INTRODUCTION

The majority of eukaryotic mRNAs terminate in a poly(A) tail, which is required for efficient translation and message stabilization (reviewed in Refs. 1 and 2). As regulators of translation, the cap (m⁷GpppN) and the poly(A) tail are functionally codependent (3), suggesting that these two regulatory elements, in conjunction with associated proteins, communicate to direct efficient translation. Only a few mRNAs have been found that do not terminate in a poly(A) tail, including several plant viral mRNAs (reviewed in Refs. 4 and 5), metazoan cell cycle-regulated histone mRNAs (6), and flaviviral mRNAs (7). Tobacco mosaic virus (TMV)¹ is an RNA virus with genomic RNA that also serves as mRNA and with a life cycle that takes place in the cytoplasm. Subgenomic mRNAs generated during the viral life cycle are colinear with the genomic mRNA at the 3-terminus, and consequently, all TMV mRNAs contain the same 205-base 3 untranslated region (UTR). The TMV 3 UTR is the functional equivalent to a poly(A) tail in that it enhances the translational efficiency and increases the stability of reporter mRNAs and does so in the absence of the virus or any viral gene product (8, 9). The 3 UTR is composed of two domains. A 105-base tRNA-like domain, located at the 3 terminus, mimics the three-dimensional shape of true tRNAs to such an extent that many tRNA-specific enzymes will also specifically recognize and modify it (4, 5). Immediately upstream of the tRNA-like domain is a 72-base domain composed solely of three RNA pseudoknots (10). This upstream pseudoknot domain (UPD) is responsible for the regulation of translation associated with the TMV 3 UTR (8). The pseudoknots contain a phylogenetically conserved primary sequence and higher-ordered structure. Mutational analysis revealed that the phylogenetically conserved primary sequence within the third pseudoknot is essential for facilitating translation in vivo (11) (Fig. 1). Moreover, the structure of the second pseudoknot in the UPD was also needed, probably to increase the stability of the third pseudoknot through coaxial stacking of base pairs within the RNA duplex of the UPD.

Sequence and structure of the TMV 5 leader and 3 UTR showing the primary sequence required for regulating translation.

In addition to the UPD-mediated regulation, the 5 leader of TMV, called , substantially enhances the translational efficiency of mRNAs containing this sequence in both plant and animal species (9, 12, 13, 14, 15). Recently, the subsequence within  essential for function has been delineated to a 25-base poly(CAA) region (15) (Fig. 1). Like the synergy between a cap and poly(A) tail, the cap and the UPD are codependent as regulators of translation (11), suggesting that communication between the termini of TMV RNA is the basis for the efficient translation of this mRNA. Consequently, although TMV mRNA has evolved an alternative to a poly(A) tail, it has nevertheless maintained the requirement for an interaction between the termini of the mRNA. In addition to the functional codependence between the cap and the UPD, the TMV 5 leader and the UPD synergistically increase translation (11).

Previously, we have shown that both  and the UPD are specifically recognized by and compete for the same binding activity from carrot suspension cells and wheat germ (11). In this study, we report the purification and characterization of a 102-kDa protein (p102) from wheat that specifically binds to the poly(CAA) region within  and the 3-distal RNA pseudoknot of the UPD. This binding activity is conserved in all plant species tested. Furthermore, p102 polyclonal antibodies were used to demonstrate the conservation of this protein throughout the higher plant kingdom.

MATERIALS AND METHODS

Purification of p102 followed the procedure in Fig. 2. All steps were carried out at 4 °C, and the purification was followed by monitoring RNA binding activity using gel retardation. 50-g batches of wheat germ (Arrowhead Mills) were ground in a blender for 45 s, to which was added 75 ml of grinding buffer (5 mM, HEPES pH 6.9, 120 mM potassium acetate, 5 mM magnesium acetate, 1 mM dithiothreitol (DTT), 1 mM phenylmethylsulfonyl fluoride, and 1 µg/ml leupeptin and pepstatin). Following centrifugation at 30,000 × g at 4 °C, the supernatant was supplemented with 0.1 volume of 500 mM HEPES, pH 7.5, and centrifuged again for 15 min. 10 g of crude protein extract was brought to 40% ammonium sulfate saturation. The 3.6-g pellet was dissolved and dialyzed against Buffer B (20 mM Tris-HCl, pH 7.5, 50 mM KCl, 5 mM MgCl₂, 1 mM DTT, 0.5 mM EDTA, 1 mM phenylmethylsulfonyl fluoride, and 1 µg/ml leupeptin and pepstatin) and stored at 80 °C. 10-ml aliquots (400 mg) were dialyzed overnight against 20% glycerol to prepare for preparative isoelectric focusing. A 55-ml sample containing 2% (w/v) ampholytes (pH 6-8) and 20% glycerol was focused for 4 h at 12  in a Rotoford IEF Cell (Bio-Rad). Exceeding 400 mg of protein lead to excessive precipitation of binding activity. Fractions containing RNA binding activity (pH range, 7.21-7.35) were pooled and dialyzed against Buffer B and loaded onto a heparin-Sepharose column (Pharmacia). RNA-binding activity was step eluted from the column using Buffer B with 300 mM KCl and dialyzed overnight against the same buffer with 10% glycerol. The activity was subsequently loaded onto a 5/5 MonoQ column (Pharmacia) and eluted from the column at 0.25 M KCl using a linear gradient. The active fraction was dialyzed against 50 mM HEPES, pH 7.0, 20 mM KCl, 5 mM MgCl₂, 10% glycerol, 1 mM DTT, 0.5 mM EDTA, and 0.5 mM phenylmethylsulfonyl fluoride and loaded onto a 5/5 MonoS column. The RNA binding activity eluted at 0.17 M KCl using a linear gradient. This protein was used for the binding characterization.

The T7-based constructs containing either the TMV 3 UTR, the UPD, , or control sequences have been described previously (11). In vitro transcription was carried out as described previously (16) using 40 mM Tris-HCl, pH 7.5, 6 mM MgCl₂, 100 µg/ml bovine serum albumin, 0.5 mM each ATP, CTP, UTP, and GTP, 10 mM DTT, 0.3 units/µl RNasin (Promega), and 0.5 units/µl T7 RNA polymerase. Radiolabeled probes were made as uncapped RNAs using either [³²P]ATP or [³²P]CTP, and the full-length transcripts were resolved and electroeluted on 4% polyacrylamide gels. Quantitation of RNA yields was determined spectrophotometrically.

Antisense oligonucleotides (Cruachem Inc.), designed to anneal to regions within  and the UPD, were used to map the protein binding sites. Oligonucleotides annealing to pseudoknots (PKs) or the stem loop within the TMV 3 UTR were: anti-PK1, GTGATTACGGACACAATCC; anti-PK2, TGCGTTATCGTACGCACCAC; anti-PK3, CCTTCGATTTAAGTGGAGGGAAAAACAC; anti-PK4, GCGATCCAAGACA; anti-PK5, TGGGCCCCTACCGGGGGTAACGGGG; anti-stem loop, ATGTATATGAACCATATACATTTGAC; and anti-UPD; TTCGATTTAAGTGGAGGGAAAAACACTATGCGTTATCGTACGCACCACGTGTGATTACGGACACA. Antisense oligonucleotides annealing with  were: 1, GTAAAAATAC; 2, GTAATTGT; 3, TGTTGTTTGTTGTTTGTTGTTGTT; 4, GTAATTGTTGTAAAAATAC; 5, GTAATTGTAAATAGTAATTGTAA; and 6, TGTAATTGTAAAT. Sense oligonucleotides representing subsequences within  were: 7, GTATTTTT; 8, GTATTTTTACAACAATTAC; 9, ACAATTAC; and 10, CAACAACAACAAACAACAAACAACAT.

1 ng of radiolabeled RNA and the indicated amount of extract were used for the binding reactions in a 14-µl volume containing 10 mM Tris, pH 7.5, 35 mM KCl, 1.0 mM MgCl₂, 5% glycerol, 1 mM DTT, 0.7 mg/ml total yeast RNA, and 0.5 units/µl RNasin. Following 15 min of incubation at 4 °C, heparin was added to make 5 mg/ml, and the solution was incubated for an additional 10 min. The RNA-protein complexes were resolved on native 3.5% polyacrylamide gels, dried, and analyzed by autoradiography. For competition assays, the radiolabeled and unlabeled RNAs were mixed prior to addition of protein. To map the protein binding site on the RNA, excess antisense oligonucleotide was added to the binding reaction containing all components except the protein. The reaction was heated at 85 °C for 5 min, allowed to cool to room temperature to permit the RNA and DNA to anneal, and placed on ice, and purified p102 was added. Quantitation of the gel shift was determined using a phosphorimager (Molecular Dynamics).

200 µg of  RNA was synthesized with a 25-base polyadenylate track at the 3 terminus, purified as indicated above, and annealed to oligo(dT)-cellulose (Research Genetics). After equilibrating the column with Buffer B, 1 mg of protein enriched for the binding activity (MonoQ fraction) was applied to the column. The flow-through was then supplemented with 5 mg/ml heparin (to reduce nonspecific interactions), reapplied to the column, and washed with 3 volumes of Buffer B containing 5 mg/ml heparin. Following a 10-bed volume washed with Buffer B alone, the RNA binding activity was eluted with Buffer B containing 2 M KCl.

The antisera used for Western blotting were generated in rabbits (Robert Sargeant, Ramona, CA) and were either partially purified over DEAE-Affi-Gel Blue (Bio-Rad) or were affinity purified as described previously (17) with the following changes. 25 µg of highly purified p102 protein was separated by SDS-polyacrylamide gel electrophoresis (PAGE), electroblotted onto nitrocellulose, and stained with colloidal gold, and the portion of the membrane containing the p102 was excised and fixed with 4% formalin for 10 min. The p102-containing strip was washed in phosphate-buffered saline containing 0.1% Tween-20 (TPBS), followed by a 30-min block in TPBS containing 5% dry milk, and then incubated with anti-p102 antisera for 1.5 h. After several 5-min washes with TPBS, anti-p102 antibodies were eluted with 3 M potassium thiocyanate in PBS containing 0.5 mg/ml bovine serum albumin and dialyzed against PBS in 50% glycerol.

To prepare protein extracts, seeds were ground in liquid nitrogen, resuspended in Buffer B (containing 2 mM phenylmethylsulfonyl fluoride, 3 µg/ml leupeptin, and 3 µg/ml pepstatin), and centrifuged for 30 min at 30,000 × g. Protein concentration was determined as described (18). Equal amounts of protein from extracts were resolved on 8% SDS-PAGE and transferred to a nitrocellulose membrane, which was then blocked for 30 min in TPBS with 5% reconstituted dry milk and incubated with p102 antibodies diluted in TPBS with 1% milk for 1.5 h. The blots were then washed with TPBS and incubated with goat anti-rabbit horseradish peroxidase antibody (Southern Biotechnology) for 1 h, and the p102 signal was detected using chemilumensence (Amersham Corp).

100 ng of each luciferase mRNA was translated using wheat germ lysate as described by the manufacturer (Promega), except all amino acids were unlabeled and used at 80 µM. The extent of translation was measured by luciferase activity. 5 µl of wheat germ translation lysate was added to 100 µl of luciferase assay buffer [20 mM Tricine, pH 7.8, 1.07 mM (MgCO₃)₄Mg(OH)₂·5H₂O, 2.67 mM MgSO₄, 0.1 mM EDTA, 33.3 mM DTT, 270 µM coenzyme A, and 500 µM ATP (Promega)], and the reaction was initiated with the injection of 100 µl of 0.5 mM luciferin in luciferase assay buffer. Photons were counted using a Monolight 2010 luminometer (Analytical Luminescence Laboratory). Each mRNA construct was assayed in duplicate, and the average was reported.

RESULTS

We had previously identified specific binding activity from wheat and carrot to  and UPD RNA (11). Because  enhances translation in in vitro lysates derived from wheat germ, and both  and the UPD increase the translation of reporter mRNAs in both monocot and dicot species, we chose to purify the binding activity from wheat germ. Crude wheat germ extract was precipitated with 40% ammonium sulfate, separated by preparative isoelectric focusing, fractionated over a heparin-Sepharose column, and purified over a MonoQ anion exchange column (see ``Materials and Methods'' and Fig. 2, table). Following this purification scheme, the protein profile as determined by SDS-PAGE remained complex (Fig. 2, lane 5). Therefore, an RNA affinity column was used to identify the protein that binds specifically to  (19).  RNA was synthesized with an (A)<sub>25</sub> tail and annealed to oligo(dT)-cellulose (see ``Materials and Methods''). 1 mg of a highly active MonoQ fraction was applied over the column and extensively washed with heparin to remove nonspecific binding proteins, and the column was eluted with 2 M KCl. Each fraction from the affinity column was assayed for RNA binding activity using  RNA as the probe (Fig. 3). RNA binding activity was eluted with 2 M KCl (Fig. 3, lane 7). Antibodies raised against the most active fraction following the MonoQ column in the above purification scheme were used in Western analysis to detect the proteins present in each fraction from the affinity column (Fig. 3B). The applied sample contained multiple proteins (Fig. 3B, lane 1). Most of these were found in the flow-through or were removed with subsequent washes (Fig. 3, lanes 2-6). Only a single protein of 102 kDa in size was detected in the fraction eluted with 2 M KCl (Fig. 3B, lane 7). The activity obtained from the MonoQ column was further purified using a MonoS cation exchange column, which resulted in the purification of p102 to near homogeneity. Throughout the purification, the presence of p102, as determined by Western analysis, correlated with RNA binding activity. From these observations, we conclude that p102 is the protein responsible for the specific binding to .

We have previously demonstrated that  and the UPD of TMV enhanced translational efficiency in a wide array of species, including monocots and dicots (8, 9, 12, 13, 15), observations suggesting that the protein that mediates the regulation associated with these translational enhancers is present in these species. To determine the extent of the conservation of p102, extracts were prepared from plant species representing several subclasses of flowering plants. The presence and activity of p102 in the seed was examined using Western and gel shift analyses, respectively. Activity that specifically bound to  RNA but not to control RNA was detected in nearly all plant extracts (Fig. 4A). A 58-base RNA of random nucleotide sequence was used as a control and failed to form any complex with the extracts in the gel shift assay (data not shown). Extracts prepared from carrot seed contained pigments that interfere with electrophoresis (Fig. 4A, lane 14). However, we have previously reported specific binding activity to  and UPD RNA using carrot cell culture cells (11). Extracts prepared from poppy seed have high levels of RNase activity, which accounts for the failure to detect a complex with extracts from this species (Fig. 4, lane 15). Western analysis using affinity-purified, anti-wheat p102 antibodies was performed to determine whether p102 was present in the extracts (Fig. 4B). A single protein comigrating with purified wheat p102 was detected in all extracts (Fig. 4B, lanes 2-13). The presence of this protein and gel shift activity in diverse plant species suggests that p102 is highly conserved in size, antigenicity, and RNA binding specificity throughout flowering plants.

, UPD, and control RNAs were used to evaluate the specificity of p102 binding in competitive gel shift assays (Fig. 5). Highly purified p102 protein from wheat germ was used in the binding reactions. The affinity of p102 for  is higher than it is for UPD RNA. Using unlabeled UPD RNA as a competitor RNA, a 20-fold molar excess was required for UPD RNA to begin to compete with labeled  RNA (Fig. 5B, lane 10), whereas only a 10-fold molar excess of UPD RNA was needed to effectively compete with labeled UPD RNA (Fig. 5A, lane 4). In contrast, unlabeled  RNA efficiently competes with the UPD RNA at an equal molar ratio (Fig. 5A, lane 6) and competes with labeled  RNA from equal molar to a 20-fold molar excess (Fig. 5B lane 3-6). Unlabeled control RNA does not compete with either UPD RNA (Fig. 5A, lanes 9 and 10) or  RNA (Fig. 5B, lanes 13-17) even at a 50-fold molar excess.

Analysis of p102 binding specificity to UPD and  RNA.

Although maximal p102 binding to  and the UPD was observed when the binding was performed at pH 7.5, p102 binding activity was not highly sensitive to pH. Even at pH 6.0 and 9.5, 66 and 85% binding activity to , and 80 and 82% binding activity to UPD remained, respectively (data not shown). p102 binding activity was not dependent on Mg²⁺ and was stable over a wide range of salt concentrations. Optimal binding occurred at 20 mM KCl for  and 80 mM KCl for UPD RNA, but even at 1.6 M KCl, 32 and 28% binding remained, respectively (data not shown). Heat treatment of purified p102 revealed that the binding activity to either  or UPD RNA was irreversibly abolished after heating the protein at 65 °C for 5 min and was significantly reduced after a 5-min treatment at 45 °C (data not shown). 60% of binding activity remained following a 5-min treatment at 37 °C.

Heparin is a polyanion used to eliminate nonspecific RNA-protein interactions. p102 binding was examined over a range of heparin concentrations (Fig. 6). With a low concentration of heparin in the binding reactions, it was not possible to resolve the RNA-protein complexes by native PAGE (Fig. 6). With , the specific complex remained unchanged from 1000 to 8.5 µg/ml, but as the heparin concentration was reduced below 1.7 µg/ml, there was an increasing amount of a greatly retarded band that did not even enter the gel (Fig. 6A, lanes 7 and 8). Similar results were obtained with the UPD RNA, although, at 40 and 8.5 µg/ml, several additional complexes were observed (Fig. 6B, lanes 4 and 5). These results underscore the importance of using heparin to eliminate nonspecific RNA-protein interactions.

Heparin can influence the specificity of p102 binding to  and UPD RNA.

Mapping the p102 Binding Site within  and UPD RNA

To identify the p102 binding site within  and the UPD, an antisense deoxyoligonucleotide mapping assay was developed. Radiolabeled RNA was heat denatured in the presence of excess unlabeled antisense oligonucleotides corresponding to distinct regions within the RNA and allowed to anneal by slow cooling. An antisense oligonucleotide will inhibit RNA-protein complex formation and therefore inhibit the appearance of a gel shift if the oligonucleotide anneals to the protein binding site. Highly purified p102 protein was used for the mapping studies.  has little potential for secondary structure but has a highly ordered primary sequence containing a 25-base poly(CAA) element and three 8-base direct repeats (Fig. 1). Antisense oligonucleotides to the direct repeats had no impact on the RNA binding (Fig. 7A, lane 7), whereas the antisense oligonucleotide to the poly(CAA) region p102 binding abolished p102 binding to  (Fig. 7A, lane 9). Oligonucleotide annealing to the RNA was confirmed by following a shift in the migration of the radiolabeled oligonucleotides on native PAGE gels following annealing to unlabeled target RNA (data not shown). The migration of the free  RNA was unaltered by the annealing of the oligonucleotides as a 3.5% gel was used, which provides good resolution for protein-RNA complexes but does not resolve well small changes in RNA length or structure. No other antisense oligonucleotide could inhibit p102 complex formation, suggesting that the poly(CAA) sequence alone within  constitutes the p102 binding site. p102 binding to  is specific to the sequence as RNA as unlabeled  RNA could effectively compete with the labeled  RNA for p102 (Fig. 7B, lane 5), whereas when a full-length, sense deoxyoligonucleotide representing the  sequence was used as a competitor, it failed to compete for p102 binding (Fig. 7B).

Mapping the p102 binding site within .

Antisense oligonucleotides were also used to map the p102 binding site in the TMV 3 UTR. The entire length of the TMV 3 UTR sequence could be probed for p102 binding using antisense oligonucleotides individually or in combination to pseudoknot (PK) 1, PK2, and PK3 within the UPD or PK4, PK5, and the stem loop within the tRNA-like structure (Fig. 8). In addition, an antisense oligonucleotide to the entire UPD was used as a probe. Antisense oligonucleotides annealing to PK5 or the stem loop structure within the tRNA-like structure did not reduce complex formation relative to the control, in which no antisense oligonucleotides were present (Fig. 8), suggesting that the tRNA-like structure is not required for p102 binding. Annealing the oligonucleotide complementary to PK1 reduced complex formation to 73% of the control, whereas those annealing to PK2, PK3, or PK4 reduced complex formation to 45.3, 39.9, and 44.8%, respectively, of the control. This compared with a reduction to 55.3% of the control when the full-length antisense oligonucleotide to the entire UPD was used or a reduction to 24.5% of the control when a combination of all the antisense oligonucleotides was used. A second complex that migrates just above the free RNA was observed only when protein was present in the binding reaction. As this complex was not affected by the antisense oligonucleotides, it may represent a nonspecific interaction. These data suggest that p102 interacts preferentially with the region containing PK2, PK3, and PK4.

If p102 is involved in mediating efficient translation from TMV mRNA, it may have a general function in the translation of cellular mRNAs as well. As a first step to examine whether p102 is involved in translation, we added purified p102 to an in vitro translation lysate derived from wheat germ prior to the translation of luciferase (luc) mRNA. Supplementation with purified p102 did not increase the translation of luc mRNA with or without  as the 5 leader (data not shown). This is not entirely surprising, because p102 is relatively abundant in wheat germ, the tissue from which the p102 was originally purified. We therefore added anti-p102 antibodies to wheat germ lysate to examine whether inactivating the endogenous p102 would affect the translation of luc mRNA in vitro. Translation was increasingly inhibited as the concentration of affinity-purified p102 antibodies increased (Fig. 9). Preimmune or affinity-purified antibodies raised against a second, but unrelated, RNA-binding protein of 100 kDa in size² did not substantially affect translation (Fig. 9). Translation from luc mRNA with or without  as the 5 leader was inhibited by the addition of p102 antibodies, suggesting that p102 may be required for the efficient translation of mRNAs in vitro regardless of whether  is present or not. Addition of purified p102 to lysate containing p102 antibodies reversed the inhibition (Fig. 9), a finding that suggests a role for p102 during translation.

DISCUSSION

TMV viral RNA has evolved to compete efficiently for the translational machinery of its host to synthesize the virally encoded, RNA-dependent RNA polymerase required for viral replication. The 5 leader () and the 3 UTR are responsible for enhancing the translation of this mRNA, and both function independent of any other viral gene product or sequence (8, 9, 11, 12, 13, 15). We have characterized a single 102-kDa RNA-binding protein that binds to those sequences within  and the UPD that are responsible for the translational enhancement (Fig. 1), suggesting a possible role for p102 in translational control. The addition of affinity-purified antibodies raised against p102 inhibited translation in vitro, which could be subsequently reversed by the addition of purified p102. These findings support the idea that p102 may be necessary for efficient translation. Although p102 was initially purified from wheat, p102 antibodies were used to demonstrate that the protein is conserved both antigenetically and in molecular weight throughout the plant kingdom. Moreover, functional analysis using gel shift assays suggests that its specific RNA binding activity has also been conserved throughout the evolution of higher plants. It is possible, therefore, that p102 plays a role in the translation of plant mRNAs that has been conserved throughout plant species, and that TMV has evolved to efficiently compete for this protein on entry into the host.

TMV RNA represents an exceptional mRNA in that it contains a translational enhancer in the 5 leader, and rather than terminating in a poly(A) tail, it contains an unusual 3 UTR composed of RNA pseudoknots. Although regions of poly(CAA) are found in the 5 leaders of some plant mRNAs, the regulatory pseudoknots present in the UPD have not been observed in the 3 UTR of plant mRNAs. This raises two intriguing questions regarding the function of p102. Given that p102 binds both  and the UPD, what is the mechanism by which p102 mediates the regulation associated with these RNA elements? Second, what is the cellular function of p102?

One possibility we investigated was that p102 was one of the known translation initiation factors that have been well characterized in wheat and include several RNA-binding proteins (20). None of the purified initiation factors bound , nor did antibodies raised against each factor recognize p102 in Western analysis (21),³ suggesting that p102 is not one of the known initiation factors. Poly(A)-binding protein purified from wheat failed to bind either  (composed of 50% adenosine) or the UPD, whereas it did bind (A)₅₀.⁴ Moreover, as p102 does not bind poly(A), and p102 and poly(A)-binding protein are immunologically unrelated,³ we conclude that poly(A)-binding protein and p102 have distinct sequence specificities. The functional codependence between the cap and the poly(A) tail during translation suggests communication between the termini that may involve both poly(A)-binding protein and the cap-associated initiation factors as a means to commit the translational machinery to an mRNA (21). As the cap and the TMV UPD are also functionally codependent during translation, it is possible that p102 interacts with one or more initiation factors as a means to compete efficiently for the translational machinery to the mRNA.

As an RNA-binding protein, what might be the cellular target of p102? It is not likely that plants have evolved a mechanism that only facilitates the translation of TMV mRNA. Rather, TMV mRNA has most likely evolved to mimic p102 binding sites of cellular mRNAs to compete for p102. Such mimicry has been observed with the 3 terminal, tRNA-like structure of TMV, and other viral mRNAs that mimic true tRNAs to functionally interact with tRNA-modifying enzymes. One consequence of this tRNA mimicry is that cellular tRNA nucleotidyltransferase maintains the integrity of the viral 3 terminus by replacing nucleotides lost through 3  5 exonuclease activity (22). The observation that p102 binds to two unrelated sequences within TMV, i.e. the poly(CAA) region within  and the distal two RNA pseudoknots within the UPD, suggests that p102, although exhibiting some sequence specificity, is not limited to a single RNA recognition element. Identification of p102 binding sites within cellular mRNAs will be an important step in elucidating the cellular function of p102.