Virology 327 (2004) 196 – 205 www.elsevier.com/locate/yviro

Subgenomic RNA as a riboregulator: negative regulation of RNA replication by Barley yellow dwarf virus subgenomic RNA 2 Ruizhong Shen, W. Allen Miller* Interdepartmental Genetics Program and Department of Plant Pathology, Iowa State University, Ames, IA 50011, United States Received 7 April 2004; accepted 15 June 2004 Available online 6 August 2004

Abstract Barley yellow dwarf virus (BYDV) generates three 3V-coterminal subgenomic RNAs (sgRNAs) in infected cells. Translation of BYDV genomic RNA (gRNA) and sgRNA1 is mediated by the BYDV cap-independent translation element (BTE) in the 3V untranslated region. sgRNAs 2 and 3 are unlikely to be mRNAs. We proposed that accumulation of sgRNA2, which contains the BTE in its 5V UTR, regulates BYDV replication by trans-inhibiting translation of the viral polymerase from genomic RNA (gRNA). Here, we tested this hypothesis and found that: (i) co-inoculation of the BTE or sgRNA2 with BYDV RNA inhibits BYDV RNA accumulation in protoplasts; (ii) Brome mosaic virus (BMV), engineered to contain the BTE, trans-inhibits BYDV replication; and (iii) sgRNA2 generated during BYDV infection transinhibits both GFP expression from BMV RNA and translation of a non-viral reporter mRNA. We conclude that sgRNA2, via its BTE, functions as a riboregulator to inhibit translation of gRNA. This may make gRNA available as a replicase template and for encapsidation. Thus, BYDV sgRNA2 joins a growing list of trans-acting regulatory RNAs. D 2004 Elsevier Inc. All rights reserved. Keywords: sgRNA; Barley yellow dwarf virus; BTE

Introduction Most RNA viruses replicate in the cytoplasm of their host cells. Thus, translation rather than transcription is often the major step at which viral gene expression is regulated. Many viral translational control strategies are conferred by RNA structures in cis (Gale et al., 2000; Macdonald, 2001; Mazumder et al., 2003). In recent years, diverse regulatory RNAs known as riboregulators have been discovered in prokaryotes and eukaryotes (Lease and Belfort, 2000; Rastinejad et al., 1993; Reinhart et al., 2000). Riboregulators function in trans, mainly post-transcriptionally. Only a few trans-regulatory RNAs from RNA viruses have been reported (Albarino et al., 2003; Eckerle and Ball, 2002; Sit et al., 1998). Here, we show that subgenomic RNA 2 * Corresponding author. Interdepartmental Genetics Program and Department of Plant Pathology, 351 Bessey Hall, Iowa State University, Ames, IA 50011. Fax: +1 515 294 9420. E-mail address: [email protected] (W.A. Miller). 0042-6822/$ - see front matter D 2004 Elsevier Inc. All rights reserved. doi:10.1016/j.virol.2004.06.025

(sgRNA2) of Barley yellow dwarf virus (BYDV) acts as a riboregulator to negatively regulate viral replication in trans. BYDV is the type member of genus Luteovirus in the family Luteoviridae. BYDV RNA has a complex set of primary and secondary structures that regulate many noncanonical translation events (Miller et al., 2002). These include cap-independent translation (Guo et al., 2001; Wang et al., 1997, 1999), 1 ribosomal frameshifting (Barry and Miller, 2002; Paul et al., 2001), leaky scanning (Chay et al., 1996), and stop codon readthrough (Brown et al., 1996). BYDV has a positive sense RNA genome of 5677 nt that encodes six open reading frames (ORFs) (Miller et al., 1997, 2002). Its genomic RNA (gRNA) and subgenomic RNAs (sgRNAs) have no 5V-cap and no 3V-poly(A) tail (Allen et al., 1999), yet they are translated efficiently. The 105-nt capindependent translation element (TE) in the 3V untranslated region (UTR) of BYDV RNA facilitates efficient translation initiation at the 5V-proximal AUG (Guo et al., 2000). Similar structures are present in the 3V UTRs of necrovi-

R. Shen, W.A. Miller / Virology 327 (2004) 196–205

ruses (Meulewaeter et al., 2004; Shen and Miller, 2004) and dianthoviruses (Mizumoto et al., 2003), so we now refer to this element as a member of the BYDV-like class of TE, or BTE (Shen and Miller, 2004). The BTE (i) binds translation factors (E. Allen and E. Pettit, personal communication) that presumably recruit the ribosome, (ii) is brought into proximity with the 5V end by direct base pairing to the 5V UTR (Guo et al., 2001), and (iii) functions both in the 5V UTR and in the 3V UTR (Guo et al., 2000). Most of the 869-nt 3V UTR of BYDV gRNA is required for full cap-independent and poly(A) tail-independent translation in oat protoplasts (Guo et al., 2000; Wang et al., 1999). sgRNA2 corresponds to the 869-nt 3V UTR of BYDV RNA and the BTE is at the 5V end of sgRNA2 (Fig. 1). sgRNA2 encodes a small ORF (ORF 6) that varies from 4.3 to 7.2 kDa and is poorly conserved between isolates. It is absent in BYDV-related Soybean dwarf (Rathjen et al., 1994) and Bean leafroll (Domier et al., 2002) viruses. After much effort to detect the product of ORF 6 (P6) or translatability of sgRNA2, it appears that ORF 6 is not translated in vivo (A. Rakotondrafara, personal communication). In addition to conferring cap-independent translation in cis, the BTE inhibits translation in trans, in vitro (Wang et al., 1997, 1999). In wheat germ extract, either the BTE alone or full-length sgRNA2, which harbors the BTE at its 5V end, trans-inhibit translation of BYDV genomic RNA (gRNA) and (to a much lesser extent) sgRNA1 (Wang et al., 1999). The inhibition does not require translation of ORF 6, but does require a functional BTE. Based on these in vitro data, we proposed that in later stages of the virus replication cycle, accumulation of sgRNA2 inhibits translation of RNA-dependent RNA polymerase (RdRp) from gRNA while allowing translation of structural proteins from sgRNA1. Thus, viral RNA replication would be inhibited by sgRNA2 introduced prematurely in the replication cycle. Here, we tested this hypothesis. We found that both

Fig. 1. BYDV genome organization. Boxes represent open reading frames (ORFs) with the sizes of encoded proteins indicated in kilodaltons (K). Black lines represent genomic RNA (gRNA) and subgenomic RNAs (sgRNAs). The 3V UTR needed for full cap- and poly(A)-independent translation in vivo is located between two dashed lines. Gray boxes represent the in vitro-defined BTE.

197

replicating and nonreplicating RNAs containing the BTE inhibit BYDV RNA accumulation in trans. sgRNA2 generated during BYDV infection trans-inhibits translation of a reporter gene and gene expression from BMV RNA, which suggests that BYDV sgRNA2 inhibits viral replication via inhibition of translation. We also showed that the BTE in cis increases translation of capped RNA. Our data suggest that sgRNA2, via its BTE, functions as a riboregulator to negatively control translation of the viral RdRp, thus replication of BYDV RNA.

Results Nonreplicating BTE and sgRNA2 RNAs trans-inhibit accumulation of BYDV RNAs in infected plant cells In wheat germ extract, both the 105-nt BTE and sgRNA2, which harbors the BTE at its 5V end, trans-inhibit the translation of BYDV genomic RNA (gRNA) and sgRNA1 (Wang et al., 1999). In natural infection, the molar ratio of sgRNA2 to sgRNA1 and gRNA is similar to the ratio that strongly inhibits translation of gRNA and weakly inhibits translation of sgRNA1 in vitro. Thus, we predict that addition of excess BTE or sgRNA2 during inoculation with BYDV RNA should inhibit BYDV replication via premature inhibition of translation of the RdRp from genomic RNA. To test this prediction, we co-inoculated oat protoplasts with the 105-nt BTE (TE105) or the 109-nt nonfunctional mutant BTE (TEBF) transcripts and wild-type infectious BYDV transcript, PAV6. TEBF contains a GAUC duplication in the BamHI4837 site that completely abolishes cap-independent translation mediated by the BTE in cis (Wang et al., 1997). The accumulated BYDV gRNA and sgRNA levels at 24 h post-inoculation (hpi) were detected by northern blot hybridization. When co-inoculated with PAV6 RNA into oat protoplasts, nonreplicating TE105 RNA trans-inhibited BYDV RNA accumulation including the sgRNAs (Fig. 2A, lanes 1 and 3). The defective mutant TEBF did not inhibit PAV6 replication (Fig. 2A, lane 4). The trans-inhibitory effects of the BTE were dosedependent (Fig. 2B, lanes 3–7). As low as 2.5-fold excess BTE RNA reduced BYDV RNA accumulation (Fig. 2B, lane 3). Ten-fold excess BTE almost abolished BYDV RNA accumulation (Fig. 2A, lanes 1 and 3; Fig. 2B, lane 5). Fulllength sgRNA2 and its counterpart containing the TEBF mutation, sgRNA2BF, had similar effects as TE105 and TEBF RNAs, respectively, when co-electroporated with PAV6 (data not shown). Thus, up to a 10-fold excess of nonreplicating BTE or sgRNA2 trans-inhibited accumulation of BYDV RNA as predicted. However, we were surprised to find that when we increased the molar ratio of BTE/PAV6 to z20:1, replication of BYDV RNA was partially recovered (Fig. 2B, lanes 6 and 7). These results were highly reproducible in many experiments using different RNA and protoplast preparations.

198

R. Shen, W.A. Miller / Virology 327 (2004) 196–205

monitor gene expression and to avoid complications caused by encapsidation, the BMV coat protein ORF was replaced with that of GFP (Fig. 3). The BTE or TEBF sequence from BYDV was inserted into the intergenic region between the 3a and coat protein genes of BMV RNA 3, just upstream of the GFP ORF start codon (Fig. 3). This places the BTE in the 3V UTR of the 3a gene on RNA 3 and in the 5V UTR of the GFP-encoding subgenomic RNA 4. The resulting viruses were designated BMV.TEGFP and BMV.TEBFGFP (Fig. 3). We inoculated oat protoplasts simultaneously with BYDV PAV6, and BMV RNAs 1 + 2 combined with various RNA 3 transcripts (Fig. 3) that harbor the BTE or its nonfunctional counterpart, TEBF. BMV RNAs 1, 2, and 3

Fig. 2. Effects of nonreplicating BTE on BYDV RNA accumulation. Fulllength infectious BYDV transcripts from pPAV6 were co-inoculated with in vitro transcripts of BTE or TEBF (BTE with a BamHI fill-in mutation) RNA into oat protoplasts. After 24-h incubation, total RNA was extracted and analyzed by Northern blot hybridization. gRNA and sgRNAs are indicated. (A) Effects of a 10-fold molar excess of BTE or TEBF on BYDV RNA accumulation. Inoculum in lanes 1 and 3: PAV6 + 10-fold excess BTE; lane 2: PAV6 only; lane 4: PAV6 + 10-fold excess TEBF RNA. The bottom panel shows an ethidium bromide-stained gel to indicate amount of total RNA loaded in each lane. (B) Effects of increasing molar ratios of BTE on PAV6 replication. Inoculum in lane 1: no RNA; lane 2: PAV6; lanes 3–7: PAV6 + 2.5-, 5-, 10-, 20-, and 40-fold molar excess BTE, respectively.

Replicating BTE trans-inhibits accumulation of BYDV RNA We next examined the trans-effects of the BTE in a replicating context, but still isolated from other potential regulatory elements in BYDV RNA. To do this, we developed an expression system from an unrelated virus, Brome mosaic virus (BMV). BMV is a tripartite RNA virus in the Bromoviridae family with three genomic RNAs. Only RNAs 1 and 2 are required for viral RNA replication. RNA3 encodes two ORFs including the coat protein gene. The coat protein mRNA, subgenomic RNA 4, is generated from RNA3 during infection (Miller et al., 1985). All BMV RNAs are capped (Dasgupta et al., 1975), so BMV has no apparent need for a cap-independent translation element. To

Fig. 3. Genome organization of Brome mosaic virus constructs. Boxes represent ORFs with the gene names indicated above. CP, coat protein; GFP, green fluorescent protein. Black ovals indicate 5V cap. Cloverleaves indicate the 3V tRNA-like structure. Arrows show synthesis of the subgenomic RNA (RNA 4). Gray boxes indicate the inserted BTE of BYDV. The sequence of BMV subgenomic core promoter, the secondary structure of the BTE, and start codon of GFP gene (underlined) are shown in the dotted box. The GAUC that is duplicated in the BamHI fill-in mutation is boxed.

R. Shen, W.A. Miller / Virology 327 (2004) 196–205

199

were inoculated in a molar ratio of 1:1:2. To ensure that the effects were conferred specifically by the BTE, we also included tRNA and wild-type BMV.GFP RNA (no BTE) as controls. When co-inoculated with BYDV PAV6, BMV.TEGFP and BMV.TEBFGFP RNAs replicated similarly, as revealed in northern blot hybridization (Fig. 4A, lanes 3,4). The BYDV-specific probe detected the engineered BMV RNA3 accumulation due to the presence of the

Fig. 5. (A) Effects of the BTE in cis on expression of GFP from BMV. GFP fluorescence intensities were measured by using flow cytometry. Mock: oat protoplasts were electroporated without RNA; hpi, hours post-inoculation. Vertical bars indicate standard deviation. Each value is a mean of at least three replicates. (B) Accumulation of recombinant BMV RNAs 3 and 4. Total RNA was extracted from oat protoplasts 24 h after inoculation with BMV.GFP, BMV.TEGFP, or BMV.TEBFGFP, and used for northern blot hybridization. A 32P-labeled probe complementary to the full-length GFP gene sequence was used to detect recombinant BMV RNAs 3 and 4. The bottom panel shows the RNA loading control.

Fig. 4. Effects of BTE-containing BMV replication on BYDV RNA accumulation. Northern blot analyses were done as in Fig. 2. BYDV gRNA and sgRNAs are indicated. BMV RNAs 3 and 4 were also detected by the BYDV-derived probe because they contain the BTE. Below each blot, stained gels show relative loading of total RNA in each lane. Inoculum (A) in lane 1: no RNA; lane 2: PAV6; lane 3: PAV6 + 4 Ag of BMV.TEGFP; lane 4: PAV6 + 4 Ag of BMV.TEBFGFP; lane 5: PAV6 + 4 Ag of tRNA; lane 6: PAV6 + 4 Ag of BMV.GFP. (B) Effects of increasing BMV RNA inocula on BYDV RNA accumulation. Lane 1: PAV6; lanes 2–5: PAV6 + 1, 2, 4, and 8 Ag of BMV.TEGFP, respectively.

BTE or TEBF sequences. Replicating BMV.TEGFP inhibited accumulation of BYDV RNA (Fig. 4A, lane 3), whereas BMV.TEBFGFP (Fig. 4A, lane 4), tRNA (lane 5), and BMV.GFP (lane 6) did not inhibit BYDV RNA accumulation as much, if at all. The inhibitory effects conferred by the BTE from replicating BMV.TEGFP were dose-dependent (Fig. 4B). When the co-inoculated BMV.TEGFP RNA was increased from 1 to 4 Ag, the amounts of BYDV gRNA and sgRNAs decreased (Fig. 4B, lanes 2–4). However, when co-inoculated with 8 Ag of BMV.TEGFP, BYDV RNA accumulation was inhibited less (Fig. 4B, lane 5). Subgenomic RNA 2 produced during BYDV infection trans-inhibits gene expression of BMV RNA containing or lacking the BTE Having established that the BTE and sgRNA2 transinhibit BYDV RNA replication in vivo, we set out to test the

200

R. Shen, W.A. Miller / Virology 327 (2004) 196–205

mechanism of the inhibition. Based on our in vitro data (Wang et al., 1999), we proposed that BTE or sgRNA2, when co-inoculated with BYDV RNA, attenuate replication via premature inhibition of translation of the RdRp from genomic RNA. Because of the difficulty of detecting RdRp, we used BMV.GFP and BMV.TEGFP as sensors to test whether sgRNA2 could trans-inhibit translation in vivo. Unexpectedly, we found that the BTE enhanced BMV gene expression in cis (Figs. 5A, 6C). By using flow cytometry and UV microscopy, we found that GFP expression levels from BMV.TEGFP were higher than those from BMV.GFP (Figs. 5A, 6C). The TEBF leader, which differs from the BTE by only four bases, reduced GFP expression to near background levels (Fig. 5A). This may be caused by the secondary structure in the TEBF impeding ribosome scanning to the start codon. BMV.TEGFP-infected cells fluoresced more brightly than BMV.GFP-infected cells (Fig. 6C),

and the BTE increased the number of cells expressing detectable levels of GFP. The percentage of oat protoplasts with green fluorescence was 6.5% (F3.3%) in BMV. TEGFP-inoculated cells, 2.2% (F0.7%) in the BMV.GFPinoculated group, and 0.25% (F0.12%) in the BMV. TEBFGFP-inoculated group. Standard deviations from at least three independent experiments are included in parentheses. For each independent experiment, the GFP expression levels from BMV.TEGFP were 2- to 5-fold higher than those from BMV.GFP. The insertions of BTE and TEBF sequences had little effect on BMV.GFP RNA replication and synthesis of BMV RNA 4 (Fig. 5B). This result agrees with previous reports that insertion of a foreign gene within 17 bases downstream of the RNA 4 start site did not greatly affect subgenomic RNA synthesis (French et al., 1986). Because presence of the BTE had little effect on BMV RNA accumulation (Fig.

Fig. 6. Effects of subgenomic RNA2 (sgRNA2) expressed from BYDV, in trans, on expression of GFP from BMV. (A) BYDV RNA was co-inoculated with BMV.GFP transcripts into oat protoplasts. At different time points post-inoculation, a portion of cells was collected and GFP fluorescence intensities were measured by using flow cytometry. PAV6, wild-type BYDV; PAV6DSG2, one base mutation (G4810C) of PAV6 that knocks out sgRNA2 synthesis; hpi, hours post-inoculation. Each point is the mean of at least three replicates. Vertical bars indicate standard deviation. (B) Accumulation of PAV6 and PAV6DSG2 progeny RNA in oat protoplasts. Cells were collected at the indicated time points post-inoculation, total RNA was extracted and used for northern blot hybridization. gRNA and sgRNAs are indicated. The bottom panel shows the RNA loading. The decrease in viral RNA at 72 hpi is due to a decrease in viable protoplasts. (C) Oat protoplasts infected with BMV.GFP, BMV.TEGFP alone, or with PAV6 or PAV6DSG2. Pictures were taken under UV microscopy 24 hpi. (D) RNA accumulation of BMV.GFP and BMV.TEGFP in oat protoplasts. Total RNA was extracted from protoplasts in panel C at 24 hpi and used for Northern blot hybridization. 32P-labeled probe complementary to full-length GFP gene sequence was used to detect recombinant BMV RNAs 3 and 4. The bottom panel shows RNA loading control.

R. Shen, W.A. Miller / Virology 327 (2004) 196–205

5B), but increased GFP expression, these data indicate that the insertion of the BTE in the 5V UTR of non-BYDV (i.e. BMV) mRNA increased translation. Thus, in addition to providing cap-independent translation, the BTE also increases the translation of capped mRNAs. To test whether sgRNA2 can trans-inhibit translation in vivo, we co-inoculated BMV.GFP RNA with wild-type BYDV infectious transcript PAV6, or with a mutant, PAV6DSG2 RNA. PAV6DSG2 contains a point mutation (G4810C) that knocks out sgRNA2 synthesis but has little effect on accumulation of the other BYDV RNAs (Koev and Miller, 2000). When co-inoculated with BMV.GFP RNA, wild-type PAV6 reduced GFP expression from BMV.GFP by 2- to 6-fold (Figs. 6A, C). PAV6DSG2 was less inhibitory (Figs. 6A, C). The degrees of inhibition of GFP expression by PAV6 were similar at different time points (Fig. 6A). The reason for the reduced inhibition by PAV6DSG2 was not due to reduced replication of PAV6DSG2 relative to PAV6. Both PAV6DSG2 and PAV6 replicated similarly in the presence of BMV.GFP (Fig. 6B). The experiments above show that sgRNA2 inhibits expression of RNA lacking the BTE. To examine whether sgRNA2 inhibits translation of BTE-containing RNA in a BYDV infection, we co-inoculated oat protoplasts with BMV.TEGFP RNA and either PAV6 or PAV6DSG2 RNA. PAV6 reduced GFP expression from BMV.TEGFP substantially, whereas PAV6DSG2 only slightly reduced the expression level of GFP (Fig. 6C). Northern blot hybridization revealed that PAV6 and PAV6DSG2 also inhibited accumulation of RNAs 3 and 4 of BMV.GFP and BMV.TEGFP (Fig. 6D), and that PAV6 showed higher inhibitory effects than PAV6DSG2 did (Fig. 6D). Thus, accumulated sgRNA2 in BYDV-infected cells trans-inhibits GFP expression from BMV RNAs that contain or lack a BTE. These experiments do not differentiate whether the reduced GFP levels are due to inhibition of translation, transcription, or replication of the GFP mRNA, or a combination of these events. Subgenomic RNA2 in BYDV-infected cells inhibits translation of reporter mRNA To determine whether the inhibition of gene expression by BYDV sgRNA2 is at the level of translation, we tested the effect of BYDV infection on translation of a nonreplicating reporter mRNA construct in oat protoplasts. A two-step electroporation method was developed. First oat protoplasts were inoculated with infectious BYDV PAV6 or PAV6DSG2 RNA by electroporation. After 24-h incubation to allow genomic RNA replication and accumulation of sgRNAs, protoplasts were electroporated again with reporter cap-fLuc-A(60). This is a capped and polyadenylated firefly luciferase gene lacking any viral sequence. The firefly luciferase activity was analyzed after another 4-h incubation. Inoculation of 1 pmol of PAV6 RNA in the first step caused a 60% drop in translation of cap-fLuc-A(60), whereas

201

Fig. 7. (A) Differential effects of PAV6 and PAV6DSG2 replication on translation of reporter construct cap-fLuc-A(60). Twenty-four hours after inoculation of PAV6 or PAV6DSG2 RNA, oat protoplasts were reelectroporated with 1 pmol cap-fLuc-A(60) transcript. Luciferase activities were analyzed 4 h later. (B) Northern blot analysis of replication of PAV6 and PAV6DSG2. Lane 1, PAV6. Lane 2, PAV6DSG2.

inoculation of PAV6DSG2 RNA in first step had no effect on translation (Fig. 7A). Again, PAV6 and PAV6DSG2 RNAs accumulated to similar levels (Fig. 7B). Thus, the inhibition of gene expression by BYDV sgRNA2 most likely functions at the level of translation.

Discussion Subgenomic RNA2 trans-inhibits the accumulation of BYDV RNA Subgenomic RNAs of positive-sense RNA viruses in many plant virus families and the Nidovirales class and Togaviridae and Nodaviridae families of animal viruses all have been considered to be messenger RNAs required for expression of 3V-proximal viral genes. In this report, we show that sgRNA2 of BYDV has a different function, acting as a trans-inhibitor of RNA replication via the BTE in its 5V end. In three different contexts, nonreplicating 105-nt BTE RNA, replicating BMV RNA, and expressed as sgRNA2 in natural BYDV infection, the BTE inhibited viral RNA accumulation in trans. RNAs containing the nonfunctional TEBF sequence that differs from BTE by only a four base

202

R. Shen, W.A. Miller / Virology 327 (2004) 196–205

duplication did not inhibit in trans. In a natural infection, the trans-function of BTE is fulfilled in the context of sgRNA2, which is not detectable until about 10 hpi (data not shown), after translation of RdRp from gRNA. Thus, we propose that BYDV sgRNA2 trans-inhibits translation of BYDV RNAs via its BTE to act as a switch that turns off translation of gRNA. When introduced artificially at the very beginning of infection (co-inoculated with gRNA), sgRNA2 prevents the initial translation of gRNA, blocking infection all together. These results reveal that viral subgenomic RNAs do not always serve as mRNAs, and instead can perform important regulatory functions. The inhibitory effects of the BTE and sgRNA2 on BYDV replication and transcription were dose-dependent (Figs. 2B, 4B). The dose dependency up to 10-fold excess supports our hypothesis that BTE and sgRNA2 transinhibited translation of capped and uncapped mRNAs by competing for translation initiation factor(s) (Wang et al., 1997). The BTE inhibits translation of capped, polyadenylated non-viral RNA in vitro (Wang et al., 1997) and in vivo (Fig. 7). Added eIF4F reversed the trans-inhibition effect caused by BTE in vitro (Wang et al., 1997), and the BTE specifically binds eIF4F and eIFiso4F (E. Allen, E. Pettit, and W.A. Miller, unpublished data). Moreover, our preliminary data suggest that sgRNA2 (via its BTE) inhibits translation of cellular mRNAs (R. Shen, W. Staplin, unpublished data). Surprisingly, the replication of BYDV was not inhibited as much when the molar ratio of BTE/PAV6 was increased to 20:1 and 40:1. One possible explanation for this result is that, at these higher concentrations, the BTE may base pair to the viral 5VUTR by the kissing stem-loop interaction in trans instead of in cis. Thus, high concentrations of added BTE may stimulate translation by delivering translation factors to the 5VUTR in trans. This process would be possible only if the BTE in the 105 nt or sgRNA2 5V UTR context has a higher affinity for factors than the BTE in the 3V UTR context, so that at the highest concentrations of added BTE, there are still free factors available to bind BTE105 but which are too low in concentration to bind the BTE within BYDV genomic RNA to stimulate translation in cis. A second possibility for the decreased inhibition at high BTE/sgRNA2 is that, at the highest concentrations, the BTE may not fold into a functional secondary structure, preventing trans-inhibition that occurs at lower concentrations. It is noteworthy that the dose–response curve resembles the bbell-shapedQ double-stranded RNA activated PKR antiviral response in mammals. dsRNA induces the response, but excessively high concentrations of dsRNA prevent dimerization of PKR necessary for autophosphorylation to initiate shut-off of translation (Davis and Watson, 1996; Hunter et al., 1975). A PKR system may also exist in plants (Bilgin et al., 2003) so we cannot rule out the possibility that the BTE may induce a PKR-like translational shutdown, as is the case for the highly structured 3V UTRs of some tumor-suppressing genes (Nussbaum et al., 2002).

Subgenomic RNA 2 trans-inhibits gene expression from RNAs containing or lacking the BTE Wild-type PAV6 trans-inhibited GFP expression from an unrelated virus, BMV with or without the BTE, whereas PAV6DSG2 did not (Figs. 6A and C). Thus, the decreased expression levels of GFP were caused by BYDV sgRNA2. There are at least two explanations for the differential effect of PAV6 and PAV6DSG2 on the translation of GFP from BMV. The first is that specific BTE secondary structure present only in the sgRNA2 context, but not gRNA and sgRNA1 due to position effect, and is required for the transinhibition function. A more likely reason for the lack of transinhibition by the BTE in the gRNA and sgRNA1 contexts (PAV6DSG2) is that gRNA and sgRNA1 are normally present at N10-fold lower concentrations than sgRNA2. Furthermore, most gRNA may be sequestered in virions. Feedback regulation of BYDV gene expression by its sgRNA2 The correlation between the stimulatory function of the BTE in cis and ability to inhibit virus replication in trans provides strong evidence that the same factors are used in trans-inhibition and cis-stimulation of translation. Competition studies showed that both sgRNA2 and the BTE transinhibited translation of gRNA in vitro (Wang et al., 1999). Here, we showed that sgRNA2 trans-inhibited translation of reporter mRNA (Fig. 7) and GFP expression from BMV in vivo (Fig. 6). These data suggest that the BTE and sgRNA2 trans-inhibit BYDV RNA replication and transcription by inhibiting translation of genomic RNA which prevents the production of the RdRp. However, we cannot rule out the possibility that sgRNA2 may also trans-inhibit RNA replication or transcription directly. Combined with previously reported results, we propose a feedback regulation mechanism (Fig. 8): in the earliest stage of BYDV infection, viral RdRp is translated via BTEmediated cap-independent translation of gRNA (Stage 1, Fig. 8). The RdRp then carries out viral RNA replication and sgRNA synthesis (Stage 2, Fig. 8). Viral RNAs accumulate with sgRNA2 becoming particularly abundant. Accumulated sgRNA2, via its 5V BTE, trans-inhibits translation of BYDV RdRp from gRNA (Stage 3, Fig. 8). This switches gRNA from its mRNA function to its replicase template function and also allows it to be encapsidated, that is, gRNA is available for replication by the existing RdRp and for encapsidation (Stage 4, Fig. 8). It is noteworthy that BYDV RNA accumulates in the absence of sgRNA2 (Figs. 6B, 7B, PAV6DSG2), thus sgRNA2 is not essential for virus replication at least for 72 h in protoplasts. Moreover, the mechanism requires replication before the sgRNA2-mediated switch-off of translation. Thus, sgRNA2 may serve subtle regulatory roles, such as enhancing a different translational switch-off mechanism mediated by the replicase itself, that was

R. Shen, W.A. Miller / Virology 327 (2004) 196–205

Fig. 8. Model for feedback regulation of BYDV gene expression. In the early stage of BYDV infection, viral RNA-dependent RNA polymerase (RdRp) is produced via BTE-mediated cap-independent translation of gRNA (stage 1). The RdRp then carries out viral RNA replication and sgRNA synthesis (stage 2). Viral RNAs accumulate with sgRNA2 being particularly abundant. Accumulated sgRNA2, via its BTE, trans-inhibits translation of RdRp from gRNA (stage 3). Genomic RNA switches from translation to replication and encapsidation, that is, gRNA is available for replication by the existing RdRp and for encapsidation (stage 4). ORF 1 and 2: RdRp. ORF 3: coat protein (CP). ORF 4: movement protein. ORF 5: readthrough domain (RT).

proposed previously (Barry and Miller, 2002). The effect of sgRNA2 on fitness of BYDV in plants and in the field remain to be tested. A new regulatory role for a viral subgenomic RNA We show here that RNA harboring the BTE transinhibits translation of other BTE-containing RNAs in vivo, as well as translation of RNA lacking the BTE. Thus, the BTE serves as a riboregulator, as proposed from previous in vitro translation experiments (Wang et al., 1999). Substantial evidence indicates that sgRNA2 is not translatable in vivo (A. Rakotondrafara, personal communication). In vitro data also showed that inhibition of translation by sgRNA2 does not require expression of ORF 6 (Wang et al., 1999). Thus, if sgRNA2 is not translatable, it must serve a nonmRNA function, such as a riboregulator. Other trans-regulatory RNAs of RNA viruses have been reported. A 34-nt sequence in RNA2 of Red clover necrotic mosaic virus trans-activates synthesis of sgRNA from RNA1 by base pairing to RNA1 (Sit et al., 1998). Replication of FHV RNA2 is dependent on the synthesis of subgenomic RNA from RNA1 (Albarino et al., 2003; Eckerle and Ball, 2002). FHV RNA2 then down-regulates synthesis of the subgenomic RNA from RNA1 (Zhong and Rueckert, 1993). With regard to trans-regulation of translation, Adenovirus virus-associated (VA) RNAs (Mathews and Shenk, 1991) and Epstein–Barr virus EBER RNAs (Bhat and

203

Thimmappaya, 1983; Clarke et al., 1990) protect against dsRNA-activated inhibitor (DAI)-mediated phosphorylation of eIF-2a by binding DAI (Sharp et al., 1993). Like BYDV sgRNA2, VA and EBER RNAs are nontranslated. As a 3V UTR-derived trans-inhibitor of translation, sgRNA2 sequence resembles the cellular tumor suppressor genes for which the term riboregulator was first coined. The 3V UTRs of alpha-tropomyosin (Rastinejad et al., 1993) and other (Manjeshwar et al., 2003) mRNAs alone have tumor suppressor activity. The former acts by PKR-mediated inhibition of translation (Davis and Watson, 1996). More recently, large classes of noncoding microRNAs (miRNA) have been found that regulate gene expression either by inducing mRNA degradation (RNAi) or by blocking translation (Bartel, 2004; Carrington and Ambros, 2003). Recently, Epstein–Barr virus has been shown to express miRNAs that target both host and viral mRNAs (Pfeffer et al., 2004). While its mechanism of action remains to be fully elucidated, sgRNA2 of BYDV appears to be a new example of a trans-regulatory RNA.

Materials and methods Plasmids Infectious BYDV-PAV genomic RNA was transcribed from the full-length clone, pPAV6 (Di et al., 1993). The sgRNA2 knockout mutant clone of BYDV-PAV, pPAV6DSG2, referred to previously as SG2G/C (Koev and Miller, 2000), differs from pPAV6 by a G to C mutation at position 4810, which prevents sgRNA2 synthesis. pTE and pTEBF are clones for T7 transcription of the 105-nt BTE RNA (TE105) and its nonfunctional mutant TEBF (Wang et al., 1997). pSG2 and pSG2BF allow T7 transcription of the 869-nt sgRNA2 and its nonfunctional mutant sgRNA2BF, respectively (Wang et al., 1999). Both pTEBF and pSG2BF contain a GATC duplication in the BamHI site in the BTE. This duplication destroys the cap-independent translation function of the BTE (Wang et al., 1997, 1999). BMV RNA clones were kindly provided by A.L.N. Rao (University of California, Riverside). pT7B1, pT7B2, and pT7B3 are clones for T7 transcription of BMV RNA1, RNA2, and RNA3, respectively (Dreher et al., 1989). pT7B3EGFP is a clone of BMV RNA3 with the coat protein gene replaced by an enhanced green fluorescent protein (GFP) gene (Rao, 1997). To construct pT7B3TEGFP for T7 transcription of BMV.TEGFP RNA3, the 109-nt fragment corresponding to the BTE (nt 4809–4918) was amplified from pPAV6 by PCR using the upstream primer, 5VG G A G AT C TAT G T C C TA AT T C A G C G TAT TA ATA GTGAAGACAACACCA-3V, and the downstream primer, 5V-CCTGAAGTCGAC ATTCGGCCAAACACAATACGATA-3V. The PCR products were cut with BglII and SalI (restriction sites are in italics), then ligated with pT7B3EGFP that had also been digested with BglII and SalI. The same

204

R. Shen, W.A. Miller / Virology 327 (2004) 196–205 32

strategy was used to clone pT7B3TEBFGFP, except the template for PCR was pSG2BF. The pT7B3TEGFP and pT7B3TEBFGFP constructs were verified by sequencing at the DNA Sequencing and Synthesis Nucleic Acid Facility of Iowa State University on an ABI377 sequencer (Applied Biosystems, Foster City, CA).

P-labeled probe complementary to the full-length GFP gene sequence RNA to detect recombinant BMV RNAs 3 and 4.

RNA preparation and infection of protoplasts

The authors thank A.L.N. Rao for providing pT7B3EGFP and infectious clones of BMV, and Andy Ball for helpful discussion. This research was funded by grants from USDA/ NRI (2001-35319-10011) and NIGMS (GM067104).

The capped and uncapped RNAs were synthesized by in vitro transcription by using the T7 mMESSAGE mMACHINE or MegaScript kits (Ambion, Austin, TX) as per manufacturer’s instructions. For transcription of infectious RNAs, BYDV constructs were linearized with SmaI to give a perfect genomic 3V end. pT7B1, pT7B2, pT7B3GFP, pT7B3TEGFP were linearized with BamHI. pT7B3TEGFP was linearized with Tth111I. Oat (Avena sativa cv. Stout) protoplasts were prepared and inoculated with RNA as described by Dinesh-Kumar and Miller (1993). Except when explicitly stated otherwise, 10 Ag of RNA transcript was used for BYDV inoculation and 4 Ag of BMV RNAs 1, 2, and 3 in a molar ratio of 1:1:2 were used for BMV inoculation. Two-step electroporation In the first step, oat protoplasts were inoculated with infectious BYDV PAV6 or PAV6 = C6SG2 RNA by electroporation as described by Dinesh-Kumar and Miller (1993), except that voltage was 280V. After incubation for 24 h at room temperature, cells were pelleted at 100  g for 4 min, washed in electroporation buffer, centrifuged again at 100  g for 4 min, then electroporated in the presence of the appropriate RNA at 280V. Analysis of GFP expression Oat protoplasts were analyzed for GFP expression 24, 48, 72, and 96 h after inoculation by flow cytometry by using an ELITE ESP fluorescence-activated cell sorter (BeckmanCoulter, Anaheim, CA) at the Cell and Hybridoma Facility of Iowa State University. All data presented in this report were obtained from at least three independent experiments. Northern blot hybridization Total RNA was extracted from protoplasts by using the RNeasy plant RNA isolation kit (QIAGEN, Los Angeles, CA) as per manufacturer’s instructions. Protoplasts were collected at indicated times post-inoculation, RNA extracted, and analyzed by Northern blot as described previously (Koev et al., 1999). A 32P-labeled probe complementary to the 1.5kb 3V-terminal sequence of BYDV-PAV RNA was used to detect BYDV gRNA and sgRNAs (Koev et al., 1999). Because of the low replication level of BMV RNAs 3 and 4 in oat protoplasts, we could hardly detect these two RNAs by using BMV tRNA-like structure probe. Instead, we used a

Acknowledgments

References Albarino, C.G., Eckerle, L.D., Ball, L.A., 2003. The cis-acting replication signal at the 3V end of Flock House virus RNA2 is RNA3-dependent. Virology 311, 181 – 191. Allen, E., Wang, S., Miller, W.A., 1999. Barley yellow dwarf virus RNA requires a cap-independent translation sequence because it lacks a 5V cap. Virology 253, 139 – 144. Barry, J.K., Miller, W.A., 2002. A programmed -1 ribosomal frameshift that requires base-pairing across four kilobases suggests a novel mechanism for controlling ribosome and replicase traffic on a viral RNA. Proc. Natl. Acad. Sci. U.S.A. 99, 11133 – 11138. Bartel, D.P., 2004. MicroRNAs: genomics, biogenesis, mechanism, and function. Cell 116, 281 – 297. Bhat, R.A., Thimmappaya, B., 1983. Two small RNAs encoded by Epstein–Barr virus can functionally substitute for the virus-associated RNAs in the lytic growth of adenovirus 5. Proc. Natl. Acad. Sci. U.S.A. 80, 4789 – 4793. Bilgin, D.D., Liu, Y., Schiff, M., Dinesh-Kumar, S.P., 2003. P58(IPK), a plant ortholog of double-stranded RNA-dependent protein kinase PKR inhibitor, functions in viral pathogenesis. Dev. Cell. 4, 651 – 661. Brown, C.M., Dinesh-Kumar, S.P., Miller, W.A., 1996. Local and distant sequences are required for efficient read-through of the barley yellow dwarf virus-PAV coat protein gene stop codon. J. Virol. 70, 5884 – 5892. Carrington, J.C., Ambros, V., 2003. Role of microRNAs in plant and animal development. Science 301, 336 – 338. Chay, C.A., Gunasinge, U.B., Dinesh-Kumar, S.P., Miller, W.A., Gray, S.M., 1996. Aphid transmission and systemic plant infection determinants of barley yellow dwarf luteovirus-PAV are contained in the coat protein readthrough domain and 17-kDa protein, respectively. Virology 219, 57 – 65. Clarke, P.A., Sharp, N.A., Arrand, J.R., Clemens, M.J., 1990. Epstein–Barr virus gene expression in interferon-treated cells-implications for the regulation of protein synthesis and the antiviral state. Biochim. Biophys. Acta 1050, 167 – 173. Dasgupta, R., Shih, D.S., Saris, C., Kaesberg, P., 1975. Nucleotide sequence of a viral RNA fragment that binds to eukaryotic ribosomes. Nature 256, 624 – 628. Davis, S., Watson, J.C., 1996. In vitro activation of the interferon-induced, double-stranded RNA-dependent protein kinase PKR by RNA from the 3V untranslated regions of human a-tropomyosin. Proc. Natl. Acad. Sci. U.S.A. 93, 508 – 513. Di, R., Dinesh-Kumar, S.P., Miller, W.A., 1993. Translational frameshifting by barley yellow dwarf virus RNA (PAV serotype) in Escherichia coli and in eukaryotic cell-free extracts. Mol. Plant-Microb. Interact. 6, 444 – 452. Dinesh-Kumar, S.P., Miller, W.A., 1993. Control of start codon choice on a plant viral RNA encoding overlapping genes. Plant Cell 5, 679 – 692. Domier, L.L., McCoppin, N.K., Larsen, R.C., D’Arcy, C.J., 2002. Nucleotide sequence shows that Bean leafroll virus has a Luteoviruslike genome organization. J. Gen. Virol. 83, 1791 – 1798.

R. Shen, W.A. Miller / Virology 327 (2004) 196–205 Dreher, T.W., Rao, A.L., Hall, T.C., 1989. Replication in vivo of mutant brome mosaic virus RNAs defective in aminoacylation. J. Mol. Biol. 206, 425 – 438. Eckerle, L.D., Ball, L.A., 2002. Replication of the RNA segments of a bipartite viral genome is coordinated by a transactivating subgenomic RNA. Virology 296, 165 – 176. French, R., Janda, M., Ahlquist, P., 1986. Bacterial gene inserted in an engineered RNA virus: efficient expression in monocotyledonoous plant cells. Science 231, 1294 – 1297. Gale Jr., M., Tan, S.L., Katze, M.G., 2000. Translational control of viral gene expression in eukaryotes. Microbiol. Mol. Biol. Rev. 64, 239 – 280. Guo, L., Allen, E., Miller, W.A., 2000. Structure and function of a capindependent translation element that functions in either the 3V or the 5V untranslated region. RNA 6, 1808 – 1820. Guo, L., Allen, E., Miller, W.A., 2001. Base-pairing between untranslated regions facilitates translation of uncapped, nonpolyadenylated viral RNA. Mol. Cell. 7, 1103 – 1109. Hunter, T., Hunt, T., Jackson, R.J., Robertson, H.D., 1975. The characteristics of inhibition of protein synthesis by double-stranded ribonucleic acid in reticulocyte lysates. J. Biol. Chem. 250, 409 – 417. Koev, G., Miller, W.A., 2000. A positive strand RNA virus with three very different subgenomic RNA promoters. J. Virol. 74, 5988 – 5996. Koev, G., Mohan, B.R., Miller, W.A., 1999. Primary and secondary structural elements required for synthesis of barley yellow dwarf virus subgenomic RNA1. J. Virol. 73, 2876 – 2885. Lease, R.A., Belfort, M., 2000. A trans-acting RNA as a control switch in Escherichia coli: DsrA modulates function by forming alternative structures. Proc. Natl. Acad. Sci. U.S.A. 97, 9919 – 9924. Macdonald, P., 2001. Diversity in translational regulation. Curr. Opin. Cell. Biol. 13, 326 – 331. Manjeshwar, S., Branam, D.E., Lerner, M.R., Brackett, D.J., Jupe, E.R., 2003. Tumor suppression by the Prohibitin gene 3V untranslated region RNA in human breast cancer. Cancer Res. 63, 5251 – 5256. Mathews, M.B., Shenk, T., 1991. Adenovirus virus-associated RNA and translation control. J. Virol. 65, 5657 – 5662. Mazumder, B., Seshadri, V., Fox, P.L., 2003. Translational control by the 3V-UTR: the ends specify the means. Trends Biochem. Sci. 28, 91 – 98. Meulewaeter, F., Van Lipzig, R., Gultyaev, A.P., Pleij, C.W., Van Damme, D., Cornelissen, M., Van Eldik, G., 2004. Conservation of RNA structures enables TNV and BYDV 5V and 3V elements to cooperate synergistically in cap-independent translation. Nucleic Acids Res. 32, 1721 – 1730. Miller, W.A., Dreher, T.W., Hall, T.C., 1985. Synthesis of brome mosaic virus subgenomic RNA in vitro by internal initiation on ( ) sense genomic RNA. Nature 313, 68 – 70. Miller, W.A., Brown, C.M., Wang, S., 1997. New punctuation for the genetic code: luteovirus gene expression. Semin. Virol. 8, 3 – 13.

205

Miller, W.A., Liu, S., Beckett, R., 2002. Barley yellow dwarf virus: Luteoviridae or Tombusviridae? Mol. Plant Pathol. 3, 177 – 183. Mizumoto, H., Tatsuta, M., Kaido, M., Mise, K., Okuno, T., 2003. Capindependent translational enhancement by the 3V untranslated region of red clover necrotic mosaic virus RNA1. J. Virol. 77, 12113 – 12121. Nussbaum, J.M., Gunnery, S., Mathews, M.B., 2002. The 3V-untranslated regions of cytoskeletal muscle mRNAs inhibit translation by activating the double-stranded RNA-dependent protein kinase PKR. Nucleic Acids Res. 30, 1205 – 1212. Paul, C.P., Barry, J.K., Dinesh-Kumar, S.P., Brault, V., Miller, W.A., 2001. A sequence required for -1 ribosomal frameshifting located four kilobases downstream of the frameshift site. J. Mol. Biol. 310, 987 – 999. Pfeffer, S., Zavolan, M., Grasser, F.A., Chien, M., Russo, J.J., Ju, J., John, B., Enright, A.J., Marks, D., Sander, C., Tuschl, T., 2004. Identification of virus-encoded mRNAs. Science 304, 734 – 736. Rao, A.L., 1997. Molecular studies on bromovirus capsid protein: III. Analysis of cell- to-cell movement competence of coat protein defective variants of cowpea chlorotic mottle virus. Virology 232, 385 – 395. Rastinejad, F., Conboy, M.J., Rando, T.A., Blau, H.M., 1993. Tumor suppression by RNA from the 3V untranslated region of a-tropomyosin. Cell 75, 1107 – 1117. Rathjen, J.P., Karageorgos, L.E., Habili, N., Waterhouse, P.M., Symons, R.H., 1994. Soybean dwarf luteovirus contains the third variant genome type in the luteovirus group. Virology 198, 571 – 579. Reinhart, B.J., Slack, F.J., Basson, M., Pasquinelli, A.E., Bettinger, J.C., Rougvie, A.E., Horvitz, H.R., Ruvkun, G., 2000. The 21-nucleotide let7 RNA regulates developmental timing in Caenorhabditis elegans. Nature 403, 901 – 906. Sharp, T.V., Schwemmle, M., Jeffrey, I., Laing, K., Mellor, H., Proud, C.G., Hilse, K., Clemens, M.J., 1993. Comparative analysis of the regulation of the interferon-inducible protein kinase PKR by Epstein–Barr virus RNAs EBER-1 and EBER-2 and adenovirus VAI RNA. Nucleic Acids Res. 21, 4483 – 4490. Shen, R., Miller, W.A., 2004. The 3Vuntranslated region of tobacco necrosis virus RNA contains a Barley yellow dwarf virus-like cap-independent translation element. J. Virol. 78, 4655 – 4664. Sit, T.L., Vaewhongs, A.A., Lommel, S.A., 1998. RNA-mediated transactivation of transcription from a viral RNA. Science 281, 829 – 832. Wang, S., Browning, K.S., Miller, W.A., 1997. A viral sequence in the 3Vuntranslated region mimics a 5V cap in facilitating translation of uncapped mRNA. EMBO J. 16, 4107 – 4116. Wang, S., Guo, L., Allen, E., Miller, W.A., 1999. A potential mechanism for selective control of cap-independent translation by a viral RNA sequence in cis and in trans. RNA 5, 728 – 738. Zhong, W., Rueckert, R.R., 1993. Flock house virus: down-regulation of subgenomic RNA3 synthesis does not involve coat protein and is targeted to synthesis of its positive strand. J. Virol. 67, 2716 – 2722.