Evidence for the Existence of the Loop E Motif of Potato Spindle Tuber Viroid In Vivo

Abstract

RNA motifs comprising nucleotides that interact through non-Watson-Crick base pairing play critical roles in RNA functions, often by serving as the sites for RNA-RNA, RNA-protein, or RNA small ligand interactions. The structures of viral and viroid RNA motifs are studied commonly by in vitro, computational, and mutagenesis approaches. Demonstration of the in vivo existence of a motif will help establish its biological significance and promote mechanistic studies on its functions. By using UV cross-linking and primer extension, we have obtained direct evidence for the in vivo existence of the loop E motif of Potato spindle tuber viroid. We present our findings and discuss their biological implications.

RNA motifs comprising nucleotides that interact through non-Watson-Crick base pairing play critical roles in RNA functions, often by serving as the sites for interactions with proteins, other RNA motifs, or small ligands (19). For viral and viroid RNAs, such interactions are crucial for the establishment of infection. The structures of RNA motifs of viral and viroid RNAs have been generally studied by in vitro chemical/enzymatic mapping, thermodynamic calculations using minimum free energy, biophysical characterization, and mutational analyses. It has been shown experimentally that some structural features of RNAs deduced by in vitro and in silico methods may not be identical to those found in vivo, likely because of the influence of cellular factors that interact with the RNAs (2, 26, 30). Therefore, to further establish the biological significance of findings from in vitro and in silico approaches, the in vivo existence of an RNA motif should be directly demonstrated.

We are interested in using viroid infection as a model system to investigate the structure-function relationships of RNA motifs. Viroids are the smallest known nucleic acid-based infectious agents and self-replicating genetic units. Their genomes consist of single-stranded, circular RNAs ranging in size from 250 to 400 nucleotides (12, 13, 33). Viroids do not encode proteins, do not have encapsidation mechanisms, and do not require helper viruses. Nonetheless, they can replicate efficiently and spread throughout an infected plant (13, 33). Thus, viroid infection provides an excellent experimental system to investigate the basic RNA structure-function relationships for infection as well as for RNA biology (18).

One of the best-studied viroid RNA motifs is the so-called loop E located in the central conserved region of Potato spindle tuber viroid (PSTVd). The loop E is a recurrent motif found in many RNAs, including 5S, 16S, and 23S rRNAs, group I and group II introns, bacterial RNase P, ribozyme of Tobacco ringspot virus satellite RNA (20), and lysine riboswitches (15, 31), where it plays critical roles in RNA-RNA and RNA-protein interactions. UV-induced cross-linking between G98 and U260 in vitro provided the first evidence for the existence of local tertiary structure in the loop E of PSTVd resembling that of the loop E of 5S rRNA from HeLa cells (6) (Fig. 1). A recent model depicts all of the non-Watson-Crick base pairs within the PSTVd loop E (37) (Fig. 1). This motif is involved in replication (24, 37), in vitro processing (3), pathogenicity (25), and host adaptation (24, 35, 38).

FIG. 1.

Secondary structure of PSTVd (upper panel) (14) and non-Watson-Crick base pairs in loop E (lower panel) (37). The symbols that denote each of the specific base edge-edge interactions are defined below the structure (for details, see references 21 and 37). The dashed line indicates G98 and U260, which can be cross-linked by UV irradiation. R1, R2, and R3 indicate the positions of the primers used in the primer extension experiments.

In this study we investigated the existence of PSTVd loop E in infected tomato leaves by UV cross-linking, following the protocols of Daròs and Flores (10). Pilot experiments established that 80 min of 254-nm UV irradiation (10 J/cm²; Stratalinker model 1800; Stratagene, La Jolla, CA) was optimal to detect cross-linked products derived from PSTVd. After UV irradiation, we extracted total RNAs from the leaves for RNA gel blot analyses using riboprobes specific for the (+)-strand PSTVd. Compared to nonirradiated samples (Fig. 2, lane 1), a band with a gel mobility that is intermediate between those of the circular and linear PSTVd RNAs was detected (Fig. 2, lane 2). Proteinase K treatment showed that the presence of this intermediate band was not due to protein binding to the viroid RNA (Fig. 2, lane 3).

FIG. 2.

RNA gel blot showing the appearance of a UV-cross-linked product (arrow) of PSTVd RNAs from infected tomato leaves. Treatment with proteinase K (0.5 units for 30 min; Invitrogen) indicates that this cross-linked product is not attributed to protein binding. c-PSTVd and l-PSTVd denote circular and linear PSTVd RNAs, respectively. For gel blotting, total RNAs were extracted from infected tomato leaves that were untreated or irradiated with UV for 80 min (10 J/cm²) by using the RNeasy plant mini kit (QIAGEN, Valencia, CA). RNA samples were run on 5% polyacrylamide-8 M urea gels, transferred to Hybond-XL nylon membranes (Amersham Biosciences, Piscataway, NJ) using a vacuum blotting system (Amersham), and immobilized by UV cross-linking. After overnight hybridization with [α-³²P]UTP-labeled riboprobes at 65°C in ULTRAhyb reagent (Ambion, Austin, TX), the membranes were washed twice at 65°C in 2× SSC-0.1% sodium dodecyl sulfate (1× SSC is 0.15 M NaCl plus 0.015 M sodium citrate) and 0.2× SSC-0.1% sodium dodecyl sulfate and exposed to a Storage phosphor screen (Kodak, Rochester, NY).

To determine whether the UV-cross-linked product was derived from the circular or linear PSTVd RNA or both, we gel purified the circular and linear RNAs from infected tomato leaves for UV cross-linking. As shown in lane 3 of Fig. 3, UV treatment of the circular RNAs yielded a product with a mobility similar or identical to that detected from UV-treated tomato leaves. UV treatment of the linear RNAs yielded a product of faster mobility (Fig. 3, lane 5). The different mobilities of the cross-linked products derived from the circular and linear PSTVd RNAs are likely attributable to their different conformations. These results suggest that the UV-cross-linked product from PSTVd-infected plants was mostly derived from the circular genomic RNA of PSTVd. The cross-linked product derived from the linear RNAs in tomato leaves was present at a very low level, likely because of the relatively low amounts of linear RNAs, which were barely visible only with overexposure (data not shown). The band observed between the cross-linked product and the linear substrate RNA was also observed in previous experiments, and its nature remains to be determined (28, 37).

FIG. 3.

RNA gel blot showing the origin of the in vivo UV-cross-linked product predominantly from the circular PSTVd RNAs. Total RNAs were extracted and run on 5% polyacrylamide-8 M urea gels as described in the legend for Fig. 2. The in vivo circular PSTVd (c-PSTVd) and linear PSTVd (l-PSTVd) RNAs were gel purified and irradiated with UV for 5 min in EXL buffer (600 mM NaCl, 4 M urea, 1 mM cacodylate, 0.1 mM EDTA) as described by Zhong et al. (37). In vitro transcripts of PSTVd were obtained by transcribing HindIII-linearized pRZ6-2 template (17) by utilizing a T7 MAXIscript kit (Ambion). Unit-length l-PSTVd RNAs were gel purified from the 5% polyacrylamide-8 M urea gel. Incubation of the l-PSTVd transcripts in wheat germ extract gave rise to the circular PSTVd RNAs, which were run on a 5% polyacrylamide-8 M urea gel followed by gel purification. Both c-PSTVd and l-PSTVd in vitro transcripts were subjected to UV irradiation for 5 min in the EXL buffer. Approximately 100 to 200 ng of RNA was loaded for each lane.

It has been well established, via enzymatic mapping (6), primer extension (3), and mutational (37) analyses, that in vitro UV cross-linking of PSTVd occurs within loop E. Thus, a key question is whether the observed UV-induced cross-linking in vivo also occurs within loop E. To address this question, we first compared the gel mobilities of cross-linked products from circular and linear in vitro transcripts of PSTVd with those from the in vivo PSTVd RNAs, under the assumption that cross-linking at the same sites should give rise to products of the same gel mobilities for either the circular or linear substrate RNAs. We used linearized plasmid pRZ6-2 (17) as a template to synthesize the linear PSTVd transcripts. Ribozyme-mediated self-cleavage of the transcripts gave rise to unit-length PSTVd RNAs. Incubation of the linear PSTVd transcripts in wheat germ extract yielded circular PSTVd RNAs (11, 37). Both the unit-length linear and circular RNAs were gel purified for UV treatment. As shown in lanes 7 and 9 of Fig. 3, the UV-cross-linked products derived from the in vitro circularized and linear RNAs had gel mobilities comparable to those derived from the in vivo PSTVd RNAs. As a negative control, UV irradiation of green fluorescent protein RNAs did not produce any cross-linked products (data not shown) (37). These results suggest that UV-induced cross-linking of PSTVd RNAs in vivo occurs within loop E, as is the case for UV-induced cross-linking in vitro.

To demonstrate directly UV-induced cross-linking within loop E in vivo, we conducted primer extension experiments. We gel purified the UV-cross-linked as well as the circular PSTVd RNAs from infected tomato leaves as templates for reverse transcription to map one of the candidate cross-linking sites, G98. The reverse transcription was carried out following the protocols described by Baumstark et al. (3) with modifications, using primer R1 (5′-AAACCCTGTTTCGGCGGGAATTAC-3′, complementary to position 154-179 on the PSTVd genome) (Fig. 1) and the SuperScript III RNase H- reverse transcriptase (Invitrogen, Carlsbad, CA) at 52°C for 40 min. As shown in Fig. 4, reverse transcription terminated predominantly at A99 when the cross-linked PSTVd RNA was used as the template (lane cI), compared to the situation when the non-cross-linked circular PSTVd RNA was used as the template (lane c). Identical results were obtained when two other primers (R2, 5′-TTTCGGCGGGAATTACTCCTGTCGGC-3′, complementary to position 146-171 on the PSTVd genome, and R3, 5′-ATTACTCCTGTCGGCCGCTGGG-3′, complementary to position 139-160 on the PSTVd genome) were used for reverse transcription (Fig. 4). All together, these results indicated that G98 was a cross-linking site, identical to that mapped in vitro (3, 6), providing direct evidence for UV-induced cross-linking within loop E in vivo.

FIG. 4.

Primer extension mapping of an in vivo cross-linking site to G98 of loop E, using three different primers as indicated (R1, R2, and R3). Lane c, cDNAs from reverse transcription of the circular PSTVd RNA template purified from infected tomato leaves; lane cI, cDNAs from reverse transcription of the UV cross-linked circular PSTVd RNA template purified from infected leaves. The reverse transcription was performed by following the protocols of Baumstark et al. (3) with modifications, using Invitrogen Superscript III RNase H- reverse transcriptase and ³²P-labeled primers for 40 min at 52°C. Lanes U to A, sequencing ladders generated by PCR with pRZ6-2 template and ³²P-labeled primers in the presence of ddATP, ddCTP, ddGTP, and ddUTP by utilizing the Thermo Sequenase cycle sequencing kit (USB, Cleveland, OH). All samples were run on an 8% polyacrylamide-8 M urea sequencing gel. The nucleotide sequence of the (+)-PSTVd is given on the left, and the arrows point to the bands corresponding to reverse transcription termination at A99.

Comparative sequence analysis using isostericity matrices in combination with mutational and functional studies suggested that the PSTVd loop E exists and functions in vivo (37). The results from the UV-cross-linking experiments presented here provide conclusive evidence for the in vivo existence of PSTVd loop E, establishing a firm structural foundation for further mechanistic studies on the role of this motif in the viroid life cycle. Another important ramification is that these findings support the notion that the rod-shaped native structure of PSTVd also exists in vivo. This is consistent with the observation that deletion mutations that restored the rod-shaped PSTVd structure restored infectivity (34) and supports the biological significance of the pathogenicity model based on the bending of the rod-shaped structure of PSTVd (23, 27). The model that the genomic PSTVd RNA exists as a rod-shaped structure in vivo implies that many, if not all, of the predicted loop and bulge motifs present in this genomic structure (Fig. 1) likely have distinct biological functions yet to be elucidated. It should be noted that a metastable hairpin II structure that may play a role in replication has also been detected in vivo as well as in vitro (29). Altogether, these studies bring us one significant step towards elucidating the in vivo structures and functions of viroid RNA motifs.

UV cross-linking has been used for detecting tertiary interactions in vitro in a wide range of RNAs, for example, hepatitis delta virus (5, 7, 9), tRNAs (4), small nuclear RNAs (32), hairpin ribozymes (8), ribosomal RNAs (22), and Peach latent mosaic viroid (16). As shown here, this approach may be extended to investigate the in vivo existence of these tertiary interactions. As has also been well demonstrated by studies on cellular and viral RNAs (1, 2, 26, 36), probing the in vivo structure of RNA motifs will be necessary to further establish the biological significance of cellular and viral RNA motifs deduced from in vitro and in silico studies.