Multi-tasking of nonstructural gene products is required for bean yellow dwarf geminivirus transcriptional regulation Kathleen L. Hefferon 1 , Yong-Sun Moon 2 and Ying Fan 3 1 Cornell University, Cornell Research Foundation, Ithaca, NY, USA 2 Yeungnam University, Department of Horticulture, Gyeongsan-si, Gyeongsangbuk-do, Korea 3 Cornell University, Cornell School of Veterinary Medicine, Ithaca, NY, USA Geminiviruses belong to a family of plant viruses that can be classified into four distinct genera on the basis of genomic organization, vector transmissibility and host range. These include the mastreviruses, which possess monopartite genomes, are transmitted by leaf- hoppers and infect monocotyledonous plants. Excep- tions to the rule are the Australian-derived tobacco yellow dwarf virus and South African-derived bean yellow dwarf virus (BeYDV), two distantly related mastreviruses that infect dicotyledenous plants [1]. BeYDV consists of a single-stranded circular DNA molecule of 2.6 kb in length, and contains four ORFs encoding three different genes. The coding region is divided bidirectionally by long intergenic regions (LIR) and short intergenic regions (SIR). The MP and CP genes are expressed from the virion sense- strand, while the replication-associated protein (Rep) is produced from overlapping ORFs C1 and C2 from the complementary sense-strand. An intron spans the region overlapping C1 and C2 and this is spliced dur- ing Rep expression. Both Rep, which functions as the replication-associated protein, and RepA, the gene product of ORF C1, are produced during virus infec- tion [1–3]. Keywords geminivirus; gene expression; promoter control; transactivation Correspondence K. L. Hefferon, University of Toronto, Center for Virology, 25 Willcocks St., Toronto, ON, Canada M5J3B2 Fax: +1 607 254 1015 Tel: +1 607 257 1081 E-mail: klh22@cornell.edu (Received 21 June 2006, accepted 7 August 2006) doi:10.1111/j.1742-4658.2006.05454.x Mastreviridae, of the family geminiviridae, possess a monopartite genome and are transmitted by leafhoppers. Bean yellow dwarf dirus (BeYDV) is a mastrevirus which originated from South Africa and infects dicoyledenous plants, a feature unusual for mastreviridae. Previously, the nonstructural proteins Rep and RepA were examined with respect to their independent roles in BeYDV replication. This was achieved by placing both gene pro- ducts under independent constitutive promoter control and examining their effects on replication-competent constructs. In the current study, Rep and RepA are examined further for their roles in regulating BeYDV gene expression using a series of replication-incompetent constructs. While both Rep and RepA are found to behave as equally potent inhibitors of comple- mentary-sense gene expression, they differ considerably with respect to their abilities to transactivate virion-sense gene expression. Furthermore, RepA is identified as playing more than one role in this transactivation process. A nuclear localization domain is identified in Rep which is absent in RepA, and Rep–RepA interactions are examined under in vivo condi- tions. The study concludes with an investigation into the expression strate- gies of the BeYDV capsid protein. Abbreviations BeYDV, bean yellow dwarf virus; CLE, conserved late element; GFP, green fluorescent protein; HA, hemagglutinin; LIR, long intergenic regions; MSV, maize streak virus; NLS, nuclear localization site; RBR, retinoblastoma-binding protein; Rep, replication-associated protein; SIR, short intergenic regions; WDV, wheat dwarf virus. 4482 FEBS Journal 273 (2006) 4482–4494 ª 2006 The Authors Journal compilation ª 2006 FEBS BeYDV, like other geminiviruses, replicates via a rolling circle mechanism. First, the host cell replication machinery synthesizes a complementary sense-strand from a primer located within the SIR to form a dou- ble-stranded intermediate. Next, Rep binds to the hair- pin structure located within the LIR, nicks the virion sense-strand and initiates DNA synthesis from the 5¢-terminus. As DNA synthesis progresses, the virion sense-strand is displaced and eventually is recircular- ized and religated by Rep [1,4–6]. The LIR contains sequences responsible for tran- scription of genes in both genome senses, as well as an inverted repeat sequence that forms the hair–loop structure required for replication [7]. A conserved non- anucleotide sequence, located within the loop of the hairpin structure, contains the origin of replication. Cis-acting elements, responsible for both complement- ary and virion sense gene expression, are also located within the LIR. An iteron, which contains the Rep- binding site, is located between the TATAA sequence and the transcriptional start site of the Rep gene. This enables Rep to mediate repression of its own promoter by interfering with initiation of transcription of the Rep gene. RepA, on the other hand, has been shown to function as a retinoblastoma-binding protein (RBR). RepA is involved in controlling the cell cycle, but is not required for virus replication [8]. RepA of wheat dwarf virus (WDV), a related mastrevirus, has been shown to bind to the LIR in addition to Rep, and may play a role in regulating both complementary and virion sense gene expression [5,9–11]. A number of studies have suggested that in Mastre- viridae, the virion sense promoter is transactivated by Rep gene products. Hofer et al. showed that no activ- ity was detectable from the virion sense promoter of WDV in the absence of Rep expression [12]. Further- more, a replication-deficient mutant, which still pro- duced Rep, was able to transactivate virion sense gene expression. Similarly, Zhan et al. found that Rep could enhance virion sense gene expression of chloris striate mosaic virus [13]. Further studies, in which constructs containing a frameshift mutation in ORF C2 had lost their ability to activate virion sense expression, suggested that Rep, not RepA (C1), is the transactivator. Conversely, Collin et al. showed that a cDNA form of Rep, which lacks the intron and thus could not produce RepA, was unable to promote viri- on sense gene expression from a replicating WDV construct, whereas the full-length Rep gene, with the intron intact, produced high levels, suggesting that RepA (C1) alone is required for virion sense expres- sion [14]. More recently, using maize, WDV RepA was shown to activate virion-sense gene expression in maize streak virus (MSV) and WDV, with the RBR-binding domain of RepA being essential for activation in MSV but nonessential in WDV [15]. Using RepA RBR-binding mutants, the authors of this study suggested that the interference of RepA with an RBR-dependent cellular pathway for gene expression in one virus, but not in the other, indicates that two alternative means of activating virion-sense gene expression may exist. In order to elucidate further the roles of Rep and RepA in BeYDV replication and regulation of gene expression, we separated Rep and RepA activities by individually placing them under constitutive promoter control. We cobombarded these Rep constructs, inde- pendently of one another, along with replication- incompetent BeYDV-based constructs containing the viral elements required for both virion and comple- mentary sense gene expression. We found that Rep A (C1) acts as a potent transactivator of virion sense gene expression and inhibitor of complementary sense gene expression. Rep, on the other hand, while also an inhibitor of complementary sense gene expression, had a much weaker effect on virion sense gene expression. Further studies, using RBR mutant RepA constructs, indicated that in the BeYDV system, RepA transacti- vation is still able to take place (albeit to a lesser degree) in the absence of an intact RBR-binding domain. We also demonstrated that Rep possesses a nuclear localization site that is absent from RepA, and that Rep and RepA are able to interact with each other under in vivo conditions. Finally, regulation of BeYDV CP expression was examined. The discovery of an unconventional mechanism of translational initi- ation is discussed. Results Comparison of Pc and Pv promoter strengths in NT-1 cells Previously, we had examined the effects of Rep and RepA on BeYDV replication by placing various Rep constructs individually under 35S promoter control. These constructs were then cobombarded into NT-1 cells along with a replication- competent reporter con- struct containing both BeYDV LIR and SIR sequences [8]. In the current study, we wished to examine, in greater detail, the roles of the Rep gene products in regulating BeYDV gene expression. We designed a replication-incompetent construct, pBYD–LIR, which lacks the SIR required for replication but retains the LIR from which the promoters Pc and Pv are derived (Figs 1 and 2). The GUS gene was inserted into the K. L. Hefferon et al. Geminivirus transcriptional regulation requirements FEBS Journal 273 (2006) 4482–4494 ª 2006 The Authors Journal compilation ª 2006 FEBS 4483 EcoRI site of pBYD–LIR and pBYD–LIRSIR, as described in Hefferon & Dugdale [8]. NT-1 cells were bombarded with pBYD–LIR or pBYD–LIRSIR and either pBYSK1.4 or p35SRep (Fig. 1). A Southern blot was performed to demonstrate that pBYD–LIR was replication-incompetent (Fig. 2A). Replication prod- ucts were detected from extracts of NT-1 cells cobom- barded with the reporter construct pBYD–LIRSIR and either pBYSK1.4 or p35SRep (expressing the Rep gene products Rep and RepA) (Fig. 2A, lanes 1 and 2) but not when the pBYD–LIR reporter construct was used with these Rep-expressing constructs (Fig. 2A, lanes 3 and 4). Similar results were achieved when lar- ger constructs, containing BeYDV sequence beyond the boundaries of the LIR, were used. In addition, fur- ther truncation of the constructs containing the LIR or SIR elements did not change the efficiency of the Pv1 promoter (data not shown). BeYDV promoter strengths were compared by gener- ating various constructs containing the BeYDV LIR in which the virion sense or complementary sense genes were replaced with the GUS ORF. pPcGUS was con- structed by creating an NcoI site at the initiation codon of the ORF C1 (RepA) and by substituting a GUS ORF, as well as a termination signal, in place of the C1:C2 ORF. Similarly, pPvGUS was constructed by creating an NcoI site at the initiation codon for ORF V1 (MP) and inserting the GUS ORF and termination signal in place of ORFs V1 and V2 (CP) (Fig. 2B). These reporter constructs were bombarded into NT-1 cells and the relative GUS activities were determined (Fig. 2C). In these assays, luciferase expression from pLUC was included as an internal control to normalize DNA delivery for GUS expression [16,17]. GUS under 35S promoter control (p35SGUS) was included as a positive control, and GUS in the absence of a promoter Fig. 1. Schematic diagram of the constructs used in this work. (A) Genomic organization of pSKBYD1.4. P, PstI; Xb, XbaI; S, SacI; B, BamHI; E, EcoRI; C, ClaI; Bg, BglII; C1, C2, V1 and V2 represent complementary and virion sense ORFs, respectively. The bar represents 500 bp. The intron is represented by an open box. Promoters are indicated by arrows. (B) BeYDV-derived plasmids containing various forms of Rep ORFs. Rep constructs under 35S promoter control were constructed by PCR amplification of the Rep ORFs. Portions of the Rep gene removed for 35SDintron and 35SDBRep are indicated by a ‘v’. The boxed arrow refers to the cauliflower mosaic virus (CaMV) 35S promoter. The small rectangle represents TEV leader sequences at the 5¢-end of the constitutively expressed Rep constructs. The VSP termination sequence (Tvsp) is depicted by an open rectangle at the 3¢ end of the constitutively expressed Rep constructs. (C) BeYDV-derived reporter cassettes were constructed. The solid line represents the portion corresponding to the BeYDV genome used to construct reporter cassettes. Geminivirus transcriptional regulation requirements K. L. Hefferon et al. 4484 FEBS Journal 273 (2006) 4482–4494 ª 2006 The Authors Journal compilation ª 2006 FEBS (pGUS) was included as a negative control. GUS activ- ities were determined at 6, 12, 24, 36, 48 and 72 h post- bombardment (Fig. 2C). The highest level of GUS activity was determined for p35SGUS; less than half of this level of activity was achieved from the construct, pPcGUS. While cells bombarded with p35SGUS did not reach their maximum level of GUS activity until 48 h postbombardment, NT-1 cells bombarded with the pPcGUS construct reached maximal levels of GUS activity within 12 h of expression, suggesting that this promoter is active early on in the infection cycle. GUS activities generated by the pPvGUS construct were only slightly higher than activities observed for the control construct pGUS in the absence of a promoter. The results of this study indicate that while BeYDV comple- mentary sense genes appear to be active in NT-1 cells in the absence of any additional virus-derived or virus-activated cellular factors, virion sense gene expression is minimal under these conditions. The relat- ive promoter strengths did not change significantly over a time course of 72 h, suggesting that any temporal changes in relative promoter activity may require the presence of additional factors. GUS activity from NT-1 cells bombarded with pGUS was negligible at all time points. Effect of Rep gene products on BeYDV gene expression To determine the respective roles of various BeYDV gene products in the regulation of complementary sense gene expression, we cobombarded Rep gene products independently, and in a number of combinations, with these replication-incompetent reporter constructs. The construct pPcGUS was cobombarded into NT-1 cells along with various constructs expressing BeYDV gene products, and complementary gene expression was quantified by assay for GUS activity (Fig. 3A). Con- struct p35SDBRep, containing a large deletion within the Rep ORF, was included in this study as a negative control (Fig. 3, lane 8) [8]. Cobombardment of either p35SDintron or p35SRepA with the expression cassette, Fig. 2. Comparison of Pv and Pc promoter strengths in NT-1 cells. (A) Southern blot depicting replication products observed when reporter plasmid pBYDLIR–SIR (lanes 1 and 2) and pBYD–LIR (lanes 3 and 4) are cobombarded along with pSKBYD1.4 (lanes 1 and 3) or p35SRep (lanes 2 and 4). 32 P-labelled cDNA, corresponding to the GUS ORF, was used as a probe. Double- and single-stranded DNA replication products are labelled on the left hand side. Molecular weight markers are labelled on the right. (B) Schematic diagram of pPcGUS and pPVGUS replication- incompetent constructs. Details are provided in Experimental procedures. LIR refers to the long intergenic region within the genome of BeY- DV. V1 and C1 refer to the virion-sense and complementary-sense ORFs adjacent to the LIR, respectively. NcoI refers to the restriction site, inserted, via site-directed mutagenesis, at the ATG initiation codons for C1 and V1, repectively. T35S refers to the 35S terminator. Arrows refer to the direction of transcription for both constructs. (C) Relative GUS activity (lgÆmg )1 Æmin )1 ) over a time course for the following constructs bombarded into NT-1 cells; p35SGUS, pPcGUS, pPvGUS and pGUS. Luciferase was used as an internal control in this assay and all experi- ments were repeated in triplicate. p35SGUS activity was standardized to a value of 1 and relative GUS activities were determined. K. L. Hefferon et al. Geminivirus transcriptional regulation requirements FEBS Journal 273 (2006) 4482–4494 ª 2006 The Authors Journal compilation ª 2006 FEBS 4485 pPcGUS, revealed that reporter gene expression was significantly inhibited by either gene product. Inhibition remained consistant, regardless of whether RepA, Rep or both gene products were simultaneously present (Fig. 3A, compare lanes 1–4 with lane 8). No difference in the inhibition of Pc was observed when p35SRep or p35SRepA were substituted with their respective RBR mutants (Fig. 3A, compare lanes 5 and 6 with lane 8). Cobombardment of the expression cassette, pPcGUS, with the p35SCP construct had no effect on comple- mentary sense gene expression (Fig. 3A, compare lane 7 with lane 8). Cobombardment of cells containing the reporter construct, pPvGUS, with p35SRepA revealed that RepA was capable of strongly transactivating the virion sense promoter, whereas cobombardment of pPvGUS with p35SDintron had little effect on transac- tivation (Fig. 3B, compare lanes 1 and 2 with lane 8). Cobombardment of p35SRep (from which both Rep and RepA gene products are produced) and pPvGUS into NT-1 cells also resulted in a great amount of trans- activation (Fig. 3, lane 3). However, simultaneous co- bombardment of p35SDintron and p35SRepA, along with pPvGUS, did not enhance GUS activity further (Fig. 3B, lane 4). Transactivation of virion sense gene expression was also examined when p35SRep and p35SRepA were replaced with their RBR mutant counterparts. Replace- ment of p35SDintron with p35SDintron RBR– resulted in no significant change in GUS activity (Fig. 3B, com- pare lane 1 with lane 5). However, a significant decrease in Pv activation was observed when RepA RBR– was substituted for RepA (Fig. 3B, compare lane 2 with lane 6). The effect of p35SCP on virion-sense gene expres- sion was also examined in this study. No increase in GUS activity was observed when constructs expressing either the CP from BeYDV (p35SCP) or the CP from a nonrelated plant virus (p35SPVXCP) were included (Fig. 3B, compare lane 7 with lane 8, data not shown). Transcript stability and expression levels support the GUS assay results (data not shown). Subcellular localization of Rep gene products Examination of Rep and RepA nucleotide sequences revealed that Rep, but not RepA, possesses a putative nuclear localization site (NLS) within the C-terminal half of the molecule (Fig. 4A). To determine whether this site is functional, constructs p35SDintron–green fluorescent protein (GFP) and p35SRepA–GFP were designed, creating Rep–GFP fusion products. To ensure that these fusion products were still biologically active, p35SDintron–GFP was shown (by Southern blot analysis) to promote BeYDV replication, and RepA– GFP was shown (by assay for GUS activity) to transac- tivate virion sense gene expression, (data not shown). Tobacco protoplasts were electroporated with these constructs and visualized under UV light (Fig. 4B–E). The results of this study indicated that the Rep–GFP fusion product localized exclusively to the nucleus (Fig. 4B,C), whereas the RepA–GFP fusion product was found to be distributed equally throughout both the nucleus and the cytosol (Fig. 4D,E). Interaction of BeYDV Rep and RepA in vivo Horvath et al. and Missich et al. have published con- flicting data regarding the interactions between Rep and RepA of WDV and MDV, using the two-hybrid yeast system [18,19]. To examine, in greater detail, the hetero-oligomerization properties of BeYDV Rep and Fig. 3. Effect of Rep gene products on BeYDV gene expression. Relative GUS activities are shown for constructs (A) pPvGUS and (B) pPcGUS cobombarded into NT-1 cells along with the following BeYDV-encoded gene products: lane 1, p35SDintron; lane 2, p35SRepA; lane 3, p35Srep; lane 4, p35Dintron + p35SRepA; lane 5, p35SDintron RBR– ; lane 6, p35DRepA RBR– ; lane 7, p35SCP; lane 8, p35SDBRep. Samples were collected 24 h after cobombardment. Luciferase was used as an internal control in this assay. The experi- ments were repeated in triplicate. Geminivirus transcriptional regulation requirements K. L. Hefferon et al. 4486 FEBS Journal 273 (2006) 4482–4494 ª 2006 The Authors Journal compilation ª 2006 FEBS RepA under in vivo conditions, NT-1 cells were cobombarded with both p35SHA6HISRep and p35SHARepA. p35SHA6HISRep was collected on a Ni 2+ column and removed by washing the column, then collecting the eluate into 100 lL fractions. Frac- tions were subjected to electrophoresis on a gradient gel, and western blot analysis was performed using antisera to hemagglutinin (HA) (Fig. 5). Rep and RepA were easily detected from cells bombarded with p35SHA6HISRep or p35SHARepA alone (Fig. 5, lanes 1 and 2). While RepA was detected from the first of several washed fractions derived from samples of NT-1 cells bombarded with both constructs (Fig. 5, lanes 3–5), both Rep and RepA were found in the final eluate, indicating that these gene products can interact with each other in vivo (Fig. 5, lane 6). Detec- tion of Rep and RepA in the final eluate was con- firmed by immunoprecipitation, indicating that the presence of RepA in this fraction was not the result of an artifact (Fig. 5, lanes 7 and 8). Regulation of expression of BeYDV CP While the MP of BeYDV appears to be expressed from the V1 promoter, the manner by which the coat protein (V2) is expressed is less clear. As CP expression is known to bring about an increase in single-stranded DNA replication products, replication experiments, using constructs containing the virion sense half of the BeYDV genome, were performed to examine the effect of CP expression on the replication product profile by Southern blot analysis (Fig. 6A,B) [8]. NT-1 cells were cobombarded with the replication-competent expres- sion cassette, pBYDLIR–SIR [8], p35SRep and one of several constructs that contain functional MP or CP genes. When cells were bombarded with a construct that contains both functionally active MP and CP genes (pBYV1V2), a single-stranded replication prod- uct was observed (Fig. 6B, lane 1). Previous studies have indicated that accumulation of single-stranded DNA is probably a consequence of CP accumulation [8] and, because a similar pattern of replication prod- ucts was observed when the CP was placed under 35S promoter control, the results presented in this study are suggestive of CP expression [8]. When a deletion was placed within the MP gene to prevent a functional protein (pBYXV2) from being expressed, and this con- struct was cobombarded along with the replication cas- sette, a single-stranded gene product was still observed, again suggesting that CP accumulation has taken place (Fig. 6B, lane 2). Destruction of the CP gene in con- struct pBYV1X, or elimination of it entirely from this Fig. 5. Interaction of BeYDV Rep and RepA under in vivo condi- tions. p35SHA6HISRep and p35SHARepA were cobombarded into NT-1 cells. Extracts prepared from these cells were then loaded onto a Ni+ column and p35SHA6HISRep was purified according to the protocol of Hefferon & Fan [46]. Lane 1, extracts from cells bombarded with p35SHA6HISRep alone; lane 2, extracts from cells bombarded with p35SHARepA alone; lane 3, extracts from cells bombarded with both p35SHA6HISRep and p35SHARepA after the first wash; lane 4, after the second wash; lane 5, after the third wash; and lane 6, after the elution buffer. Location of Rep and RepA are indicated by arrows. Lanes 7 and 8, Immunoprecipitation of Rep products. Lane 7, extracts of cells after elution buffer; lane 8, nonbombarded cells. E B C D PPLKKKKLKDD A p35S intron/GFP p35SRepA/GFP 35S 35S T T G FP GFP Rep A Rep Fig. 4. Subcellular localization of Rep gene products. (A) Schematic diagram of constructs p35SRep–GFP and p35SRepA–GFP. Location of the putative nuclear localization site is indicated above the Rep– GFP fusion construct. (B–E) Visualization of protoplasts electropo- rated with BeYDV constructs under either UV (B, D) or visible (C, E) light. (B, C) Protoplasts electroporated with p35SRep–GFP; (D, E) protoplasts electroporated with p35SRepA–GFP. K. L. Hefferon et al. Geminivirus transcriptional regulation requirements FEBS Journal 273 (2006) 4482–4494 ª 2006 The Authors Journal compilation ª 2006 FEBS 4487 replication assay, resulted in double-stranded DNA as the predominant replication product (Fig. 6B, lanes 3 and 4). These experiments suggest indirectly that the BeYDV CP may be expressed independently from the downstream cistron of a dicistronic transcript in the absence of a translatable MP (V1). Examination of the sequence surrounding the termination codon of V1 and the initiation codon of V2 revealed a short gap of 13 nucleotides. No obvious promoter signature is apparent within or surrounding this region. Comparison of sequences of similar regions for rela- ted geminiviruses indicates that a similar gap of 10 nucleotides also exists for MSV. WDV and tobacco yellow dwarf virus, on the other hand, possess overlap- ping V1 and V2 ORFs, suggesting that an alternative mechanism of translational initiation may exist among these geminiviruses. Furthermore, the V1 AUG codon of BeYDV is in a suboptimal context (ttgAUGg), sug- gesting that leaky scanning may be the favoured method of translation of the CP. To determine, in greater detail, whether V2 is expressed from a smaller monocistronic transcript, or as the downstream cistron of a larger polycistronic transcription unit, a northern blot was performed on NT-1 cells bombarded with pBYSK1.4, using a 32 P-labelled cDNA probe corres- ponding to the BeYDV CP gene. A 1.4 kb RNA tran- script, corresponding to the size of a full-length polycistronic transcription unit, was observed, imply- ing that CP translation takes place from the down- stream cistron of a single, dicistronic transcript (Fig. 6D, lane 2). No transcripts were observed in non- bombarded NT-1 cells (Fig. 6D, lane 1). Discussion Previous experiments, using replication-competent BeYDV-based constructs, demonstrated that the maxi- mum rate of reporter gene expression is not achieved when active replicons are used [8,20]. In this study, to examine the regulation of gene expression in detail, transcription was uncoupled from replication by the creation of replication-incompetent reporter constructs, and the relative strengths of BeYDV C1 and V1 pro- moters were examined. A reporter construct (pBYD– LIR) containing the LIR, but lacking the SIR, of BeYDV was shown (by Southern blot analysis) to be unable to support replication (Fig. 2A). The present work serves to elucidate further the roles of Rep gene products in BeYDV infection and in transcriptional regulation in general. In the system described in this article, weak expression from the virion sense promoter was attributed to an absence of virus-derived or virus- activated cellular factors from the system. Previous experiments performed with WDV and MSV demon- strated that virion sense promoter activity is greater in phloem cells, suggesting that phloem-specific Fig. 6. Regulation of expression of BeYDV CP. (A) Design of constructs. ‘X’ marks the site where each ORF was disrupted. (B) Southern blot illustrating the profile of repli- cation products collected from NT-1 cells cobombarded with the following BeYDV-based constructs. Lane 1, pBYDLIR-SIR, p35SRep and pBYV1V2; lane 2, pBYDLIR-SIR, p35SRep and pBYXV2; lane 3, pBYDLIR-SIR, p35SRep and pBYV1X; lane 4, pBYDLIR-SIR and p35SRep. Double- stranded (ds) and single-stranded (ss) DNA species are indicated on the left hand side. (C) Nucleotide sequence surrounding V1 and V2 of several mastreviruses. Start and stop codons for ORFs V1 and V2 are indicated by arrows and bold text. (D) Northern blot of total RNA from NT-1 cells cobombarded with pSKBYD1.4 using a 32 P-cDNA probe corresponding to the CP ORF of BeYDV. The RNA ladder is labelled on the left hand side. The single RNA species is indicated by an arrow. Lane 1, nonbombarded NT-1 cells; lane 2, cells bombarded with pSKBYD1.4. Geminivirus transcriptional regulation requirements K. L. Hefferon et al. 4488 FEBS Journal 273 (2006) 4482–4494 ª 2006 The Authors Journal compilation ª 2006 FEBS transcription factors play a role in activating virion sense expression in the infection cycle [21–23]. In addition to this, a cell cycle specificity has been identi- fied for both virion sense and complementary sense promoters of MSV [15]. The differential activity of pro- moters in developmental or tissue-specific cells suggests that cellular proteins may modify Rep to modulate both replication and repression activites [23,24]. The use of suspension cells in the current study would explain the low activity of the virion sense promoter reported here. Addition of Rep to the BeYDV reporter system inhibited expression from the complementary sense promoter. The ability of Rep to down-regulate expres- sion of its own promoter has been studied previously. The AL1 protein of tomato golden mosaic virus, for example, has been shown to play a dual role in tran- scription and replication and it can inhibit its own expression by 20-fold [4,22]. It is likely that either the modification of Rep, or the interaction of Rep with other viral or cellular proteins, may be involved in regulating the role of Rep as either a participant in viral replication or as a repressor of complementary sense gene expression. RepA of BeYDV is not required for BeYDV replica- tion [8,25]; however, the data presented here indicate that it plays an essential role in transactivating the viri- on sense promoter. Transactivation may take place by two mechanisms. The first involves direct binding of RepA to DNA. Similar modes of transactivation have been demonstrated in other virus systems [19,26]. For example, the E1A protein can bind to and transacti- vate the adenovirus major late promoter, and VP16 can stimulate herpes simplex virus-1 early promoters [25]. RepA binding may also be mediated through interactions between RepA and other transcription fac- tors in a manner analogous to those demonstrated for adenovirus E1A and herpesvirus VP16 [26–28]. The second mechanism by which transactivation takes place may involve the binding of RepA to the RBR [26,29]. Activation of late gene expression by RBR binding has also been demonstrated for other DNA virus systems, such as the E1A protein of adenovirus, the large T-antigen of Simian virus-40 and the E7 pro- tein of papillomavirus [26]. In each instance, the viral transactivator protein possesses an LXCXE motif that can interact within a subdomain of RBR. This site of interaction overlaps with the E2F-binding site present on the RBR protein and forces the release of the tran- scription factor. E2F. E2F can then bind to, and initi- ate, transcription from a wide variety of cellular and viral promoters and control transition from G to S phase of the cell cycle, therefore promoting cell cycle progression to one that is more environmentally per- missive for viral replication [25,30–35]. RepA, which is considered to be a functional ana- logue of animal virus oncoproteins, also contains an LXCXE motif [25]. A search revealed two potential E2F-binding sites within the LIR of BeYDV, each located on either side of the hairpin structure. The first site, GTTCCCGC, is located on the virion sense strand (nucleotides 63–68) and the second, TTG GCCGC, is located on the complementary sense- strand (nucleotides 2440–2447). Both have a one- nucleotide mismatch from the consensus sequence TTTG ⁄ CG ⁄ CCGC. Two similar binding sites have been identified within the LIR of WDV, and one of these sites has been shown to interact with human E2F [15]. The same authors further showed that when this sequence was fused as a trimer to a minimal 35S pro- moter controlling GUS, an enhancement of GUS activity was observed in the presence of RepA, but not in the presence of a RepA RBR-binding deficient mutant, indicating that this viral sequence motif is a binding site for E2F and is activated by RepA. It was therefore postulated that RepA can stimulate virion sense gene expression by interfering with a cellular pathway involving both cellular RBR and E2F. The results of work presented in the present article suggest that a similar pathway of gene regulation may occur for BeYDV. However, the fact that transactivation of gene expression could still be observed, although at a lower level, when wild-type RepA was substituted with an RBR-binding mutant, suggests that the RBR-bind- ing pathway is not the exclusive means by which trans- activation occurs. It is more likely that direct RepA binding also plays a role in BeYDV virion-sense gene expression. Besides the E2F-binding sites, two additional con- served late elements (CLEs), each deviating from the consensus GTGGTCCC in one position, were also found to lie 123 and 88 nucleotides upstream of the V1 initiation codon within the BeYDV LIR, respect- ively. CLEs, which had originally been identified as evolutionally conserved DNA sequences present in several different Geminivirus and Nanovirus species, have been shown to have intrinsic enhancer activity in the absence of viral gene products. In begomoviruses, the CLE has been implicated in AC2-mediated trans- activation of the rightward promoter [36,37]. It is possible that the CLEs identified in the current study may contribute, in some way, to transactivation of virion sense gene expression. Our studies indicate that while RepA activates virion sense gene expression, Rep has little effect [37–40]. As both Rep and RepA contain the same LXCXE motif K. L. Hefferon et al. Geminivirus transcriptional regulation requirements FEBS Journal 273 (2006) 4482–4494 ª 2006 The Authors Journal compilation ª 2006 FEBS 4489 for RBR binding, the question of how each performs such different functions in transcriptional activation arises. Secondary structural predictions of WDV Rep and RepA have been made to analyze in detail the region around the LXCXE motif of both gene prod- ucts. Different hydrophobicity patterns, and a differen- tial distribution of L-helices and M-strands between the two proteins, suggest a difference in predicted sec- ondary structures within the same area of the two pro- teins [1,11,19]. The fact that Rep does not interact with RBR suggests that the C-terminus of Rep hinders its ability to bind its LXCXE motif to the appropriate site in RBR. These steric differences between Rep and RepA may also explain how each have overlapping, but different, binding sites within the LIR of WDV [19,22]. While differential binding may also play a role in the inability of Rep to transactivate virion sense gene expression, the altered binding site of RepA may still have the same effects as Rep binding to inhibit complementary sense gene expression. Therefore, in the BeYDV system, inhibition of the complementary sense promoter by either Rep or RepA may differ ster- ically, but the overall effects are similar. The greatest transactivation levels for virion sense gene expression were found when construct p35SRep, which expresses both Rep and RepA gene products, was used in this study. However, placing Rep and RepA each independently under 35S promoter control resulted in a decrease of gene expression. These results are in agreement with earlier experiments using a repli- cation-competent construct [8]. It is possible that alter- ations in the ratios of Rep ⁄ RepA affects the ability of RepA to bind to the LIR, once again suggesting that binding of RepA to the LIR is, at least partially, responsible for transactivation of virion sense gene expression. To understand, in greater detail, the roles of Rep and RepA in regulating BeYDV gene expression, we explored the subcellular localization properties of these gene products by constructing Rep and RepA– GFP fusion proteins and electroporating them into tobacco protoplasts. The exclusivity of Rep in the nucleus, and diffuse pattern of RepA throughout both the nucleus and cytoplasm, support the hypothe- sis that the NLS identified within the Rep ORF is indeed functional. Using AC1 of the begomovirus African cassava mosaic virus in a PVX expression vector, Hong et al. found that mutant AC1–GFP fusion proteins, with an altered nuclear localization site, were also not particularly restricted to the nuclei of cells, but occurred in equal proportions throughout the cytoplasm in a pattern resembling the results des- cribed in the present article [41,42]. It is possible that RepA is small enough to passively enter the nucleus in the absence of an NLS. It has been suggested previously that Rep may inter- act to form hetero-oligomers with RepA to assist in its entry into the nucleus [11]. Indeed, Rep–RepA interac- tions have been observed, with varying degrees of suc- cess, by using the two-hybrid yeast system [18,19,43]. As an alternative to the two-hybrid yeast system, we further examined the ability of these two proteins to interact by copurification of RepA with 6His-tagged Rep from plant extracts from a Ni 2+ column. The strength of the interactions found in this study add another layer of complexity to the roles of Rep and RepA in transcription and replication. The expression of BeYDV CP from the downstream cistron of a single transcript is not unique [44]. A num- ber of plant viruses use unconventional translational initiation mechanisms to express proteins [45]. These mechanisms include leaky scanning, ribosomal frame- shifting, ribosomal shunting, transactivation and cap- independent ribosome binding at internal ribosome entry sites. Future research should shed some light regarding the underlying molecular mechanisms behind CP expression of BeYDV. From a biotechnology per- spective, such knowledge may serve as a powerful tool to enhance or direct the translation of foreign proteins in plants to more desirable levels within BeYDV-based expression vector systems [46,47]. Such a system is cur- rently being used to produce foreign proteins from a plant virus expression vector [48]. The results described in the present article assist in completing a general picture of the multiple roles of Rep and RepA during the BeYDV life cycle. We have demonstrated that Rep and RepA perform different functions with respect to regulating BeYDV bidirec- tional promoter activity. In the early stages of BeYDV infection, both gene products are expressed from pro- moter Pc, apparently in the absence of other virus gene products. While high levels of Rep and RepA result in a shut-off of promoter Pc, RepA alone is responsible for transactivating late genes V1 and V2 as a single dicistronic transcription unit from promoter Pv. This transactivation takes place at least partially via a dis- tinct RBR-binding pathway. The identification of an NLS that resides within Rep, but not RepA, further defines the different roles of these two gene products. Furthermore, their ability to form hetero-oligomers with one another illustrates the intimate associations which exist between Rep and RepA during BeYDV infection. We have suggested, in the current study, that CP expression may take place by a mechanism alternative to conventional scanning. It is thought that the Geminivirus transcriptional regulation requirements K. L. Hefferon et al. 4490 FEBS Journal 273 (2006) 4482–4494 ª 2006 The Authors Journal compilation ª 2006 FEBS Mastrevirus CP may sequester single-stranded DNA molecules for assembly and encapsidation into nascent virus particles, and therefore dictates the ratio of sin- gle-stranded DNA to double-stranded DNA produced during replication [49]. Therefore, transactivation of the virion-sense genes V1 and V2 by RepA during the later stages of BeYDV infection ultimately results in the formation of a pool of virus particles that are ready to be transported to neighbouring cells. The work presented here, in combination with the results of other studies, will assist in the future design and improvement of Geminivirus vectors for expression of foreign proteins in plants and will provide valuable information regarding the biology of this virus. Experimental procedures Cells and viruses NT-1 tobacco cell suspensions were maintained in NT-1 liquid medium as shaker cultures, as described previously [8,50]. The NT-1 suspensions were prepared for biolistic DNA delivery by pipetting a 10-day-old culture onto NT-1 agar plates and preincubating the cells for 3–4 days prior to bombardment. pBYD1.4mer and pDintron were generously provided by J. Stanley (John Innes Centre, Norwich, UK). For bombardments, one micron gold particles (Bio-Rad, Hercules, CA) were used at 800 psi ( 5.52 Mpa) with the Bio-Rad Model PDS-10000 ⁄ He Bioloistic Particle Delivery System, to deliver 2 lg of plasmid DNA prepared according to the Qiagen maxiprep kit protocol (Qiagen, Valencia, CA). Construction of plasmids A schematic diagram of the constructs made is shown in Fig. 1. pSKBYD1.4 contains 1.4 copies of the BeYDV gen- ome cloned into pSK and was generously provided by J. Stanley (John Innes Center) [2]. Construction of p35SRep, p35SDintron, p35SDBRep, p35SRepA and p35SCP are des- cribed by Hefferon & Dugdale [8]. pBYD–LIRSIR was pre- pared by removing the XbaI–SacI fragment of pSKBYD1.4 (encompassing the Rep gene, LIR and SIR) and subcloning the fragment into pBluescriptSKII+. The construct was ren- dered replication deficient by BamHI digestion to release a 727 bp Bam HI fragment within the Rep gene, followed by religation, as in the construction of p35SDBRep [8]. pBYD– LIR was constructed by digestion of pSKBYD1.4 with XbaI and BamHI and subcloning the released fragment, contain- ing the LIR only, into pBluescriptSKII+. The p35SGUS reporter cassette was inserted into the EcoRI site, as des- cribed by Hefferon & Dugdale [8]. pPcGUS and pPvGUS were constructed by introducing an NcoI site at the ATG initiation codon, corresponding to C1 or V1 ORFs of pBYD–LIR, by site-directed mutagen- esis (BRL, Nimes – Cedex 5, France) using primers NcoC1 (CAACACCATGGCTTCTGC) or NcoV1 (GGTATTC CATGGAGCG). An NcoI–HindIII fragment, isolated from the plasmid pGUS2 and containing the GUS gene and 35S terminator, was then inserted into these constructs to gener- ate pPcGUS and pPvGUS reporter constructs, respectively [48]. To create the Rep and RepA–GFP fusion constructs, fragments containing Rep and RepA ORFs were PCR amplified using primers 5¢-NcoRep (GGGCCCCCATGG CTTCTGC) and 3¢-SacRepA (GCAGGTATATGAGCT CCCCGGG), and subcloned into pXbaGFP [49]. pLUC, the luciferase vector, was kindly provided by T. Delaney (Cornell University, Ithaca, NY). Construction of plasmids p35SHA6HISRep and p35SHARepA are described in Hefferon & Dugdale [8] and Hefferon et al. [51], respectively. pBYV1V2 was generated by a BamHI digest of pBYD1.4mer to release a 2.5 kb frag- ment containing the virion-sense genes of the BeYDV genome and subcloned into the plasmid vector pBluescript- SKII+. pBYXV2 was generated by PstI digestion, blunt- ended with mung bean exonuclease (New England Biolabs, Ipswich, MA) to disrupt the V1 ORF and religated with T4 ligase (New England Biolabs). pBYV1X was generated by SalI digestion, blunt-ended with mung bean exonuclease (New England Biolabs) and religated with T4 ligase (New England Biolabs) [52]. Southern blot analysis Two micrograms of each plasmid DNA was cobombarded into a thick slurry of NT-1 cells that had been slowly pipetted onto Petri dishes containing NT-1 cells media plus 8 g of agar-1 (Sigma, St Louis, MO). Plates were then incubated for up to 8 days at 28 °C, depending on the experiment per- formed, and DNA was extracted from cells using the proce- dure described by Wilke [53]. Ten micrograms of total DNA of each sample was digested with HindIII (for replication competency studies) or BamHI (for CP studies) and loaded onto a 1% agarose gel. DNA was transferred onto nitrocellu- lose by capillary action [19]. A 0.5 kb fragment containing the LIR of BeYDV was labelled with 32 P by random priming, according to the conditions recommended by the manufac- turer (Life Technologies, Invitrogen, Carlsbad, CA) and used as a probe for hybridization in 25 nm Tris ⁄ HCl, pH 7.2, 1mm EDTA and 5% SDS at 65 °C, and the signal was detected and quantified by the STORM Optical Scanner sys- tem (Molecular Dynamics, Sunnyvale, CA). GUS assays NT-1 cells, cobombarded with pBYGUS constructs, were analyzed for GUS activity using the protocol of Jefferson [54]. Briefly, 1 g of NT-1 cells was crushed using a micro- pestle, resuspended in GUS extraction buffer (50 nm K. L. Hefferon et al. Geminivirus transcriptional regulation requirements FEBS Journal 273 (2006) 4482–4494 ª 2006 The Authors Journal compilation ª 2006 FEBS 4491 [...]... Organization of the cis-acting element required for wheat dwarf geminivirus DNA replication and visualization of a Rep protein-DNA complex Virology 243, 119–129 11 Gutierrez C (2000) Geminiviruses and the plant cell cycle Plant Mol Biol 43, 763–772 12 Hofer JM, Dekker EL, Reynolds HV, Woolston CJ, Cox BS & Mullineaux PM (1992) Coordinate regulation of replication and virion sense gene expression in wheat dwarf. .. Determination of the origin, cleavage and joining domain of geminivirus Rep proteins Nucleic Acids Res 23, 910–916 8 Hefferon KL & Dugdale B (2003) Independent expression of Rep and RepA and their roles in regulating bean yellow dwarf virus replication J Gen Virol 84, 3465– 3472 9 Sunter G, Hartitz MD & Bisaro DM (1993) Tomato golden mosaic virus leftward gene expression: autoregulation of geminivirus. .. Identification of a sequence element involved in C2-mediated transactivation of the pepper huasteco virus coat protein gene Virology 253, 162–169 38 Sunter G & Bisaro DM (2003) Identification of a minimal sequence required for activation of the tomato golden mosaic virus coat protein promoter in protoplasts Virology 305, 452–462 39 Hung H-C & Petty ITD (2001) Functional equivalence of late gene promoters in bean. .. transcripts and regulatory proteins of FEBS Journal 273 (2006) 4482–4494 ª 2006 The Authors Journal compilation ª 2006 FEBS 4493 Geminivirus transcriptional regulation requirements 41 42 43 44 45 46 47 48 K L Hefferon et al Mungbean Yellow Mosaic Geminivirus J Virol 79, 8149–8163 Hong Y, Stanley J & van Wezel R (2003) Novel system for the simultaneous analysis of geminivirus DNA replication and plant... Hefferon et al 13 Zhan X, Richardson KA, Haley A & Morris BAM (1993) The activity of the coat protein promoter of Chloris Striate Mosaic Virus is enhanced by its own and C1–C2 gene products Virology 193, 498–502 14 Collin S, Fernandez-Lobato M, Gooding PS, Mullineaux PM & Fenoll C (1996) The two nonstructural proteins from wheat dwarf virus involved in viral gene expression and replication are retinoblastoma-binding... analysis of the bidirectional African cassava mosaic virus promoteractivity using two different lucifierase genes Virus Genes 22, 231–242 25 Liu L, Saunders K, Thomas CL, Davies JW & Stanley J (1999) Bean yellow dwarf virus RepA, but not Rep, binds to maize retinoblastoma protein, and the virus tolerates mutations in the consensus binding motif Virology 256, 270–279 Geminivirus transcriptional regulation. .. been presented here as lmolÆmin)1 (per mg of total protein) Luciferase from plasmid pLUC was included as an internal control in this assay, and statistical analysis was performed according to Breyne et al [55] All GUS experiments were repeated in triplicate Northern blot analysis Plant total RNA was isolated by the method of Chomczynski & Sacchi, and 10 lg of purified total RNA per lane was loaded onto... with Indian mung bean yellow mosaic virus Rep and downregulates Rep activity J Virol 78, 11890–11903 33 Castillo AG, Collinet D, Deret S, Kashoggi A & Bejarano ER (2003) Dual interaction of plant PCNA with geminivirus replication accessory protein (Ren) and viral replication protein (Rep) Virology 312, 381–394 34 Liu L, Davies JW & Stanley J (1998) Mutational analysis of bean yellow dwarf virus, a Mastrevirus... Fernandez-Lobato M & Fenoll C (2003) Regulation of MSV and WDV virion-sense promoters by WDV nonstructural proteins: a role for their retinoblastoma protein-binding motifs Virology 306, 313–323 16 Alam J & Cook JL (1990) Reporter genes; applications to the study of mammalian gene transcription Anal Biochem 188, 245 17 Wood KV (1990) Firefly luciferase: a new tool for molecular biologists Promega Notes 28, 1 18... cell-directed control of virion sense gene expression in wheat dwarf virus Nucleic Acids Res 27, 1709–1718 21 Dinant S, Ripoll C, Pieper M & David C (2004) Phloem specific expression driven by wheat dwarf geminivirus V-sense promoter in transgenic dicotyledenous species Physiol Plantarum 121, 108–116 22 Eagle PA & Hanley-Bowdoin L (1997) Cis elements that contribute to geminivirus transcriptional regulation and . Multi-tasking of nonstructural gene products is required for bean yellow dwarf geminivirus transcriptional regulation Kathleen L points. Effect of Rep gene products on BeYDV gene expression To determine the respective roles of various BeYDV gene products in the regulation of complementary