Viroids

Authors

  • Efthimia Mina Tsagris,

    Corresponding author
    1. Institute of Molecular Biology and Biotechnology, Foundation for Research and Technology, PO Box 1385, 71110 Heraklion, Greece.
    2. Department of Biology, University of Crete, PO Box 2208, 71409 Heraklion, Greece.
    Search for more papers by this author
  • Ángel Emilio Martínez de Alba,

    1. Institute of Molecular Biology and Biotechnology, Foundation for Research and Technology, PO Box 1385, 71110 Heraklion, Greece.
    Search for more papers by this author
    • Present addresses: Instituto de Biología Molecular y Celular de Plantas, Universidad Politécnica de Valencia, Consejo Superior de Investigaciones Científicas, Valencia, Spain;

  • Mariyana Gozmanova,

    1. Institute of Molecular Biology and Biotechnology, Foundation for Research and Technology, PO Box 1385, 71110 Heraklion, Greece.
    2. Department of Biology, University of Crete, PO Box 2208, 71409 Heraklion, Greece.
    Search for more papers by this author
    • University of Plovdiv, Department of Plant Physiology and Molecular Biology, 24, Tsar Assen St., 4000 Plovdiv Bulgaria.

  • Kriton Kalantidis

    1. Institute of Molecular Biology and Biotechnology, Foundation for Research and Technology, PO Box 1385, 71110 Heraklion, Greece.
    2. Department of Biology, University of Crete, PO Box 2208, 71409 Heraklion, Greece.
    Search for more papers by this author

*E-mail tsagris@imbb.forth.gr; Tel. (+30) 2810 394367; Fax (+30) 2810 394404.

Summary

Viroids are small, circular RNA pathogens, which infect several crop plants and can cause diseases of economic importance. They do not code for proteins but they contain a number of RNA structural elements, which interact with factors of the host. The resulting set of sophisticated and specific interactions enables them to use the host machinery for their replication and transport, circumvent its defence reactions and alter its gene expression. Although found in plants, viroids have a distant relative in the animal world: hepatitis delta virus (HDV), a satellite virus of hepatitis B virus, which has a similar rod-like structure and replicates in the nucleus of infected cells. Viroids have also a cellular relative: the retroviroids, found in some plants as independent (non-infectious) RNA replicons with a DNA copy. In this review, we summarize recent progress in understanding viroid biology. We discuss the possible role of recently identified viroid-binding host proteins as well as the recent data on the interaction of viroids with one part of the host's defence machinery, the RNA-mediated gene silencing and how this might be connected to viroid replication and pathogenicity.

Introduction

Studies on viroids have led to the discovery of some of the most interesting principles of the biology of RNA: the fact that a non-coding, non-translatable RNA can cause a disease (Diener, 1971), the extraordinary small size of their genome (Gross et al., 1978) and their circularity (Sänger et al., 1976), which enables them to circumvent the problems of linear genome replication such as the accurate replication of linear ends (Diener, 1989). One of the first self-cleaving structures, the hammerhead ribozyme, was found in a satellite RNA virus and a viroid RNA (Prody et al., 1986; Forster and Symons, 1987; Forster et al., 1987). Prior to the molecular characterization of the hepatitis delta virus (HDV), a circular (‘viroid-like’) RNA infecting human liver cells and associated with hepatitis B virus (Lai, 2005; Taylor, 2006) viroids remained as an interesting, but exotic example of a plant pathogen, investigated by some researchers in the field of plant pathology and molecular plant virology, together with a group of biophysicists studying RNA structure. They recognized in this relatively abundant, natural RNA a perfect object to study RNA structure transitions. Viroids have been the basis on which new experimental and computational methods have been developed (Riesner, 1991; Steger and Riesner, 2003). A recent landmark of scientific breakthrough originating from viroid research is the discovery of RNA-mediated de novo DNA methylation, which was first described in transgenic plants carrying copies of the viroid cDNA (Wassenegger et al., 1994). Although this was an ‘artificial’ transgenic system, it was quickly recognized that RNA-mediated DNA methylation is a mechanism that is a part of a whole battery of responses that plants have towards environmental changes and developmental programmes (Wassenegger, 2005; Henderson and Jacobsen, 2007).

Several reviews have been published recently on viroids, showing the increasing scientific interest in these molecules as a model system and plant pathogen. In this review, we will discuss some of the most recent articles on viroids, and present some possible models concerning their replication, biogenesis and evolution.

Basic properties and mode of replication of viroids

Viroids have a size of ca 250–400 nucleotides. They infect several crop plants, causing symptoms of differential severity, which range from mild effects such as hardly visible growth reduction, up to deformation, necrosis or chlorosis and severe stunting (Singh et al., 2003). Some viroid strains do not cause symptoms at all and seem to behave as simple RNA replicons rather than pathogens. However, symptoms depend very much on environmental conditions and may change during infection as has also been found with plant viruses (Semancik, 2003).

Viroids do not have coding capacity and so far no viroid-coded protein has been detected in infected plants. Viroids are not encapsidated. They replicate in the plant hosts by RNA–RNA transcription, in a ‘rolling-circle’ mode (Branch and Robertson, 1984). During replication, oligomeric linear RNAs of plus polarity are formed, which are cleaved to monomers and ligated into circles (Fig. 1A). Oligomeric forms of (−) polarity are also present as replicative intermediates. Circular molecules of minus polarity have been detected in plants only in some members of the family Avsunviroidae, which bear self-cleaving ribozyme structures in their RNA genomes. Viroid localization is either nuclear (family Pospiviroidae) or chloroplastic (family Avsunviroidae), where they replicate with the aid of host-encoded DNA-dependent RNA polymerases. Viroids can therefore be considered as parasites of the transcriptional machinery of the organelles (nucleus or chloroplast), in contrast to most plant RNA viruses, which replicate in the cytoplasm and can be therefore regarded as parasites of the translational machinery of the cell.

Figure 1.

A. A rolling-circle model has been proposed for replication of the circular RNA genome in order to explain the generation of oligomeric linear forms (Branch and Robertson, 1984). RNAs of sense polarity are shown in red and those of antisense polarity in blue. Processing structures are in yellow or light blue and mark the length of the linear unit. Pospiviroidae replicate in the nucleus of the plant cell. The replication cycle is asymmetric, because only circular RNAs of (+) polarity are produced. Pospiviroidae do not contain known self-cleaving structures. Avsunviroidae replicate in the chloroplast. They contain active self-cleaving and possibly ligating hammerhead ribozyme structures on the RNAs of both polarities (yellow or light blue lines). Their replication cycle is classified as symmetric, because a circular RNA of both polarities is formed.
B and C. Schematic representation of systemic infection of nuclear viroids. In the circles, a cross-section of a leaf is shown (according to figs 15.9 and 21.15 of Buchanan et al., 2000; and fig. 3c of Lucas and Gilbertson, 1994). Different cell types are represented by different forms and colours. EC: epidermal cells; MC: mesophyl cells: BSC: bundle sheath cells; CC; phloem companion cells; SE: sieve elements. The nuclear membrane is represented by a dotted red circle. Black lines represent the difference in number of plasmodesmata connecting different cells. Sieve elements are degenerated cells which form the vascular tubes and do not contain nuclei. The viroid RNA is represented by open circular forms with different colours. Each colour indicates a replication cycle in the next cell (brown, orange, red, yellow, magenta), indicating possible differences in the RNA–protein complexes. For simplicity, intermediate oligomeric forms are not shown in this scheme.
B. Events occurring on the first infected (or inoculated) leaf. After damage of epidermal tissues, the viroid moves slowly from cell to cell. One possibility would be that the viroid undergoes a replication cycle in each nucleus before moving to neighbouring cells. Replication in the inoculated leaf remains at low levels.
C. Events occurring in the newly developing leaves. Viroid RNA has to cross the plasmodesmata from phloem tissues (SE, CC) to bundle sheath and mesophyl cells (BSC, MC). A specific structure of PSTVd has been shown to be necessary for trafficking of the RNA from phloem companion cells (CC) to bundle sheath cells (BSC) (Zhong et al., 2007) and from bundle sheath to mesophyl cells (Qi et al., 2004). Some cells may remain viroid-free due to effective defence by the silencing mechanism.

In some plants, non-infectious circular RNA transcripts of similar size to viroids have been detected, which contain hammerhead ribozyme structures (Darós and Flores, 1995; Hegedus et al., 2001). These transcripts exist in the genome of their hosts as extra chromosomal or integrated concatameric DNA copies; if they replicate, they do so via oligomeric forms of plus and minus polarity. As they are usually found in plants associated with a pararetrovirus they may use the machinery of the virus to copy and integrate their cDNA in the host's genome. They have been named ‘retroviroids’ (Darós and Flores, 1995). One can speculate on whether viroids evolved from retroviroids, or vice versa, and what the intermediate steps have been.

Viroids infect the epidermis of their hosts after mechanical damage of the plant cell wall. Unlike most other plant viruses, they do not have important natural vectors from the animal kingdom (Singh et al., 2003). After entry to the first cell layers of the leaves and initial replication, they are transported to neighbouring cells, reaching the vascular tissues, from which they travel to newly emerging young leaves or other sink tissues like roots, together with photoassimilates (Ding et al., 2005). Infected plants contain viroid RNA in the conductive plant organ, the phloem and its surrounding cells (Fig. 1B and C). For cell-to-cell and long-distance transport, viroids use the supracellular structure of vascular plants, where groups of cells are connected via special channels, called plasmodesmata. Therefore, like plant viruses but unlike animal viruses, viroid RNA does not have to cross the plasma membrane for long-distance transport. It does, however, cross the borders of the nucleus via the nuclear pore or the chloroplast membranes. The precise mechanism by which this transfer is executed is not known.

The entire replication and spread of viroids depends on host factors. DNA-dependent RNA polymerase II has been shown to replicate nuclear localized viroids (reviewed in Tabler and Tsagris, 2004), while it has been suggested that the nuclear-encoded, but chloroplast-localized DNA-dependent RNA polymerase (NEP) replicates the chloroplastic viroids (Flores et al., 2004). Viroids and HDV seem to be able to delude specific DNA-dependent RNA polymerases of their host cells, offering genomic and antigenomic RNAs as a competitive template to the normal DNA templates of these enzymes. Or, alternatively, they just ‘remind’ to these enzymes their ‘RNA World’ attributes. In order to be able to compete with the normal DNA templates, these RNA replicons must have a competitive advantage: either a high-affinity binding site, or the independence of accessory trans-acting factors for recognition of the primitive ‘promoter’ by the DNA-dependent RNA polymerase, or both.

The viroids Potato spindle tuber viroid (PSTVd) and Citrus exocortis viroid (CEVd) are both replicated in the nucleus, most probably by the same enzyme. PSTVd has been shown to be transcribed in vitro and in vivo by DNA-dependent RNA polymerase II (Mühlbach and Sänger, 1979; Rackwitz et al., 1981) and CEVd exists in a complex with at least the large subunit of the DNA-dependent RNA polymerase II (Warrilow and Symons, 1999). Putative initiation sites for PSTVd have been determined, using different methods with varying results. Earlier work has identified nucleotide G168, localized at the right terminal part (Tabler and Tsagris, 1990), as transcription initiation site but nucleotides U359, C1 and others (Kolonko et al., 2006) have also been identified as potential transcription initiation sites. The methods used did not, however, allow the identification of a genuine 5′-triphosphate transcript.

For Avocado sunblotch viroid (ASBVd), a viroid RNA pathogen replicating in the chloroplast, it has been proposed that a nuclear-encoded DNA-dependent RNA polymerase (NEP) is involved in its replication (Navarro et al., 2000). NEP is a nuclear-encoded enzyme, which is transported to the chloroplast and functions there during chloroplast development. It is also a candidate protein for transporting the Avsunviroid RNAs into the chloroplast. Recent findings however question whether indeed NEP or PEP (a plastid-encoded DNA-dependent RNA polymerase) is the replicating enzyme for avsunviroids (Motard et al., 2008). Regardless of which RNA polymerase is involved, the initiation sites for RNA replication for two chloroplast-replicating viroids have been determined using an elegant method of labelling the newly generated transcripts with guanylyltransferase (Navarro and Flores, 2000; Delgado et al., 2005). The initiation sites for both Peach latent mosaic viroid (PMLVd) RNA polarities reside at a stem of the conserved hammerhead ribozyme structure of the molecule, adjacent to the ribozyme cleavage sites (Delgado et al., 2005). A conserved hexanucleotide has been shown to be a characteristic motif just before the initiation site, not only of the genuine viroid template, but also for an in vitro selected shorter RNA template (Motard et al., 2008) thus resembling most probably the primitive RNA promoter (Pelchat et al., 2002).

Comparing the replication steps of viroids and HDV, common but also distinct mechanisms can be identified. One of the features that all these circular RNA pathogens have in common is that they replicate in DNA-containing organelles (nucleus or chloroplasts), so that their RNA must therefore be transported to the corresponding organelle. HDV is larger than viroids, and contains in the antigenomic strand the open reading frame for the hepatitis delta antigen (small and large forms, HDAg) (Taylor, 2006). A homologue protein to HDAg has been reported to exist in the human genome. It is conceivable that HDV satellite RNA had originally a smaller size and acquired through RNA recombination the RNA encoding the HDAg (Brazas and Ganem, 1996). Reverse transcription has not been reported to be important for either viroids or HDV; no DNA intermediate forms exist, and no integration in the genome of the host has been shown to occur. Plant retroviroids are circular RNA replicons which can be transmitted only through cell division; they are not pathogenic and cannot replicate autonomously (Darós and Flores, 1995). Most probably, they are reverse transcribed from a pararetrovirus reverse transcriptase (with which they are often associated). Another possibility would be that reverse transcriptase activities of endogenous retroelements, which are very abundant in many crop plants, participate in their replication. Retroviroids and some viroids contain hammerhead ribozymes which are responsible for steps during the rolling-circle replication (Côtéet al., 2003). Specific Hop stunt viroid (HSVd) strains contain a conserved sequence element which has partial homology to the hammerhead ribozyme structure. This sequence element has not been reported to have self-cleaving activity and possibly resembles an evolutionary link between viroids still possessing the hammerhead structure and those having lost it (Amari et al., 2001). Hepatitis delta genomic and antigenomic RNAs contain ribozymes forming other characteristic hairpin structures (Taylor, 2006). A model for the generation and evolution of circular viroids from simpler RNA molecules is presented in Fig. 2A.

Figure 2.

A. A model for the generation of imperfect self-complementary helical circular RNAs. An RNA molecule has, for example, three different domains (black, green, red) and contains the structure that an RNA polymerase (represented by a light blue full circle) recognizes as a promoter. At the 3′-OH end, the structure of the RNA allows extension by the polymerase, templated by the rest of the RNA (direction of new synthesis indicated by the red arrow, complementary parts shown as dotted lines in the same colours, black, green and red). The result is an imperfect helix connected by a loop (similar to miRNA precursors). RNA–RNA transcription by DNA-dependent RNA polymerases is error-prone and results in the generation of bulged structures in the helix. This RNA acquires by recombination or through evolution a self-cleaving structure (blue line). The RNA can undergo cleavage to monomers (forward reaction), circularization or oligomerization (reverse reaction of the ribozyme).
B. A possible role for Virp1. Virp1 (magenta full ellipses) is involved in import of viroid RNA (brown open circular structures) to the nucleus. Other proteins (black full circles) may participate in intercellular transport or export from the nucleus.

Another similarity between the plant RNA replicons and HDV is that they both infect organisms (plants and humans respectively) for which DNA methylation is an important epigenetic control mechanism. Viroid-like RNAs have not been found in other organisms such as invertebrates or unicellular organisms and most probably they do not exist there as pathogens (because they would have been detected).

Recent advances in viroid research

Host factors involved in viroid replication and transport

In recent years, focus on viroid research has moved on from structure–function analysis of the viroid RNAs to the search of host factors which interact directly or indirectly with viroids and facilitate their replication and transport (Tabler and Tsagris, 2004; Flores et al., 2005b). Different methods have been used to isolate host proteins interacting with the viroid RNA: direct RNA ligand screening method (Sägesser et al., 1997), cross-linking and/or biochemical isolation (Wolff et al., 1985; Darós and Flores, 2002) and genomic approaches, using microarrays to identify which genes are induced and which ones repressed after infection by viroids (Hammond and Zhao, 2000; Itaya et al., 2002; Tessitori et al., 2007). Biochemical identification after cross-linking has the advantage of isolating complexes of viroid RNA with host proteins which exist in vivo, but transient and short-lived complexes or complexes existing in minute amounts may escape from this approach. A ‘direct’ approach of screening for RNA-binding proteins in cDNA libraries is a powerful method (Werner et al., 1995; Sägesser et al., 1997), which has helped to identify proteins interacting with the whole viroid RNA genome (Martínez de Alba et al., 2003), but which can potentially also be used for proteins interacting with subdomains of the viroid RNA, stabilized RNA structures or short tertiary RNA structure motifs. However, direct screening methods identify interactions in another biological system and must be verified by biochemical or immunological methods to confirm whether they indeed exist in planta.

Martínez de Alba et al. (2003) have used the direct screening method for isolation of PSTVd RNA-binding proteins. One of these proteins, Viroid-binding protein 1 (Virp1), contains nuclear localization signals, an RNA-binding domain and a bromodomain (a protein domain found in many chromatin remodelling factors). It has been suggested that bromodomain proteins interact with other proteins in vivo; they bind to histones containing acetylated lysines. It has also been proposed that they keep themselves and their interacting partners in the vicinity of chromatin, even at phases where the nuclear membrane does not exist (during cell division); they have the ‘chromatin association mark’ (Yang, 2004). The carboxy-terminal domain of Virp1, which contains the RNA-binding domain, was shown to interact specifically with the right terminal domain of the viroid RNA in vivo and in vitro (Maniataki et al., 2003; Martínez de Alba et al., 2003). A dual domain motif was proposed to be responsible for the interaction (Gozmanova et al., 2003). Indeed, mutagenesis of one of the motifs resulted in reduction of the binding interaction and mutagenesis of both motifs destroyed the interaction down to 5%. All mutants were not infectious in mechanical inoculation assays or reverted to wild-type PSTVd sequence. Kalantidis et al. (2007) recently showed that Nicotiana benthamiana plants in which expression of Virp1 is suppressed cannot be infected with PSTVd when this is mechanically inoculated. Interestingly, protoplasts from these plants were also not infected with PSTVd RNA, indicating that Virp1 has a major role in viroid infection at the cellular level. A Virp1–GFP fusion protein was found to be localized in the nucleus, thus Virp1 is a candidate for translocating with the viroid RNA in the nucleus. Whether Virp1 has also a direct role in replication (e.g. assisting DNA-dependent polymerase II in replication) or whether it has an additional role in RNA-mediated methylation, or export of viroid RNA from the nucleus, is not known (Fig. 2B). From a plant pathology view, Virp1 is a candidate resistance gene for nuclear-replicating viroids. As N. benthamiana plants in which Virp1 is suppressed, when grown in tissue culture and in the greenhouse, did not show any obvious phenotype, it will be interesting to test whether Virp1 orthologue genes can be suppressed in crop plants for creating transgenic plants resistant to viroids. Other host genes conferring resistance to viroids when suppressed or expressed have not been described and no classical ‘resistance genes’ (R-genes) have been described so far for these RNA pathogens.

Using immunoprecipitation experiments, Pallás' and Owens' groups showed that phloem protein 2 from cucumber (CsPP2) is in a complex with HSVd RNA (Gómez and Pallás, 2001; 2004; Owens et al., 2001) and can be translocated with the RNA as an infectious entity in grafting experiments between host and non-host plants. The analysis of the primary structure of CsPP2 revealed the existence of a potential double-spaced RNA-binding motif, previously identified in a set of proteins that bind to highly structured RNAs. It was proposed that this phloem protein is assisting the long-distance movement of the viroid RNA within the plant, and that this type of interaction might operate in other viroid/host combinations. It will be interesting to test whether downregulation of this protein inhibits systemic viroid translocation.

Ribonucleoprotein complexes were isolated and characterized from Avocado sunblotch viroid (ASBVd)-infected plants. In vivo cross-linking and tandem mass spectrometry has been used to analyse the components of the most abundant cross-linked protein species on ASBVd RNA. Two closely related chloroplast RNA-binding proteins (PARBP33 and PARBP35) have been identified. Members of this protein family are involved in stabilization, maturation and editing of chloroplast transcripts. PARBP33 behaves as an RNA chaperone that stimulates in vitro the hammerhead-mediated self-cleavage of the multimeric ASBVd transcripts, indicating that this reaction, despite its RNA-based catalysis, is assisted by proteins (Darós and Flores, 2002).

Mechanism of systemic trafficking of viroid RNAs

Plants are ‘supracellular’ organisms, in which groups of cells are connected by special structures called plasmodesmata. Viroids are a mobile, systemically spreading RNAs; thus, they can be excellent models to study RNA signalling throughout the plant vasculature (Ding and Itaya, 2007; Kalantidis et al., 2008).

PSTVd mutants with N. benthamiana acting as a host served in two studies which elucidated steps of systemic trafficking of this viroid. In the first study, the difference between two viroid mutants was examined, using in situ hybridization of infected plants, and transgenic N. benthamiana plants expressing the corresponding viroid RNAs. Mutant PSTVd-NT arose from transgenic tobacco plants expressing PSTVd KF440-2, a viroid that infects tomato, but does not replicate in tobacco (Wassenegger et al., 1996). In contrast to KF440-2, PSTVd-NT replicates in tobacco, although at lower levels. Mutant PSTVd-NB arose from PSTVd-NT, and replicates at even higher rates in tobacco. In situ hybridization revealed that PSTVd-NT remains restricted to phloem cells in systemic leaves, while PSTVd-NB is capable of exit from phloem cells in systemically infected leaves and enters the mesophyl of newly infected young leaves. Subsequent mutational analysis of the five different nucleotides between strain PSTVd-NT and PSTVd-NB revealed that the ‘systemic trafficking’ motif consisted of two parts, with mutations lying in the ‘pathogenicity’ and the terminal right domains (Qi et al., 2004). Using biolistic infection on epidermal cells of young ‘sink’ leaves, Qi et al. concluded that the systemic trafficking of the strain is directional, from the bundle sheath to the mesophyll cells and not vice versa.

Zhong et al. (2007) created a number of PSTVd mutants which were defective in systemic trafficking. One mutant differs in only two nucleotides from the normal systemically infecting PSTVd strain, at positions U43 and C318. If one of the nucleotides is altered to create from the wild-type bulge a helix, systemic trafficking is lost and the RNA cannot cross the boundary between phloem parenchyma (PP) and bundle sheath cells (BS) (see also Fig. 1B and C). This group analysed the tertiary structure and geometry of this region of the strains, comparing them with other mutants and other viroids. They showed that this motif is, in its tertiary structure, conserved in the genus Pospiviroid, and that all systemically infecting mutants belong to an equivalent isostericity group. They designed specific mutations for isosteric structures and predicted correctly the behaviour of these strains concerning systemic spread. It will be interesting to test whether addition of this motif alone is capable of changing an unrelated RNA molecule into a systemic trafficking RNA, and how large this RNA signal finally is. However, a recent report from the same laboratory revealed that the mechanism of systemic movement is more complicated than anticipated and that more structural elements of the rigid PSTVd RNA rod-like structure are important for systemic spread (Zhong et al., 2008).

Interaction of viroids with the host's defence mechanisms and the interrelation with replication

The mechanism of coordinated sequence-specific RNA degradation was first proposed as a model in studies of transgenic plants expressing viral genes (Lindbo and Dougherty, 2005). It was also recognized early that RNA-mediated mRNA degradation is an antiviral mechanism in higher eukaryotes (Baulcombe, 2004; 2006). It was therefore interesting to study whether viroids, RNA pathogens replicating in the nucleus or in the chloroplasts of plants, would be targets and triggers of this defence mechanism, which is predominantly cytoplasmic. The overall structural similarity of viroids with the precursor transcripts of miRNAs (the pre-miRNAs), a field emerging at the same time, was recognized early (Maniataki et al., 2003). Two questions arose: are viroids substrates for degradation by the silencing machinery and is this interaction responsible for pathogenicity?

Viroid infection results in production of viroid-specific siRNAs

Early work showed that nuclear and (surprisingly) also chloroplastic viroids are targets of the silencing machinery and that viroid-specific siRNAs are produced after infection (Itaya et al., 2001; Papaefthimiou et al., 2001; Martínez de Alba et al., 2002). In these first reports, there was no correlation observed between amount of siRNAs and severity of the viroid strain used. Interestingly, the siRNAs of PSTVd were found to be localized predominantly in the cytoplasmic fraction (Denti et al., 2004). Later, it was found that blotched leaf areas contain much more siRNAs of ASBVd than apparently symptomless areas, and that CEVd isolates, which differed in pathogenicity, differed also in their siRNA abundance in the host plant Gynura aurantiaca (Markarian et al., 2004). These differences might be due to the host/viroid systems used, or to subtle differences in the experimental set-up.

Itaya et al. (2007) characterized the viroid-specific siRNAs by cloning them and found that for the genomic RNA of (+) polarity there are mainly three ‘hot spots’ of siRNAs but that the (−) polarity siRNAs are distributed more or less all over the genome of PSTVd. This implies that the predominant RNA substrate for siRNA production is most probably not a true double stranded viroid RNA, but the genomic (+) RNA. The (−) RNA is to a lesser extent a substrate, possibly as a replicative intermediate. Interestingly, there were nearly no siRNAs found from the so-called ‘pathogenicity’ region, which is surprising, because mutations in this region are known to induce changes in pathogenicity on tomato (Schnölzer et al., 1985). siRNAs were produced also in plant protoplasts of N. benthamiana, indicating that the degradation mechanism does not need the ‘supracellular’ structure of the whole plant. Using GFP–PSTVd fusion sequences, Itaya and co-workers could show that PSTVd siRNAs are active in cleaving target RNAs in RISC complexes, but the stable secondary structure of PSTVd RNA resists RISC-mediated cleavage. Parallel to this, Gómez and Pallás (2007) have shown that circular forms of HSVd are not decreased in amount when translocating through a grafted transgenic tissue which produces HSVd siRNAs.

Machida et al. (2007) have shown that viroid siRNAs of 21 nucleotides length precede the appearance of siRNAs of 24 nucleotides length in the examined period of 45 days and are consistently present throughout this time. The siRNA profile may, however, differ dramatically from leaf to leaf if a single plant is examined. The siRNA profile of averaged samples changed during infection from size-specific, to heterogeneous and finally again to more size-specific. They identified an additional ‘hot spot’ of siRNA production derived from the complementary (−) RNA of PSTVd, taking samples from a single leaf, which may explain why this ‘hot spot’ was not detected by Itaya et al. who had (most probably) used ‘averaged’ samples for cloning.

Martin et al. (2007) found that viroid siRNAs were phosphorylated at the 5′-terminus and protected at the 3′-terminus, similar to genuine endogenous plant-derived siRNAs and miRNAs. Landry and Perreault (2005) have shown that (+) and (−) full-length forms of the PLMVd can be substrates for DCL(s) activity in wheat germ extract. A minimal length is required for DCL binding and activity and the best substrate contained a perfect dsRNA helix.

Viroid interaction with components of the RNA-silencing mechanism might contribute to symptom appearance

The discovery that viroids interact with the silencing machinery (as targets and triggers) led to the hypothesis that viroid-specific siRNAs might be responsible for symptoms caused by viroids, via RISC-mediated translational inhibition, or cleavage of specific host mRNAs containing short stretches of homologies (Tabler and Tsagris, 2004; Flores et al., 2005b). Viroid-derived siRNAs are loaded on RISC complexes and might inhibit the expression of specific host genes. Wang et al. (2004) have shown that expression of dsRNA derived from PSTVd sequences from nucleotides 16–359 can cause viroid-like symptoms in tomato. This shows that in this host/viroid combination, PSTVd siRNAs may be the triggers of pathogenicity, and that this is independent of viroid replication. Also the results of Carbonell et al. (2008) indicate the existence of viroid-specific RISC complexes operating in infected plants, as symptoms were reduced when the viroids were co-inoculated with homologous dsRNAs and not with heterologous ones. Viroid symptom development can be very specific, depending on host and viroid genotype. This has been described for tomato/PSTVd (Schnölzer et al., 1985), tobacco/PSTVd (Qi and Ding, 2003), peach/PLMVd (peach calico) (Malfitano et al., 2003) and other host–viroid combinations. In peach calico, it has been shown that a specific hairpin structure of PLMVd is responsible for the symptoms observed (Rodio et al., 2006; 2007). In all these cases, a RISC complex loaded with a specific viroid-derived siRNA might be the cause of the symptoms.

Symptoms might be induced if viroid replication or transport recruits limiting factors of the RNA-silencing machinery, thus interfering with miRNA (and/or endogenous siRNA) biogenesis and function. Martin et al. (2007) have investigated this possibility; however, they could not find indications of this type of general interference. Gas et al. (2007) have proposed that a Dicer-like RNase might be involved in viroid processing.

Future directions

A possible role for RNA–RNA transcription performed by DNA-dependent RNA polymerases

Viroids are used as templates for DNA-dependent RNA polymerase II (reviewed in Sänger and Tabler, 1987) and it has been shown that T7 phage DNA-dependent RNA polymerase can use RNA as a template too (Konarska and Sharp, 1989). DNA-dependent RNA polymerase II can use its own transcript of HDV RNA as a template to produce small hairpin-shaped longer RNA transcripts (Filipovska and Konarska, 2000). In Fig. 2A, the generation of such self-complementary short RNA transcripts is shown. Production of such ‘aberrant’ transcripts originating from this type of reaction might be inhibitory to the normal DNA-dependent RNA transcription of the polymerases and in this case it should be suppressed in the cell. Another scenario could be that this type of reaction has a biological role. As genome-wide low-level basal transcription has been shown to occur (Kapranov et al., 2007) and the fast biogenesis of miRNAs (frequent birth and death of miRNA genes) has also been described (Fahlgren et al., 2007), one could consider the possibility that de novo RNA–RNA-dependent transcription of short hairpin transcripts is an underestimated part of the whole genome transcriptome. This type of transcript has a structural similarity to the miRNA precursor transcripts, which have a length of c. 70–80 nucleotides, an imperfect stem-loop secondary structure and look similar to the terminal parts of viroids (see also Fig. 2A). Such low-level ‘error-prone’ RNA–RNA transcription, in combination with reverse transcription, possibly contributes to fast adaptation of organisms with novel sequences. A third possibility would be that small, ‘aberrant’ RNA transcripts derived from DNA-dependent RNA polymerases play a role in RNA signalling in the cell. Viroid replication can serve as a model for elucidation of the mechanism of RNA-templated RNA transcription.

Interaction of viroids with the silencing mechanism

Many interesting questions concerning the interaction of viroids with the silencing mechanism remain open: How (and where) do the nuclear and chloroplastic viroids interfere with the (mostly cytoplasmic) operating RNA-silencing machinery? Are viroid siRNAs produced in the cytoplasm in the (short) transit phase of viroids during cell-to-cell transport? Or are viroids targeted in phases where nuclei are disrupted (cell division) or chloroplasts are first formed and increasing in size (during early development of meristematic and organ precursor tissues)? In this context it is interesting to mention that biolistic infection of cotyledons with PSTVd led to a complete ‘stalling’ of development and severe stunting of N. benthamiana infected with PSTVd (which usually is a symptomless host (Matousek et al., 2007). Which enzymes are involved in the production of viroid siRNAs and is there a time frame or tissue tropism? Are viroid siRNAs products of DCL-1 (which is in the nucleus and accepts ‘imperfect’ hairpins as substrates) and/or other DCL proteins (which are generally thought to accept perfect dsRNA as substrate), and how does this correlate with their intracellular localization? An interaction between the (nuclear) replicating viroids and the silencing machinery can be envisaged also at the level of pericentromeric repeat regulation. Pericentromeric repeats have conserved elements (i.e. retrotransposons and similar repeat elements, repeats of 5S rDNA transcriptional units), but also contain spacers which are species and cultivar specific. Pericentromeric repeats are silenced through the action of siRNAs from these sequences. These siRNAs require the activity of an enzyme, called DNA-dependent RNA polymerase IV, which is a fourth DNA-dependent RNA polymerase, unique to plants (Baulcombe, 2006). This protein, together with RNA dependent RNA polymerase 2 (RDR2) and Argonaute 4 (AGO 4), two other proteins involved in nuclear RNA silencing, leads to endogenous siRNA production, which in turn methylates the DNA and remodels the chromatin for silencing (Henderson and Jacobsen, 2007). It is possible that nuclear viroid replication either interferes with steps of this regulation or else uses any of its factors for own purposes.

The two models of interference of viroid RNA with this regulatory and defence machinery of the host are not mutually exclusive: the general interference of the structured viroid RNA with components of the silencing and microRNA machinery could coexist with the very specific interactions of viroid strain X with host genotype Y through a RISC loaded with a viroid-specific siRNA. The possibility that viroid RNA uses any of the proteins involved in silencing in order to complete its replication or spread (as suggested by (Gas et al., 2007)) is the other side of this interesting interaction: the viroid needs components of the silencing machinery, but it is also attacked by the same machinery.

Acknowledgements

We thank all our former collaborators who have been working in the viroid field: P. Arabatzi, M.A. Denti, K. Karademiris, E. Maniataki, E. Marinou, I. Papaefthymiou, A. Prombona, R. Sägesser, E. Stylianou and rotation students. We thank S. Tzortzakaki and M. Providaki for excellent technical assistance. We apologize to colleagues whose work was not cited due to space limitations. We are grateful to Dr Kathryn Melzak for editing the English of the manuscript. This review was supported by University of Crete Grant No. 2078-2192-2161 and IMBB Grant No. 005120 (FOSRAK). For Martin.

Note added in proof

While this article was in proof, the following related articles were published:

Abraitiene, A., Zhao, Y., and Hammond, R. (2008) Nuclear targeting by fragmentation of the Potato spindle tuber viroid genome. Biochem Biophys Res Commun368: 470–475.

Di Serio, F., and Flores, R. (2008) Viroids: Molecular implements for dissecting RNA trafficking in plants. RNA Biol5: in press. http://www.landesbioscience.com/journals/rnabiology/article/6638

Ancillary