Nucleic Acids Research

RNA numbering system and free energy minimization

The numbering system used for IRNA differs from that used in previous publications (21,23) because the vector sequences that contribute to the full-length RNA molecule are considered here for structural purposes. Previously, IRNA was described as being 60 nt long. Here, IRNA derived from the pGEM3Z vector has 10 additional vector-derived nucleotides at the 5[prime]-end and one additional vector-derived nucleotide at the 3[prime]-end for a total of 71 nt. IRNA or mutants of IRNA derived from the pCDNA3 vector are 73 nt long. cIRNA, derived by transcription in the opposite direction to IRNA from the pGEM3Z vector, is 75 nt in length. Secondary structures of the RNA molecules were predicted by free energy minimization analysis using the RNA folding program, MFOLD (24). For each RNA, all structures within 40% of the minimum predicted free energy were retained as possible candidates; these candidates were then compared with chemical and enzymatic data to generate the most likely structures.

Plasmid construction and in vitro transcription

The pcMut1IRribo and pcMut2IRribo clones consist of the mut1IRNA or mut2IRNA sequences followed by the hepatitis delta virus ribozyme sequence (22) inserted into the pcDNA3 1.1 vector (Stratagene). The mut1IRNA and mut2IRNA oligonucleotides were generated by synthesizing a plus-strand oligonucleotide with a HindIII overhang at the 5[prime]-end and a negative strand oligonucleotide containing an EcoRI overhang at its 5[prime]-end (IDT Inc.). The two strands were annealed, then placed in a tripartite ligation reaction with pcDNA3 (cut with HindIII and XhoI) and the ribozyme (with 5[prime] EcoRI and 3[prime] XhoI overhangs). The various mRNAs were transcribed in vitro from gel-purified, linearized plasmids with the appropriate (T7 or SP6) RNA polymerase as previously described (21).

Nuclease probing and chemical modification

Chemical modification by DMS was performed essentially as described elsewhere (25). Nuclease probing was performed as follows. RNAs were equilibrated by heating to 65°C for 5 min and slow cooling to room temperature. Digestion with RNase V1 (Pharmacia) (in 20 mM Tris-HCl, pH 7.2, 200 mM NaCl and 10 mM MgCl₂) and RNase T1 (Boehringer Mannheim) (in 30 mM Tris-HCl, pH 7.8, 20 mM MgCl₂, 300 mM KCl) were done for 30 min on ice. Digestion with nuclease S1 (Promega) (in 50 mM sodium acetate, pH 4.5, 280 mM NaCl and 4.5 mM ZnSO₄) was performed at 37°C for 10 min after equilibrating the RNA with buffer only for 10 min at 37°C (26-29).

Oligonucleotide hybridization followed by RNase H digestion

Two micrograms of in vitro transcribed RNA was added to RNase H buffer (40 mM Tris-HCl, pH 7.9, 4 mM MgCl₂, 1 mM DTT, 30 mg/ml BSA). In each reaction mixture, the appropriate complementary oligonucleotide was added, followed by incubation at 55°C for 3 min and equilibration at 32°C for 30 min. An aliquot of 1.5 U of RNase H (Pharmacia) was then added to each reaction and cleavage was perfomed for 30 min at 32°C (31).

Primer extension of digested RNAs

Primer extension was performed essentially as described elsewhere (26). For position markers, sequencing ladders were generated from plasmid DNA using the same primers and a Sequenase v.2.0 kit (US Biochemicals).

3[prime]-End labeling of RNA

RNAs were labeled at 3[prime]-termini using T4 RNA ligase and [5[prime]-³²P]pCp. Partial alkaline hydrolysis and denaturing T1 ladders were generated as described elsewhere (30).

In vitro translation and UV-induced cross-linking

HeLa cell extracts were prepared as previously described (21). In vitro translation of monocistronic constructs in HeLa cell extracts and UV-induced crosslinking were performed essentially as described elsewhere (21). In vitro translation of the bicistronic constructs was performed essentially as described previously (22). Non-specific RNA used in competitive UV crosslinking experiments is derived from the polylinker of pspLUC as previously described (22).

Both IRNA and cIRNA inhibit IRES-mediated translation in vitro

Previous studies from our laboratory have shown that IRNA specifically inhibits IRES-mediated translation (21). We speculated that IRNA could exert its inhibitory effect by two possible mechanisms. IRNA could function as an antisense RNA and bind to complementary sequences in IRES elements or it could bind to and compete for certain protein factors that are necessary for IRES-mediated translation. Previous results had established that IRNA binds cellular proteins similar to those bound by the poliovirus 5[prime]-UTR, suggesting the likelihood of the latter possibility being correct (21,23). In addition, the IRNA sequence is neither homologous nor complementary to the viral 5[prime]-UTR and was unable to hybridize with poliovirus 5[prime]-UTR (data not shown).

To further confirm that IRNA's actions are not mediated through an antisense effect, we studied the effect of complimentary IRNA (cIRNA) on IRES-mediated translation. Surprisingly, as shown in Figure 1A, cIRNA possesses similar inhibitory activity to IRNA in an in vitro translation assay using a bicistronic template consisting of poliovirus 5[prime]-UTR flanked by the CAT and luciferase reporters (compare lanes 4 and 5 with lanes 2 and 3). In this assay, CAT is synthesized by cap-dependent translation, while luciferase synthesis is internally initiated by the PV IRES element. Poliovirus IRES-mediated translation of luciferase is substantially inhibited in lanes 2-5 as compared with the control in lane 1, while cap-dependent translation of the CAT gene is reduced only slightly in lanes 2 and 3. We have previously observed that IRNA does slightly inhibit cap-dependent translation, thus accounting for the drop in CAT levels (lanes 2 and 3). Quantitation of luciferase bands, which were normalized with respect to CAT bands in order to reflect degree of inhibition of internally initiated translation as compared with inhibition of capped mRNA translation, shows that while IRNA, at the highest concentration, specifically inhibits IRES-mediated translation by 80% of the control (compare lanes 1 and 3), specific inhibition by cIRNA ranged from 60 to 65% of the control (compare lanes 1 and 5). Similar results were obtained whether the bicistronic translations were performed in rabbit reticulocyte lysate supplemented with HeLa S10 extract (Fig. 1A) or in HeLa S10 extract alone (data not shown). In addition, an in vitro translation assay was performed in HeLa S10 extract using a monocistronic translational template, p2CAT, which consists of the poliovirus 5[prime]-UTR preceding the CAT reporter gene. Translational inhibition by IRNA, depicted in quantitated form in Figure 1B, is comparable with that seen in Figure 1A. cIRNA also inhibited CAT translation significantly; however, the degree of translation inhibition by cIRNA in Figure 1B was lower than that seen in Figure 1A. This is due to the fact that different preparations of cIRNA, with differing specific inhibitory activities, were used in the two assays. The same concentrations of IRNA or cIRNA do not significantly inhibit in vitro translation from pG3CAT, a capped monocistronic RNA (21; data not shown).

A    B

Figure 1. (A) Effects of IRNA and cIRNA on in vitro translation of a bicistronic construct. A bicistronic construct containing CAT and luciferase genes flanked by the PV type 2 5[prime]-UTR was translated in vitro in the absence (lane 1) or presence of 1 (lane 2) or 2 µg IRNA (lane 3) or 1 (lane 4) or 2 µg cIRNA (lane 5). Products were analyzed on a SDS-14% polyacrylamide gel. Migration of a molecular weight marker, in kDa, is shown on the left and the position of the CAT and luciferase (LUC) gene products are shown on the right. Quantitation of Luc and CAT bands is indicated below each lane, as is the Luc/CAT ratio. (B) Effects of IRNA and cIRNA on internal initiation of translation in vitro. A monocistronic construct consisting of the CAT gene preceded by the PV type 2 5[prime]-UTR was translated in the presence of varying concentrations of IRNA, cIRNA or a non-specific RNA (yeast tRNA). Products were analyzed on a SDS-14% polyacrylamide gel and intensities of CAT bands were quantitated. The percentage of CAT translation with respect to control (no inhibitory RNA added) was plotted against concentration of inhibitory RNA.

IRNA and cIRNA bind many of the same cellular proteins

UV crosslinking experiments were performed to determine whether cIRNA and IRNA bind similar proteins. Uniformly [³²P]UTP-labeled IRNA, cIRNA or poliovirus 5[prime]-UTR (which contains the PV IRES element) was first incubated with HeLa S10 extract and then crosslinked by UV irradiation. The resulting protein-nucleotide complexes were subjected to RNase treatment and analyzed by SDS-PAGE. Previous results have indicated that IRNA binds a number of polypeptides, including p100, p70, p57, p52 and p38, that are also bound by the poliovirus 5[prime]-UTR (21). Some of these proteins, including p57 (PTB) and p52 (La), have been found to be important for translation. Experiments in which unlabeled poliovirus 5[prime]-UTR was used as a competitor for binding have shown that the binding of many of these proteins to IRNA is specific and can be competed out by 5[prime]-UTR but not by a non-specific RNA (21). In Figure 2A, the protein binding profile of IRNA and cIRNA can be compared (IRNA, lanes 6 and 9; cIRNA, lanes 5 and 8). Both IRNA and cIRNA bind proteins of similar sizes, such as p110, p97, p70, p57 (binding of p57 by IRNA is apparent upon overexposure; data not shown), p52 and p38, although the intensities of the UV crosslinked bands differ. The major differences in protein binding between IRNA and cIRNA were that cIRNA bound p57 more strongly than IRNA while binding p110 more weakly than IRNA. Binding of poliovirus 5[prime]-UTR to cellular polypeptides is shown in lanes 4 and 7. Both IRNA and cIRNA bind many proteins with identical molecular masses to those bound by the viral 5[prime]-UTR.

A    B

Figure 2. (A) IRNA and cIRNA exhibit a similar binding profile. ³²P-labeled IRNA (lanes 3, 6 and 9), cIRNA (lanes 2, 5 and 8) and PV 5[prime]-UTR RNA (lanes 1, 4 and 7) were UV crosslinked to cellular polypeptides, using 0 (lanes 1-3), 30 (lanes 4-6) or 60 µg (lanes 7-9) of HeLa S10. Numbers to the left correspond to the migration in kDa of marker proteins. Numbers to the right correspond to the approximate molecular masses in kDa of the polypeptides indicated. (B) Competitive UV crosslinking. The triangles at the top represent the molar excess (100- or 250-fold) of each unlabeled RNA over ³²P-labeled PV 5[prime]-UTR RNA. NS is a non-specific competitor from the pspLuc+ polylinker (22). Lane N, no competitor added.

To demonstrate that the many similar sized proteins bound by IRNA, cIRNA and PV 5[prime]-UTR RNA are in fact the same proteins, competitive UV crosslinking experiments were performed. In Figure 2B, the results of such an experiment are shown. HeLa S10 extract was preincubated with unlabeled PV 5[prime]-UTR RNA (lanes 8 and 9), IRNA (lanes 6 and 7), cIRNA (lanes 4 and 5) or a non-specific RNA of similar size to IRNA (lanes 2 and 3) before addition of uniformly [[alpha]-³²P]UTP-labeled poliovirus 5[prime]-UTR followed by UV irradiation. Previously, we have shown that both unlabeled PV 5[prime]-UTR and IRNA, but not a non-specific RNA, can almost completely compete out binding of p52 (La) to labeled PV 5[prime]-UTR (21). Here, we have used more HeLa S10 protein (and, consequently, more La) in the UV crosslinking reaction to allow us to monitor binding and competition of fainter crosslinked bands. Under these conditions, only ~65% of La bound to [[alpha]-³²P]UTP-labeled PV 5[prime]-UTR was competed out by IRNA (Fig. 2B, lane 7) and 50% of bound La was competed out by unlabeled PV 5[prime]-UTR (Fig. 2B, lane 9). As is apparent from Figure 2B, both IRNA and cIRNA significantly compete for the binding of most PV 5[prime]-UTR-bound cellular proteins, including p97, p70, p57, p52 and p38. In addition, some of the previously mentioned differences in band intensities between IRNA and cIRNA binding found in Figure 2A are also reflected in Figure 2B; IRNA competes out p110 binding more strongly than does cIRNA, while cIRNA competes out p57 binding more strongly than does IRNA (compare Fig. 2B, lanes 5 and 7). Thus, competitive UV crosslinking experiments establish that both IRNA and cIRNA specifically bind many of the same proteins that are bound by the poliovirus 5[prime]-UTR RNA.

Elucidation of the secondary structures of IRNA and cIRNA

Computer modeling of the secondary structure of IRNA and cIRNA using the free energy minimizing Zuker MFOLD algorithm indicated that the structures of these two molecules may be similar, suggesting a structural basis behind the similarities of their inhibitory actions and protein binding profiles (24). To establish the secondary structure of IRNA, we applied two different enzymatic approaches. In the first approach, we digested in vitro transcribed IRNA with RNase T1, nuclease S1 and RNase ONE to identify single-stranded regions and RNase V1 to identify double-stranded regions (26-29). Our second enzymatic approach involved oligonucleotide hybridization with IRNA followed by RNase H cleavage (31). A third approach, involving chemical modification of single-stranded adenosines and cytosines by DMS, was also used to probe the structure of IRNA (25). In all of these approaches, sites of cleavage or modification were identified by primer extension with reverse transcriptase using a radiolabeled oligonucleotide primer complementary to the 3[prime]-end of the RNA, followed by analysis of the resulting cDNA on an 8 M urea-12% polyacrylamide gel. Alongside these and all subsequent primer extension reactions described, a sequencing ladder was run to determine the exact nucleotide positions of extended products.

The results of nuclease S1, T1 and V1 cleavage of IRNA are shown in Figure 3A. When these nuclease probing data were fitted to alternative MFOLD-predicted secondary structures, the model shown in Figure 4A emerged as the most likely secondary structure of IRNA. Results from RNase ONE cleavage and DMS modification interference experiments are consistent with this proposed structure (Fig. 4A and data not shown). Upon inspection, the major structural features of this molecule are a stem comprised of nt 20-27 base paired with nt 62-69, two loops between nt 6 and 15 and 36 and 43 and a large bulge between nt 46 and 61. Consistent with this model is the finding that the strongest V1 hits occur between nt 20 and 26, an area predicted to be a stem region. Significantly, no bases in this stem were cleaved by T1 and only one base in the stem, the guanosine at position 21, was weakly cleaved by S1. In addition, neither of the cytosines in the stem, at positions 20 and 27, are modified by DMS (Fig. 4A). The presence of one of the main predicted loops of IRNA, between positions 36 and 43, and the bulge between positions 46 and 61 were also substantiated by the nuclease probing data presented in Figure 3A. S1 and/or T1 hits occurred at seven of the eight bases in the 36-43 loop and at seven of the nine bases analyzed in the putative 46-61 bulge. Two bases within these two regions, the adenosine at position 41 and the guanosine at position 48, were strongly cleaved by the double-strand-specific nuclease V1. However, both of these bases were also strongly cleaved by S1 and the guanosine was also strongly cleaved by T1. The ability of V1 nuclease to cleave the phosphate backbone of a stacked base independent of the base's involvement in a Watson-Crick pairing may explain the cleavage of these two bases by V1 (32,33). DMS modification confirmed the nuclease digestion findings that nucleotides of IRNA between positions 36 and 43 and 46 and 61 are in single-stranded regions, as substantial modifications of cytosines and adenosines in both the loop and the bulge were observed upon addition of DMS. Taken together, the enzymatic and chemical cleavage data is consistent with the presence of a loop between nt 36 and 43 and a bulge between nt 46 and 61. Predicted loop 6-15 was more difficult to analyze by nuclease probing. Two bases within the loop were modified weakly by V1 and none were modified by either T1 or S1. DMS modifications were observed at three bases within this putative loop (Fig. 4A). We believe that this loop is less reactive to nuclease cleavage than the other single-stranded regions within the molecule.

Figure 3. (A) Structural analysis of in vitro synthesized IRNA by nuclease probing or chemical modfication followed by primer extension. IRNA was treated with nucleases or with DMS as described in Materials and Methods, annealed to a ³²P-5[prime]-end-labeled oligonucleotide complementary to the terminal 16 nt of IRNA and then transcribed with reverse transcriptase. Elongation products were separated on a urea-containing polyacrylamide gel. The nucleotide sequence of the RNA was deduced by dideoxy sequencing; representative lanes are shown on the left, marked C and U. Nucleotide positions, as determined by dideoxy sequencing, are shown on the right. Triangles at the top represent increasing amounts of each nuclease. Lane P, reverse transcription primer only. Positions of free primer and full-length reverse transcribed IRNA are indicated by arrowheads on the right. (B) Structural analysis of in vitro synthesized IRNA by oligonucleotide hybridization and RNase H digestion followed by primer extension. IRNA was annealed to various complementary oligonucleotides and treated with RNase H as described in Materials and Methods and was reverse transcribed and analyzed as described in Figure 2A. Nucleotide positions as deduced by dideoxy sequencing are shown on the right, as are positions of free primer and full-length reverse transcribed IRNA. Above each lane are shown the nucleotide positions of IRNA that the oligonucleotide in that reaction is complementary to; in lane 1, which serves as a control, no oligonucleotide has been added. Lane P, reverse transcription primer only.

Figure 4. (A) Proposed secondary structure model of IRNA with an enzymatic digestion map. Triangles represent nucleotides reactive to nuclease S1, squares represent nucleotides reacive to RNase T1 and circles mark nucleotides reactive to RNase V1. Solid symbols represent strong reactivities and open symbols represent weak reactivities. D marks bases reactive to DMS and R marks bases reactive to RNase ONE. The arrow marks the 3[prime]-most nucleotide analyzed by S1, T1 and V1 digestion, due to annealing of the primer for reverse transcription. (B) Proposed secondary structure model of cIRNA with an enzymatic digestion map. Symbols are explained in the legend to (A). Open arrows mark strong reverse transcription pause sites. (C) Proposed secondary structure model of mut1IRNA as predicted by MFOLD free energy minimization and confirmed by RNase H digestion. (D) Proposed secondary structure model of mut2IRNA as predicted by MFOLD free energy minimization and confirmed by nuclease S1, T1, V1 and RNase H digestion. Symbols are explained in the legend to (A).

The secondary structure of IRNA depicted in Figure 4A was confirmed by RNase H probing (Fig. 3B), a technique in which complementary oligonucleotides are annealed to various portions of the RNA molecule. Single-stranded areas are free to anneal to these oligonucleotides, whereas double-stranded RNA regions are not capable of forming additional hydrogen bonds. This step is followed by treatment with RNase H, an enzyme that specifically cleaves DNA-RNA hybrids comprised of, in general, [ge]4 bp. As expected, putative loop areas within IRNA were cleaved upon addition of the appropriate complementary oligonucleotide. For example, in lane 5, the loop between nt 36 and 43 was free to anneal to an oligonucleotide complementary to positions 35-43; as a result, cleavage by RNase H at positions 42 and 43 is observed. Cleavage of IRNA depicted in lanes 6 and 7 upon addition of oligonucleotides complementary to positions 44-53 and 54-60 confirms the S1, T1 and V1 data that suggest that nt 44-61 are present in a single-stranded form. In addition, cleavage in lane 2 suggests the presence of a loop area 5[prime] of the 20-27 stem; an alternate oligonucleotide, complementary to positions 8-15, also induced cleavage of IRNA upon addition of RNase H (data not shown), again consistent with the Zuker MFOLD prediction of a loop between positions 6 and 15. The presence of the 20-27/62-69 stem was also detected here. Very little cleavage of IRNA occurs in lanes 3 and 8, where oligonucleotides complementary to predicted helical regions 20-27 and 62-68, respectively, have been added prior to RNase H digestion. When an alternate set of oligonucleotides that overlaps with the set shown in Figure 3B was annealed to IRNA, followed by digestion with RNase H, the resulting cleavage patterns further supported the secondary structure model of IRNA shown in Figure 4A (data not shown).

cIRNA was also investigated through nuclease S1, T1, V1, RNase ONE and RNase H probing in order to determine its secondary structure. The results of the RNase H experiments performed on cIRNA are shown in Figure 5 and a best fit Zuker MFOLD model incorporating both the RNase H data and the schematically depicted RNase ONE, S1, T1 and V1 cleavages is shown in Figure 4B. The main area of V1 cleavage occurs between nt 42 and 48, which is predicted to be a stem with a 2 nt bulge. Since RNase V1 is known to sometimes cleave bases of a single-stranded region that are adjacent to a double-stranded region (32), cleavage of the 2 nt that comprise the bulge (positions 44 and 45) is not inconsistent with the predicted structure. Also consistent with the proposed structure of cIRNA is the almost total lack of T1 cleavage throughout the entire molecule, given the scarcity of unpaired guanosines. In addition, RNase ONE cleavage data suggest that the two areas indicated as `loops' in Figure 4B are, indeed, single stranded. The RNase H data shown in Figure 5 are also consistent with the secondary structure model of cIRNA depicted in Figure 4B. The two main loops of cIRNA, between positions 20 and 28 and 49 and 57, are cleaved upon addition of the corresponding complementary oligonucleotides (Fig. 5, lanes 3 and 7, and Fig. 4B). As was previously observed with double-stranded regions of IRNA, addition of oligonucleotides to the main stem of cIRNA (comprised of nt 39-48 and 58-69) did not result in a significant degree of cleavage (lanes 5, 6, 8 and 9). In lane 5, strong bands are observed at positions 42 and 44. However, the presence of a highly intact full-length cIRNA band in lane 5, similar to that observed in the no oligonucleotide control (lane N), and the fact that addition of the same oligonucleotide to uniformly labeled cIRNA followed by RNase H digestion did not yield any cleavage products (data not shown) indicates that no significant cleavage of cIRNA occurs in lane 5. We instead believe that the normally weak reverse transcription pause sites at positions 42 and 44 become strong stops in lane 5, presumably due to bypass of the normally strong pause site observed at position 52 (compare lanes N and 5). Thus, the cleavage data represented in lanes 5-9 support the free energy minimization prediction of cIRNA structure between nt 39 and 69. In comparing the structure of cIRNA with IRNA, a structural motif that appears similar is the stem-bulge-loop configuration present between nt 20 and 69 of IRNA and nt 39 and 69 of cIRNA. To further confirm that cIRNA does indeed possess such a stem-bulge-loop structure, alternate oligonucleotides were used to demonstrate the existence of this structure between positions 39 and 69. In the inset of Figure 5, results of RNase H cleavage from a set of 10 overlapping oligonucleotides complementary to the cIRNA region spanning nt 37-66 are presented. These results further demonstrate the existence of a stem-bulge-loop configuration in this region of cIRNA.

Figure 5. Structural analysis of in vitro synthesized cIRNA by oligonucleotide hybridization and RNase H digestion followed by primer extension. (Inset) Summarized results of additional oligonucleotide hybridization/RNase H cleavage experiments. In addition to the oligonucleotides used in the experiments depicted in Figure 5, oligonucleotides complementary to the positions indicated in the inset were used to probe the structure of cIRNA. No significant cleavage is indicated by (-) and significant cleavage is indicated by (+).

Upon comparison of the proposed structures of IRNA and cIRNA (Fig. 4A and B), it appears that although the nucleotides that form the particular stems and loops of IRNA do not correspond to the nucleotides of cIRNA that form its secondary structure elements, similarities exist between the overall structures of the two molecules. Both molecules possess stems, loops and bulges that are oriented with respect to each other in similar ways. The main structural difference between IRNA and cIRNA is the substitution of a small bulge in cIRNA (nt 61-64) for a large bulge in IRNA (nt 45-61); this difference appears to account for the lesser predicted thermodynamic stability of IRNA (-12.6 kcal/mol) as compared with cIRNA (-13.8 kcal/mol) (24).

Site-directed mutagenesis of IRNA disrupts secondary structure and affects translation inhibitory activity

In order to create a disruption in the structure of IRNA, Zuker's free energy minimization algorithm was used to generate a structure in which the hairpin loop between nt 36 and 43 would no longer exist. By changing bases 44-46 from 5[prime]-GCA-3[prime] to 5[prime]-UUC-3[prime], such a change was predicted to occur (proposed structure in Fig. 4C). This RNA molecule, called mut1IRNA, was cloned and its structure was studied through RNase H assays (Fig. 6A). The region of the newly created stem, between bases 45 and 59, was studied in particular detail by using oligonucleotides complementary to positions 45-52, 53-59 (Fig. 6A, lanes 6 and 7), 46-54 and 56-63 (data not shown). As predicted, the only of these oligonucleotides capable of base pairing to mut1IRNA and directing cleavage by RNase H is the oligonucleotide complementary to positions 56-63, for bases 60-63 are in a single-stranded region. Thus, the RNase H experiments depicted in Figure 6A suggest that the 36-43 loop of IRNA is abolished upon making a three base change at nucleotide positions 44-46.

Figure 6. (A) Structural analysis of in vitro synthesized mut1IRNA by oligonucleotide hybridization and RNase H digestion followed by primer extension. Nucleotide positions, as deduced by dideoxy sequencing, are shown on the right. (B) Structural analysis of in vitro synthesized, 3[prime]-end-labeled mut2IRNA by nuclease S1, T1 and V1 digestion. Lane A, alkaline hydrolysis ladder; lane T, T1 denaturing ladder; Triangles above lanes indicate increasing amounts of each nuclease. Nucleotide positions, as deduced by alkaline hydrolysis and the T1 denaturing ladder, are shown on the right.

In an effort to change the sequence of bases within IRNA while maintaining the same structure, three bases (positions 24-26) from one side of the main stem of IRNA were exchanged with three bases (positions 63-65) from the other side of the stem. This RNA molecule, called mut2IRNA, was cloned and its structure was studied through S1, T1, V1 and RNase H cleavages. Unexpectedly, the structure of mut2IRNA was altered drastically; the mutation allowed the formation of a new, more stable helix between nt 53-61 and 65-73 as predicted by free energy minimization and as shown by cleavage assays (Figs 4D and 6B). Since reverse transcription of mut2IRNA yielded too many strong stops to allow analysis of cleavage sites (data not shown), mut2IRNA was 3[prime]-end-labeled prior to subjecting it to nuclease cleavage and running on a denaturing sequencing gel. This method revealed, in particular, significant V1 cleavage between positions 53 and 57 and between 65 and 73, suggesting the existence of an extensive helical region between positions 53 and 73 (Fig. 6B). In addition, bases between 53 and 73 are insensitive to cleavage by RNase H after addition of the corresponding oligonucleotides (data not shown). The results of these cleavages, along with free energy minimization using MFOLD, suggest that mut2IRNA assumes a quite different structure than IRNA (compare Fig. 4D with 4A).

Since these mutants were cloned into a different vector, pCDNA3, than IRNA, which was cloned into pGEM3Z, the possibility existed that the bases at the 5[prime]- and 3[prime]-ends of the mutant RNA molecules contributed by the vector sequence was affecting the structure of these mutants. To address this possiblity, IRNA cloned into pCDNA3 in exactly the same way as the mutants was also studied. IRNA derived from pCDNA3 exhibited the same in vitro translation inhibitory activity and a similar binding profile as IRNA derived from the pGEM3Z vector (data not shown). In addition, nuclease digestions and oligonucleotide hybridization/RNase H digestions demonstrated that the structure of the 60 nt of IRNA, whether derived from the pCDNA3 vector or the pGEM3Z vector, was identical (data not shown).

To determine whether alteration of the structure of IRNA affects its inhibitory activity, an in vitro translation assay using the bicistronic reporter construct was performed. As shown in Figure 7A, neither mut1IRNA (lanes 4 and 5) nor mut2IRNA (lanes 6 and 7) significantly inhibit IRES-mediated translation of luciferase in this assay. Quantitation of the luciferase bands, normalized with respect to the CAT bands, reveals that although IRNA at the maximal concentration shown here specifically inhibits IRES-mediated translation by 70%, neither structural mutant specifically inhibits IRES-mediated translation by >1%. Similar results were obtained in a monocistronic assay using the p2CAT reporter construct. Quantitation of the CAT bands from the monocistronic assay using the PV 5[prime]-UTR-CAT template, depicted in the form of a graph in Figure 7B, shows that the inhibitory activity of both structural mutants is far less than that of IRNA. Thus, it appears that alteration of the structure of IRNA can lead to the abrogation of its inhibitory activity.

A    B

Figure 7. (A) Effects of IRNA, mut1IRNA and mut2IRNA on in vitro translation of a bicistronic construct. A bicistronic construct containing CAT and luciferase genes flanked by the PV type 2 5[prime]-UTR was translated in vitro in the absence (lane 1) or presence of 1 (lane 2) or 2 µg IRNA (lane 3), 1 (lane 4) or 2 µg mut1IRNA (lane 5) or 1 (lane 6) or 2 µg mut2IRNA (lane 7). (B) Effects of IRNA, mut1IRNA and mut2IRNA on internal initiation of translation in vitro. A monocistronic construct consisting of the CAT gene preceded by the PV type 2 5[prime]-UTR was translated in vitro in the presence of varying concentrations of IRNA, mut1IRNA, mut2IRNA or a non-specific RNA (yeast tRNA). As in Figure 1B, the percentage of quantitated CAT translation with respect to control (no inhibitory RNA added) was plotted against concentration of inhibitory RNA.

Site-directed mutants of IRNA exhibit an altered protein binding profile

We also attempted to determine whether the UV crosslinked protein binding profiles of these structural mutants were different from that of IRNA and cIRNA. Figure 8 shows the results of such a UV crosslinking experiment in which [[alpha]-³²P]UTP-labeled IRNA (lanes 1-3), mut1IRNA (lanes 4-6) or mut2IRNA (lanes 7-9) were incubated with increasing amounts of HeLa S10 protein. Unexpectedly, mut1IRNA binds La even more strongly than does IRNA or cIRNA. Using purified La protein in UV crosslinking studies, we have found that mut1IRNA, IRNA and cIRNA all specifically bind the La protein and these protein-nucleotidyl complexes co-migrate with the p52 band seen in Figures 2 and 8 (data not shown). In addition, mut1IRNA binds purified La more strongly than does IRNA, a result consistent with that observed in Figure 8 (data not shown). Previous studies have shown that La can stimulate IRES-mediated translation (12,23). Thus, it is surprising that mut1IRNA, which binds La even more strongly than IRNA, inhibits IRES-mediated translation to a much lesser degree than does IRNA. However, these results would be consistent with the idea that La as well as other proteins are necessary for IRES-mediated translation; perhaps mut1IRNA cannot bind one of these other proteins effectively and thus cannot significantly inhibit IRES-mediated translation. In fact, significant reduction of binding of other proteins (p110, p57 and p38) to mut1IRNA is observed compared with IRNA (compare Fig. 8, lanes 2 and 3 with lanes 5 and 6).

Figure 8. IRNA, mut1IRNA and mut2IRNA exhibit different binding profiles. ³²P-labeled IRNA (lanes 1-3), mut1IRNA (lanes 4-6) and mut2IRNA (lane 7-9) were UV crosslinked to cellular polypeptides using either 30 (lanes 2, 5 and 8) or 60 µg (lanes 3, 6 and 9) of HeLa S10 fraction. In lanes 1, 4 and 7, which serve as negative controls, no HeLa S10 fraction was added. Numbers to the left correspond to the migration of marker proteins. The inset shows the results of direct loading of RNA-HeLa S10 complexes onto a 8% acrylamide-8 M urea denaturing gel.

The binding of mut2IRNA to proteins in HeLa extract is also very different from that of IRNA. There is a global decrease in protein binding as measured by UV crosslinking when the 3 nt of the main stem of IRNA are switched with each other (compare Figure 8, lanes 2 and 3 with lanes 8 and 9). The global decrease in binding could be due to a decrease in stability of mut2IRNA under binding conditions as compared with IRNA, rather than an alteration of structure. To rule out this possibility, a parallel binding reaction was set up in which the RNA-HeLa S10 compexes were formed, but instead of being subjected to UV crosslinking the labeled RNA probe was separated from the proteins and analyzed under denaturing conditions in an 8% acrylamide-8 M urea gel. As the inset to Figure 8 demonstrates, there is no significant difference in the stability of IRNA and either of the two mutants during the binding reaction. Thus, it appears that mut2IRNA, although just as stable as IRNA, does not bind HeLa proteins as well as IRNA. The UV crosslinking experiment was repeated with [[alpha]-³²P]CTP-labeled IRNA, mut1IRNA and mut2IRNA in order to rule out the possibility that changes in binding profiles are a result only of changes in the position of U residues between IRNA and the two mutants. [[alpha]-³²P]CT-P-labeled RNAs exhibited a similar binding profile to the [[alpha]-³²P]UTP-labeled RNAs (data not shown). Thus, it appears that the structural differences between IRNA and the two mutants contribute to the differences in protein binding.

DISCUSSION

Through nuclease digestions and oligonucleotide hybridization assays, we have established the secondary structure of a naturally occurring small yeast RNA (IRNA), which was shown previously to block IRES-mediated translation programmed by various viral mRNAs. The IRNA secondary structure appears to consist of two loops, a seven base long stem and a large bulge region. In addition, we have established the secondary structure of cIRNA, which is also capable of blocking IRES-mediated translation and have shown that its overall structure resembles that of IRNA. Both IRNA and cIRNA are capable of preferentially inhibiting IRES-mediated translation, while mutations in IRNA that alter secondary structure disrupt IRNA's inhibitory activity. Maintenance of the established secondary structure correlates well with ability to bind many of the same proteins that bind the poliovirus 5[prime]-UTR and ability to prevent IRES-mediated translation.

The functional similarity between IRNA and cIRNA was surprising to us, as was the ability of both molecules to bind many of the same cellular proteins. Upon inspection of the sequences of IRNA and cIRNA, it was apparent that there are three areas of sequence homology of six or more bases between the two molecules. These areas are 5[prime]-CGCGCG-3[prime] between nt 17 and 22 of IRNA and 64 and 69 of cIRNA, 5[prime]-CGGGUU-3[prime] between nt 20 and 25 of IRNA and 31 and 36 of cIRNA and 5[prime]-CCCGGG-3[prime] present between nt 51 and 56 of IRNA and 29 and 34 of cIRNA. However, it is significant that a mutation in a region of IRNA that does not possess any sequence homology whatsoever to cIRNA (mut1IRNA, where 5[prime]-GCA-3[prime] between nt 44 and 46 was altered) abolished the inhibitory activity of IRNA. Moreover, none of the three 6 nt stretches are present in the PV 5[prime]-UTR. These observations led us to explore the possibility that secondary structure plays an important role in the activities of both IRNA and cIRNA.

The nucleotides that form the particular secondary structure elements of IRNA do not correspond to the nucleotides of cIRNA that form its stems, loops and bulges. For instance, IRNA loop 35-44 is comprised of the sequence 5[prime]-CAGAACAGCG-3[prime]. The complementary sequence to this, 5[prime]-CGCUGUUCUG-3[prime], is present in cIRNA not in a loop region but predominantly in a helical region, spanning nt 41-50. However, upon inspection of the secondary structures of IRNA and cIRNA (Fig. 4A and B), it appears that the overall structures of the two molecules are similar, with the various structural elements oriented with respect to each other in similar ways. It should be pointed out that although the overall structures of IRNA and cIRNA are similar, various differences exist upon closer inspection of the molecules. Specifically, the main bulge of IRNA is much larger than that of cIRNA. In addition, there is a 4 nt bulge just upstream of the main stem in cIRNA (nt 35-38), while the bulge upstream of IRNA's main stem is comprised of only 1 nt (nt 19).

How might the structures of IRNA and cIRNA allow these molecules to bind similar proteins to the PV 5[prime]-UTR? One possibility is that the secondary structure of IRNA mimics a single portion of the PV 5[prime]-UTR, thereby allowing it to bind similar proteins to that particular region of the 5[prime]-UTR. Alternatively, IRNA may assume a completely different structure from the PV 5[prime]-UTR that allows it to interact with a variety of proteins that are bound by many different structural elements over the entire PV 5[prime]-UTR. We are currently attempting to distinguish between these possibilities by probing the structures of various regions of the PV 5[prime]-UTR that are predicted, by free energy minimization and other structure analyses (3), to assume a secondary structure similar to that of IRNA.

IRNA and cIRNA both inhibit IRES-mediated translation, but to differing degrees. Figure 1A shows that IRNA is a more potent inhibitor of IRES-mediated translation, but that it also affects cap-dependent translation more than does cIRNA. For instance, at a concentration of 1 µg of IRNA and 2 µg of cIRNA (lanes 2 and 5, respectively), both RNAs inhibit IRES-mediated translation, as measured by the intensity of the luciferase bands, to similar extents. At these same concentrations of RNAs, though, IRNA appears to inhibit cap-dependent translation, as measured by intensity of the CAT bands, more than does cIRNA (compare lanes 2 and 5). These subtle differences in translational inhibition by the two RNAs may be due to the presence of slight structural differences between IRNA and cIRNA (for instance, the larger bulge in IRNA as compared with cIRNA). These structural differences may also account for the fact that the protein binding profiles of the two molecules are similar, but not identical. However, we do not rule out the possibility that in addition to secondary structural elements, sequence-specific interactions may play a role in the ability of these molecules to inhibit cap-dependent and cap-independent translation; the differences in sequence between IRNA and cIRNA, then, may also cause some differences in both translational inhibition and protein binding.

Mut1IRNA lacks the small loop (nt 36-43) and binds La efficiently, but is defective in interacting with other polypeptides (p110, p57 and p38; Figs 4C and 8). It is possible that this loop is not involved in La binding but may play an important role in binding other relevant proteins. However, we cannot differentiate between this scenario and the possibility that the altered structure of mut1IRNA enables it to bind so much La that the binding of other proteins is impaired. Indeed, such an `exclusionary' role for La has been posited in a different context, where coating of mRNAs with La has been proposed to prevent binding of ribosomal initiation factors (34). Both of these possibilities are consistent with the idea that although La binding is important for IRNA's ability to inhibit IRES-mediated translation (23), the binding of other factors to IRNA is also necessary.

In the case of mut2IRNA, the swapping of complementary sequences (5[prime]-UUU-3[prime] and 5[prime]-AAG-3[prime]) within the stem of IRNA totally alters its structure due to the formation of a new helix. mut2IRNA is highly defective in binding almost all polypeptides, including La, that normally interact with IRNA. One possibility for this global decrease in protein binding is that the structure has been altered so extensively that these proteins are no longer capable of readily recognizing the RNA molecule. However, another possibility is that mut2IRNA is deficient in binding a protein that normally recruits the other proteins to IRNA; thus, a deficiency in binding of this protein leads to a global decrease in binding by mut2IRNA. Taken together, these results suggest that while interactions of La with the IRES (or IRNA) may be important, additional factors are almost certainly involved in IRES-mediated translation. Additionally, these results underscore the importance of the overall secondary structure of IRNA in protein binding and consequent inhibition of IRES-mediated translation. Although we have shown that the stem and one loop of IRNA are important for protein binding and inhibitory activity, we are currently attempting to more finely dissect IRNA structures involved in binding of each of the polypeptides seen in Figure 2A. Future studies directed at cloning these additional polypeptides and exploring their interactions with IRNA and IRES elements should aid in the elucidation of the mechanism of IRES-mediated translation.

We have not found, to date, small mammalian RNAs possessing sequences similar to yeast IRNA. What, then, is the function of IRNA in yeast? It is tempting to speculate that IRNA is involved in regulation of translation in yeast. In fact, we have recently found that IRNA specifically interacts with a number of yeast proteins, all of which are involved in translation. Future studies, including determination of IRNA expression patterns in yeast as well as knockout of IRNA, will shed light on the physiological role of this molecule in yeast.

ACKNOWLEDGEMENTS

This work was supported by NIH grant AI38056 to A.D. A.V. was a Howard Hughes Medical Institute Medical Student Research Training Fellow and was also supported by NIH Medical Scientist Training Program grant NRSA/GM08042-15. We are grateful to Mr Weimin Tsai for his superb technical help.

REFERENCES

1. Pelletier,J. and Sonenberg,N. (1988) Nature, 334, 320-325. MEDLINE Abstract

2. Jang,S.K., Krausslich,H.G., Nicklin,M.J.H., Duke,G.M., Palmenberg,A.C. and Wimmer,E. (1988) J. Virol., 62, 2636-2643. MEDLINE Abstract

3. Jackson,R.J. (1996) In Hershey,J.W.B., Mathews,M.B. and Sonenberg,N. (eds), Translational Control. Cold Spring Harbor Laboratory Press,Cold Spring Harbor, NY.

4. Ehrenfeld,E. (1996) In Hershey,J.W.B., Mathews,M.B. and Sonenberg,N. (eds), Translational Control. Cold Spring Harbor Laboratory Press,Cold Spring Harbor, NY.

5. Poyry,T., Kinnunen,L. and Hovi,T. (1992) J. Virol., 66, 5313-5319. MEDLINE Abstract

6. Haller,A.A. and Semler,B.L. (1992) J. Virol., 66, 5075-5086. MEDLINE Abstract

7. Hoffman,M.A. and Palmenberg,A.C. (1996) J. Virol., 70, 6425-6430. MEDLINE Abstract

8. Le,S.-Y., Siddiqui,A. and Maizel,J.V. (1996) Virus Genes, 12, 135-147. MEDLINE Abstract

9. Meerovich,K., Svitkin,Y.V., Lee,H.S., Lejbkowicz,F., Kenan,D.J., Chan,E.K., Agol,V.I., Keene,J.D. and Sonenberg,N. (1993) J. Virol., 67, 3798-3807.

10. Ali,N. and Siddiqui,A. (1997) Proc. Natl Acad. Sci. USA, 94, 2249-2254. MEDLINE Abstract

11. Hellen,C.,Witherell,G.W., Schmid,M., Shin,S.H., Pestova,T.B., Gil,A. and Wimmer E. (1993) Proc. Natl Acad. Sci. USA, 90, 7642-7646. MEDLINE Abstract

12. Ali,N. and Siddiqui,A. (1996) J. Virol., 69, 6367-6375.

13. Niepmann,M., Petersen,A., Meer,K. and Beck,E. (1997) J. Virol., 71, 8330-8339. MEDLINE Abstract

14. Blyn,L.B., Swiderek,K.M., Richards,O., Stahl,D.C., Semler,B.L. and Ehrenfeld,E. (1996) Proc. Natl Acad. Sci. USA, 93, 11115-11120. MEDLINE Abstract

15. Gamarnik,A.V. and Andino,R. (1997) RNA, 3, 882-892. MEDLINE Abstract

16. Schultz,D.B., Hardin,C.C. and Lemon,S.M. 1996 (1996) J. Biol. Chem., 271, 14134-14142. MEDLINE Abstract

17. Svitkin,Y.V., Lee,H.S., Dholakia,J.N., Kenan,D.J., Agol,V.I. and Sonenberg,N. (1994) J. Virol., 68, 1544-1550. MEDLINE Abstract

18. Kaminski,A., Hunt,S.L., Patton,J.G. and Jackson,R.J. (1995) RNA, 1, 924-938. MEDLINE Abstract

19. Blyn,L.B., Towner,J.S., Semler,B.L. and Ehrenfeld,E. (1997) J. Virol., 71, 6243-6246. MEDLINE Abstract

20. Coward,P. and Dasgupta,A. (1992) J. Virol., 66, 286-295. MEDLINE Abstract

21. Das,S., Coward,P. and Dasgupta,A. (1994) J. Virol., 68, 7200-7211. MEDLINE Abstract

22. Das,S., Ott,M., Yamane,A., Tsai,W., Gromeier,M., Lahser,F., Gupta,S. and Dasgupta,A. (1998) J. Virol., 72, 5638-5647. MEDLINE Abstract

23. Das,S., Kenan,D.J., Bocksai,D., Keene,J.D. and Dasgupta,A. (1996)J. Virol., 70, 1624-1632. MEDLINE Abstract

24. Zuker,M. (1989) Science, 244, 48-52. MEDLINE Abstract

25. Moazed,D., Stern,S. and Noller,H.F. (1985) J. Mol. Biol., 187, 399-416.

26. Stern,S., Moazed,D. and Noller,H. (1988) Methods Enzymol., 164, 481. MEDLINE Abstract

27. Knapp,G. (1989) Methods Enzymol., 180, 192-212. MEDLINE Abstract

28. Jacobson,S., Konings,D.A. and Sarnow,P. (1993) J. Virol., 67, 2961-2971. MEDLINE Abstract

29. Tabernero,C., Zolotukhin,A.S., Valentin,A., Pavlakis,G.N. and Felber,B.K. (1996) J. Virol., 70, 5998-6011. MEDLINE Abstract

30. Guo,H.C.T. and Collins,R.A. (1995) EMBO J., 14, 368-376. MEDLINE Abstract

31. Donis-Keller,H. (1979) Nucleic Acids Res., 7, 179-192. MEDLINE Abstract

32. Lowman,H.B. and Draper,D.E. (1986) J. Biol. Chem., 261, 5396-5403. MEDLINE Abstract

33. Celander,D.W. and Cech,T.R. (1990) Biochemistry, 29, 1355-1361. MEDLINE Abstract

34. Svitkin,Y.V., Ovchinnikov,L.P., Dreyfuss,G. and Sonenberg,N. (1996) EMBO J., 15, 7147-7155. MEDLINE Abstract

---

*To whom correspondence should be addressed. Tel: +1 310 206 8649; Fax: +1 310 206 3865; Email: dasgupta@ucla.edu

---

This page is run by Oxford University Press, Great Clarendon Street, Oxford OX2 6DP, as part of the OUP Journals
Comments and feedback: www-admin{at}oup.co.uk
Last modification: 23 Dec 1998
Copyright©Oxford University Press, 1998.
CiteULike    Connotea    Del.icio.us    What's this?

This article has been cited by other articles:

				E. Martínez-Salas, R. Ramos, E. Lafuente, and S. López de Quinto Functional interactions in internal translation initiation directed by viral and cellular IRES elements J. Gen. Virol., May 1, 2001; 82(5): 973 - 984. [Full Text]

				A. Venkatesan and A. Dasgupta Novel Fluorescence-Based Screen To Identify Small Synthetic Internal Ribosome Entry Site Elements Mol. Cell. Biol., April 15, 2001; 21(8): 2826 - 2837. [Abstract] [Full Text]

				C. B. Buck, X. Shen, M. A. Egan, T. C. Pierson, C. M. Walker, and R. F. Siliciano The Human Immunodeficiency Virus Type 1 gag Gene Encodes an Internal Ribosome Entry Site J. Virol., January 1, 2001; 75(1): 181 - 191. [Abstract] [Full Text]

				M. Gale Jr., S.-L. Tan, and M. G. Katze Translational Control of Viral Gene Expression in Eukaryotes Microbiol. Mol. Biol. Rev., June 1, 2000; 64(2): 239 - 280. [Abstract] [Full Text] [PDF]

This Article Abstract Print PDF (564K) Alert me when this article is cited Alert me if a correction is posted Services Email this article to a friend Similar articles in this journal Similar articles in ISI Web of Science Similar articles in PubMed Alert me to new issues of the journal Add to My Personal Archive Download to citation manager Search for citing articles in: ISI Web of Science (8) Request Permissions Commercial Re-use Guidelines for Open Access NAR Content Google Scholar Articles by Venkatesan, A. Articles by Dasgupta, A. Search for Related Content PubMed PubMed Citation Articles by Venkatesan, A. Articles by Dasgupta, A. Social Bookmarking          What's this?