matzke et al., 2009). At some RdDM loci, siRNAs are not detected in the absence of DRM2 which would suggest that a positive feedback loop is required for efficient siRNA dependent de novo methylation and gene silencing (Cao et al., 2003).
Post transcriptional gene silencing
Post-transcriptional gene silencing was first observed as an unexplainable co-suppression of the transgene and the homologous endogenous gene(s) in petunia and tomato (Napoli et al., 1990; Smith et al., 1990; van der Krol et al., 1990). Although both sense and antisense transgenes can lead to PTGS, inverted repeat transgenes more frequently induce PTGS (Beclin et al., 2002; Vaucheret, 2006). In fact, the inverted repeat transgenes produce RNA that are able to fold back directly and form a long dsRNA while sense transgenes need the action of RDR6 and SGS3 to be converted into a dsRNA (Brodersen and Voinnet, 2006; Vaucheret, 2006). Transgene-derived and endogenous dsRNA are diced into different size of small RNAs, but only 21- and 22-nt small RNAs have been identified as effectors of PTGS (Brodersen and Voinnet, 2006; Chapman and Carrington, 2007; Hamilton and Baulcombe, 1999; Vaucheret, 2006). 21- and 22-nt siRNAs can act either in cis or in trans if they exhibit complementarity with mRNA targets produced from loci that are distinct from the locus of origin. Post-transcriptional regulation in plants involves either RNA cleavage or translational repression (Brodersen et al., 2008).
RNAi as an Antiviral Response
Since the discovery of small RNAs, their vital functions became more and more clear. It may prove difficult to find a biological function or process that is not influenced at least to some degree or in some cell type, by small RNAs. Indeed, small RNAs participate to plant development, to adaptation to biotic and abiotic stress and to transposon control (Law and Jacobsen, 2010; Navarro et al., 2006; Robert-Seilaniantz et al., 2007; Vaucheret, 2006; Zhang et al., 2011). Moreover, RNA silencing participates to antiviral defense in plants.
Actors of the antiviral RNA silencing
Most plant viruses have an RNA genome and replicate through dsRNA intermediates, which can trigger RNA silencing (Ding and Voinnet, 2007). Indeed, DCLs likely process these intermediate dsRNA into primary viral siRNAs (v-siRNAs), which loaded onto a RISC complex, can mediate the cleavage of viral transcripts (Ruiz-Ferrer and Voinnet, 2009; Shimura and Pantaleo, 2011). DNA viruses do not replicate through dsRNA intermediates, therefore the v-siRNAs precursors are likely produced from transcription of both strands of the viruses (Chellappan et al., 2004).
All four plant DCLs participate to the production of v-siRNAs (Blevins et al., 2006; Moissiard and Voinnet, 2006), but mostly DCL2 and DCL4 produce v-siRNAs that target viral RNA for destruction (Bouche et al., 2006; Deleris et al., 2006). DCL2 antiviral activity is only a replacement of DCL4 in case of genetically compromising or suppressing of DCL4 by dedicated proteins called viral suppressors of RNA-silencing (VSRs) (see below).
AGO1 and AGO2 are able to bind v-siRNAs and thus are key component of antiviral RISC (Takeda et al., 2008; Zhang et al., 2006). Indeed, ago1 mutants are hyper-susceptible to cucumber mosaic virus (CMV) (Morel et al., 2002), and AGO1 is up-regulated upon virus infection (Zhang et al., 2006). In addition, AGO2 provides a secondary antiviral mechanism that is important when the primary AGO1-mediated layer is not active (Harvey et al., 2011; Wang et al., 2011).
RDR proteins also participate to the antiviral response. Indeed, RDR6 and RDR1 are required for efficient antiviral response by converting cleaved viral transcripts into dsRNA for the production of secondary v-siRNAs (Garcia-Ruiz et al., 2010; Wang et al., 2010).
Viral suppressors of RNA silencing
Plant viruses encode suppressor proteins of RNA silencing as counter-defense mechanisms that influence their accumulation and spread in their host plants (Bivalkar-Mehla et al., 2011; Burgyan and Havelda, 2011; Harvey et al., 2011; Zamore, 2004). The various VSRs are able to target many effectors of the silencing pathway involved in viral RNA recognition, dicing, RISC assembly, RNA targeting and amplification. Most of the identified VSRs are multifunctional. Indeed, they often perform essential roles for viruses such as coat proteins, replicases, movement proteins and transcriptional regulators (Gillet et al., 2013; Hao et al., 2011; Sansregret et al., 2013).
The V2 protein of tomato yellow leaf curl gemnivirus (TYLCV) competes with SGS3 for RNA binding, thus preventing the production of secondary v-siRNAs (Glick et al., 2008).
The pothos latent aureusvirus P14 protein inhibits the processing of dsRNAs into siRNAs by binding to dsRNA in a size-independent manner (Merai et al., 2006; Merai et al., 2005).
The cauliflower mosaic virus (CaMV) P6 is a viral transcriptional trans-activator protein, which has been recently described as a VSR factor. In fact, P6 is able to suppress RNA silencing by interacting with dsRNA-binding protein 4 (DRB4), which is required for DCL4 functioning (Haas et al., 2008).
The cucumber mosaic virus (CMV) 2b protein is able to physically interact with the PAZ domain and part of PIWI domain of AGO1, which inhibits AGO1activity (Zhang et al., 2006).
The turnip crinckle virus (TCV) P38 protein is also able to directly interfere with AGO1 function by binding to the PIWI domain through its GW motif (Azevedo et al., 2010).
The P0 protein of the phloem-limited poleroviruses can perturb AGO1 activity in an indirect manner. This protein contains an F-box motif, which hijacks the host SCF ubiquitin-protein ligase (E3) to promote AGO1 degradation (Baumberger et al., 2007; Pazhouhandeh et al., 2006).
The tombuvirus P19, which is one of the most studied VSRs, prevents RNA silencing through binding double-stranded siRNA (Silhavy et al., 2002).
The Beet yellows virus p21 protein can suppress silencing by binding directly to v-siRNAs (Ye and Patel, 2005).
At last, sweet potato chlorotic stunt virus (SPCSV) encodes an RNaseIII that likely cleaves siRNA, including v-siRNAs, thus inhibiting antiviral silencing activity (Cuellar et al., 2009; Kreuze et al., 2005).
A summary of different types of VSRs and their effects on RNAi pathways can be found in Fig. 1.5.
Presentation of the objectives of the thesis
Dicer-independent small RNAs
In eukaryotes, the biogenesis of small RNAs generally requires Dicer. However, a few Dicer-independent small RNAs have been reported in fungi, invertebrates and mammals (Cheloufi et al., 2010; Halic and Moazed, 2010; Lee et al., 2010).
In mammals, the conversion of pre-miR-451 to mature miR-451 requires the action of Ago 2 and not Dicer (Cheloufi et al., 2010). The mammalian Ago 2 protein is the only Ago that has endonuclease activity, and this activity is required for the production of miR-451(Liu et al., 2004). Indeed, in vitro incubation of the wild type, but not catalytically inactivated, Ago2 with 5’end-labelled pre-miR451 produces mature miR-451(Cheloufi et al., 2010).
In Neurospora crassa, a filamentous fungus, two dicer enzymes process dsRNAs into small RNAs, which are loaded onto the Argonaute protein QDE-2 (QUELLING DEFFICIENT-2). QDE-2 interacts
Fig. 1.5 Plant antiviral defense and viral suppressors of RNA silencing.
A fraction of viral sense RNAs is translated into viral proteins. Another fraction is transformed into dsRNA by viral RNA-dependent RNA polymerases during viral replication. Viral dsRNA are diced by DCLs, mostly DCL2 and DCL4. Viral siRNAs associate with AGOs, mostly AGO1 and AGO2, to guide the cleavage of viral sense RNA. Cleaved viral products are protected from degradation by SGS3 and copied into dsRNA by cellular RNA -dependent RNA polymerases, mostly RDR6 and RDR1. dsRNA are diced by DCL2 and DCL4 to form secondary viral siRNAs, which reinforce the cleavage of viral sense RNAs.
Viruses encode proteins that act as suppressors of RNA silencing (VSR). These VSRs are able to inhibit many steps of the plant antiviral silencing pathway, including the protection of cleavage products by SGS3 (V2), the dicing of dsRNA by DCL4 (P14) and DRB4 (P6), the loading of siRNA duplexes onto AGO1 (P19, P21 and RNaseIII), the stability of AGO1 (P0), and the cleavage of viral sense RNA by AGO1 (2b).
with QIP, an exonuclease that removes the siRNA* strand, resulting in an active RISC complex (Maiti et al., 2007). Sequencing results of QDE-2 associated small RNAs allowed identifying a distinct classe of miRNA-like small RNAs (milRNAs). Both pre-milR and mature miRL remain constant in the dcl-1/dcl-2 double mutant, indicating that the biogenesis of milR is Dicer independent but requires QDE-2 (Lee et al., 2010).
In Caenorhabditis elegans, a substantial fraction of siRNA does not derive directly from input dsRNA. Instead, a population of antisense siRNAs called secondary siRNAs derives from the direct action of a cellular RNA-directed RNA polymerase (RdRP) on mRNAs that are targeted by primary siRNA(Sijen et al., 2007).
In the fission yeast S. pombe, a distinct class of small RNAs, called primal small RNAs (priRNAs), exists beside Dicer-dependent siRNAs. priRNAs are degradation products of abundant transcripts that result from bidirectional transcription of DNA repeats (Halic and Moazed, 2010)
Up to now, Dicer-independent small RNAs have been only found in fungi, invertebrates and mammals, but not in plants. However, beside its four Dicer proteins, the Arabidopsis genome potentially encodes five non-Dicer proteins that are referred to as RNASE THREE-LIKE (RTL) proteins because they carry at least one RNase III domain (Fig. 1.6.B). So far, it has not been determined if these proteins participate to the small RNA repertoire, either by playing a direct role in small RNA production or an indirect role in regulating small RNA biogenesis or degradation.
The RNaseIII family
Double-stranded (ds)RNA-specific endoribonucleases referred to as RNaseIII enzymes, are found in all eukaryotes and bacteria (Comella et al., 2008). All members of the RNase III family are characterized by a highly conserved stretch of nine amino acid residues referred to as the RNaseIII signature motif (Comella et al., 2008; Du et al., 2008). RNaseIII enzymes range in length from approximately 200 to 2000 amino acid residues depending on the number of additional domains that they carry. Accordingly, they are divided in four classes based on their molecular weight and structural complexity (Fig. 1.6.A)(Filippov et al., 2000).
Class I has single RNase III and dsRNA-binding (DRB) domains (Kharrat et al., 1995). Enzymes of this class are found in bacteria and viruses. Escherichia coli and Aquifex aeolicus RNase III are representative enzymes of this class. Class II differs from Class I by the presence of a highly variable N-terminal domain extension of approx. 200 amino acid residues. Saccharomyces cerevisiae Rnt1 and S.pombe Pac1 enzymes are both members of class II (Lamontagne et al., 2004; Rotondo et al., 1997; Wu et al., 2004). Class III contains two RNase III domains, one DRB domain and a large N-terminal extension. Drosha, which processes pri-miRNAs to pre-miRNAs in animals, is a member of this class (Du et al., 2008; Filippov et al., 2000). Class IV is by far the largest and structurally most complicated A) Canonical and atypical RNaseIII: Escherichia coli and Aquifex aeolicus RNaseIII are representatives of class I (one RNaseIII + one DRB). Saccharomyces cerevisiae Rnt1 and S.pombe Pac1 are representatives of class II (one RNaseIII + one DRB + N-terminal domain). Animal Drosha is representative of class III (two RNaseIII + one DRB). Animal and plant Dicer are representatives of class IV (one RNA helicase + one PAZ + two RNaseIII + one or two DRB). Atypical Dicer proteins lacking some of the Class IV domains are found in S. castelli and G. intestinalis.
B) Plant RNaseIII: Arabidopsis contains four DICER-LIKE (DCL) and five RNASE-THREE-LIKE (RTL) proteins. RTL1 resembles Class I proteins. RTL2 resembles atypical S. castelli Dicer. RTL3 exhibit an atypical organisation of multiple RNaseIII and DRB domains. RTL4 and RTL5 contain only one RNaseIII domain but no DRB domains.
RNaseIII family and corresponds to small RNA-producing Dicer enzymes, which generally contain two RNase III domains, one or two DRB domains, one PAZ domain, one RNA Helicase domain and one domain of unknown function (DUF283).
Class I and Class II enzymes, which carry a single RNase III domain, form a catalytic RNase III site by intermolecular homodimerization. The enzymatic studies on Escherichia coli RNase III have suggested a “single processing center model” for RNase III cleavage. In this model the action of RNaseIII proteins relies on the formation of a catalytic dimer of RNase III domains, which can be obtained by intermolecular homodimerization. Moreover, studies on the Aquifex aeolicus RNase III protein revealed that four patches of residues, also known as RNA-binding motifs1-4 (RBM1-4) are involved in protein-RNA interactions. RBM1 and RBM2 are located in the DRB domain whereas RBM3 and RBM4 are located in RNaseIII domain (Gan et al., 2006). The RNaseIII activity on dsRNA requires divalent cations such as Mg+2 or Mn+2 (Kreuze et al., 2005).
For the most part bacterial and eukaryotic RNase III cleavage of dsRNA precursors is performed in a non-sequence-specific manner. Nevertheless, the presence of AGNN or AAGU terminal tetraloops is a strong determinant for S.cerevisiae Rnt 1 binding and cleavage (Gaudin et al., 2006; Wu et al., 2004). RNaseIII products contain 5’ phosphate and 3’hydroxyl termini and a 2-nt overhang at the 3’ ends (Du et al., 2008). Generally, RNA cleavage mediated by Class I or Class II enzymes is performed in a non-sequence-specific manner. Nevertheless, the presence of AGNN or AAGU terminal tetraloops is a strong determinant for S.cerevisiae Rnt 1 binding and cleavage (Gaudin et al., 2006; Wu et al., 2004).
Class III and Class IV, which carry two RNase III domains, form a catalytic RNase III site by intramolecular heterodimerization between their two RNaseIII domains (Zhang et al., 2004). Class III Drosha proteins, which are involved in the first cut of animal miRNA precursors, are incapable of producing small RNAs. In contrast, Class IV Dicer proteins produce small RNAs. Dicer proteins contain one or two DRB domains. They also contain one PAZ, one RNA Helicase and a DUF283 domain. The PAZ domain recognizes the 3’ 2-nt overhang of dsRNAs, while the RNA helicase act as a molecular ruler to measure the distance from the dsRNA end (recognized by the PAZ domain) (Liu et al., 2009). The DUF283 domain has recently been suggested to play significant roles for partner protein selection in small RNA processing (Fig. 1.6.B) (Qin et al., 2010).
Remarkably, some Dicer proteins from lower eukaryotes have simpler domain structures. For example, the Giardia intestinalis Dicer protein only contains a PAZ and two RNase III domains (Macrae et al., 2006). The PAZ domain is directly connected to the RNase IIIa domain by a long α helix that runs through the handle of the molecule. This “connector” helix is encircled by the N-terminal residues of the protein, which form a platform domain composed of an antiparallel β sheet and three α helices. Measuring from the active site of the first RNase III domain of Giardia Dicer to the 3′ overhang-binding pocket in the PAZ domain gives a distance of ∼65 Angstrom, which matches the length of RNA fragments produced by Giardia Dicer, which are 25 to 27 nucleotides long (Macrae et al., 2006).
Moreover, S. castelli Dicer has one RNAseIII domain and two DRB domains, and thus differs from Class II Rnt1 by an additional DRB domain (Drinnenberg et al., 2009; Weinberg et al., 2011). Despite significant differences in domain structure between the budding-yeast and canonical Dicers, the RNase
III signature motif and active site residues important for dsRNA cleavage are conserved in Dcr1, suggesting that the RNase III cleavage activity is conserved in Dcr1. In fact, S. castellii extract contains an activity that produces 22- to 23-nt RNAs from dsRNA precursors. Moreover, in Δdcr1 S.castellii mutants, siRNAs failed to accumulate, suggesting that the Dcr1 functions in Dicer-specific activities (Drinnenberg et al., 2009). However, budding-yeast Dicers produce siRNAs through a mechanism different from that of canonical Dicers. Instead of successively removing siRNA duplexes from the dsRNA termini, Dcr1 starts in the interior and works outward (Weinberg et al., 2011).
Functions of plant RNaseIII
The Arabidopsis genome encodes four Dicer-Like (DCL) proteins with known functions. DCL1 produces miRNAs from imperfectly double-stranded stem-loop RNA transcripts which are generated from MIR genes (Park et al., 2005). DCL2, DCL3 and DCL4 process near-perfect long dsRNA precursors to generate either siRNAs or young miRNAs. 24-nt p4/p5-siRNAs are derived from the action of DCL3 (Daxinger et al., 2009). DCL4 produces 21-nt young miRNA, ta-siRNAs and v-siRNAs during viral infections (Bouche et al., 2006; Gasciolli et al., 2005; Xie et al., 2005). DCL2 produces 22-endoIR-siRNAs and can replace DCL4 and DCL3 in their absence (Bozorov et al., 2012; Deleris et al., 2006; Dunoyer et al., 2010).
Beside the four DCL proteins, the Arabidopsis genome potentially encodes five RNASE THREE-LIKE (RTL) proteins, which have not been well studied yet. RTL1 contains only one RNase III and one DRB domain. RTL1 is only expressed in Arabidopsis roots at very low levels, and its function is unknown. RTL2 contains one RNaseIII domain and two DRB domains. RTL2 is ubiquitously expressed. RTL2 cleaves the 3’External Transcribed Spacer (EST) of ribosomal 45S-pre-rRNA in vitro and in vivo (Comella et al., 2008). An in vitro assay has revealed that RTL2 processes dsRNA into 21-25-nt small RNAs (Kiyota et al., 2011). rtl2 mutants do not show any obvious developmental defects (Comella et al., 2008). RTL3 is the biggest member of the RTL family and has two RNase III domain and three DRB domains. RTL3 mRNA has not been detected in any tested tissue (Comella et al., 2008). RTL4 contains only one RNaseIII domain but no DRB domain, which makes it similar to the Bacillus subtilis Mini-III structure (Redko et al., 2008). RTL4 is expressed in almost every tissue. rtl4 mutants are impaired in male and female gamethophyte formation; however, RTL4 function remains unclear (Portereiko et al., 2006). RTL5 contains only one RNaseIII domain but no DRB domains. RTL5 is expressed in almost every tissue. RTL5 is similar to maize RNC1, which splices several chloroplastic group II introns (Watkins et al., 2007).
The aim of this project is to expand our understanding of RTL1 and RTL2 functions and to shed light on their potential interaction with plant small RNA pathways.
Table of contents :
1 – Introduction
1.1 – RNA Silencing: Definition, History & Key factors
1.2 – RISC: the main actor for RNA silencing
1.2.1 – Argonaute proteins
1.2.2 – Small RNAs
1.3 – Modes of action
1.3.1 – Transcriptional gene silencing
1.3.2 – Post transcriptional gene silencing
1.4 – RNAi as an Antiviral Response
1.4-1 – Actors of the antiviral RNA silencing
1.4.2 – Viral suppressors of RNA silencing
1.5 – Presentation of the objectives of the thesis
1.5.1 – Dicer-independent small RNAs
1.5.2 – The RNaseIII family
1.5.3 – Functions of plant RNaseIII
2 – Materials and Methods
2 .1 – Plants and viruses
2.1.1 – Plant material
2.1.2 – Plant transformation
2.1.3 – Virus and Bacterial inoculation
2 .2 – Plasmid constructs
2.2.1 – RTL1 and RTL2 tag fusion
2.2.2 – RNaseIII mutation
2.2.3 – DRB deletion
2.2.5 – amiR-RTL1
2.3 – DNA, RNA and protein analysis
2.3.1 – RNA extraction and Northern analyses
2.3.2 – RT reactions
2.3.3 – Protein extractions and immunoblotting
2.3.4 – GUS enzymatic activity test
3 – Results
3.1 – RTL1
3.1.1 – Induction and suppression of RTL1 by viruses (article)
3.1.2 – Supplementary Results
3.2 – RTL2
3.2.1 – RTL2 enhances the accumulation of 21- and 24-nt siRNAs
3.2.2 – Mutations in the RTL2 RNaseIII domain mostly affect the accumulation of 24-nt siRNAs
3.2.3 – The second DRB domain is required for RTL2 activity
3.3 – Epistatic relationship between RTL1 and RTL2
4.1 – RTL1
4.2 – RTL2
5 – Possibilities for future research
5.1 – Short-term goals
5.1.1 – RTL specificity
5.1.2 – RTL1 localization
5.1.3 – Complimentary viral trails with RTL1
5.1.4- RTL2 overexpression in dcl mutants
5.2 – Long term goals
5.2.1- Functional studies of other RTLs
5.2.2 – RTL interactions with other proteins/RNAs
6 – References