Differential Responses to Virus Challenge of Laboratory and Wild Accessions of Australian Species of Nicotiana, and Comparative Analysis of RDR1 Gene Sequences

Nicotiana benthamiana is a model plant utilised internationally in plant virology because of its apparent hyper-susceptibility to virus infection. Previously, others showed that all laboratory accessions of N. benthamiana have a very narrow genetic basis, probably originating from a single source. It is unknown if responses to virus infection exhibited by the laboratory accession are typical of the species as a whole. To test this, 23 accessions of N. benthamiana were collected from wild populations and challenged with one to four viruses. Additionally, accessions of 21 other Nicotiana species and subspecies from Australia, one from Peru and one from Namibia were tested for susceptibility to the viruses, and for the presence of a mutated RNA-dependent RNA polymerase I allele (Nb-RDR1m) described previously from a laboratory accession of N. benthamiana. All Australian Nicotiana accessions tested were susceptible to virus infections, although there was symptom variability within and between species. The most striking difference was that plants of a laboratory accession of N. benthamiana (RA-4) exhibited hypersensitivity to Yellow tailflower mild mottle tobamovirus infection and died, whereas plants of wild N. benthamiana accessions responded with non-necrotic symptoms. Plants of certain N. occidentalis accessions also exhibited initial hypersensitivity to Yellow tailflower mild mottle virus resembling that of N. benthamiana RA-4 plants, but later recovered. The mutant Nb-RDR1m allele was identified from N. benthamiana RA-4 but not from any of 51 other Nicotiana accessions, including wild accessions of N. benthamiana, demonstrating that the accession of N. benthamiana used widely in laboratories is unusual.


Introduction
The genus Nicotiana, family Solanaceae, comprises 64 described species, the best known and perhaps most infamous of which are N. tabacum and N. rustica from the Americas, which form the basis of the tobacco industry. The genus is divided into 13 sections, the largest of which is Suaveolentes. Nicotiana section Suaveolentes holds 30 species, 27 of which are endemic to Australia, two to the Pacific Islands and one to Africa [1,2]. The species within Suavalentes best known to the plant science community is N. benthamiana, an allotetraploid thought to originate from diploid parents [3,4]. N. benthamiana occurs sporadically across approximately 3000 kilometres of northern Australia from longitudes 114°E to 140°E and between latitudes 14°S and 26°S. Like Nicotiana species from the Americas, the Australian members of the genus contain nicotine and other alkaloids that stimulate the human central nervous system, and these compounds made members of the genus important to the Aboriginal peoples of Australia [5]. From the leaves of various Nicotiana species and the related genus Duboisia they made Pituri (also known as Tjuntiwari, Muntju, Pinkaraangu, Mingulba and other names) [5,6], a product made of dried or baked leaves and wood ash to form a wad that was held in the mouth between the teeth and gums [5,7,8]. N. benthamiana was not the most favoured species for making Pituri, but it was used when more desirable species were unavailable [8].
N. benthamiana is valued today not only because of its susceptibility to over 500 plant viruses [9], but also because of its susceptibility to infection by bacteria, fungi, oomycetes and nematodes [10,11,12,13]. It is used for transient expression of transgenes through agroinfiltration, where Agrobacterium tumefaciens harbouring a T-DNA plasmid is introduced locally into a leaf. Transient local expression of genes from the T-DNA region by the plant enables studies in protein expression and regulation in the infiltrated leaf without the need to express transgenes stably in a whole plant [9,12,14,15,16]. As a reflection of its importance in the plant sciences, two draft genome sequences and transcriptome sequences of N. benthamiana have been released [16,17,18].
The evolutionary basis of the apparent hyper-susceptibility to viruses by N. benthamiana is unclear. Being highly susceptible to many viruses would seem at first glance to place the species on a fast track to extinction, but recent research has shown that viruses with long associations with wild plants are seldom severe pathogens [19]. Under experimental conditions, only a minority of plant viruses actually kill N. benthamiana plants; most infect with mild to moderate symptoms and often the plant is able to reproduce. The climate in which N. benthamiana grows may offer protection from severe virus-induced symptoms. N. benthamiana grows in Australia's north where daytime temperatures can reach above 40°C (>104°F). The earliest report we could find in the scientific literature describing N. benthamiana and responses to virus infection showed how high growing temperatures ameliorated symptoms of Tobacco mosaic virus (TMV) infection [20]. Another suggestion is that wild populations of N. benthamiana live in zones relatively free of plant virus incidence, making resistance to viruses an unnecessary trait. Although virus surveys of wild plants have not been undertaken in most of the natural range of N. benthamiana, there is no reason to suppose that viruses are not present in the flora there. Further south in Western Australia, new viruses are regularly encountered in the indigenous flora [21,22,23,24,25,26,27,28,29,30]. If natural populations of N. benthamiana are indeed highly susceptible to virus infection, might infection confer an evolutionary advantage under certain environmental conditions and/or at some stages of the life cycle? In the controlled conditions of the laboratory, N. benthamiana plants infected with Cucumber mosaic virus (CMV) lived about 20% longer under drought conditions than did uninfected plants [31], probably because virus-infected plants accumulate glycol, myo-inositol and other water stressrelated protectants [32]. If the same phenomenon occurs amongst wild plants living in regions of variable water availability and seasonally arid conditions, such as occur in the areas that N. benthamiana naturally inhabits, it is conceivable that sub-lethal virus infections in later stages  of the life cycle may enable plants to tolerate drought longer than uninfected plants, perhaps  long enough to complete the life cycle. A possible genetic basis to virus susceptibility in N. benthamiana was provided by Yang et al. [33] who identified that the RNA-dependent RNA polymerase I (Nb-RDR1) involved in small interfering RNA synthesis and virus resistance, contained a 72 nucleotide insertion mutation that introduced tandem stop codons. The mutant allele was referred to as Nb-RDR1m (Nicotiana benthamiana RNA-dependent RNA polymerase I mutant).
Using AFLP analysis Goodin et al. [12] showed that N. benthamiana accessions used in laboratories have a very narrow genetic basis. They named the five accessions gathered from laboratories around the world Research Accession (RA) 1-5 and concluded they were probably all derived from a single source. The source of the original laboratory accessions is not published. In collections of wild N. benthamiana lines held by Australian herbaria there exist specimens collected from different habitats, and these show variation in plant size and structure, leaf and flower shape, and other traits [12].
Other Australian Nicotiana species have been utilized by science to a much lesser extent than has N. benthamiana. The best known is N. occidentalis, where several accessions, for example B37 (also known as 37B), P-1, P12, and N1, are identified and have been used in virusrelated studies [34,35,36,37,38]. These accession codes probably refer to members of N. occidentalis ssp obliqua, the most widespread subspecies in Australia. Although N. hesperis was described in 1960 [39] and accession 67A of this species has been cited in scientific reports up until the present day [38,40,41,42], N. hesperis has not existed as a species since 1981 when it was reclassified as N. occidentalis subspecies hesperis [43]. This subspecies has a limited natural distribution, being restricted to northern coastal regions of Western Australia. We could find no records of scientific use of the third subspecies, N. occidentalis ssp occidentalis, also restricted to northern coastal and island sites in Western Australia. Despite recommendations that N. cavicola, N. rosulata, N. ingulba (syn N. rosulata ssp ingulba), and N. rotundifolia are useful experimental hosts in the diagnosis of plant viruses [10], apparently none of these Australian species have been widely adopted for this purpose. The reason is unclear, but perhaps it is because of limited availability of their seed or because of the broader availability of N. benthamiana, N. occidentalis and non-Australian species such as N. tabacum, N. clevelandii, and N. glutinosa [44].
Here, we tested responses of laboratory and wild accessions of N. benthamiana, accessions of the three subspecies of N. occidentalis, accessions of 19 other Australian Nicotiana species, a South American Nicotiana species, and the sole African Nicotiana species to plant viruses. Partial RDR1 gene sequences were obtained from some accessions, and we speculate further on the role of this gene in virus susceptibility and symptom development. N. glutinosa [46] N. goodspeedii SL13 Port Augusta, SA 1955 [46] (Continued) Accessions referred to as 'Seed Lines' (SL) of 20 other Nicotiana species indigenous to Australia and one from Namibia (N. Africana, section Suaveolentes) were kindly provided as seed by Dr Edward Newbigin, University of Melbourne ( Table 1). Plants of a laboratory accession of N. benthamiana, which we designated RA-4 after Goodin et al. [12], N. glutinosa (section Undulatae, naturally occuring from Bolivia to Peru), and Chenopodium amaranticolor (local lesion host native to South America) were already available. All plants were grown in a rotted bark and sand mix to which 5 g each of lime and dolomite and 40 g of slow release NPK fertiliser was added per 40 litres of potting mix. When the germination rate was low or uneven, seed was soaked overnight at room temperature in a solution of 100 mM gibberellic acid (GA 4 ) to stimulate germination [45].

Viruses
Virus isolates used to challenge plants. Anthocercis littoria was used to inoculate a plant of N. benthamiana RA-4, and the isolate was subcultured every 20-30 days to a new plant before the previous host died.

Inoculation of Nicotiana plants
After germination, seedlings were grown to the 4-leaf stage before they were subjected to either mock inoculation with 0.1M phosphate buffer (pH7) and diatomaceous earth (Sigma Corp.) or inoculation as above with the addition of macerated leaf material from a virus-infected plant. Five to ten plants of each accession were used for each treatment, and an equal number were used for mock inoculations. Treated plants were grown in climate-controlled, insect-free greenhouses where they were provided with optimal growing conditions (22°C day and 17°C night temperatures, daily watering, weekly nutrient feeds).

Symptom category index and statistical analysis
Symptom development was recorded on infected plants every two to five days until 35 days postinoculation (dpi). All plants were tested for the presence of systemic spread of virus in uninoculated young leaves at 35 dpi using virus-specific primers (S1 Table) in RT-PCR assays (below). Plant symptoms and infections were also scored 35 dpi using a simple assessment of systemic infection and symptom severity indices as follows (Fig 2):

Confirmation of infection status
Inoculated plants were screened at 35 dpi for presence of the virus using virus-specific primers (S1 Table) in RT-PCR assays. The MyTaq One-Step RT-PCR kit (Bioline) was used to synthesise cDNA and amplify fragments of virus genomes in the presence of virus-specific forward and reverse primers from total RNA extracted from plants using the RNeasy Plant Mini kit (Qiagen) or a dsRNA enrichment method [47] modified by replacing Whatman CF11 cellulose powder with Machery Nagel MN100 cellulose powder. Virus-specific primers were used for YTMMV [30], BYMV [48], CMV [49], and TSWV [50] (S1 Table). Subsequent Sanger sequencing of amplicons was done using the amplification primers and with BigDye terminator V3.1 chemistry, and analysed with an Applied Biosystems/Hitachi 3730 DNA Analyzer. Sequences were edited manually in FinchTV (Geospiza) and aligned using ClustalW [51].

RDR1 gene sequencing
DNA was extracted from 100 mg young leaf tissue using DNeasy Plant Mini kit columns (Qiagen). Primers were designed to flank the region of the Nb-RDR1m allele 72 nt insertion mutation [33]. Three sets of primers were used. Primers RP1 and RP2 [33], which generated an amplicon of approximately 327 nt or 255 nt, depending on the presence of the insertion mutation. Primers RP120614 and RP220614 generated amplicons of 351 nt or 279 nt, and primers RP1new and RP2new generated amplicons of 389 nt or 317 nt (S2 Table). Amplification was done using GoTaq DNA Polymerase (Promega Corp). The parameters used for subsequent Sanger sequencing of amplicons were as described above.

YTMMV
There were significant differences (p<0.05) among the infected plants by YTMMV. This virus killed members of some of the Nicotiana species tested, and it had the highest overall mean         Table, S4  Table). N. benthamiana laboratory accession RA-4 became wilted and chlorotic within 20 days of infection by YTMMV (Fig 3C), then died between 21 and 35 dpi In contrast, none of the wild accessions of N. benthamiana died upon infection with YTMMV. Instead, they exhibited moderate symptoms of mosaic and leaf distortion ( Table 2, Fig 2B, Fig 3D).  (Fig 2D), the two N. rosulata accessions tested, consisting of subspecies ingulba and rosulata (SL18 and SL51, respectively), two accessions of N. simulans (SL19, SL29), three of N. rotundifolia (KB-1, HH-1, RG-1), N. umbratica (Wea-1, Fig 2C), and a Nicotiana accession not identified to the species level (BrH-1). Of those that did not die, plants of N. occidentalis accessions CB-1, Ft-1, and Cl-1, MtA-9, MtA-11, MtA-12, UK-1, UK-3, SC2-A grew new, often chlorotic and deformed shoots from axillary buds, some of which produced flowers and seed. Most of the other species tested exhibited moderate symptoms, the exceptions being accessions of N. heterantha (Ft-2, Ft-3, SL33) and N. africana (SL6), which exhibited mild symptoms. N. glutinosa plants  responded with small necrotic local lesions on infected leaves but there was no systemic spread of the virus. Seeds of N. benthamiana accessions VL552B2.1, 17.24 and 17.26 were collected from YTMMV-infected parent plants and from three mock-infected plants of the same accessions and germinated. Visual assessment was done at the two-leaf stage of approximately 300 seedlings from each batch. Seedlings derived from infected and uninfected parents appeared indistinguishable and none exhibited symptoms typical of YTMMV infection that were evident on parents. RT-PCR assays using YTMMV-specific primers were carried out RNA extracted from bulked leaf samples from each group of seedlings, and these failed to detect YTMMV.

BYMV
There were significant differences (p<0.05) among the infected plants by BYMV. All inoculated Nicotiana plants became infected, but none of the infected plants died. The range of responses expressed by systemically infected plants was from asymptomatic to severe (symptom indices 1-4) ( Table 2, S2 Table, S3 Table, Fig 2A). Most N. benthamiana accessions tested, including RA-4, responded with similar mild to moderate symptoms. Plants of N. benthamiana accession Kx-1 and of N. occidentalis ssp hesperis (Nt-1, Nt-4, Nt-5) were unusual in that they exhibited small necrotic lesions on inoculated leaves (Fig 3A and 3E). Plants of N. occidentalis ssp hesperis accessions NT-1, Nt-4 and Nt-5 became severely symptomatic, but N. occidentalis ssp obliqua-VL552B1.1 plants remained mildly symptomatic, and no necrotic lesions were present. Another unusual symptom of BYMV infection was chlorotic ring patterns that occurred on leaves of all infected N. simulans SL19 plants (Fig 3B), but not on N. simulans SL29 plants. In plants of severely affected accessions of N. goodspeedii SL13 and N. heterantha SL33, the majority of leaves and apical meristems died. These showed signs of recovery when axillary buds emerged by about 35 dpi, but these remained deformed and chlorotic and few flowers, if any, were produced. None of the following exhibited visible symptoms, although they were all systemically infected: N. amplexicaulis SL7, N. forsteri SL5, N. excelsior SL11, N. gossei SL14, N. rotundifolia SL20, N. umbratica Wea-1, N. africana SL6, and Nicotiana species 'Corunna' SL23 (Table 2).

CMV
There were significant differences (p<0.05) among the infected plants by CMV. All Nicotiana plants tested became infected with CMV. In most cases, infected plants showed similar patterns of symptom development to those infected with BYMV (the mean difference in overall symptom index between BYMV and CMV was 0.56) (S2 Table). The notable exception was N. occidentalis ssp hesperis Nt-1, where all plants died, whereas the other two N. occidentalis ssp hesperis accessions exhibited moderate symptoms. The three accessions of N. occidentalis ssp hesperis tested (Nt-1, Nt-4, Nt-5) developed necrotic lesions. N. occidentalis ssp obliqua SL17 plants were asymptomatic. N. cavicola SL9 plants that were strongly symptomatic under BYMV infection responded with mild symptoms to CMV infection ( Table 2). There were no rings induced by CMV infection on any plant tested.

TSWV
There were significant differences (p<0.05) among the infected plants by TSWV. All Nicotiana plants inoculated became infected, and plants of some accessions were killed by infection with TSWV (symptom index 5), and on all plants tested, overall symptoms of TSWV were more severe (mean symptom index of 3.49) than with BYMV or CMV infection (S2 Table, S3 Table). Generally, plants reacted with symptoms about one category higher on infection with TSWV than they did with BYMV or CMV. Exceptions were plants of N. rotundifolia SL20 and Nicotiana 'Corunna' SL23 that had symptoms three categories higher. Infected plants N. occidentalis ssp hesperis Nt-1, Nt-4, Nt-5 and SL15 initially exhibited local necrotic lesions on inoculated leaves and later symptoms were severe or plants died. N. occidentalis ssp obliqua accessions VL552B1.1 and SL17 did not show local lesions on inoculated leaves but later symptoms resembled those of N. occidentalis ssp hesperis accessions. Plants of N. megalosiphon SL1, N. simulans SL19 and SL29, N. occidentalis ssp hesperis SL15, N. cavicola SL9, N. rotundifolia SL20, and Nicotiana 'Corunna' SL23 all exhibited severe symptoms. TSWV did not spread systemically in N. glutinosa, instead plants responded with small necrotic local lesions on inoculated leaves.

RDR1 gene sequence
Fifty-one partial RDR1 gene fragments were sequenced and GenBank accessions assigned ( Table 2), 19 of which were from N. benthamiana accessions and the rest from accessions of other Nicotiana species. Fifty sequences shared >94% nt identity with one another. The notable exception was laboratory accession N. benthamiana RA-4, which contained the identical 72 nt insertion mutation reported by Yang and colleagues [33]. The wild accessions of N. benthamiana tested did not contain this insertion or other insertions or deletions or translation stop codons in this gene region, nor did accessions of the other species tested ( Table 2).

Discussion
Responses to infection by YTMMV were significantly different between new wild accessions of N. benthamiana and laboratory accession RA-4. Although both groups were equally susceptible to infection by YTMMV and to the other viruses tested, as defined by the ability of the virus to replicate and move systemically within the plant, the differences between them were in symptom responses. Every N. benthamiana RA-4 plant infected with YTMMV was dead by 21-35 dpi, whereas all plants of the new wild accessions responded with moderate symptoms, and none died. Additionally, YTMMV-infected plants of the wild N. benthamiana accessions produced viable seed. YTMMV was not detected from seedlings grown from seed from three infected N. benthamiana parent plants, indicating that seed transmission of YTMMV does not occur or it is rare. The insertion mutant Nb-RDR1m allele [33] was present in plants of our laboratory accession RA-4, but was absent in all wild Nicotiana accessions tested, including all those of N. benthamiana. It is tempting to speculate that the absence of Nb-RDR1m in wild accessions of N. benthamiana was responsible for preventing systemic necrosis induced by YTMMV infection. Yet, YTMMV-infected plants of N. occidentalis, N. rosulata, N. excelsior, N. forsteri, and N. cavicola accessions tested also lacked the mutant RDR1 allele and all developed symptom indices of 4 or 5, in some cases resembling those observed in N. benthamiana RDR1m plants. Notably, the response of some N. occidentalis plants to YTMMV infection initially resembled those observed from Nb-RDR1m plants (rapid chlorosis and plant collapse), but later they partially recovered by generating new shoots, and in some cases flowers and seed. The sequences of complete RDR1 genes were not obtained here, so is conceivable that additional loss-of-function mutations exist in RDR1 genes of Nicotiana species.
Previously, two groups [33,52] [33] showed that expression of Nt-RDR1 did not suppress expression of endogenous Nb-RDR1m, Nb-RDR2 or Nb-RDR6 in the transgenic N. benthamiana lines developed. In the two studies [33,52], responses to virus infection differed widely. After Cucumber mosaic virus (CMV, Cucumovirus), PVX or Potato virus Y (PVY, Potyvirus) infection, Yang et al. [33] demonstrated that Mt-RDR1 plants and Nb-RDR1m control plants responded similarly in terms of virus accumulation, viral RNA expression, and symptom severity. In contrast, Ying and colleagues [52] found that Nt-RDR1 plants infected with CMV, PVY and Plum pox virus (Potyvirus) displayed more severe symptoms, had higher virus accumulation, and greater viral RNA expression than did Nb-RDR1m control plants. Like the earlier study [33], our study showed that infection by the non-tobamoviruses we tested (BYMV, CMV, TSWV) did not induce greater symptom expression in Nb-RDR1 than in Nb-RDR1m plants. When challenged with tobamoviruses, Mt-RDR1 plants accumulated less TMV, Turnip vein-clearing virus and Sunn hemp mosaic virus, whereas Nb-RDR1m control plants were severely symptomatic [33]. Again, the responses reported by Ying and colleagues [52] differed markedly; transgenic Nt-RDR1 plants responded in a similar manner to Nb-RDR1m control plants after infection with TMV and Tomato mosaic virus. Our experiment with Nb-RDR1m and Nb-RDR1 plants and the tobamovirus YTMMV gave results more in line with those of Yang et al. [33] than of Ying et al. [52]. We found that plants with an apparently functional Nb-RDR1 were protected against tobamovirus-induced severe symptomology. We are aware that the Nb-RDR1 and Nb-RDR1m plants we used have slightly different genetic backgrounds, so traits other than RDR1 may be present, and these may also influence symptom response. Other genes known to be involved in antiviral RNA silencing include DCL2, DCL3, DCL4, DRB4, RDR6, SGS3, HEN1, AGO1 and AGO2 [18,53], and these were not examined in the current study.
It is unclear why plant responses reported by us and by Yang et al. [33] differed so much from those of Ying et al. [52]. Possible reasons are the differences observed may relate to the sources of the RDR1 (M. truncatula vs N. tabacum vs N. benthamiana), or differences in virulence in the virus strains used. The latter explanation does not easily account for the differences seen in CMV symptomology; both Yang et al. [33] and Ying et al. [52] used genetically similar CMV subgroup I isolates (CMV-Fny and CMV-SD, respectively), whereas in this study we used a relatively dissimilar subgroup II isolate (CMV-SW-11).
Here we determined that there was no difference between Nb-RDR1 and Nb-RDR1m plants in terms of susceptibility to all the viruses tested, as defined by the ability of the virus to infect the plant systemically. Both Yang et al. [33] and Ying et al. [52] determined that Mt-RDR1 or Nt-RDR1 were still susceptible to the tobamoviruses they analysed. Therefore, from all these results it appears that symptom responses and possibly virus titre and viral RNA accumulation, but not susceptibility as such, is associated with Nb-RDR1 function.
Yang et al. [33] proposed that Nb-RDR1m is a recent mutation because its nucleotide sequence (other than the 72 nt insertion) closely resembles the N. tabacum Nt-RDR1, and because Nb-RDR1m is still inducible by phytohormone application and virus infection. Our finding that Nb-RDR1m was absent in the wild populations we tested tends to support the hypothesis that Nb-RDR1 is a recent mutation that has not had time to expand its range. On the other hand, it is possible that it was once more widely distributed but now remains in localised populations, perhaps because of tobamovirus epidemics. The sporadic distribution of N. benthamiana populations in northern Australia may help protect wild Nb-RDR1m populations from tobamovirus infection. Tobamoviruses have no known arthropod vectors, but because they have extremely stable particles [54], potentially any creature that eats or contacts them might vector them. The existence of YTMMV, a Solanaceae-infecting tobamovirus, and Clitoria yellow mottle virus, a legume-infecting one [55], makes it possible that other tobamoviruses exist in the Australian flora, and that over time contact with N. benthamiana populations is likely to occur. Systemic necrosis responses in Nb-RDR1m plants and in other Nicotiana species to YTMMV infection may be examples of 'field resistance', where susceptible plants die quickly, thereby removing themselves as sources of infection and slowing virus spread within the population, but we find this scenario improbable because it should lead to the extinction of Nb-RDR1m. If the location of the Nb-RDR1m population that yielded the original laboratory accession were determined from historical documents, it would be of interest to elucidate natural distribution of Nb-RDR1m within it and surrounding populations to establish if its range is expanding or contracting. Similarly, analysis of the dynamics of the acute and persistent viral flora infecting natural Nb-RDR1m and Nb-RDR1 populations would be of great scientific interest in revealing co-evolutionary strategies and ecological roles in natural plant/virus systems.
Ying et al. [52] proposed that Nb-RDR1 has a dual role: that of silencing viral transcripts and of suppressing Nb-RDR6-mediated inhibition of systemic virus spread. Their hypothesis that the dysfunctional Nb-RDR1m evolved during a prolonged host-virus arms race to favour up-regulation of an Nb-RDR6-induced antiviral system seems unlikely given the apparent rarity of Nb-RDR1m in the wild, but this hypothesis can be tested by carrying out the distribution and ecology studies of natural Nb-RDR1m populations proposed above.
We propose two hypotheses to account for the prevalence in laboratories of the apparently rare-in-the-wild Nb-RDR1m. The first hypothesis is that the insertion mutation happened recently in a laboratory during the 70+ years that N. benthamiana has been utilised by plant biologists. Its discovery in a natural population would immediately discount this hypothesis. The second hypothesis is that seed from a wild Nb-RDR1m plant was collected from a natural population, perhaps amongst Nb-RDR1 plants, and biologists selected the former because of its greater symptom responses to some viruses and perhaps other traits. Indeed, when we compare Nb-RDR1m plants with Nb-RDR1 plants, the former appear to express several unusual traits, some of which may be considered as 'domestication' traits. For example, seed from Nb-RDR1m plants germinated quickly and evenly 1-3 weeks after harvest, whereas that of wild accessions usually required 6-20 weeks of storage (or GA 4 treatment) after harvest before it germinated evenly. Uneven physical or physiological seed dormancy is a common survival strategy against desiccation in wild annual plants [56]. The leaves of Nb-RDR1m plants are relatively glabrous (hairless), have thin laminas, are paler green, smaller, and with petioles that lacked wings. Thin glabrous laminas enable more efficient virus inoculation. Leaves of wild accessions from inland populations were often covered with hard prickly hairs, had thicker laminas, larger leaf sizes, and sometimes they had winged petioles (e.g accessions MtA 3-7). Leaves from plants collected at coastal sites were glabrous with reflective, rugose, waxy laminas, and without winged petioles (e.g. accession VL552B2.1). Experiencing the same growing conditions under glass, Nb-RDR1m plants grew to about a metre in height, whereas plants of some wild accessions (e.g. accession PPM1) grew to two metres. Nb-RDR1m plants quickly produced many small white flowers, while wild accessions were generally slower to flower, had fewer but relatively larger flowers (e.g. accession PPM-1), and some were cream to yellow in colour, or tinged with purple pigment. Smaller, faster developing plants are more suited to scientific experiments done in confined spaces. We are cautious to attribute all the unusual traits seen in N. benthamiana RA-4 plants to human selection. RDR1 influences small RNA expression, which in turn may influence expression of other genes [57,58,59,60,61].
The genetic homogeneity of the single available accession of N. benthamiana has enabled meaningful extrapolation of experiments done in different laboratories over 70 years. The presence of Nb-RDR1m has undoubtedly stimulated its acceptance as a model plant for virology and transient gene expression studies. The discovery reported here that the widely used laboratory accession is probably not representative of the species as a whole is surprising and exciting. This finding opens up possibilities of comparisons between accessions, the most obvious being comparative expression analysis of Nb-RDR1m and Nb-RDR1 plants. There appears to be no sexual incompatibility between the few accessions we have already crossed using N. benthamiana-RA-4 as the female partner, and this is being tested more broadly within the species. Sexual incompatibilities do occur between some accessions of some species, for example Arabidopsis thaliana [62,63].
Any effective biological model system should have available a collection of accessions with a range of genotypical and phenotypical characteristics. For example, A. thaliana, the most widely used model plant system, has over 300 natural accessions available from across Eurasia [64]. Until recently, N. benthamiana probably had only one. Although N. occidentalis has played second fiddle to N. benthamiana in virus research and expression studies, we envisage that availability of a broader range of subspecies and accessions will enable it too to become a useful model species.  Table. Comparison symptom severity induced by four viruses on 75 Nicotiana plants using severity of symptoms as the dependent variable. This is a measure of relative symptom severity induced by each virus. Thus, mean differences >1 indicates overall greater symptom severity is induced by virus I than by virus J, while mean differences <1 indicates overall milder symptoms induced by virus I compared to virus J.