RNA structural features responsible for potato spindle tuber viroid pathogenicity

VIROLOGY

222, 144–158 (1996) 0405

ARTICLE NO.

RNA Structural Features Responsible for Potato Spindle Tuber Viroid Pathogenicity ROBERT A. OWENS,*,1 GERHARD STEGER,† YI HU,‡ ANDREAS FELS,† ROSEMARIE W. HAMMOND,* and DETLEV RIESNER† *Molecular Plant Pathology Laboratory, Plant Sciences Institute, U.S. Department of Agriculture/Agricultural Research Service, Beltsville, Maryland 20705; †Institut fu¨r Physikalische Biologie, Heinrich-Heine Universita¨t, Du¨sseldorf, Federal Republic of Germany; and ‡Department of Plant Biology, University of Maryland, College Park, Maryland 20742 Received April 5, 1996; accepted June 5, 1996 The native structure of potato spindle tuber viroid (PSTVd) contains a series of short double helices and small internal loops that are organized into five structural domains. Nucleotides within the pathogenicity domain are known to play a critical role in modulating PSTVd symptom expression, and it has been suggested that disruption of a comparatively unstable ‘‘premelting region’’ within the pathogenicity domain may be required for disease induction. We have used a combination of quantitative bioassays, temperature gradient gel electrophoresis of circularized RNA transcripts, and thermodynamic calculations to compare the biological and structural properties of 12 representative PSTVd sequence variants. Certain mutations appeared to act indirectly, downregulating pathogenicity by suppressing the rate of PSTVd replication/accumulation. The effects of other mutations appeared to be more direct, but there was no consistent correlation between symptom severity and melting temperature. Taking into account the three-dimensional shape of RNA helices, comparison of the optimal secondary structures for these variants point to major differences in the geometry of their pathogenicity domains; i.e., variants producing intermediate symptoms possess a linear arrangement of three consecutive helices, whereas for variants producing mild or severe symptoms this domain is bent in opposing directions. Such alterations in RNA structure together with concomitant alterations in RNA–protein interaction(s) may be the primary cause of viroid pathogenicity. q 1996 Academic Press, Inc.

INTRODUCTION

short double-helical regions and small internal loops; (2) three comparatively unstable ‘‘premelting (PM) regions’’ present at conserved positions within the native structure; and (3) formation of a series of alternative structural interactions during thermal denaturation in vitro (see Fig. 1; for reviews, see Riesner, 1990, and Riesner and Steger, 1990). Pairwise sequence comparisons suggest that the rod-like native structure of PSTVd and several related viroids contains five structural domains whose boundaries are defined by sharp changes in sequence similarity (Keese and Symons, 1985), i.e., a conserved central domain believed to contain the site where multimeric viroid RNAs are cleaved and ligated to form circular progeny (Baumstark and Riesner, 1995), flanking pathogenicity and variable domains, and two terminal domains. Structural elements within the pathogenicity domain appear to play an especially important role in modulating PSTVd symptom expression. Only four nucleotide substitutions are required to convert the Intermediate strain of PSTVd into a severe strain, and these changes are confined to a ‘‘virulence modulating (VM) region’’ that overlaps PM 1 within the pathogenicity domain. For certain naturally occurring strains of PSTVd, there appears to be an inverse correlation between the calculated stability of PM 1 and virulence in tomato (Schno¨lzer et al., 1985); Visvader and Symons (1985) failed to find a similar relationship among a number of citrus exocortis viroid (CEVd) sequence variants, however. More recently, a combina-

Viroids are the smallest known agents of infectious disease—small (246–463 nt), single-stranded, and highly structured RNA molecules which lack both a protein capsid and detectable mRNA activity (see reviews by Diener, 1987, and Semancik, 1987). Despite their exceedingly small size and lack of mRNA activity, viroids are able to replicate autonomously and induce disease in susceptible plant hosts. A variety of molecular approaches have been used in efforts to identify structural features which allow viroids to replicate, move from cell to cell, and induce disease, but the molecular mechanisms responsible for these processes remain largely unknown. The absence of viroid-encoded proteins implies that replication and pathogenicity result from direct interactions between certain host cell constituents and either the viroid RNA itself or complementary RNAs generated during its replication. Physical studies of potato spindle tuber viroid (PSTVd) and related viroids have shown that these RNAs share several unusual structural features in addition to their small size and circularity. These include: (1) a rod-like native structure composed of an alternating series of 1 To whom correspondence and reprint requests should be addressed at Room 252, Bldg. 011A, Beltsville Agricultural Research Center, Beltsville, MD 20705. Fax: 504-5449. E-mail: ROWENS@ASRR. ARSUSDA.GOV (Internet).

0042-6822/96 $18.00

144

Copyright q 1996 by Academic Press, Inc. All rights of reproduction in any form reserved.

AID

VY 8043

/

6a1b$$$341

07-17-96 13:21:58

viral

AP: Virology

PSTVd PATHOGENICITY

tion of chemical mapping studies and structural calculations has suggested that PSTVd symptom expression may be modulated by either RNA conformation or sequence-specific interactions with host factors rather than structural stability per se (Hammond, 1992). Previous studies of PSTVd pathogenicity have focused on either selected natural isolates of this viroid (e.g., Schno¨lzer et al., 1985; Gruner et al., 1995) or derivatives of the Intermediate strain created by site-directed mutagenesis (Hammond, 1992; Owens et al., 1995). A number of sequence variants differing markedly in their biological properties have been identified, but comparatively little is known about possible differences in their secondary/ tertiary structures. In order to identify more precisely the structural features within the pathogenicity domain of PSTVd which modulate symptom expression, we have compared the biological and structural properties of a set of 12 representative PSTVd sequence variants. Structural effects of individual mutations were monitored by temperature gradient gel electrophoresis (TGGE) of circularized RNA transcripts as well as thermodynamic calculations, and their biological effects were assessed by quantitative bioassays. MATERIALS AND METHODS Viroid strains and cDNA clones Three naturally occurring isolates of PSTVd were included in the quantitative comparisons of pathogenicity. The nucleotide sequence of each isolate—the Intermediate strain (Gross et al., 1978), a mild strain isolated from cultivated potato (Schno¨lzer et al., 1985), and RG1, a severe strain that spontaneously appeared in PSTVdinfected tomato seedlings growing in a greenhouse (Zimmat et al., 1990)—has been published. Eleven other PSTVd variants were produced by site-directed mutagenesis of the Intermediate strain (Loss et al., 1991; Hammond, 1992; Owens et al., 1995). Construction of a full-length (359 nt) PSTVd cDNA clone whose BamHI termini are derived from the upper portion of the central conserved region was described by Cress et al. (1983). This cDNA, derived from the Intermediate strain of PSTVd and cloned in a ‘‘sense’’ orientation within the BamHI site of pUC9 (Owens et al., 1986), was subcloned into plasmid pRZ6-2, a derivative of pTZ18R (U.S. Biochemicals) in which a full-length PSTVd cDNA is flanked by specially modified versions of hammerhead and paperclip ribozymes derived from satellite tobacco ringspot virus RNA (P. A. Feldstein, unpublished data). Transcription of the resulting plasmid DNA by T7 RNA polymerase produces an RNA that spontaneously self-cleaves to release a precisely full-length PSTVd RNA whose 5*-hydroxyl and 2*,3*-cyclic phosphate termini are derived from positions 88 and 87, respectively. A complete set of recombinant plasmids was generated by replacing the 294-bp EagI–Eco47III fragment of PSTVdIntermediate cDNA (positions 145–359/1–79) with the

AID

VY 8043

/

6a1b$$$341

07-17-96 13:21:58

145

corresponding fragments from the mutant cDNAs. Recombinant plasmids were propagated in Escherichia coli strains JM83 or DH5a using 11 yeast-tryptone media supplemented with 50–100 mg/ml ampicillin (Sambrook et al., 1989). Synthesis and analysis of PSTVd RNAs by TGGE Plasmid DNAs were linearized by digestion with HindIII before transcription with T7 RNA polymerase. Transcription reactions (20 ml) containing T7 RNA polymerase were incubated for 3–4 hr at 377 as described by the manufacturer (Promega) except for the addition of [a32 P]UTP (1.5–5 mCi, 3000 Ci/mmol—Amersham) and an increase in the MgCl2 concentration to 15 mM. Radioactively labeled PSTVd RNAs to be used for TGGE analysis were purified by electrophoresis in 6% polyacrylamide gels containing 8 M urea–11 TBE buffer (89 mM Tris, 89 mM boric acid, 2.5 mM EDTA) followed by elution in 0.5 M NH4 acetate–0.1%SDS and ethanol precipitation. The purified linear RNAs were circularized by incubation with a cell-free wheat germ extract. Two microliters of 51 ligase buffer [100 mM Tris–HCl (pH 8.0), 30 mM Mg acetate, 1 mM spermidine, 2 mM EDTA) and 5 ml of wheat germ extract (Promega) were added to 3 ml of 32Plabeled PSTVd RNA (£400,000 cpm), and the resulting mixture was incubated at 377 for 30–90 min. Because the wheat germ extract contains a phosphocreatine/ phosphocreatine kinase energy generating system, RNA ligase activity is not dependent on the addition of exogenous ATP. Following the addition of 90 ml of water, the mixed population of linear and circular PSTVd was recovered by phenol–chloroform extraction and ethanol precipitation. TGGE analysis of 32P-labeled circular (and linear) PSTVd RNAs was carried out as described by Owens et al. (1995). The horizontal 5% polyacrylamide gel and buffer reservoirs contained 0.21 TBE–5 mM NaCl and mixtures of wild-type and mutant PSTVd RNAs (ca. 40– 60,000 cpm each) were applied to the single 12-cm sample slot. Following electrophoresis, the gel was fixed in 10% ethanol–1% acetic acid, dried, and subjected to autoradiography. Quantitative comparisons of PSTVd pathogenicity Development of disease symptoms in tomato seedlings (Lycopersicon esculentum L. cv ‘‘Rutgers’’) infected by each of the 12 infectious PSTVd variants was quantitatively assessed as described elsewhere (Sano et al., 1992; Owens et al., 1995). Unfractionated linear T7 RNA transcripts were used as inocula, and the concentration of full-length PSTVd RNA in individual transcription reactions was estimated by polyacrylamide gel electrophoresis followed by analysis of the fixed and dried gel with a Betascope image analyzer (Betagen). Aliquots (10 ml) of appropriately diluted inocula containing approximately 0.5 ng PSTVd RNA in 20 mM Na phosphate (pH 7.0) were

viral

AP: Virology

146

OWENS ET AL.

rubbed on carborundum-dusted cotyledons of 1-weekold tomato seedlings with a sterilized glass rod. One group of inoculated plants (six plants/treatment) was maintained in growth chambers under conditions suitable for viroid replication and symptom development (i.e., 267, 18 hr mixed fluorescent/incandescent illumination). At the end of the bioassay (48 days p.i.), the total height of each plant as well as the lengths of its individual internodes was measured. Between 2 and 6 weeks p.i., symptom severity in the glasshouse-grown plants was evaluated at weekly intervals using a specially developed numerical index (Sano et al., 1992). PSTVd titers in individual leaves of infected plants were determined by periodic quantitative dot blot hybridization analysis. Eighteen, 25, 32, and 46 days p.i., a 4mm disc was removed from each true leaf. Discs collected from equivalent leaves were combined within treatments, homogenized in 150 ml of AMES buffer (Palukaitis et al., 1985), and extracted with 150 ml of chloroform. Aliquots (20 ml) of the resulting supernatants were added to 350 ml of 101 SSC–7% HCHO–20 mg/ml yeast tRNA and denatured by incubation at 607 for 15 min. Aliquots (50 ml) of the resulting nucleic acid preparations as well as 6- and 36-fold dilutions prepared with the denaturant solution were filtered through Nytran Plus membranes (Schleicher & Schuell) previously equilibrated with 101 SSC. After UV crosslinking, hybridization with a 32P-labeled, full-length RNA probe specific for PSTVd was carried out as described (Owens and Diener, 1984). Serial dilutions of denatured nonradioactive PSTVd RNA transcripts were included as standards, and radioactivities of individual spots were quantitated using a Betascope.

of the resulting circular RNA:RNA duplexes was carried out at 157 in 5% acrylamide–0.12% bisacrylamide gels containing 8 M urea–11 TBE buffer for 18 hr at 160 V in a thermostatted apparatus (Model SE600—Hoefer Scientific). The consensus nucleotide sequence of selected preparations of PSTVd progeny was determined from enzymatically amplified viroid cDNAs as previously described (Owens et al., 1990, 1995). After phenol–chloroform extraction and elution from Prep-A-Gene matrix (Bio-Rad), sequence analysis of double-stranded PCR products was carried out using the fmol Sequencing System (Promega) and 5*-32P-labeled sequencing primers. Thermodynamic calculations Calculations of secondary structure distributions were performed with LinAll (Steger et al., 1984; Schmitz and Steger, 1992). In addition to the published version of LinAll, the program used here takes into account the extra stability of ‘‘tetraloop’’ hairpins (Groebe and Uhlenbeck, 1988; Antao et al., 1991; Antao and Tinoco, 1992). The program calculates the optimal structure plus a defined number of suboptimal structures which are representative of the thermodynamic structure distribution. Structure distributions are presented in the form of a dot plot (see Fig. 8) in which the area of each dot at position (i,j) is proportional to the base-pair probability of the corresponding nucleotides (i:j) (compare McCaskill, 1990). Furthermore, the program predicts optical denaturation curves (i.e., melting curves) from structure distributions calculated at different temperatures (Schmitz and Steger, 1992). RESULTS

Characterization of PSTVd progeny Progeny were examined for possible sequence heterogeneity using a combination of nondenaturing polyacrylamide gel electrophoresis (ndPAGE) of PSTVd/(0)PSTVd duplexes (Zimmat et al., 1990) and PCR-mediated nucleotide sequence analysis (Owens et al., 1990). Starting material for both types of analysis was total cellular RNA extracted from infected leaf tissue, and concentrations of circular PSTVd were estimated by polyacrylamide gel electrophoresis under denaturing conditions (Riesner et al., 1987). PSTVd-complementary RNAs required for analysis of RNA:RNA duplexes by ndPAGE were synthesized by transcription of monomeric, circularly permuted PSTVd cDNA templates, i.e., plasmids pRH704 and 714 (Hecker, 1989; Hecker et al., 1988) plus pPL7233 and 7318 (Loss, 1989; Loss et al., 1991). Total cellular RNA containing 100 ng of circular PSTVd was allowed to hybridize with 100 ng of (0)PSTVd RNA transcripts in 50 ml hybridization buffer (100 mM NaCl, 0.1 mM EDTA, 1 mM Na-cacodylate, pH 6.8) by heating the samples to 957 and then allowing them to cool to 507 over 2–3 hr. Electrophoretic analysis

The biological and structural studies described below involved a total of 14 PSTVd sequence variants. All variants contain 359 nt, and Fig. 1 illustrates the potential range of structural variation within the pathogenicity domain. For the Intermediate strain, the VM region contains two short helical regions separated by a 4-nt symmetric internal loop. Note the differences in both the size and nature of this internal loop for a naturally occurring mild strain. RG1 is the most severe isolate of PSTVd yet described, and the three sequence changes present in this variant appear to destabilize the 6-bp helix on the left side of the VM region. Table 1 describes the location of each sequence change present in these variants with respect to the sequence of PSTVd-Intermediate (Gross et al., 1978). All but one of these changes (i.e., an AA r U substitution at positions 120–121 that is present in several naturally occurring mild strains of PSTVd) are located in the pathogenicity domain. Two of the variants included in our studies were noninfectious, i.e., inoculation of susceptible host plants with RNA transcripts containing mutations at positions 310–311—either alone or in combination with

AID

viral

VY 8043

/

6a1b$$$341

07-17-96 13:21:58

AP: Virology

PSTVd PATHOGENICITY

147

FIG. 1. Location and structure of the PSTVd pathogenicity domain. (Above) Schematic representation of the complete structure of PSTVdIntermediate including the locations of left terminal (TL , 1-46/315-359), pathogenicity (47-73/286-314), central conserved (74-120/240-285), variable (121-148/212-239), and right terminal (TR , 149-211) domains proposed by Keese and Symons (1985), premelting (PM) regions 1–3 (Steger et al., 1984), and secondary hairpins I–III. (Below) Portions of the pathogenicity domains of three naturally occurring PSTVd isolates. Horizontal arrows show the locations of PM 1 and the ‘‘virulence modulating’’ (VM) region proposed by Schno¨lzer et al., 1985; asterisks indicate sequence differences between PSTVd-Intermediate and other variants listed in Table 1. Sequences involved in formation of secondary hairpin II are underlined. Note the presence of an additional U residue between positions 311 and 312 in PSTVd-Mild.

a mutation at position 43—was not followed by systemic infection. For the infectious variants, the stability of each mutation in their respective progeny was monitored using a combination of ndPAGE and PCR-mediated sequence analysis. All changes were stably maintained during the first passage in vivo. As previously reported (Owens et al., 1995), replication of variants containing a C r U substitution at position 311 was accompanied by the spontaneous appearance of a U r A change at position 309.

Several earlier studies have addressed the relationship between PSTVd structure and pathogenicity (Schno¨lzer et al., 1985; Hammond, 1992; Owens et al., 1995; Gruner et al., 1995). None of these studies, however, has compared the biological and structural properties of a full range of PSTVd sequence variants, and the bioassays were carried out under environmental conditions that were only partially controlled. Data presented in Figs. 2 and 3 were collected from plants maintained in growth chambers under a moderate temperature and light regime favoring the expression of a full range of

PSTVd symptoms. The specific infectivity of the precisely full-length, ribozyme-derived, linear PSTVd RNAs used as inocula is equivalent to that of naturally occurring circular PSTVd and approximately 10,000-fold higher than conventional SP6 or T7 RNA transcripts (P. A. Feldstein, unpublished data). Figure 2 illustrates the range of symptoms observed in Rutgers tomato (a traditional indicator host for PSTVd) approximately 7 weeks p.i. Note that the severe stunting and epinasty (a downward curling of leaf lamina and petioles) visible in plants inoculated with RG1 was virtually absent from plants inoculated with PSTVd-Mild. Less dramatic differences in symptom expression were also apparent; these include (i) a variable reduction in stunting associated with mutations at positions 43 or 318 (Fig. 2c) and (ii) the increasing severity of the symptoms induced by variants M1, M2/M1, and M3/M1 (Fig. 2d). The severity of symptoms induced by each of the variants was also quantitatively assessed using the methods described by Sano et al. (1992), and the results of these analyses are presented in Figs. 2e and 2f. For most variants, there appeared to be a strong correlation between the degree of stunting and the number of leaves exhibiting epinasty/rugosity. A possible exception to this generalization is 310G, a PSTVd variant which induced the

AID

viral

Effects of sequence variation on PSTVd symptom expression

VY 8043

/

6a1b$$$341

07-17-96 13:21:58

AP: Virology

148

OWENS ET AL. TABLE 1 Mutations and Compensatory Changes in PSTVd Sequence Variants Variant

Replication rate

Initial mutation(s)

‘‘Compensatory’’ mutations

Intermediate Mild RG1 M1 M2/M1 M3/M1 310G 318A 318G 43G 311U 43G / 311U 310 0 311UU 43G / 310 0 311UU

/// /// //// /// /// //// // /// /// / / / 0 0

None A310 r U, / U311a , AA120 – 121 r U G46 r C, C47 r A, U317 r C G46 r C, C47 r A G46 r C, C47 r A, C315 r U G46 r C, C47 r A, C315 r U, U317 r C A310 r G C318 r A C318 r G U43 r G C311 r U U43 r G, C311 r U A310 r U, C311 r U U43 r G, A310 r U, C311 r U

— — — — — — — — — — U309 r A U309 r A (Noninfectious) (Noninfectious)

appearance of epinasty/rugosity in the foliage with almost the same kinetics as the Intermediate strain but caused a much milder degree of stunting (see Fig. 2f). This stunting became visible only during the later stages of infection.

The secondary structure and structural transitions of PSTVd have been studied in considerable detail (reviewed by Riesner, 1987, 1990; Riesner and Steger, 1990). Thermal denaturation begins within PM regions 1 – 3 (see Fig. 1), and the rod-like native structure un-

dergoes several transitions during its conversion to a single-stranded circle lacking all intramolecular basepairing. In the highly cooperative main transition, all base-pairs in the native structure are disrupted, and new stable interactions (e.g., secondary hairpins I, II, and III) form. This switch from an extended to a branched structure results in a drastic reduction in the electrophoretic mobility. At higher temperatures, PSTVd secondary hairpins I – III dissociate independently in the order of their individual thermal stabilities. A combination of TGGE (Rosenbaum and Riesner, 1987) and thermodynamic calculations of base-pairing probabilities (Schmitz and Steger, 1992) was used to compare the structural effects of mutations within the pathogenicity domain. Prior to TGGE, precisely fulllength PSTVd RNA transcripts were purified by polyacrylamide gel electrophoresis and circularized by incubation with wheat germ extract. The presence of an uncharacterized nuclease activity in the wheat germ extract resulted in cleavage of a portion of the transcripts within the lower central conserved region, and each sample analyzed by TGGE thus contained a mixture of several different PSTVd-related RNAs, i.e., linear and circular ‘‘half molecules’’ derived from both the left and right halves of PSTVd as well as linear and circular forms of the full-length transcript. Although RNAs circularized in vitro presumably contain a 2*-phosphate group at the site of ligation, we have not observed any difference in the behavior of in vitro and in vivo circularized PSTVd RNAs with the same sequence (data not shown). Figure 4a shows the effects of a single U r G substitution at position 43 on the behavior of circularized PSTVd RNA transcripts. This mutation is predicted to close a small internal loop located at the left side of the VM region and immediately adjacent to nucleotides involved in the formation of secondary hairpin II

AID

viral

Effects of sequence variation on PSTVd replication Viroid titer is known to play an important role in modulating symptom expression (Sano et al., 1992), and in the case of PSTVd, several mutations which stabilize the pathogenicity domain have been shown to inhibit replication/accumulation (Owens et al., 1995). Progeny concentration in individual leaves of infected plants was monitored throughout the course of the bioassay, and data for selected variants collected during the mid–late stages of infection are presented in Fig. 3. As expected, PSTVd variants such as RG1 and M3/M1 which induced the most severe symptoms accumulated more rapidly and to higher titers than the less virulent ones (compare Figs. 3a and 3b). Later during infection, titers of the more severe variants tended to drop while those of milder variants continued to increase. Mutations which stabilize the pathogenicity domain (e.g., those at position 43, 318, or 311) led to a moderate–dramatic decrease in PSTVd replication/accumulation—especially early in infection. In other cases, however, sequence changes with no demonstrable structural effect were also seen to inhibit both PSTVd replication and symptom expression. One example of such a mutation is the A r G substitution at position 310 (see Fig. 3d). Effects of sequence variation on PSTVd denaturation

VY 8043

/

6a1b$$$341

07-17-96 13:21:58

AP: Virology

PSTVd PATHOGENICITY

149

FIG. 2. Symptoms of PSTVd infection in Rutgers tomato approximately 7 weeks after inoculation. Seedlings were inoculated with 0.04 ng/ml ribozyme-derived PSTVd RNA, and all 12 variants were compared in the same bioassay. (a) Naturally occurring PSTVd isolates. (b–d) Variants produced by site-directed mutagenesis of PSTVd-Intermediate. M3/M1 is identical in sequence to the naturally occurring PSTVd isolate KF440-1 (Schno¨lzer et al., 1985). (e–f) Quantitative assessment of stunting and epinasty/rugosity. All values are averages (six plants/treatment), and the lengths of individual segments within each bar in (e) represent the distance between successive nodes. The shaded portions of the bars shown in (f) denote leaves exhibiting epinasty/rugosity.

AID

VY 8043

/

6a1b$$8043

07-17-96 13:21:58

viral

AP: Virology

150

OWENS ET AL.

FIG. 3. Accumulation of PSTVd progeny in leaves of systemically infected tomatoes. Tissue samples were collected from individual leaves of infected plants at weekly intervals and pooled within treatments. Data from the mid–late stages of infection are presented.

(see Fig. 1). The effect of loop closure is clearly visible as an increase in the melting temperature (Tm) of the main transition. Closing this internal loop should also prevent disruption of base-pairing within PM 1 from spreading leftward, thereby releasing nucleotides required for the formation of secondary hairpin II. Formation of secondary hairpin II is responsible for the high cooperativity of the PSTVd-Intermediate main transition. Indeed, the denaturation of variant 43G can be seen to be less cooperative than that of PSTVd-Intermediate — especially at higher temperatures (see arrow). The structural effects of several other mutations were also readily apparent from the behavior of the respective full-length RNA transcripts. For example, Fig. 4d compares the denaturation profile of PSTVdIntermediate with that of a highly stabilized mutant containing a single A r G substitution at position 135 in the variable domain. In this case, the presence of an additional G:C base-pair raises the Tm of the circu-

larized full-length transcripts by 3 – 47. Note that this stabilizing effect increases by nearly twofold when half-molecules derived from the right half of PSTVd are compared. Figures 4b – 4d compare the denaturation profiles of PSTVd-Intermediate and PSTVd-Mild, two naturally occurring variants with very different biological properties. Even though comparison of their lowest freeenergy structures suggests that the VM region of PSTVd-Mild should be significantly more stable than that of PSTVd-Intermediate (see Fig. 1), we were unable to distinguish between these two variants by TGGE. Pairwise comparisons with an internal standard (compare Figs. 4c and 4d) showed their structural properties to be very similar/identical, and Fig. 4b shows that TGGE analysis under the same conditions failed to resolve a mixture of the full-length RNAs. The denaturation profiles of half molecules derived from the left half of PSTVd-Mild or -Intermediate were also indistinguishable (results not shown). In all, Tm values

AID

viral

VY 8043

/

6a1b$$$341

07-17-96 13:21:58

AP: Virology

PSTVd PATHOGENICITY

151

FIG. 4. Structural effects of sequence variation within the PSTVd pathogenicity domain. Samples containing approximately equal amounts of two circularized RNA transcripts were analyzed by TGGE. The wedge at the bottom of each panel indicates the position of the 25–657 temperature gradient, and the filled triangles denote the main transition of PSTVd-Intermediate. (a) PSTVd-Intermediate and 43G. Note the decreased cooperativity of the 43G transition, especially at higher temperatures (see arrow). (b) PSTVd-Intermediate and PSTVd-Mild. (c, d) Comparison of PSTVd-Mild (c) and PSTVd-Intermediate (d) with 135G, a highly stabilized mutant derived from PSTVd-Intermediate. As shown in (d), each sample contained six PSTVd-related RNAs; in addition to the full-length linear (FL) and circular (FC) transcripts, cleavage of FL near position 273 by a nuclease activity in the wheat germ extract produced two half-molecules (LL and RL) that were ligated to form LC and RC. Note the increased thermal stability and cooperativity associated with transcript circularization.

07-17-96 13:21:58 AID

VY 8043

/

6a1b$$8043

viral

AP: Virology

152

OWENS ET AL.

for a total of 12 infectious and noninfectious PSTVd variants were determined by TGGE. Quantitative relationship between pathogenicity domain structure and symptom expression To determine how PSTVd structure might modulate symptom expression, we sought to establish a quantitative relationship between these two properties. A thermodynamic approach to this question involved two steps, i.e., comparison of Tm values determined experimentally by TGGE with calculated Tm values followed by comparison of Tm values with disease severity. A second, more structural approach compared the optimal as well as suboptimal secondary structures for the pathogenicity domain of each variant. Schmitz and Steger (1992) have described a computer algorithm which can predict the equilibrium distribution of optimal and suboptimal RNA secondary structures at a variety of temperatures. Because the temperature dependence of these equilibrium distributions reflects denaturation behavior, it is possible to calculate the denaturation curve of an RNA. The accuracy of this methodology has been confirmed by comparison of the experimental and calculated denaturation curves for PSTVd-Intermediate (see Schmitz and Steger, 1992), and the same approach was used to compare the Tm values for the different transitions of our infectious variants. Results of these comparisons are presented in Figs. 5–8. In Fig. 5a, the calculated Tm values for the main transition have been plotted against Tm values determined experimentally by TGGE of circularized full-length PSTVd transcripts. The good correlation between experimental and calculated values emphasizes the accuracy of the predicted structural distributions which provide the basis for calculation of Tm values. The structures themselves as well as the extraordinary behavior of variants 311U and 310-311UU will be discussed below. In Fig. 5b, the calculated Tm values have been plotted against the average height of the diseased plants, a value used as an inverse measure of pathogenicity. Calculated Tm values were used for this plot in order to allow inclusion of variants such as 309A / 311U for which experimentally determined Tm values were not available. Although the relationship between thermal stability and pathogenicity appears to be reasonably linear, note that the difference between the Tm values for the least (RG1) and most stable (43G) variants is only 57 (approximately). Note also that, in agreement with the results of TGGE analysis, several pairs of variants (e.g., Intermediate and Mild or RG1 and M1) are predicted to have nearly identical Tm values despite large differences in pathogenicity. Consequently, a simple linear relationship between stability and pathogenicity seems unable to account for the behavior of all variants, and the fact that slowly replicating variants like 309A / 311U or 43G / 309A / 311U appear to exhibit such a correlation is

AID

VY 8043

/

6a1b$$$341

07-17-96 13:21:58

FIG. 5. Denaturation temperatures (Tm) for the main transition of circularized PSTVd transcripts. Correlations between (a) Tm values determined experimentally by TGGE and calculated Tm values as well as (b) calculated Tm values and heights of diseased plants are shown. The continuous line is the least squares fit of all data points, and only the differences between the experimentally determined Tm values for the different variants and that of PSTVd-Intermediate are shown in order to avoid extrapolation of Tm values from the low ionic strength conditions of TGGE (0.21 TBE–5 mM NaCl) to those used for thermodynamic calculations (1 M NaCl). For variants 318A, 43G, and 43G / 309A / 311U, the cooperative main transition is predicted to split into two separate, poorly resolved transitions; the mean value of both transitions is shown. Solid circles, replication more rapid than PSTVdIntermediate; half-filled circles, replication similar to PSTVd-Intermediate; open circles, replication slower than PSTVd-Intermediate; asterisks, noninfectious mutants.

probably fortuitous. Similar plots of Tm values versus percentage of leaves exhibiting epinasty or rugosity also failed to reveal a consistent correlation (results not shown). Use of the Tm value for the main transition to estimate the stability of the pathogenicity domain is not straightforward because the main transition is influenced by a competition between the rod-like native structure and a structure containing secondary hairpins I to III. For example, a C r G substitution at position 318 stabilizes the native structure by loop closure within PM 1; it also stabilizes secondary hairpin II by addition of a G:C base-pair. Such

viral

AP: Virology

PSTVd PATHOGENICITY

153

four sequence variants (i.e., Mild, Intermediate, and M3/ M1 plus PSTVd-Severe) provided the basis for an earlier proposal that PSTVd pathogenicity is inversely related to structural stability (see Schno¨lzer et al., 1985). Obviously, this correlation does not hold for all variants examined here, because several variants possess VM regions that are very similar in stability yet induce quite different symptoms (e.g., RG1 and M1 or Intermediate and 318G). The large deviations of 309A / 311U and 43G / 309A / 311U from the interpolated line in Fig. 7 can easily be explained by their slow replication rate. Suboptimal structures for the pathogenicity domain

FIG. 6. Denaturation temperature (Tm) of circularized PSTVd ‘‘lefthalf’’ molecules. For a description of symbols and lines, see Fig. 5. (a) Correlation between Tm values determined experimentally by TGGE and calculated Tm values. Only the differences between the experimentally determined Tm values for the different variants and that of PSTVdIntermediate are shown in order to avoid extrapolation of Tm values from the low ionic strength conditions of TGGE (0.21 TBE–5 mM NaCl) to those used for calculations (1 M NaCl). (b) Correlation between calculated Tm values and height of diseased plants.

The thermodynamic distribution of optimal and suboptimal secondary structures at 257 was also calculated for the pathogenicity domain of each variant. Selected results from this analysis are presented as ‘‘dot plots’’ in Fig. 8. Note that (i) these calculations were carried out on less-than-full-length RNAs in which the nucleotides comprising the pathogenicity domain (positions 36–67 and 292–325) are flanked by GC-rich ‘‘clamps’’ and (ii) dot sizes are proportional to the probability of base-pairing (i.e., only major species are shown). Figure 8a clearly shows that for PSTVd-Intermediate the lowest free energy structure (see Fig. 1) is the predominant species. From left to right, this structure contains a series of three short (i.e., 4-, 6-, and 4-bp) helices separated by internal loops containing, respectively, 2 and 4 nt. Several suboptimal structures are also visible; the most common alternative conformation involves a rearrangement of nucleotides within the leftmost 4-bp helix (see gray arrows in Fig. 8b). Figures 8b and 8c illustrate how sequence variation within the pathogenicity domain favors several additional supoptimal structures. The secondary structure distributions for the Mild and Intermediate strains appear very similar. For variant RG1, note that the left side of the

difficulties were circumvented by comparing the behavior of circularized half-molecules derived from the left side of PSTVd (see Fig. 6). The denaturation behavior of these truncated molecules more directly reflects the stability of the pathogenicity domain than does that of the fulllength circles. As shown in Fig. 6a, the correlation between experimental and calculated Tm values for the circularized half-molecules is quite good. Data presented in Fig. 6b, however, show that the relationship between stability and pathogenicity is not monotonic. In addition to the discrepancies involving RG1 and M1 or Intermediate and Mild noted above, three other variants that are more stable than PSTVd-Mild (i.e., 43G, 318A, and 318G) produced symptoms that were more (rather than less) severe than those of the Mild strain. Once again, these discrepancies were not resolved by using alternative measures of disease severity. Figure 7 compares the calculated stability of the VM region with pathogenicity. A similar plot involving only

FIG. 7. Correlation between calculated denaturation temperature (Tm) of the VM region and height of diseased plants. Calculations were performed as described by Schno¨lzer et al. (1985). The continuous line is the least squares fit of all data points except 309A / 311U and 43G / 309A / 311U; the dotted line includes all data points.

AID

viral

VY 8043

/

6a1b$$$341

07-17-96 13:21:58

AP: Virology

154

OWENS ET AL.

larger internal loop within PM 1, for both PSTVd-Intermediate and the progeny of 311U (i.e., 309A / 311U) contain a 4-nt loop. The dark arrows in Figs. 8b and 8c point to an internal loop separating the middle and right helices of PM 1; this loop may be either symmetrical (PSTVdIntermediate) or asymmetrical (Mild and RG1). Possible noncanonical interactions within the pathogenicity domain

FIG. 8. Base-pair probability plots for the pathogenicity domains of selected PSTVd variants. The structure distribution of each variant was calculated from the optimal and 30 suboptimal secondary structures at 257. The probability of each possible base-pair (i:j) is proportional to the area of the dot at the corresponding position (i,j) of the matrix. Calculations were carried out for subsequences of the pathogenicity domain (positions 36–67 and 292–325) flanked by G:C-rich clamps. These clamps simulate the stability of missing portions of the PSTVd native structure (see Riesner, 1990). (a) Base-pair probability plot for the pathogenicity domain of PSTVd-Intermediate. Only the upper portion of the matrix is shown; the lower portion is a mirror image of the upper. (b, c) Corresponding portions near the diagonal (top left to bottom right) from probability plots of selected variants. Shaded arrows indicate differences in the distributions of suboptimal structures; dark arrows point to the position of the loop separating the two helices within the VM region. Note the differences in loop type among the different PSTVd variants.

Although hydrodynamic and ligand binding studies suggest that PSTVd contains little significant tertiary structure (reviewed by Riesner, 1990), ultraviolet crosslinking has identified a unique region of local tertiary structure in the CCR similar to loop E in 5S ribosomal RNA (Branch et al., 1985; Wimberley et al., 1993). The behavior of certain PSTVd variants during TGGE as well as comparisons of their suboptimal structures provides circumstantial evidence for the existence of additional ‘‘noncanonical’’ interactions within the pathogenicity domain. For example, TGGE analysis was unable to distinguish the stabilizing effects of a single C r U substitution at position 311 from those of a double AC r UU substitution at positions 310–311 (see Figs. 5a and 6a). As shown in Fig. 8c, in 311U the 4-nt loop of PSTVd-Intermediate is reduced to a 2-nt loop with an A51rA310 opposition; formation of a second A:U base-pair in 310–311UU results in the fusion of the two short helices, thereby creating the longest uninterrupted helix in the molecule. Failure of an A r G substitution at position 310 to affect TGGE behavior (see Figs. 5a and 6a) argues against the existence of an interaction involving A51 and A310 in PSTVd-Intermediate, but reducing the loop size to 2 nt via a C r U substitution at position 311 might facilitate such an interaction. The resulting stabilization of the pathogenicity domain in variant 311U by pairing of A51rA310 would explain the only deviation in the correlation of experimental and calculated Tm values. If the pairing does exist, this ArA interaction would appear to inhibit PSTVd replication, for the presence of a C r U substitution at position 311 consistently induced the appearance of a spontaneous U r A change at position 309. One effect of the spontaneous mutation might be to disrupt the putative ArA interaction. DISCUSSION

pathogenicity domain may assume any of three different conformations. Figure 8c compares the structures accessible to the PSTVd variant with a C r U substitution at position 311 with those of its progeny which contain an apparently compensatory U r A change at position 309. Comparison of these plots suggests that the additional mutation has little or no effect on the relative probabilities of different secondary structures. Evidently, selective pressures in vivo may strongly favor the presence of a

As discussed by Diener (1987), the range of symptoms associated with viroid infection is virtually identical to that produced by conventional plant viruses. In the apparent absence of viroid-encoded proteins to act as elicitors of the host response, the nature of the signal(s) mediating viroid–host interaction is a matter of considerable interest. Variation in the severity of symptoms accompanying PSTVd infection is correlated with sequence changes in a portion of the VM region (Schno¨lzer et al., 1985), an observation that has led to the assumption that pathogenicity must be regulated by nucleotides within this region.

AID

viral

VY 8043

/

6a1b$$$341

07-17-96 13:21:58

AP: Virology

PSTVd PATHOGENICITY

Although several mechanistic models have been proposed (cf. below), it is far from clear how this regulation is accomplished on the molecular level. In the present work, we have characterized 12 PSTVd sequence variants with respect to their pathogenicity and time course of progeny accumulation as well as certain structural features of their VM regions. All but one of the sequence differences among these variants are located within the pathogenicity domain. For some parameters, significant correlations with the pathogenicity could be obtained; for others, correlations were either imperfect or missing altogether. Pathogenicity and the rate of PSTVd replication To date, strain-dependent differences in the rate of PSTVd replication or cell-to-cell spread have received comparatively little attention, and mechanistic models for PSTVd pathogenicity assume equal concentrations of each variant. This is undoubtedly an oversimplification, even for naturally occurring strains of PSTVd. For example, RG1 (and other severe variants) could be shown to accumulate more rapidly than milder variants early in infection (see Fig. 3). For slowly replicating variants such as 43G or 309A / 311U, the primary effect of the mutation(s) may be on replication, and the effects on symptom expression are likely to be indirect. The seven variants which accumulate at relatively high rates (i.e., M1, M2/ M1, Intermediate, 310G, Mild, 318A, and 318G), produced quite divergent disease symptoms, however.

155

calculated Tm values seen in Fig. 6a provides assurance that the thermodynamic calculations for not only the left half of PSTVd but also for the VM region are reasonably accurate. Figure 7b contains a plot of Tm values for the VM region vs plant height (the inverse of pathogenicity). Note that there is only a weak correlation between pathogenicity and thermodynamic instability, and several pairs of variants such as Intermediate and 318G or RG 1 and M1 do not fit the correlation. Omitting the slowly replicating variants 309A / 311U and 43G / 309A / 311U would not improve the correlation. The strong correlation between experimental and theoretical thermodynamic data seen in Fig. 6a indicates that the problem lies not with inaccuracies in the thermodynamic calculations but rather calls the model itself into question. Sequences outside the pathogenicity domain as site(s) for pathogenic interaction

An apparent inverse correlation between symptom severity and the calculated stability of the VM region initially suggested that interaction with the host might require the breakdown of normal secondary structure within the pathogenicity domain (Schno¨lzer et al., 1985). Extensive complementarity between nucleotides in that region and a segment of tomato 7S RNA led to the suggestion that viroids might inhibit the incorporation of 7S RNA into signal recognition particles (Haas et al., 1988). More recently, however, the same group determined the sequence of 7S RNA from several different tomato cultivars and reported that symptom severity was not correlated with the degree of complementarity (Riedel et al., 1995). Whatever the nature of the host molecule interacting with the VM region in the dissociated state, we wished to test the proposed inverse correlation with a larger series of PSTVd variants. In fact, two types of correlations could be tested. From a purely physical perspective, it was important to experimentally confirm the reliability of the thermodynamic predictions. This could not be done with the isolated VM region, but half-molecules containing the entire pathogenicity domain were accessible to TGGE analysis. The good correlation between experimentally determined and

By exchanging individual structural domains between CEVd and tomato apical stunt viroid, Sano et al. (1992) have shown that sequences within the TL , pathogenicity, and variable/TR domains all contribute to viroid pathogenicity and replication. The recent demonstration by Gruner et al. (1995) that mutations within the VM region of PSTVd can indeed affect the stability of the whole molecule provides further evidence that sequences within the pathogenicity domain may not act alone to regulate symptom expression. Their demonstration that more severe strains of PSTVd are able to outcompete milder (i.e., less easily denaturable) strains during mixed infections suggested an alternative hypothesis for viroid pathogenicity, namely, that the VM region modulates the accessibility of a second, probably nonoverlapping region in the molecule which acts as the primary site of viroid–host interaction. Like the model proposed by Schno¨lzer et al. (1985), this hypothesis minimizes the possible role of the native structure in regulating viroid– host interaction; instead, it focuses on metastable structure(s) generated during synthesis and active in replication as well as pathogenesis. A more stable rod-like native structure would depopulate the active (but metastable) structure more rapidly, thus explaining the correlation between pathogenicity and instability of the native structure. When Gruner et al. (1995) argued that PSTVd replication and pathogenicity are governed by the same or closely related metastable structures within a replication intermediate, they predicted a faster conversion of the metastable state into the rod-like native structure for milder strains. In the absence of kinetic data, the relative stability of the native structure was used as a rough approximation of kinetic behavior. Our more complete biological data, together with additional thermodynamic experimentation and theoretical calculations, do not support such a mechanistic model for PSTVd–host interaction. Despite a surprisingly good correlation between the

AID

viral

A denatured VM region as the site for pathogenic interaction

VY 8043

/

6a1b$$$341

07-17-96 13:21:58

AP: Virology

156

OWENS ET AL.

FIG. 9. Schematic representations of the PSTVd pathogenicity domain (positions 36–67 and 292–325) flanked by G:C-rich clamps. A standard A-type RNA helix is assumed, and the helix axis (indicated by dotted lines) is bent to compensate for missing nucleotides in asymmetrical loops. As described by Riesner (1990), the G:C-rich clamps simulate the stability conferred by the missing portions of the native structure. (a) Thermodynamically optimal secondary structures of selected PSTVd variants. (b) Two-dimensional representations of the corresponding three-dimensional structures. For each variant, the vertical arrows denote corresponding positions in the two- and three-dimensional structures.

experimentally determined and calculated Tm values (cf. Fig. 5a), PSTVd pathogenicity (as measured by plant height) was only weakly correlated with instability of the whole molecule. Furthermore, certain pairs of variants (e.g., Intermediate and 318G or RG1 and M1) did not obey the proposed inverse relationship.

Other models attempting to explain PSTVd pathogenicity assume specific interactions between host components (most probably proteins) and certain elements of the native structure. Several complexes containing PSTVd and plant nuclear proteins (Mr Å 33–92 kDa) have been isolated from infected tomato leaf tissue (Klaff et al., 1989). These interactions appear to be primarily electrostatic in nature, but no information is available about the site(s) of interaction. More recently, Diener et al. (1993) have suggested that the triggering event in PSTVd pathogenesis may be its interaction with a host cell protein kinase. Preliminary evidence suggests that PSTVd infection may activate a plant homolog of the interferon-

induced, dsRNA-activated mammalian protein kinase known to interact with small RNA molecules encoded by adenovirus or human immunodeficiency virus (see also Hiddinga et al., 1988). Like PSTVd, adenovirus VA RNAI is a highly structured, single-stranded RNA, and its approximately 160 nt are arranged in two base-paired stems connected by a complex stem-loop structure known as the ‘‘central domain.’’ Although the relative contributions of the apical stem and central domain to the ability of VA RNA to bind to the kinase and inhibit its activation are controversial, VA RNA function both in vivo and in vitro appears to be critically dependent upon subtle alterations in the secondary or tertiary structure of the central domain (e.g., Clarke et al., 1994; Ghadge et al., 1994; Schmedt et al., 1995; Clarke and Matthews, 1995). The strength of the primary interaction between PSTVd and a host cell component might depend on similar, subtle variations in the three-dimensional geometry of the VM region. We compared the secondary structures of our PSTVd variants in an effort to derive general rules for the threedimensional structure of the VM region. Sequence differ-

AID

viral

Three-dimensional structure of the pathogenicity domain

VY 8043

/

6a1b$$$341

07-17-96 13:21:58

AP: Virology

PSTVd PATHOGENICITY

ences are restricted to nucleotides 42–52 and 319–309, and except for variants 43G, 318A, and 318G where the left loop has been replaced by a base-pair, these nucleotides are organized into a central 4- to 7-bp double-helical segment flanked by internal loops. The size and the symmetry of these flanking loops determine the relative orientation of the remaining portions of the rod-like native structure with respect to the central helix. Figure 9a contains two-dimensional representations of the thermodynamically most stable structure for PSTVd-Mild, -Intermediate, -Severe (see Schno¨lzer et al., 1985), and RG1. In comparing these structures, it is important to note that the central helix forms about half a helical turn and, therefore, the portion of the structure containing the left terminal loop is rotated ca. 1807 with respect to that containing the central conserved region and right terminal loop. This is shown schematically in Fig. 9b. Although the absolute angles associated with different asymetrical loops are unknown, certain characteristic differences between the variants are evident. Relative to the central helix, the right ‘‘arm’’ of all variants whose symptoms are more severe than PSTVd-Intermediate can be seen to bend ‘‘downward.’’ For RG1 (the most severe variant) both the left and right arms are bent downward. The presence of symmetrical loops in the VM region of PSTVd-Intermediate results in a coaxial arrangement of helical segments, and the right arm of PSTVd-Mild can be seen to bend ‘‘upward.’’ The same tendencies were also observed among the other variants included in our study as well as the six additional variants described by Herold et al. (1992). This would suggest that the structure of RG1 and other severe variants is optimal for pathogenic interaction, possibly involving the central helix and both arms, and that such interaction(s) is either greatly inhibited or completely suppressed for PSTVd-Mild. Of the three properties examined for their correlation with pathogenicity (i.e., instability of the VM region, instability of the whole molecule, and three-dimensional structure of the VM region), our results clearly point to the three-dimensional structure of the VM region as the best predictor of PSTVd pathogenicity. In certain situations, however, thermodynamic features of PSTVd structure may also be mechanistically relevant. For example, the decreased stability of the more pathogenic strains could facilitate binding of a ‘‘bent’’ molecule to the interacting surface of a host protein. In addition, optimum replication of PSTVd may also require a certain degree of instability in the loop immediately to the right of the central helix. In the most slowly replicating of our variants (i.e., 311U), purine:purine interactions similar to those within the Eloop of 5S RNA (Wimberly et al., 1993) and involving A51 and A310 may stabilize the structure of the VM region and inhibit the rate of PSTVd replication. The spontaneous U r A change which consistently appeared at position 309 may relieve this inhibition by destabilizing the neighboring 4-bp helix. Lone pairs (i.e., isolated interactions that are not suffi-

AID

VY 8043

/

6a1b$$$341

07-17-96 13:21:58

157

ciently stable to exist in their own right) often serve as recognition sites for RNA-binding proteins (Puglisi et al., 1992; Battiste et al., 1994), and phylogenetic evidence points to an important role for these and other unusual structural elements in maintaining the overall structure of rRNA (reviewed by Gutell et al., 1994). In a similar fashion, the strength of the interaction between PSTVd and a host component appears to depend on subtle variations in the three-dimensional structure of the VM region. Further consideration of the interactions mediating PSTVd pathogenicity awaits direct evidence for the interaction of host proteins with nucleotides within the pathogenicity domain. ACKNOWLEDGMENTS The first two authors contributed equally to these studies. Y.H. was supported by a grant from the USDA/NRI Competitive Grants Program (Grant 91-373303-6649). Additional support was provided by grants from Deutsche Forschunggemeinschaft and Fonds der Chemischen Industrie as well as an agreement between the USDA and the Bundesminister fu¨r Landwirtschaft. We thank S. M. Thompson for technical assistance and J. N. Culver, A. Mattoo, W. O. Dawson, and P. A. Feldstein for helpful discussions and critical review of the manuscript.

REFERENCES Antao, V. P., and Tinoco, I., Jr. (1992). Thermodynamic parameters for loop formation in RNA and DNA hairpin tetraloops. Nucleic Acids Res. 20, 819–824. Antao, V. P., Lai, S. Y., and Tinoco, I., Jr. (1991). A thermodynamic study of unusually stable RNA and DNA hairpins. Nucleic Acids Res. 19, 5901–5905. Battiste, J. L., Tan, R., Frankel, A. D., and Williamson, J. R. (1994). Binding of an HIV Rev peptide to Rev responsive element RNA induces formation of purine-purine base pairs. Biochemistry 33, 2741–2747. Baumstark, T., and Riesner, D. (1995). Only one of four possible secondary structures of the central conserved region of potato spindle tuber viroid is a substrate for processing in a potato nuclear extract. Nucleic Acids Res. 23, 4246–4254. Branch, A. D., Benenfeld, B. J., and Robertson, H. D. (1985). Ultraviolet light-induced crosslinking reveals a unique region of local tertiary structure in potato spindle tuber viroid and HeLa 5S RNA. Proc. Natl. Acad. Sci. USA 82, 6590–6594. Clarke, P. A., and Matthews, M. B. (1995). Interactions between the double-stranded RNA binding motif and RNA: definition of the binding site for the interferon-induced protein kinase DAI on adenovirus VA RNA. RNA 1, 7–20. Clarke, P. A., Pe’ery, T., Ma, Y., and Matthews, M. B. (1994). Structural features of adenovirus 2 virus-associated RNA required for binding to the protein kinase DAI. Nucleic Acids Res. 22, 4364–4374. Cress, D. E., Kiefer, M. C., and Owens, R. A. (1983). Construction of infectious potato spindle tuber viroid cDNA clones. Nucleic Acids Res. 11, 6821–6835. Diener, T. O. (Ed.) (1987). The Viroids. Plenum, New York. Diener, T. O., Hammond, R. W., Black, T., and Katze, M. G. (1993). Mechanism of viroid pathogenesis: Differential activation of the interferon-induced, double-stranded RNA-activated, Mr 68,000 protein kinase by viroid strains of varying pathogenicity. Biochimie 75, 533– 538. Ghadge, G., Malhotra, P., Furtado, M. R., Dhar, R., and Thimmapaya, B. (1994). In vitro analysis of virus-associated RNA I (VAI RNA): Inhibition of the double-stranded RNA-activated protein kinase PKR by VAI RNA mutants correlates with the in vivo phenotype and the structural integrity of the central domain. J. Virol. 68, 4137–4151.

viral

AP: Virology

158

OWENS ET AL.

Groebe, D. R., and Uhlenbeck, O. C. (1988). Characterization of RNA hairpin loop stability. Nucleic Acids Res. 16, 11725–11735. Gross, H. J., Domdey, H., Lossow, C., Jank, P., Raba, M., Alberty, H., and Sa¨nger, H. L. (1978). Nucleotide sequence and secondary structure of potato spindle tuber viroid. Nature 273, 203–208. Gruner, R., Fels, A., Qu, F., Zimmat, R., Steger, G., and Riesner, D. (1995). Interdependence of pathogenicity and replicability with potato spindle tuber viroid. Virology 209, 60–69. Gutell, R. R., Larsen, N., and Woese, C. R. (1994). Lessons from an evolving RNA: 16S and 23S rRNA structures from a comparative perspective. Microbiol. Rev. 58, 10–26. Hammond, R. W. (1992). Analysis of the virulence modulating region of potato spindle tuber viroid (PSTVd) by site-directed mutagenesis. Virology 187, 654–662. Haas, B., Klanner, A., Ramm, K., and Sa¨nger, H. L. (1988). The 7S RNA from tomato leaf tissue resembles a signal recognition particle RNA and exhibits a remarkable sequence complementarity to viroids. EMBO J. 7, 4063–4074. Hecker, R. (1989). Thesis, Heinrich-Heine Universita¨t, Du¨sseldorf. Hecker, R., Wang, Z., Steger, G., and Riesener, D. (1988). Analysis of RNA structure by temperature-gradient gel electrophoresis: Viroid replication and processing. Gene 72, 59–74. Herold, T., Haas, B., Singh, R. P., Boucher, A., and Sa¨nger, H. L. (1992). Sequence analysis of five new field isolates demonstrates that the chain length of potato spindle tuber viroid (PSTVd) is not strictly conserved but as variable as in other viroids. Plant Mol. Biol. 19, 329–333. Hiddinga, H. J., Crum, C. J., Hu, J., and Roth, D. A. (1988). Viroid-induced phosphorylation of a host protein related to a dsRNA-dependent protein kinase. Science 241, 451–453. Keese, P., and Symons, R. H. (1985). Domains in viroids: Evidence of intermolecular RNA rearrangement and their contribution to viroid evolution. Proc. Natl. Acad. Sci. USA 82, 4582–4586. Klaff, P., Gruner, R., Hecker, R., Sa¨ttler, A., Theissen, G., and Riesner, D. (1989). Reconstituted and cellular viroid-protein complexes. J. Gen. Virol. 70, 2257–2270. Loss, P. (1989). Thesis, Heinrich-Heine-Universita¨t, Du¨sseldorf. Loss, P., Schmitz, M., Steger, G., and Riesner, D. (1991). Formation of a thermodynamically metastable structure containing hairpin II is critical for infectivity of potato spindle tuber viroid RNA. EMBO J. 10, 719–727. McCaskill, J. S. M. (1990). The equilibrium partition function and base pair binding probabilities for RNA secondary structure. Biopolymers 29, 1105–1119. Owens, R. A., and Diener, T. O. (1984). Spot hybridization for detection of viroids and viruses. In ‘‘Methods in Virology’’ (K. Maramorosch and H. Koprowski, Eds.), Vol. VII, pp. 173–187. Academic Press, New York. Owens, R. A., Hammond, R. W., Gardner, R. C., Kiefer, M. C., Thompson, S. M., and Cress, D. E. (1986). Site-specific mutagenesis of potato spindle tuber viroid cDNA. Plant Mol. Biol. 6, 179–192. Owens, R. A., Candresse, T., and Diener, T. O. (1990). Construction of novel viroid chimeras containing portions of tomato apical stunt and citrus exocortis viroids. Virology 175, 238–246. Owens, R. A., Chen, W., Hu, Y., and Hsu, Y-H. (1995). Suppression of potato spindle tuber viroid replication and symptom expression by

mutations which stabilize the pathogenicity domain. Virology 208, 554–564. Palukaitis, P., Cotts, S., and Zaitlin, M. (1985). Detection and identification of viroids and viral nucleic acids by ‘‘dot-blot’’ hybridization. Acta Horticult. 164, 109–118. Puglisi, J. D., Tan, R., Calnan, B. J., Frankel, A. D., and Williamson, J. R. (1992). Conformation of the TAR RNA-arginine complex by NMR spectroscopy. Science 257, 76–80. Riedel, L., Pu¨tz, A., Hauser, M. T., Luckinger, R., Wassenegger, M., and Sa¨nger, H. L. (1995). Characterization of the signal recognition particle (SRP) RNA population of tomato (Lycopersicon esculentum). Plant Mol. Biol. 27, 669–680. Riesner, D. (1987). Structure formation. In ‘‘The Viroids’’ (T. O. Diener, Ed.), pp. 63–98. Plenum, New York. Riesner, D. (1990). Structure of viroids and their replication intermediates. Are thermodynamic domains also functional domains? Semin. Virol. 1, 83–99. Riesner, D., and Steger, G. (1990). Viroid and viroid-like RNA. In ‘‘Landolt-Bo¨rnstein—Group VII Biophysics’’ (W. Saenger, Ed.), Vol. 1— Nucleic Acids, Subvol. d—Physical Data, II—Theoretical Investigations, pp. 194–243. Springer-Verlag, Berlin. Riesner, D., Klaff, P., Steger, G., and Hecker, R. (1987). Viroids: Subcellular location and structure of replication intermediates. In ‘‘Endocytobiology III’’ (J. J. Lee and J. F. Fredrick, Eds.), Vol. 503, pp. 212–237. N.Y. Acad. Sci., New York. Rosenbaum, V., and Riesner, D. (1987). Temperature gradient gel electroporesis: Thermodynamic analysis of nucleic acids and proteins in purified form and in celllular extracts. Biophys. Chem. 26, 235–246. Sambrook, J., Fritsch, E. F., and Maniatis, T. (1989). ‘‘Molecular Cloning—A Laboratory Manual.’’ Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY. 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. Schmedt, C., Green, S. R., Manche, L., Taylor, D. R., Ma, Y., and Matthews, M. B. (1995). Functional characterization of the RNA-binding domain and motif of the double-stranded RNA-dependent protein kinase DAI (PKR). J. Mol. Biol. 249, 29–44. Schmitz, M., and Steger, G. (1992). Base-pair probability profiles of RNA secondary structures. Comp. Appl. Biosci. 8, 389–399. Schno¨lzer, M., Haas, B., Ramm, K., Hofmann, H., and Sa¨nger, H. L. (1985). Correlation between structure and pathogenicity of potato spindle tuber viroid (PSTV). EMBO J. 4, 2181–2190. Semancik, J. S. (Ed.) (1987). ‘‘Viroids and Viroid-Like Pathogens.’’ CRC Press, Boca Raton. Steger, G., Hofmann, H., Fo¨rtsch, J., Gross, H. J., Randles, J. W., Sa¨nger, H. L., and Riesner, D. (1984). Conformational transitions in viroids and virusoids: Comparison of results from energy minimization algorithm and from experimental data. J. Biomol. Struct. Dyn. 2, 543–571. Visvader, J. A., and Symons, R. H. (1985). Eleven new sequence variants of citrus exocortis viroid and correlation of sequence with pathogenicity. Nucleic Acids Res. 13, 2907–2920. Wimberly, B., Varani, G., and Tinoco, I., Jr. (1993). The conformation of loop E of eukaryotic 5S ribosomal RNA. Biochemistry 32, 1078–1087. Zimmat, R., Gruner, R., Hecker, R., Steger, G., and Riesner, D. (1990). Analysis of mutations in viroid RNA by non-denaturing and temperature gradient gel electrophoresis. In ‘‘Proceedings of the 6th Conversation in Biomolecular Stereodynamics’’ (R. Sarma and M. Sarma, Eds.), Vol. 3, pp. 339–357. Adenine Press, Schenectady, NY.

AID

viral

VY 8043

/

6a1b$$$341

07-17-96 13:21:58

AP: Virology