Citrus tristeza virus (CTV), a member of genus Closterovirus, is one of the more economically important plant viruses. Almost 100 million trees propagated on sour orange (Citrus aurantium L.) rootstocks died worldwide from different tristeza epidemics, and presently, many millions more propagated on decline-tolerant rootstocks are debilitated by severe CTV isolates inducing stem pitting in commercial citrus varieties regardless of the rootstock used (Bar-Joseph and Dawson, 2008; Moreno and Garnsey, 2010).

CTV virions (2000 × 10–12 nm) are composed of a single-stranded, positive-sense genomic RNA (gRNA) of about 20 kb and two coat proteins of 25 (CP) and 27 (CPm) kDa that encapsidate about 97 and 3% of the gRNA, respectively (Bar-Joseph and Lee, 1989; Satyanarayana et al., 2004; Gowda et al., 2009). The CTV gRNA has 12 open reading frames (ORF) and untranslated regions (UTR) of 107 and 273 nt at the 5′ and 3′ termini, respectively (Karasev et al., 1995). ORFs 1a and 1b, encompassing the 5′ half of the genome, encode replicase-related proteins that are translated from the gRNA and contain papain-like protease, methyltransferase-like, helicase-like and RNA-dependent RNA polymerase domains. The ten 3′- proximal ORFs encode proteins p33, p6, p65, p61, p27, p25, p18, p13, p20, and p23, which are expressed via 3′ coterminal subgenomic RNAs (sgRNAs) (Hilf et al., 1995), promoted by internal controller elements (Gowda et al., 2001). Proteins p6, p65, p61, p27, and p25 are part of a module conserved among closteroviruses that is involved in virion assembly and movement (Satyanarayana et al., 2000, 2004; Dolja et al., 2006; Gowda et al., 2009; Tatineni et al., 2010). The p33, p18, and p13 proteins are dispensable to systemically infect some citrus species (Tatineni et al., 2008), but they are required to invade others like grapefruit (C. paradisi Macf.) and sour orange (Tatineni et al., 2011). Moreover, p33 is required in CTV-infected plants to exclude superinfection by isolates of the same strain (Folimonova et al., 2010; Folimonova, 2012), and the expression ratio between p33 and p13 or p18 seems to determine the stem pitting symptom (Tatineni and Dawson, 2012). Proteins p25, p20, and p23 have been shown to act as silencing suppressors in Nicotiana benthamiana (NB) and N. tabaccum plants (Lu et al., 2004), with p23 being also a pathogenicity determinant (Ghorbel et al., 2001; Fagoaga et al., 2005; Albiach-Martí et al., 2010).

In nature, the CTV host range is restricted to species of a few genera within the subfamily Aurantioideae, and within infected plants, the virus invades only phloem tissues. Although CTV was experimentally transmitted to Passiflora gracilis and P. caerulea (Müller et al., 1974; Roistacher and Bar-Joseph, 1987), two perennial vines, trials to mechanically transmit it to herbaceous and other non-rutaceous woody species, including NB and other Nicotiana species, were unsuccessful (Müller and Garnsey, 1984; our unpublished results). Moreover, although citrus can be mechanically inoculated by slashing citrus stems with CTV virions (Garnsey et al., 1977), all attempts to mechanically inoculate them with virion RNA or RNA transcripts from a cDNA clone were unsuccessful (Satyanarayana et al., 2001). These limitations and the large size of the CTV genome that hindered preparation of full-length cDNA clones and of intact RNA transcripts for inoculation greatly delayed development of a genetic system for this virus. After Navas-Castillo et al. (1997) showed that CTV virions were able to replicate in NB protoplasts, Satyanarayana and associates (1999) developed a full-length cDNA clone of the CTV isolate T36 (CTV-T36), from which they synthesized in vitro RNA transcripts that infected NB protoplasts and produced normal CTV virions. Due to the large size and fragility of the RNA transcripts and the difficulty to inoculate protoplasts with such large RNAs, the protoplast infection rate was only about 10⁻⁴ and the amount of virions produced was insufficient to infect citrus plants by mechanical inoculation. Virion amplification by successive cycles of protoplast inoculations yielded amounts of virions which were able to systemically infect citrus plants and incite the symptoms characteristic of the wild T36 isolate (Satyanarayana et al., 2001). However, the above limitations and frequent failures in virion amplification and transfer between protoplast batches made this genetic system very tenuous.

To overcome these problems we developed an improved genetic system based on agroinfiltration of NB leaves with binary plasmids carrying a cDNA of the CTV-T36 genome and different silencing suppressors and performed a time-course analysis of CTV accumulation in those leaves (Ambrós et al., 2011). Unexpectedly, we found that agroinoculated plants of this species, presumed to be “non-host,” were systemically invaded by CTV-T36 with high viral titers, particularly in the upper leaves in which the virus eventually invaded some non-phloem tissues and incited typical disease symptoms. Citrus plants mechanically inoculated with virions produced in NB became systemically infected, displayed the symptoms characteristic of the wild CTV-T36 isolate, and restricted the virus to the phloem, suggesting that replication and movement in NB tissues does not alter CTV-T36 properties.

This new genetic system based on the use of NB as an intermediate host was simpler and more reliable than the former system based on protoplast transfection but it still had at least three potential limitations:

Direct agroinoculation of citrus plants would be an easier and faster procedure than using NB as an intermediate host for producing virions. Although previous efforts to agroinfect citrus with CTV using binary plasmids and different A. tumefaciens strains were unsuccessful (Gowda et al., 2005, and our unpublished data), virulence and transformation efficiency of A. tumefaciens strains are widely determined by experimental conditions and specific interactions with the plant that make some Agrobacterium strains more virulent than others in a particular host species, including members of the Rutaceae (Cervera et al., 1998). Transient and stable transformation of herbaceous and woody plants has been performed mainly with non-oncogenic (disarmed) strains in which the oncogenes of the wild-type T-DNAs were removed (Gelvin, 2005). In this work we tried agroinoculation of citrus plants using two oncogenic strains, the virulent strain C58 (pTiC58) and the supervirulent strain A281 (pTiBo542), transfected with binary vectors carrying a plant expression marker gene and a cDNA of CTV-T36.

We found that the oncogenic Agrobacterium strains efficiently induced tumors expressing GUS in different plant species, including citrus, and that systemic CTV infection developed in some NB plants. However, CTV virions were not detected in tumors of citrus plants. Mechanical inoculation of CTV virions on NB plants agroinfiltrated previously with a silencing suppressor resulted in systemic infection with CTV-T36, but not with CTV T318A, even though this latter isolate replicated and accumulated in NB protoplasts to the same extent as CTV-T36. CTV-T36 was readily graft transmitted from infected to healthy NB plants after agroinfiltrating the receptor plants with a silencing suppressor. The possibilities and limitations of the new genetic system are further discussed.

A new genetic system based on agroinfiltration of NB leaves with disarmed Agrobacterium cultures transfected with binary vectors, carrying the CTV-T36 cDNA, was recently developed (Ambrós et al., 2011). Although this system was easier, faster and more reliable than the former protoplast system (Satyanarayana et al., 1999, 2001), it still showed limitations that we tried to overcome in this work. The most obvious was, why can we not directly agroinoculate citrus plants? A. tumefaciens has been widely used for efficient delivery of viral genomes into different plants including a citrus-infecting virus and several phloem-limited viruses some of which belong to the Closteroviridae family (Grimsley et al., 1986; Prokhnevsky et al., 2002; Chiba et al., 2006; Vives et al., 2008; Liu et al., 2009; Wang et al., 2009). Although transient or stable transformation of citrus species has also been demonstrated (Cervera et al., 1998; Tzfira et al., 2004; reviewed in Peña et al., 2004a), previous trials to agroinfect citrus plants with CTV using different binary plasmids, silencing suppressors and disarmed Agrobacterium strains were unsuccessful (Gowda et al., 2005, and our unpublished data). Infiltration of citrus leaves with Agrobacterium is usually very inefficient for bacterial penetration and usually leads to low level of T-DNA expression, as observed with other recalcitrant species (Wroblewski et al., 2005). However, the ability of wild oncogenic strains to incite tumours on some citrus species proved very efficient for citrus transformation (Cervera et al., 1998). Here the virulent strain C58 was as efficient as the supervirulent strain A281 inducing tumor formation in Mexican lime plants, albeit tumours induced by the latter grew earlier, were larger in size and showed a necrotic phenotype, in agreement with the supervirulence reported by Cervera et al., (1998). Both oncogenic strains, carrying the empty or the CTV expressing vector, produced good vascularized tumours and high GUS transient expression in NB and lime, with C58 giving better results than A281 in the latter host species. Using a gus-intron marker gene engineered in the T-DNA of the binary vector guaranteed that GUS expression occurred only in planta and, therefore, it was indicative of cell transformation. Most tumours incited by C58 in NB became CTV infected at 1–2 mpi, as detected by ELISA and qRT-PCR assays, and 23% of the plants became systemically infected by CTV at 2–3 mpi, as expected from the high vascularization of these tumours. Remarkably, systemic infection of these plants was not associated with highest CTV titers in tumours. Moreover, some plants developed secondary tumorigenic tissues with high CTV titer, but systemic infection of these plants never occurred, further supporting our previous suggestion that systemic infection of NB results from CTV invasion of the vascular system and not from Agrobacterium migration within agroinfiltrated plants (Cubero et al., 2006; Ambrós et al., 2011). Contrasting with NB, essentially no CTV coat protein or viral RNA was detected in N. occidentalis tumours in spite of the high GUS expression observed, suggesting limited virus replication if any in cells of this host, in agreement with the lack of CTV replication observed previously in protoplasts of this species (Navas-Castillo et al., 1997; Satyanarayana et al., 2000). Neither CTV coat protein nor viral RNA was detected in Mexican lime tumours in spite of the high proportion of cells showing intense GUS activity, indicative of total or partial T-DNA integration. Moreover, PCR detection of different T-DNA regions in individual tumours suggested that integration of the full CTV cDNA in the Mexican lime cells occurs at an effective rate. Therefore, failure to detect CTV infection in lime tumours does not seem to depend on the oncogenic strain, tumor formation or cell transformation efficiency. Indeed co-agroinoculation with a binary plasmid expressing the p19 silencing suppressor (Voinnet et al., 2003) slightly increased GUS activity and the CTV systemic infection rate in NB tumours, but it had no effect in lime tumours. Since Mexican lime is known to be highly susceptible to CTV infection, our results suggest a strong silencing response against CTV at the very early steps of infection, or more likely, failure to get enough functional RNA transcripts reaching the cytoplasm of transformed lime cells. Moreover, alemow and Mexican lime leaves pre-infiltrated with a silencing suppressor and then with the vector carrying CTV-T36 developed tumours, but only trace amounts of CTV gRNA could be detected by qRT-PCR in some of them at 14–28 dpi, suggesting occasional restricted infections that never progressed. These results indicate that a CTV genetic system based on direct agroinoculation of citrus hosts presently is unworkable and that, at least in the near future, the use of NB as an intermediate host to produce CTV virions will be necessary.

Since the present genetic system relies on the ability of CTV-36 to replicate in NB cells and to eventually move cell-to-cell and long distance, developing a similar system with new CTV genotypes requires testing previously the ability of their virions to replicate in NB cells. Previous observations indicated that not all CTV genotypes can replicate in NB protoplasts. Here we confirmed that while isolate T318A (Ruiz-Ruiz et al., 2006) replicated and accumulated in protoplasts to the same extent as CTV-T36, the isolate T305 replicated at low level and the isolate T385 did not replicate at all. Phylogenetic comparison of the full-genome sequence of different CTV isolates has revealed that these belong to at least six different strains, with most genetic differences being located in the 5′ moiety of the gRNA (Harper, 2013). Differences in the replicase components may affect their interactions with host factors and thus determine the ability of each strain to replicate in NB protoplasts. To avoid the need for protoplast preparation we tried to mechanically inoculate NB plants by rubbing leaves with CTV-T36 and T318A virion extracts from citrus. While no infection was observed in plants without pre-treatment with a silencing suppressor, up to 39% infection rate was achieved in pre-treated plants mechanically inoculated with CTV-T36 virions. This is the first time that CTV is mechanically transmitted to a new host by leaf rubbing. Although co-infiltration of NB plants with a silencing suppressor was not essential for CTV agroinfection with disarmed A. tumefaciens strains, in line with results reported for the crinivirus Lettuce infectious yellowing (LIYV) (Wang et al., 2009), silencing suppressors expedited systemic infection and often increased infectivity (Ambrós et al., 2011). This finding and the need for a silencing-suppressed receptor plant to achieve infection in mechanically inoculated plants underlines the importance of the antiviral silencing reaction in the early stages of CTV infection. Unfortunately no infection was obtained in pre-treated plants inoculated with T318A virions, even though this genotype replicated like CTV-T36 in NB protoplasts, suggesting that viral factors other than the replicase interact differently with host factors in T36 and in T318 genotypes. Previous studies have shown that resistance to CTV infection is often strain dependent (Yoshida, 1985, 1993, 1996; Garnsey et al., 1987; Gmitter et al., 1996; Mestre et al., 1997a,b,c). Furthermore, this resistance was often due to inability of the virus to move cell-to-cell or long distance, since CTV replicated and accumulated in protoplasts of resistant citrus varieties to the same extent as in susceptible varieties, and normal infectious virions were produced (Albiach-Martí et al., 2004). It is possible that, although the CTV T318A gRNA efficiently replicates in N. benthamiana protoplasts and likely in cells, virions may not be correctly assembled, or virion proteins may not interact properly with host factors, thus impairing virus movement. These results are a major concern for the actual possibilities to use the present genetic system with CTV genotypes other than CTV-T36. Additional CTV isolates replicating in NB protoplasts should be tested for infectivity on NB plants to ascertain if systemic infection of this species is T36-specific or it can be achieved by other virus genotypes. In any case, availability of CTV genotypes capable to produce systemic infection of this host species, others able to replicate but not to spread systemically, and still others unable to replicate, may be helpful to dissect the CTV-NB interactions at the genetic level.

A final limitation of the new genetic system based on agroinoculation of NB plants was the need for new plant agroinoculation cycles to maintain in this host the CTV-T36 or other interesting hybrid constructs that could be produced. This limitation was overcome by developing an efficient graft inoculation system to transmit CTV from infected to healthy NB plants. Pre-treatment of the receptor plants by agroinfiltrating a silencing suppressor, proper selection of the inoculum for long-term survival and incubation conditions to avoid inoculum desiccation were critical factors that increased the transmission rates. Although side grafts allow direct contact of the donor and the receptor phloem tissues, pre-treatment with a silencing suppressor raised transmission rate from less than 2% to about 50% at 3 mpi, underlining again the importance of antiviral silencing defense of the plant in the early stages of CTV infection. Several reports have documented that virus loading into the vascular system is a complex process involving a strong bottleneck for the virus population (Ding et al., 1995; Gilbertson and Lucas, 1996; Wintermantel et al., 1997; Cruz, 1999; Li and Roossinck, 2004; Ali and Roossinck, 2010). Reduction in the effective population size and virus silencing by the new host might explain in part the long delay necessary for systemic infection. Finding that higher infectivity was obtained pre-treating with p19, a silencing suppressor acting at cellular and systemic levels (Voinnet et al., 2003), than with p23, a CTV-encoded suppressor acting only at cellular level (Lu et al., 2004), support the idea that virus loading in the vascular system is an important step for systemic infection and that suppressing long-distance silencing of the receptor plant helps to overcome this obstacle. Moreover, plants pre-treated with p19 not only showed a higher rate of infection at 2 mpi but, contrasting with those pre-treated with p23, this rate further increased at 4 mpi.

CTV distribution and symptom expression in NB plants systemically infected after agroinoculation with oncogenic strains, mechanical inoculation of virions or graft transmission, mimic those reported previously on this host (Ambrós et al., 2011), supporting the notion that essential host-viral interactions leading to viral movement and symptom development remain unaltered. The main difference between agroinoculation and mechanical or graft inoculation was the significant delay in systemic CTV infection observed with the two latter methods, particularly with mechanical inoculation that sometimes had a lag period of up to 4–6 months, probably due to the low number of CTV virions initiating infection. Biological characteristics of CTV virions from NB systemically infected by either procedure remain unaltered and upon slash inoculation to alemow plants these displayed the symptoms characteristic of the CTV-T36 isolate.

Summarizing, agroinoculation of NB with CTV-T36-based vectors by either agroinfiltration or stem agroinoculation are presently the best procedure to assay new CTV hybrid constructs in citrus plants. These new constructs can now be easily maintained in NB plants for future transmissions to citrus or for eventual studies on CTV stability or evolutionary adaption to this non-natural host after successive passages. Infection of NB by leaf rubbing with virion extracts is to our knowledge the first report on how a non-natural host species may become susceptible to mechanical inoculation with a virus that is naturally phloem-restricted, after pre-treatment with a silencing suppressor. This procedure may potentially help developing future infectious CTV clones if virus genotypes other than CTV-T36 replicating in protoplasts are found capable of invading NB plants.