Response of tomato and its wild relatives in the genus Solanum to cucumber mosaic virus and satellite RNA combinations

Cucumber mosaic virus (CMV; genus Cucumovirus, family Bromoviridae), a ubiquitous pathogen with a very wide host range, has polyhedral virions that encapsidate three linear, plus-sense, single-stranded genomic RNAs (RNAs 1, 2 and 3) and the subgenomic RNAs 4 and 4A (Palukaitis & Garcia-Arenal, 2003). In addition, some isolates of CMV encapsidate a satellite RNA (satRNA), which is a small RNA molecule that is dependent on the virus for its replication, encapsidation and spread, but does not supply the helper virus with any essential function. CMV satRNAs exist as different variants, ranging from 330 to 405 nt in size, which apparently do not contain any functional open reading frames and can sometime attenuate or aggravate disease symptoms induced by the helper virus. This effect is particularly relevant in tomato, where the presence of satRNA variants is sometime associated with lethal necrosis and severe diseases in the field (Gallitelli, 2000; Garcia-Arenal & Palukaitis, 1999). Other satRNA variants have been described whose effects on CMV infection result in symptom amelioration. These variants, also referred to as benign satRNAs, have been proposed as biocontrol agents for the prevention of CMV-induced diseases of tomato, using either the preinoculation of plants with mild strains of CMV supporting a benign variant of satRNA for cross-protection (Gallitelli, 1998) or the expression of a benign satRNA sequence in transgenic plants (Cillo et al., 2004).

The control of plant viral diseases in the field is usually achieved through a combination of prophylactic measures and crop-management practices, whose effects are only partially able to prevent damage and economic losses. The availability of crop varieties carrying virus-resistance genes is one of the strategies of choice in order to obtain effective virus control at relatively low costs (Lecoq et al., 2004). Resistance to CMV has been reported in cultivated varieties of tomato (Solanum lycopersicum) (Stoimenova et al., 1992; Weber et al., 1989; Yasui & Yamakawa, 1978). Because of the limited genetic diversity within the cultivated tomato genotypes, CMV resistance has also been sought in some wild tomato species (Solanum section Lycopersicon), which have often been employed by breeders as sources for introgressing disease resistance for tomato genetic improvement. Variable levels of CMV resistance have been reported in Solanum habrochaites (formerly Lycopersicon hirsutum) (Abad et al., 2000; Gebré-Selassié et al., 1990; Laterrot, 1990), Solanum peruvianum (Ciccarese et al., 1987; Kuriyama et al., 1971; Laterrot, 1980; Sotirova et al., 1992; Stamova et al., 1990), Solanum pimpinellifolium (Ciccarese et al., 1987; Stoimenova et al., 1992), Solanum chmielewskii (Abad et al., 2000), Solanum lycopersicoides (Phills et al., 1977) and Solanum chilense (Stamova & Chetelat, 2000; Stamova et al., 1990; Stoimenova & Sotirova, 1991).

The genetic bases of these resistances remained unexplored or were apparently due to quantitative traits (Weber et al., 1989). Only in the case of tomato breeding lines with introgressions from S. chilense was the resistance inherited as a single dominant gene, mapping to chromosome 12, that showed limited penetrance and sensitivity to environmental factors (Stamova & Chetelat, 2000). All of these studies contributed to the widening of genetic resources applicable to breeding programmes aiming to introduce CMV resistance into tomato (Marchoux, 1990) but, despite all efforts, no resistant cultivars or breeding lines are yet available.

Resistance to CMV infection has been tested with different viral isolates, which could account for observed differences reported by different research groups in the response of host plants to the virus. In two cases, results of resistance screening are reported that made use of CMV supporting necrogenic satRNA variants and, in both cases, accessions of S. peruvianum and S. habrochaites that showed some levels of resistance to satRNA-free CMV were susceptible to lethal necrosis (Abad et al., 2000; Jacquemond & Laterrot, 1981). Wild accessions that supported infections by CMV/satRNA strains, which were necrogenic on tomato cultivars without displaying lethal necrosis, have also been reported (White & Kaper, 1987).

In this paper, we report the results of an investigation of the differential response of S. lycopersicum varieties and selected Solanum spp. accessions to an aggressive CMV strain, alone or in combination with three different satRNA variants that can worsen or attenuate symptoms.

Collection of Solanum genotypes.
All Solanum genotypes used in this study are listed in Table 1. S. lycopersicum varieties were mostly from collections stored at the University of Bari. UC82 and Rutgers are processing tomato varieties that were chosen as references for their susceptibility to the employed CMV/satRNA combinations. Diaz is a F₁ hybrid that carries the Sw-5 gene for resistance to tomato spotted wilt virus (TSWV). Edkawi and Saladette, kindly provided by the C. M. Rick Tomato Genetics Resource Center (TGRC, University of California, Davis, CA, USA), were selected as sources of tolerance to salt/alkali and heat stress, respectively. Super Marmande, L250 and Moneymaker were chosen as examples of fresh tomato cultivars. E6203, Micro-Tom and M82 were chosen for their importance in applications of molecular genetics and genomics.

Table 1. Solanum spp. cultivars and accessions, site and country of origin, and symptoms observed upon infection with four CMV/satRNA combinations

All accessions of Solanum section Lycopersicon were provided by the TGRC. One or two accessions per species were used in the study, with the exception of S. chilense, which had been reported to be a source of resistance to CMV, for which more accessions were considered. S. chilense G1.1556 was kindly provided by Dr Yuling Bai (Wageningen University and Research Centre, The Netherlands).

A set of 78 introgression lines (ILs), containing defined segments or more complex introgressions from the S. habrochaites LA1777 genome in an S. lycopersicum E6203 background (Monforte & Tanksley, 2000), was kindly provided by Professor Steven Tanksley (Cornell University, Ithaca, NY, USA). These ILs provided a coverage of about 80 % of the wild parent genome. S. lycopersicumxS. chilense LA1932 back-cross introgression lines, derived from a back-cross (BC₁F₂) population previously selected for resistance to tomato mottle virus (ToMoV) (Griffiths & Scott, 2001), were obtained from Professor J.W. Scott (University of Florida, FL, USA). The BC₁F₃ progeny were derived from seed collected from tolerant BC₁F₂ plants.

Virus isolates and inoculation.
For resistance screening, the aggressive isolate CMV-Fny from New York, USA, belonging to subgroup IA (Owen & Palukaitis, 1988), was used.

Three satRNA variants found in association with three natural CMV strains in tomato fields of southern Italy were converted into cDNA clones by RT-PCR and infectious transcripts were obtained according to previously described procedures (Cillo et al., 2004). The following four CMV preparations were used in this study: (a) CMV-Fny, obtained from in vitro transcription of the three CMV genomic cDNA clones (Rizzo & Palukaitis, 1990), maintained on tomato plants; (b) CMV-Fny/Tfn-satRNA, derived by the co-inoculation of CMV-Fny with the satRNA variant found in association with CMV-Tfn (Crescenzi et al., 1993); (c) CMV-Fny/TTS-satRNA, derived from CMV-Fny plus the satRNA variant found associated with CMV-TTS (Cillo et al., 1994); (d) CMV-Fny/77-satRNA, derived from CMV-Fny plus the satRNA variant found associated with CMV-77 (Gallitelli et al., 1997). The three satRNA transcripts were inoculated onto young tomato plants by adding 100 ng purified CMV-Fny ml^–1 directly to the transcription mix and rubbing on cotyledon leaves dusted with celite. All CMV/satRNA combinations were maintained and purified from tomato plants at 12 days post-inoculation (p.i.) according to standard protocols (Lot et al., 1972).

For resistance tests, at least 10 plants per genotype were inoculated mechanically at the cotyledon stage with sap from systemically infected tomato leaf tissues at 12–15 days p.i. homogenized in 20 vols 0.1 M phosphate buffer, pH 7.2.

After inoculation, plants were kept in a controlled environment in the glasshouse at 22–24 °C (±2 °C) with a 16 h photoperiod, and were observed for disease symptoms every week for a minimum of 2 months.

Nucleic acid extraction and analysis.
Tissues (approx. 0.1 g) from virus-inoculated and mock-inoculated plants were harvested from systemically infected leaves at 14 days p.i., and total nucleic acids were extracted with phenol–choloroform (White & Kaper, 1989). Equal amounts (1 µg per sample) of total nucleic acids were electrophoresed through a denaturing 1.2 % agarose gel, transferred to positively charged nylon membranes and hybridized to CMV RNA (Gal-On et al., 1994)- and satRNA (Cillo et al., 2004)-specific probes. Probes were labelled by in vitro transcription in the presence of digoxigenin (DIG)–11-UTP and membranes were hybridized according to the instructions of the manufacturer of the DIG non-radioactive system (Roche Applied Science). ChemiDoc system apparatus and Quantity One software (Bio-Rad) were used to detect and quantify the chemiluminescent signal. For viral RNA quantification, the accumulation levels of CMV-Fny RNA were used to estimate the relative amount of each viral RNA segment in infections with inocula containing the three satRNA variants. Quantification figures for each segment of genomic RNA were calculated by densitometric measurements of gel-blot hybridization signals of three individual RNA preparations extracted from three plants per treatment, and expressed as the mean percentage of the corresponding CMV-Fny RNA segment signals.

Symptom expression in Solanum spp. genotypes infected with CMV/satRNA combinations
Twenty-nine different Solanum spp. genotypes, including 11 cultivated tomato varieties and hybrids (Table 1), were tested for resistance to CMV-Fny by mechanical inoculation with the virus, alone or in combination with the variants TTS-satRNA, Tfn-satRNA and 77-satRNA, known for their ability to co-determine different disease phenotypes in tomato.

In tomato plants, CMV-Fny infection produced the so-called shoestring symptoms, consisting of the severe reduction of leaflet blades and whole plant growth. The symptoms induced by CMV-Fny were exacerbated when inocula contained either the TTS- or the 77-satRNA. The TTS-satRNA variant co-determined the severe shortening of apical internodes and leaf distortion, leading to a bushy appearance of the plant and very poor flowering, whereas the 77-satRNA co-determined systemic leaf and stem necrosis that caused plant death, usually within 21 days p.i. In contrast, addition of the Tfn-satRNA variant to CMV-Fny inoculum led to an asymptomatic infection, with plants growing with vigour and size comparable to healthy controls.

Thus, the four inocula delineated a disease phenotype pattern in tomato that was reproduced in all S. lycopersicum genotypes used here, which is referred to as disease phenotype pattern 1 (Table 1). Disease phenotype pattern 1 was also induced in all the plants tested of one accession each of S. pimpinellifolium, S. peruvianum and Solanum sitiens. In wild accessions, the symptoms induced by CMV-Fny were usually less severe, i.e. leaf blade narrowing rather than shoestring (Table 1).

S. habrochaites, Solanum pennellii and S. chilense accessions inoculated with the four combinations of CMV/satRNA reacted with two distinct disease phenotype patterns, denoted 2 and 3. Disease phenotype pattern 2 was characterized by latent infections co-determined by Tfn- and TTS-satRNA (a mild form of chlorosis preceded the symptomless phenotype in S. pennellii inoculated with CMV-Fny/TTS-satRNA), whereas disease phenotype pattern 3 was characterized by latent infections co-determined by all three satRNA variants (Table 1). All CMV/satRNA combinations co-determined latent infections in S. chilense accessions LA1930 and LA1932, but not in LA1958. In plants of accession LA1958, CMV-Fny/77-satRNA co-determined chlorosis of the basal leaves that faded away gradually in newly grown vegetation by 21 days p.i. Finally, although the majority (seven of ten) of plants of accession LA2931 exhibited phenotype pattern 3, three of ten plants challenged with CMV-Fny/77-satRNA developed systemic necrosis, suggesting that genetic heterogeneity is present in the population of this accession.

A fourth phenotype pattern was observed on three S. lycopersicum var. cerasiforme accessions, which differed from the typical tomato phenotype pattern in that all plants inoculated with CMV-Fny/77-satRNA initially showed leaf necrosis, but later underwent a progressive recovery, with new vegetation showing no symptoms by 21 days p.i. (Table 1).

Viral RNA and satRNA accumulation in Solanum spp. genotypes infected with CMV/satRNA combinations
In infections by CMV-Fny alone, viral RNAs accumulated at 14 days p.i. to high levels in all the Solanum genotypes (Fig. 1, (a), lane 1; (b), lanes 1 and 5; (c), lanes 2 and 6).

(30K): Fig. 1. Analysis of viral RNA and satRNA accumulation levels in Solanum genotypes inoculated with four CMV/satRNA combinations. (a) S. lycopersicum UC82 inoculated with CMV-Fny (F), CMV-Fny/TTS-satRNA (TS), CMV-Fny/Tfn-satRNA (TF) or CMV-Fny/77-satRNA (77), and healthy control (H). CMV RNA accumulation levels are shown in the upper panel and the positions of the four genomic RNAs are shown on the left; satRNA accumulation levels are shown in the middle panel, and the positions of the satRNAs of different lengths are shown on the left; methylene blue staining of rRNA after blotting on membrane was used to confirm equal loading of RNA in all lanes (bottom panel). (b) S. habrochaites LA1777 and S. pennellii LA0716 inoculated with CMV/satRNA combinations as in (a). (c) S. chilense LA1930 and LA1932 inoculated with CMV/satRNA combinations as in (a).

In tomato, the hybridization signals of genomic RNAs 1 and 2 were reduced by approximately 95 % by Tfn-satRNA and by approximately 75 and 55 % by TTS- and 77-satRNA, respectively (Table 2). Accumulation of RNA3 was poorly affected by TTS- and 77-satRNA variants, whereas its signal was reduced by approximately 70 % when the inoculum contained Tfn-satRNA and a similar effect was also observed for RNA4, whose signal was nearly undetectable. Thus, Tfn-satRNA showed a more evident ability to downregulate the helper virus genomic RNAs in tomato than TTS- and 77-satRNAs. These effects on viral RNA accumulation appeared to be consistent with disease phenotype pattern 1 observed in tomato, where the addition of Tfn-satRNA to the virus inoculum determined a suppression of symptoms, whilst the other two satRNA variants co-determined phenotypes characterized by specific symptoms. Therefore, the condition in which a correlation was observed between latent infection and a sharp downregulation of viral RNA accumulation is referred to as LI–DR (latent infection and downregulation) for the purpose of this study.

Table 2. Relative viral RNA accumulation levels in selected Solanum cultivars and accessions upon infection with four CMV/satRNA combinations

In S. habrochaites, satRNA variants TTS and Tfn strongly reduced hybridization signals of all CMV RNAs, whilst 77-satRNA downregulated RNAs 1 and 2 to about 24 % and reduced the accumulation of RNAs 3 and 4 by only 20 and 35 %, respectively (Fig. 1b; Table 2). This condition was apparently consistent with disease phenotype pattern 2, in which a symptomatic infection developed only when the inoculum contained the 77-satRNA variant, whilst TTS- and Tfn-satRNA variants co-determined LI–DR expression. In S. pennellii, CMV RNAs 1 and 2 were downregulated by about 35 %, whereas RNAs 3 and 4 remained substantially unaffected upon infection with CMV-Fny/77-satRNA. Tfn-satRNA strongly affected accumulation of RNAs 1 and 2 and reduced that of RNAs 3 and 4 by approximately 60 %. A strong downregulation of all CMV RNAs was induced by the TTS-satRNA variant in this host (Fig. 1b, lane 6; Table 2). Therefore, in S. pennellii, LI–DR expression was obtained with Tfn-satRNA, whereas TTS-satRNA downregulated CMV RNAs, but co-induced a latent infection only after a recovery from the initial mild chlorosis, and the presence of 77-satRNA co-determined the appearance of lethal necrosis and the accumulation of CMV-Fny RNAs at high levels.

In S. chilense LA1930, Tfn-satRNA reduced accumulation of RNAs 3 and 4, and 1 and 2 by about 60 and 35 %, respectively. Notably, in this host, all viral RNAs were downregulated by both TTS- and 77-satRNA (Fig. 1c, lanes 3 and 5; Table 2). TTS-satRNA also completely eradicated the accumulation signals of all viral RNAs in S. chilense LA1932, whilst both Tfn- and 77-satRNA variants that showed similar RNA accumulation patterns downregulated RNAs 1 and 2 to about 8–32 %, RNA 3 to about 0–20 % and RNA 4 to about 36–63 %. In the two accessions of S. chilense, LI–DR was induced by all CMV/satRNA combinations.

In the case of disease phenotype pattern 4, the recovery from lethal necrosis observed in plants infected with CMV-Fny/77-satRNA did not correlate with the downregulation of viral RNA (Table 2). Therefore, all CMV/satRNA combinations accumulated in S. lycopersicum var. cerasiforme at approximately the same levels as were observed in tomato.

On the whole, the results suggest that LI–DR expression can be determined (with the partial exception of S. pennellii) by any of the tested satRNA variants in those Solanum genotypes where they are able to reduce the hybridization signals of CMV-Fny RNAs 3 and 4 by at least 50 %. In the host genotypes tested in this study, satRNA variants TTS and 77 exacerbated CMV-induced symptoms only when the downregulation of RNAs 3 and 4 was not below the apparent threshold of a 50 % reduction of their hybridization signals. However, the extent to which each satRNA variant differentially affected the accumulation of CMV RNAs 3 and 4 appeared to be strictly dependent on the host genotype. CMV RNA 1 and 2 accumulation levels were reduced substantially in both the aggressive (necrogenic and stunting) and the benign (asymptomatic) CMV/satRNA combinations, although three exceptions to this general observation (S. pennellii LA0716 infected with CMV-Fny/77-satRNA, S. chilense LA1930 infected with CMV-Fny/Tfn-satRNA and S. lycopersicum var. cerasiforme LA1230 infected with CMV-Fny/77-satRNA) occurred.

Partial mapping of the genomic loci determining tolerance to the stunting phenotype co-determined by CMV-Fny/TTS-satRNA
S. habrochaites LA1777 showed the described LI–DR condition upon infection with CMV-Fny/TTS-satRNA. Therefore, as a first approach to the identification of genomic loci correlated with the determination of this phenotype, we used a set of ILs containing in the genetic background of the susceptible host, tomato cultivar E6203, defined segments of the genome of the tolerant host S. habrochaites LA1777.

Tomato E6203 and most of the ILs responded to the CMV/satRNA inocula according to disease phenotype pattern 1, whereas one IL showed a disease phenotype that did not match that of the parental cultivar. In the case of this IL, designed 3944, carrying two introgressions on chromosomes 3 and 6 (Fig. 2), infection of CMV-Fny/TTS-satRNA determined a latent infection during the first 60 days p.i., whilst typical stunting symptoms appeared only after that period (Fig. 3a). This transitory tolerance did not correlate with LI–DR, as gel-blot analysis did not show a downregulation of viral RNA accumulation levels in tolerant versus susceptible genotypes (Fig. 3b). As the LA1777 chromosome 3 segment introgressed in line 3944 is also contained in ILs 3928 and, partially, 3929, and as neither of these ILs showed a delay in the appearance of the stunting phenotype when infected with CMV-Fny/TTS-satRNA, this observation provided preliminary and indirect evidence that a genetic trait involved in transitory tolerance to stunting might be located on the chromosome 6 segment of S. habrochaites LA1777.

(17K): Fig. 2. Chromosomal location and size of introgressions in S. lycopersicumxS. habrochaites LA1777 introgression line 3944. Numbers at the top are chromosome numbers; black boxes represent S. habrochaites introgressions. Positions of RFLP (restriction fragment-length polymorphism) markers used to map the introgressions are shown to the right of each chromosome.

(49K): Fig. 3. Disease phenotype and viral RNA and satRNA accumulation levels in S. habrochaites LA1777 introgression lines (ILs) inoculated with CMV/satRNA combinations. (a) Plants of ILs 3944 (left) and 3946 (right) inoculated with CMV-Fny/TTS-satRNA at 28 days p.i. Line 3944 shows tolerance, whereas line 3946 shows the usual stunting symptoms. (b) RNA gel-blot of S. lycopersicum UC82 and S. habrochaites ILs 3944 and 3946 inoculated with CMV-Fny (F) and CMV-Fny/TTS-satRNA (TS) at 28 days p.i., and healthy control (H). The positions of the four genomic RNAs are shown on the left side; satRNA accumulation levels are shown in the middle panel; ethidium bromide staining of rRNA, to confirm equal loading of RNA in all lanes, is shown in the bottom panel (inverted image).

Inheritance of the satRNA-mediated tolerance to lethal necrosis derived from S. chilense
From the limited screening of Solanum germplasm presented previously, S. chilense LA1930, LA1932 and LA1958 were identified as potential sources of tolerance to aggressive CMV/satRNA combinations. Therefore, we tested a set of five S. lycopersicumxS. chilense LA1932 BC₁F₂ back-cross introgression lines (BILs) for tolerance to CMV-Fny and CMV-Fny/77-satRNA. These BILs were derived from a previous selection for resistance to ToMoV.

Like tomato, plants of all five BILs inoculated with CMV-Fny showed typical shoestring and growth reduction. When inoculated with CMV-Fny/77-satRNA, three of the five BILs, namely BIL 12, 17 and 18, showed tolerance to lethal necrosis (Table 3). Plants of the three BILs, in a proportion varying between 36 and 63 %, remained free of symptoms throughout the vegetative cycle until fruit ripening and collection of seed, whilst the remainder displayed typical lethal necrosis (Table 3, Fig. 4a). Molecular analysis by RNA gel-blot indicated that, as observed in the parental S. chilense accession, latent infection always correlated with the downregulation of viral RNA accumulation levels (Fig. 4b), thus indicating that LI–DR also occurred in this instance.

Table 3. Response of BC1F2- and BC1F3-inbred progeny of S. lycopersicumxS. chilense hybrids and susceptible controls to CMV-Fny/77-satRNA LI–DR, Latent infection and downregulation of viral RNA accumulation levels; SD, symptoms delayed; LN, lethal necrosis.

(50K): Fig. 4. Disease phenotype and viral RNA and satRNA accumulation levels in S. lycopersicumxS. chilense LA1932 back-cross introgression lines (BILs) inoculated with CMV/satRNA combinations. (a) Plants of BC1F2 BIL 12 inoculated with CMV-Fny/77-satRNA at 21 days p.i. Left panel, non-inoculated healthy control (H); centre panel, tolerant individual showing latent infection (77-Tol); right panel, susceptible individual showing systemic necrosis symptoms that will develop later into plant death (77-Susc). (b) RNA gel-blot of S. lycopersicum UC82 and S. lycopersicumxS. chilense BILs inoculated with CMV-Fny (F) and CMV-Fny/77-satRNA (77) at 21 days p.i., and healthy control (H). CMV RNA accumulation levels (upper panel) in two tolerant BIL individuals (lanes 4 and 6) are compared with a susceptible BIL individual (lane 5) and a CMV-Fny/77-satRNA-inoculated UC82 control (lane 3). The positions of the four genomic RNAs are shown on the left; satRNA accumulation levels are shown in the middle panel; methylene blue staining of rRNA confirms equal loading of RNA in all lanes (bottom panel).

To follow the segregation of the genetic traits that might be involved in controlling LI–DR expression, the BC₁F₃ progeny derived from selfing four BIL 12 and five BIL 18 tolerant plants were used for new tests employing CMV-Fny/77-satRNA as the inoculum (Table 3). As shown in Table 3, the proportion of plants showing lethal necrosis was 100 % in susceptible tomato UC82 controls, in two lines derived from BIL 12 (named BILs 12/04 and 12/06) and in one line derived from BIL 18 (BIL 18/03). In all other BC₁F₃ BILs, the typical lethal necrosis that killed plants within 20–25 days p.i. was observed on 20–90 % of individuals, but a number of plants ranging from about 6 to 80 % displayed different forms of systemic necrosis, usually consisting in a delay of 7–10 days in the appearance of leaf and stem necrosis, but no death. The only plants of the BC₁F₃ progeny that responded to CMV-Fny/77-satRNA with a latent infection were two of 30 plants from BIL 12/01 (Table 3). Once again, RNA gel-blot analysis showed a direct correlation between symptomless infection and downregulation of viral RNA, whereas in plants showing necrosis, CMV RNA accumulation levels were reduced only slightly in comparison with control plants inoculated with CMV-Fny alone (not shown). Altogether, these results suggest that symptomless infection and suppression of CMV RNA accumulation in the presence of a co-infecting satRNA variant are responses that are controlled by quantitative traits, which are not fixed in the progeny tested here and can be inherited simultaneously at low rates. In this article, we present the screening of Solanum spp. germplasm response to the infection of the well-characterized and virulent strain CMV-Fny, alone or in combination with three variants of satRNA that worsen or attenuate the pathogenesis of CMV helper virus-induced symptoms. The differential response of 29 Solanum genotypes to the four inoculum combinations revealed four disease phenotype patterns, each including plants that developed very severe symptoms and plants that remained asymptomatic. Our analysis showed that, under the same environmental conditions, such a differential response appeared to be dependent on the ability of a given satRNA variant to diversely regulate viral RNA accumulation in distinct Solanum spp. genotypes, so that even a satRNA that is necrogenic on tomato may be ameliorative in another specific host. In turn, this condition led to the preliminary identification of Solanum spp. germplasm that could be useful in conferring resistance to the very harmful inocula consisting of CMV plus necrogenic satRNA variants.

Resistance to CMV in tomato and its wild relatives had been reported previously, whereas our tests did not reveal genotypes tolerant or resistant to CMV-Fny alone, i.e. in the absence of satRNA co-infection. This observation, already reported by other authors (Abad et al., 2000), might be due to the virulence of CMV-Fny itself, which was shown to overcome resistances assessed by infecting plants with other CMV isolates. In our study, for instance, S. chilense LA0458 proved to be susceptible to CMV-Fny, like other accessions of the same species, although it has been described as being resistant to eastern European and American CMV isolates and was used as a resistant parent in genetic studies and breeding programmes (Stamova & Chetelat, 2000; Stamova et al., 1990). Moreover, we could not find CMV resistance in any of the S. lycopersicum var. cerasiforme, S. pimpinellifolium, S. peruvianum or S. habrochaites accessions tested, despite the fact that resistant genotypes have been reported for all these species (Abad et al., 2000; Ciccarese et al., 1987; Gebré-Selassié et al., 1990; Laterrot, 1990; Nitzany, 1992; Stoimenova et al., 1992).

A particular kind of response to CMV in some hosts is the well-known satRNA-mediated suppression/exacerbation of disease symptoms that was described almost three decades ago (Collmer & Howell, 1992; Tien & Wu, 1991; Waterworth et al., 1979). The disease symptom modulation is particularly evident in tomato, where it is considered to be the outcome of a trilateral interaction, determined by the satRNA sequence itself, the helper virus and host factors and, with certain CMV strains and satRNA variants, also dependent on environmental conditions (Kaper & Tousignant, 1977; Garcia-Arenal & Palukaitis, 1999; Palukaitis et al., 1992).

Two reports are available, to our knowledge, about the screening of Solanum genotypes of the tomato clade for tolerance to aggressive CMV/satRNA combinations. In an earlier work (White & Kaper, 1987), most S. habrochaites and S. habrochaites f. glabratum accessions, and also one each of S. chmielewskii, S. peruvianum and Solanum neorickii (formerly Lycopersicon parviflorum) accessions, showed no necrotic response but only mosaic when inoculated with the necrogenic combination CMV-D/D-satRNA, whereas other authors (Abad et al., 2000) did not find forms of resistance/tolerance to necrosis in any of the S. habrochaites, S. peruvianum or S. pimpinellifolium accessions infected with the necrogenic combination CMV-Fny/Ix-satRNA. The use of different host and viral genotypes and different inoculation methods might account for the observed differences within the only two host genotypes (i.e. S. habrochaites and S. peruvianum) that were used in both instances.

In our screening, we dealt not only with a necrogenic satRNA variant, but also with a tomato top stunting co-inducing satRNA and a well-characterized benign satRNA. All tests were carried out under the same environmental conditions to guarantee comparability. The most important result was that, with the CMV-Fny/satRNA combinations used, the suppression of viral symptoms in each host genotype appeared to be associated with a reduced accumulation of helper virus RNAs, leading to what we defined as LI–DR. As a general rule, it is well documented that the presence of the satRNA results in depression of the accumulation of CMV, but this does not necessarily imply that the symptoms will be attenuated (Escriu et al., 2000; Garcia-Arenal & Palukaitis, 1999). Indeed, our tests showed that the satRNA-mediated LI–DR was not peculiar to the benign satRNA variant (i.e. Tfn-satRNA), but that it was also reproducible with the aggressive satRNA variants (i.e. TTS- and 77-satRNA) in those Solanum genotypes where the accumulation levels of CMV RNAs 3 and 4 were reduced to below approximately 50 % of the hybridization signal of CMV-Fny (Table 2). Thus, our data suggest that, in a given host, and with a given helper virus/satRNA combination, the extent to which symptoms are attenuated or suppressed might depend on the extent to which the accumulation of viral RNAs 3 and 4 is depressed. The effects of the reduction of RNAs 1 and 2 accumulation levels might be less important for symptom determination, as CMV-Fny/Tfn-satRNA still produces an asymptomatic infection in S. chilense LA1930, although RNAs 1 and 2 are poorly downregulated. Apparently, our data do not completely match those of Escriu et al. (2000), who reported that, in tomato, (i) the necrogenic satRNA variants caused a higher reduction of CMV-Fny accumulation levels than non-necrogenic variants, and (ii) RNAs 1 and 2 accumulated at higher levels in the presence of necrogenic satRNA variants than with the non-necrogenic variants but, differing from what we showed in this work, the reverse was true for RNA 4, whilst the amount of RNA 3 did not differ. This might be true with certain host genotypes, but not with others, and Escriu et al. (2000) may have dealt with satRNA variants that were non-necrogenic (like the variant TTS-satRNA used in our work), but not truly ameliorative, like Tfn-satRNA.

Our results are in agreement with the proposed role of viral coat protein in symptom determination in specific CMV/host combinations, as this protein is encoded by RNA 3 and expressed via the subgenomic RNA 4, whose downregulation seems to be necessary for symptom suppression (Palukaitis & Garcia-Arenal, 2003). Additionally, the symptom suppression might be due to a reduced ability of the virus to move locally or systemically in the plant as a result of the downregulation of RNA 3, which encodes the 3a protein that is involved in cell-to-cell and long-distance movement (Palukaitis & Garcia-Arenal, 2003).

As symptom appearance was dependent on the host genotype that the CMV strain and the satRNA variant interact with, it was possible to identify four types of disease phenotype pattern, depending on the ability of the individual CMV/satRNA combinations to induce or suppress specific symptoms in specific hosts (Table 1). Disease phenotype pattern 2 was characterized by LI–DR upon inoculation with CMV-Fny, supporting either Tfn-satRNA or TTS-satRNA, with tolerance to stunting in S. habrochaites, S. habrochaites f. glabratum and S. pennellii and in four S. chilense accessions. As one of the tolerant accessions, S. habrochaites LA1777, was also the parental genotype of a collection of S. habrochaites ILs (Monforte & Tanksley, 2000), the use of these lines in our experiments allowed us to investigate tentatively the genomic location of the host factors that could be involved in the development of the top stunting in tomato. However, none of the 78 tested ILs showed LI–DR when inoculated with CMV-Fny/TTS-satRNA, and only one individual IL with an introgression on chromosome 6 showed a significant delay of symptom manifestation. The inefficiency of the ILs at reproducing the wild parental phenotypes, found previously by other authors (Parrella et al., 2004), can be explained by the partial coverage of the LA1777 genome, but another possible explanation is that the tolerance to the stunting phenotype is controlled genetically by more than one locus, one of which is probably located on chromosome 6 and confers only partial tolerance.

Disease phenotype pattern 3 was observed on the remaining four S. chilense accessions, showing susceptibility to CMV-Fny, but LI–DR upon infection with any of the three CMV-satRNA combinations, including the tomato necrosis-inducing CMV-Fny/77-satRNA. BILs deriving from an interspecific hybridization via embryo rescue between S. lycopersicum and S. chilense LA1932, selected previously for ToMoV resistance, were used for a preliminary approach to the identification of the trait that might be involved in the tolerance to the necrosis induced by CMV-Fny/77-satRNA. Although 17 of 51 plants (33 %) belonging to BC₁F₂ BILs responded with LI–DR upon inoculation with CMV-Fny/77-satRNA, the same phenotype was observed in only <1 % (two of 241 plants) of the BC₁F₃ progeny. This seems to suggest a low inheritance rate of such a character, probably due to quantitative trait loci that were not fixed in the hybrid populations that we tested, and more efforts to select inbred lines that contain the desired loci at a higher level of homozygosity are currently in progress.

A variation on the typical tomato-like disease phenotype pattern was observed on all three S. lycopersicum var. cerasiforme accessions (disease phenotype pattern 4), which showed a clear tendency to recover from initial leaf necrosis and re-establish healthy-looking new vegetation. This tolerance could potentially be the basis for valuable field resistance to CMV/satRNA-induced lethal necrosis and, furthermore, this genotype is genetically very close to cultivated tomato, and loci of interest would be less difficult to introgress into tomato breeding lines.

The results of this study need to be complemented by assays to separate genetic from environmental effects that might have led to the observed differential response of Solanum genotypes to the four CMV/satRNA inoculum combinations. Nonetheless, the present study revealed the existence of potentially useful tolerance traits to CMV-Fny, supporting aggressive satRNA variants in the genetic background of S. chilense through the establishment of LI–DR. If characterized by molecular and genetic analyses on more segregating progeny, these traits might be worth being introgressed into commercial tomato varieties. This would not solve the problem of infections sustained by CMV alone, but would be very useful to control the necrogenic CMV/satRNA strains in those Mediterranean areas where they are present as endemic and aggressive pathogens.

This work was supported by CNR grant DG.RSTL.107.002 and by MURST grant PRIN 2006. We thank C. M. Rick TGRC, S. Tanksley, Y. Bai and J. W. Scott for providing seeds for the experiments described in this work.