Viroids are the smallest autonomously replicating pathogenic agents yet described--small, highly structured, circular RNAs which lack the protective protein capsid and mRNA activity characteristic of almost all viral RNAs. The term viroid was introduced by Diener (1971) to emphasize the fundamental differences between potato spindle tuber viroid (PSTVd) and conventional plant viruses. Nearly a dozen different viroids have been characterized since PSTVd was first identified as the causal agent of potato spindle tuber disease, all from higher plants. Many of the symptoms commonly associated with viral infection--stunting, epinasty, localized chlorosis or necrosis, and death of the plant--have also been observed with one or another viroid disease (for reviews, see Diener, 1987; Semancik, 1987). Although viroids are responsible for significant losses of food and fiber, no naturally-occurring sources of genetic resistance to these diseases are available to plant breeders.

As illustrated in Fig. 1, the rod-like "native" structure of PSTVd and related viroids is believed to contain five structural domains--a conserved central region (CCR), flanking pathogenicity and variable domains, and left (T_L) and right (T_R) terminal loops. Mutational analysis has shown that replication, cell-to-cell movement, and disease induction are regulated by discrete structural elements located within these domains (e.g., Sano et al., 1992; Hammond, 1994). Several years ago, we began to explore the possibility of engineering disease resistance via constitutive expression of "disabled" RNAs from full-length viroid cDNA templates inserted into the host genome. Ideally, such disabled RNAs would be able to initiate replication but unable to move from cell-to-cell or induce disease. A major obstacle to the widespread implementation of such a "cross protection" strategy is the unknown potential for the disabled RNA to regain the ability to cause severe disease.

Figure 1. Locations of disabling mutations within the secondary structure of PSTVd. Most naturally-occurring strains of PSTVd contain 359 nucleotides, and the boundaries of the five structural domains proposed by Keese and Symons (1985) are shown. A, a 3-nucleotide substitution in the left terminal loop (T_L); B, a 2-nucleotide insertion in the upper portion of the pathogenicity domain; C, a 4-nucleotide deletion in the lower portion of the central conserved region; and D, a 2-nucleotide substitution in the right terminal loop (T_R). Inserts, locations of several other single and multiple nucleotide substitutions within the pathogenicity and variable domains that were also examined for their tendency to revert to wild-type are indicated by arrows.

Based upon preliminary results suggesting that the presence of disabled PSTVd RNAs can make an otherwise susceptible plant resistant to viroid infection, our original proposal contained two specific experimental objectives: i. to assess the ability of several disabled PSTVd RNAs to spontaneously regain the capacity for systemic replication in N. tabacum cv. SR1, and ii. to test selected mutant or wild-type viroid RNAs for their ability to recombine with or otherwise support the replication/systemic spread of disabled PSTVd RNAs. More conclusive evidence concerning the durability of this resistance was also necessary.

Durability of engineered resistance to viroid disease. Several lines of transgenic tobacco (N. tabacum cv. SR1) in which PSTVd cDNA transcription is under the control of the cauliflower mosaic virus 35S promoter were available at the beginning of the granting period. In addition to a control line containing only polylinker sequence between the 35S promoter and 3' polyadenylation site, there were five lines containing PSTVd cDNAs. Three of these lines contained slightly-longer-than-unit-length PSTVd cDNAs and were designed to express "sense" transcripts, each containing a different mutation known to abolish infectivity (see Fig. 1). One of the two lines expected to express PSTVd "antisense" RNA contained a wild-type PSTVd cDNA dimer. Although the precise molecular mechanism is still under investigation, a majority of such plants eventually become infected with PSTVd. Figures 2 and 3 summarize data which indicate that the presence of either wild-type or disabled PSTVd sense RNA is sufficient to confer a marked degree of resistance to subsequent viroid infection.

Young transgenic tobacco seedlings expressing wild-type PSTVd were resistant to challenge inoculation with tomato apical stunt viroid (TASVd), a closely-related viroid (Fig. 2). Resistance was effective against both mechanical and Agrobacterium-mediated challenge (Gardner et al., 1986) and appeared to be correlated with the level of PSTVd replication at the time of challenge. Note that our hybridization assays were specific for PSTVd or TASVd as demonstrated by an almost complete lack of cross reaction in the samples from plants containing only the empty cassette. These results are entirely consistent with the well-known phenomenon of cross protection among related viroids (e.g., Niblett et al., 1978; Branch et al., 1988). They do not, however, directly address the more important question: Will the expression of a disabled PSTVd RNA also confer resistance upon a susceptible host?

Figure 2. Constitutive expression of wild-type PSTVd protects tobacco against superinfection by TASVd. Transgenic tobacco seedlings at the 3- to 5-leaf stage were tested for the presence of replicating PSTVd using a dot-blot hybridization assay (Podleckis et al., 1993). Groups of plants containing either low (+), intermediate (++), or high (+++) levels of PSTVd were then challenged with TASVd and maintained in a greenhouse under conditions suitable for viroid replication. Six to eight weeks p.i., tissue discs collected from each leaf above the site of inoculation were pooled within treatments and tested for the presence of either PSTVd (upper panel) or TASVd (lower panel) by dot-blot hybridization. Some plants contained as many as 13 leaves above the site of inoculation. Heights of individual bars provide a semi-quantitative estimate of the amount of PSTVd and/or TASVd present in plants from different treatments. EC, empty cassette; WT(--), wild-type PSTVd cDNA dimer expected to produce antisense transcripts.

Data from a subsequent experiment showing that disabled PSTVd RNAs are also able to confer a similar degree of resistance to viroid infection is presented in Fig. 3. Two different disabled PSTVd RNAs were tested in this experiment: "CCR(-4)" contains a 4-nucleotide deletion in the lower central conserved domain; "T_R" contains a double substitution in the right terminal loop. Both mutations abolish the ability of PSTVd to replicate systemically, and their locations (designated as "C" and "D", respectively) are shown in Fig. 1. Because the disabled PSTVd RNAs present in these lines are unable to replicate, it was possible to test their resistance to superinfection by both PSTVd and TASVd. Comparison of the upper and lower panels in Fig. 3 shows that resistance to PSTVd infection was much stronger than resistance to TASVd infection. TASVd induces the appearance of much more severe symptoms in Rutgers tomato than does the PSTVd-intermediate strain and is generally considered to be much more "aggressive" than PSTVd. Comparison of the data from plants expressing sense and antisense T_R RNAs (upper panel) suggests that resistance requires the synthesis of PSTVd (+)strands.

Figure 3. Constitutive expression of disabled PSTVd RNAs protects tobacco against Agrobacterium-mediated superinfection by PSTVd or TASVd. Very similar in design to the experiment shown in Fig. 2, this experiment examined only the response to Agrobacterium-mediated challenge inoculation. SR1, nontransformed plants; CCR(-4), PSTVd RNA containing a 4-nucleotide deletion in the lower portion of the central conserved region; and T_R, PSTVd RNA containing a 2-nucleotide substitution in the right terminal loop (Hammond, 1994).

Generation of wild-type PSTVd via sequence reversion. Having established the ability of disabled PSTVd RNAs to confer resistance to viroid infection, we have used two different approaches to examine the stability of such RNAs in planta. Four different mutations were studied in transgenic SR1 tobacco; i.e., a triple substitution in the left terminal loop and a 2-nucleotide insertion in the pathogenicity domain (designated as "A" and "B" in Fig. 1) as well as the 4-nucleotide deletion in the lower central conserved region and double substitution in right terminal loop just described. Only transcripts from the PSTVd cDNA containing a 2-nucleotide insertion in the pathogenicity domain were observed to give rise to wild-type PSTVd. Leaf sap from 55 of 100 such plants reacted strongly with an RNA probe specific for PSTVd, and R-PAGE analysis of RNA extracted from several plants revealed large amounts of circular PSTVd. PCR-mediated sequence analysis showed that these molecules no longer contained the duplication of nucleotides 43-44 present in the PSTVd transgene. The presence of an additional mutation (i.e., a C/U substitution at position 1) in the PSTVd progeny isolated from one infected plant indicates that removal of the extra nucleotides from the pathogenicity domain may proceed via a multi-step pathway.

The stability of several single and multiple nucleotide substitutions within the pathogenicity and variable domains of PSTVd (see Fig. 1) has also been examined using a ribozyme-based system for the in vitro synthesis of highly infectious, precisely full-length PSTVd RNA. The essential features of this expression system--a full-length PSTVd cDNA flanked by specially-designed "hammerhead" and "paperclip" ribozymes derived from tobacco ringspot virus satellite RNA and preceded by a T7 RNA polymerase promoter--are shown in Fig. 4. Progeny isolated from tomato seedlings inoculated with such altered RNAs often contain one or more compensatory mutations. In addition to nucleotide substitutions, some PSTVd molecules also contained small insertions or deletions. Our data do not yet allow us to accurately predict the in vivo stability of a particular disabled viroid RNA, but RNAs that are capable of at least a low rate of replication appear to be under intense selective pressure to regain a near-wild-type structure. Certain types of mutations--small deletions as well as multiple substitutions in critical regions of PSTVd--appear to be stably maintained, however. One example of such a stable mutation is a 3-nucleotide substitution in the left terminal loop (mutation "A" in Fig. 1).

Generation of wild-type PSTVd via RNA recombination. Mutations "A", "B", and "D" (see Fig. 1) were identified in our original proposal as possible "markers" for viroid recombination. As described above, RNA transcripts containing a 2-nucleotide insertion in the pathogenicity domain readily gave rise to wild-type PSTVd in vivo. Use of Agrobacterium-mediated inoculation techniques rather than transgenic plants has shown that a second possible marker mutation (i.e., mutation "D") is also unstable in vivo (Hammond, 1994). Thus, we have generated a number of new transgenic N. benthamiana lines containing the third potential marker for viroid recombination, a triple substitution in the left terminal loop. Each of the 10 independent transformants characterized to date expresses the PSTVd transgene at a level that is easily detectable by dot blot hybridization. None of these plants show any signs of RNA-dependent PSTVd replication.

Figure 4. Transcription and processing of PSTVd ribozyme constructs. The cassette shown is designed for the in vivo production of precisely full-length PSTVd RNAs and contains a cauliflower mosaic virus (CaMV) 35S promoter. Replacement of the CaMV 35S promoter with a T7 promoter yields a template suitable for the in vitro synthesis of highly infectious PSTVd RNAs. HH and PC, abbreviations denoting specially-modified versions of the "hammerhead" and "paperclip" ribozymes present in tobacco ringspot virus satellite RNA.

In contrast to the cDNAs in our transgenic SR1 plants, the PSTVd cDNAs in these N. benthamiana plants are flanked by ribozyme sequences derived from tobacco ringspot virus satellite RNAs (see Fig. 4). We have shown that the precisely full-length PSTVd RNA released from the initial transcript by ribozyme-mediated cleavage is 10,000-fold more infectious than the longer-than-full-length transcripts present in the transgenic SR1 lines (Feldstein et al., manuscript in preparation). This dramatic increase in specific infectivity is almost certainly related to the structure of the termini generated by ribozyme cleavage; i.e., the 5'-hydroxyl and 3'-cyclic phosphate termini allow circularization of the 359-nucleotide PSTVd RNA by RNA. Although these experiments were carried out with RNAs synthesized in vitro from wild-type PSTVd cDNA, we would expect circularization to significantly enhance the ability of disabled PSTVd RNAs to confer resistance and/or undergo recombination in vivo.

Additional studies now in progress. Experiments are currently underway to compare the concentrations of disabled PSTVd RNA transcripts in various lines of transgenic SR1 tobacco by dot blot and Northern analysis. If necessary, a ribonuclease protection strategy (Ambion-Austin, TX) will be used to increase the sensitivity of these assays. A PCR-based assay for the presence of circular PSTVd (a likely prerequisite for the initiation of replication) has also been designed and tested. Of particular interest are RNA molecules derived from a PSTVd transcript containing a double nucleotide substitution in the right terminal loop. As described above, we know that this mutant is able to replicate under certain circumstances (albeit inefficiently).

Efforts to demonstrate recombination between disabled PSTVd RNAs will concentrate on lines of transgenic N. benthamiana plants which express high levels of circular PSTVd RNA containing a triple nucleotide substitution in the left terminal loop. Seed is now being collected from the initial R^o plants. R¹ plants will be challenged with using either Agrobacterium-mediated or mechanical inoculation as described in our original proposal.

Now that adequate technical assistance has been identified, we are confident that we will be able to successfully fulfill our original objectives by the end of FY95.

Branch, A.D., Benenfeld, B.J., Franck, E.R., Shaw, J.F., Varban, M.L., Willis, K.K., Rosen, D.L., and Robertson, H.D. (1988). Interference between coinoculated viroids. Virology 163, 538-546.

Diener, T.O. (1971). Potato spindle tuber "virus." IV. A replicating, low molecular weight RNA. Virology 45, 411-428.

Diener, T.O., Ed. (1987). The Viroids. Plenum, New York.

Keese, P., and Symons, R.H. (1985). Domains in viroids: Evidence of intramolecular RNA rearrangements and their contribution to viroid evolution. Proc. Natl. Acad. Sci. USA 81, 4582-4586.

Gardner, R.C., Chonoles, K., and Owens, R.A. (1986). Potato spindle tuber viroid infections mediated by the Ti plasmid of Agrobacterium tumefaciens. Plant Mol. Biol. 6, 221-228.

Hammond, R.W. (1994). Agrobacterium-mediated inoculation of PSTVd cDNAs onto tomato reveals the biological effect of apparently lethal mutations. Virology 201, 36-45.

Niblett, C.L., Dickson, E., Fernow, K.H., Horst, R.K., and Zaitlin, M. (1978). Cross protection among four viroids. Virology 91, 198-203.

Podleckis, E.V., Hammond, R.W., Hurtt, S.S., and Hadidi, A. (1993) Chemiluminescent detection of potato and pome fruit viroids by digoxigenin-labeled dot blot and tissue blot hybridization. J. Virol. Methods 43, 147-158.

Sano, T., Candresse, T., Hammond, R.W., Diener, T.O., and Owens, R.A. (1992). Identification of multiple structural domains regulating viroid pathogenicity. Proc. Natl. Acad. Sci. USA 89, 10104-10108.

Semancik, J.S., Ed. (1987). Viroids and Viroid-Like Pathogens. CRC Press, Boca Raton.

Symons, R.H. (1994). Ribozymes. Current Opinion Struc. Biol. 4, 322-330.

Owens Figure 1

Owens Figure 2

Owens Figure 3

Owens Figure 4