Suppression of Viral RNA Recombination by a Host Exoribonuclease

RNA viruses of humans, animals, and plants evolve rapidly dueto mutations and RNA recombination. A previous genome-wide screenin Saccharomyces cerevisiae, a model host, identified five hostgenes, including XRN1, encoding a 5'-3' exoribonuclease, whoseabsence led to an $~$ 10- to 50-fold enhancement of RNA recombinationin Tomato bushy stunt virus (E. Serviene, N. Shapka, C. P. Cheng,T. Panavas, B. Phuangrat, J. Baker, and P. D. Nagy, Proc. Natl.Acad. Sci. USA 102:10545-10550, 2005). In this study, we foundabundant 5'-truncated viral RNAs in xrn1 ${Delta}$ mutant strains butnot in the parental yeast strains, suggesting that these RNAsmight serve as recombination substrates promoting RNA recombinationin xrn1 ${Delta}$ mutant yeast. This model is supported by data showingthat an enhanced level of viral recombinant accumulation occurredwhen two different 5'-truncated viral RNAs were expressed inthe parental and xrn1 ${Delta}$ mutant yeast strains or electroporatedinto plant protoplasts. Moreover, we demonstrate that purifiedXrn1p can degrade the 5'-truncated viral RNAs in vitro. Basedon these findings, we propose that Xrn1p can suppress viralRNA recombination by rapidly removing the 5'-truncated RNAs,the substrates of recombination, and thus reducing the chancefor recombination to occur in the parental yeast strain. Inaddition, we show that the 5'-truncated viral RNAs are generatedby host endoribonucleases. Accordingly, overexpression of theNgl2p endoribonuclease led to an increased accumulation of cleavedviral RNAs in vivo and in vitro. Altogether, this paper establishesthat host ribonucleases and host-mediated viral RNA turnoverplay major roles in RNA virus recombination and evolution.

INTRODUCTION

Top

Introduction

Human-, animal-, and plant-pathogenic RNA viruses are frequentlysubjected to RNA recombination (1, 3, 20, 51), a process thatjoins noncontiguous RNA segments together. The resulting novelcombinations of genes, sequence motifs, and/or regulatory RNAsequences could cause dramatic changes in the infectious propertiesof RNA viruses that can potentially lead to the emergence ofnew viruses or strains. Therefore, RNA recombination can helpviruses to "jump species," escape natural resistance mechanisms,and/or render antiviral methods ineffective. Indeed, there isa growing number of examples where RNA recombination likelycontributed to viral outbreaks, including outbreaks of caliciviruses(15), astroviruses (48), poliovirus (12, 22), dengue virus (14,52), enteroviruses (21, 29), influenza virus (17), bovine viraldiarrhea virus (12), and the recombinant severe acute respiratorysyndrome (SARS) coronavirus, a newly emerged viral pathogenof humans (4, 42, 46). RNA recombination can also increase virusfitness by facilitating virus genome repair, a process thatleads to the correction of mistakes introduced during viralRNA replication by the error-prone viral RNA replicases or dueto damage caused by host ribonucleases (1, 3, 20, 27, 51).

Viral RNA recombination is thought to occur when the viral replicaseaccidentally switches templates during complementary RNA synthesis(6, 20, 27, 28). Indeed, mutagenesis of viral replicase proteinsaffected the sites and frequency of recombination (10, 11, 25,33). Moreover, purified viral replicases were used to demonstrateefficient generation of RNA recombinants in vitro in the presenceof added RNA templates (5, 6, 18). Some recombination eventsmight be due to a promiscuous RNA ligation process, as demonstratedfor a small number of RNA viruses (8).

Similar to other processes that take place during virus propagationin infected cells, RNA recombination is possibly affected byhost factors. Accordingly, a systematic, genome-wide analysisof individual host genes based on the Saccharomyces cerevisiaesingle-gene deletion library identified five host genes whoseabsence led to a 10- to 50-fold enhancement in viral recombinantRNA accumulation compared to the parental yeast strain (43).The above study was based on a model RNA virus, tomato bushystunt virus (TBSV), which is a small, nonsegmented, plus-strandRNA virus of plants (50). Yeast has been shown to be a highlysuitable host for studies of the replication and recombinationof a small TBSV replicon RNA (30). The replicon RNA, which isa naturally generated defective interfering (DI) RNA, consistsof four noncontiguous segments of the TBSV genomic RNA, namedRI to RIV. Coexpression of the replicon RNA with the two essentialtombusvirus replication proteins (i.e., p33 and p92 of cucumbernecrosis virus [CNV], which is closely related to TBSV [50])leads to robust RNA replication in yeast cells, which showedsimilar characteristics to plant cell infections (30, 35). Inaddition, small amounts of RNA recombinants, which were shownto be similar to those occurring in plants and plant protoplasts,were also detected in yeast, establishing that it is a suitablehost for viral recombination studies. The identification ofyeast genes that greatly affected the accumulation of TBSV recombinantsopened up the possibility of dissecting the roles of the identifiedhost genes in viral RNA recombination.

For this study, we have tested the role of the XRN1 gene, oneof the identified host genes from the genome-wide genetic screenwhose deletion increased TBSV recombination (43). Because XRN1encodes a 5'-3' exoribonuclease (16, 36), we hypothesized thatXrn1p could be involved in viral RNA recombination via affectingviral RNA degradation, which could potentially influence theamounts of viral RNA substrates available for recombinationevents. Accordingly, we demonstrate that Xrn1p is involved inrapid degradation of 5'-truncated viral replicon RNAs, whichthen leads to suppression of RNA recombination in the parentalyeast strain. In contrast, in an xrn1 ${Delta}$ mutant strain, the partiallydegraded (i.e., 5'-truncated) replicon RNAs accumulate and serveas efficient templates for RNA recombination. In addition, weshow that the partially degraded replicon RNAs are likely generatedby Ngl2p and other endoribonucleases. Based on our results,we propose that Xrn1p RNase suppresses viral RNA recombinationby quickly removing partially degraded replicon RNA templatesused for RNA recombination events. Altogether, our results establishthat host-mediated viral RNA turnover is a major factor in viralRNA recombination.

MATERIALS AND METHODS

Top

Materials and Methods

Yeast strains and expression plasmids. Saccharomyces cerevisiae strain BY4741 (MATa his3 ${Delta}$ 1 leu2 ${Delta}$ 0 met15 ${Delta}$ 0ura3 ${Delta}$ 0) and a haploid deletion series (BY4741 strain background)were obtained from Open Biosystems (Huntville, AL). The expressionplasmids pGBK-His33 (carrying the p33 gene of CNV, a closelyrelated virus to TBSV, behind the ADH1 promoter and HIS3 marker),pGAD-His92 (containing the CNV p92 gene behind the ADH1 promoterand LEU2 marker), and pYC/DI-72 (expressing TBSV DI-72 RNA underthe control of the GAL1 promoter and containing the URA3 markergene) have been previously described (31, 35). Each yeast strainwas cotransformed with all of the above three plasmids (exceptfor the RNA stability experiments) using a lithium acetate/single-strandedDNA/polyethylene glycol method (13), and transformants wereselected by complementation of auxotrophic markers. The transformedyeast strains were grown under inducing conditions as describedearlier (31, 35). Xrn1p and Ngl2p were expressed from plasmidspAJ10 (16) and pESC/DI-NGL2HisFLAG, respectively.

Protoplast inoculation. Three different DI RNA constructs with different 5'-end deletionswere generated by PCR and T7 transcription as described previously(35). The same amount of DI RNA transcripts (1 µg/sample),together with CNV genomic RNA transcripts (5 µg), wasused for electroporation of 5 x 10⁵ Nicotiana benthamiana protoplasts,prepared as described previously (32). Protoplasts were incubatedin the dark, followed by RNA extraction as described previously(32, 44).

Viral RNA analysis. Total RNA isolation and Northern blot analysis were done aspreviously described (30). For detection of plus-strand DI RNAs,a ³²P-labeled RNA probe representing RIII (Fig. 1B) was preparedby T7 transcription from a PCR-generated cDNA, which was amplifiedwith primers 1165 (AGCGAGTAAGACAGACTCTTCA) and 23 (GTAATACGACTCACTATAGGGACCCAACAAGAGTAACCTG).Total RNAs obtained from selected strains were also analyzedby Northern blotting using a probe representing RI of DI-72(Fig. 1A) obtained from a PCR template (primers 15 [GTAATACGACTCACTATAGGGCATGTCGCTTGTTTGTTGG]and 20 [GGAAATTCTCCAGGATTTCTC] were used for amplification).

View larger version (41K):
[in this window]
[in a new window]
FIG. 1. Accumulation of partially degraded replicon RNAs in xrn1 ${Delta}$ mutant yeast. (A) Schematic representation of DI-72 replicon RNA and newly generated recombinant RNAs. The recombinant RNAs contain partially duplicated sequences from the 3' half of the DI-72 replicon RNA joined in a head-to-tail fashion. (B) Ethidium bromide-stained agarose gel of total RNA extracts obtained from two yeast strains showing the accumulation of DI-72 replicon RNA (white arrowhead), recombinant RNA (open arrowhead), and putative partially degraded replicon RNAs (closed arrowheads). The yeast cells coexpressed DI-72 RNA and the p33 and p92 replication proteins. (C and D) Northern blot analysis of total RNA extracts with a ³²P-labeled RNA probe specific for RIII (C) or RI (D). The samples are the same as those used for panel B.

RT-PCR analysis of junction sites. We used both yeast total RNA extracts and gel-isolated recombinantsfor reverse transcription-PCRs (RT-PCRs) to specifically amplifyregions covering the junction sites (Fig. 1). First, the RTreaction included primer 14 (GTAATACGACTCACTATAGGGTTCTCTGCTTTTACGAAG)for cDNA synthesis, followed by PCR with primers 168 (TCGTCTTATTGGACGAATTCCTGTTTACGAAAG)and 270 (TTGGAAATTCTCCTTCAGTCTGAGTTTGTGGA). The PCR productswere cloned into the pGEM-T Easy vector (Promega) and sequencedwith the M13 reverse primer (5).

5' RACE and 3' RACE of recombinants. The 5' sequences of truncated RNAs were determined by using5' rapid amplification of cDNA ends (5' RACE). We followed aGC-rich sequence protocol for 5' RACE (Invitrogen). Briefly,total RNA samples obtained from yeast (43) were loaded in 4%denaturing polyacrylamide gel electrophoresis (PAGE) gels forgel isolation of the truncated RNAs. After recovery of the RNAsfrom gels, we performed 5' RACE. For amplification of first-strandcDNAs, we used primer 116 (GTAATACGACTCACTATAGGACACCTAACTTTCGT),which anneals to the 5' end of RIV(+) of TBSV DI-72. Second-strandcDNAs were synthesized by using a 3' RACE anchor primer (GGCCACGCGTCGACTAGTACTTTTTTTTTTTTTTTTT).A nested primer (291 [GTAATACGACTCACTATAGGAACCTGTATGCTATGCC]),which anneals to RIII(+) of DI-72, and an abridged universalamplification primer (AUAP) (GGCCACGCGTCGACTAGTAC) were usedfor the first PCRs, and the second PCRs were done by using theAUAP primer and the second nested primer 14 (GTAATACGACTCACTATAGGGTTCTCTGCTTTTACGAAG),which anneals to RII(+). The resulting products were directlycloned into pGEM-T Easy (Promega) and sequenced with the M13reverse primer.

The 3' end sequences of the truncated DI RNAs were identifiedusing 3' RACE as described previously (5, 43), based on a 3'RACE kit (Invitrogen). Purified RNAs were first polyadenylated[poly(A) polymerase; USB], followed by phenol-chloroform extractionprior to cDNA synthesis. First-strand cDNAs were synthesizedwith the 3' RACE anchor primer, followed by PCRs using the AUAPprimer and primer 20 (TBSV/CNV/1F+G [GGAAATTCTCCAGGATTTCTC]),which anneals to the 5' end of RI(–) of DI-72. The PCRproducts were cloned and sequenced as mentioned for 5' RACE.

Purification of recombinant Xrn1p from yeast. We used a yeast strain carrying tandem affinity purification(TAP)-tagged Xrn1p from a TAP fusion library (Open Biosystems)to obtain recombinant Xrn1p in a single-step purification withcalmodulin affinity resin (Stratagene) (39). Briefly, the yeastlysate containing TAP-tagged Xrn1p was loaded on 300 µlcalmodulin resin that was equilibrated with 5 ml IPP150 calmodulin-bindingbuffer (10 mM ß-mercaptoethanol, 10 mM Tris-HCl, pH8.0, 150 mM NaCl, 1 mM magnesium acetate, 1 mM imidazole, 2mM CaCl₂, 0.1% NP-40). After washing of the resin three timeswith 10 ml of IPP150 calmodulin-binding buffer in a cold room,recombinant Xrn1p was eluted with 1 ml IPP150 calmodulin elutionbuffer (10 mM ß-mercaptoethanol, 10 mM Tris-HCl, pH8.0, 150 mM NaCl, 1 mM magnesium acetate, 1 mM imidazole, 2mM EGTA, 0.1% NP-40). The presence and quality of Xrn1p (200kDa) were checked by 10% sodium dodecyl sulfate-PAGE (SDS-PAGE)and silver staining. In some of the in vitro experiments, wealso used purified recombinant Xrn1p-His6 provided by A. W.Johnson (16), with comparable results to those obtained withTAP-based purification (not shown).

RNA preparation for in vitro assays. Full-length DI-72(+) and DI- ${Delta}$ RI(+) RNAs with 5' triphosphateor 5' monophosphate were synthesized by T7 polymerase and labeledwith [³²P]UTP. RNAs with 5' monophosphate were generated byT7 polymerase with 10 mM ATP, 10 mM CTP, 10 mM GMP, and 0.5mM GTP. All transcripts were purified on MicroBio-Spin P-30columns (Bio-Rad) prior to the in vitro assay with Xrn1p.

In vitro Xrn1p assay. The standard assay was based on the procedure of Stevens, using1 to 10 µl of purified Xrn1p (47). Briefly, we used 2pmol of ³²P-labeled RNA transcripts in 70 µl RNA-dependentRNA polymerase buffer (50 mM Tris-HCl, pH 8.2, 10 mM MgCl₂,10 mM dithiothreitol, 100 mM potassium glutamate) (26) in thepresence of 10 U of RNase inhibitor (Amersham) and bovine serumalbumin at 37°C for various lengths of time (see Fig. 7),followed by phenol-chloroform extraction. The control preparationfor Xrn1p was obtained using the same calmodulin-based procedure(see above), except that the yeast extract was derived fromthe parental yeast strain (lacking TAP-tagged Xrn1p).

View larger version (38K):
[in this window]
[in a new window]
FIG. 7. Degradation of replicon RNAs by Xrn1p in vitro. (A) Demonstration of the exoribonuclease activity of affinity-purified Xrn1p in vitro. The template was ³²P-labeled DI- ${Delta}$ RI carrying a monophosphate at the 5' end. A comparison of the sensitivities of the full-length DI-72 and DI- ${Delta}$ RI replicon RNAs carrying either 5' monophosphate (B) or 5' triphosphate (C) to degradation by Xrn1p in vitro was performed. The time of treatment with Xrn1p is shown in minutes at the top. (D) Silver-stained SDS-PAGE analysis of affinity-purified Xrn1p preparations obtained from yeast expressing Xrn1p-TAP.

Purification of recombinant Ngl2p. Recombinant FLAG-tagged Ngl2p was affinity purified from yeaststrain BY4741 transformed with pESC/DI-NGL2HisFLAG. Briefly,the yeast extract was loaded on a column containing the FLAGM2 resin (Sigma) equilibrated with Tris-glycine (TG) buffercontaining 0.01% NP-40 (Nonidet P-40). After careful washingof the resin two times with 600 µl TG buffer (50 mM Tris-HCl,pH 7.5, 15 mM MgCl₂, 10 mM KCl, and 10% glycerol), Ngl2p waseluted with TG buffer containing 100 mM ß-mercaptoethanol.The quality of the purified Ngl2p ( $~$ 70 kDa) preparation was checkedby 8% SDS-PAGE, staining with Coomassie brilliant blue, andWestern blotting using anti-His antibody (35).

In vitro Ngl2p assay. One microgram of RNA transcripts (see above) was used with 1µl of purified Ngl2p in buffer (50 mM Tris-HCl, pH 7.5,10 mM MgCl₂, 100 mM NaCl, 0.1 mg/ml bovine serum albumin, andyeast tRNA [0.1 µg]). Reactions were carried out at 25°Cfor 10 min, followed by phenol-chloroform extraction. Some ofthe samples were further treated with 1 µl of Xrn1p-His6(a gift from Arlen W. Johnson) and 10 units of RNase inhibitorat 37°C for 20 min, followed by phenol-chloroform extraction.The obtained RNAs were used in RT reactions with a ³²P-labeledprimer (1163 [CCCGAAGCTTCCCAACAAGAGTAACCTGTATGCT]), using SuperScriptII reverse transcriptase (Invitrogen). After 60 min at 50°C,the RT reactions were stopped by adding 2x loading dye, andthe RT samples were loaded into 5% denaturing PAGE gels.

RESULTS

Top

Results

Rationale. A previous systematic, genome-wide screen of individual yeastgenes led to the identification of five host genes whose separatedeletion enhanced the accumulation of novel viral RNA recombinants>10-fold compared to that in the parental yeast strain (43).One of the identified genes was XRN1, which codes for a cytoplasmic5'-3' exoribonuclease (16, 36). Interestingly, the isolatedrecombinant RNAs had 5' deletions and 3' duplications of repliconRNA sequences, which resulted in incomplete dimeric recombinantRNAs (43) (shown schematically in Fig. 1A). Based on the lackof Xrn1p 5'-3' exoribonuclease and the nature of junction sequences,we hypothesized that viral RNA degradation might be slow inxrn1 ${Delta}$ mutant cells, which could then lead to an accumulationof 5'-truncated viral RNA degradation products. This in turnmight promote viral recombination if the incompletely degradedreplicon RNAs could serve as efficient templates for recombinationevents, resulting in the generation of abundant dimeric recombinantviral RNAs. This model was tested by using in vivo and in vitroapproaches.

Characterization of partially degraded viral RNAs in xrn1 ${Delta}$ mutant cells. To dissect the role of XRN1 in recombination of the TBSV replicon,we first characterized partially degraded viral replicon RNAs,which are suggested to serve as recombination intermediatesin xrn1 ${Delta}$ mutant yeast. We performed Northern blotting analysiswith probes specific for 5' and 3' replicon sequences, usingtotal RNA samples obtained from cells actively replicating DI-72replicon RNA. As predicted, we observed the accumulation ofabundant amounts of partially degraded 3' fragments of viralRNAs (termed 3'Fr), which represented two different sizes (theabundant $~$ ${Delta}$ 170 and the minor ${Delta}$ 240 RNAs, which were 170 to 240 nucleotides[nt] shorter than DI-72 replicon RNA), in xrn1 ${Delta}$ mutant cells(Fig. 1B to D). In contrast, the accumulation of 3'Fr RNA wasbarely detectable in the parental yeast cells (Fig. 1C). Interestingly,we also detected 5' fragments of DI-72 replicon RNAs (5'Fr RNA, $~$ 170 nt in length), which were detectable in the parental cellsand abundant in xrn1 ${Delta}$ mutant cells (Fig. 1D).

To find out the possible relationship between these partiallydegraded replicon RNAs and the recombinant RNAs, we determinedthe sequences of the truncated RNAs, using gel-isolated RNAsafter cloning of RT-PCR, 3' RACE, and 5' RACE products (43).The 3'Fr RNAs had 5' deletions ranging from 172 to 376 nt fromthe 5' end of the replicon RNA (Fig. 2). Interestingly, thedeletions removed the entire RI sequence and part of the RIIsequence of DI-72 replicon RNA. Altogether, the 5'-terminalsequences in many 3'Fr RNAs (i.e., 5' deletions of 68 to 70nt in RII; Fig. 1A) resembled the sequences present at the 5'ends and at the recombination junctions in the most abundantrecombinant RNAs, supporting the model that some of the partiallydegraded 3'Fr RNAs are likely involved in RNA recombinationevents as intermediates. This is based on the observation thatrecombination-driven joining of two partially degraded 3'FrRNAs in a head-to-tail fashion could give rise to the recombinantsidentified in xrn1 ${Delta}$ mutant cells (Fig. 1A).

View larger version (37K):
[in this window]
[in a new window]
FIG. 2. Terminal sequences of partially degraded replicon RNAs. (Top) The original replicating DI-72 replicon RNA and the partially degraded RNA products are shown schematically. (Bottom left) 3'-Terminal sequences in the 5'Fr RNA products were obtained by 3' RACE from the xrn1 ${Delta}$ mutant and parental yeast strains. The terminal sequences represent either RI or RII sequences, as indicated. (Bottom right) 5'-Terminal sequences in the 3'Fr RNA products were obtained by 5' RACE from xrn1 ${Delta}$ mutant strains. The lengths of the missing 5' sequences are indicated with numbers behind the " ${Delta}$ " symbols, whereas the 5' sequences in the 3'Fr RNAs are shown. Lowercase letters indicate nonviral sequences.

Sequence analysis of the 5'Fr RNAs from both wild-type (wt)and xrn1 ${Delta}$ mutant cells by 3' RACE revealed the presence of RIsequences and short stretches of RII (Fig. 2). The corresponding5' sequences were missing in the recombinant RNAs (Fig. 1D),suggesting that the observed 5' fragments of DI-72 repliconRNA did not participate in viral recombination events (see Discussion).Overall, sequences present in 3'Fr and 5'Fr RNAs did not overlapsignificantly (Fig. 2). The presence of nonoverlapping 5'Frand 3'Fr RNAs in xrn1 ${Delta}$ mutant cells suggests that they were generatedvia cleavage of the replicon RNA within RII sequences (or inthe vicinity of the junction between RI and RII) by a putativeendoribonuclease instead of by incomplete degradation from eitherthe 5' or 3' end by exoribonucleases in parental and xrn1 ${Delta}$ mutantcells.

3'Fr RNAs serve as substrates for RNA recombination in yeast cells and plant protoplasts. To test if the 3'Fr RNAs could participate in recombination,we constructed two yeast plasmids for the expression of TBSVRNAs (i.e., DI- ${Delta}$ RI [Fig. 3A] and DI- ${Delta}$ 69RII [Fig. 3B]) that represent3'Fr RNAs (Fig. 2). Then we expressed DI- ${Delta}$ RI and DI- ${Delta}$ 69RII, togetherwith the p33 and p92 replicase proteins, in wt and xrn1 ${Delta}$ mutantyeast to see if these RNAs can replicate and recombine. As predictedabove, total RNA extracts obtained from xrn1 ${Delta}$ yeast cells expressingDI- ${Delta}$ 69RII RNA contained abundant amounts of recombinant RNAs(Fig. 3B). Sequencing the junctions in these recombinants confirmedthat the expressed DI- ${Delta}$ 69RII RNA (a representative 3'Fr RNA species;Fig. 2) did indeed participate in RNA recombination (not shown)by generating head-to-tail dimers similar to that shown schematicallyin Fig. 1A. Interestingly, expression of a longer 3'Fr RNA (i.e.,DI- ${Delta}$ RI RNA; Fig. 3A) led to the occurrence of shorter 3'Fr RNAswith additional 60- to 150-nt deletions from the 5' end of RII(Fig. 3A [sequences are not shown]). Sequence analysis of thegenerated recombinants obtained with yeast expressing DI- ${Delta}$ RIrevealed that the shorter 3'Fr RNAs (which quickly evolved tosizes similar to that of DI- ${Delta}$ 69RII) participated in the generationof recombinants with higher frequencies than the originallyexpressed DI- ${Delta}$ RI RNA (not shown), suggesting that the short version(similar to DI- ${Delta}$ 69RI) is the more active substrate in recombination.

View larger version (35K):
[in this window]
[in a new window]
FIG. 3. Efficient generation of viral recombinants from 3'Fr RNAs in yeast. Two representative 3'Fr RNAs, DI- ${Delta}$ RI RNA (A) and DI- ${Delta}$ 69RII RNA (B), were separately expressed in yeast (together with p33 and p92) from the GAL1 promoter. Northern blot analysis of the total RNAs obtained 24 h after induction with galactose revealed the expressed 3'Fr RNAs (open arrowheads). The newly generated recombinants are bracketed. The black arrowhead indicates that DI- ${Delta}$ RI RNA also generated a partially degraded RNA (possibly 5'-truncated, DI- ${Delta}$ 69RII-like RNA) in the parental and xrn1 ${Delta}$ mutant strains.

Unlike the case with full-length DI-72 replicon RNA, the controlparental yeast strain also generated abundant recombinants whenit expressed either of the two 3'Fr RNAs (Fig. 3B). The sizesof the recombinants, however, were shorter in the parental yeastthan in xrn1 ${Delta}$ mutant yeast due to $~$ 100-nt 5' truncations in therecombinants (Fig. 3B [sequencing data are not shown]). Nevertheless,the occurrence of abundant recombinant RNAs in the parentalstrain suggests that the presence of plentiful 3'Fr RNAs promotesRNA recombination events. Thus, the parental strain is alsorecombination competent if a sufficient amount of recombinationsubstrates is available. Overall, these data support the modelthat some of the 3'Fr RNAs can serve as recombination intermediates(substrates) during RNA recombination events, leading to theformation of partially dimeric recombinant RNA.

To test if 3'Fr RNAs can also serve as recombination intermediatesin plant cells, we separately electroporated three different3'Fr RNAs into Nicotiana benthamiana protoplasts. The wt CNV(a closely related virus to TBSV) genomic RNA was also coelectroporatedinto these cells to serve as a helper virus (which producesthe replicase proteins) for the replication/recombination ofthe 3'Fr RNAs (44). Total RNA extracts obtained from protoplasts24 h after electroporation contained not only the replicating3'Fr RNAs but also novel dimeric recombinant RNAs in the caseof two of the three 3'Fr RNAs tested (Fig. 4). The amounts of3'Fr RNAs and recombinant RNAs increased over time (Fig. 4).Sequencing of the junctions in these recombinants revealed thatthey consisted of head-to-tail dimers, similar to the recombinantsobtained in yeast cells (not shown). Similar to its replicationin parental yeast, DI- ${Delta}$ RI also generated a shorter, DI- ${Delta}$ 69RII-sizedreplicating RNA in N. benthamiana protoplasts (indicated bya black asterisk in Fig. 4), suggesting that the evolution andreplication of DI- ${Delta}$ RI RNA are similar in yeast and plant hosts.Construct DI- ${Delta}$ 179RII did not replicate and generate recombinantsin N. benthamiana protoplasts (Fig. 4), consistent with previousfindings that the central portion of RII includes an essentialcis-acting element which is involved in template selection bythe tombusvirus replication protein (23, 38). The stabilitiesof all three 3'Fr RNAs were comparable in N. benthamiana protoplasts(Fig. 4), supporting the model that the differences among theseconstructs in the presence of the helper virus are due to theirdifferent abilities to participate in replication/recombination.Overall, these experiments demonstrated that some 3'Fr RNAscan participate in similar recombination events in both yeastand plant cells.

View larger version (82K):
[in this window]
[in a new window]
FIG. 4. N. benthamiana protoplasts also support the generation of viral recombinants from 3'Fr RNAs. Three representative 3'Fr RNAs (missing RI and portions of RII, as indicated in their names) were coelectroporated with CNV helper virus into protoplasts. Northern blot analyses of the total RNAs obtained at 20 h (top) and 40 h (bottom) based on RIII-specific probes revealed two of the electroporated 3'Fr RNAs (open arrowheads) replicated in the presence of the helper virus (left side of the image). The newly generated recombinants are indicated by black arrowheads. Asterisks indicate partial degradation products (DI- ${Delta}$ 69RII-like RNAs). Note that the electroporated 3'Fr RNAs show similar levels of stability in the absence of replication (no helper virus; right side of the image).

The above data support the model that 3'Fr RNAs are generatedvia partial degradation of the full-length DI-72 replicon RNAin cells and that some of the 3'Fr RNAs can efficiently participatein RNA recombination. Due to the central role of 3'Fr RNAs inviral recombination, below we examine the proposed roles ofhost endoribonucleases and Xrn1p in the generation and degradationof 3'Fr RNAs and their effects on tombusvirus recombination.

Ngl2p endoribonuclease can generate 3'Fr RNAs in vivo and in vitro. Based on the detection of nonoverlapping 5'Fr and 3'Fr RNAsfrom the full-length DI-72 replicon RNA (Fig. 1 and 2), we predictedthat a putative endoribonuclease might be involved in selectivecleavage of the full-length replicon RNA. To this end, we selectedNgl2p and Rny1p from a small group of known cytoplasmic endoribonucleasesfor further studies, because they are abundant and located inthe cytoplasm (9). The expression of Ngl2p in ngl2 ${Delta}$ mutant yeastwith replicating DI-72 replicon RNA resulted in an increasedlevel of RI-containing degradation products (i.e., 5'Fr), bysevenfold (Fig. 5), and of the 3'Fr products, by twofold (Fig.5B), whereas the expression of Rny1p endoribonuclease had nosignificant effect on DI RNA degradation (not shown). The observedpartially degraded products in the Ngl2p-expressing strain werecomparable in size to the previously characterized 5'Fr RNAs(Fig. 2 and 5), supporting the model that Ngl2p is involvedin selective degradation of DI-72 replicon RNA that resultsin 5'Fr and 3'Fr RNAs (Fig. 5). Overexpression of Ngl2p in xrn1 ${Delta}$ mutant cells increased the total amount of 3' degradation products(3'Fr) of DI-72 replicon RNA more than threefold. The increasedlevels of 5'Fr and 3'Fr RNAs in yeast expressing Ngl2p are consistentwith the model that Ngl2p is involved in TBSV RNA degradationin yeast. In addition, cleavage of the full-length DI-72 repliconRNA by Ngl2p likely generates 3'Fr RNAs, which could then stimulateRNA recombination if saved from rapid degradation (i.e., inxrn1 ${Delta}$ mutant cells).

View larger version (60K):
[in this window]
[in a new window]
FIG. 5. Expression of Ngl2p endoribonuclease leads to increased levels of 5'Fr and 3'Fr RNAs in yeast. (A) Northern blot analyses of total RNA extracts obtained from either the ngl2 ${Delta}$ or xrn1 ${Delta}$ mutant strain not expressing (–) or expressing Ngl2p (in addition to p33, p92, and DI-72 RNA). The probe was specific for RI (i.e., 5'Fr RNA). The positions of the expressed DI-72 replicon RNA and the generated 5'Fr RNAs are shown. (B) Northern blot analyses of the above RNA samples with a probe specific for RIII (i.e., 3'Fr RNA). Part of the gel is shown after a longer exposure to visualize less intense bands on the blot. See above for additional description.

To test if Ngl2p can also cleave DI-72 RNA in vitro, we affinitypurified Ngl2p from yeast cells overexpressing FLAG-tagged Ngl2p(see Materials and Methods). After the nuclease assay (see Materialsand Methods), we performed in vitro reverse transcription witha ³²P-labeled primer, which was complementary to RIII(+), onthe Ngl2p-treated DI-72(+) RNA template (Fig. 6, lane 1). Basedon the sizes of the cDNA fragments separated in sequencing gels,we concluded that Ngl2p could cleave DI-72(+) RNA in vitro,resulting in three major and additional minor cleavage productswithin RI and RII (Fig. 6, lane 1). In conclusion, these invitro and in vivo experiments support the model that the Ngl2pendoribonuclease can specifically cleave DI-72(+) RNA, whichcould then result in 3'Fr RNA substrates for RNA recombination.

View larger version (85K):
[in this window]
[in a new window]
FIG. 6. Ngl2p-cleaved DI-72(+) replicon RNA is sensitive to Xrn1p. The full-length DI-72 replicon RNA carrying 5' triphosphate was treated with affinity-purified Ngl2p, followed by treatment with Xrn1p for 10 and 30 min (lanes 3 and 4, respectively). Reverse transcription in the presence of a ³²P-labeled primer was used to detect partially degraded RNAs with different 5' termini. The partial Ngl2p cleavage products are marked with arrowheads, whereas RT pausing sites are shown with asterisks. The three samples on the right are molecular size markers.

Redundant endoribonucleases are involved in cleavage of the replicon RNA in yeast. To test if Ngl2p is the only endoribonuclease of yeast capableof cleaving the replicon RNA, we tested viral RNA degradationin ngl2 ${Delta}$ mutant cells (Fig. 5). These experiments demonstratedthat ngl2 ${Delta}$ mutant cells, similar to the parental yeast cells,produced the characteristic RI-sized cleavage products of theDI-72 replicon RNA, suggesting that yeast has functionally redundantendoribonucleases (possibly Ngl1p, Ngl3p, and others). In addition,we have previously detected viral RNA recombinants at closeto the parental level in ngl2 ${Delta}$ mutant cells (43). Therefore,we concluded that Ngl2p and additional endoribonucleases arelikely involved in viral RNA degradation/recombination in yeast.The combined activity of these endoribonucleases produces 5'Frand 3'Fr RNAs (the latter are the recombination substrates),leading to rapid recombinant accumulation in xrn1 ${Delta}$ mutant cells.

Xrn1p exoribonuclease can rapidly degrade 3'Fr RNAs in vitro and in vivo. Based on (i) the abundance of the full-length DI-72 RNA andits 5' fragments generated by endoribonuclease cleavage (Fig.5) in both ngl2 ${Delta}$ and xrn1 ${Delta}$ mutant cells and (ii) the abundanceof 3'Fr RNAs only in xrn1 ${Delta}$ mutant cells, but not in the parentalcells (Fig. 1C), we predicted that Xrn1p is likely involvedin rapid degradation of 3'Fr RNAs but that the full-length DI-72RNA might only be slowly degraded by Xrn1p.

To test the activity of Xrn1p on DI-72(+) RNA templates, weaffinity purified Xrn1p as a TAP fusion protein from yeast cells(Fig. 7D) (39). In addition, we also used metal-affinity-purifiedXrn1p provided by A. W. Johnson. The obtained Xrn1p preparationswere used to treat ³²P-labeled, in vitro-transcribed DI-72-derivedRNAs (Fig. 7). These in vitro experiments demonstrated that3'Fr RNA, which carried a monophosphate (the expected productof an endonuclease cleavage) at its 5' end, was quickly degradedby Xrn1p (Fig. 7A). In contrast, full-length DI-72(+) RNA, regardlessof the presence of either mono- or triphosphate at its 5' end,was more resistant to Xrn1p under the conditions tested (Fig.7B and C). The significant degree of resistance to Xrn1p-mediateddegradation by the full-length DI-72(+) RNA and the 5'Fr RNAcould be due to the highly structured nature of RI(+) (40),which is present in both RNAs but not in 3'Fr RNA. It is establishedthat Xrn1p activity is inhibited by stable secondary structuresin the RNA substrate (47). We concluded based on the in vitroexperiments described above that Xrn1p is capable of rapidlydegrading the 3'Fr RNA, but not the full-length DI-72 RNA.

To test if Xrn1p could degrade DI-72(+) RNA after cleavage withNgl2p, we treated DI-72(+) RNA with purified recombinant Ngl2pand Xrn1p, followed by primer extension with RT (Fig. 6, lanes3 and 4). As expected, the amounts of Ngl2p-cleaved DI-72(+)RNA products were decreased in the presence of Xrn1p. Overall,the data from the in vitro experiments are consistent with themodel that Xrn1p is capable of rapidly degrading endoribonuclease-cleavedreplicon RNA products.

To examine if 3'Fr RNA is more stable in xrn1 ${Delta}$ mutant cells thanin parental yeast cells, we expressed DI- ${Delta}$ 69RII RNA from thegalactose-inducible GAL1 promoter in the absence of p33/p92replication proteins in the above yeast strains (43). The stabilityof DI- ${Delta}$ 69RII was then determined by measuring the half-livesof the RNA transcripts in the absence of replication. As expected,the half-life of DI- ${Delta}$ 69RII RNA increased almost threefold inxrn1 ${Delta}$ mutant cells compared to the parental yeast cells (Fig.8). The increased half-life of the recombination substrate inxrn1 ${Delta}$ mutant cells supports our model that the abundant recombinationallyactive templates are more stable and thus more available toincrease the frequency of recombination in xrn1 ${Delta}$ mutant cellsthan in the parental yeast cells.

View larger version (16K):
[in this window]
[in a new window]
FIG. 8. Increased stability of 3'Fr RNAs in xrn1 ${Delta}$ mutant cells. The 5'-truncated DI- ${Delta}$ 69RII RNA, which represents a partially degraded DI-72 RNA product (Fig. 2), was expressed in the xrn1 ${Delta}$ mutant and parental (wt) yeast strains from the GAL1 promoter (in the absence of p92), followed by the suppression of RNA synthesis by glucose. RNA samples were taken at various time points (as shown) and analyzed by Northern blotting. The intact DI- ${Delta}$ 69RII RNA was quantified and shown as a percentage of the RNA level detected at time zero.

Xrn1p is a suppressor of viral RNA recombination in vivo. To test if Xrn1p can suppress TBSV RNA recombination, we expressedXrn1p from a plasmid in xrn1 ${Delta}$ mutant cells that also coexpressedDI-72(+) replicon RNA and the p33/p92 replication proteins (Fig.9A). The expression of Xrn1p in xrn1 ${Delta}$ mutant cells reduced theamount of recombinant RNAs three- to fourfold as well as decreasingthe amount of 3'Fr-like, partially degraded replicon RNAs three-to fourfold (Fig. 9B). In contrast, Xrn1p did not reduce thelevel of DI-72 replicon RNA. Similarly, the expression of Xrn1pinhibited the accumulation of recombinant RNAs 12-fold and ofpartially degraded 3'Fr products 6-fold in xrn1 ${Delta}$ mutant cellsexpressing the highly recombinogenic DI-AU-FP RNA (Fig. 9B).Altogether, these data establish that Xrn1p inhibits the accumulationof both viral recombinant RNAs and partial viral RNA degradationproducts, thus acting as a suppressor of viral RNA recombination.

View larger version (40K):
[in this window]
[in a new window]
FIG. 9. Expression of Xrn1p in xrn1 ${Delta}$ mutant strain inhibits recombination. (A) The coexpressed replicon RNA was either DI-72 or DI-AU-FP replicon RNA, as shown schematically. These cells also expressed p33 and p92. (B) Northern blot analysis was done as described in the legend to Fig. 1C with an RIII-specific probe. The DI-72 and DI-AU-FP replicon RNAs, the recombinants, and partially degraded 3'Fr RNAs are marked. Note that the recombinant RNAs are shorter than the DI-AU-FP replicon RNA due to the deletion of the 5'-terminal $~$ 350 to 400 nt in the 3'Fr recombination substrate (bracketed).

DISCUSSION

Top

Discussion

In contrast to recent advances in our understanding of the rolesof host factors in RNA virus replication (2, 19, 31), the rolesof host proteins in viral RNA recombination are currently unknown.However, host factors might affect RNA recombination in multipleways by (i) influencing the quantity of viral RNA substratesavailable in cells, which is known to affect the frequency ofrecombination (5, 18, 24); (ii) changing the characteristicsof the viral replicase, which are proposed to drive most RNArecombination events (10, 11, 25, 33); or (iii) other stillundiscovered mechanisms.

In this paper, we demonstrate that the XRN1 gene, which wasidentified previously during a systematic genome-wide screenof yeast genes affecting tombusviral RNA recombination (43),plays a central role in viral RNA recombination by controllingthe amounts of partially degraded RNA substrates available forrecombination in cells. Based on multiple in vivo and in vitroassays, we propose that Xrn1p reduces the recombination frequencyby rapidly degrading the preferred recombination substrates,which are 5'-truncated replicon RNAs, termed 3'Fr RNAs. Indeed,previous studies revealed that (i) RII of DI-72 RNA is a recombinationhot spot (7, 33, 44), (ii) 5'-truncated TBSV RNAs are activein RNA recombination in plant protoplasts (49), and (iii) thetombusvirus replicase favors end-to-end template switching invitro (5). All of the above data support our model that partiallydegraded 3'Fr RNAs promote efficient RNA recombination by thetombusvirus replicase due to the exposure of the RII recombinationhot spot in the 5'-truncated RNAs. Accordingly, overexpressionof the 3'Fr RNAs in yeast or in plant protoplasts, which likelyoverwhelmed the host RNA degradation machinery (which is basedon Xrn1p in yeast [16, 36] and Xrn4p in plants [45]), led toefficient viral recombination (Fig. 3 and 4). This suggeststhat high levels of recombination substrates are sufficientto produce viral RNA recombinants in a wt or mutant host background.In contrast, we found that the full-length DI-72 replicon RNAwas more resistant to Xrn1p degradation in vitro, suggestingthat Xrn1p selectively degrades partially cleaved and/or incompletereplicon RNAs. In addition, the above data suggest that thehighly structured RI sequence present at the 5' end of the TBSVreplicon RNA (40, 41) seems to be partially resistant to Xrn1p-mediated5'-3' RNA degradation. Therefore, we propose that Xrn1p affectsreplication/recombination of the TBSV replicon by (i) preferablydegrading partial RNA degradation products, (ii) reducing theirchances of participating in RNA recombination, and (iii) maintainingthe most stable full-length DI-72 replicon RNA.

The vastly increased accumulation of viral RNA recombinantsin xrn1 ${Delta}$ mutant cells in comparison with parental yeast cellsand the inhibition of recombinant RNA accumulation in xrn1 ${Delta}$ mutantcells expressing Xrn1p strongly establish that Xrn1p is a potentsuppressor of the viral RNA recombination process. Because asimilar 5'-3' exoribonuclease, named Xrn4p, with comparablefunctions has been identified and characterized from Arabidopsis(45), it is possible that the Xrn4p exoribonuclease plays analogousroles in plants in tombusvirus recombination and possibly inthat of other RNA viruses, too.

This work also reveals that Xrn1p does not work alone in viralRNA degradation and recombination. This conclusion is supportedby the abundance of partially degraded replicon RNA products,which represented both 5' and 3' fragments, in xrn1 ${Delta}$ mutant cells(Fig. 1). The nonoverlapping nature of the 5' and 3' fragmentsaccumulating in xrn1 ${Delta}$ mutant cells supported the involvementof a host endoribonuclease(s) that might cleave the full-lengthreplicon RNA prior to further digestion by Xrn1p. Accordingly,overexpression of the Ngl2 endoribonuclease resulted in increasedamounts of partially degraded replicon RNAs in vivo (Fig. 5).Moreover, purified Ngl2p was able to cleave the replicon RNAin vitro, generating cleavage products of the expected sizes.These observations are consistent with the role of Ngl2p intombusviral RNA degradation. However, Ngl2p is not essentialin the viral RNA degradation and recombination processes, becausepartially degraded products were also observed in ngl2 ${Delta}$ mutantcells (Fig. 5B). This is likely due to the redundancy of cytoplasmicendoribonucleases, such as Ngl1p and Ngl3p, in yeast. Overall,the presented data promote central roles for host exo- and endoribonucleasesin viral RNA recombination.

Model for the roles of host ribonucleases in viral RNA recombination. Based on our in vivo and in vitro studies, we propose that Ngl2pand similar endoribonucleases, together with the Xrn1p 5'-3'exoribonuclease, affect viral RNA recombination. First, selectivecleavage of the full-length DI-72(+) RNA within the RII sequenceby Ngl2p produces 5' and 3' fragments, the latter of which isimportant for RNA replication/recombination. Subsequently, the3' fragment of DI-72 RNA could be quickly degraded by Xrn1pin the parental yeast strain, thus reducing its chance to participatein RNA recombination events (Fig. 10). Thus, Xrn1p basicallyacts as a suppressor of RNA recombination. In the absence ofXrn1p, however, the 3' fragment (i.e., 5'-truncated RNA) ofthe replicon RNA accumulates to higher levels due to (i) ongoingNgl2p cleavage of the full-length replicon RNA, (ii) replicationof the generated 5'-truncated replicon RNA, and (iii) a lackof Xrn1p-mediated degradation of the 5'-truncated replicon RNA.The resulting increased amount of 5'-truncated replicon RNAcould then serve as an efficient template for RNA recombinationdue to exposure of the RII recombination hot spot. Overall,our model predicts that the combined activities of the Ngl2pand Xrn1p ribonucleases control the amounts of available recombinationsubstrates and thus the frequency of RNA recombination eventsand the in vivo stabilities of the recombinants (43).

View larger version (34K):
[in this window]
[in a new window]
FIG. 10. Model of the suppression of tombusviral RNA recombination by Xrn1p in yeast. (A) DI-72 replicon RNA replicates in the parental strain (left panel) or in the xrn1 ${Delta}$ mutant strain (right panel). (B) During or after the replication of DI-72 replicon RNA, a host endoribonuclease such as Ngl2p cleaves some of the viral RNA (marked by arrows), producing 5'Fr and 3'Fr RNAs. (C) These cleaved RNA products will then be rapidly removed by the host Xrn1p exoribonuclease in the parental strain. In contrast, the 5'Fr and 3'Fr RNAs are only slowly degraded in the xrn1 ${Delta}$ mutant strain (right panel). (D) Rapid removal of the recombination substrates inhibits viral RNA recombination in the parental strain. However, the abundant 3'Fr RNAs will participate in replicase-driven recombination events in the xrn1 ${Delta}$ mutant strain. Note that 5'Fr RNAs lack important cis-acting replication sequences (34, 37, 38) and do not seem to participate in RNA recombination.

General conclusions. We have dissected for the first time the roles of host factorsin viral RNA recombination, which is a major process in virusevolution. This study establishes that the Xrn1p 5'-3' exoribonucleaseacts as a suppressor of viral RNA recombination. In addition,the combined activities of the Ngl2p endoribonuclease and Xrn1pcontrol the amounts of partially degraded replicon RNAs, whichaffect RNA recombination by serving as substrates. Thus, ourresults suggest that host-mediated viral RNA turnover is a majorfactor in viral RNA recombination.

ACKNOWLEDGMENTS

We thank Judit Pogany, Andy White, and Saulius Serva for valuablecomments and A. W. Johnson for the purified Xrn1p preparationand plasmid pAJ10.

This work was supported by the National Science Foundation (EPS-0132295)and the Kentucky Tobacco Research and Development Center.

FOOTNOTES

* Corresponding author. Mailing address: Department of Plant Pathology, University of Kentucky, Plant Science Building, Lexington, KY 40546. Phone: (859) 257-7445. Fax: (859) 323-1961. E-mail: pdnagy2{at}uky.edu.

These authors contributed equally.

REFERENCES

Top