A Short Double-Stranded RNA Motif of Peach Latent Mosaic Viroid Contains the Initiation and the Self-Cleavage Sites of Both Polarity Strands

The transcription initiation sites of viroid RNAs, despite theirrelevance for replication and in vivo folding, are poorly characterized.Here we have examined this question for Peach latent mosaicviroid (PLMVd), which belongs to the family of chloroplasticviroids with hammerhead ribozymes (Avsunviroidae), by adaptingan RNA ligase-mediated rapid amplification of cDNA ends methodologydeveloped for mapping the genuine capped 5' termini of eukaryoticmessenger RNAs. To this aim, the characteristic free 5'-triphosphategroup of chloroplastic primary transcripts from PLMVd-infectedyoung fruits was previously capped in vitro with GTP and guanylyltransferase.PLMVd plus and minus initiation sites map at similar double-strandedmotifs of 6 to 7 bp that also contain the conserved GUC tripletpreceding the self-cleavage site in both polarity strands. Withinthe branched secondary structures predicted for the two PLMVdstrands, this motif is located at the base of a similar longhairpin that presumably contains the promoters for a chloroplasticRNA polymerase. The transcription templates could be the circularviroid RNAs or their most abundant linear counterparts, assumingthe involvement of an RNA polymerase able to jump over templatediscontinuities. Both PLMVd initiation sites were confirmedby applying the same methodology to two purified PLMVd subgenomicRNAs and by primer extension, and they therefore likely reflectthe in vivo situation. The location of the PLMVd initiationsites provides a mechanistic view into how the nascent strandsmay fold and self-cleave during transcription. The approachdescribed here may be extended to other chloroplastic RNA repliconsand transcripts accumulating at low levels.

INTRODUCTION

Top

Introduction

Viroids represent the lowest step of the biological scale. Theirgenome is composed exclusively by a small (246 to 401 nucleotides[nt]) circular RNA without any apparent protein-coding capacity(16, 20, 21, 49), a key aspect that sets a demarcating differencewith virus genomes, which do code for one or more proteins oftheir own. Despite their structural simplicity, viroids areable to replicate autonomously in their host plants and to elicitsymptoms in most instances. Therefore, viroids offer uniqueopportunities for studying RNA replication without interferencewith the concurrent transcription and translation accompanyingthe replication of viral RNAs.

Viroids replicate through a rolling-circle mechanism, with twoalternative pathways mediated by three catalytic activities:RNA polymerase, RNase, and RNA ligase (6). The most abundantviroid circular RNA, arbitrarily considered as having a pluspolarity, is transcribed into oligomeric minus strands that,by themselves (asymmetric pathway) or after processing to theirmonomeric circular counterparts (symmetric pathway), serve astemplates for a second RNA-RNA transcription round leading tooligomeric plus strands that are finally cleaved and ligatedto the monomeric plus circular forms. There is empirical evidencesupporting the asymmetric pathway for Potato spindle tuber viroid(14, 25) and for other nuclear viroids of the family Pospiviroidae(6, 7, 18, 30, 44) and the symmetric pathway for Avocado sunblotchviroid (ASBVd) (48) and other chloroplastic viroids of the familyAvsunviroidae (8, 11, 28). Remarkably, cleavage in members ofthe family Avsunviroidae is mediated by hammerhead ribozymes(22, 29, 43) embedded in both polarity strands (17, 19).

Viroids do not necessarily require a transcription startingfrom specific initiation sites, because reiterative copyingof their circular genomes would ensure complete transcription.Although this may have represented an advantage in the primitiveprecellular RNA world-of which viroids are considered "molecularfossils"—by precluding the need of genomic tags (15),the available evidence indicates that in the present cellularhabitat, the polymerization of viroid strands starts at definedsites. Specifically, the labeling of the free 5'-triphosphategroup characteristic of chloroplastic primary transcripts with[ ${alpha}$ -³²P]GTP and guanylyltransferase (in vitro capping), togetherwith RNase protection assays, has mapped the initiation of ASBVdplus and minus RNAs isolated from infected avocado at similarA+U-rich terminal loops in their predicted quasirodlike secondarystructures (38). Data on the initiation sites for the familyPospiviroidae, obtained by in vitro transcription of the Potatospindle tuber viroid monomeric plus circular RNA either witha potato nuclear extract or with purified RNA polymerase IIfrom wheat germ and tomato, are restricted to the minus polaritystrand and are not coincidental (49). This discrepancy may inpart reflect the difficulties in reconstituting in vitro aninitiation complex reproducing the in vivo situation. Thereare also data on the initiation sites of a second member ofthe family Avsunviroidae, Peach latent mosaic viroid (PLMVd)(27), derived from primer-extension analysis of the 5' terminiof certain viroid subgenomic RNAs isolated from infected tissuethat are presumed to be replication byproducts and from in vitrotranscriptions with truncated PLMVd RNAs and the RNA polymeraseof Escherichia coli (41). However, the presence of free 5'-triphosphategroups in the subgenomic RNAs was not examined, and the problemsinherent in reconstituting a bona fide initiation complex invitro are here enhanced by using a eubacterial RNA polymerasethat may have a specificity very different from that of itshomologous chloroplastic RNA polymerase (PEP, from plastid-encodedpolymerase) (47). Moreover, a second RNA polymerase class existsin chloroplasts (NEP, from nuclear-encoded polymerase) (1),with the available evidence suggesting that it is an NEP-likeenzyme that is involved in ASBVd replication (39).

Here we report a reassessment of the PLMVd initiation sitesbased on mapping the 5'-triphosphate termini of PLMVd plus andminus RNAs isolated from infected peach tissue. Our resultshave identified another initiation site for each PLMVd strandlocated in the vicinity of the self-cleavage site of its complementarystrand. The quasisymmetrical distribution of both initiationsites and, particularly, their close proximity to the self-cleavagesites are intriguing and indicate that the 6- to 7-base pair(bp) RNA motif in which they are embedded is involved in a dualfunction.

MATERIALS AND METHODS

Top

Materials and Methods

RNA extraction, partial purification, and Northern blot hybridization. Nucleic acids from young expanding fruits of a PLMVd-infectedpeach tree (Prunus persica Batsch cv. ‘Springcrest’)were extracted (12) and fractionated on nonionic cellulose (CF11;Whatman) with STE (50 mM Tris-HCl, pH 7.2, 100 mM NaCl, 1 mMEDTA) containing 35% ethanol (38). Control preparations fromhealthy peach fruits and from gynura leaves (Gynura aurantiacaDC) infected by Citrus exocortis viroid (CEVd) were obtainedfollowing the same protocol. RNAs were separated by denaturingpolyacrylamide gel electrophoresis (PAGE) in 5% gels containing1x TBE (Tris-borate-EDTA) and 8 M urea (45), electroblottedto nylon membranes, and hybridized with strand-specific riboprobes.For preparative purposes, following electrophoresis the gelwas stained with ethidium bromide, and the monomeric linearand subgenomic RNAs, identified with appropriate markers, werecut and eluted.

Analysis by primer extension. Different PLMVd-specific primers (Table 1) were used to determinethe 5' ends of plus and minus PLMVd RNAs by primer extensionassays. Primers (100 ng) were annealed to purified linear PLMVdRNAs (100 ng) in sterile water at 80°C for 3 min and snap-cooledon ice. Extension reactions (in 20-µl final volumes) werecarried out in 50 mM Tris-HCl, pH 8.3, containing 75 mM KCl,3 mM MgCl₂, 10 mM dithiothreitol (DTT), 1 mM each of dATP, dGTP,and dTTP, 0.1 mM of dCTP supplemented with 3 µCi of [ ${alpha}$ -³²P]dCTP(3,000 Ci/mmol; Amersham), 125 µg/ml actinomycin D, and200 U of reverse transcriptase (SuperScript II RNase H^–;Invitrogen). After incubation at 42°C for 1 h, the reactionmixture was heated at 70°C for 15 min and the cDNAs wereanalyzed by denaturing PAGE in 6% sequencing gels. The precisesizes of the extension products were determined by running themin parallel sequence ladders obtained with the same primer anda recombinant plasmid with a complete PLMVd-cDNA insert.

View this table:
[in this window]
[in a new window]
TABLE 1. Primers used for extension analysis and RLM-RACE

In vitro capping. The two subunits of the vaccinia virus mRNA-capping enzyme (guanylyltransferase)were coexpressed in bacteria after proper modification by additionsof N-terminal histidine tags for affinity purification by Ni-agarosechromatography (QIAGEN) (33). CF11-fractionated RNA preparations(3 µg) from healthy and PLMVd-infected peach tissues and,when indicated, from individual RNAs eluted from denaturinggels were in vitro capped in a reaction mixture (30-µlfinal volume) containing 50 mM Tris-HCl, pH 7.9, 1.25 mM MgCl₂,6 mM KCl, 2.5 mM DTT, 10 U of human placental RNase inhibitor(HPRI) (Roche Applied Science), 0.3 mM GTP, and 2.5 µlof a purified guanylyltransferase preparation (which in previousexperiments was sufficient to efficiently label with [ ${alpha}$ -³²P]GTP200 ng of a PLMVd synthetic transcript; data not shown). Followingincubation at 37°C for 45 min, the addition of the sameamount of fresh enzyme, and incubation for another 45 min at37°C, the RNAs were extracted with phenol-chloroform, recoveredby ethanol precipitation, and resuspended in sterile distilledwater.

RNA ligase-mediated rapid amplification of cDNA ends (RLM-RACE). After in vitro capping, the RNAs were denatured (95°C for1.5 min) and incubated for 30 min at 50°C with 10 U of calfintestinal phosphatase (CIP) (Roche Applied Science) in a reactionmixture (50-µl final volume) containing 50 mM Tris-HCl,pH 8.5, 0.1 mM EDTA, and 10 U of HPRI. Following denaturation(95°C for 1.5 min), the incubation was repeated with freshCIP and HPRI, and the RNAs were extracted with phenol-chloroformand recovered by ethanol precipitation (this deproteinatingtreatment was repeated after each subsequent enzymatic step).For decapping, the RNAs (in a final volume of 10 µl) wereincubated for 1 h at 37°C with 0.5 U of tobacco acid pyrophosphatase(TAP) (Epicenter Technologies) in 50 mM sodium acetate, pH 6.0,containing 1 mM EDTA, 0.1% ß-mercaptoethanol, 0.01%Triton X-100, and 10 U of HPRI. For ligation, the decapped RNAswere mixed with 250 ng of an RNA adaptor (5'-CGACUGGAGCACGAGGACACUGACAUGGACUGAAGGAGUAGAAA-3')that was previously denatured at 65°C for 5 min and incubatedfor 1 h at 37°C with 10 U of T4-RNA ligase (10-µlfinal volume) in the buffer recommended by the supplier (RocheApplied Science) supplemented with 1 mM ATP and 10 U of HPRI.For reverse transcription (RT), the ligated RNAs (500 ng) weremixed with 100 ng of the corresponding complementary primer,heated at 85°C for 5 min, and incubated for 1 h at 42°Cin a reaction mixture (20-µl final volume) containing50 mM Tris-HCl, pH 8.3, 75 mM KCl, 3 mM MgCl₂, 0.5 mM dNTP,10 mM DTT, 10 U HPRI, and 300 U of reverse transcriptase. One-tenthof the RT reaction volume was used for PCR amplification with0.4 mM each of the dNTPs, 400 ng of each primer, and 1 U ofExpand High Fidelity DNA polymerase in the buffer recommendedby the supplier (Roche Applied Science). After initial denaturationat 94°C for 2 min, the amplification profile consisted of30 cycles at 94°C for 30 s, 55 to 65°C for 30 s, and72°C for 2 min. In some cases, a second (nested) PCR amplificationwas performed using 1 µl of a 1/100 dilution of the firstPCR mixture and a pair of internal primers.

RESULTS

Top

Results

PLMVd plus and minus monomeric linear RNAs from infected tissuehave complex populations of 5' termini. PLMVd RNAs accumulatein vivo at amounts considerably lower than those of ASBVd RNAs(27). Since this could represent a limiting factor in the characterizationof the PLMVd initiation sites, we performed some preliminaryexperiments aimed at identifying the plant organ with the highestcontent of viroid RNAs. This organ proved to be the young expandingfruits, for which analysis of four different extracts by denaturingPAGE and Northern blot hybridization with strand-specific riboprobesrevealed a reproducible series of PLMVd RNAs of both polarities(Fig. 1). The specificities of the signals were shown by thelack of hybridization of the two riboprobes with CEVd, an unrelatednuclear viroid (21) and of each one with the PLMVd in vitrotranscript of the same polarity (Fig. 1, lanes 1, 2 and 5).In agreement with previous results (8), the concentrations ofthe monomeric linear plus and minus forms were significantlyhigher than those of their circular counterparts, with the predominantband being generated by the monomeric linear plus RNA; no larger-than-unitPLMVd RNAs could be detected, but several subgenomic RNAs wereclearly visible (Fig. 1). The pattern of these viroid-specificRNAs was analogous in PLMVd-infected leaves collected at differentdevelopmental stages, but their concentration was significantlylower than in young fruits (data not shown). RT-PCR amplification,cloning, and sequencing revealed that variants of the PLMVdisolate here used had sequences similar to that of the referencevariant (2, 27).

View larger version (45K):
[in this window]
[in a new window]
FIG. 1. RNA analysis by denaturing PAGE and Northern blot hybridization with full-length radioactive riboprobes for detecting PLMVd plus (A) and minus (B) strands. Lanes 1 and 2, primary transcripts and their self-cleavage products resulting from in vitro transcription of a recombinant plasmid containing a dimeric head-to-tail PLMVd-cDNA insert (numbers on the left refer to their sizes in nucleotides). Lanes 3, 4, 6, and 7 show CF11-fractionated RNA preparations from different PLMVd-infected peach fruits. Lane 5, CF11-fractionated RNA preparation from CEVd-infected gynura leaves. The positions of the PLMVd monomeric circular (C) and linear (L) forms and of certain viroid subgenomic (SG) RNAs are indicated on the right. The acid-precipitable counts of riboprobes in panels A and B were the same, and the exposure time for panel B was three times that for panel A.

We anticipated that a fraction of the PLMVd monomeric linearforms should have the characteristic 5' termini resulting fromthe hammerhead-mediated self-cleavage of both polarity strands.To confirm this point and to get insight into how complex thepopulation of 5' termini was, the PLMVd monomeric linear formswere eluted and reverse transcribed with the minus polarityprimers RF-43 and RF-132. Whereas RF-43 led to six cDNAs, RF-132generated only one well-defined cDNA (Fig. 2A, lanes L). Thecorresponding 5' termini mapped at residues U290, C3, C22, C51,C80, and A113 of the PLMVd plus RNA and are referred to as termini1, 2, 3, 4, 5, and 6, respectively. Only the cDNA with a lengthconsistent with terminus 1 was clearly visible with primer RF-132,most likely because of the relative positions and abundancesof the other termini. Extension controls using as a templatethe linear monomeric PLMVd plus RNA that results from self-cleavageduring in vitro transcription of the linearized pPL30 plasmidcontaining a dimeric viroid-cDNA insert identified only terminus1 (Fig. 2A, lanes SC). A functional origin could thus be establishedexclusively for terminus 1, the most represented in the populationof the monomeric plus linear RNAs isolated from PLMVd-infectedtissue, with the other 5' termini probably reflecting sitesin the monomeric plus circular RNA particularly sensitive todegradation or the specific site at which the transcriptionof PLMVd plus RNAs starts.

View larger version (32K):
[in this window]
[in a new window]
FIG. 2. (A) Primer extension analysis with primers RF-43 and RF-132 using as templates the natural monomeric linear PLMVd RNAs (lanes L) or the PLMVd plus monomeric linear RNA generated by self-cleavage during in vitro transcription of a linearized recombinant plasmid with a dimeric PLMVd-cDNA insert (lanes SC). (B) Primer extension analysis with primers RF-44 and RF-26 using as templates the natural monomeric linear PLMVd RNAs (lanes L), or the PLMVd minus monomeric linear RNA generated by self-cleavage during in vitro transcription of a linearized recombinant plasmid with a dimeric PLMVd-cDNA insert (lanes SC). In both panels, the sequencing ladders were obtained with the same primers and recombinant plasmids containing a monomeric PLMVd-cDNA insert. For panels A and B, the positions in the 6% sequencing gel of the extension products produced by primers RF-43 and RF-132 are indicated with Arabic numerals, and the positions of the products produced by RF-44 and RF-26 are indicated with Roman numerals. (C) Locations of the 5' termini mapped by primer extension on the secondary structure predicted for the plus strand of the PLMVd reference variant (2, 27); there are minor sequence discrepancies, because the sequencing ladders derive from a PLMVd variant slightly different from the reference variant. Note that Roman numerals refer to the same positions in the PLMVd minus strand. Continuous lines denote positions to which RF-43 and RF-132 are complementary, and discontinuous lines denote the sequences of RF-26 and RF-44. Sequences involved in PLMVd plus and minus self-cleaving domains are delimited with flags, the 13 nucleotides conserved in most natural hammerhead structures are denoted by bars, and the self-cleavage sites are shown with arrows. Black and white symbols refer to plus and minus polarities, respectively. Dashed lines between positions GCGG178 to 181 and CCGC211 to 214 indicate a pseudoknot (9).

Parallel extension experiments with the plus polarity primersRF-44 and RF-26 revealed four 5' termini mapping to residuesU66, U48, G3, and U286 of the PLMVd minus RNA (numbers correspondto the plus polarity), referred to as termini I, II, III, andIV, respectively (Fig. 2B, lanes L). Control extensions withthe same primers and the linear monomeric PLMVd minus RNA resultingfrom self-cleavage during in vitro transcription of the linearizedpPL30 plasmid identified U48 as the self-cleavage site of PLMVdminus strands (Fig. 2B, lanes SC). Therefore, any of the remaining5' termini could correspond to the initiation site of PLMVdminus RNAs.

Setting up a sensitive methodology for mapping the initiation sites of PLMVd RNAs. Because chloroplast primary transcripts possess a 5'-triphosphategroup that can be specifically capped in vitro with guanylyltransferaseand GTP (3) and because PLMVd RNAs have been localized in chloroplasts(8), we first attempted to label the monomeric PLMVd linearRNAs with a 5'-triphosphate isolated from PLMVd-infected tissue.However, even with RNA preparations from young fruits, the radioactivesignal was too faint for applying the RNase protection strategy(38) used previously for identifying the initiation sites ofASBVd RNAs (data not shown). Therefore, we focused our attentionon more-sensitive approaches and, particularly, on an RLM-RACEmethodology developed for mapping the genuine 5' termini ofeukaryotic messenger RNAs, which is based on a specific property:such termini are capped (35, 46). We reasoned that by introducingan additional step of in vitro capping, this strategy couldbe adapted for mapping the initiation sites of PLMVd RNAs. Figure3 summarizes the sequential steps. Only RNAs with a 5'-triphosphateare specifically capped in vitro with guanylyltransferase andGTP; other RNAs, such as those resulting from processing ordegradation, have a 5'-monophosphate or a hydroxyl group notsusceptible to in vitro capping (13, 36). Subsequent digestionwith CIP removes the 5'-monophosphate groups and impedes theligation of these RNAs to an RNA adaptor. Further treatmentwith TAP eliminates the cap structure of the genuine 5' terminiand leaves a free 5'-monophosphate suitable for ligation tothe RNA adaptor with T4-RNA ligase. The resulting product isthen reverse transcribed with a viroid-specific primer, andthe cDNA obtained was PCR-amplified (with primers specific forthe viroid and the RNA adaptor), cloned, and sequenced. To testits reliability, this methodology was applied to two artificialmixtures formed by 10 ng of an in vitro PLMVd plus or minustranscript, therefore containing 5'-triphosphate termini, combinedwith an excess of a nucleic acid preparation from healthy peach,which were used to simulate the situation existing in PLMVd-infectedtissue. After completing the enzymatic treatments, RT-PCR productswith the lengths predicted for both PLMVd transcripts were obtained(Fig. 4A and B, lanes 2), and cloning and sequencing confirmedthat they had the expected 5' termini (data not shown). In contrast,two controls in which the incubations with guanylyltransferaseor the RNA adaptor were omitted failed to generate the sameRT-PCR products (Fig. 4A and B, lanes 3 and 4, respectively),thus showing that the experimental approach developed was sensitiveand specific.

View larger version (10K):
[in this window]
[in a new window]
FIG. 3. Scheme of the adapted RLM-RACE strategy used for mapping the transcription initiation sites of PLMVd. RNAs with a 5'-triphosphate are specifically capped in vitro with guanylyltransferase (GT) and GTP, while other RNAs with a 5'-monophosphate or a hydroxyl group are not susceptible. The digestion of the 5'-monophosphate groups of noncapped RNAs with CIP and the removal of the cap structure with a subsequent treatment with TAP leaves in the 5' terminus of the primary transcript a 5'-monophosphate suitable for ligation to the RNA adaptor with T4-RNA ligase. The resulting product is retrotranscribed with RT and a viroid-specific primer, with the cDNA then being PCR amplified (with primers derived from the viroid and the RNA adaptor), cloned, and sequenced. In some instances, a second (nested) PCR amplification was performed.

View larger version (37K):
[in this window]
[in a new window]
FIG. 4. PAGE analysis of the RT-PCR products resulting from the adapted RLM-RACE strategy (A to D) and localization of the 5' termini of the corresponding RNA templates on the secondary structure predicted for the plus strand of the PLMVd reference variant (E). (A and B) RT-PCR (single) products from plus RNA templates (using primer RF-129 for reverse transcription and primers RF-43 and RF-553 for PCR amplification) and minus RNA templates (using primer RF-26 for reverse transcription and primers RF-446 and RF-553 for PCR amplification). Lanes 1, DNA markers (100 bp and of multiples thereof). Lanes 2, artificial reaction mixtures containing in vitro PLMVd plus or minus transcripts starting at position 1 (the same numbers are used for both polarities) combined with excesses of a CF11-fractionated RNA preparation from healthy peach fruits. Lanes 3, as in lanes 2 but without the guanylyltransferase. Lanes 4, as in lanes 2 but without the RNA adaptor. Lanes 5, as in lanes 2 but without the in vitro PLMVd plus or minus transcript. Lanes 6, CF11-fractionated RNA preparation from PLMVd-infected peach fruits. (C and D) RT-PCR (nested) products from plus RNA templates (using primer RF-555 for reverse transcription, RF-129 and RF-553 for the first PCR amplification, and RF-45 and RF-544 for the second PCR amplification) and from minus RNA templates (using primer RF-37 for reverse transcription, RF-623 and RF-553 for the first PCR amplification, and RF-26 and RF-544 for the second PCR amplification). Lanes 1, DNA markers (100 bp and multiples thereof). Lanes 2, artificial reaction mixtures containing in vitro PLMVd plus or minus transcripts starting at position 1 combined with excesses of a CF11-fractionated RNA preparation from healthy peach fruits. Lanes 3, as in lanes 2 but without the guanylyltransferase. Lanes 4, as in lanes 2 but without the RNA adaptor. Lanes 5, as in lanes 2 but without the in vitro PLMVd plus or minus transcript. Lanes 6, CF11-fractionated RNA preparation from PLMVd-infected peach fruits. Lanes 7, reaction mixtures containing RNAs from healthy peach fruits eluted from the region of the denaturing gel corresponding to the PLMVd plus and minus subgenomic RNAs of approximately 240 nt. Lanes 8, reaction mixtures containing gel-eluted PLMVd plus and minus subgenomic RNAs of approximately 240 nt. Arrows on panels A to D denote the PLMVd-specific amplification products. Other details are as described in the legend to Fig. 2.

The initiation site of each PLMVd strand maps in the vicinity of the self-cleavage site of its complementary strand. Application of the RLM-RACE methodology to CF11-fractionatedRNA preparations from PLMVd-infected peach tissue led for eachviroid polarity strand to one specific RT-PCR product that wasabsent in parallel controls in which the CF11-fractionated RNApreparation was from healthy peach tissue (Fig. 4 A and B, lanes5 and 6). More specifically, an RT-PCR product of approximately200 bp was obtained for the plus polarity strand (Fig. 4A, lane6). Cloning and sequencing revealed that the 5' terminus ofthe corresponding PLMVd plus RNA mapped at position 51 (Fig.4E), which coincides with terminus 4, the major 5' terminusafter that produced by the hammerhead-mediated self-cleavage,in the primer-extension experiment (Fig. 2A). Thus, the resultsfrom both experimental approaches (RLM-RACE and primer extension)were mutually consistent. Intriguingly, this position 51 islocated only a few nucleotides away from the self-cleavage siteof the minus polarity strand (Fig. 4E).

The specific RT-PCR product obtained for the minus polaritywas of approximately 150 bp (Fig. 4B, lane 6). The 5' terminusof the corresponding PLMVd minus RNA mapped at position 286(Fig. 4E), which coincides with terminus IV of the primer-extensionexperiment (Fig. 2B). The results from both experimental approacheswere again mutually consistent, and position 286 is also locatedonly a few nucleotides away from the self-cleavage site of theplus polarity strand. Therefore, the transcription of each PLMVdRNA starts at almost symmetrical positions in the base of thelong hairpin that contains the sequences forming the hammerheadstructures, the so-called hammerhead arm (2). Because of theirhigh thermodynamic stability, the branched conformations ofminimal free energy predicted for both PLMVd polarity strandscontain the long hairpin or a cruciform variation thereof (2,27). Nuclease probing (9) and natural sequence variability (2,40) also support the existence of this long hairpin in vitroand in vivo, respectively.

Analysis of PLMVd subgenomic RNAs also confirms the proposed initiation sites. To provide further experimental support for these initiationsites, we explored the existence in vivo of PLMVd subgenomicRNAs with sizes consistent with their initiations at the proposedsites and subsequent self-cleavings at the sites predicted bythe corresponding hammerhead structures (before the first replicationround was completed). Examination of the patterns resultingfrom denaturing PAGE and Northern hybridization of RNA preparationsfrom PLMVd-infected fruits showed that, in addition to the prominentbands generated by the viroid linear and circular forms of bothpolarities, bands generated by PLMVd subgenomic RNAs of theexpected size (about 240 nt), also of both polarities, werediscernible (Fig. 1). This size is significantly different fromthat reported previously (about 280 nt) for what seem to besimilar or even the same RNAs (41). We ignore the reasons forthis discrepancy, but our size estimation appears reliable becausethe plus subgenomic RNA essentially comigrated with one of thesize markers of 239 nt (Fig. 1).

Next, to confirm that the $~$ 240-nt PLMVd subgenomic RNAs indeedhad 5'-triphosphate termini, they were eluted from preparativegels and analyzed with the RLM-RACE strategy developed previously(Fig. 3). To increase its sensitivity, a nested instead of asingle PCR amplification was applied. For the plus polaritystrand, an RT-PCR product of approximately 110 bp was obtainedusing as templates the eluted PLMVd subgenomic RNAs and alsoa CF11-fractionated RNA preparation from PLMVd-infected tissue(Fig. 4C, lanes 8 and 6, respectively). This was the expectedsize for a PLMVd plus subgenomic RNA with the same 5' terminusdetermined previously (Fig. 4A), and cloning and sequencingthe amplified product showed that this was the case. The reliabilityof the nested RLM-RACE was verified with an in vitro PLMVd plustranscript (mixed with an RNA preparation from healthy peach),which generated the predicted amplification product of 160 bp(Fig. 4C, lane 2), and by the lack of the same product in twoadditional controls without guanylyltransferase and withoutthe RNA adaptor (Fig. 4C, lanes 3 and 4, respectively). Similarexperiments for the minus polarity strand, using the elutedPLMVd subgenomic RNAs and a CF11-fractionated RNA preparationfrom PLMVd-infected tissue as templates, led to the RT-PCR productof approximately 200 bp (Fig. 4D, lanes 8 and 6, respectively)expected for a PLMVd minus subgenomic RNA with the same 5' terminusdetermined previously (Fig. 4B), and cloning and sequencingof the amplified product confirmed this point. Again, an invitro PLMVd minus transcript control (mixed with an RNA preparationfrom healthy peach) produced the predicted amplification productof approximately 250 bp (Fig. 4D, lane 2), which was absentin two parallel controls in which the guanylyltransferase orthe RNA adaptor was omitted (Fig. 4D, lanes 3 and 4, respectively).Collectively, the coincidence of the results obtained with variations(single and nested RLM-RACE) of the methodology applied eitherto a complex RNA mixture or to purified components thereof providesfirm support for the functional relevance of the proposed transcriptioninitiation sites.

DISCUSSION

Top

Discussion

How viroids switch the specificities of certain host RNA polymerasesto transcribe RNA instead of their physiological DNA templatesis one of the most interesting questions that remains to besolved, particularly considering that viroids have evolved thisability with both nuclear and chloroplastic RNA polymerases.A way to begin addressing this issue is to identify where thetranscription of viroid RNAs starts, because the adjacent regionmay provide hints about the nature of the promoters involved.Moreover, since RNA folding occurs during transcription, theinitiation sites of nascent viroid strands may determine theadoption of transient metastable structures that are functionallyrelevant (see below). We previously developed an in vitro capping/RNaseprotection methodology for mapping the 5' transcription initiationsites of ASBVd plus and minus RNAs isolated from infected avocado,which, like the 5' termini of other chloroplastic primary transcripts,are chemically labeled with a triphosphate (38). Here we reportan extension of this methodology for other chloroplastic viroidsand host transcripts accumulating in vivo at concentrationsmuch lower than that of ASBVd. An example of these viroids isPLMVd, which differs from ASBVd in key aspects, including G+Ccontent, secondary structure, and morphology of the hammerheadribozymes (27). Moreover, the accumulation in vivo of both polaritystrands does not differ very much in PLMVd (reference 8 andthis work), whereas the ratio of plus strands to minus strandsin ASBVd is considerably higher (11, 28); on the other hand,the PLMVd monomeric linear RNAs are more abundant than theircircular counterparts (reference 8 and this work), whereas theopposite occurs with ASBVd (11, 28). Finally, longer-than-unitviroid strands accumulate in ASBVd-infected avocado (11, 28),but they are absent in PLMVd-infected peach (Fig. 1) or accumulateat very low concentrations (8); this difference may result fromthe higher catalytic efficiency of PLMVd ribozymes, which arestable and operate as single hammerhead structures (4, 27),compared to those of ASBVd, which are unstable and operate asdouble-hammerhead structures (23). Overall, these significantdivergences between ASBVd and PLMVd made unlikely the existenceof similar transcription initiation sites and created the stimulusto determine those of the latter.

After checking with proper controls the reliability of the adaptedRLM-RACE methodology, its application to RNA preparations fromPLMVd-infected peach fruits led to the mapping of the initiationsite of each PLMVd strand in the vicinity of the self-cleavagesite of its complementary strand. The existence in these preparationsof monomeric linear PLMVd RNAs with 5' termini correspondingto both initiation sites was further confirmed by primer extensionanalysis. Moreover, the application of a nested modificationof the adapted RLM-RACE methodology to two PLMVd subgenomicRNAs of both polarities led to the same transcription initiationsites. We believe that these sites, located at the base of thehammerhead arm (Fig. 4E), reflect the in vivo situation betterthan those reported previously at terminal loop A (Fig. 4E)of this long hairpin, which were inferred from primer extensionanalysis of subgenomic PLMVd RNAs isolated from infected tissueand from in vitro transcriptions with a nonphysiological system(41, 42). Moreover, the 12- to 13-nt insertion observed in certainPLMVd variants, inducing peach calico (an extreme chlorosis)(34), introduces an important sequence and structural variabilityin loop A that argues against this loop containing the in vivoinitiation sites.

Therefore, in contrast with the situation observed for ASBVdinitiation sites, which are located at A+U-rich terminal loops(38), those from PLMVd map at a motif of 6 to 7 bp with lowvariability, because it contains the conserved triplet GUC thatprecedes the self-cleavage site in both polarity strands. Moreover,preliminary experiments with the same adapted RLM-RACE methodologyindicate that at least the initiation site of the plus strandof Chrysanthemum chlorotic mottle viroid, a viroid that is clusteredwith PLMVd within the genus Pelamoviroid on the basis of theirsimilar branched conformations (21, 37) stabilized by pseudoknots(9, 24), also maps at a similar region (data not shown). Altogether,these results suggest that the hammerhead arm, which adoptssimilar long hairpin foldings in the two polarity strands ofPLMVd, contains the corresponding promoters that, by themselvesor via the binding of a transcription factor, would recruitthe chloroplastic RNA polymerase. The sequence and structuralsimilarity between both promoters might account for the comparableaccumulations in vivo of PLMVd plus and minus strands, althoughalternative explanations cannot be dismissed. On the other hand,the location of the PLMVd initiation sites at the base of thelong stable hairpins entails that the sequences of the hammerheadstructure of the nascent strand are synthesized before thoseforming the hammerhead structure in the complementary polarity.Therefore, nascent plus and minus strands can self-cleave duringtranscription, a possibility that is disfavored for the alternativeinitiation sites in loop A reported previously (41), becausethe adoption of the hammerhead structure by the nascent strandcompetes with folding into the stable, catalytically inactive,hairpin. The particular location of the initiation sites hereproposed and the morphology of PLMVd hammerhead structures,which as indicated above operate as single hammerhead structures,would account for the low amount or absence of longer-than-unitviroid strands in PLMVd-infected peach.

The templates for transcription could be the circular viroidRNAs or even their linear counterparts, which accumulate invivo at higher concentrations (Fig. 1), assuming the involvementof an RNA polymerase with the ability to recognize a 2',3' cyclicphosphodiester terminus in the template and to jump reiterativelyover the template discontinuity. The resulting oligomeric viroidstrands would then be processed into the mature circular RNAby the concerted action of the corresponding hammerhead ribozymeand either an RNA ligase or autocatalytically (20). Such "jumpingpolymerases" have been proposed to play a role in the recombinationand evolution of viroid RNAs (31) as well as in the in vivoreplication of an RNA sharing certain structural and functionalproperties with viroid RNAs, the RNA of human hepatitis deltavirus (10), for which linear and circular species can contributeto the overall process of hepatitis delta virus genome replication(26).

With regard to the specific chloroplastic RNA polymerase involved,in vitro studies with PLMVd and E. coli RNA polymerase suggestthe participation of the PEP (41), although results obtainedwith such heterologous systems should be viewed with care asto whether or not they reproduce the physiological context.Moreover, the synthesis (and accumulation) of PLMVd is particularlyactive in leaf areas displaying peach calico, in which the developmentof proplastids into chloroplasts and the processing of certainchloroplastic rRNA precursors (and, consequently, the translationof plastid-encoded proteins) appear to be impaired (M. E. Rodio,R. Flores, and F. Di Serio, unpublished results). These observationsare more consistent with the involvement in PLMVd replicationof an NEP-like enzyme, which could be the same as that proposedto mediate ASBVd replication on the basis of its sensitivityto the inhibitor tagetitoxin (39) or another NEP-like enzyme,because more than one RNA polymerase of this class has beenreported (5). NEP appears more flexible than initially presumed,because recent results have shown that all plastid genes, ratherthan only the housekeeping genes needed early in development,are transcribed by NEP in PEP-deficient tobacco plastids (32).

ACKNOWLEDGMENTS

Work in R.F.'s laboratory has been supported by the Ministeriode Ciencia y Tecnología (MCyT) (BMC2002-03694) and theGeneralidad Valenciana (GV) of Spain. During this work, S.D.was supported by postdoctoral fellowships from the MCyT andthe GV, and A.E.M.D.A. was supported by postdoctoral fellowshipsfrom the Basque Government and the MCyT.

We thank S. Shuman for kindly providing the recombinant plasmidfor coexpression of the two subunits of the vaccinia virus mRNA-cappingenzyme and A. Ahuir for excellent technical assistance.

FOOTNOTES

* Corresponding author. Mailing address: Instituto de Biología Molecular y Celular de Plantas (UPV-CSIC), Universidad Politécnica de Valencia, 46022 Valencia, Spain. Phone: 34-96-3877861. Fax: 34-96-3877859. E-mail: rflores{at}ibmcp.upv.es.

We dedicate this paper to the memory of our dear colleague andfriend Martin Tabler.

REFERENCES

Top